American Journal o 26 Wist

Se
aa
Sea
3
a,
Sareea
Cy
$
shah
eae
i
pairs)
=
=
fat at
aS
RSaReo
ee
==:
as
oe
ee
:
iNtate
f
adentagndon
7 ahh?
=
yrorten
ua}ty AU yh
Highs
ne
ennwemtonn
eSth)y ttt
ara
‘ tte)gepahe
pat
Lt) iY wert) rf
iP
>
Patiny Hoa
dsabpeepye N
I
oe
yo
r
ens
:
ee
=
;
eanemnnanis
: ‘
‘
: :
aes
faieimREE : 2
Sie.
yet
:
Solos -getey
y
ai
= ae
be ' yet
pesth
eS
SSeS
ae
ppithrieye
as
ate
Varfhcarysceee aa
ee
Paced
Hotesetebratessty ey
; Uetys eed
Laude prtaztipyinss
Uprebieat if TERED:
gyhthy
Hei
yeatis
tree
3D
tn
puateathe
—
By) = =
eRe
{
min) "‘ ¥
\( jf! Wy,
Woy
iPW
his Wea} WA
Devan
SMa
y " HI ee MRI Oy i
st
jHib)
f
PNET.
et
=;
Ss
oS
g
Mi
ny Ne
eS TN, BAY i
THE AMERICAN JOURNAL
OF
ANATOMY
EDITORIAL BOARD
Cuarues R. BARDEEN G. Cart Huser J. Puayrain McMurrica

University of Wisconsin University of Michigan : University of Toronto
Henry H. DoNALDSON GEORGE 8S. HUNTINGTON ’ GrorGe A. PIERSOL

The Wistar Institute Columbia University University of Pennsylvania
Simon H. GaaGeE Henry Mck. KNowir, SECRETARY

Cornell University University of Cincinnati
VOLUME 26
SEPTEMBER, 1919—JANUARY, 1920
THE WISTAR INSTITUTE OF ANATOMY AND BIOLOGY

PHILADELPHIA, PA.
a
} 1
'
: :
}¢ : P
yy
b
ae
ean
i)
'
;
ry j
A iis A
AY nis aed J Wieepiagali ae ae

(mires MA doy : ! (ligay 2% <b
7 ' ‘ =
Pa tO CRD ets py Whe? a C

TCE! Gansu Yo qin nagh y ; 1) plow ’ »!aaa
pe 45 Roe A ©eit) : nit
SR) oh ‘Queen
a g a i n s p i : oe
lie
—— «CT
' i +: : ,> 5
ee
CONTENTS
No. 1. SEPTEMBER, 1919
H. E. Jorpan. The histology of the umbilical cord of the pig, with special reference to
the vasculogenic and hemopoietic activity of its extensively vascularized connective
tissue. Fifteen figures.. , 1
Henry A. Murray, JR. The jenalansen a the Gaediac lageinAahe ane oe epee
reference to the bulboventricular groove and origin of the faitemyontn cular septum.
Seven figures........ 29
FRANK Buatr Hanson. The pntocens, onl1 phylogeny es dite een mele nlates
(forty-nine figures).. Ef Meilegen oie > Al
GrorGcE W. Corner. On che origin oiSheraT mie oF iesSOW on both eenlon
BuGeuneca mternae. 1 wenty=s1X NEUES... o> 3..ccaes cele e mega 4 ques acmenscteonsanan
cc: LLG
No. 2. NOVEMBER, 1919

Harowtp J. Cooper. The hypophysis cerebri of the California Sa eas Citellus
beechyi (Richardson). Eleven figures.. x . 185
RautpH Dovueatt Litiur. The early histogenesis a ihe Bleed in Baro falophitns Bard
and Girard. Seven figures. ee: .. 209
Henry H. Dona.pson. Cronica oindiiag on he aoe of fhe peieion ai he siishine
PP PRICENIG=EHTCE |CHALUSH 65 «5.2.15 1-1Se ance cis oe ocdss ars ovate aval sais Fens ne oiatee aepetoke gable Lou
No. 3. JANUARY, 1920

Lestiz B. Arzy. The origin, growth, and fate of osteoclasts and their relation to bone
FCNOLPULOH ae PnWeMEeeLOUN TOUTES, «35. sate salal iene eka eee chat Ae as ate se neal da adk oc 315
S. Saacucur. Studies on the glandular cells of the frog’s pancreas. Five plates........ 347
FrepeEric T. Lewis. The course of the Wolffian tubules in mammalian embryos. Thir-
SeEMEA SUITCS <5 fo RN eels co's iesShoo gpsn SuldeSoets Gusrseayn ENT atsSteal abrSa [oN was de 423
Wiuuiam H. F. Appison. Histological study of the spleen of the rabbit under height-
ened phagocytic activity. Six figures (one plate)...............cccccceceecteeeeees 437
!
'
i : ‘
a
j t
4] pear) i f 4 ( Ni bod)
3 aed
f 1 i wir?
=) ae 4
r 5 L
; i i | ue
ae es
i Lar
eens hs
4 ?
; F ' i bn
i J hipaa oll iter Apri te } ert re , pe,
ay ea bf + BA shea
valtanae a
mini 7 fei Taka manip Ewes sds A tke iF “ill a iymera '
int ‘ ; as! ‘is ‘ A ‘ f ‘ if Be A
MA, ey Pe ok Me nace Pane
‘af werk ane le yi.
if 4
o) Sel, ihre cular,wit

AUTHOR’S ABSTRACT OF THIS PAPER ISSUED
BY THE BIBLIOGRAPHIC SERVICE, JULY 7
THE HISTOLOGY OF THE UMBILICAL CORD OF THE

PIG, WITH SPECIAL REFERENCE TO THE VASCULO-
GENIC AND HEMOPOIRTIC ACTIVITY OF ITS EXTEN-
SIVELY VASCULARIZED CONNECTIVE TISSUE
H. E. JORDAN
Department of Anatomy, University of Virginia
FIFTEEN FIGURES
The umbilical cord of the pig differs markedly from the human
cord in that it is extensively vascularized, its connective tissue
maintains throughout the gestation period largely its original
embryonal character, and vasculogenic and hemopoietic activity
persist to full term.
The material upon which this investigation is chiefly based is
a complete cord, 4 cm. in length, from a fetus of 16-cm. length.
This fetus was secured in utero at the abattoir and preserved in
10 per cent formalin. Reckoned by length, it lacked between
one and two weeks of full term. Portions of the cord from the
proximal end, the middle, and the distal end, were imbedded
both in celloidin and in paraffin. Serial sections were cut from
the paraffin blocks and stained with hematoxylin and eosin.
Some of the celloidinsections were similarly stained; others were
stained with resorcin-fuchsin and counterstained with picric-
acid-fuchsin, for a study of the elastic and collagen fiber content.
A second specimen of a nearly full-term cord, for which I am
indebted to Prof. George S. Huntington, was used for comparison.
Cords of pig embryos from 9 to 21-mm. length and five full-term
and three fetal (three to seven months) human cords, variously
fixed and stained, were also employed for comparative study.
Though primary interest centers upon the vasculogenic and
hemopoietic activity of the connective tissue, it seems desirable
to preface the description and discussion of these phenomena
1
2 H. E. JORDAN
with a brief description of the comparative histology of this

cord. Compared with the human cord it is very short, of con-
siderably lesser girth, and only slightly twisted. It has the
same light gray, pearly appearance, and feels of about the same
consistency. In transverse section it has an irregularly oval
shape (fig. 1), measuring 5 by 7 mm. Its three main blood-
vessels have an approximately equal caliber and thickness of
wall. It contains a large open allantoic duct and remnants of
the occluded yolk-stalk. The connective tissue contains many
arterioles, venules, and capillaries. Only one of the post-em-
bryonic human umbilical cords in my collection, a full-term
specimen, contains any blood-vessels besides the usual umbilical
arteries and vein. In this cord occurs a venule of considerable
size, lying near the surface and completely filled with red blood-
corpuscles. The human umbilical cord is typically non-vascular
except for an occasional capillary at the extreme proximal end.
None of my sections of these eight human cords contains any
vestige of the allantois. One cord contains a small, still patent
yolk-stalk; three contain a double, occluded yolk-stalk remnant.
One of the full-term cords of the pig also contains a double oc-
cluded yolk-stalk (figs. 1 and 3), the other only a single, small,
occluded remnant in only a few sections. In one of the human
cords the persistent, double, occluded yolk-stalk is enveloped
by a double layer of smooth muscle, an inner longitudinal and
an outer thinner circular layer.
THE UMBILICAL ARTERIES AND VEIN
In the cord of the pig the wall of the umbilical arteries only
contains circularly disposed smooth-muscle cells, more compact
centrally; the vein in one of the two specimens contains also
scattered, longitudinally placed cells beneath the intima. The
disposition of the muscle differs from that in the human cord
where the arteries and the vein contain, in addition to the chief
circular layer, also a thin internal longitudinal layer and scat-
tered bundles of longitudinally arranged cells externally. With
regard to the elastic tissue content also, the pig’s cord differs
sharply from the human cord. The arteries of the latter lack
UMBILICAL CORD OF THE PIG 3
an internal elastic membrane, but several muscle layers beyond

the tunica intima the elastic tissue forms complete fenestrated
membranes through many layers; toward the periphery of these
vessels the elastic tissue only occurs as scattered delicate fibrils.
The vein, on the contrary, contains a very robust internal elastic
membrane, while through the central half of the wall occur rel-
atively coarse scattered fibers. In the pig’s cord neither arteries
nor vein contain an internal elastic membrane. ‘The elastic
fibers are practically limited to the inner half of the wall, only
very delicate and widely scattered fibrils occurring peripherally.
In the arteries the elastic tissue forms membranes for from three
to five layers beyond the central two or three layers. Consider-
able variation occurs with regard both to the amount and the
disposition of both the smooth muscle and the elastic tissue
constituents of the wall of the umbilical vessels both in the pig’s
cord and the human cord.
ECTODERM
The covering ectoderm constitutes a stratified epithelium of

generally four layers of cells (fig. 2). In certain restricted areas
the epithelium only consists of three layers of cells, in others of
as many as eight layers of cells. This epithelium resembles the
transitional rather than the stratified squamous type. It is
comparable to the thicker portions of the epidermis of the three-
month human fetus, which consists of from four to six layers of
cells, including a superficial periderm. It differs from the ecto-
derm of the full-term human cord, which includes only two or
three layers of flattened keratized cells. It differs also from the
continuous abdominal ectoderm in that the latter includes about
eight layers of cells, all of which, except the basal cuboidal layer,
consist of greatly flattened cells. Small patches of partially
fused, keratized cells occur in five different regions of the section
here shown. The lowermost layer of the epithelium consists
of cuboidal cells; the outermost variously of thick, rectangular,
flattened or dome-shaped peridermal cells; the intervening layers
include polyhedral and stout fusiform types. In those portions
4 H. E. JORDAN
where the epithelium consists of more than four layers, the one
next the basal cuboidal layer is generally composed of more
flattened cells. The basal layer seems to rest directly upon the
adjacent connective tissue, without the intervention of a definite
basement membrane. After picric-acid-fuchsin counterstain,
however, a narrow subepithelial layer of the connective tissue
stains more deeply red.
None of the cells contains mitotic figures. An occasional cell
of the superficial layer contains two or even four nuclei or a
nucleus in process of fission. The nuclei of the cells of the basal
layer are of spheroidal shape and contain a clear, lightly staining
nucleoplasm and a distinct granular reticulum. ‘These cells are
completely filled with a slightly basophilic cytoplasm. The
nuclei of the several outer layers are irregular in shape and they
have a homogeneous, cloudy, more deep-staining character, and
the nuclear wall is generally wrinkled. The cytoplasm of the
cells of the intermediate layers is aggregated next the cell wall so
that the cell appears hollow. This condition is probably a fix-
ation artifact. The nucleus almost invariably lies next the
outer wall, as if moved by currents passing toward the surface
of the ectoderm. The cells of the outermost layer are again
completely filled with an acidophilic cytoplasm. The latter is
keratized to an extent which could resist the action of the fixation
currents that caused the peripheral shrinkage and central excava-
tion of the cells of the intermediate layers.
YOLK-STALK
The yolk-stalk remnant of one of the two practically full-term

specimens of the pig’s cord, and of four of the specimens of human
cord used for comparative study, is a double structure. In the
specimen of the pig’s cord the two portions are of nearly equal
size (fig. 3). They are approximately circular in outline and
perfectly solid. The cells of the peripheral layer are squamous
or very low cuboidal. The more central cells are polyhedral er
spheroidal. The cells next the outermost layer are generally
flattened and appear ,fusiform in longitudinal section. About
twelve cells occupy the diameter of each of the two elements.

Most of the central cells appear hollow, the nuclei having become
moved to one side, almost invariably to that side nearest the
center of the structure. These cells have a superficial resem-
blance to fat cells. The hollow condition is probably the result
of the coagulative action of the fixing fluid upon the delicate
cytoplasm. A few of the cells appear keratized, and are acido-
philic in staining reaction, and solid. The nucleiof the outer-
most cuboidal cells are deep-staining, granular, greatly elongated
bodies. Those of the more central cells have a generally irregu-
lar shape with wrinkled contour and a generally non-granular
homogeneous nucleoplasm. No mitotic or amitotic figures can be
detected. Each member of the double structure is enveloped by
a very thin inner connective-tissue theca, forming a delicate,
fibrillar, nucleated basement membrane. Both are inclosed in
a common, more external theca. The intervening partition only
consists of the fused basement membranes.
The fact that four of the five specimens of human cords used
in connection with this study also contain a double yolk-stalk
indicates that this doubling is a common condition. ‘Two ex-
planations suggest themselves: 1) That the doubling is due to
the partition of the originally single stalk by the ingrowth of a
connective-tissue septum related to the regressive changes by
which the stalk becomes obliterated. 2) That the ‘doubling’
is only apparent, it being due in section to a sharp turning of
the stalk in certain regions. The latter interpretation is sup-
ported by the evidence from one of the human cords: here one
of the ‘halves’ is cut transversely, the other half is very
obliquely cut, and the two are in continuity. In other words,
the condition is such as would result if the section passed ob-
liquely through the proximal end of one of the limbs and the
connecting loop of a U-shaped structure. Opposed to the latter
interpretation, however, is the fact that in the specimen of the
pig’s cord here described a common connective-tissue sheath
envelops both moieties.
6 H. E. JORDAN
ALLANTOIC DUCT
The allantoic duct (fig. 1, All.) is shown highly magnified in

figure 4. It has an irregularly oval shape in cross-section. Its
wall is thrown into deep folds. Within the depth of the folds
the epithelium consists of a single layer of cuboidal or even
squamous cells, comparable to its condition throughout in the
21-mm. fetus; over the crests the epithelium is of the stratified
columnar type, consisting of from three to four layers of cells.
No distinct basement membrane can be discerned other than as
indicated by the deeper red color of the immediately subjacent
connective tissue after picric-acid-fuchsin counterstain. The
nuclei have an irregularly oval form; their wall is wrinkled, and
the nucleoplasm generally lacks a distinct network or granules.
A number of the cells are hollowed out centrally. The cells next
the lumen have a more condensed, probably slightly keratized,
broad distal border. The enveloping connective tissue is less
differentiated and denser than in any other portion of the section.
It resembles early embryonic connective tissue or young mesen-
chyma. It contains many vasofactive cells (‘angioblasts’), and
one large blood-island (B. J.), which will be described below.
THE CONNECTIVE TISSUE
The character of the connective tissue varies in different

portions of the transverse section (compare figs. 2, 3,4, and 5).
In the region immediately surrounding the allantoic duct (fig. 1)
it is least differentiated. Here it is compact and resembles
embryonic connective tissue or young mesenchyma (fig. 4). In
this region also are numerous vasofactive cells (fig. 15). Along
the peripheral border of the specimen the connective tissue is
somewhat more differentiated and represents an older type of
mesenchyma (fig. 2). Here the cells have generally a stellate
or fusiform shape and are more widely separated. In this region
also the capillaries are most abundant; these for the most part
lie along the radii of the section and appear to be growing towards
the ectodermal covering. Here also are found abundantly initial
stages in the formation of the blood channels (figs. 8, 9, and 10).
In the area between the lower umbilical artery on the left and the
vein on the right, the connective tissue is in part of the type
characteristic of the human cord; that is, it is typical mucous
connective tissue, but with an occasional capillary. In this
region also occur small bundles of collagen fibrils. The area
around the yolk-stalk contains vascular connective tissue of
an intermediate type, with occasional collagen fibers (fig. 5).
In the narrow space between the two umbilical arteries, extending
to a point below the allantoic duct, there occur several blood-
islands (figs. 6 and 7). About midway between the central
umbilical blood-vessels and the periphery occur numerous arte-
rioles and venules. Occasionally these are arranged in pairs
(fig. 5). These vessels terminate in capillaries. The arterioles
are enveloped by a thin layer of smooth muscle; the wall of the
venule only consists of endothelium resting upon the slightly
more condensed enveloping connective tissue.
Certain of these smaller blood-vessels can be traced into con-
nection with the umbilical arteries and the vein at their proximal
(fetal) end. This is true both in the case of the younger cords
(of embryos from 9 to 21 mm.) and in those near full term. In
the 21-mm. fetus branches from the proximal end of the umbilical
vein can be seen entering the body wall as well as the connective
tissue of the cord. It may be confidently assumed that all of
the blood-vessels, including the numerous capillaries, connect with
vessels which ultimately connect with the main umbilical vessels
proximally. But not all of these vessels are properly interpreted
as vasa vasorum. Undoubtedly many function thus, as is indi-
cated by the numerous capillaries directed towards the walls of
the umbilical arteries and vein; but others are equally certainly
nutrient vessels for the allantoic duct, the ectodermal covering,
and the general connective tissue. Nor can the assumption be
properly made that all of these vessels arise by sprouting from
originally direct umbilical branches. If the capillaries grew ex-
clusively by sprouting, their tips should show mitotic figures.
Mitotic figures are practically absent in these capillary terminals.
On the contrary, these tips seem to fuse with the general con-
nective tissue, the cells of which become hollowed out, arrange
8 H. E. JORDAN
themselves in line with the capillaries and ultimately become

incorporated as part of the capillary network. In the full-term
condition of the cord, blood-vessels arise in a manner identical
with their primary origin in the original body-stalk, and become
secondarily connected with the preexisting vascular net. The
connective tissue of the full-term cord maintains the primitive
vasculogenic mode by which the primitive blood-vessels were
formed. ‘The full-term cord is relatively much more extensively
vascularized, and it consists of connective tissue largely of a
less differentiated type than the cord of the 21-mm. fetus. Since
the larger blood-vessels extend to the distal end of the cord, it
may be inferred that they supply also the proximal pole of the
allantois. _
VASCULOGENESIS AND HEMOPOIESIS
In the description of the mode of vasculogenesis illustrated

in this cord, we may begin most conveniently with the stage
represented. by the cell of figure 8. This cell has the general
characteristics of an irregular mesenchymal cell. Three vacuoles
can be seen to the right of the nucleus. <A later stage may be
represented by the upper cell of figure 9. Here the cell is binucle-
ated, and the originally smaller discrete vacuoles have presumably
coalesced to form a single large vacuole, the precursor of the
initial capillary lumen. The appearance of the lumen has ef-
fected a modification of one of the nuclei so that it begins to
assume endothelial features. In figure 10 is shown a still later
phase of vasculogenesis. Here, moreover, the more typical endo-
thelial ‘cell’ has taken on hemoblast features and has differentiated
an erythroplastid (ep.) intracellularly. Figure 11 may be con-
ceived to represent a transection of the cell of figure 9 or 10.
Cells like those of figures 8, 9, 10, and 11 are very numerous in
the more peripheral regions of the cross-section. Figure 12
illustrates a binucleated cell with essentially mesenchymal fea-
tures, which has differentiated an erythroplastid and a lumen,
centrally. The cell shown in figure 13 has the nuclear and cyto-
plasmic characteristics more of a young hemoblast. This cell
represents a mesenchymal cell which has rounded up and differen-
tiated into a hemoblast, and subsequently differentiated an

erythroplastid intracellularly.
The mesenchymal cells of this primitive ‘mucous’ connective
tissue may apparently undergo any one of several types of dif-
ferentiation: 1) They may separate from the mesenchymal
syncytium, round up and differentiate into potential hemoblasts,
which may le freely among the undifferentiated mesenchymal
cells, but apparently never in this condition directly metamor-
phose into erythroplastids; but grouped into blood-islands, about
which the adjacent mesenchyme differentiates into endothelium,
they develop into erythroblasts (fig. 7). 2) They may become
bi- or multinucleated and, as hemogenic giant-cells, differentiate
erythrocytes intracellularly (figs. 6, e, and 15, f). 3) A mesen-
chymal cell may acquire a lumen and join with other cells to
form an initial capillary, incidentally differentiating also
erythroplastids intracellularly (fig. 12). Erythroplastids may
originate intracellularly also in young endothelial cells (fig. 10).
Hemoblasts can, therefore, apparently differentiate into erythro-
cytes only when inclosed by endothelium; or in the multinucleated
condition, hemoblasts can differentiate intracellular erythrocytes.
The latter phenomenon is essentially like that where a hemo-
blast is inclosed by endothelium.
Figures 6 and 7 illustrate an earlier and later stage, respectively,
in the differentiation of a blood-island. In figure 6 the blood-
island is still largely a syncytium. However, endothelium can
be seen forming on its surface, and several of its cells are taking
on erythroblast (‘megaloblast’) features (a, 6, and d). The cell
d has developed a large vacuole at one pole; this vacuole may
form part of the subsequent lumen. Cell e has differentiated
an erythrocyte. Several intercellular spaces have appeared in
the syncytium; these are the forerunner of the subsequent lumen,
to which certain intracellular spaces may also contribute. In
figure 7 the endothelium and the lumen are developed further,
and the hemoblasts are mostly in the erythroblast stage and
are generally separated from each other by cell membranes.
It seems desirable at this point to indicate the chief differ-
ences, nuclear and cytoplasmic, between the young mesenchymal
10 H. E. JORDAN
cell, young endothelial cell, hemoblast, erythroblast (‘megalo-

blast’), and the erythrocyte. The mesenchymal cell has in
general a light-staining, spheroidal or oval nucleus with a delicate
reticulum; its cytoplasm contains delicate fibrillae. The young
endothelial cell has a similarly light-staining, but generally more
elongated nucleus; and its cytoplasm is less distinctly fibrillar.
The hemoblast is generally spheroidal in shape, but it may
assume various forms due to its ameboid capacity; its nucleus
also generally has a spheroidal shape, but it contains a more
distinct and more granular network, and the cytoplasm appears
homogeneous. The nucleus of the young erythroblast (‘megalo-
blast’) has a spheroidal shape, a robust chromatic membrane, a
generally deeper-staining nucleoplasm; and it contains one or
several small nucleoli and numerous granules scattered over its
delicate reticulum. Its cytoplasm is distinctly granular, and
it stains deeply in eosin. The granular condition of the cyto-
plasm of the erythroblast is the most distinctive mark of this
cell. This cell corresponds to the megaloblast of certain writers.
The erythrocyte has a considerably smaller, generally deeper-
staining, spherical nucleus, and a clear cytoplasm delimited by
a distinct membrane (figs. 6, e, and 15,b ande). The erythro-
plastid has in contrast a brownish-yellow color.
In figure 15 (a to 7) are illustrated various vasofactive and
hemogenic cells. In fact, as these figures clearly indicate, vaso-
factive and hemogenic activities are intimately associated. The
cell a may be regarded as at the stage of a late hemoblast or a
young erythroblast. The cell 6 is similar, but has produced an
intracellular erythrocyte. Cell c has developed a lumen, and
it has differentiated an inclosed erythroplastid. Cell d is es-
sentially a young endothelial cell with vestiges of cytoplasmic
hemoblast features. Cell e has become essentially a young
endothelial cell with an included small erythrocyte and an
erythroplastid. Cells g andi are essentially hemoblasts (‘angio-
blasts’) which have become differentiated into binucleated
endothelial cells. Cells f and h should be interpreted together;
h is essentially a multinucleated hemoblast or small ‘giant-cell,’
one of whose nuclei is apparently undergoing amitotic division;
UMBILICAL CORD OF THE PIG ti
f may be regarded as a later stage in the intracellular eryth-

rocytogenic function of h, in which two of the nuclei and their
enveloping cytoplasm have differentiated into erythroplastids
(ep). Cells h, f, 6, and e of figure 15 and e of figure 6, when
considered in common, demonstrate that erythroplastids in these
vasofactive cells do not arise as such out of the cytoplasm, but
under the direct influence of a nucleus of a bi- or multinucleated
hemoblast (‘hemogenic giant-cell’). The absence of free eryth-
rocytes and erythroplastids in the regions from which these cells
are taken contravenes any suggestion that the intracellular red
blood-corpuscles should be interpreted in terms of phagocytosis.
Figures 12 and 14 supply similar evidence. The erythro-
plastid of figure 12 may appear to have arisen directly from the
cytoplasm of this vasofactive cell. But figure 14 shows an
essentially similar cell at a slightly earlier stage of differenti-
ation, in which the chief nucleus has liberated a small bud.
About this bud an erythrocyte and a lumen may be conceived
to originate in the manner shown in e of figure 15, and so lead
to a condition like that of figure 12. In other words, the cells
of figures 12 and 14 are essentially multinucleated hemoblasts.
In this sense multinucleated hemoblasts, hemogenic giant-cells,
and blood-islands (‘angioblasts’) are fundamentally and poten-
tially alike; that is, they are essentially multiple erythroblasts
enveloped by a layer of potential endothelium.
DISCUSSION
From the foregoing description it will be clear that the con-

nective tissue of the full-term umbilical cord of the pig is exten-
sively vascularized and that it is actually for the most part still
in the condition of young mesenchyma or embryonal connective
tissue. The conditions are essentially similar to those described
for the body-stalk of very young human embryos. The question
arises whether the connective tissue of this cord is in the prim-
itive mesenchymal condition because it is vascularized or
whether it is vascularized because the connective tissue is in the
condition of undifferentiated mesenchyma. Since the blood-
vessels have apparently arisen to a considerable extent in situ,
12 H. E. JORDAN
the latter interpretation would seem to be the correct one; that

is, that this cord has maintained early embryonic conditions,
like that of its anlage, the body-stalk, and in consequence re-
tained its original capacity for vasculogenesis and erythrocyto-
genesis. The cause of this maintainance of early embryonic
vasculogenic and hemopoietic potentialities, especially singular
in connection with the advanced developmental condition of the
umbilical arteries and vein, and the several small areas of fully
differentiated mucous connective tissue, remains for the present
undertermined. It is most probably associated with the large
functional allantois, but the nature of this association is not
clear. The relatively highly developed character and healthy
condition of the covering ectoderm may be secondary to the
presence of the large number of capillaries in the subjacent
connective tissue.
Though this study can throw no light on the cause of the
vascularized condition of the umbilical cord of the pig, the in-
tense hemopoietic activity of its connective tissue supplies valu-
able data with respect to the initial steps in vasculogenesis. This
is the chief point of value in this specimen. In this connection
interest centers upon the mesenchymal cell, which becomes
hollowed out to form an endothelial cell and at the same time
differentiates erythrocytes (figs. 6, 12, 13, and 15). This cell
combines the functions of an endothelioblast and an erythroblast.
The process appears to be quite similar to that first described by
Ranvier® (’74) in the mesentery of the seven-day rabbit and in the
great omentum of the cat, and independently by Schaefer!? (74)
in the subcutaneous tissue of the new-born rat, and subsequently
confirmed by other workers on other forms. Ranvier named the
cells concerned in the process ‘vasoformative cells.’ According
to these investigators, mature (non-nucleated) ‘erythrocytes’ of
greatly varying sizes are formed directly within the protoplasm
of connective-tissue cells (vasoformative cells) by a process in-
volving the coalescence of scattered granules of hemoglobin into.
condensed globules, which then come to lie in vesicles within
the cells, the precursors of the capillary lumen.
UMBILICAL CORD OF THE PIG i
(3
The method of erythrocytogenesis here described for this

specimen of umbilical cord of the pig, however, differs radically
from that described by Ranvier and Schaefer, in that the erythro-
plastid in this case differentiates from a nucleated portion of a
vasofactive cell (figs. 6, e; 15, b and e, and 14). That is, the
erythroplastid differentiates from a typical erythroblast in the
usual mode. The nucleus of the erythrocyte disappears by
karyolysis (fig. 15, e). This is apparently a very rapid process,
since it can be detected in only relatively few cells. If one
considered only cells like those of figures 12 and 13, the process
would appear to be identical with that described by Ranvier
and Schaefer; but figures 15, b, e, h, and f demonstrate the essen-
tial difference.
Certain investigators (Spuler, Fuchs,’ et al.) have expressed
dissent from Ranvier’s and from Schaefer’s interpretation of
their observations; they explain these phenomena, the occurrence
of which they confirm, in terms of regressive changes and phago-
cytosis. They believe that the so-called ‘vasoformative cells’
are either isolated portions of a disintegrating embryonic vas-
cular plexus or erythrophagic connective tissue cells. It is
obvious that since the vasofactive phenomena here described
for the umbilical cord of the pig are fundamentally different,
while superficially apparently identical with those described by
Ranvier and Schaefer, the criticisms of Spuler and Fuchs have
no pertinancy to this case. Moreover, the red cells involved in
this process in the umbilical cord of the pig show no distinct
nuclear or cytoplasmic marks of degeneration. This cord, except
for the almost complete absence of mitotic figures, appears in
perfectly healthy condition. No free erythrocytes are available
for phagocytosis in the regions here described. There is no
indication of a disintegration of blood-vessels; on the contrary,
the full-term cord is relatively more extensively vascularized
than the cord of the 21-mm. fetus. Finally, and most. signifi-
cantly, this intracellular mode of erythrocytogenesis is strictly
comparable to that described for other hemopoietic organs, e.g.,
yolk-sac of 10-mm. pig embryo,’ yolk-sac of mongoose embryos,®
and red bone-marrow.’
14 H. E. JORDAN
The matter may be summed up with figures 15, a, b, and e,

and 15, h and f. A connective-tissue cell becomes transformed
into a hemoblast (erythroblast) with vasofactive capacity. This
cell may become bi- or multinucleated. One or several of the
nuclei with their enveloping cytoplasm may differentiate into
erythrocytes. Meanwhile a lumen appears within the cell, and
one or two of the original nuclei may persist as the nuclei of the
peripheral cytoplasm of the differentiating cell, which now forms
the endothelial wall of the initial capillary. In later stages in
the yolk-sac and in the red bone-marrow generally, the peripheral
‘endothelial’ layer of the original ‘vasofactive’ cell disappears,
thus freeing the intracellularly differentiated erythrocytes into
the confining blood spaces. The mesenchymal cell thus appears
endowed with divers hemogenic potentialities: it may become
an endothelial cell or a hemoblast (erythroblast). The endothe-
lial cell may secondarily differentiate into a hemoblast. These
hemoblasts may differentiate into erythroblasts or, as multinu-
cleated cells, they may differentiate both intracellular erythro-
cytes and a potential endothelial cell. These facts demonstrate
the very close relation between mesenchyme, endothelium, and
hemoblasts.
Sabin® records a similar vacuolization of mesenchymal ‘angio-
blasts’ in the living blastoderm of the two-day chick embryo
grown in Locke’s solution, by which the blood-vessel lumen
forms. But these observations do not justify her conclusion
that they prove “‘that the lumen of a blood-vessel is intracellular”
(p. 200). The data supplied by the umbilical cord of the pig
show that the definitive lumen of the blood-vessel derived from
a blood-island is of both inter- and intracellular origin.
This brings us to the matter of the factors which determine
whether the mesenchymal cell shall become an endothelial cell
or a hemoblast, and relates this investigation to the discussion
regarding theories of hemogenesis, that is, whether blood de-
velopment proceeds according to the monophyletic or the poly-
phyletic mode. . This much seems certain regarding this tissue:
single hemoblasts, freed from the mesenchyme and wandering
within its meshwork, do not differentiate into erythrocytes. It
UMBILICAL CORD OF THE PIG >
is only when such a cell becomes inclosed by endothelium that

it differentiates into a red blood-cell. Thus a group of such cells
may form and in consequence produce pressure upon the sur-
rounding mesenchyme, which then becomes transformed into
endothelium. Under these conditions the enveloped hemoblasts
become erythrocytes (figs. 6 and 7). Such endothelium is simply
an adaptive form of mesenchyme, as originally maintained by
Huntington! and by Schulte," and it may subsequently return
to mesenchyme, remain as endothelium, or differentiate hemo-
blasts either intra- or extraluminally. Endothelium, accordingly,
develops originally by two different methods, both clearly repre-
sented in the specimen under consideration: 1) By adaptation
of mesenchyme about a blood island; 2) by vacuolization of
vasofactive mesenchymal cells (‘angioblasts’).
A point of special interest concerns the fact that the nucleated
periphery of a multinucleated hemoblast supphes the same
favorable conditions or factors for determining erythrocytogenic
differentation as an endothelial wall. This phenomenon becomes
intelligible when we consider that both endothelial cells and
hemoblasts are only slightly modified mesenchymal cells, and
that the latter, as vasofactive cells, may become hollowed out
to form the lumen of an original capillary or differentiate intra-
cellular erythrocytes. The central fact here pertains to the
obviously very minute difference between the environmental
conditions or stimuli that determine whether the same cell (the
potential hemoblast, ‘vasoformative cell’, or ‘angioblast’) shall
become an endothelial cell or an erythroblast. This suggests
that also the factors which determine whether the hemoblast
shall become a leucocyte or an erythrocyte, in accordance with
the monophyletic theory of blood-cell origin, are similarly rel-
atively subtle and of minute degree. Original confinement by
endothelial walls furnishes the stimulus which determines
erythrocytogenesis; extravascular differentiation leads to granu-
lopoiesis. As shown by the recent experiments of Danchakoff,'?
the original polyvalency of the hemoblast, however, is lost by
the erythroblast, and this degree of differentiation is irrever-
sible. An erythroblast freed from its endothelial confines and
THE AMERICAN JOURNAL OF ANATOMY, VOL. 26, NO. 1

16 H. E. JORDAN
thrown into the surrounding mesenchyme, as in the allantoic

spleen grafts of Danchakoff, will not differentiate into a leucocyte,
but into an erythrocyte. A mature endothelial cell does not
normally, as originally, differentiate into a hemoblast. And an
extravascular hemoblast which is in process of differentiation into
a granulocyte apparently cannot redifferentiate into an erythro-
cyte after it has wandered into the blood-vessel lumen.
SUMMARY
The results of this study of the umbilical cord of the pig,

which maintains to full-term largely the embryonic condition
of the original body-stalk, emphasize the polyvalent capacity
of the mesenchymal cell and its hemoblast derivative, and supply
further evidence in agreement with the monophyletic view of
hemogenesis. The multinucleated hemogenic giant-cell fur-
nishes the same essential stimuli for the differentiation of erythro-
cytes as does an inclosing endothelium. It 1s comparable to a
blood-island, and produces erythrocytes intracellularly in a
manner similar to that by which erythrocytes separate out of
a blood-island syncytium. This tissue demonstrates also the
origin of endothelium both by adaptation of mesenchyme about
a blood-island and by vacuolization and fusion of vasofactive
mesenchymal cells. It shows, moreover, that the lumen of the
original blood-vessels includes both inter- and intracellular
contributions.
UMBILICAL CORD OF THE PIG L7/
LITERATURE CITED
DaANcCHAKOFF, VERA 1918 Cell potentialities and differential factors con-

sidered in relation to erythropoiesis. Am. Jour. Anat., vol. 24, p. 1.
bo 1918 Equivalence of different hematopoietic anlages (by method of stimu-
lation of their stem-cells). II. Grafts of adult spleen on the allantois
and response of the allantoic tissues. Am. Jour. Anat., vol. 24, p. 127.
Fucus, H. 1903 Uber die sogenannte ‘‘intracellulire’’? Entstehung der
roten Blutkérperchen junger und erwachsener Siuger. Anat. Hefte,
Bye, 22. Sh VE.
Huntineton, G. &. 1914 The development of the mammalian jugular
lymph sac, of the tributary primitive ulnar lymphatic and of the
thoracic ducts from the viewpoint of recent investigations of verte-
brate lymphatic ontogeny, together with a consideration of the
genetic relations of lymphatic and hemal vascular channels in embryos
of Amniotes. Am. Jour. Anat., vol. 16, p. 259.
Jorpan, H. E. 1916 The microscopic structure of the yolk-sac of the pig
embryo, with special reference to the origin of the erythrocytes. Am.
Jour. Anat., vol. 19, p. 277.
1917 Hemopoiesis in the mongoose embryo, with special reference to the
activity of the endothelium, including that of the yolk-sac. Pub. 251
of the Carnegie Institution of Washington, p. 291.
1918 A contribution to the problems concerning the origin, structure,
genetic relationship and function of the giant-cells of hemopoietic and
osteolytic foci. Am. Jour. Anat., vol. 24, p. 225.
Ranvier, L. A. 1874 De développement et de l’accroisement des vaisseaux
sanguins. Arch. de Physiol., T. 6, p. 429.
Sapin, F. R. 1917 Preliminary note on the differentiation of angioblasts
and the method by which they produce blood-vessels, blood plasma
and red blood cells as seen in the living chick. Anat. Rec., vol. 13,
p. 199.
10 ScHaEFer, E. A. 1874 The intracellular development of blood corpuscles
inmammals. Mon. Mier. Jour., vol. 11, p. 261.
11 ScuHutte, H. von W. 1914 Early stages of vasculogenesis in the cat (Felis
domestica) with special reference to the mesenchymal origin of endo-
thelium. Memoir, Wistar Inst. Anat. and Biol., no. 3, pp. 1-92.
12 Sputer, A. 1892 Uber die ‘‘intracellulire’’ Entstehung roter Blutkér-
perchen. Arch. f. mikr. Anat., Bd. 40, 8. 530.
PLATE 1
EXPLANATION OF FIGURE
1 Photomicrograph of transverse section of umbilical cord of pig. The cord is
covered with a stratified epithelium from three to eight (generally four) layers
thick, resembling somewhat the transitional type. Tufts of keratized cells
occur at certain points (#). To the right of the allantoic duct (All) are the
umbilical arteries. The umbilical vein lies below the double remnant of the
occluded yolk-stalk (Y.S.). At B.V. is one of the larger arterioles of the ex-
tensively vascularized connective tissue. In the area between the lower umbilical
artery and the vein, the connective tissue resembles the mucous type; elsewhere
it resembles more young mesenchyme, and contains many capillaries, arterioles
and venules (A.V.), and also numerous hemoblasts and several typical blood-
islands. (Photos. by W. S. Dunn, Cornell University Medical College, N. Y.
City. The illustrations were made from the Columbia specimen.) XX 18.
18
UMBILICAL CORD OF THE PIG PLATE 1
H. E. JORDAN
19
PLATE 2
EXPLANATION OF FIGURES
2 Photomicrograph of the ectodermal covering of the cord in the region,

Som., of figure 1. It includes four or five layers of cells and resembles transitional
epithelium. The lowermost layer is composed of cuboidal cells; the outermost
layer includes dome-shaped, rectangular, and flattened peridermal cells; the
intermediate layers include spheroidal and polyhedral cells. > 300.
3 Photomicrograph of remnant of yolk-stalk (fig. 1, Y.S.). It is double and
occluded, and each moiety includes about twelve cells in its diameter. Art.,
arteriole. X 300.
H. E. JORDAN
21
PLATE 3
4 Photomicrograph of allantoic duct. The lining epithelium is thrown into

folds. In the troughs of the folds the epithelium consists of a single layer of
cuboidal or flattened cells; over the crests, of from three to four layers, consti-
tuting a stratified columnar epithelium. Art., arteriole; B.I., “blood-island’ of
erythroblasts. X 235.
5 Photomicrograph of pair of blood vessels (the branching venule cut
obliquely) and the surrounding mesenchyme (region A.V. of fig. 1). > 300.
22
H. B. JORDAN
< a ego OLE re

“gs me _o/
; 2
23
PLATE 4
6 Drawing of blood-island, from region just below the allantoic duct (fig. 1,
All.). Peripherally the cells are becoming differentiated into an endothelium.
Centrally the syncytial mass is becoming vacuolated through the appearance of
intercellular spaces, and certain of the cells have entered the early erythroblast
(‘megaloblast’) stages (a and 6b). One erythroblast (d) contains a large vacuole.
A hemoblast (e) has differentiated an intracellular erythrocyte. A hemoblast
(h) is separating from the differentiating endothelium. Between the two endo-
thelial cells below, appears another hemoblast. Figures 6 to 9 are magnified
1500 diameters.
7 Blood-vessel in process of differentiation from the mesenchyme. This
drawing is from the region to the right of the allantoic duct between the two
umbilical arteries (fig. 1), and includes approximately the middle third of this
entire vascular anlage. The forming lumen contains seven young erythroblasts
(‘megaloblasts’), separating out of an originally syncytial mass.
8 Young vasofactive cell, with generally mesenchymal features and three
vacuoles, the precursors of a capillary lumen.
9 Slightly older vasofactive cell with large vacuole, the underlying nucleus
assuming endothelial features.
24
H. E. JORDAN
PLATE 5
10 Vasofactive cell with three nuclei. The nucleus at the right has original
mesenchymal features; the endothelial ‘cell’ below the lumen has assumed hemo-
blast features, and has differentiated an intracellular erythroplastid (ep).
Figures 10 to 15 are magnified 1500 diameters.
11 Vasofactive cell differentiating a lumen. This figure corresponds to a
transverse section of figure 10.
12 and 13 Vasofactive cells with an erythroplastid in the lumen.
14. Vasofactive cell in early stage of differentiation from mesenchyme. The
cell has in general hemoblast features. The larger nucleus has produced a small
bud at the left. From such nuclear buds and their enveloping cytoplasm
develop intracellular erythroplastids.
15 (ato7) Vasofactive mesenchymal cells at various stages of differentiation:
a) Typical young erythroblast (‘megaloblast’). 6) Hemoblast that has differ-
entiated an erythrocyte intracellularly. c) Cell with vascular lumen and an
intracellularly differentiated erythroplastid. d) Cell with lumen, having as-
sumed endothelial features. e) Cell with lumen, containing an erythrocyte
(ec, ‘normoblast’) and an erythroplastid (ep.). jf) Binucleated cell with two
intracellularly differentiated erythroplastids (ep). g) Cell with lumen and two
nuclei, both with endothelial features. A) Cell with four nuclei, one apparently
in process of amitotie division. The centrally located nuclei with their en-
veloping cytoplasm may differentiate into erythrocytes. 7) Binucleated cell,
endothelial in character, with lumen containing an erythroplastid.
H. E. JORDAN
Resumen por el autor, Henry Alexander Murray, Jr.
Colegio de Médicos y Cirujanos, Columbia University.
El desarrollo del asa cardiaca en el conejo, con especial mencion

del surco bulbo-ventricular y el origen del
tabique inter-ventricular.
E] autor ha estudiado varios embriones de conejo para deter-

minar si la formaci6n del asa cardiaca en esta especie es Semejante
a la deserita por Schulte en el gato. Los procesos de formacion
soh semejantes en ambas especies en lo referente al origen del
asa primaria, que se produce a consecuencia del hundimiento de
la hendidura bulbo-ventricular izquierda, y la fusién de los tubos
miocirdicos se verifica por la intervencién de la placa cardiaca
media, que esta representada temporalmente por una cresta bien
distinta situada en la pared interna del miocardio. En las series
de embriones de conejo, sin embargo, esta elevacién se oblitera
algunas veces por completo y no susministra prueba alguna
sobre su intervencién en la formacién del surco inter-ventricular
primitivo.
Translation by José F. Nonidez
Carnegie Institution of Washington
AUTHOR'S ABSTRACT OF THIS PAPER ISSUED
BY THE BIBLIOGRAPHIC SERVICE, JULY if
THE DEVELOPMENT OF THE CARDIAC LOOP IN THE

RABBIT, WITH ESPECIAL REFERENCE TO THE
BULBOVENTRICULAR GROOVE AND ORIGIN OF
THE INTERVENTRICULAR SEPTUM
HENRY A. MURRAY, JR.
College of Physicians and Surgeons, Columbia University, N. Y.
SEVEN FIGURES
The process of union of the two lateral cardiac vessels to form

the heart has recently been described in detail by Doctor Schulte
as exemplified in the series of young cat embryos of the Columbia
Collection. It was at Doctor Schulte’s suggestion and under
his supervision that I undertook to ascertain whether fundamen-
tally the same processes took place in the embryo rabbit.
Earlier investigators had described how the lateral plexuses
of blood-vessels, forming into two longitudinal endothelial tubes
in the splanchnopleure, unite in the midline to create a cylin-
drical median structure—the heart; and they considered that
during subsequent growth the heart became coiled to accommo-
date itself within the pericardium. Doctor Schulte’s investi-
gation, previously mentioned, showed that it was not such a
simple process, but that a number of very interesting factors
were responsible for the changes that took place. Through the
kindness of Dr. F. T. Lewis in putting at my disposal the beauti-
ful Harvard series of rabbit embryos and by his guidance and
suggestions, I was able to study the fusion and subsequent
history of the heart in another species. Approximately forty
embryos from the Harvard rabbit and Columbia cat series were
examined under the microscope and a dozen and a half models
were constructed in wax, according to the Born method. The
cat embryos were cut in 13.34 sections, whereas the rabbits were
cut in 6u or 10u sections, mostly transverse. The thicker sections
are less apt to be damaged, pile into better models, and under
29
30 HENRY A. MURRAY, JR.
some conditions are to be recommended. For the finer details,

however, thin sections are naturally the more desirable. The
embryos were variously stained. My observations lead me to
believe that the development of the rabbit heart tallies in every
important respect with that of the cat.
In the accompanying diagram (fig. 1), the schema usually
presented to portray the formation of the cardiac loop may be
contrasted with a parallel series of figures representing Doctor
Schulte’s findings. Note that the initial and final stages in each
WUC OC
VINE? OO
Fig. 1 A. Schema of cardiac loop formation; as presented in most modern
text-books, based on the His models. B. Schema representing the same period
of development, as observed by Doctor Schulte in the Columbia Laboratory.
(Note that the bulboventricular clefts are formed in both hearts before fusion
takes place, that the left groove together with the left shoulder of the ventricle
becomes accentuated, that there is a corresponding obliteration of the right
groove and shoulder, and that the venous end of the heart migrates to the left.
These are the principal factors in the formation of the loop.)
case are the same, but that the intervening processes are dis-
similar. Instead of reiterating Doctor Schulte’s conclusions, I
will ask the reader to examine carefully figure 2 before reading
the following explanation. It is a model of the endocardial
cavity in a nine-day rabbit. The myocardial mantles, not repre-
sented in the model, have completely fused, but the endothelial
tubes have not as yet entirely coalesced. The picture presents
a condition previous to the complete amalgamation of the vessels
to form a common cavity. Those points where the endocardia
CARDIAC LOOP IN RABBIT 31
are still separate mark out the line of fusion of the two primitive
tubes, and it is thus quite evident what portions of the cavity
are derived from the right cardiac vessel and what portions from
the left.
The elements represented are: a) the sinus venosus at the
confluence of the vitelline veins; b) the canal between the sinus
venosus and the ventricle, which, as the atrium develops at this
point, we may call the atrial canal; c) the common ventricle,
Fig. 2. Model representing a cast of the cavities within the heart and the
connecting vessels in a rabbit embryo of nine days. Ventral view. Harvard
Embryological Collection, Series 619. 1, aortic branches; 2, right and left bulbs;
3, apertures formed by the septum dividing the bulboventricular canal; 4,
right shoulder; 5, left shoulder; 6, aperture formed by the septum dividing the
common ventricle; 7, atrial canal; 8, right vitelline vein; 9, left vitelline vein.
X 100.
‘The valuable study of the development of the pericardium in ferrets by

Professor Robinson (Journ. of Anat. and Phys., 1902, vol. 37) has recently been
extended by his pupil, Doctor Wang, with results which deserve more critical
consideration than can be given here. Being concerned especially with early
stages, Wang does not discuss the questions raised by Schulte, and his inter-
pretation of the model shown in figure 2 would differ from ours, as may be seen
by comparing it with his figure 31 representing the heart of a ferret of 13 to 14
segments. Wang’s most interesting observation is of a ‘primary heart rudiment,’
a vessel crossing the median line and subsequently dividing into two endothelial
tubes. The lateral vessels thus formed, or others somewhat posterior and lead-
ing to them, then reunite to make a ‘secondary’ heart. In the rabbits of the
Harvard Collection, as Doctor Lewis informs me, there may be seen a strand of
presumably angioblastic tissue in the region of the primary heart of Wang, but
nothing which should be interpreted as a heart. At present, therefore, we are
not inclined to recognize a heart previous to the union of the lateral cardiac
vessels.
THE AMERICAN JOURNAL OF ANATOMY, VOL. 26, No. l

with a right and left shoulder; d) the restricted portion between

ventricle and bulb, which may be called the bulboventricular
canal; e) the bulb, and f) the arterial branches. The following
points should be particularly noted: 1) the atrial canal has been
forced to the left, and that portion of the canal contributed by
the right heart has become relatively much reduced; 2) the left
shoulder of the ventricle is elevated, the right depressed; 3) the
left bulboventricular cleft is pronounced, the right is obscure,
and 4) on account of the greater impression made by the left
bulboventricular cleft that portion of the bulbar canal contri-
buted by the left heart is diminished. Later, the right bulbo-
ventricular cleft disappears, the left becomes more pronounced
and vertical, the atrium develops from the atrial canal growing
cephalad behind the ventricles, and the cardiac loop is then
complete. My observations commence at a stage when the
original lateral tubes have become ventrally placed and are
united through the intervention of a middle cardiac plate. The
process by which this change is effected—a subject upon which
Wang speculates at some length—will not be discussed. At-
tention will be focused on what may be considered the most
fundamental aspects of the succeeding modifications, namely,
1) the middle cardiac plate with a consideration of its future
history and possible connection with the interventricular septum,
and 2) the bulboventricular groove.
MIDDLE CARDIAC PLATE AND INTERVENTRICULAR SEPTUM
In figure 3 observe the middle cardiac plate connecting the

two hearts. It will be noticed that it is narrow cephalad, con-
necting the bulbs, and broad caudad between the ventricles.
This mesothelial element gradually becomes incorporated into
the myocardial walls of the enlarging ventricular cavity and is
later represented in the rabbit by a ridge or series of ridges mark-
ing the original line of fusion. The ridges which are quite ap-
parent on the inner aspect of the ventral wall are well shown in
figure 4; they are placed opposite the septa which still remain
between the two endothelial tubes (compare fig. 2). The early
embryos of the cat in the Columbia series show similar ridges
and also some intermediate stages. The rabbit differs from the
eat in that this ental protrusion of the myocardia is a simple
ridge and is not surmounted by a groove. Has this ridge any
definite relation to the future interventricular septum? I think
not. After modeling a number of rabbit and cat hearts between
this stage and the stage when the interventricular septum is
first apparent, I find no connection between the two. As shown
in figure 4, when last observed the median ridge is directed to-
wards the atrial canal. As the latter does not change its position
until a later date, if the interventricular septum were a product
of this ridge we should expect to find it at first obliquely placed
Fig. 3 Model of the myocardium (ventral view) from a rabbit embryo of nine
days; H. E. C., Ser. 620. 1, right bulb; 2, left bulb; 3, right. bulboventricular
groove; 4, left bulboventricular groove; 5, right ventricle and shoulder; 6, left
ventricle and shoulder; 7, middle cardiac plate. > 100.
and in line with thé canal. This is not the case. On the con-
trary, we find a septum arising apparently as a thick muscular
ridge from the most caudal portion of the ventricular loop, cor-
responding to a groove on the exterior.2, Both septum and groove
are sagittally placed and are not at this early stage directed to-
wards the atrial canal (compare fig. 6). Furthermore, the sep-
tum appears at a considerably later date, after the common
atrium is well formed and the ventricular wall has undergone
great expansion and growth with considerable trabecular for-
mation. In both the rabbit and cat embryos the increased
? As Mall says, it is more correct to speak of the downward growth of the

apices of the two ventricles than the upward growth of the septum.
growth in thickness of the ventricular wall is first manifest on

the ventral wall and gradually spreads laterally and dorsally.
In my series, the actual height® of the heart in the rabbit when
the last sign of a middle cardiac plate can be determined is 0.32
mm., whereas the earliest sign of an interventricular septum is
found in an embryo with an actual heart length of approximately
0.75 mm. Doctor Schulte, in the latest stage in which a middle
cardiac plate was present in his material, found it continued
caudad by a sulcus which he regarded as the beginning septum
ventriculorum. Beyond this stage there was a gap in his ma-
terial to the period of a well-developed septum with no remnant
of the middle cardiac plate. My reason for dissenting from his
conclusion that the median plate gives rise to the septum is that
in the rabbit models and in embryos of the cat subsequently
obtained the sulcus mentioned is found to disappear and the
ventricle becomes evenly convex without a trace of indentation
referable to the middle cardiac plate. Only later does the ventri-
cular septum arise in the manner J have described. The interior
of the model portrayed in figure 5 shows a smooth wall with no
sign of a ridge, although at two points the endothelial tubes have
not yet completely united. Considerably later, as the atrial
canal is moved to the right while the interventricular septum
remains fixed, the canal comes to be immediately dorsal to the
septum; the septum will then be seen to extend toward the centre
of the canal and still later to become fused to the endocardial
cushion. The subsequent development of these parts is beyond
the scope of this paper.
THE BULBOVENTRICULAR GROOVES
These clefts first appear on each side before the lateral cardiac
vessels have fused. They have been given this name because
in their primary position they separate primitive bulb from
primitive ventricle. Later, as we shall see, the bulb contributes
to the right ventricle, and the left bulboventricular cleft may
then be termed an ‘interventricular groove.’ In figure 3 the
3 As measured from the most cephalic to the most caudal points, regardless
of what portions of the heart these may be.
CARDIAC LOOP IN RABBIT ao
Fig. 4 Model of the myocardium (dorsal view of the ventral wall) from a
rabbit embryo of nine days; H. E. C., Ser. 619. 1, bulboventricular canal; 2,
left shoulder; 3, left bulboventricular groove; 4, right shoulder; 5, right bulbo-
ventricular groove; 6, inward protrusions of the myocardium in a line directed
toward the center of the atrial canal; 7, sinus yvenosus. X 100.
Fig. 5 Model of the myocardium (ventral view) from a rabbit embryo of
nine and one-half days; H. E. C., Ser. 565. 1, aortic branches; 2, bulb; 3, bulbo-
ventricular groove; 4, shoulder; 5, common ventricle; 6, right vitelline vein; 7,
left vitelline. X 100.
grooves are horizontal. The left groove is already assuming a

more significant réle and the asymmetry forecasts future changes.
In the cat this prominence of the left cleft is not accentuated until
somewhat later. A subsequent stage in the rabbit is represented
in the endothelial model, figure 2, and in figure 5 a fully developed
cardiac loop is seen. The bulboventricular cleft (there is now
only one, the right having become obliterated) is more oblique.
From this stage onward there is a continual progressive ex-
tension of this furrow, and as it develops, its plane is modified
so that in figure 6 we find it very nearly vertical. Notice in
this drawing its relationship to the primitive septum. It is not
in line with the latter structure. In the next period, however,
the bulboventricular groove, formerly horizontal, is now vertical,
protrudes into the ventricular chamber, becomes continuous
with the septum, and together with it divides the cavity into
right and left portions. This is well shown in figure 7. The
division of the common ventricle then seems to be the result of
four processes: 1) the interventricular septum growing cephalad
from the floor of the loop; 2) the bulboventricular groove becom-
ing vertical and forming the ventral portion of the septum; 3)
the migration of the atrial canal] to the right, allowing the endo-
thelial cushions to play their part, and finally, as His has shown,
4) the downgrowth of the pulmonoaortic septum which fuses with
the above-mentioned elements so as to form a continuous par-
tition between the right and left hearts.
Fig.6 Model of the myocardium (dorsal view of the ventral wall) from a rabbit
embryo of ten and one-half days; H. E. C., Ser. 559. 1, bulb; 2, ridge extending
into the common ventricular cavity and corresponding to the bulboventricular
cleft; 3, shoulder; 4, interventricular septum (this is the first indication of the
ridge found at the apex of the ventricular loop). X 100.
Fig. 7 Model of the endocardial cavity (ventral view) from a cat embryo of
7 mm.; Columbia Collection, Series 266. By kind permission of Dr. A. J. Brown.
In the Harvard Laboratory there is a very similar model of the cavities in the
heart of a 4.4-mm. pig embryo, made in 1909, under Dr. Minot’s direction, by Mr.
A. E. Meyers. 1, cleft made by the ridge growing upward from the caudal
extremity of the loop, which is continuous with 2, the impression made by the
bulboventricular groove. Together they partially subdivide the ventricular
cavity; 3, atrial canal; 4, left atrium. X 50.
SUMMARY
The processes in the early development of the rabbit’s heart

are fundamentally the same as in the cat. 1) The primary loop
is due to the deepening of the left bulboventricular cleft and a
disappearance of the right, accompanied by a reduction on the
part of the right shoulder of the ventricle and a very marked
growth of the left. 2) The middle cardiac plate, marked tem-
porarily after myocardial fusion by a distinct ridge which corre-
sponds precisely to the line of fusion of the endothelial tubes,
eventually becomes entirely obliterated. 3) The primitive inter-
ventricular septum arises de novo from the floor of the loop in
a sagittal plane. 4) The left bulboventricular cleft, at first
horizontal, becomes oblique and then vertical; protruding into
the common ventricular cavity as a well-marked ridge, it meets
the septum developing in the apical portion of the heart and
contributes to the formation of the interventricular septum. of
the adult.
BIBLIOGRAPHY
Bremer, J. L. 1912 Development of the aorta and aortic arches in rabbits.

Am. Journ. Anat., vol. 13, no. 2.
Lewis, F. T. Intraembryonic blood vessels of rabbits from 8} to13 days. Am.
Journ. Anat., vol. 3.
Mauu, F. P. 1912 On the development of the human heart. Am. Jour. Anat.,
vol. 13.
1912 Bifid apex of human heart. Anat. Rec., vol. 6.
1912 Aneurysm of membranous septum. Anat. Rec., vol. 6.
Parker, K. M. The early development of the heart and anterior vesselsin
marsupials with special reference to Berameles. Proc. Zool. Soc.,
London, 1915, Pt. 3.
Ropinson 1902 Early stages of development of the pericardium. Journ. of
Anat. amd Physiol., vol. 37.
Rovvibre, H. 1904 Etudes sur le developpement du péricarde chez le lapin.
Jour. de l’anat et de la physiol., vol. 11.
Scuutte, H. W. 1916 The fusion of the cardiac anlages and the formation of
the cardiac loop in the cat. Am. Journ. Anat., vol. 20, no. 1.
_ 1914 Early stages of vasculogenesis in the cat with especial reference
to the mesenchymal origin of endothelium. Memoirs of Wistar Insti-
tute of Anat. and Biol., no. 3.
STRAHL AND Carius 1899 Beitrige zur Entwickelungsgeschichte des Herzens
und der Kérperhéhlen. Arch. f. Anat. und Physiol., Bd. 15.
TANDLER, J. Keibel and Mall. Manual of human embryology, vol. 2.
Wana, Cuuna-Cuina 1917 Earliest stages of development of the blood-vessels
and of the heart in ferret embryos. Journ. of Anat., vol. 52, part 1,
Resumen por el autor, Frank Blair Hanson.
Universidad Washington, San Luis.
La ontogenia y filogenia del esternon.
Existen en la literatura tres teorfas diferentes sobre el orfgen

del esternén, suponiéndose generalmente que este hueso no es
homologo en los amniotos e ictiépsidos. El autor demuestra que
el estern6n de los vertebrados esta’ mas intimamente ligado con
la cintura escapular que con las costillas, ‘y describe diferentes
estados de desarrollo en los embriones de un cierto ntimero de
mamiferos—gato, rata, raton, cerdo y hombre—en los cuales las
barras esternales son estructuras bien manifiestas antes de su
unién con las costillas, cuyo hecho es contrario a la teoria de Ruge
sobre el origen costal des esternén. El presternén esté intima-
mente asociado con los coracoides en todas las clases de los ver-
tebrados, incluso los monotremas. En los embriones jévenes del
raton y el hombre se forma una cintura mesenquimatosa continua
comparable a la cintura pectoral del tiburén, de la cual se derivan
la cintura escapular y el manubrio. Por consiguiente, el ele-
mento anterior del estern6n tiene un origen comun con la cintura
escapular y, en el embrién o durante toda la vida del animal,
esta en intima relacion con los coracoides. Las bandas esternales
son derivados del rudimento anterior medio, pudiéndose asociar
secundaria pero no genéticamente con las costillas. El esternén
es una estructura homdloga en todos los grupos de vertebrados
y se presenta en las formas comprendidas entre el tibur6n Hexan-
chus hasta el hombre.
Translation by José F’. Nonidez
BY THe BIBLIOGRAPHIC SERVICE, AUGUST 4
THE ONTOGENY AND PHYLOGENY OF THE STERNUM

FRANK BLAIR HANSON
Zoological Laboratory of Washington University, St. Louis, Missouri?
TWELVE PLATES (FORTY-NINE FIGURES)
CONTENTS
ems ROCCEOT wre tert Mascot aioe, wha ried SRR eM oe:To RR ia ee ented yates 41
Mien Cmtrcaleestimeateol existiney: theorlesa.qsecsee ssa see esas cls ee elas 42
Peiwiges theory ofcostal origms sre A PA SS 42
DMP ALCTSONLS COLACOLGaleth eOLy7.& aes ESO ok aoe Dee eaes Shla ty oe 43
Se WOLaOtbarkersan d oEROWOS acc :5ni5 seein kins oe telocktens ites 45
AmWihiteheaa andewaddell’s “in situ? stheony-s... «2.4.20 +4.00ee eee 47
5. Work of Rathke, Kravetz, Mueller, etc........ LG Mee oe Or ng oi 52
MRiThe ontogeny offthestermum. «$02... VIPAT a 58
URN herrallabyATG species vated Sacer oie pee pam oet erew aie taeopin Pesca ode 58
2 iheranternonmedran sternal rudiments: 90.cse sos is eed cus ook 60
Stoo hid ENSUE
x8]Oi ese enna 20Soe ¢ Utes Deane ei ate cn iit a ica i 61
4, Stages in the ontogeny of the mammalian sternum............... 63
PRA ONG IUISTONS HRN a aie suKa eee AVEO RED RIORT IS ye Sekt ORISEL MCPS TS vis hasan 63
ven themmyloreny: of the sternum. : 1.22232 245, 225..0 2. Dee 3 ee 64
ETE RN A yore frp st sal kes Aide Le vie ea a hy els dal 64
PE STRRE TERI GE ET 93,5 oe ee Thc VOTES os Soe eats feoscee wie Het 66
S18) 2CTLGETs Sas EOI Re eae ee NO J ee ae me eee 70
AUB 3 ee SRT i ee RR ae 76
Dh CORRE MRE DMMS iS 0. 0 Sc JeRUM WS se Cae Satng: bts 2 onl ere de Paani sie 78
GrelVistrsuplalomme te: | <\eie Sermo tes a Athen ee oe ccs me nem eoa 78
TS PirieReprrrerrennarase 8990), RCA Pee ia, SS) OEE gra eka 81
8; Thetadubiinumanystennum«.,.1 4925. oF td <2) mse speeintoe a cipe cee ok oe 83
G50 SiC Is Mma I DEO. 5 dct Ee wae littae hen Pee RS bicieee sh §,anes A did Spek: 87
Ppp ULTMIR ELIS. 5) ee MEN coc ls ugg she aa He nan ateoea octet) aiairard ore lae ote 88
I. INTRODUCTION
The shoulder-girdle complex presents some of the most fasci-

nating and difficult problems of vertebrate morphology. In all
the lower vertebrates the sternum, because of its intimate re-
lation to the coracoids, enters into and constitutes one of these
problems. Its origin, development, and homologies have been
41
42 FRANK BLAIR HANSON
the subject of numerous contributions for a century past, yet

to-day there is no general agreement upon many of the points
involved.
For several years the writer has been studying the shoulder-
girdle region in the vertebrates. This included also a restudy of
the origin of the sternum, the results of which are embodied in’
this paper. The fact that my conclusions are at variance with
the usually accepted theory of sternum origin only adds to the
interest of the undertaking. If this paper settles the points at
issue or stimulates further investigation upon the part of others,
it will in either case not have been in vain.
One of the recognized deficiences of much of the previous work
is that general and far-reaching conclusions have been deduced
from the study of only one or two forms, and these most often
the more highly specialized ones. The author has attempted
herein to bring together corroborating lines of evidence from
both ontogeny and phylogeny, believing that a theory of sternal
origin only so demonstrated can command consideration.
I desire to express my deep appreciation for the constructive
criticisms and helpful suggestions given by Prof. J. Sterling
Kingsley during the course of this investigation.
II. CRITICAL ESTIMATE OF EXISTING THEORIES
1. Ruge’s theory of costal origin
Ruge (’80) was the first investigator to work up and present

a well-developed and illustrated paper containing a theory of the
origin of the sternum. His fifty pages of text and twenty-two
figures gave his theory a commanding place in the literature.
Most books on human and comparative anatomy, until the
present day, copy his figures and accept his view that the sternum
arises as a product of the ventral costal cartilages. Asanexample
of this, Keibel and Mall in their two-volume Embryology give
the following statement concerning the origin of the sternum:
“The cartilage of the sternum arises mainly from the cartilage
of the ribs, from which it is secondarily separated by the for-
mation of the costosternal joints.”” So completely has Ruge’s
ONTOGENY AND PHYLOGENY OF THE STERNUM 43
work dominated the field that even in the latest editions of

Human Anatomy texts his work is alone mentioned, or at most,
a footnote is added to the effect that this view has of late been
questioned by some.
However, there are at least two later views as to the origin of
the sternum. Paterson (’00, ’02, 04), on the one hand, and
Whitehead and Waddell (’11), on the other, have propounded
theories which are at once contradictory of Ruge’s view and also
antagonistic to each other. Thus there are at the present time
three distinct-and opposing theories concerning the origin of the
sternum, and it was with a view to clearing up this confusion
and also to give the prominence deserved to this later work that
the present investigation was undertaken. For while a study of
the papers representing these theories may not convince one of
the validity of any one of them as opposed to the others, the
latter two mentioned do point out very clearly that there are data
which Ruge did not consider; and further, that our commonly
accepted view concerning the origin of the sternum, held for
nearly forty years, must be greatly modified and possibly cast
aside altogether.
2. Paterson’s coracoidal theory
Professor Paterson in a series of papers (’00, ’02, ’04) was the

first to attempt to overthrow Ruge’s theory of the origin of the
sternum from the ventral ends of the ribs. His observations
were made upon the rat, rabbit, and man. He describes a single
median rudiment which is directly continuous with the mass of
cells destined to form the shoulder-girdle. From this median
mass two strands of cells grow caudally to form the sternal
bands. Whitehead and Waddell (’11) say, ‘‘thus in the final
analysis, according to Paterson’s view, the sternum is derived
from the shoulder-girdle.’’ This seems an unwarranted state-
ment. It would be as correct to say that, according to Paterson,
the shoulder-girdle is derived from the sternum as to say, as
do Whitehead and Waddell, that Paterson makes the sternum to
be a derivative of the shoulder-girdle. What Paterson does
succinctly say is, ‘‘that the presternum and shoulder-girdle are
originally derived from the same (italics mine) element; a primi-

tive band of cellular tissue which crosses the midline.”’ Paterson
has no interest in deriving the sternum from the shoulder-girdle
or vice versa; his contention being, first, that the sternum is not
a product of the costal cartilages, and, second, that it is yielded
from a common, continuous, mesenchymatous element which
gives rise to the shoulder-girdles and the sternum.
Paterson (’02) compares this continuous cellular element in
the rat to the girdle in the elasmobranchs. He exhibited before
the British Medical Association his sections of rat embryos side
by side with embryos of Acanthias vulgaris to demonstrate that
‘essentially the same method of development occurs in the dog-
fish and in the rodent. But a marked difference is produced in
the process of development.. Instead of a jointed and highly
differentiated structure such as is characteristic of mammals, a
simple continuous bar of cartilage is formed, across the middle
line and below the heart, which gives rise laterally to the primi-
tive shoulder-girdle.”’
Paterson (’00) also points out, and gives several figures in sub-
stantiation, that the parts of the sternum opposite the costal
attachments remain longest in a cellular condition. His point
being, of course, that if the sternum were ossified from the ribs,
these regions should ossify first, and not last as is actually the
case.
This comparison of the girdles in the shark and rat embryos
is very suggestive. Many other structures of present-day mam-
mals may be traced directly back to homologous structures in
the elasmobranchs, and since in the cartilaginous girdle of the
shark we have all the necessary material and in proper position
for differentiation into scapulae, coracoids, and sternum, we might
even upon a priori grounds expect to find in the higher groups of
vertebrates an embryonic stage in which the rudiment of the
girdles and sternum might be represented by such a “‘contin-
uousbar . . . . reaching across the middle line” as Paterson
found in the rat.
3. Work of Parker and Howes
Parker (’91) claims to have found a sternum in the shark

Notidanus indicus. A small blunt process is set in between the
two cartilages which unite later to form the girdle. This struc-
ture was earlier described in the same shark by Haswell (’84) who
says ‘‘the intercepted cartilage is temptingly like a presternal,
but the absence of such an element in the skeleton of any group
nearer than the Amphibia seems to preclude this explanation.”’
Parker’s (’91) figures 1 and 2 would indicate that this was a
presternum, and that Haswell was more nearly correct in his
observation than in his deduction therefrom. Had Paterson used
Hexanchus rather than Acanthias, he might have found an even
more striking resemblance of stages between the rat and shark
than he did.
By the courtesy of the officials of the U. 8. National Museum,
I was permitted to examine their type specimen of Hexanchus.
The body wall had been laid open along the ventral side to allow
the preserving fluid to bathe the viscera. By means of a short
anterior and two lateral incisions I was enabled to lay bare the
median ventral portion of the pectoral girdle without otherwise
disturbing the value of the specimen as a type. The girdle
(fig. 1) was exactly as described by Haswell (’84) and Parker
(91). The median cartilage in a young specimen was dis-
tinctly marked off from the coracoids and in general appearance
was not unlike the fetal girdle of the marsupial (fig. 35). Later
I dissected two specimens of Acanthias that measured 4 inches
and 7 inches, respectively, in order to confirm Paterson’s
statement of its likeness to the early embryonic girdle of the rat.
In both these specimens the girdle was approximately the same
as in the adult. Figure 2 shows the girdle of the 7-inch speci-
men, and if compared to the marsupial girdle (fig. 35) and the
mouse girdle (fig. 5), the morphological relations are apparent.
In these early stages of the marsupial and mouse no suture has
as yet appeared between coracoids and presternum, giving the
resulting shark-like girdle, complete across the midventral line.
Parker (’91), accepting in common with others the validity of

Ruge’s theory of a costal sternum in reptiles, birds, and mammals,
but being unable to relate a sternum of such derivation with the
sterna of the Ichthyopsida, suggested that there must be two
distinct types of sterna: 1) a costal sternum, characteristic of the
Amniota, and 2) a coracoidal or clavicular sternum, character-
istic of the Ichthyopsida.
This classification of the sternum was adopted by Howes
(91) who says, ‘‘the distinction indicated by the two terms
‘costal sternum’ and ‘coracoidal sternum’ is but the expression of
a fundamental morphological difference between the two struc-
tures.”’? Howes slightly altered the terminology of Parker. He
would distinguish between a ‘coracoidal archisternum’of the Ich-
thyopsida and a ‘haemocoracoidal neosternum’ of the Amniota.
This latter term was based upon his idea that the ‘‘interclavicle
may be, throughout, the vanishing vestige of the coracoidal
sternum of the Ichthyopsida.”” The acceptance by Parker
(91) and by Howes (’91) of this division of two morphologically
different sterna in the group of the vertebrates indicates how
completely Ruge’s theory dominated their thoughts, and the
thought and teaching of that day concerning the origin of the
sternum. If, however, the conclusions of later workers regarding
Ruge’s theory prove valid, and all the facts at the present time
seem to substantiate their validity as we shall later see, it is no
longer necessary to divide the sterna of the various classes of
vertebrates into coracoidal and costal, for no sternum is costal in
origin, the union of ribs and sternum being but a late and sec-
ondary stage in development. This is obviously an important
item, if proved, for it enables us to homologize all vertebrate
sterna. Heretofore it has been impossible to homologize the
sterna of the Ichthyopsida and the Amniotes because of their
supposed dual origin. One of the objects of the present paper
is to determine this matter of a single or dual origin for the
sternum and the solution of its homology throughout the
vertebrates. .
4. Whitehead and Waddell’s ‘in sitw? theory

In 1911 Whitehead and Waddell undertook to settle the whole
vexed question by a reexamination of all the evidence and by a
study of younger stages than had hitherto been used. Their
work was based on observations made upon three forms: the pig,
cat, and man. ‘These studies, however, instead of settling the
dispute between the theories of Ruge and Paterson, led the
authors to reject both of them and to propound an entirely new
one. So, as a result of this latest paper, there are at the present
time three, instead of two, rival theories of sternal origin. For
neither of these two latter theories had been able of its own
weight effectually to settle the points in question; Ruge’s theory
of costal origin still maintains its hold upon the minds of most
morphologists; but nevertheless, new evidence produced by later
work throws very strong doubt upon the conclusions of Ruge.
Ruge apparently had no stage prior to that in which the sternal
bands were united with the costal cartilages; but it must be ad-
mitted, that his conclusions, based upon the material that passed
through his hands, are clearly valid for the stages described—
in fact, the only ones that could possibly be deduced therefrom.
It is probably this fact that gave Ruge’s theory its persistence
through the years. The later workers, however, have had as
their goal stages much earlier than Ruge’s, and, while they have
succeeded in finding them, are still very far apart in the inter-
pretation thereof.
As Whitehead and Waddell’s paper is of considerable im-
portance and has never been reviewed, a brief summary of its
contents and conclusions is necessary here.
They studied first the pig, because the absence of clavicles in
this form tends to simplify matters at the cranial end of the
sternal rudiment; next the cat, for here the clavicle is a rudi-
mentary bone and does not articulate with the sternum, and
finally the human embryo, where the clavicle reaches its fullest
development.
These authors began with pigs of 24 mm. and worked through
successively smaller stages, until the sternal rudiment was very
THE AMERICAN JOURNAL OF ANATOMY, VOL. 26, No. 1

feeble, while in specimens smaller than 15 mm. no rudiment could

with certainty be detected.
In a pig 24 mm. long, to follow the description of Whitehead
and Waddell, the sternal rudiment is an aggregation of mesen-
chymal cells lying transverse to the median plane of the body.
In cross-section it is triangular, with the apex directed ventrally,
and each lateral angle of the base connected with the corre-
sponding first rib, there being a perfect and direct continuity of
tissue between the rudiment of the first rib and that of the
sternum. Proceeding backward in the series of sections two
bands of mesenchymal cells separate from the mass and extend
as far back as the level of the seventh rib, all seven ribs being
connected with and shading off into the sternal bands without
any definite demarkation. This is the stage in which Kravetz
(’05) found that the first ribs did not reach the sternal rudiment,
and the junction of the other six ribs was too feeble to have any
morphological significance. Whitehead and Waddell think this
is not a tenable conclusion in light of the intermediate position
occupied by this specimen, i.e., between older stages in which
no doubt of the absolute continuity of ribs and sternum exists,
and younger stages in which they seek for new facts concerning
the earliest relation between these two structures.
In a 20-mm. pig the pericardial cavity extends into the neck,
the ventral ends of the ribs are wide apart, and, in the region
anterior to the level of the first rib, the sternal rudiment is present
and is composed of three parts: the two sternal bands and a plate
of more diffused mesenchymal cells connecting the two across the
middle line. This stage presents two facts worthy of note:
the sternal bands are well defined separate structures at a level
considerably anterior to that of the first rib, and, further, in this
anterior region the two sternal bands are connected by a median
aggregate of cellular tissue. Posteriorly, in the region of the
ribs, there is the same continuity between costal extremity and
sternal band as in the 24-mm. stage.
Just as the older stage of Whitehead and Waddell (24-mm.
pig) corresponded to the youngest studied by Kravetz, so does
this 20-mm. stage in the pig correspond essentially to the youngest
stage in man found by Mueller (06), yet the interpretations of

these stages by the three authors are totally at variance.
The next stage was a 22-mm. pig, longer than the preceding
by 2 mm., but distinctly a younger stage in so far as develop-
ment of these parts was concerned. ‘The sternal bands and con-
necting bridge of mesenchymal cells are still anterior to the
first ribs; the pericardial cavity reaches far into the neck and
separates the sternal band and first rib of each side. Passing
caudally in the series of sections, the first rib falls just short of
reaching the sternal band; the extremity of the second rib ap-
proaches more nearly to the sternal band, and the remaining five
present the same continuity as before.
In the next younger stage, the 18-mm. pig, the heart is far for-
ward in the neck, with the consequent wide separation of the sternal
bands in this region; the median connecting portion is now absent;
the remaining parts of the sternal bands were traced from a point
150 « anterior to the level of the first ribs back to the level of
the ventral extremity of the seventh; the first two pairs of ribs
do not reach the sternal bands, but the other five are firmly
fused with it.
The earliest stages in which any sternal rudiment could be
detected were in 15- and 16-mm. pigs. The bands are not very .
clearly defined and stop at the level of the third and fourth
ribs, respectively; again the first and second ribs fail to reach the
sternal bands, and it is the opinion of the authors that, ‘judging
from the behavior of the ventral extremities of the first and second
ribs in somewhat older stages, we think it probable that a stage
exists in which no rib is connected directly with the sternal
bands, but we were unable to detect such a stage.”’
The cat embryos studied by Whitehead and Waddell (from the
Princeton Embryological Collection) ranged in size from 25 mm.
to 10mm. From the description given, they are essentially the
same as the several corresponding stages of the pig. In the
12-mm. cat the first three ribs clearly did not reach the sternal
bands, the fourth was uncertain, while the three posterior pairs
made the connection. Since, from the account given, the early
stages in the pig and cat are practically identical, nothing further
need be said of the sternum in the cat.
Several human embryos from the Johns Hopkins Embryo-

logical Collection, ranging in size from 17.2 to 10.5 mm. are de-
seribed in the paper now under review. The 13-mm. human
embryo corresponds essentially to the 18-mm. pig and 13-mm.
cat. The pericardial cavity extends far forward; the sternal
bands are traceable to the level of the ventral extremities of the
fifth ribs; the ventral tips of the ribs are non-cartilaginous, and
any possible continuity between them and the sternal bands is
much too slight to suggest that the latter is a derivative of the
former; neither median sternal rudiment nor clavicles were de-
tected, though this was probably due to the loss of an entire
slide of sections from the very region in which one would
expect such a structure to occur, if present. —
In a 10.5-mm. human embryo, the sternal bands reach pos-
teriorly only as far as the fourth rib; and although the cells com-
posing the bands are not sharply differentiated from the sur-
rounding tissue, they are still recognizable and ‘“‘it is evident that
they are not continuous with the tips of the ribs, but are con-
nected with them only by loose mesenchymal cells. ”’
In these stages which are far earlier than any Ruge describes,
there is a history quite different from that which later stages
- had led us to expect. In the first place, there was no indication
of segmentation in the sternal bands, which might be expected
were these bands due to proliferated cells from the tips of the
costal cartilages, and, second, in the pig the first two ribs in the
earliest stages studied did not reach the sternal bands at all;
three did not do so in the cat, while in the 10.5-mm. human em-
bryo none of the ribs reached the band.
These results also contradict the statement of Paterson that
the median rudiment is a part of the scapular arch, and that the
sternal bands are derivatives of the single median blastema.
Whitehead and Waddell do not mention the condition or extent
of the girdles in any of the forms studied by them, merely stating
that at no point was there any connection discernible between
sternum and girdles; and that in every case ‘‘the appearance of
the paired portion of the rudiment, the sternal bands, antedates
that of the median portion.”
Since the observations of Whitehead and Waddell accord with

neither those of Ruge nor of Paterson, the formulation of a new
theory of sternal origin was necessary.
Their discovery of the single median rudiment would not allow
of Ruge’s conclusion, for he thought the sternum a product of
the costal extremities; neither would they square with Pater-
son’s view, for he derived the sternal bands from the median
rudiment; and further, it was apparently not a product of the
sternal ends of the clavicles. So recourse was had to either of
two other theories, each possible so far as their observations go:
“first, it may be formed ‘in situ,’ or, second, it may be derived
from the anterior ends of the sternal bands by each of them send-
ing a prolongation medialward to join its fellow in the median
plane. The fact that we never found this rudiment in a paired
condition, but always as a single band of cells uniting the an-
terior ends of the sternal bands leads us to believe that the first
interpretation is the more probable.”’
Thus Whitehead and Waddell say there are two _ possi-
bilities remaining for the formation of the anterior median rudi-
ment, either that it arises ‘in situ’ or as a derivative of the sternal
bands. They overlooked another possibility, its derivation from
and relation to the shoulder-girdle. This may not have been an
unnatural error in view of the fact that they studied chiefly the
pig where the clavicle and that associated coracoidal mesen-
chymatous material of the early embryo is lacking. Their
view-point was derived from the developmental stages in one or
two mammals only, and they paid little attention to the com-
parative anatomical and the phylogenetic side.
This median rudiment is considered by Whitehead and Wad-
dell to be the homotype of the presternum of monotremes, but
no reasons or arguments for such a belief are set forth. This con-
clusion, however, is clearly invalid, for it is impossible to reconcile
the theory of ‘in situ’ for this anterior rudiment with their state-
ment that it finds its homologue in the presternum of lower forms
or the so-called prosternum of monotremes. If these be homol-
ogous, then the presternum and prosternum also arise ‘in situ,’
and no morphologist believes that they do.
The paired rudiments, or sternal bands, also arise very early,

according to Whitehead and Waddell, ‘in situ,’ one on either
side of the body and unattached in the earlier stages to the
ventral extremities of the ribs. The paired rudiments ante-
date the appearance of the median rudiment.
While Whitehead and Waddell and Paterson are very far from
being in accord as to the origin of the sternal rudiments, they
agree in demonstrating that the attachment of ribs and sternum
is a secondary fusion of parts, and that Ruge, while essentially
correct in his description of the stages he had under observation,
did not have the earlier stages and therefore was not in a position
to frame a theory of sternal origin. Without stopping now to
consider the relative merits of these two later theories, it must
be pointed out that their united efforts in overthrowing the Ruge
theory are of great value because of the hitherto widespread,
almost universal acceptance of this view.
5. Work of Kravetz, Rathke, Mueller, etc.
Kravetz (’05) worked on the pig. His youngest stage (24 mm.)
was also the oldest stage of Whitehead and Waddell. In the
24-mm. pig he found that the first ribs did not reach the sternal
rudiment, and, from the conditions in a series of later stages,
came to the conclusion that primarily there is no connection
whatever between the sternal rudiment and the costal cartilages
at their ventral ends.
Bruch (’52) describes the early stage of the sternum as two
longitudinal rods, one on either side, which later unite with each
other and with the ribs of their respective sides. He thus in-
directly denies a costal origin, but fails to indicate just what his
views were in this respect. It is highly probable that the ques-
tion was never raised in his mind at that early date. Whitehead
and Waddell add but little to the description of this early worker,
except that they have a theory of ‘in situ’ origin for the struc-
ture in question.
Rathke (’48) has an early, but very important paper in con-
nection with this discussion. His views are set forth in two short
paragraphs which are quoted in full as follows:
According to the researches I have made on the development of the

sternum in mammals, birds, and batrachians, this bone (sternum) may
be formed in a two-fold fashion. In mammals and birds it occurs
under the form of two very slender, long rods (italics mine), divided
_ into two lateral halves, and already at an early period consisting of
cartilaginous tissue, each of which rods unites itself with the extremities
of several ribs of its own side (italics mine) when these project them-
selves through a small part of the lateral wall of the body. ‘The two
halves, therefore, at first, lie at a considerable distance from each
other. Gradually, however, these two rods are approximated to one
another by the extension and development of the ribs, until, at length,
they come into contact throughout their whole length, and ultimately
coalesce, forming the sternum.
As regards the Batrachia, even in those which possess ribs, there is
never at any time two rods which unite the ribs and coalesce with one
another to form the sternum, but in some of these Amphibia there
originates a single cartilaginous lamina; in others a row of two or three
cae laminae quite independent of the lateral rays of the vertebral
column. aes
Rathke’s account of the origin of the sternum in birds and
mammals gives us a description of a stage far earlier than Ruge’s
youngest sternum. Ruge saw these ‘rods, long and slender’
only after they had been united with ribs, and therefrom made
his deductions that they originated from ribs. Over thirty
years before Ruge’s paper, Rathke accurately described this
very early stage, which had only been rediscovered in a few
mammals very recently. Ruge mentions Rathke’s work, but,
strange to say, makes no reference to this earlier stage in his
account.
This earlier account by Rathke is hidden in a discussion of
the homologies of the Chelonian plastron, and has not been
mentioned by any writer since Ruge. Its confirmation by Pat-
erson, Whitehead and Waddell, as well as by my own obser-
vations, renders Ruge’s theory of the sternum untenable.
However, accurate as this description of Rathke’s is, we must
not forget that he did not recognize its significance; in fact,
it is introduced into a paragraph describing and maintaining
that the sternum of the mammals is different in origin from that
of batrachians, not realizing as we do to-day that in separating
it completely in its genesis from the costal cartilages, he made
unnecessary a dual theory of sternal origin.
Charlotte Mueller (’06) worked out the development of the

thorax in a series of human embryos. She modeled the entire
thoracie framework, including vertebral column, ribs, and
sternum. ‘The series of models as pictured in her paper are fine
examples of the possibilities in careful modeling of large and
complicated structures. However, her youngest stage had the
sternal bands, though remote from each other, firmly fused on
either side with the costal cartilages, and, following Ruge, she
deseribes the sternal bars as arising from the ribs. She makes
little other contribution to the subject of sternal genesis, for her
material, like Ruge’s, was far too advanced for the early history
of this bone.
It is at once apparent that the theory of Parker, Ruge, Mueller,
etc., on the one hand, and those of Paterson, Whitehead and
Waddell on the other are mutually exclusive. But it would
seem that those of Paterson and Whitehead and Waddell have
only an apparent incompatibility, and that at least in so far as
their observations go, each was correct in reporting what he saw,
but the fact that they worked on very different forms and in-
terpreted their results in a widely different manner, leads to the
belief that they were looking on opposite sides of the same shield,
and that it is possible to reconcile the two.
The ‘in situ’ theory of Whitehead and Waddell is hardly to be
taken seriously. So ancient a structure as the sternum, dating
back to the Elasmobranchs, as we have seen, and found in every
higher group of vertebrates, including at least one or two teleosts,
and having an enormous and highly complex development in
many of the groups, can scarcely be accounted for in this simple
fashion. The ‘in situ’ theory took form from the material upon
which its authors worked. ‘The pig is the basis of their main con-
clusions and this form is peculiarly unsuited for this work,
because of the degenerate character of the shoulder-girdle, which
is lacking at once in both clavicle and coracoid process (the so-
called subcoracoid now being thought to be an epiphysis).
For a proper interpretation of the sternum and shoulder-girdle
in the mammals those groups with well-developed clavicles,
episternals, suprasternals, and other appurtenances of this
region are the only ones which can be lined up with the lower
groups of vertebrates. The rat was used by Paterson and the
mouse is the basis of the present paper, and this would seem to
be the ideal form; for the rodents, while highly specialized in
some respects, are primitive in others, and are to be grouped with
the Edentates and Insectivores somewhere near the monotreme
stem. They are also small enough so that when sectioned, com-
paratively high powers of the microscope may be used, and it is
possible to section all stages up to the ripe fetus.
Disecarding, then, Whitehead and Waddell’s theory of sternal
origin, while retaining an appreciative memory for their valuable
work in combating one of the remaining theories, we can reduce
the great mass of papers and discussion on this subject to Just
two absolutely irreconcilable theories of sternal origin, which
may for convenience in treatment be designated as Ruge’s
‘theory of costal origin,’ and Paterson’s ‘theory of coracoidal
origin.’ All other workers, except Whitehead and Waddell, have
supported one or other of these theories or modifications of
them.!
1 Through the courtesy of Doctor Kingsley, I have just received two papers
on the sternum, one of which requires mention. This is by Albrecht: Sur les
Copulae Intercostoidales et les Hemisternoides du Sacrum des Mammifers.
Bruxelles, 1883. It contains a most curious modification of Ruge’s theory of
costal origin. Albrecht’s idea is that the first and second ribs of each side at
first are united by an arch of cartilage, giving, according to his schematic figures,
a structure similar to a horseshoe magnet, the two arms of the magnet being the
ribs, the arch connecting them the sternal band. Then by a union of the two
sides in the midline, and the fusion of the consecutive pairs of such magnet-
shaped structures, a sternum is derived. This is ingenuous and is the only
theory of costal origin which gives the sutures between the sternebrae their
proper position, i.e., opposite the ends of the ribs, making the sternebrae inter-
costal as they actually are. However, the arguments used to overthrow Ruge’s
theory apply equally here. This theory does not account for the anterior and
posterior extension of the sternal bands for a considerable distance beyond the
region of ribs; it does not explain the appearance of the bands prior to their
union with ribs; and fatal to Albrecht’s hypothesis is the fact that the bars are
continuous, unsegmented structures throughout their entire length from their
earliest appearance in the mesenchyme, and never occur in short, semicircular
segments connecting the ends of the ribs. Albrecht was evidently unable to
find any stages in actual material in support of his theory, for his figures with-
out exception are diagrammatic, and do not fit the observed facts of sternal
development.
To sum up, then, there are extant at the present time in the
literature three opposing theories as to the origin of the sternum
in the Mammalia. The oldest and most generally accepted of
these is that proposed by Ruge.in 1880, which in substance
states that the sternum is a direct derivative of the ventral ends
of the costal cartilages.
In 1900 and more fully in 1902 and 1904, Paterson was led to
doubt the validity of Ruge’s theory, claiming that there was an
earlier history than that of which Ruge was aware. Paterson
derived the presternum from the same element which gives rise
to the shoulder-girdle, describing a continuous cellular element
crossing the midline in the rat. He derived the sternal bands
from this presternum as backward prolongations, which later and
secondarily are fused with the ventral ends of the ribs.
Whitehead and Waddell (11) agree with Paterson that Ruge
did not have the earliest stages, and that his theory is therefore
untenable, but they disagree with Paterson as to the interpre-
tation of these early stages. They deny any connection or re-
lation between sternum and shoulder-girdle, believing that both
presternum and sternal bands arise ‘in situ.’
This discussion of the literature is one of selected papers
which supports one or other of the different theories of sternal
origin and is fairly representative of the literature. However,
only a few papers are mentioned in comparison with the vo-
luminous literature extant. The author has collected a bibliog-
raphy of about one hundred titles on the sternum, but has con-
sidered it necessary to treat only a few of the more prominent
ones in this connection, with the assurance that those omitted
contain nothing new or affect the situation as outlined here.
It has added greatly to the confusion existing between these
opposing theories that most of the more important papers
(Ruge’s excepted) are very inadequately illustrated. Paterson
(02) does not give a single figure in this paper and his figures in
the 1900 paper are small and inadequate. Important stages are
described in the Whitehead and Waddell paper, but those upon
which they base their chief conclusions are not supported by any
figures. If we had clear-cut drawings of Paterson’s continuous
cellular element extending across the middle line, and could

compare this with equally well-drawn stages from the material
of Whitehead and Waddell, an unprejudiced worker, from a
study of the figures, supplemented by his own observations,
might bring the whole tangled mass into harmony. As it is, the
present author has but little to start with except the verbal
statements of the opposing theorists.
Table 1 gives at a glance the position of several of the leading
workers in their attitude toward the problem.
TABLE 1
HOMOLOGY
ORIGIN OF STERNUM FORMS STUDIED
OF PRESTERNUM
Ruge, G. (’80) Ventral ends of Man

ribs
Paterson (’00) Shoulder girdle Rat, rabbit, man, Middle part shark
Paterson (’02) dogfish girdle
Paterson (’04)
Kravetz (95) Two longitudinal Pig No morphological

bars importance
Mueller (’06) Ventral ends of Man Episternum

ribs
Parker (’91) Coracoidal and Hexanchus Ap- Omo-sternum in

costal teryx shark; omo-ster-
num, | in. am-=
phibia
Whitehead-Wad- ‘In situ’ Pig, cat, man Episternum; pro-

dell (711) sternum
Rathke (’48) Two longitudinal Batrachia, chick,

bars pig
Ill. THE ONTOGENY OF THE STERNUM
The earliest development of the sternum in a number of mam-

mals has been worked out by Paterson, Kravetz, Whitehead
and Waddell, and myself. An attempt is made to compare the
steps of development in ontogeny with those in the phylogeny
of the sternum, or in other words to make a practical demonstra-
tion of Haeckel’s recapitulation theory as applied to sternal
development.
1. The sternal bands
One decisive result of this investigation has been to demon-

strate the existence of the sternal bands as independent struc-
tures far earlier in development than Ruge and the older workers
suspected. Hence a new theory of sternal origin was demanded,
and as above indicated, this has taken two directions: White-
head and Waddell do not relate the sternal rudiment genetically
to any preexisting structure, while Paterson identified it with
the coracoidal girdle of lower forms.
The conclusive evidence against Ruge’s theory of costal origin
led the author to examine the material of Paterson and White-
head and Waddell, in an effort to confirm one or other of these
workers or reject both, as the case might be. Paradoxically
enough, I have been able to corroborate Paterson’s account
of the shark-like girdle, found by him in the rat, both in
the mouse and human embryos; and in the identical slides?
of cat and human embryos used by Whitehead and Waddell
have been equally able to confirm their observations. In pig _
embryos of 24-mm. length, Kravetz found that the first rib did
not reach the sternal band, and the connections of the remain-
ing six ribs were too feeble to have any morphological importance.
Whitehead and Waddell studied a 24-mm. pig in which the first
rib did just reach the sternal rudiment and the union of the other
six ribs was marked. My 24-mm. stage agrees with that of
Kravetz in that the first rib does not reach the band, and with
Whitehead and Waddell’s in that the connection of the other
2 My sincere thanks are tendered Dr. C. W. F. McClure, of Princeton Uni-
versity, and Dr. George L. Streeter, of The Johns Hopkins University, for the
loan of the series of cat and human embryo studied by Whitehead and Waddell.
six is marked. No special significance attaches to these varia-

tions since individual differences in development as well as in
measurements and amount of shrinkage gives form for consid-
erable variation. However, in stages of the pig 22 mm., 20
mm., 18 mm., and smaller, there is no doubt of the failure of
several pairs of the anterior ribs to meet the sternal rudiment.
Figures 3 and 4 are of two cat embryos from the Princeton
Embryological Collection. Camera-lucida drawings were made
of the sternal bands and ventral ends of the ribs. These struc-
tures were then plotted upon millimeter-ruled paper, which gives
a graphic reconstruction made to seale. Several wax models
were also made from the early pig and mouse embryos. Figure
3, of the 12-mm. cat embryo, shows clearly that in this stage the
first three pairs of ribs do not extend to and unite with the ster-
nal bands. It is also apparent that no anterior sternal rudiment
or presternal rudiment is present at this age. It is undoubtedly
true, as claimed by Whitehead and Waddell, that in the ontog-
eny of the mammalian sternum the two sternal bands antedate
in appearance the median and anterior rudiment. However, in
the phylogeny of the sternum, as will be shown further on, the
presternum is the first to arise, and from this come the sternal
bars. I am unable to account for this discrepancy by any ob-
served facts, but think the history from phylogeny must take
precedence over that from ontogeny; explaining the rise of the
sternal bands in the mesenchyme of the mammal as the result
of protoplasmic memory, which dates back to the early reptilian
ancestor in which the presternum grew backward as two pro-
-longations that became the mesosternum and the xiphisternum,
In human embryos from the F. P. Mall Collection studied
by Whitehead and Waddell and myself, is found the best evi-
dence of the complete separation of ribs and sternal bars in the
early stages of development. In embryos 10.5 mm. and 13 mm.
long none of the ribs reaches the sternum, the presternum has
not yet appeared, and no clavicles are apparent. ‘These stages,
if graphically represented, would appear similar to figure 3 of
the cat, except that all the ribs would be in the same relation
to the sternum as are the first three in the stage of the cat
figured.
2. The anterior median sternal rudiment
In the mouse, rat, and human embryos occurs a stage in

which a mesenchymatous girdle appears, in shape and rela-
tions comparable to the pectoral cirdle of the shark. Figures
5 and 6 show this girdle in the mouse and human embryos. It
is composed throughout of mesenchyme cells, and the structural
development of each part may be followed in later stages. In
the mouse girdle the two dorsally extending wings on either side
are the rudiments of the scapulae; the medial and ventral ex-
tensions are the coracoids and clavicles; the enlarged portion in
the ventral midline is the fundament of the presternum. ‘This
is conclusive evidence that the presternum is intimately asso-
ciated with the shoulder-girdle in the earliest ontogenetic stages
in the mammals, just as they are also phyletically bound to-
gether in the evolution of the vertebrate shoulder-girdle (infra).
The mesenchymatous material extending from the scapulae to
the presternum (fig. 5) is the track in which the clavicles will
soon develop. In the human embryo (fig. 6, cl.) this has already
commenced on one side. According to Gegenbaur and his fol-
lowers, the core of the clavicles is the old cartilaginous precora-
coid of the Amphibia. If it be true that the clavicles do have a
precoracoidal core of cartilage, as Gegenbaur thought, here is the
coracoidal extension in the human embryo reaching the prester-
num in the ventral midline, just as it does in Hexanchus, Amphibia,
Reptilia, Aves, Monotremes, and fetal Marsupialia (infra).
Gegenbaur’s clavicle containing a precoracoidal cartilaginous
core has been attacked in several papers by Broom, who denied
the presence of any cartilage in the earliest stages of the clavicle.
However, Broom admits that cartilage does appear at a later
stage in the development of the clavicle, and it may be assumed
that cartilage appearing either as a clavicular basis (Gegenbaur)
or at some later stage (Broom) would in this region in highest
probability be coracoidal tissue. This position is strengthened
when it is recalled that in the Anura a precoracoid actually
functions as the core of the dermal clavicle. Huntington* is
‘From a private communication containing Huntington’s views on several
shoulder-girdle problems, kindly prepared and sent to the author October 30, 1918.
the latest defender of the Gegenbaurian hypothesis, and, going

a step further even than Gegenbaur, declares that in the case of
the frog’s girdle there is a ‘“‘preparatory action on the part of the
coracoid cartilage directed toward the reception and assimila-
tion of the corresponding dermal accession of the clavicle.”
In later stages of the mouse and human embryos, after the
mesenchymatous girdle has become broken up into its component
parts, the coracoid process is relatively much larger than in the
adult and has a medial and ventral extension. I have observed
this repeatedly in embryos of pigs, mice, cats, and man. In one
mouse embryo, 7.75 mm., there seemed to be a distinct thicken-
ing of the mesenchyme between the very large coracoid process
and the yet partly mesenchymatous clavicle. This was a strik-
ing spectacle in the pig, because of the well-known fact that in
the adult no coracoid process is present, but only the subcoracoid,
glenoid-sharing portion.
3. Sternebrae
The segmentation of the sternum into sternebrae occurs late

both in phylogeny and ontogeny. A glance through the figures
in the literature assures one that sternebrae are unknown in
four of the five classes of vertebrates, being found only in the
mammals. Hence they play little or no part in the origin or
development of the sternum phyletically.
Likewise, they are a secondary and- acquired character in
ontogeny. It is contrary to all expectation, if sternal bands
are derived from ribs, to find that the sternebrae are invariably
intercostal, and not at the point opposite the ends of the ribs.
Figures 7 to 10 show what is the true condition in all fetal mam-
mals as regards the formation and ossification of its sternebral
elements. The center of ossification always occurs at a point
midway between two ribs, while the line of transverse division
crosses the sternum exactly in the center of the area of union
of ribs and sternum. According to Ruge’s theory, this should
be just the reverse. The sutures between the sternebrae should
be intercostal, the sternebrae themselves opposite the costal
cartilages.
Furthermore, if the sternum is ossified from the ribs, seg-

mentation of that structure should be apparent at its earliest
appearance. Qn the contrary, however, the sternal bands ex-
hibit no trace of segmentation until a late period of develop-
ment. The bands may be followed carefully from section to
section, or the parts reconstructed in wax, but no one has ever
reported the slightest indicationof an early division of the
bands into segments.
The sternebrae may be interpreted as arising by a process of
segmentation in response to the demand for as great a measure
of elasticity on the ventral side of the animal as is allowed by
the more or less flexible vertebral column on the dorsal side.
Sutures arising in this manner, as a response to strain, will
naturally appear at the weakest parts along the sternum. At the
points of attachment for the ribs the sternum is often deeply
notched, weakening this region, and here, as expected, occur
the lines of divisions of the sternum into segments or sternebrae.
That this is the cause and manner is indicated by the fact that
there are always the same number of sutures as there are pairs
of ribs attached to the sternum. By cutting a typical sternum
out of cardboard or a wax plate, and notching the sides for the
reception of ribs, it is possible by applying a lateral strain fo
produce sutures or cracks across the cardboard or wax sternum,
dividing it into sternebrae exactly as in the actual sternum.
In many of the reptiles and such animals as the cats among
the mammals, where a long, lithe body in making its way through
thick undergrowth or over rough ground is often twisted into
almost an S-shape, the advantage of a segmented sternum is
obvious. How large a part this plays in breathing is not so
apparent, but doubtless has some bearing.
In the Primates where the semi- and upright position obtains,
there is less need of flexibility, and the sternebrae tend to be-
come fused into the three typical parts of the primate sternum.
That the entire sternebral development is a secondary and late
acquisition and has no bearing on the origin of the sternum is
quite apparent.
So far as I am aware, this will constitute the only explanation

in the literature of the rise of the sternebrae, other than the
statement that they represent the original costal contributions,
which is, as we have shown, absolutely untenable.
4. Stages in the ontogeny of the mammalian sternum
1. Appearance of two laterally situated sternal bands, inde-

pendent of ribs.
2. Appearance of a single median anterior rudiment, inti-
mately associated with the shoulder-girdle.
3. Gradual approximation and union of sternal bands with the
anterior sternal rudiment on the one hand, and with the ventrally
growing tips of the ribs upon the other.
4. Gradual approach and fusion of sternal bands in midline of
body to form a sternum.
5. Division of sternum into a number of sternebrae. Lines
of division (sutures) always appearing opposite the ends of each
pair of ribs.
6. Ossification of the intercostal sternebrae by the appearance
of one or more centers for each segment.
7. Fusion of the sternebrae in Primates into three parts:
manubrium, gladiolus, and xiphisternum.
5. Conclusions
1. That the sternal bands arise and remain as two unseg-

mented structures until the relatively: late process of ossification
begins.
2. That the sternebrae are invariably intercostal; arise by
reason of functional demands for greater freedom of movement,
and play no part in the origin of the sternum.
3. That in the mouse and rat embryos a mesenchymatous
horseshoe-shaped girdle extends across the ventral midline;
from this material are derived presternum, coracoids, and
scapulae; this girdle being the homologue of the adult cartilag-
inous pectoral girdle of Hexanchus.
THE AMERICAN JOURNAL OF ANATOMY, VOL. 26, NO 1

4. That in early stages of the cat, pig, mouse, and human

embryos, the sternal bands exist as well-defined, separate,
mesenchymatous entities, prior to their union with the costal
cartilages, thus indicating their independence of, and the sec-
ondary nature of their relation to, the ribs.
IV. THE PHYLOGENY OF THE STERNUM
Paterson’s comparison of the ‘‘continuous bar :

across the middle line” in the rat with the cartilaginous scapular
arch of Acanthias, suggested the idea of following this structure
found by Haswell, and later independently by Parker, in the
middle ventral line of the shark Notidanus and identified by
them as a presternum, up through the various groups of verte-
brates to see how nearly it could be carried up in a phylogenetic
series to the rodents, in which Paterson thought he had detected
it again. For this purpose recourse was had to all available
figures extant in the literature, and especially to that monu-
mental monograph on the shoulder-girdle and sternum, by
Parker (’68). From these sources a series of figures has been
adapted, beginning with Parker’s (’91) figures of the presternum
in the shark, and including one figure from the Ganoids, and
one each from the Teleosts and Dipnoi; then numerous figures
from the Amphibia, Reptilia, Birds, and Mammals. Such a
search through the literature, though wearisome, has rewarded
the labor far beyond any expectations.
It is believed that the evidence presented in this phylogenetic
survey would alone go far toward convincing any one of the
truth of the conclusions arrived at in this paper, even though
it were not preceded by the corroborating evidence of the sec-
tion on the ontogeny of the sternum. It is the hope of the
author that it may lead to a general agreement as to the homology
and origin of the mammalian sternum.
1. Fishes
It is pretty clearly demonstrated by Haswell (’84) and also

by Parker (’91) that at least in one shark there is a presternum
derived from the ends of the coracoidal portions of the usually

single continuous pectoral girdle characteristic of sharks. Figure
11 shows the girdle of Notidanus from the ventral side with the
“intercepted cartilage . . . . temptingly like a presternal”
(Parker, 91). Figure 43, the first of the series of girdles shown
in plate 12, is another drawing from the same form. It would
seem that here is the initial material from which all later pre-
sterna might possibly be derived.
The author is of course aware that neither Notidanus nor any
other living shark is the direct ancestor of the vertebrates, and
that the following phylogenetic series of figures does not neces-
sarily mean that the successively higher animals are direct
descendants of those immediately lower which are used for illus-
trating the points.
Among the teleosts a complete girdle across the midventral
line is ordinarily lacking. However, several do have the clavicles
prolonged toward the center, and when so, there is quite uni-
formly a cartilaginous element in the midline which Parker
(68) calls the ‘epicoracoid.’* This lies between the coracoids,
in the identical position of a presternum. It is interesting to
note from figures 12, 13, 14, and 15, that this condition is so
closely alike in a ganoid, Polypterus (fig. 12); a teleost, Gobius
niger Linn. (fig. 13); and a dipnoan, Lepidosiren (figs. 14 and
15). A description of the condition in these three fishes may be
given in the account of Parker (’68) for that of Lepidosiren
annectens: ‘‘Lepidosiren agrees with the elasmobranchs in a
well developed epicoracoidal belt. Originally the epicoracoid
mass must have been double, and perhaps in a very early stage
each moiety was continuous with the coracoid proper, but a
wide transverse cleft was soon formed which separates epicoracoid
and coracoid.”’
‘In this paper the term ‘epicoracoid’ is employed to designate the cartilag-
inous ventral ends of the posterior coracoids. The author is following here the
usage of Parker and others from whom many of the figures have been borrowed.
That this is not the correct term, he is well aware, and in another paper is sug-
gesting the term ‘infracoracoid’ as probably a more suitable one for these parts.
The term ‘epicoracoid’ was applied to the anterior element of the monotremes
by Cuvier, and this use should be retained.
Figures 14 and 15 show the epicoracoid to have a large mid-

ventral portion, rounded out anteriorly and posteriorly (com-
parable directly with the so-called presternum of the shark and
the omosternum and sternum of the Amphibia), while laterally
extend two bars, the parts which originally were continuous
with the coracoids. Here there is cut ‘off from the coracoids
that material which nature will use in all higher classes of verte-
brates in the construction of a presternum. And if it be said
that this dipnoid fish is not the direct ancestor of the Tetrapoda,
it may be replied that every consideration points to the same
condition in the direct ancestor, whatever form it was. Both
Paterson, in the rat, and the author, in the mouse, found stages
in early embryo of these mammals where the clavicles, extend-
ing toward the midline, are closely invested with mesenchy-
matous bars which unite into a single, median mass, compar-
able at once to this epicoracoid in the dipnoid fish and to the
‘presternum’ of Haswell’s shark, and actually in the rodents
uniting to form the manubrium.
2. Amphibia
A similar sternum is found in the Amphibia. As Howes re-

marks, ‘‘that the Amphibian sternum is for the most part, if
not wholly, a derivative of the shoulder-girdle, there can be no
longer a question; and although the researches of Goette (’77)
leave us in doubt concerning the hypo (post-omosternum) they»
show that that can be no derivative of the costal apparatus.”
Parker (’91), believing in the diverse origins of the sternum
in the Ichthyopsida and Amniota, also describes a dual origin
for the sternum in the Amphibia. He says, ‘‘a pair of narrow
strips are separated off from the posterior borders of the cora-
coids,” also ‘‘a pair of cartilaginous bands appear in the in-
seriptiones tendineae of the mm. recti abdominis. From these
four elements the sternum is produced.” Ruge considers that
these cartilaginous bands are to be looked upon as vestigial
ribs. The narrow strips are admitted to be from the shoulder-
girdle.
This description of two cartilaginous bands appearing in the

inscriptiones tendineae is strikingly similar to the account of
the sternal bands in the pig by Whitehead and Waddell. There
seems to be no antagonism anywhere to the view that in the
Amphibia the anterior part of the sternum is the product of the
shoulder-girdle.
Nor is it necessary to stretch any morphological relations
unduly to see this coracoidal sternum reproduced in the Dipnoi,
Gobius, Polypterus, and Notidanus. We must agree with
Parker (’91), Haswell (’84), and others, that the sternum in the
Ichthyopsida is coracoidal in origin and homologous throughout
that group.
From here on, however, we part company with all early
workers, and most later ones as well, for all of them accept
Ruge’s conception of a costal sternum in the Amniotes, and
deny any homology between the sterna of Ichthyopsida and
Amniota. One of the designs of this present phylogenetic
sketch is to’show from figures and data already in the literature
that the two are one and the same thing in origin and develop-
ment, and therefore homologous.
As practically all are agreed that the sternum in the Amphibia
is coracoidal in origin, while but few are agreed that it is so in
the Amniotes, we are brought to the necessity of bridging in
some manner the alleged gulf between the amphibian and
reptilian sternum.
In order later to make direct comparisons between amphibians
and reptiles, when treating the latter group, it is necessary to intro-
duce here a number of figures and remarks thereon for several
representative amphibians. Starting with the frog, figures 16
to 19, inclusive, give a very good idea of the development of the
coracoids and sternum in this form. It is hardly necessary to
point out the close relation between coracoid and sternum nor
to suggest that the figures can lead only to one conclusion, that
of a coracoidal origin for the sternum.
Pipa (fig. 20) has a sternum which, as Parker says, chal-
lenges attention. A little study of this figure will convince one
that the sternal region must have been at one time a part of, and
continuous with the epicoracoids. ‘The ossification in each is

at exactly the same stage; the amount of soft cartilage around
the edges is the same in coracoid and sternum. At the anterior
end is differentiated a small region that corresponds to the
larger omosternum in the frog. We shall come back to Pipa
when treating some of the reptiles, and by a comparison attempt
to show that this subreptilian creature has a sternum essentially
like certain reptiles.
Does Bufo (fig. 21) have a sternum? Kingsley (17) would
say that since no ribs are in this region, no one can say. On
the theory of Ruge, this would be true, but if we compare the
sternum of Bufo with that of Pipa and the early stages in the
frog, it is hard to believe that there is any essential difference
between the two structures, although the connection of the
sternum with the epicoracoids in Bufo is not so extensive as in
many other Amphibia. The entire body of evidence in this
paper and many others on the shoulder-girdle in the Amphibia
can only lead to the conclusion that Bufo does have a sternum
and that it can only be derived from the coracoidal portions of
the shoulder-girdle.
Siredon (fig. 22) has large coracoids and a considerable over-
lapping of their epicoracoidal edges. The interesting feature to
us is that the sternum lying immediately behind the overlapping
epicoracoids also shows very distinctly two grooves correspond-
ing exactly to those made by the overlying edges of the epi-
coracoids. It is the condition precisely to be expected of the
sternum if it were in an early stage a posterior continuous ex-
tension of the cartilage, sharing in the overlapping, and then later
had been cut off by sutures from the main element, but retaining
these evidences of its formation from the two epicoracoids.
Little need be said concerning such a girdle as that of Dacty-
lethra (fig. 23). There is no overlapping of the coracoids and no
sutural evidence in the sternum of the union of the two sides;
however, as this is an adult specimen, none need be expected.
Nevertheless, the intimate relation of sternum and girdle is
evident.
In Calamites (fig. 24) we see a sternum that for the first time
has two backward prolongations, or sternal bars. ‘Tims sternum

is very suggestive of several in the reptiles, and embryos of the
mammals, where a single median anterior rudiment is continued
backward as two rods or bars. This is an adult specimen and
therefore the permanent form in this species; if, however, we com-
pare this with several of the reptiles, such as shown in figures 26,
27, and 28, and with descriptions by Rathke, Bruch, Paterson,
ete., of the early sternum in the mammals, there appears to be
more than a mere resemblance—there is genetic relationship and
homology. If in Calamites the coracoids were to retreat to a
mere process attached to the scapula, leaving only the clavicle
and sternum in this region, and the sternal bars of the latter
fused together in the midline, leaving a fossa at the upper end of
the union as in many reptiles, and the posterior ends were but
incompletely fused leaving two small blunt laterally projecting
horns, we would have a sternum such as is actually found in
Chirotes (fig. 30) and the embryos of mammals.
Wilder (’03) describes several cartilaginous rudiments found in
Necturus and related by him to the sternal apparatus. These
are a series of thin cartilages located in the myocommas of the
pectoral region. One of them is usually larger than the rest and
situated near the posterior part of the overlapping coracoids.
This element is identified by Wilder as the homologue of the
sternum of the higher Urodeles. ,
If Wildér’s theory be correct, Necturus presents an exception
to the rule established in this paper that the presternum is a
derivative of the coracoids, for obviously this element in Necturus
could not possibly be derived from that source. My own dis-
sections of Necturus, however, do not bear out Wilder’s hypoth-
esis. My interpretation of these cartilages is, that they are
simply chrondifications of the outer part of the connective tissue
of the intermuscular septa and have no relation whatever to the
formation of the presternum in higher forms.
They are rather to be looked upon as subcutaneous splints
and find their homologues in the inscriptiones tendineae of other
forms, and also possibly in the abdominal ribs of Chamaeleo and
Polychrus. It is significant to note in this connection that many
animals have both the inscriptiones tendineae and also a complete

sternal and costal apparatus. If the latter be the derivative of
the former, why this persistence of the two structures side by side
in nearly all groups above the lower Urodeles?
3. Reptiles
Fossil reptiles constitute the next group in which we have

looked for a sternum or any part of one which either originates
independently of the ribs or which has intimate relations with the
shoulder-girdles. The author has studied figures and plates of
much of the recent work done on fossil reptiles, of which the in-
vestigations of Credner (’81—’93), Gregory (15), Seeley (’92, ’94),
Woodard (’98), and Zittell (’00, 718) are typical, and, in addition
has examined a number of mounted specimens in the U. S.
National Museum. It may be said in general that in those
skeletons which have the shoulder-girdles preserved there is a
strong tendency for the coracoids to grow around the side of the
body ventrally, though never meeting in the midline, for the soft
epicoracoids are not preserved, and that the coracoids are enor-
mously developed in size.
Gunther (’67), Schauinsland (’00), and Howes and Swinnerton
(01) worked on the development and anatomy of that primitive
and now almost extinct reptile from New Zealand, Hatteria
punctata or Sphenodon. In one of Gunther’s figures the
shoulder-girdle and sternal bars are shown in their natural
relation at thatstage. The coracoids do not have a large ventral
extension, but are capped on their medial ends by slender proc-
esses which later uniteto form the presternum. Three pairs of ribs
reach each sternal bar, so that the sternal and costal connections
are much too far advanced for any statement as to the origin
of the sternal bars in Sphenodon. So far as I am aware, this is
the youngest stage of the Sphenodon sternum figured in the
literature.
From the condition in the embryos of living reptiles, as well as
in many of the adult species, it may be assumed that in the early
condition of fossil reptiles the coracoids were in intimate relation
to the sternum.
a
ONTOGENY AND PHYLOGENY OF THE STERNUM ral
When we come to consider the living reptiles the material and

evidence is abundant and of the greatest significance.
Among the recent forms, Goétte (’77) has studied Cnemido-
phorus. While he does not show conclusively that the two an-
terior triangular rudiments are products of the nearby coracoids,
he demonstrates that only the first ribs have reached and attached
themselves to the sternum, while that part of it opposite the
second and third ribs is formed independently of them, as a
backward prolongation of the anterior paired rudiments.
These results of Goétte showing that the sternum is the result
of a backward growth of tissue from an anterior portion are in
striking anticipation of what Paterson claimed to see in the rat,
where it is stated that the anterior median portion of the sternum
is derived from the same element as is the shoulder-girdle, and
this in turn yields the sternal bands as posterior prolongations.
Anguis fragilis (fig. 25) has a sternum that in many respects is
typically amphibian in character, as may be seen at once by a
comparison of this form with that of Pipa. As in Pipa, there is
only present an anterior or presternum, separated by sutures only
from the epicoracoids, and as in Pipa, ossification and the amount
of soft cartilage around the borders are the same for sternum and
coracoids. Of this form Parker (’68) says, ‘“‘there is a well devel-
oped sternum, not continuous with the ribs;’’ and Rathke, describ-
ing two embryos of Anguis, says, ‘‘ with the coracoids the sternum
was intimately united, but it was not very closely connected with
the neighboring ribs, lying at a much greater distance from them
than in adult Blindworms. There cannot be any doubt that in the
Blindworms the two latter halves of the sternum do not :
originate under the ribs, and unite with them, but develop at a
distance from the ribs.”’ Barring the presence of the interclavicle
in Anguis, it would be difficult to recognize this shoulder-girdle
as being that of a reptile, for in the relations of its girdle and pre-
sternum and in the absence of ribs it is characteristically
amphibian.
Stellio cordylinus (fig. 26) has a sternum in which the anterior
part is enormously enlarged; is in intimate relation to the epi-
coracoids, and in addition has two greatly extended xiphisternal
horns. No ribs are attached to these horns, but three pairs reach
the anterior portion, to which they are but feebly attached. If it
be assumed that the presternum here is of necessity derived either
from the coracoids or from the ribs, the answer can only be that
it must have come from the former. This sternum with its
great horns reappears in Manis longicauda (Parker, ’68).
A series of three figures (figs. 27, 28, and 29) gives an idea of
how the mesosternum and xiphisternum are formed. In figure
27 the presternum is as before, and the two bars extend caudally.
This is very similar to the amphibian Calamites (fig. 24), and the
suggested series of changes outlined in the description of Cala-
mites necessary to make of it a typical reptilian or mammalan
sternum are progressively illustrated in these three reptiles.
In figures 28 and 29 the xiphisternal bars, by a fusion along
their medial surfaces, have formed a middle sternal piece or
mesosternum. The posterior ends of the coalesced bars remain
apart in the xiphisternum.
In the mesosternum is a sternal fossa, where the union was not
complete. This may persist throughout life in many forms
(Varanus, Crocodilia) or, as in others, close up later, leaving a
whitish streak to indicate the line of fusion. This fontanelle is
also common in mammals, but there it is usually located in the:
xiphisternum, and I have also repeatedly observed the whitish
streak of hyaline cartilage in the mesosternum in fetuses of pigs
and mice. It would seem from this reptilian material, and avian
and mammalian material agree, that the presternum is a product
of the coracoids, and this in turn gives off two backward prolon-
gations, which, fusing throughout a greater or lesser part of their
extent, form the mesosternum and xiphisternum. It is hardly
necessary to again point out the feeble relation of ribs and ster-
num in these last three figures.
Chirotes (fig. 30) gives the completion of the series; it is a
sternum of the utmost importance in the consideration of the
problem. Parker’s description is so trenchant that a part is
quoted:
In the whole range of vertebrate morphology there is nothing more
beautiful or more instructive than the relatively large sternum of Chi-
rotes; for if the sternum of the human embryo were to be demonstrated

apart from the costal girdles, one diagram would serve to explain both
that and what we find in this little snake-like lizard . . . . and
if ribs had not been arrested we might have seen the counterpart of
the ribs of the mammalian embryo.
The figure is of an adult, and, as Parker says, it might also be de-

scribed as that of the mammalian embryo, except that no ribs are
_ present and the sides of the presternum are closely applied to the
coracoids. Now, if Paterson and the author be correct in their con-
tention that the presternum and coracoids in the rat and mouse
are continuous at an early stage, and this stage precedes the
fusion of ribs and sternum, then the adult Chirotes sternum is a
structure that far more closely approximates the early embry-
onic sternum of mammals than Parker suspected when he made
the comparison above quoted between the two. This is a more
striking parallel than is usually met between the adult structure
in a lower group and an embryonic stage of the same structure
in a higher group.
While probably all will admit the above argument, since Beth
are Amniotes, the question may be raised as to whether there is
any evidence for relating such a sternum as occurs in Chirotes
with the amphibian sternum, for the crux of the whole matter of
sternal homology lies between these two groups. It would seem
that a direct comparison might be made with the Calamites
sternum, and this structure, as has already been indicated, by a
fusion of the sternal bars, and a retention of the sutural relation
to the epicoracoids might be metamorphosed directly into the
adult sternum of Chirotes, and this is especially strong evidence
when we consider that in neither the reptile nor amphibian com-
pared does the question of ribs enter at all, since there are none
in either form in the region of the sternum.
Using the crocodile (fig. 31) as a contrast to Chirotes in show-
ing the extreme of fusion of ribs and sternum in reptiles, it may
be remarked at once that aside from the presence of ribs this
sternum is directly comparable with the last one considered. The
difference is due to the ventral growth and attachment of ribs to
the sternum, otherwise it is essentially the same as more primitive
reptilian sterna having a presternum in close conjunction with the

coracoids; a middle piece composed of the union of two longi-
tudinal bars, with the line of fusion clearly evident, and the
xiphisternal horns wide spread. A close comparison may be
made between this and the later stages of the mammalian ster-
num, except that in them all connection with coracoids is early
lost.
A Chamaeleo vulgaris adult sternum (fig. 32) is the last reptile
considered here. It is mammalian-like and well ossified for a
lizard. The presternum is a large, lobate structure, bearing two
strong notches on either side on its posterior end, ‘this constric-
tion answering to the transverse cleft so constant in the mam-
malian sternum” (Parker, ’68). In the mesosternum the line of
fusion of the two halves is well marked and extending also into
the presternum. The ribs articulate by synovial joints with a
series of enlargements on either side of the mesosternum. In de-
scribing the xiphisternum, Parker (’68) says, ‘“‘The xiphisternum
has a bilobate extremity that is quite mammalian in character
and no ribs ever reach this part . . . . the horns being
free from ribs, grew not only towards each other and fused, but
also grew backwards, so as to form a free, single xiphisternum,
exactly like that of an ordinary mammal. That there is no real
difference between these two classes in the formation of the xiphi-
sternum, I feel certain. : |
The interesting fact about "Ghaitaeles is the statement of
Parker that behind the xiphisternum there are seven pairs of
floating ribs which later become fixed by growing toward each
other and unite by suture at the midline. This is significant in
the light of our contention of the non-relationship between pleural
ribs and sternum. ‘There is a distinct tendency in all vertebrates
with ribs for these to grow ventrally. Now, if a sternum be
present, it is likely that they would form an articulation with it;
if none is present, that they should either remain free (floating)
or unite (true ribs) with each other in the midline. In the
chamaeleon both conditions are present in one animal; those an-
terior thoracic ribs which, growing ventrally, met the sternum
and articulated with it, while those ribs immediately behind the
sternum kept on growing until the respective pairs met and fused
in the midventral line. Chamaeleo is considered to present
strong evidence of the secondary character of the relation between
ribs and sternum, and it is an important intermediate stage in
the development between the typical reptilian sternum and the
same structure in the mammals.
The preceding account of the development and anatomy of
the sternum in the Reptilia and a comparison with the same struc-
ture in the Amphibia must inevitably lead to the conclusion
that, if the presternum be coracoidal in origin in the Amphibia,
it is equally so in the Reptilia. For, beginning with that most
primitive reptilian sternum in Anguis, and comparing with Pipa
and other Amphibia, the gulf was bridged between these two
phyla, and then by a series of successively more highly developed
sterna in the reptiles a stage is reached (Chirotes) which spans
the divide between the reptiles and mammals. We have also
seen how in one amphibian (fig. 24) the beginning of sternal
bands arises, and in the reptiles these are developed in the same
way, and in higher reptiles fuse to form the mesosternum and,
xiphisternum, preparing the way for the typically mammalian
sternum, soon to be considered.
The last fact concerning the Reptilia is in regard to the ex-
tremely variable relation of ribs and sternum, both as to number
and position. Rathke (53) had an interesting paragraph on this
which is quoted in part:
In typical sealy lizards several ribs are always in relation with the
sternum; still . . . . it may be either only the anterior divi-
sion (manubrium) which is connected with ribs, or it may be exclusively
the posterior part. But, generally speaking, the number of ribs which
are intimately connected with the sternum, and to which the name of
true ribs can be applied, not only varies with the genus, but is also
very variable in different species.
Both Rathke and Parker give long lists of species of reptiles

showing this variability in the number and position of ribs
reaching the sternum. Their lists comprise some fifty species,
but only a few are mentioned here as indicating the range of
variation found by them. Their figures show that in some species
76 FRANK BLAJR HANSON
with the largest number and most intimate union of ribs to

breast bone the sternum is but feebly developed; while, on the
other hand, some of the largest sterna have the fewest number
of ribs and feeblest connection of the two; or, as in Chirotes, a
fully developed sternum showing the three typical divisions is
present, but no ribs reach the sternum—conditions hardly to be
expected if the ribs contribute the sternal materials.
The following table is a composite one from several authors and
shows but a few of the numerous forms they list.
RIBS TO RIBS TO
MANUBRIUM MESOSTERNUM
G@hurotes, canaliculatusae i dork acetals foe

Anpipis: {a etis: 6:2. eR Pee og ot are Sek Race
Chanracleo wurailisy (eer. jee - Wye sees Stes oo (om)
leh
Monitor ‘dracvenaien ane Pons ote. Seema ts a

Draco tvinidisie es Abia eee ean eee ears
Stellio: aol aris: Ohare eis ae See AS ee ae
alates MiGbust geen Milas .te Recto elas etna eet
Horan; hmoerCUlatans. ace) St tai. peepee ess
CRGcodiie Aes Ie! NS. IY Re EER eer ete
Gantialis,cellegeliat aes b.c 3 3) AR id aed op ed eae eR W&
Ww
Be
Nw
Wb OF
NJIOnNWOrF
ooo
4. Birds
Birds, although not in the line of descent of the mammals and

also having a sternal apparatus highly modified for purposes of
flight, are still not difficult to bring into line in this argument.
However, it is hardly necessary to do this as we are following
successively more complex and highly differentiated groups in
their phylogenetic course. Nevertheless, to show that there is
nothing contradictory to our thesis among the birds, one figure
of a bird is introduced. This is Vanellus cristatus (fig. 33) and
is of a stage at the end of the first third of the incubation period.
Parker’s remarks upon this sternum give us the pertinent facts:
““The longitudinal bands are long and wide, and in great contrast
to the very slender pairs of ribs attached to them. On the other
hand, a transverse cleft between the epicoracoids and the antero-
lateral margins of the longitudinal bands would give us the con-
ONTOGENY AND PHYLOGENY OF THE STERNUM b ~“I
dition as here found. The sternum of Vanellus is not a highly

complex structure, as such structures go in the higher birds, but
shows many affinities to the reptilian stock.”’
It would be absurd to think that these large, heavy sternal
bands were originally derived from the feeble and loosely attached
pairs of ribs, while it is entirely plausible to suppose that in an
earlier stage a continuous sheet of cartilage on either side was
subsequently differentiated into sternal bands, coracoid, and
scapula. That this is actually the state of affairs in early embryos
of several mammals will soon be shown.
That the development of the sternum is largely independent of
ribs in the birds, as in reptiles, is further shown by Parker’s ob-
servations on Apteryx:
In the earliest stage in which the sternum is present it extends

backwards to the level of the third thoracic rib; the first two ribs are
united to it by joints, the third loosely attached by connective tissue.
In the next stage, the first three ribs are attached by joints, and the
fourth by connective tissue, that is, as it appears to me, the portion of
the sternum corresponding to the third and fourth ribs is formed by
a backward growth of the anterior region and quite independently of the
last two ribs (italies mine), the union of which with it is a secondary
process.
This is just about what Paterson says concerning the devel-

opment of the sternum in the rat, except that it leads him to
conclude that the sternum is not of costal origin, while Parker,
giving all the evidence necessary to substantiate Paterson’s
view, is nevertheless himself oblivious of the logic of his own work,
and makes the surprising statement that “the relation of the
shoulder girdle to the sternum is altogether secondary, and forms
no part of the axial skeleton, as the Transcendentalists vainly
teach.”
The above account of conditions in the birds seems to be suf-
ficient to relate genetically the sternum in birds to that of rep-
tiles and mammals, and while of little or no philogenetie value
in this connection, yet the evidence shows the avian sternum to
be homologous with the sterna throughout the entire vertebrate
series, one of the theses for which this paper contends.
|a FRANK BLAIR HANSON .
5. Monotremes
The monotremes (fig. 4) have long been held to have a sternum

closely reptilian in character. In addition to the coracoids,
which are firmly attached to the sternum, there is also an an-
terior paired element, usually called the epicoracoid. In these
there is an overlapping in the midline much as in certain amphib-
ians, and in a recent paper Watson (717) holds that the epicora-
coids of the monotremes are nothing other than the precoracoids
of the lower forms. That typically reptilian structure, the inter-
clavicle, is also present. As seen in the figure, there is notice-
able a compactness of the elements of the shoulder-girdle and
sternum, as if these might have been in the early embryo or in
the ancient progenitor a single shield-like plate, such as occurs
in Pipa (fig. 20) or in many of the reptiles.
6. Marsupials
In 1897 Broom discovered in the marsupial Trichosurus vul-

pecula, measuring 17 mm., a well-developed coracoid, which was
at birth ‘‘structurally continuous with the sternum.” Figure 35
is an anterior view of the entire scapular arch which in general
outline is strikingly like that of the shark. Its ventral middle
portion is a part of the sternum, yet the parts are, as Broom says,
not jointed, but constitute a single bar of cartilage.
In a smaller specimen (fig. 36) of 8.5 mm. of the same animal
both cartilaginous and mesenchymatous elements are present in
the coracoid; the cartilaginous part being that nearest the glenoid
cavity, and the mesenchymatous spreading out in a fan-shape
portion, which is continued without any interruption into the
mesenchymatous sternum. It is apparent that this marsupial
has a complete scapular arch crossing the middle line; and that,
from the history of the later stages, the sternum as well as the
girdles are known to be its derivatives. Furthermore, just in
front of the coracoid and posterior to the clavicle, there was ‘‘a_
thin, feebly developed continuous sheet of mesenchymatous
cells;”’ lying therefore in the exact position of that anterior cora-
coidal element of the monotremes generally called the epvicoracoid,
but by some the precoracoid. That the two structures, 1.e., the
epicoracoid of the monotremes and this “‘sheet of mesenchy-
matous cells’? are homologous is the belief of Broom. In the
larger specimens (37 mm. and over) the coracoid becomes de-
tached from the sternum by a process of degeneration, and this
continues until the well-known adult condition is reached (fig. 37)
where coracoids and sternum are far apart.
However, in a mammary fetus 23 mm. long an intermediate
condition was found. As Broom describes it:
The coracoid process is similar to that in the large foetus, but from
it there is produced backwards and inwards a small cartilaginous proc-
ess, which nearly meets the outer process of the presternum. It may
thus be concluded that during the later intra-uterine development of
Trichosurus, and probably of other marsupials (later verified in other
marsupials), there is a well developed coracoid, which, as in the adult
Monotremes, most reptiles, birds, and amphibians, articulates with the
sternum, and that shortly after birth, the coracoid loses its attachment
with the sternum, and becomes rapidly absorbed, only the anterior
part remaining as the coracoid process. .
From this description it appears that the degeneration of the

coracoid begins in its middle part and absorption progresses
toward each end. In the marsupials that part of the coracoid
attached to the sternum is completely absorbed and no trace of
it is found in the adult, while the half connecting with the scapula
is represented in the adult by the coracoid process. In this con-
nection mention may be made of a peculiar structure I find in
the mouse and rat, and an almost constant structure in rodents,
as figured by Parker (’68). In most of his figures of the Rodentia,
there is a small bony process on either side of the presternum
between the juncture of the clavicle and the first rib with the
sternum (fig. 38). Parker calls this process the epicoracoid and
says it was left by the retreating coracoids of the lower forms. It
would seem from my observations on the mouse, described above,
that Parker is correct to the extent that this is the median end
of the coracoid, but wrong in making it simply an hereditary
rudiment. It is, rather, the end of a complete embryonic cora-
coid, which in the rat, mouse and man, as in the marsupial,
extends across to the sternum in an early stage. Whereas in the
.
marsupial and most other mammals, all trace of this sterno-

coracoidal piece is lacking in the adult, it persists in the rodents,
and also in the narwhal (Monodon monocerus), and the black-
fish (Globicephalus melas), and possibly other animals. In
other words, in the rodents there is at first a complete coracoid
reaching from the scapula to the sternum; later, degeneration of
the coracoid sets in at its middle portion, and working toward
each end, stops just in time to leave the coracoid process attached
to the scapula and the epicoracoid to the sternum.
From the foregoing it is apparent that Trichosurus passes
through a stage in its development, as regards girdles and sternum,
which is directly comparable with that of the adult condition in
the monotremes.
In aseries of later papers (’98,’99’02,’08) Broom studied a num-
ber of marsupials, both Polyprotodonts and Diprotodonts, and
found that conditions throughout the Marsupialia were as described
for Trichosurus. In a mammary fetus of the common phalanger,
14 mm. long, the well-developed coracoid articulates with the
sternum almost exactly as in the adult monotremes. In the
earliest stage of Dasyurus (fig. 39) studied by Broom, the cora-
coid is still large and reaches nearly to the sternum. It would
seem that here absorption had commenced at the medial end
rather than in the middle of the cartilage. Figure 48 (petrogale)
shows a complete, unjointed arch similar to that of Trichosurus.
Pseudochiurus and permales were studied in various stages and
agree in general with other marsupials, so that this is obviously
a normal and constant phenomenon of the marsupial embryonic
girdle, as it is of the adult girdle in the monotremes.
Prior to Broom’s work, it was difficult, practically impossible
in fact, to pass from the monotreme shoulder-girdle to that of
the marsupial, albeit the relationship of monotreme girdle to that
of the reptile was apparent. Since Broom’s discovery, however,
of the complete girdle in the early marsupial, and also in the rat
by Paterson and the mouse and man by the present author, it
is clearly seen that the girdles and sternum of the higher mammals
are directly comparable to that of the marsupial fetus, and this
in turn to the girdle of monotremes, which are unquestioned in
their reptilian affinities. It is thought that this makes a strong

case for the homology of the sternum in the Amniota and its
origin in connection with the coracoids; and since this same struc-
ture is undoubtedly coracoidal in the Ichthyopsida, the two,
Amniote and Ichthyopsidan sternum, are homologous.
As this work of Broom’s is of considerable importance to our
argument, it may be stated that Broom in his later papers veri-
fied his first discoveries in representatives of all groups of mar-
supials and made graphical reconstructions of the parts. His work
has been checked up by Watson (’17), who made wax models
of the girdles and sternum, and completely confirmed Broom’s
results.
7. Aquatic mammals
Among the aquatic mammals there are several interesting

sterna. In the adult Manatus americanus (fig. 40) the sternum
is moderately large and is typically divided into the usual
three parts. This animal has seventeen pairs of ribs, but only
three come anywhere near the sternum, and Parker (68) says that
only the second pair of ribs reaches it. The first and third pairs
are connected with it by ligament only. Here in this adult
mammal is a stage comparable to the embryo of the pig, man,
etc., where ribs either do not reach the sternum or are connected
with it by fibrous tissue. It is difficult to believe that this entire
sternum was derived from the costal ends of three pairs of ribs,
of which only one pair even approaches it.
According to W. K. Parker (’68) the sternum of the dolphin
embryo (fig. 41) has reached its highest development in aquatic
mammals. In the stage figured the sternal bands have fused,
leaving a prominent longitudinal groove to mark the line of
fusion, several centers of ossification are present, and in the pre-
sternum is an oval fontanelle such as is common in the lizards.
In fact, this entire structure is very reptilian in character, as is
seen by a comparison with Trachydosaurus (fig. 28), even to the
number of ribs.
In orders of mammals higher than those already mentioned
the literature is scanty. Practically all papers based upon the
82 FRANK BLAIR‘ HANSON
human embryo accept Ruge’s view, and the reason is apparent

when we consider that in working exclusively with the highest
mammals, one is at the disadvantage of not knowing what stages
may have been suppressed in evolution. It would seem that,
beginning with the lowest forms and working up through each
successive group, as is attempted in this present paper, a foun-
dation is laid upon which to interpret the greatly reduced history
in the higher mammals.
Does the complete girdle, sternum, coracoid, and scapula, cross
the midline in forms higher than the marsupial? Since the per-
manent adult coracoid articulating with the sternum in the mono-
treme is reduced to only a fetal stage in the marsupial, we might
expect this stage to be very transitory or entirely suppressed
higher in the scale. Since the marsupials, edentates, rodents,
and insectivores are all ancient orders and probably lie not far
from the monotreme stem, among them such a stage might be
found, if present. Paterson was the first to discover this in an
early embryo of the rat, and I found the same thing (section on
ontogeny) in the mouse of 7.75-mm. and 17.2-mm. human em-
bryo. This is shown in figures 5 and 6, where the girdles are
very similar indeed to that of a shark (fig. 2). The next stages
indicate that scapulae, coracoids, and sternum are all derivatives
of this single mesenchymatous element. This stage is exceedingly
brief in the mouse, as compared with the marsupial, for it cannot
be detected with certainty in a 6-mm. mouse, and is hardly rec-
ognizable in the 8.75-mm. mouse. This rapid suppression would
lead us to suppose that possibly no such stage is present in the
human embryo, yet figure 6 shows man to have retained it iden-
tically as in the rodents.
Huntington (’18) identifies the costocoracoid ligament of man -
as indicating ‘‘the original path of the sternal extension of the
coracoid.”” It is interesting in this connection that the costo-
coracoid ligament often contains fibrocartilaginous nodules.
Figure 42 shows this ligament in relation to the other parts in
man on the left side of the figure, while on the right side, the fun-
damental plan of the vertebrate girdle as illustrated in the Anura
is shown. Huntington does not say whether he considers the
YNTOGENY AND PHYLOGENY OF THE STERNUM 83
costocoracoid ligament to be the remains of an embryonic cora-

coid occurring in man or only the rudimentary indication of
man’s phylogenesis. If this interpretation of the costocoracoid
ligament as the evidence of a coracoidal connection with the
sternum be correct, then from the elasmobranch to man the
closest relation between coracoids and sternum is maintained.
On the other hand, if Huntington’s hypothesis is not sustained
by later investigation, no harm is done to our attempt to over-
throw Ruge’s theory, for Rathke, Paterson, Whitehead and Wad-
dell, and myself have already shown that in man and other
mammals the sternum is an established structure prior to the
union of ribs with it.
8. The adult human sternum
In the literature of the adult sternum in man there is evi-

dence of a secondary, but interesting and corroborative sort.
It may be related back to the tendency of the ribs to grow toward
the ventral median line and fuse or articulate with each other, as
was noted in Chamaeleo. This has been shown by several
workers, Cunningham (’90), Dwight (’90), Tredgold (’97), Pat-
erson (’09), and Lickley (’04), to be especially prominent in the
human subject. Lickley, whose work is typical, studied a
series of fifty-one human sterna with special reference to the
relations of the seventh and eighth ribs to the sternum.
The number of ribs reaching the sternum in Lickley’s material
is shown in the following table:
RIGHT LEFT
SISONET Ser oe te ee ee Re ed ca I 2 1
Seventies TID Sey ees yea ee: ee ey See Bes 43 43
RETIN ceMbe vs. onicre comes pepae ade take dia Dolce ey 6 4
las Ws Se LG 51 651
This constitutes a considerable percentage of variation at the

posterior end of the thorax, yet whether six or eight ribs enter
into union with the sternum, that structure is apparently not
84 FRANK BLAIR HANSON "
affected in length or shape. Advocates of a costal origin for the

sternum derive the xiphisternum from the ends of the seventh
or eighth ribs or both. However, when these are absent, the
formation of the xiphisternum does not seem to be affected in
the least. As in the reptiles, it is a constant feature regardless of
the absence or presence of ribs.
In the work of Paterson (’09) and Lickley (04) an even more
suggestive fact is clearly apparent. This relates to the tendency
of ribs to grow to the ventral side and fuse as was also seen in
Chamaeleo and noted in the discussion of reptiles. In man the
seventh rib normally reaches the sterrium, but in its mode of
attachment we have the remarkable statement of Lickley that
in over 50 per cent the seventh costal cartilages are either
fused or articulate with each other in a plane anterior (ventral)
to the xiphisternum. That is to say, these ribs which are anterior
to the posterior end of the sternum reach the median line and fuse
just as did the ribs posterior to the posterior end of the sternum
in Chamaeleo. In all the accounts here under review the ribs
act as if entirely independent of and irresponsible for the being
or well-being of the sternum.
Paterson’s results are not so striking in large percentages as
Lickley’s, but sufficiently so to give us pause, even if considered
alone. He examined 236 human fetal sterna and found that the
seventh costal cartilages articulated in front of the sternum in
14.4 per cent.
In this connection three special cases may be cited, the first
two reported by Lickley and the last by Dwight:
1. The sixth costal cartilages articulated dorsally with the
lower end of the mesosternum and ventrally with one another.
The seventh cartilages articulated with the lower borders of the
sixth cartilages and by their extremities with one another.
2. Girl, eighteen years. Mesosternum terminated at level of
insertion of fifth ribs. The extremities of the sixth and seventh
cartilages on the left side were fused together, those on the right
were closely united by fibrous tissue. The two bars formed in
this way. articulated with the mesosternum above, and with one
another in front.
3. The body of the sternum ends at the level of the fourth

ribs. The fifth pair is attached to its lower end. The fifth,
sixth, and seventh pairs meet one another and fuse.
Kirchner (’98) and Adolphi (’05) also describe sterna in large
numbers and find many cases of variation in the posterior tho-
racic ribs much in accord with the above. The results of these
papers on the adult sterna of the human subject, together with
similar observations on the lower forms, which are pertinent to
our thesis are four:
1. Ribs posterior to the level of the sternum may grow ven-
trally and meet by fision or articulation in the midline
(Chamaeleo).
2. Ribs at the level of the sternum may articulate firmly with
it, as is the usual case in man.
3. Ribs may pass on the ventral side of the posterior end of
the sternum and fuse in the midline much as they do when farther
back where no sternum is present (Chamaeleo, man).
4. All ribs may fail to meet the sternum at any point: without
affecting the full development of that structure (Amphibia,
Chirotes, Manatus, human embryo).
Plate 12, figures 43 to 49, inclusive, is designed to give at a
glance a series of shoulder-girdles representing the main groups
of vertebrates, in an effort to show how from the elasmobranch
to the rodent, either in the embryo or throughout life, there is an
intimate relation in all between shoulder-girdle and sternum, and
in some it is demonstrated that they are for a time one struc-
turally continuous element. In some (shark, lissotriton, lizard,
and monotreme) this relation is maintained throughout life; in
others (marsupial, mouse, rat, and man) it is of short duration in
embryonic life. In all, however, it indicates .a phylogenetic
relationship between the shoulder-girdle and sternum in all stages
of vertebrate evolution that cannot possibly be duplicated or
even remotely approximated by a similar comparison of sterna
and costal cartilages.
In most adult Amniotes the sternum is in close relation with
both ribs and shoulder-girdle; in some, however, the connection
with either ribs or shoulder-girdle may be lost (Chirotes, fig. 30,
being an example of thefirst, and the pig of the second) ; but none
is known in which the relation to both ribs and shoulder-girdle
is lacking. Assuming that the sternum must be either a deriv-
ative of the shoulder-girdle or of the ribs, it is clearly evident
from a phylogenetic viewpoint that since the costae verae ex-
tending to the ventral side of the body were not acquired by the
vertebrates uutil the rise of the Reptilia, whereas the sternum and
shoulder-girdle, in an ever-increasing closeness of relation and
association, may be traced back to the very beginning of the
Ichthyopsida in the elasmobranchs, that we cannot hope to find
in the ribs any clue to sternal origin. If the sternum be homol-
ogous throughout, as the conclusions of this present investigation
seem to warrant, then its origin may be sought in a structure
which is coexistent with it and also in the closest possible rela-
tion to it in the lowest forms. In the shoulder-girdle of the Ich-
thyopsida we seem to meet with both of these requirements,
while in each of them the ribs fail us.
In Huntington’s (18) paper, which appeared after this work
was practically completed, are certain fundamental conceptions
of the shoulder-girdle and its phylogenetic relationships which in-
directly corroborate my conclusions. In the first place, Hunt-
ington recognizes the elasmobranch pectoral girdle as ‘‘the
primordial fundament upon which all other vertebrate modi-
fications are built.’ By a dual process of segmentation and re-
placement by bone, all structures of the complicated girdles of
the higher classes of vertebrates are derived from this simple
continuous unsegmented bar of cartilage found in the dogfish,
sharks or rays. Huntington was not, however, primarily inter-
ested in the sternum and did not see in the midventral portion
of the elasmobranch girdle the fundament of the presternum.
Like other investigators, he finds the ‘‘first appearance of the
sternal apparatus” in the amphibian girdle, but notes its intimate
association with the epicoracoids, which latter cartilages are begin-
ning to loosen by sutures on either side of the ventral midline.
However, it must be pointed out that these structural relations
of the Amphibia (clavicle excepted) are also present in Hexanchus,
where a suture on either side of the ventral midline gives the
‘intercepted cartilage’ interpreted by Howes (’91), Parker (’91),

and myself as the rudiment of the presternum. If two dotted
lines be added to the Hexanchus girdle (fig. 1), a sternum com-
parable to that of Rana is produced, with epicoracoids meeting
in the midline and the fundaments of the omosternum and ster-
num present.
9. Conclusions
This somewhat lengthy review and discussion of the phylogeny

of the sternum in the different classes of vertebrates, together
with the accompanying figures, which have been adapted and
modified from various sources, has revealed very clearly several
facts of outstanding importance in relation to the problem of
sternal origin:
1. That there is present a median ventral rudiment, derived
from the coracoids, which may be identified as a presternum as
far back in the vertebrate series as the shark, and can be followed
up through a ganoid, a teleost, a dipnoan, and from there on
through the Tetrapoda.
2. That in all cases of vertebrates, and as high as man in the
Mammalia, there is in the embryo or throughout life a contin-
uous girdle across the ventral side and connecting the two scap-
ulae above.
3. That this girdle in its ventral aspect is in the most intimate
relation to the anterior part of the sternum; sometimes the mes-
enchymatous material passes over insensibly from one structure
into the other without any line of demarkation; or at most,
adults of Amphibia and Reptilia, there being but a suture between
the two parts.
4. That in all these forms from the lowest to the highest, the
relation of sternum and ribs is purely secondary and the result of
a comparatively late fusion of the two structures in the embryo.
The presence or absence of ribs does not seem to affect the devel-
opment or size of the sternum in any degree.
5. That plural ribs extending to the ventral side of the body
are a recent acquisition of the vertebrates, while the shoulder-
girdle and sternum are coexistent and intimately related from
the earliest appearance of the Gnathosomes.
6. That the number of ribs reaching the sternum varies from zero
to a large number ; sometimes the ribs are attached to the anterior
part of the sternum, again exclusively to the posterior part, but
apparently whatever the number or relation of ribs, the sternum
remains unaffected, indicating strongly its independence of the
costal cartilage.
7. That the evidence presented seems to bear out the homology
of the sterna throughout the vertebrates; therefore, to classify
them as coracoidal for the lower groups, and costal for the higher
is unnecessary and artificial, for in the Amniota the sternum is as
truly coracoidal in origin as it is in the Ichthyopsida.
8. That the mesosternum and xiphisternum are two back-
ward prolongations of the coracoidal presternum, sometimes
uniting in the midline (some reptiles, birds, and mammals),
again: remaining distinct as horns or bars (some Amphibia, some
reptiles).
V. SUMMARY
At the close of the section on the ontogeny of the sternum a

number of conclusions were listed, and likewise at the close of the
section on phylogeny. These results may now be gathered up
in the three main theses of this paper, which are stated thus:
1. That the sternum is an homologous structure throughout all
groups of vertebrates, and occurs in forms ranging from Hexanchus
up to the highest mammals.
2. That the anterior element of the sternum has its origin in
common with the shoulder-girdle, and in the embryo or through-
out life is in intimate relation to the coracoids.
3. That the sternal bands are derivatives of the anterior median
rudiment, and may be secondarily, but never genetically, as-
sociated with ribs.
o
LITERATURE CITED
Apoupni, H. 1905 Uber die Variation des Brustknorbes und der Wirbelsiule
des Menschen. Morph. Jahrb., Bd. 33.
AntHony, R. 1901 Notes sur la Morphogenie du sternum chez Mammiferes a
propos de l’etude de Paterson sur le developpement de cet os. Bull.
Soc. Anthrop. Paris (5), T. 2, pp. 19-43.
Broom, R. 1897 On the existence of a sternocorocoidal articulation in a fetal
marsupial. Jour. of Anat. and Physiol., vol. 31.
1898 Description of the shoulder-girdle in an 8.5-mm. embryo of
trichosurus (not exact title). Proce. Linn. Soc. N. 8. W., 1898.
1899 On the development and morphology of the marsupial shoulder-
girdle. Trans. Roy. Soc. Edinb., vol. 39, pt. 3, pp. 749-770.
1902 On the early condition of the shoulder-girdle in the Polyproto-
dont Marsupials Dasyurus and Permales. Linn. Soc. Jour. Zool.,
vol. 28.
1908 On the nomenclature of the elements of the amphibian shoulder-
girdle. Report S. Afr. Assoc. Adv. Se., 6th. Meet., pp. 162-166.
Brucu, Cari 1852 Beitrige zur Entwicklungeschichte des Knochensystems
von Dr. Carl Bruch. Neue Denkschriften der allgemeinen Schweizer-
ischen Gesellschaft fiir die gesammten Naturwissenschaften. Bd. 12,
Zurich (quoted by Ruge).
CREDNER 1881-1893 Stegocephalen und Saurier. Zeitschr. deutsch. Geolog.
Gesellsch., 1881-1893.
CunnNINGHAM, D, J. 1890 The occasional eighth, true rib in man and its rela-
tion to right-handedness. Jour. Anat. and Physiol., vol. 24, p. 127.
Dwicut, THomas 1890 Irregular union of the first and second pieces of the
sternum in man and the apes. Jour. Anat. and Physiol., vol. 24, pp.
536-542.
Goittr, A. 1877 Morphologie des Skelettsystems der Wirbeltiere: Brustbein
und Schultergiirtel. Arch. f. Mikr. Anat., Bd. 14.
GoopricH 1909 Vertebrate Craniata (first fasicle): Cyclostomes and fishes.
A treatise on zoology, pt. 9, by E. Ray Lankester. London.
Grecory, W. K. 1915 Present status of the problem of the origin of the Tet-
rapoda. Ann. N. Y. Acad. Sce., vol. 26.
GuntrHER 1867 Anatomy of Hatteria (Sphenodon). Phil. Trans., 1867.
Hasweu, W. A. 1884 Studies on the elasmobranch skeleton. Proc. Linn.
Soc. N. S. W., vol. 9.
Horrmann, C. K. 1879 Zur Morphologie des Schultergiirtels und des Brust-
beins bei Reptilien, Végeln, Siugetieren, und dem Menschen. Nieder-
land. Archiv. f. Zool., Bd. 5.
Howes, G. B. 1891 Morphology of the sternum. Nature, vol. 48.
Howes AND SwInNERTON 1901 Development of the skeleton of Sphenodon.
Trans. Zool. Soc. Lon., vol. 16.
Hountineton, G. 8. 1918 Modern problems of evolution, variation, and in-
heritance in the anatomical part of the medical curriculum. Anat.
Rec., vol. 14, no. 6.
Kerset and Mati 1910-1912 A manual of human embryology. J. B. Lippin-
cott Co., Phila.
Kinastey, J. S. 1917 Comparative anatomy of vertebrates. 2nd ed. P.

Blakiston’s Sons & Co., Phila.
Kircuner, A. 1898 Das obere Brustbeinende und das ligamentum interclavi-
culare nebst zusammenstellungen itiber das Verhiltnis des oberen
sagittalen Brustdurchmessers und der Brustbeinlinge zur K6rper-
linge. Anat. Hefte., Bd. 10, S. 127-149.
Kraverz, L. P. 1905 Entwicklungsgeschichte des Sternum und des Epister-
nalapparatus der Siugetiere. Bull. Soc. Imper. Natur. Moscow,
1905, 1906.
Licxtey, J. D. 1904 On the relations of the seventh and eighth ribs to the
sternum. Anat. Anz., Bd. 24.
Marxowskl, J. 1905 Sollte der Verknécherungsprozess des Brustbeins von
keiner morphologischen Bedeutung sein? Aus anlass einer Publika-
tion von Paterson. Anat. Anz., Bd. 26, 8. 248-269.
Mvruuer, CHartotre 1906 Zur Entwicklung des menschlichen Brustkorbes.
Morph. Jahrb., Bd. 35.
Parker, W. J. 1868 Structure and development of the shoulder-girdle and
sternum in the vertebrates. Ray Soc. Pub., London.
ParRKER, T. J. 1891 On the presence of a sternum in Notidanus indicus.
Nature, vol. 43, p. 142.
1891 On the origin of the sternum. Trans. New Zealand Instit.,
vol. 23.
Parerson, A. M. 1900 The sternum; its early development and ossification
in man and mammals. Jour. Anat. and Physiol., vol. 35, N. 8., vol.
15, pt. 1.
1902 Development of the sternum and shoulder-girdle in mammals.
Brit. Med. Jour., vol. 2.
1904 The human sternum. Liverpool.
1909 Two abnormal sterna in living subjects. Jour. Anat. and
Physiol., vol. 43, pp. 322-323.
Ratuke, H. 1838 Sur le devellopment dusternum. Arch. Anat. et Physiol.
1848 Uber die Entwicklung der Schildkréten. Brunswick.
1853 Uber den Bau und die Entwicklung des Brustbeins der Saurier.
K6nigsberg.
Ruae, G. 1879 Uber die Entwickelung des Sternum. Morph. Jahrb., Bd. 5.
1880 Untersuchungen iiber Entwickelungsvorgiinge am Brustbein
und der Sterno-clavicularverbindung des Menschen. Morph. Jahrb.,
Bd. 6. °
Sapatinr, A. 1897 Morphologie du sternum et des clavicules. C. R. Acad.
Se. Paris, T. 124, no. 15.
1902 Du systeme sternal des vertebres. C. R. Ass. Anat., 4me, pp.
99-102.
ScHAUINSLAND 1900 Entwicklungsgeschichte der Hatteria (Sphenodon). Arch.
f. Mikr. Anat., Bd. 56.
Seetey, H. G. 1892 Researches on the structure, organization and classifi-
cation of the fossil Reptilia. Phil. Trans., vol. 183-B, pp. 334-338.
1894 Researches on the structure, organization, and classification of
the fossil Reptilia. Phil. Trans., vol. 185-B. :
TrepGoLp 1897 Variations of ribs in the Primates, with especial reference to

the number of ribs in man. Jour. Anat. and Physiol., vol. 31.
Watson, D. M. S. 1917 The evolution of the tetrapod shoulder-girdle and
forelimb. Jour. of Anat., vol. 52, pt. 1.
WHITEHEAD AND WaApDpELL 1911 Development of the human sternum. Am.
Jour. Anat., vol. 12, pp. 89-106.
Witper, H. H. 1903 The skeletal system of Necturus maculatus Rafinesque.
Mem. Bost. Soc. Nat. Hist., vol. 5, no. 9.
Wiuuiston, S. W. 1903 North American Plesiosaurs. Field Col. Mus. Pub.
fan (Geol. Ser., vol. 2;no. 1.
1903 On the osteology of Nyctasaurus (Nyctodactylus). Field Col.
Mus. Pub. 78. Geol. Ser., vo.. 2, no. 3.
1911 American Permian vertebrates. Chicago.
1914 Water reptiles of the past and present. Chicago.
Woopwarp 1898 Vertebrate paleontology. Cambridge.
PLATE 1
1 Median portion of pectoral girdle of Hexanchus. Note the medial ‘inter-
cepted cartilage’ which bears all the relations of a presternum in higher forms.
The dotted lines indicate how this girdle may be transformed into that of the
Amphibia with pre- and post-omosternum. Drawing from a dissection made
by the author in the U. 8. National Museum.
2 Pectoral girdle of Acanthias vulgaris. Note that all the parts of the
girdle of higher forms are present, including a sternum. Compare with figure
35 of the fetal marsupial.
Cr, coracoid Sc, scapula
POSt, pre-omosternum r SSc, suprascapula
PSt, presternum St, sternum
PiOSt, post-omosternum
92
ONTOGENY AND PHYLOGENY OF THE STERNUM PLATE 1
FRANK BLAIR HANSON
Cr Los
A
iQ
be, - <t
it
PLATE 2
3 Cat embryo, 12mm. Sternal bands are far apart. No presternum pres-
ent at this stage. The first three pairs of ribs fail to reach the sternal bands.
From the Princeton Embryological Collection, series no. 401. Graphic recon-
struction.
4 Cat embryo, 14 mm. Sternal bands nearer to each other. Presternum
has arisen and connects the anterior ends of sternal bands. All ribs’reach the
sternum. From the Princeton Embryological Collection, no. 37. Graphic
reconstruction.
PSt, presternum R,* fourth rib
PR}, first rib R,’ seventh rib
R?, third rib StB, sternal bands
94
FRANK BLAIR HANSON :
95

PLATE 3
5 Mouse girdle. Scapulae, coracoids, clavicles, and presternum, all in one

continuous mesenchymatous girdle, which is similar to that of the dogfish (fig. 1)
and the marsupial fetus (fig. 35). From the Washington University School of
Medicine Embryological Collection, series no. 102, slide 4, section 16.
6 Human embryo girdle. Mesenchymatous stage. First appearance of
clavicle; presternum and coracoids present. From The Johns Hopkins Em-
bryological Collection, series no. 424, slide 5, section 16.
Cl, clavicle PSt, presternum
Cr, coracoid Sc, scapula
96
FRANK BLAIR HANSON
a
or
~
Se,
68
©,
97
PLATE 4
7 Sternum and ends of costal cartilages in pig two weeks old. Note in
this and later figures that centers of ossification are intercostal.
8 Fetal sternum of Bradypus. After Hoffmann.
9 The ape sternum. After Anthony.
10 Sternum of Dasypus. After Hoffman. From the lowest to the highest
mammals the sternebrae and centers of ossification are between the ribs, and
not at their ends as would be expected according to Ruge’s theory.
Oc, ossifie center
98
FRANK BLAIR HANSON
PLATE 5
11 Ventral portion of shoulder-girdle of the shark, Notidanus indicus, show-

ing the presternum fn the midline. After Parker.
12 Ventral part of the shoulder-girdle of Polypterus. Note epicoracoid at
junction of the clavicles. From Gregory, after Goodrich.
13 Claviecles and epicoracoid of the tropical fish, Gobius niger Linn., after
Parker. q
14 to 15 Upper and lower views of the girdle of the Dipnoid, lepidosiren. ~
After Parker.
16 to 19 A series of stages in the development of the sternum of the frog,
Rana temporaria, indicating that the sternum is coracoidal in origin. After
Parker.
Cl, clavicle PtOmSt, post-omosternum
Cr, coracoid OSt, omosternum
ECr, epicoracoid Sc, seapula
PrOmSt, pre-omosternum St, sternum
100
FRANK BLAIR HANSON
Pea O inst
101
PLATE 6
20 Shield-like plate on ventral side of Pipa dorsigera. Adult female, upper
view. Would seem to indicate that sternum and coracoids were at one time
structurally one; omosternum but feebly separated from coracoidal portion.
After Parker.
21 Bufo vulgaris. First summer. Lower view. After Parker.
22 Large specimen of Siredon pisciformis. Sternum cut off from coracoids
in adult, but retain evidences of having come from the overlapping epicoracoids.
After Parker.
23 Dactylethra capensis. Adult female. After Parker.
24 Note the beginning of sternal bands in Calamites cyaneus. Compare
with same stage in low-type reptile. After Parker.
Cl, clavicle OSt, omosternum
Cr, coracoid PCr, precoracoid
Crf, coracoid fossa Sc, scapula
ECr, epicoracoid SSc, suprascapula
Gl, glenoid St, sternum
FRANK BLAIR HANSON
Ost
103
PLATE 7
25 An old specimen of Anguis fragilis. Very similar to amphibian girdle.
No ribs attached. Same intimate relation of sternum and coracoidal part as
in the Amphibia. Modified after Parker.
26 In Stellio cordylinus long sternal bars make their appearance. No ribs
are attached to them. Adapted from Parker.
27 Laemanctus longipes. After Parker.
28 Trachydosaurus rigosus adult. Sternal bars are fused and ribs are
approaching. Adapted from Parker.
ECr, epicoracoid R', first rib
Gl, glenoid Sc, scapula
TCl, interclavicle St, sternum
PCr, precoracoid Stf, sternal fossa
PSt, presternum XSt, xiphisternum
104
FRANK BLAIR HANSON
ihe
A
ay) --XSt---
; t
f hod
Vy
105
PLATE 8
29 Cyclodus nigroluteus. Adult, lower view. After Parker.
30 Chirotes canaliculates. Adult male. Upper view. Mammalian-like
sternum. No ribs ever reach sternum. After Parker.
31 Crocodilus acutus. Ripe embryo. Lower view. After Parker.
32 Chamaeleo vulgaris. Adult, lower view. After Parker.
Cr, coracoid R', first rib
ECr, epicoracoid R?, second rib
ICl, interclavicle ' Stf, sternal fossa
MSt, mesosternum XSt, xiphisternum
PSt, presternum
106
FRANK BLAIR HANSON
IC!
IC!
PLATE 9
33 Vanellus custatus. One-third of incubation period. Lower view. Note
slight attachment of ribs, but only sutural separation of sternum and coracoid.
After Parker.
34 Echnidna histrix. Upper view of adult specimen. Drawn from a speci-
men in Washington University, Department of Zoology, and in part after Parker.
35 Shoulder-girdle of a marsupial embryo, Trichosurus. Scapulae, cora-
coids, and sternum are continuous parts as in theshark embryo. After Broom.
36 Anterior view of girdle in an 8.5-mm. Trichosurus embryo. Dotted
portions are mesenchymatous. After Broom.
Ac, acromian Pro, pre-omosternum
Cl, clavicle R}, first rib
Cr, coracoid R?, second rib
ECr, epicoracoid Sc, scapula
Gl, glenoid Sp, spine
ICl, interclavicle SSc, suprascapula
OSt, omosternum St, sternum
PCr, precoracoid XSt, xiphisternum
108
FRANK BLAIR HANSON
109
PLATE 10
37 Petrogale xanthopus, 3 inches long. Right scapula, outer view; sternum,

inner view. After Parker.
38 Mus musculus, adult, inner view. Anterior end of sternum and medial
end of clavicles, showing omosterna, and epicoracoids. After Parker.
39 Shoulder-girdle seen in a reconstruction of Dasyurus viverrinus. Front
view of mammary fetus.
40 Sternum of adult Manatus americanus. Left half, inner view. After
Parker.
41 Sternum of the embryo of the Dolphin. Inner view. After Parker.
Ac, acromian ' PST, presternum
Cl, clavicle R!, first rib
Cr, coracoid R?, second rib
ECr, epicoracoid R3, third rib
F, fontanelle R$, eighth rib
MSt, mesosternum SSc, suprascapula
OSt, omosternum XSt, xiphisternum
110
FRANK BLAIR HANSON
Past

Td “LV Il
NOILYNV1Id
JO xXa SHYNSId
zp uostiedwoyg
Jo oY} [vyUsUEpuNy
odyJo 9y81qg9310A PIIS-Iap[Noys
9] 8B
poyeaysnqyt
Ur oy vaNUYYIM yeyYJo “UvUr poyipoyy
10438 ‘UOYSUIFUNFY
‘OY UBIUIOIOR ‘ssc— vsso vI[vulojseidns
‘99 PIOIV10I0}S09JUOUIVSI]T 4 WMUIE}SOUIO
112
99 spIABp ‘ug ploovio9erd
‘49 prloovi09 Tay 4ysagqt
“ugq plooviooide ‘ay eindvos
‘da stsAydide
Jo Q[OIABTO ‘ogy eindvoseidns
‘277 snisumny oy umuse4ys
‘97 IB[NITAV[OIE}UI
JUOUIVSI]
PLATE
11
HANSON
BLAIR
FRANK
ONTOGENY
OF
PHYLOGENY
AND
STERNUM
THE
PLATE 12
43 Girdle of Hexanchus. Pre- and post-omosterna. After Parker.

44 Diagrammatic transverse section through shoulder-girdle of adult frog.
45 Section through edges of shoulder-girdle and sternum of adult Lissotriton
punctatus. After Parker.
46 Diagrammatic transverse section of arch in the Australian lizard, Trachy-
dosaurus rugosus.
47 Diagrammatic transverse section of Echnidna histrix, from a mounted
specimen in Washington University.
48 Anterior view of cartilaginous girdle of petrogale, 21 mm. in length.
The different parts of the girdle are outlined, but the whole is a continuous piece
of cartilage. After Broom.
49 Transverse section of mouse embryo, 7.75 mm. long. Washington Uni-
versity School of Medicine series no. 102, slide 4, section 16. X 38.
Ac, acromian PrOmSt, pre-omosternum
Cr, coracoid PtOmSt, post-omosternum
ECr, epicoracoid Sc, scapula
Gl, glenoid SSc, suprascapula
Hu, humerus St, sternum
114
a
FRANK BLAIR HANSON
PtOmst wo OSC
| \
fi --=--8¢
115
Resumen por el autor, George W. Corner
Universidad de California.
Sobre el origen del cuerpo amarillo de la cerda a expensas de la

eranulosa y teca interna.
E] ovario de la cerda presenta algunas ventajas para la solucién

de este problema. El presente trabajo esta basado en la inspec-
cién de una larga serie en la cual se determiné el estado del ciclo
reproductor por la observacién de los animales vivos y sus 6vulos.
Los resultados obtenidos pueden resurmirse del siguiente modo:
En la cerda la membrana granulosa persiste intacta después de la
ruptura del foliculo de Graaf. Sus células aumentan de tamafio
sin dividirse; su citoplasma se carga de substancias lipoides y
finalmente se transforman en los grandes elementos del cuerpo
amarillo completamente formado, llamados comunmente ‘‘célu-
las luteinicas.”’ Los capilares sanguineos procedentes de la teca
interna invaden la membrana granulosa ramificandose para for-
mar un extenso plexo vascular en la nueva estructura. Las
grandes células cargadas de lipoides, presentes en la teca interna,
aumentan en nimero a consecuencia de divisiones mit6ésicas per-
diendo muchas o la mayor parte de sus inclusiones grasas y pa-
sando al cuerpo amarillo, alojandose entre las células de la gran-
ulosa, en toda la extensién de esta tiltima. No hay pruebas de
la transformacién de la células de la teca interna en fibroblastos de
tipo fusiforme corriente ni tampoco de su participacién en la for-
macion de las fibrillas del reticulo de mallas angostas que existe
en el cuerpo amarillo. El autor ha encontrado pruebas sobre la
persistencia de algunas de las células de la teca interna durante la
prefiez, las cuales forman elementos bien patentes en el cuerpo
amarillo; pero su destino no puede reconocerse por los métodos
actuales, a causa de la semejanza entre algunos de los derivados
de la teca y la granulosa.
BY THE BIBLIOGRAPHIC SERVICE, AUGUST 11,
ON THE ORIGIN OF THE CORPUS LUTEUM OF THE

SOW FROM BOTH GRANULOSA AND
THECA INTERNA
GEORGE W. CORNER
From the Anatomical Laboratory, University of California
CONTENTS
Rept OGHERG PLOKED 5,crete st stSMA ats miaiace aN eaten, oorLRA Oe ALR NE Aa eae obnle ek 117
Penn ala ES WISAGLONE f3 eit La hss eras dcos\s SS EG ee ree shaw feed 122
Previous work.on the corpus luteum) of the sOw..:.s0.2...02o0+<--.+0-54-s 127
pee TUM VIC MIVGLINOOS 5 Meyer Lie a ict) «cs «+ seaueiahare ciseine Rea ookGhetend obotece ke ean eae 6 131
Special cytology of the lutein cells of the sow..................0. 000000000 137
CRIME C TOMICle reg. Hye Rees Uist os 4 OLR poe iets. ots, Jaret cautiye .& 140
Miteartesty CU PUmeChOlliG] Gx eo. .: 5 alas apn che eoa sibSpyatebetoine apeiclwei in a anh eelelr 148
re reeat a LAT TCLS T ATUL OSD recta oi 5 a wc, «5 sed Sal ak REDAB AR oagh OTe waa? oudka a ete.2 157
The fully formed corpus luteum, until the termination of pregnancy....... 168
etrafression of the: corpus lubeam: 20s Ba a ee eel 177
VST SST Ta Eg TE 9S Ss ST a ee Roc LORY 2 ea a ee ee 178
MAINE TRY CERES) cal e Mee mE 2 sgt, hah ae epEAR Berrie Ts CCT acieglkti acura cite 180
INTRODUCTION
The history of the discussion, now of more than seventy years’

standing, as to the origin of the corpus luteum, has been repeated
so many times that it has become traditional, and the names of
von Baer and Bischoff have been passed down to us as the original
proponents of the two chief doctrines in question. It is said
that the former, in his monograph ‘‘De ovi mammalium genesi’”
(27) first stated that the corpus luteum is derived from the theca
interna of the Graafian follicle, and that Bischoff first discarded
this view in favor of the membrana granulosa as the site of origin.
I have not been able to see von Baer’s work, but judging at least
from Bischoff’s account of the early embryology of the rabbit
(’42), there was no such clear-cut opposition of view as tradition
declares, for Bischoff considered himself, rather, as an upholder
of von Baer (and was so quoted by contemporary investigators).
117
118 GEORGE W. CORNER
It must be remembered that the first of these monographs ap-

peared a decade before Schleiden and Schwann’s enunciation of
the cell theory, and the other not five years after; histology was
studied with pincettes and the needle rather than by sections,
and the first nuclear stain was not discovered. The layers of
the follicle were as yet imperfectly differentiated, and the early
descriptions are so vague that it is difficult to interpret them in
present-day terms. New steps toward the solution of this prob-
lem have always followed fast upon the development of histo-
logical technique, and thus it is in the writings of Wilhelm His
(65) and Waldeyer (’70) that we first find opinions and descrip-
tions approaching those of recent years.
The studies of His led to the complete formulation of the view
that the corpus luteum is derived from the theca interna of the
Graafian follicle, which in the next two decades was supported
by a number of investigators and still holds a place in the field
against strong opposition. The chief arguments in favor of this
view are that, first, the membrana granulosa of large follicles is
often degenerated, and is believed to be cast off at the time of
rupture; second, as the Graafian follicle ripens, the cells of the
theca interna show marked changes—they swell in volume, be-
come rounded, in some species they acquire granules of a yellowish
pigment, and in short come to present a striking resemblance to
the large cells of the corpus luteum; third, this resemblance is en-
hanced by the fact that not only are the large cells of the theca
interna folliculi and the corpus luteum similar, but the presence
of many blood-capillaries and connective-tissue cells causes a re-
semblance as well in the general structure of the two tissues;
and, fourth, such follicles as do not rupture lose their granulosa
by degeneration, become obliterated by proliferation of the theca
interna, and in this process of atresia attain also a resemblance to
the corpus luteum.
None of the contributions disagreeing with this view in favor
of the granulosa origin of the lutein cells were at all convincing,
until the appearance in 1895 and 1896 of Sobotta’s first researches,
which mark the beginning of modern work upon the question.
Here again the chief contribution was‘one of method. Sobotta
ORIGIN OF THE CORPUS LUTEUM 119
pointed out that the arguments quoted above are based merely
upon analogies between the layers of the follicle and the fully
formed corpus luteum, and that from the writings of his prede-
cessors it is apparent that few had actually seen corpora lutea in
process of formation. Even when descriptions are given of
mature follicles or supposed early corpora lutea, there is usually
no proof that the structures in question actually represent the
results of normal follicular development or recent ovulation. The
problem should be worked out from a series of specimens gathered
at known periods after rupture of the follicle; and in order to
avoid confusion with atresia or other irrelevant processes, each
follicle or corpus luteum studied should be certified as to its
normal condition and stage of development by comparison with
the fertilized ovum or embryos proceeding therefrom. To fulfill
these high requirements calls for long and tedious labors—the
investigator must spend hours and days in observation of his
animals; the reproductive cycle of the species used must be known
well enough to acquaint him with the time of ovulation, the ani-
mals must be killed at definite times thereafter, and the ova must
then be sought in the ovary, the oviducts, or the uterus. Sobotta
himself chose the mouse, in which he had found that an ovulation
takes place about twenty-one days after the birth of a litter, and
in which the small size of the animal permits serial sectioning
of the entire ovaries and Fallopian tubes. It must be admitted
that his own postulates could not be followed to the full; the in-
dividual corpus luteum corresponding to a given ovum cannot be
identified, because many ova are extruded at one ovulation in
this species; the exact time of ovulation may vary by hours, and
again there is so much variation of the interval between ovula-
tion and the entrance of the spermatozo6én into the egg that the
condition of the ovum cannot be used as an exact measure of the
age of the corpus luteum. Study of the ova merely provides
assurance that the corpora lutea are normal and gives a rough
means of determining their ages. The judgment of the investi-
gator must finally be used to rank the corpora lutea in an orderly
series. It is beyond denial, however, that Sobotta possessed
such a series, collected from about 200 mice and based upon the
study of nearly 1500 ova during the stages of maturation, fertili-

zation, and the segmentation of the blastomeres. He found no
degeneration of the membrana granulosa; instead the cells of
this layer remain and undergo hypertrophy (without division),
finally becoming the characteristic large cells of the corpus luteum.
Meanwhile the cells of the theca interna undergo mitotic division,
are converted into spindle-cells and invade the granulosa to
form the connective-tissue reticulum of the corpus. In this
process all the theca interna cells are used up, and the layer
therefore disappears. Capillary blood-vessels grow in from the
vessels of the theca interna, ultimately providing the corpus
luteum with a rich circulation.
There will be no need to enter here upon a detailed account of
the debate which began immediately upon the publication of
these epoch-making studies. A full analysis of the literature
upon the origin of the corpus luteum up to 1901 will be found in
the papers just quoted and in the two reviews of Sobotta in the
Ergebnisse der Anatomie (’99a, 702). In 1897 Sobotta himself
studied corpus luteum formation in the rabbit, and found the
process in all important points exactly as in the mouse. A year
later, Stratz (’98) printed a research upon which he had been
engaged before the publication of Sobotta’s work, in which he
had followed the early stages of the corpus luteum on ovaries of
an insectivore, Tupaia javanica, the lemuroid ape Tarsius spec-
trum (the specimens being those of Hubrecht’s well-known Java-
nese collections,) and of the ‘Spitzmaus’ or shrew, Sorex vulgaris.
Although he did not have large numbers of cases, all were checked °
by the study of the ova or embryos. As far as the persistence
of the granulosa cells and their direct conversion into the lutein
cells was concerned, Stratz agreed fully with Sobotta, but with
regard to the fate of the theca interna there is a minor difference.
If I understand Stratz correctly, he considers the theca interna
of the mature follicle merely as a zone of blood-vessels, in which
all the cells are either constituents of the vascular wall or of the
adventitia. Considered in this light, it is easy to see how this
angioma-like thecal tissue would enter into the growing corpus
luteum to form its blood-vessels and its connective tissue without
the necessity of transformation or regression of the specialized

theca cells into fibroblasts, as described by Sobotta. So far as is
known to me, no subsequent investigator has confirmed the view
of Stratz, all others being agreed that the theca interna consists
of a distinct layer of highly specialized cells derived from the
mesenchymatous elements of the ovarian stroma, and containing
a network of blood-vessels, supported by cells and fibrils of con-
nective tissue.
A third theory as to the fate of the theca interna is proposed
in the important papers of O. Van der Stricht, of which the first
appeared in 1901. The study was carried out upon the ovaries
of large numbers of European bats, chiefly Vesperugo noctula.
As with the two previously cited investigations, the animals were
collected primarily for the study of the ova and the early embryos
of the species used, and the series is therefore accurately con-
trolled by the condition of the ova. Wan der Stricht demon-
strates beyond doubt that in these species the granulosa layer
persists in situ after rupture of the follicle, and that its cells en-
large, acquire granules of lipoids staining black with osmium
tetroxide, and finally become the typical lutein cells. Contrary
to Sobotta, he thinks that mitotic division may occasionally occur
in these cells, so that the filling of the follicular cavity is brought
about by a slight increase in their number as well as by the vast
increase in their individual bulk. The connective tissue of the
corpus luteum arises chiefly as Sobotta described it in the mouse.
After rupture of the follicle the membrana propria disappears,
and fibroblasts invade the metamorphosing granulosa layer. Van
der Stricht thinks that these are the spindle-cells of the theca
interna or their descendants. Of the distinctive cells of the
theca, some seem to disappear, but others remain, chiefly about
the periphery of the new corpus luteum, or enter a short distance
into the granulosa, and here they remain almost in their original
condition. After a few days, however, when the deposition of
fatty droplets in the granulosa cells has progressed, the two types
of cells so closely resemble each other that Van der Stricht could
no longer distinguish them in his Flemming-fixed tissue. He
believes, in fact, that they have become identical, and therefore
$23 GEORGE W. CORNER
that in the fully formed corpus luteum most of the lutein cells
are of granulosa origin; a few of them, however, are from the
theca interna.
A nearly identical theory is that proposed by Rabl (’98), who
studied human corpora lutea (the youngest estimated at ten
days), and found in them about the periphery a layer of cells
differing from the rest of the lutein tissue; this, he suggested,
might be the theca interna, which he supposed to persist in its
original position until its cells were lost to view, some by becoming
converted into lutein cells, others by degenerating.
During these five years from 1896 to 1901 there was no lack
of publications restating the total loss of the granulosa before
rupture, in opposition to the descriptions of Stratz, Sobotta, and
Van der Stricht. A few of these investigations were carried out
upon the ovaries of swine, and will therefore be discussed more
fully later in this paper. As Sobotta pointed out in his résumé
of 1902, not one writer among those who taught the non-partici-
pation of the granulosa in corpus luteum formation had been
able to prove that his specimens were normal mature follicles
and corpora lutea by presenting the ova which had proceeded
therefrom.
RECENT INVESTIGATIONS
From 1901 to 1917 there have been about thirty-five more

publications upon the question, of which some twenty-five rep-
resent actual original investigations. As the subject has not been
brought up to date in any publication in English, it may be as
well to take up in some detail the work of the past sixteen years,
especially as the old differences of opinion still persist.
Jankowski (’04) reports studies upon a series of ovaries of sows
and guinea-pigs, collected without an attempt to learn the repro-
ductive cycle of the animals or to test the normal conditions of
the specimens according to the postulates of Sobotta (which he
says he had found impracticable to apply and whose value he
questions). He believes the granulosa to be intact until the
rupture of the follicle and even afterward, but to degenerate
before the ingrowth of the theca interna, to which he ascribes the
_—
e
origin of the lutein tissue. The value of his results with the
pig will be discussed below; Sobotta has presented directly op-
posite evidence and a vigorous criticism with regard to his work
on the guinea-pig (’07).
The contributions of Pottet (10) on the human corpus luteum
and Delestre (’10) on that of the cow are based on evidence which
can hardly be considered conclusive. Delestre had no bovine cor-
pora lutea of pregnancy at an earlier stage than two and a half
months. He had twelve corpora lutea from non-pregnant animals,
four of which he thought to be in the first stages of formation,
but there was no effort, by observing the animals alive or by
searching for the ova, to determine that ovulation had actually
been recent. Pottet studied twenty-two human corpora lutea
of pregnancy, the youngest already six weeks old. Both of these
authors speak for the degeneration of the granulosa before
rupture.
We cannot judge the work upon the human ovary very criti-
cally until the relation of ovulation to menstruation is better
known or some other method of estimating the age of young
human corpora lutea and of obtaining really young specimens is
at hand. Biihler (’00) collected ovaries of rabbits according to
Sobotta’s methods, but found the distinction between theca and
granulosa so difficult that he turned to the human corpus luteum.
The only specimen of importance described by him is one from
an operative case, without menstrual history or other means of
estimating its age, except that it showed a point of rupture in
process of healing (see below, p. 179, as to the possibility of error on
this point). In this supposedly early corpus luteum the granulosa
is degenerating and a ‘typical lutein tissue’ appearing in the place
of the theca interna. Cristalli (03), a pupil of Paladino, whose
peculiar views will be quoted below, believes also in the total
degeneration of the granulosa layer before rupture, but gives no
data as to his specimens. ‘Teacher, in discussing the Teacher-
Bryce-Kerr case of early ovarian pregnancy (’08), states that he
had been studying corpus luteum formation in the human, and
interprets his preparations to indicate quite clearly that “‘what-
ever the source of the cells (of the corpus luteum) may be in the
lower animals, they do not in man arise from the membrana

granulosa,’ which latter membrane he thinks is probably shed
with the ovum. Hegar, in 1910, reported an examination of
six human ovaries removed four and two days before the onset
of menstruation, all containing corpora lutea which he supposes
to be fairly young. While admitting the epithelial origin of the
structure in mammals lower than man, he is inclined to view his
preparations as indicating a thecal origin of the lutein cells in
the human species. J. Whitridge Williams, whose text-book of
obstetrics (’03—’17) is based to so great an extent upon original
study that it is regularly quoted in scientific literature as author-
itative, retains in his last edition the views of the preceding
writers, of whose correctness he feels convinced by the study of
several hundred corpora lutea.
This completes the list of recent authors who see in the corpus
luteum a structure of connective-tissue origin alone. All other
investigators of the past sixteen years uphold in general the epi-
thelial origin of the lutein cells, but among themselves they vary
according to their views as to the fate of the theca interna. The
rabbit has been studied in 1897 by Sobotta, who found the process
exactly as in the mouse, but Honoré, three years later, in the same
animal, found that not all the theca interna cells are converted
into fibroblasts, but that some of them linger about the periph-
ery of even the fully formed corpus luteum. Cohn, in 1908,
repeated the work, apparently without study of the ova, but
dating his specimens from an observed copulation, (in the rabbit
ovulation occurs only after coitus), and confirmed the results of
Sobotta. Marshall, in the next year, described the corpus
luteum of the sheep, dating his specimens from observed copula-
tion. He did not seek the ova, but as in this species ovulation
and coitus can occur at no time except during a short oestral
period, so that coitus dates the time of ovulation within a very
few hours, the presumption is great that Marshall possessed
normal corpora lutea of ages accurately known. He found the
granulosa to persist and to be vascularized by sprouts from the
theca interna, all the cells of which were finally used up, having
been converted into spindle-cells of connective tissue. <A part of
a
the connective tissue of the corpus is contributed also by the

theca externa, which is drawn inward in places by the folding of
the follicular walls. O’Donoghue (’12, 714, ’16) has given a con-
firmation of Sobotta’s views for the marsupials (which I believe
were first studied by Sandes (’03), whose paper was not acces-
sible to me). Strakosch (’15) is the last to repeat the Sobotta
theory in its original purity, basing his statements upon the
human ovaries which were used by Robert Schroeder in his
study of the time relation between ovulation and menstrua-
tion (14).
Van der Stricht’s belief that the theca interna cells are not con-
verted into fibroblasts, but remain in the corpus luteum, no
longer distinguishable from other lutein cells, found support in
the study of L. Loeb (06) upon the guinea-pig. The specimens
were collected in a series dated from copulation without study of
the ova. The theca interna cells, after a few hours, could no
longer be distinguished from the granulosa lutein cells. This
work is open to the criticism that haematoxylin and eosin (the
only staining combination used) do not accentuate differences
between cells of the types met with in this problem. In 1908
Van der Stricht himself repeated his ideas as the result of re-
searches upon the ovaries of dogs, carefully checked up by exam-
ination of the ova, and in 1912 he repeated his findings in the bat.
A somewhat different view has found exposition in the very
careful work of Volker’ (05) upon Spermophilus citellus, a Euro-
pean marmot allied to the gophers of the western United States.
The ova and embryos were recovered and examined in all cases,
and there appears to have been a sufficient number of stages,
though the author does not state the number of specimens studied.
The theca interna cells were found to persist unchanged between
the granulosa and the theca externa, even until the end of preg-
nancy. The few spindle-cells found in the fully formed corpus
luteum are said to proceed from the theca externa. Practically
the same view is presented in the thesis of Niskoubina (’09), who
worked on the rabbit, but does not give an account of the methods
used. Cohn (’09) studied the human ovary, but appears to
- have seen no really young stages. His descriptions agree with
those of Rabl, and he thinks the layer of theca cells is not des-
tined to persist, but is a ‘matrix’ or source of origin for the newly
forming connective tissue of the corpus luteum.
With this group should be placed one of the most ambitious
of the recent attempts to work out the origin of the human corpus
luteum, that of R. Meyer (11a). The paper describes five
corpora lutea in process of formation, of which one is claimed by
the author to be the youngest ever obtained in the human. The
appearance of the structures and the menstrual histories were
the only guides to their age. The specimens show first a prolif-
erative stage, during which the granulosa cells swell and acquire
granules of a fatty substance, and, second, a stage of ‘glandular
metamorphosis’ through vascularization of the granulosa layer.
The first spindle-cells seen in the lutein layer arise from the blood-
vessels, which are sprouting inward. The wall is thrown into
folds, in which the larger fat-infiltrated theca interna cells are
crowded. Here they remain until the pressure of the swelling
lutein tissue crushes them out of existence, an event which may
be early or late according to the internal conditions of pressure.
Groups of them, serving as sources of nutrition for the growing
organ, may be seen about the periphery of the corpus luteum
and in the folds of its wall, until fairly late in the life of the corpus
luteum. The name theca-lutein cells has been given them.
As an example of the difficulty of proving anything about the
origin of the corpus luteum by specimens whose age can only be
guessed, it may be mentioned that the genuineness of Meyer’s
first and supposedly youngest corpus luteum has been sharply
attacked. Ricker and Dahlmann (’12) have hinted that it is not
even a naturally ruptured follicle, and J. W. Miller (’11) believed
it was an atretic follicle, because Meyer had stated the granulosa
cells to contain ‘Fett,’ while, according to Miller, neutral fat is
never found in the normal fresh corpus luteum. It must be ad-
mitted that this criticism was rescinded when Meyer (’11 b) stated
that ‘Fett’? meant merely lipoids in general, and that Miller is
himself no opponent of Meyer’s views. But after all we shall
never be certain of the early human corpus luteum until skill and
good fortune enable someone to obtain the tubal ovum with the
ovary.
ORIGIN OF THE CORPUS LUTEUM WATS
Four recent writers upon the human ovary have repeated the
same views as Meyer with but slight modifications. Elizabeth
Wolz (12), from the study of a few specimens, believes that none
of the theca interna cells suffer change into connective tissue.
Some degenerate by atrophy, others remain in situ a long time.
Timofeiev (713)! and Wallart (14) appear to have given as ac-
curate and modern a description as is possible in the face of the
particular diffculties of the human material. Careful menstrual
histories are given, and both used varied and interesting histo-
logical methods. According to both, the theca interna cells
remain in groups about the peripheryof the corpus luteum for a
long time, as described by Meyer, and they slowly atrophy.
None of them are converted into spindle-shaped connective-tissue
cells. Timofeiey describes also the deposition of lipoid bodies
in the granulosa lutein cells during the first days of the new corpus
luteum. Lastly, Novak (’16) reports five early corpora similar
to those of Meyer, whose conclusions he follows.
PREVIOUS WORK ON THE CORPUS LUTEUM OF THE SOW
It is said that von Baer’s celebrated monograph (’27) announcing

the discovery of the mammalian ovum, contains a description of
the early corpus luteum of the sow. ‘The first account of the
histological development in swine ‘which has come into my hands,
however, is that of Zwicky (’44), entitled ‘De corporum luteo-
rum origine.”” Zwicky was a medical student who was set to
work by the distinguished Henle to study the formation of con-
nective tissue in fibrin clots, Henle being under the mistaken 1m-
pression that the corpus luteum represented the conversion of the
clotted follicular haemorrhage into scar tissue. The student was
acute enough to correct his master’s error, even though he found
haemorrhage into the follicle in two-thirds of the early corpora
lutea of swine. As a result of his studies, he announced himself
as on the side of von Baer and Bischoff in favor of the granulosa
1 This Russian dissertation seems to me the best contribution to the origin
of the human corpus luteum yet presented. As it was abstracted for me by a
Russian-speaking scientific colleague, not especially acquainted with histological
methods, I quote it with some slight hesitation, but I believe our interpretation
is correct.

origin of the corpus luteum, but his description shows that he

was probably including the theca interna as part of the granulosa.
The little dissertation must remain as a not uninteresting example
of the effect produced upon the work of an active student by the
recently announced cellular theories, rather than as an important
contribution to its subject.
The next to use the sow for study was Paladino (’79, ’80, ’81,
87), who collected about 500 corpora lutea of 100 sows. His
extensive papers propose a peculiar theory of his own, namely,
that the entire granulosa is lost before rupture of the follicle, and
that the theca externa, carrying blood-vessels, proliferates in-
ward, to form the corpus luteum tissue, pushing the theca interna
before it to form a central connective-tissue core. In 1900 and
1905 he repeated his views of twenty years before in criticism of
Sobotta, who pointed out in return that Paladino’s writings con-
tain no evidence at all, in text or plates, that he had ever seen
developing stages of the corpus luteum; a Just criticism, as study
of the originals has convinced me.
Far different is the work of Benckiser (’84), who has given a
very careful description of a small series of stages. He had,
without doubt, normal and mature Graafian follicles just prior
to rupture. These contained the membrana granulosa intact.
In his recently collapsed follicles, containing large clots, the gran-
ulosa had been torn off in places; when there was no hemorrhage,
this layer remained very largely in situ. His next stage is much
further developed, its wall showing only one homogeneous layer,
and the gap was bridged by the assumption that the granulosa
had degenerated during the interval. The account is clearly
written, its author was describing what we can now state to be
normal specimens, and it was only lack of sufficient intermediate
stages that led him into an error of interpretation.
One of the most frequently quoted works is that of Clark
(98), whose material consists of ovaries collected at random in
the slaughterhouse. It is said that among the sows used by the
butchers there were many undergoing cestrus, but it is not stated
that any of those ovaries used in the research were known to be
from the animals in heat. Clark gives a good description of the
immature follicle at growing stages. For study of the mature

follicle he selected large follicles, without discovering whether or
not they contained normal maturing ova. From these large
follicles all the granulosa cells had disappeared. The author was
willing to consider the possibility that they might be atretic, but
inclined to rate them as normal because they seemed logical pred-
ecessors of his next stage. Much is made of the fact that the
theea interna cells of ripening follicles contain granules of yellow-
ish fat, which are taken to be the ‘lutein’ already present in the
future lutein cells before rupture. This assumption rather over-
reaches itself, as one glance at fresh corpus luteum tissue of the
sow will show that there is no microscopic yellow pigment present
in the so-called lutein cells. In the one pair of ovaries next
described, some follicles were ruptured, others were not. In the
latter, the granulosa was no longer visible, except for a few cells
lying in the cavity. The theca interna was thickened and closely
resembled lutein tissue. In a later stage there was a central
cavity rimmed by connective tissue, supposed to represent the
membrana propria pushed before the thickening theca interna.
Sobotta (99 b) has given a vigorous criticism of Clark’s specimens,
explaining his so-called mature and just-ruptured follicles as cases
of atresia. Volker (’05) has also pointed out what he considered
errors of interpretation of the specimens.
However, Doering (99) came to the defense of Clark, also
using material collected at random. He states that the wall of
a recently ruptured follicle shows no granulosa. His principal
evidence, however, is from one corpus luteum of the sow, which
shows, near the center of the section he figures, a flattened circle
of granulosa cells. This he interprets as the granulosa of the
same follicle in which the corpus luteum formed, which had been
pushed inward by the proliferating theca interna, and which for
some reason had not degenerated. I have seen very much the
same appearance when a young growing follicle had crowded itself
into the side of an old corpus luteum, so that a tangential sec-
tion appeared to show a follicle within a corpus luteum.
Jankowski (’04), using the same method of collection, would
seem to have had normal mature follicles of swine; at any rate,
the granulosa was intact. The ova were not seen. His very
early corpora lutea, showing stigmata at the point of rupture,
also contain the granulosa in situ and completely preserved,
except that the cells are swollen, irregular in form, and contain
vacuoles. The theea interna cells are large, contain lipoid gran-
ules, and resemble lutein cells. No specimens between this and
the solid corpus luteum are presented. Upon such evidence he
confirms Clark’s account.
Kopsch (’01) demonstrated at a meeting of the Anatomische
Gesellschaft certain preparations by Menzer of the corpora lutea
of swine three, six, and ten days after copulation. This contri-
bution appeared by title only, and our sole information as to its
nature is the statement of Sobotta that Menzer’s specimens are
in general agreement with his own views.
It is fair to say that the theory of the origin of the corpus

luteum from the theca alone, though it still holds a place in
current literature, has no good evidence in its favor. Every in-
vestigator whose methods assure us that accurately dated speci-
mens of a sufficient number of stages were in his hands has de-
clared the persistence of the granulosa cells and their transfor-
mation directly, with little or no mitotic division, into the
characteristic large ‘lutein cells’ of the corpus luteum. The
problem has shifted during the sixteen years whose progress I
have reviewed; the present point of interest is as to the fate of
the theca interna. Are its cells all converted into connective
tissue; do they persist as a special peripheral layer of the corpus
luteum; do they assume an impenetrably close resemblance to
the granulosa lutein cells; or can we find some other and clearer
explanation of the problem of their disappearance?
The following pages contain the results of an attempt to answer
these questions. Choice of the species to be studied was influ-
enced by several reasons. There is a frequently expressed idea
that perhaps the larger and smaller animals differ in the forma-
tion of their corpora lutea, as they do in many other features of
their reproductive cycles; the sow is large and has an ovulation
cycle not unlike the human species. Other considerations in-
ORIGIN OF THE CORPUS LUTEUM Leal
clude previous experience of the author with the ovaries of swine;

certain presumed histological advantages of the species, to be
explained later, and, above all, the fact that the previous work
on swine has been much quoted by writers in confirmation of the
theeal-origin theory. Here, if anywhere, the application of
modern methods of research should settle the old difference once
for all.
MATERIAL AND METHODS
As it has been pointed out that the only hope of trustworthy

results in this problem depends upon the possession of an un-
broken series of normal specimens of known ages, a description
of the material in the author’s hands and the methods of obtain-
ing it will be given in detail. The first step was a preliminary
investigation to obtain exact knowledge of the period in the
reproductive cycle at which the ova are shed from the ovary in
order that mature follicles and very early corpora lutea might
be obtained. The results of this study have already been pub-
lished (Corner and Amsbaugh, °17) and will not be repeated in
detail here. We were able to confirm the current supposition
that in swine ovulation is coincident with the oestral period,
and by this fact we are at once provided with the means of
obtaining the desired stages of corpus luteum formation.
The females of the wild swine of Europe are monoestrous,
according to Kaeppeli (08), having but one period of heat in the
year; but under domestication the sow becomes polyoestrous,
coming in heat at intervals of two to four weeks, usually about
every twenty-one days, as all breeders agree. The pericd of
heat commonly lasts three days and is characterized by sexual
excitement and in some individuals by swelling, reddening, and
slight eversion of the vulva, or even at times by a serous, mucous,
or partially sanguineous discharge from the genital orifice. Ifa
boar be present, the sexual excitement is made apparent by
ready acceptance of coitus (commonly on the second or third
day of oestrus) ; if none but females are in the pen, the sow in heat
will be seen to sniff at the genitals of her neighbors and ‘ride’
them in imitation of coitus. Frequently the sow is the recipient
toe GEORGE W. CORNER
rather than the donor of these attentions. The period is not ter-
minated by coitus, but continues until the end of three days.
For the purpose of the present investigation, the condition of
oestrus was observed while the animals were alive in the yards of
the packinghouse. The sows were marked, and on the day of
killing they were traced through the processes of the abattoir and
the internal genitalia received from the hands of the eviscerator.
The Fallopian tubes were then removed by cutting across the
upper portion of the uterine horns, were carried to the laboratory
in 0.7 per cent saline solution, and there washed out by inflating
them with salt solution through a slit in the wall near the fim-
briated extremity. After inflation with the fluid, the tubes were
gently ‘milked’ into a Syracuse dish, and the washings examined
with the dissecting microscope. This simple and almost infal-
lible method of finding the ova was suggested to us by Professor
Evans as an improvement upon Martin Barry’s practice of milk-
ing the tube without injected fluid (39). As we have subse-
quently found, it had been used by Sobotta (in the rabbit) and
no doubt by others as well.
We found that ovulation occurs on the first or second day of
oestrus, and that the stimulus of copulation is not necessary to
cause rupture of the follicles. The ovaries of all sows killed
during heat contain mature Graafian follicles ready to rupture or
just ruptured, in which latter case the ova are in the tubes and
may be recovered therefrom for study by the method described
below. Little or nothing has been known of the mature ovum
of the sow, and we have found no record of any previous observa-
tion of the unsegmented ovum from the tube. We measured
fourteen fresh tubal ova from nine sows and found the diameter,
including the zona pellucida, to vary from 155y to 165y, the zone
being about 10u in thickness. The ova are plainly visible to the
naked eye if placed against a strong light. We have not noticed
a radial striation of the zona pellucida either in fresh or fixed
ova. Theovumis filled with yolk granules of varying sizes, usually
about 34 to 5u in diameter, which are so numerous and so
refractile that they quite conceal the nucleus.
The author has presented (17 ¢) a brief study of the matura-

tion of the pig’s ovum, based upon some of these specimens,
which indicate that the sequence of events is the same as in other
mammals. The first polar body and the second polar spindle are
formed in the ovary just before rupture. After the entrance of
the spermatozoon, which occurs in the tube, the second spindle
completes its division, and the presence of two polar bodies is
therefore a sign of fertilization. If the ova are not fertilized, they
degenerate in the tube with the second spindle undivided. Just
how long they survive is not known, but by analogy with the
smaller and better-known mammals, we may assume that after
three or four days they are no longer capable of segmentation;
the degenerating ova may be found in the tubes a few days longer.
It is said that pregnancy is more likely to result when the sow is
served on the second day of oestrus. The number of ova ex-
truded at one ovulation, and consequently the number of fresh
corpora lutea in one animal, may be quite large. One prolific
sow is known to have given birth to twenty-three pigs in one
litter. However, in the mixed stock, not especially adapted for
breeding, which is found in the abattoirs, small litters are the
rule. Records of 128 sows raised in Maryland, presented in my
paper of 1915, show that the corpora lutea of pregnancy in both
ovaries numbered one to sixteen, averaging eight, and that the
number of foetuses in the uteri of the same sows varied from one
to ten, averaging six. Failure of fertilization, abortion, and re-
sorption of embryos dying in utero account for the fact that not
all the eggs of one ovulation proceed to full development.
About 133 embryos of the sow younger than two weeks, taken
from twenty-three sows, have been observed and described in the
literature. The youngest of all are the three ova found in three
different sows by the present writer and Amsbaugh (717), in which
conjugation of the pronuclei had not occurred. It was not possible
to know the exact time of insemination in these animals, but in one
case it is believed that the animal had not been in heat, and conse-
quently had not copulated, more than forty hours before kill-
ing. R.Assheton (’99) studied about 100 specimens during the first
ten days, the youngest stage being that of two blastomeres. It
would seem that fertilization may occur about the end of the first
day or may be postponed until two or three days after copulation—
a conclusion which he draws from finding embryos of the same
stage in two sows killed on the fourth, fifth, and sixth day post
coitum. Likewise, embryos in the same uterus may vary rather
markedly as to their state of development, for instance, one uterus
contained ova of two segments, of nine segments, and completed
morulae. For this reason it is possible to give only an approx-
imate time schedule of early development. The ova pass down
the tube rapidly and enter the uterus about the fourth day post
coitum. Assheton didnot find any stage further advanced than
four blastomeres in the Fallopian tubes. (A specimen found by
the present writer and Mr. Felix H. Hurni contained ova of two,
four, and six blastomeres, all in the tube.) Assheton found that
various sows killed on the sixth day presented uterine embyros
from the stage of six blastomeres to fairly well-developed blasto-
dermic vesicles. By the seventh day the zona pellucida has
usually disappeared and the inner cell mass of the early vesicle
has differentiated into two layers, the epiblast and the hypoblast.
By the twelfth day the great elongation of the blastodermic
vesicle which is so characteristic of the pig, is well under way and
the vesicle is already 10 to 12 mm. long. By the fourteenth day
each vesicle may measure 20 em.; in the embryonic area the prim-
itive streak is well developed and there are from one to three so-
mites. In addition to Assheton’s studies, thirty embryos of the
ninth, tenth and eleventh days have been described by Weysse
(94), and from the fourteenth day to about the twenty-fifth we
have the accurate tables of Keibel (97). For older (foetal)
stages, no good age-length ratios have been determined. ‘The
period of gestation is usually 116 to 120 days. It is stated that
sows undergo oestrus and may become pregnant again five weeks
after littering.
During the progress of this investigation the ovaries and uteri
of several thousand sows have been examined macroscopically,
and the corpora lutea of about 300 have been studied under the
microscope. The permanent preparations upon which the fol-
lowing description is based comprise sections from the Graafian
follicles and corpora lutea of 171 sows of which there are records
sufficient to determine the stage of the reproductive cycle. In
162 of them the ova, developing embryos, or foetuses were
examined and recorded.
Twenty-four were killed during the oestral period or within
the first week after the onset of heat. Some of the tubal ova
found were unfertilized, others were fertilized and were in stages
from the one-celled to the six-celled embryo. Five of the twenty-
four sows mentioned were obtained before a method of discover-
ing the ova had been acquired, and the ova were therefore not
sought, but as the dates of copulation were noted at the Univer-
sity of California Farm by Professor Thompson, it seems proper
to include them, since their corpora lutea agree with the others
in structure.
Six sows were taken in the second week after ovulation. As
the ova were unfertilized, they had degenerated, and were not
found, except shriveled eggs in two of the sows. It chanced that
none of those sows which had copulated were killed during this
period, and thus the opportunity to obtain embryos of the second
week did not fall to my lot.
Fifteen contained embryos of the third week, from five to
thirty-nine somites. The ovaries of the eleven youngest of these
were given me by Prof. F. R. Sabin; some of the embryos to
which they were related are described and pictured in her recent
contribution to the early vasculogenesis of the pig (Carnegie
Institution of Washington, Contributions to Embryology, No.
18, 1917).
One hundred and twenty-four compose a complete series from
animals containing embryos of the fourth week to the end of
pregnancy, the embryos or foetuses being measured in each case.
Two were obtained from sows which had littered seven and ten
days before killing, respectively. Most of the older corpora
lutea of pregnancy were prepared in the Anatomical Laboratory
of Johns Hopkins University, and formed part of the material for
my previous monograph (’15). They have been restudied in the
light of the results gained from the specimens of the first fourteen
days after ovulation, all of which were obtained in California.
The collection of this material would have been impossible

without the special and unusual codperation which has been ex-
tended to this laboratory by the Western Meat Company of San
Francisco. I refer to their donation of permanent laboratory
quarters in their West Berkeley plant (Oakland Meat and Pack-
ing Company) to the Anatomical Laboratory of this institution.
I owe especial thanks to Mr. J. O. Snyder, general superintendent
of the Western Meat Company, and to Mr. Ralston B. Brown,
superintendent of the Oakland Meat and Packing Company,
and to many other members of the staffs and employes of both
these establishments; to Dr. H. H. Hicks, U. 8. Supervising In-
spector, Dr. G. R. Ward, and other members of the U. 8. Inspec-
tion Service at the South San Francisco plant, and to Dr. Thomas
Presst, of the California State Inspection Service. To Mr. R.
B. Brown in particular I owe the opportunity of observing living
animals and of obtaining their pelvic organs, often at the cost,
I fear, of some inconvenience to the routine of his plant. The
permanent laboratory space provided by him at the packing-
house has been invaluable during the prosecution of this work.
I am further indebted to Professors Evans and Sabin for the con-
tribution of ovaries with the corresponding early embryos; to
Prof. J. I. Thompson, of the Department of Agriculture of the
University of California, for observing and marking five animals,
and to Messrs. A. E. Amsbaugh and Felix H. Hurni for assistance
in the collection and preparation.
In general the younger specimens were fixed in Bouin’s fluid,
the older in 10 per cent formol, these fluids being selected to
secure the advantage of fixation in slow aqueous coagulants, as
will be explained in the next section; small pieces of many
ovaries were placed in osmium tetroxide for study of the lipoids.
Blocks were imbedded in paraffin and celloidin. The chief
stains used with the specimens herein described comprised
haematoxylin and eosin, Heidenhain’s iron haematoxylin, Mal-
lory’s triple connective-tissue stain, Van Gieson’s mixture, and
several lipoid-soluble dyes (Nile-blue sulphate, Sudan III, Schar-
lach R), besides many special procedures applied to fresh and
fixed tissues.
ORIGIN OF THE CORPUS LUTEUM ewe
SPECIAL CYTOLOGY OF THE LUTEIN CELLS OF THE SOW
Four years ago the writer undertook, at the suggestion of

Professor Mall, to study the corpus luteum at different stages of
pregnancy, with the aim of learning through the varying appear-
ances to standardize the stages as a means of determining the
ages of embryos and foetuses (Corner, 715). It was very good
fortune that led to the choice of the pig for the first studies, for
a useful peculiarity of cytoplasmic structure was found to occur
in this species. If we take a section of the corpus luteum of a
pregnant sow whose foetuses are perhaps 100 mm. long, fixed in
formol, and stain it with any strong cytoplasmic stain, study of
the lutein cells shows that the cytoplasm contains unstained
areas which are roughly concentric to the nucleus, and which
appear to form canal-like paths in the cell (fig. 1). In the younger
corpora the canals grow more and more complex, assuming the
form of wide V-shaped spaces, long clefts, and circles in the
cytoplasm, so extensive that the nucleus is surrounded only by
a narrow zone of endoplasm. But it is in the corpora lutea of
pregnancies under 30 mm. that the highest development of the
exoplasmic zone is found. Here the entire outer part of the cell
is occupied by a curiously elaborate system of vacuoles, almost
every one of them in turn containing a spherule of substance
which, although it takes the same stain as the cytoplasm, yet has
a more hyaline appearance, and is seen in the section as a bright
ring. Within many of the spherules is found another and tiny
vacuole (fig. 2, 7). Corpora lutea of pregnancies with foetuses
more than 140 mm. long contain no trace of this system (figs. 23
and 24), and by careful attention to the degree of its develop-
ment it is possible, therefore, to estimate the age of the cor-
responding embryo with some accuracy. Taking other histo-
logical features into consideration, I find myself able to detect
the stage of pregnancy within close limits by examination of the
corpus luteum alone. The same bodies are present in the corpora
lutea of dogs, and were seen in the lutein cells of rabbits by Cohn
(03), who undertook certain microchemical studies upon their
nature, which I have been able to extend. It was tentatively
suggested, in my former paper, that they represent an elaborate
modification of the Golgi-Holmgren intracellular apparatus.

This view I have had to discard as a result of work which had
led to the correct interpretation (Corner, ’17 a,b). The spher-
ules are due to the presence of a lipoid, probably of phosphatid
nature, which is sufficiently oily to round up in the presence
of water. The round droplets thus produced usually surround
the preéxisting globules of neutral fat present in consider-
ence
¢ Laos ~
Fig. 1 Cells of corpus luteum of pregnant sow (foetuses 100 mm. long),
showing spaces in cytoplasm. Mallory’s connective-tissue stain. Formol fix-
ation. X 810.
Fig. 2. Cells of corpus luteum of pregnant sow (embryos 20 mm. long). For-
mol fixation. Mallory’s connective-tissue stain. > 810. 7, vacuoles in lutein
cell; th.l.c.1, theca lutein cell, type 1; th.l.c.2, theca lutein cell, type 2.
Fig. 3 Cells of corpus luteum of pregnant sow (embryos 20 mm. long). For-
mol fixation folloaved by osmium tetroxide. » 810.
able numbers in the early corpus luteum cells of swine (fig. 3).
After immersion in alcohol, xylol, ether, or other lipoid solvents,
both the fatty center and the phosphatid substance of the spheri-
cal droplet are dissolved out, leaving only a hollow sphere (ap-
pearing as a ring in thin sections), probably composed of proteid
constituents of the cytoplasm precipitated in the spherules during
fixation. The bodies are not seen in fresh tissues nor in material
fixed with very rapid coagulants like osmium tetroxide, which
precipitate the proteids before the oil droplets round up. The
microchemical evidence of these conclusions is given in the
articles cited.
The appearances in question, therefore, are simply the result
of methods of fixation which do not preserve certain obscure
lipoids in’ their natural diffused state. But the artifact is a
useful one. In the first place, it enables us to follow the changes
in amount of the phosphatid substance during the advance of
pregnancy. It also allows us to estimate the age of a corpus
luteum of pregnancy from the histological appearance alone, and
it gives us a constant (even though artificial) cytological char-
acteristic of the cell which can be used in determining the early
history of the lutein cells. Due no doubt to different physical
state of the cell lipoids, the phenomenon does not occur in the
human and bovine corpora lutea (a former statement of the
author to the contrary).
THE MATURE FOLLICLE
As the reader has perceived, one of the crucial points in this

debate has been as to the condition of the granulosa of the
mature follicle. Some investigators think that this layer de-
generates before rupture, others that it remains intact. It
would seem, offhand, an easy matter to obtain mature follicles
and settle the question at once. To be certain that a given
follicle is really mature is very difficult, however, and particu-
larly so in some species. Mere size is no criterion, for full-sized
follicles are not infrequently in a state of advanced atresia. The
presence of maturation processes in the ovum is no more certain
a sign, for the formation of the polar bodies, as was pointed out
by Flemming (’85), is a frequent occurrence in early atresia.
As atresia may set in at any time in the life of a follicle, even up
to the last, it is obvious that we can never state with complete
assurance whether a given Graafian follicle is doomed to degen-
eration or is about to rupture and give rise to a corpus luteum.
We shall probably not be in error, however, in assuming that a
follicle is normal and mature if it is taken from the animal at a
time when ovulation is known to be imminent, and if it contains
a normal ovum in which the process of maturation is under way.
To satisfy these requirements is easy when the animal is small
enough to be under observation in the laboratory, when an im-
pending ovulation can be predicted (as, for instance, in the rat
and mouse, which are now known to ovulate about eighteen
hours after littering) and when the small size of the ovaries and
tubes readily permits serial sectioning. In animals like the hog,
however, it is more difficult to observe these two criteria of the
mature follicle, and no previous investigators of this species
have watched the animal during life in order to determine the
imminence of ovulation, nor have any taken the pains to find
and study the ova of the follicles which they described as ripe.
In the author’s material, of sixteen animals known to have
been in heat when killed, only two were taken early enough in
oestrus to contain unruptured follicles. In one, all the follicles
were still unruptured. Three of them were successfully sectioned;
two of them contained ova with nuclei presenting ‘germinal
vesicles,’ the third showed the first polar body and the second
polar spindle. In the second sow, one of the follicles had rup-
tured; the tubal ovum could not be found; one of the remaining
follicles, upon sectioning, showed its ovum to be in the matured
state, with the second polar spindle formed. ‘There can be little
doubt, therefore, that the follicles in question were perfectly
normal and would have immediately shed their ova and devel-
oped into corpora lutea had the sows not been slaughtered.
These follicles possessed clear, translucent, almost spherical
walls, protruding a great part of their bulk from the ovary, as is
characteristic of the species. They all measured about 7 mm.
in diameter, the measurement ranging from 6 to 8 mm. in some

which were distorted by crowding.2. The surface presented no
‘stigma’ or other sign of impending rupture. On section, they
were found to possess the usual three layers, and the membrana
granulosa was present and intact, showing no sign of degenera-
tion. The wall of that part of the follicle lying deepest in the
ovary presents a slightly wavy contour toward the cavity.
The cells of the granulosa (fig. 4, a) form a layer about six to
nine cells deep, or about 0.13 to 0.17 mm. thick. Those cells
nearest the membrana propria form an irregular columnar layer,
but the upper cells show less semblance of order in their arrange-
ment. The cells are round or polyhedral, from 8 x 8y to 10 x 16u
in diameter (in celloidin sections), with round nuclei 5 or 6u
in diameter. The cytoplasm appears homogeneous after the
usual fixing reagents, except for a few vacuoles due to the
presence of lipoid substances, as will be explained later. Often
the cells possess short processes which meet those of the neigh-
boring cells so as to make the tissue resemble a syncytium.
The theca interna is about 0.09 to 0.10 mm. thick, or a little
more than half as thick as the granulosa. Its most striking
characteristic is the presence of three to five layers of large
‘epithelioid’ cells, usually from 10 x 17 to 12 x 17 in diameter,
but occasionally reaching larger sizes, up to perhaps 16 x 24y
(fig. 4, b). On section, they are oval, spindle-shaped, or almost
rectangular, with their long axes in the circumference of the
follicle, so that they usually lie at right angles to the columnar
layer of the membrana granulosa. Between the larger cells, and
especially along the inner border of the layer formed by them,
are others of small size (though still somewhat larger than the
granulosa cells), which are similar in appearance to the large
theca cells. In material fixed in Bouin’s fluid the theca cells
2In a previous publication (715) I fell more or less into the same error of
method which I am now imputing to others, and attempted to determine the
size of the ‘ripe’ follicle without knowing the state of the enclosed ovum. The
larger follicles there mentioned, however, agree histologically with the accu-
rately known specimens described in these pages, and it is therefore likely that
my previous conclusions were correct, namely, that the normal mature follicle
may attain a diameter of 10 mm.
th, int:-B-. Qs."
a.
Fig. 4 a, Portion of wall of unruptured Graafian follicle (sow in heat, ova

maturing). Iron haematoxylin. X 380. b, A few cells from theca interna of
same specimen. X 800. gran., membrana granulosa; th.int., theea interna;
th.ext., theca externa; sp.c.z., spindle-cell zone of theea interna; b.v., blood-
vessels.
are notable for the presence in their cytoplasm of a number of

vacuoles, giving them a striking honeycombed appearance (fig.
4,6). These vacuoles are due, at least in part, to the presence
in the fresh tissue of granules of fat-like substance, packed closely
into the theca cells, whose chemical nature has not been deter-
mined (fig. 5, b). In some species they are quite yellow, since
they hold in solution some of the lipochromes common in the
ovaries of certain animals, but in the pig they are practically
colorless. It is of course the appearance of these large fatty
cells of the theca which has helped establish the belief that they
are the precursors of the ‘lutein cells’ of the corpus luteum. The
granules are soluble in alcohol; in osmium tetroxide they take a
color varying from gray to deep black; they take a decided red-
dish color with Herxheimer’s alkaline Scharlach Rot, but appear
not to stain at all with Nile-blue sulphate; they are not aniso-
tropic. From these reactions we may assume that the substance
is of lipoid nature, but is perhaps not a neutral fat. The gran-
ules are variable in diameter, from 0.54 to 1.54, a few even
reaching 2.5u. Many of the theca cells contain, instead of
lipoid granules, vacuoles which are not stained even in osmium
preparations, and which therefore must contain either a modi-
fied form of the lipoid or some other substance which is not ren-
dered insoluble by combining with OsQ, (fig. 5, 6). The smaller
cells of the theca interna mentioned above usually have finer
granules, but there are all transitions between the large and small
types. The cells of the granulosa contain a few very small
granules, uniformly black after osmium fixation; these are usually
more numerous in the basal layer of the follicular epithelium.
Between the granulosa and the cellular layer of the theca in-
terna just described is a narrow zone (0.03 to 0.04 mm.), which
contains chiefly spindle cells without fatty inclusions or vacuoles
(figs. 4, a, and 5, a). These cells appear to be of two types:
first, the endothelium of the blood-vessels which lie in this zone
and, second, the fibroblasts of the perivascular tissue, which are
part of a light network of connective-tissue reticulum support-
ing the theca and forming a base for the granulosa, for I agree
with J. G. Clark that the membrana propria is nothing more

(GE
Fig. 5 a, Portion of wall of unruptured Graafian follicle (sow in heat, ova

maturing). Osmium tetroxide fixation without further stain, showing distribu-
tion of fats. > 380. 6b, A few cells from theca interna of same specimen. X 800.
gran., Membrana granulosa; th.int., theea interna; th.ext., theca externa; sp.c.z.,
spindle-cell zone.
144
nor less than a network of these fibrils applied closely to the

base of the granulosa layer.
The theca externa consists of a layer of long spindle-shaped
cells, shading off into the stroma of the ovary (or into the cap-
sular connective tissue, over that part of the follicle which is
jutting out from the ovary). It is composed chiefly of collage-
nous fibrils and their associated fibroblasts, but it is highly in-
teresting to note in connection with the subsequent collapse of
the follicle, that there are also a good many smooth muscle
fibers, as is readily seen by the use of Van Gieson’s stain. There
are no elastic fibers, except in the walls of the larger blood-
vessels.
Just before rupture there are many mitotic figures in the cells
of the theea externa, but only occasional signs of cell division
in the theca interna and the granulosa.
Injected specimens show the blood-vascular distribution to be
as described by His and J. G. Clark (fig. 6). Large vessels form
a network in the theca externa, sending twigs inward to form
a generously anastomosing plexus which les in the above-
described spindle-celled zone of the theca interna.
I have found a curious arrangement of the blood-vessels in
the ovum-bearing area of the follicle. The discus proligerus is a
cone-shaped or rounded projection of the granulosa bearing the
ovum near its apex, which until shortly before maturation of the
ovum is composed of densely packed granulosa cells. At its
base, in this particular species, are found a number of little
vascular loops sprouting up from the vessels of the theca interna
well into the granulosa of the discus, and pushing before them the
cells of the basal columnar layer (fig. 6, loops). The basal cells
appear as if radiating from the loops, and like all the cells of
the area occupied by the loops are enlarged and have a much less
dense cytoplasm than the other granulosa cells. Those loops
which are near the center of the discus are longer than those
toward the periphery. Such vascular loops penetrating the
granulosa have apparently not been mentioned previously.
They are not to be found in the rat and mouse, the only other
species which } have studied in this regard. It would seem that
146
0 SE
GEORG E W.
CORNE
vy
R
JO
som
sjassoa poyooful
I UJIM r
wl BIPUT “YUL re*¢Qz 1 ‘osip i [
“hi0ud snostp I Ssnaostoaid
“woub I
vuBriquiour JULY}
Sesoynuvis
‘ourvy)
T[VA
‘BLT
WNAO
ATIBOU
-poolq
BOOTY SBUAOJUL
YZ “7rd BOOYY !BUIO}XO‘sdo007 ABPNOSBA
SCOOT SUIIO}UN “BSO[NUBIS
QoINyvUt
UOTWIOg
‘oOT[FOF
COPISOA
BUIMOYS
URYBRUD
OATPVULUTIOS,
the large size of the discus proligerus (as large as the entire
follicle of the mouse) places the ovum at such a distance from the
vascular bed that special vessels are needed for its nutriment.
Be this as it may, by the time maturation of the ovum is in
progress, the character of the discus proligerus has become con-
siderably modified (fig. 7). Its cells are very much swollen by a
Fig. 7 Portion of wall of mature Graafian follicle (sow in heat), showing

mature ovum in situ, and discus proligerus in process of dissociation. X 50.
vacuolization of the cytoplasm, so that they stand farther apart

from each other, and in many places seem to be no longer in
contact. About the ovum the cells of the corona radiata still
hold together, but the rest of the discus has nearly crumbled
away, and the slightest disturbance must complete the freeing
of the ovum, bound as it now is to the parietal granulosa only
by a few strands of cytoplasm. ‘This rearrangement of the discus

proligerus was long ago shown by Bischoff (78) to be a trust-
worthy sign of impending rupture of the normal follicle. Cer-
tain subsequent observers, wondering how the ovum could be
freed from its apparently secure moorings, were led to conjecture
that there is a total desquamation of the granulosa—an error in
which they were confirmed by the fact that for lack of the proper
stages they did not see the mechanism for cutting off the ovum
described by Bischoff, but did see, on the other hand, the com-
plete dissolution of the granulosa, in follicles which we now know
to have been atretic.
THE FRESHLY RUPTURED FOLLICLE
Four animals of my series contained follicles which had rup-

tured very recently. One of these sows had shown the first
signs of heat at some time between thirteen and twenty-two
hours before killing, another between sixteen and thirty-nine
hours before killing, and the other two were in the second or
third day of oestrus (probably the second). As we have shown
on page p. 132, rupture of the follicle occurs on the first or second
day of oestrus. In all four of these cases, unfertilized ova were
found in the tubes, there having been no copulation.
Apparently the act of rupture begins by the production of a small
slit in the exposed part of the follicle, through which the ovum
escapes to enter the tubal fimbria (figs. 8 and 9). A varying
amount of the follicular fluid, usually a considerable portion, is
extruded with the egg, and the follicle collapses as the volume
of its contents suddenly lessens. It seems that an important
part in this collapse must be taken by the fibers of involuntary
muscle which le in the theca externa; through their contrac-
tion the follicle is greatly diminished in all dimensions. At the
point of rupture, the muscle fibers draw the theca externa away
from the torn area, and the result is a slight eversion of the
wound, through which the theca interna and granulosa protrude,
forming a small reddish papule 1 mm. or less in diameter,
the so-called stigma (fig. 12). The eversion of the inner layers
appears at times to close the orifice immediately, but in other

cases the follicular walls do not come together about the opening
at once; the tiny slit is first plugged by fibrin, and later closed
permanently by proliferation of its edges, much as described by
Strakosch (15) in the human corpus luteum. From the minute
blood-vessels whose torn ends le in the stigma there is often a
slight oozing of serum or of blood, so that the surface of the
ovaries at this time may be roughened by tags of pale or bloody
exter
sRaud Reis
Point of Tupture
Fig. 8 Diagram of ruptured Graafian follicle (sow in heat, ova in tubes),

illustrating partial collapse without great infolding of walls. 14. (Compare
with figure 9.)
fibrin, sometimes forming temporary adhesions to the fimbriae of

the tubes. The entire ovaries and the tubes are usually much
congested during oestrus.
On section, the follicular cavity is collapsed to a mere slit
in some cases, in others it is still partially distended, owing to the
continued presence of more or less of the follicular fluid. For this
reason, the size of the structure varies, but in general it is much
smaller than the mature unruptured follicle, its diameters vary-
ing from 3.5 to 6.5 mm. when not distended by hemorrhage.

~ Most frequently the ruptured follicles are ovoid in form, about
4x4x5to5x5x6mm. in diameter. Owing to the collapse of
the follicle and to the contraction of the theca externa, the inner
walls of the cavity are no longer smooth, but are thrown into
folds whose complexity varies from that of low ridges in those
Point of rupture
Cavity of
corpus luteum
Unripe
follicle.
Spee
~Groan ulosa
Theca externa 13 .
interior of cavity”
edematous spaces intheca interna
Fig.9 Diagram of ruptured Graafian follicle (sow in heat, ova in tubes),

illustrating complete collapse with much infolding of walls. X 14. (Compare
with figure 8.)
follicles where much follicular fluid still lingers (fig. 8) to elab-

orately interwoven folds such as those shown in figure 9.
Microscopically, the follicular wall is found to consist of the
same layers as before rupture (fig. 10). There is no sign of any
degeneration of the granulosa, which is now somewhat thicker,
a
e
ORIGIN OF THE CORPUS LUTEUM We
since it lines a smaller cavity than before. The individual cells

are about the same size as former ’y, but in many places are now
elongated into oval or spindle forms by the stresses of the col-
lapse, appearing to have slid upon each other as the granulosa
thickened. The theca interna is the layer most affected, in a
mechanical way, by the sudden collapse, for in some of the
Fig. 10 Portions of wall of recently ruptured Graafian follicle (sow in first

day of oestrus, ova in tubes). X 330. a, Mallory’s connective-tissue stain,
Bouin’s fixation. 6b, Osmium tetroxide after formol fixation, showing distribu-
tion of fats. gran., membrana granulosa; th.int., theca interna; th.ext., theca
externa; memb. prop., membrana propria.
folds it is violently torn apart, so that there are many wide

spaces either within the theca interna or between the two thecae
(fig. 11, tear). These spaces are occupied either by networks of
fibrin, which may be altogether devoid of cells, or contain an
occasional theca interna cell, or a leucocyte; or the space may
ae GEORGE W. CORNER
not be an actual tear, but merely an oedematous area in which

the cells, connective-tissue fibers, and blood-vessels of the theecae
interna and externa are held apart by the tissue fluids. How-
ever, over many of the folds, and always in the depressions
between the folds, the theca interna is neither torn nor oedema-
Fig. 11 Portion of wall of recently ruptured Graafian follicle (sow in heat

first day of oestrus, ova in tubes; same animal as in figures 9 and 10), showing
torn area in theca interna in a fold of the wail. X 80. gran., membrana gran-
ulosa; th.int., theca interna; fear, torn area in theca interna; dep., depression
or recess ‘between folds of wall.
tous (fig. 11, dep.). The tears do not separate the theca interna
from the granulosa; these two layers are everywhere in apposition,
and the boundary between them is still marked at most points
by the slight wall of condensed connective-tissue fibers, origin-
ating in the innermost layer of the theca interna, the so-called
ORIGIN OF THE CORPUS LUTEUM ie
membrana propria (fig. 10). The theca externa is of course

drawn into the folds of the wall, and where these folds are very
deep, long spindle-cells of the externa may thus penetrate almost
to the center of the former follicular cavity—though of course
they are walled out by the inner layers (fig. 9). Dividing cells
are found not infrequently in the theca externa, but are quite
rare in the inner layers. The distribution of lipoid substances,
as indicated by the use of osmium tetroxide and Herxheimer’s
stain, is exactly as in the mature unruptured follicle (fig. 10, 6).
Leucocytes are found in the walls of all developing corpora lutea.
The blood-vessels are exactly as in the unruptured follicle, the
picture presented by them being modified only by the elaborate
infolding of the walls (fig. 12). At the point of rupture the torn
vessels of the thecal plexus present to the outside, and within a
few days of rupture have sprouted into a little rosette of capillaries
about the stigma, which helps to make this spot conspicuous by
its redness. In the production of the curious torn spaces of the
theca interna, described above, vessels of the theca interna are
not infrequently ruptured, with resultant haemorrhage into the
theca. If the loss of blood is very slight, the broken-down blood
is taken up by the large cells of the theca interna, in which the
phagocyted golden-brown pigment may remain for some days at:
least (fig. 13). In one of my cases there was a single local haemor-
rhage into the theca externa, and the nearest cells of the theca
interna were full of blood-pigment granules. However, when the
thecal haemorrhages are large, the resultant haematomata may
burst through the granulosa into the cavity. I am inclined to
think that we have here the source of most of the bleeding into
the early corpus luteum cavity. The now almost forgotten doc-
trine of Henle and Paterson, that the corpus luteum is formed
from the blood clot of the newly ruptured follicle, naturally led
_ to investigations into the importance and constancy of the haem-
orrhage in various species, which have been summed up by So-
botta in his paper of 1896. In the pig, Zwicky (’44) held that
bleeding is frequent, Paladino (’80) that it occurs in two-thirds of
the cases, Benckiser (’84) that it is inconstant, Spiegelberg (’65)
that it is important, and Bonnet (’91) that it is constant and
erte
th. ext.-*#,
Fig. 12 Recently ruptured Graafian follicle (ova found in tubes), blood-

vessels injected with India ink. X 15. gran., membrana granulosa; th.int.,
theca interna; th.ext., theca externa; sfig., stigma (blood-vessels at point of
rupture).
marked in extent. Sobotta (’96), reviewing the evidence, is

inclined to the last view. Pfliger (63), in an experimental
investigation, found that in cats and rabbits killed violently
there was much more frequent bleeding into young corpora
lutea than in animals killed without struggle and very carefully
autopsied. In my own specimens, out of sixteen sows whose
ovaries contained very early corpora lutea, dressed at a packing-
house using the relatively gentle method of scraping by hand,
four showed more or less blood in the corpora lutea and twelve
Fig. 13 Cells from theca interna of recently ruptured follicle (sow in heat,
ova in tubes). Iron haematoxylin stain, showing pigment and broken-down
erythrocytes in theca cells. > 1000.
were entirely free of macroscopic haemorrhage. In another

establishment, where the carcasses are conveyed 150 feet dang-
ling from a chain and are scraped by engine-driven revolving
vanes (so that in the bodies of pregnant sows, young foetuses
frequently suffer an effusion of blood into the amniotic sac),
there chanced to be a somewhat higher proportion of haemor-
rhagic follicles; but even there, in spite of such excessive vio-
lence, it is common enough to see delicate corpora lutea one or
two days old come through with no blood at all in their cavi
ties. Again, trauma at the time of killing does not explain away
all the haemorrhages; for instance, in those cases in which among
a number of solid, bloodless corpora lutea several days old, one
or two others are found distended with dark clotted blood to a
size exceeding the normal corpora. I feel that the present evi-
dence indicates that haemorrhage into the corpus luteum of the
sow, while not uncommon, is the exception rather than the rule,
and is of no anatomical or physiological importance. Indeed,
the arrangement of the follicle seems well adapted to prevent
any considerable loss of blood into the cavity, for the tiny ves-
sels at the place of rupture are promptly directed outward toward
the peritoneal cavity, while the follicle is provided with smooth
muscle, which keeps the walls tensely contracted, even after
rupture. When small haemorrhages occur, undoubtedly they
are readily resorbed, and the corpus luteum then goes on to de-
velop normally. When great enough to distend the follicle and
compress the growing wall, inhibition of corpus luteum forma-
tion presumably occurs, and we have here one of the causes of
corpus luteum cysts, which are very common in swine.
INVASION OF THE GRANULOSA
The next stage is represented by seven animals in my collection,

all of which were killed during oestrus, as normal ova were found
in the tubes. Moreover, four of them were observed during life,
and were actually seen to be in the second or third day of
oestrus. In three, copulation had not occurred; in three others,
fertilization had taken place, the ova showing the pronuclei
approaching conjugation; and in the seventh, the ova were seg-
mented into two, four and six blastomeres.
The first sign of an advance upon the previous stage consists
of the breaking-down of the membrana propria, at first at the
apices of some of the folds, later over the entire follicle, so that
the former sharp line of division between granulosa and theca
interna is no longer present (fig. 14). Wherever the membrana
propria is disappearing, slender spindle-cells are seen to be
insinuating themselves between the still closely packed granu-
ORIGIN OF THE CORPUS LUTEUM LEY
losa cells (fig. 14, sp.c.). The nature of the inwandering cells is
difficult to decide. In places there can be no doubt that they
are endothelial in nature and represent the first sprouts from
Fig. 14 Portion of wall of young corpus luteum (ova found in tubes), show-
ing swollen cells of granulosa with inwandering spindle-shaped cells. Mallory’s
connective-tissue stain, formol fixation. > 810. gr.l.c., granulosa lutein cells;
th.int., theea interna; th.ext., theca externa; sp.c., spindle-shaped cells.
the walls of the thecal capillaries, growing inward to the granu-

losa. In the endothelial cells mitoses are not uncommon at this
time. It seems quite likely that all the early invading cells are
of endothelial nature, but at some few points, however, it is

impossible to convince oneself that the spindle-cells have any
connection with the vessels, for they are not always arranged in
tubular form and are sometimes well disseminated throughout
the granulosa in advance of any circulation of blood. It can-
not be denied absolutely, therefore, that some of them may be
inwandering cells of the perivascular spindle-cell zone of the
theca interna. During these early changes the large cells of the
theea interna remain in their place, and I have never seen con-
vincing evidence of their conversion into spindle cells.
Practically all of those observers who have been convinced of
the persistence of the granulosa have described such an early
invasion of the innermost layer by spindle-cells, a stage which
was called by Robert Meyer the stage of proliferation, but as
in the pig there is doubt as to the interpretation of the observed
facts. Sobotta (’96) holds that all the cells of the theca interna
are converted into spindle-cells (fibroblasts), and wander into
the granulosa, dividing frequently, to form the connective-tissue
framework of the corpus luteum. In this view he is supported
by Marshall (’04) in his work on the sheep, and by O’ Donoghue
(16), who studied the marsupial ovary; but several authors,
including Vélker (05), Loeb (’06), and R. Meyer (11), working,
respectively, with the corpora lutea of the marmot, the guinea-
pig, and man, are inclined to consider the first inwandering cells
as endothelial, and deny the conversion of the theca interna cells
into fibroblasts.
It has been mentioned that the breaking-down of the mem-
brana propria and the invasion of the granulosa by spindle-cells
does not take place at once over the entire inner surface of the
collapsed follicle, but begins first at the apices of the folds, where
the structure has presumably been subjected to the greatest
mechanical strain. Because of this very important fact, we are
able to observe a definite stage at which, while in places there
is an actual intermingling of the two layers going on in part of
the structure (fig. 15 a, X), in other parts the two inner layers
of the wall maintain their original relations. During the same
period there is a marked and rather sudden change in the granu-
losa cells (fig. 15 6). Their cytoplasm increases in volume so

that the cells are now much larger in size, varying from 9.5 x I1y
to 14x 21» in diameter in sections, or half again as large in
diameter as before rupture. The nuclei become rather more
vesicular. Most important is the fact that many of the larger
ye. xe x, Le
at) fo
-°
e.
Fig. 15 a, Part of wall of developing corpus luteum in stage of spindle-cell

invasion (ova in tubes). Formol fixation, Mallory’s connective-tissue stain.
x 110. X, area of active invasion by spindle-cells (at apex of fold in wall).
cells now have in their cytoplasm large spaces containing rounded

bodies resembling rings in section, which we have already seen
(pp. 137, 139) to be characteristic of the so-called ‘lutein cells’
of the young corpus luteum in swine and to be due to the pres-
ence in the cytoplasm of an oily lipoid substance.

Furthermore, these changes in the granulosa cells are found to

oecur in all parts of the wall, as well in those areas as yet unin-
vaded as in those where granulosa and spindle-cells are already
intermingled. To sum up the evidence, there is a time in the
development of the corpus luteum, about three days after the
thc
Fig. 15 a, Enlarged view of portion of same as shown by rectangle. X 1000.

gr.l.c., granulosa lutein cells; th.c., theea cells; b.v., blood-vessel.
rupture of the follicle, when changes of structure within the

organ are already under way, and when many of the cells have
begun to acquire adult characteristics, but in some parts of the
organ the original relations of granulosa and theca interna are
still intact; in these areas we find that it is the granulosa cells

alone which have assumed the appearance of the large cells of
the corpus luteum, commonly called lutein cells.
The breaking-down of the membrana is followed by a rapid

sprouting and branching of the blood-capillaries throughout the
entire granulosa (figs. 16 and 17). This stage is represented, in
my material, by seven sows, of which one contained fertilized
ova (some unsegmented and some with two blastomeres); one
was killed about five days after the onset of heat, no ova being
found, probably having degenerated; one was killed about six
days after the onset of heat, one degenerate ovum being found;
the other four were among those received from the University
Farm School, in which the ova were not sought, but in which the
dates of copulation were accurately known, in two on the third
day and in two on the fifth day before killing.
Coincidently with the spread of the blood-vessels in a network
throughout the granulosa, there continues a marked swelling of
the cells of this layer, which double or more than double in
diameter, thus making an eightfold increase in volume; some of
them reach dimensions of 30u to 35u. The nuclei are larger
and more vesicular. I have never seen a mitotic figure in a
cell of the granulosa at this or later stages, and feel sure that the
generalization of Sobotta on this point is correct. In the
formol- or Bouin-fixed specimens, the periphery of the cells is
studded with the striking ringlike phosphatid artifacts of fixation
(fig. 18, gr.l.c.). The rest of the cytoplasm is thin and contains
irregular vacuolar spaces, due partly perhaps to the shrinking
away, during fixation, of the cell-substances which form the ring
bodies, and partly to the solution in the alcohols, xylol, or ether,
of other lipoid substances, which osmic preparations show as
small black globules in the center of each ring and also scattered
about the nucleus or throughout the cytoplasm. I have not
been able to apply other microchemical tests to the tissues at
this stage, but the globules are morphologically like the neutral
fat of later stages (fig. 3), and I assume that they are indeed
the neutral fat or its forerunner.
Fig. 16 Young corpus luteum (segmenting ova with one and two blastomeres
found in tubes); blood-vessels injected with India ink. Compare with figure 12.
x 15.
ee
Meanwhile, changes have also been taking place in the large

cells of the theca interna. Mitoses are more common, and the
former definite internal limit of this layer has been blurred by
the breaking-down of the membrana propria and the ingrowth
of the capillaries at all points of the follicular wall. Within the
Fig. 17 Enlarged view of small part of figure 16 as indicated by rectangle,

showing blood-vessels of theea interna branching throughout granulosa, > 45.
cells there are changes in the lipoid inclusions which are their
chief distinguishing characteristic. In some cells of osmic
preparations the granules are larger, in others smaller than before;
in some they do not form an insoluble black compound with
ie
mium tetroxide, but leave vacuoles of varying sizes, between

which a few stained granules may remain giving a character-
istic foamy appearance to the cytoplasm (fig. 19); and in
others practically all the fatty bodies and vacuoles have dis-
appeared, leaving a smooth homogeneous cytoplasm. In or-
dinary stained sections, then, the theca interna cells are of
about the same size as before rupture, their nuclei are perhaps
shghtly more vesicular, and the cytoplasm is either homogeneous
or contains many densely packed vacuoles, usually uniform in size
within any one cell, resulting in the foamy appearance.
Fig. 19 Theea interna cells of corpus luteum in stage of invasion, osmium

tetroxide fixation without further staining, showing varying degrees of fatty
inclusion. X 1000.
It will be obvious that the previous clean-cut distinction be-

tween the two layers is now lost. Heretofore we have been
able to distinguish them by position, size, and content; there
has been a wall of connective-tissue fibrils between the layers;
the cells of the granulosa have been smaller than those of the
theca interna; and the former have contained but small numbers.
Fig. 18 a, Part of wall of developing corpus luteum in stage of invasion (ova

in tubes). Bouin fixation. Mallory’s connective-tissue stain. X 110. 6, En-
larged view of portion of same as indicated by rectangle. Mallory’s connective-
tissue stain. X 1000. gr.l.c., granulosa lutein cells; th.l.c., theca lutein cells;
th.int., portion of original theea interna, near a fold in wall of corpus luteum;
b.v., blood-vessel.
of lipoid granules, the latter considerable amounts. But now

the membrana propria is done away with, the granulosa cells
have increased in size and are becoming rich in lipoids, while the
theca interna cells are losing their lipoids. It is not strange
that investigators have become involved in uncertainty regarding
the further fate of the theca cells. I have seen the abrupt end-
ing, at stages similar to those now being described, of two careful
attempts, by students in our histological courses, to follow the
theca cells of the mouse and rat by means of their osmium-
staining inclusions, owing to failure to observe further distine-
tions between the two cell types. In the pig, however, we pos-
sess a peculiar advantage in the tendency of the phosphatid
material to form the previously mentioned cytoplasmic rings
when fixed with slow aqueous fixatives, giving the granulosa
cells a distinctive appearance. There are other less regularly
present criteria, which when added together afford the prac-
ticed observer means of partially distinguishing the two cell
types; these are a tendency of the cytoplasm of the theca cells
to take acid stains somewhat more deeply than the granulosa
cells (perhaps this indicates merely a denser cytoplasm) and also
the regularity in size and closely packed disposition of the lipoid
granules or the vacuoles left when they disappear or are dissolved.
Following these clues, we find that many of the theca interna
cells remain in their original location about the periphery of the
follicle, running into the interior of the folds produced by the
collapse; but also that many of them, as the blood-vessels grow
inward, are carried or wander with the vessels, and become dis-
seminated among the cells of the membrana granulosa, where
they are finally lodged, either singly or in small groups, fre-
quently along the capillaries (fig. 18). It must be remembered
that a general scattering of the theca cells among the granulosa
in this way will not require a longer journey for any single cell
than the thickness of the inner layer, which is not more than
0.2mm. The cells thus immigrating resemble in every way their
mates left behind at the periphery and in the folds, some of
them containing large granules, staining black with osmium
tetroxide, others showing almost no fatty inclusions (fig. 19).
They are often broadly spindle-shaped or irregular in form,

sometimes compressed between the granulosa cells or applied
demilune-fashion to one of them.
Further changes are brought about by the great swelling of the
granulosa cells, which proceeds so far that the contents of the
follicle begin to equal and finally to exceed the capacity of the
contracted theca externa. The first effect of the internal pres-
sure is to fill whatever remains of the original follicular cavity
solidly with new tissue, then to compress the thecal cores of the
folds of the walls so that all fibrin-containing cavities and
Pr.
Fig. 20 Diagram showing outline of section of a corpus luteum about four

~ days after ovulation. X 5. pr., ‘Pfropf’ or hernia-like bulging of contents
through point of rupture.
oedematous spaces are obliterated, and the folds become merely

connective-tissue septa containing the remains of the theca in-
terna in the shape of a diminished number of theca cells en-
meshed by reticular fibrils; in the bases of the folds the blood-
vessels of the young corpus luteum enter, usually accompanied
by fibroblasts and fibrils proceeding from the theca externa. In
many young corpora lutea the swelling of the granulosa cells
finally causes a bulging of the contents through the outer pole
of the wall, at the point previously weakened by the rupture,
which in the prolific ovaries of the sow may be exaggerated by
the pressure of the many neighboring corpora lutea of the same

crop, until there is produced a knoblike hernia of corpus luteum
tissue sometimes containing a tenth or more of the whole corpus
luteum (fig. 20). This appearance is sometimes called by the
handy German name ‘Pfropf.’ In some species, as, for instance
the cow, it seems to occur invariably and to persist throughout
pregnancy, but in swine the hernia is not always produced, the
whole wall of the corpus distending evenly instead; and later it
seems to subside, as in most corpora lutea in more advanced
pregnancy there are no ‘Pfropfen.’
THE FULLY FORMED CORPUS LUTEUM AND ITS MORPHOLOGICAL

CHANGES UNTIL THE TERMINATION OF PREGNANCY
Invasion of the granulosa by the thecal vessels and cells begins

about the third day after the onset of oestrus (or about the sec-
ond or third day after rupture of the follicle) and is completed
about the sixth or seventh day. My series contains five sows
killed during the second week after ovulation and a large num-
ber from all stages of pregnancy from fifteen days after ovula-
tion on to full term and into the period of lactation, so that
altogether there is an unbroken series representing almost every
day of the entire reproductive cycle.
By the seventh day the corpus luteum may be considered to
have completed the first stage of its metamorphosis. It is solid
(unless there has been a decided haemorrhage into the cavity),
and it is already larger than the follicle in which it arose, reach-
ing diameters of 8 and 9 mm., although a slow increase in size
is yet to go on until the second or third week, by which time the
full diameter, 10 to 11 mm., is reached. The blood-vessels
have grown into a very narrow-meshed plexus, reaching every
cell. The remains of the former great folds of the walls are seen
as thin septa of connective-tissue fibrils running radially into the
corpus luteum, carrying the larger blood-vessels of the organ.
In some specimens, just inside the theca externa capsule and
along the septa is a layer of theca interna cells or sometimes a
few scattered clumps of them which have not chanced to in-
vade the granulosa. These clumps may be found as late as the
second month of pregnancy or may disappear long before (fig. 21).

In appearance the cells of these clumps resemble the theca cells
of the stage of invasion. In osmium-tetroxide preparations,
Fig. 21 Portion of corpus luteum from second month of pregnancy (foetuses

35 mm. long), showing a clump of theca interna cells still in their original posi-
tion. Formol fixation. Mallory’s connective-tissue stain. 450.
some retain many lipoid globules, others have lost all of their
fatty inclusions; but many present the previously noted foamy
appearance of the cytoplasm, due to the presence of many
small evenly packed vacuoles with a few tiny black granules

interspersed.
In the substance of the corpus luteum are now found two types
of cells whose differing characteristics become well marked after
the beginning of the third week. One type is that whose iden-
tity with the granulosa is fully demonstrated by the presence
of the peculiar lipoid spherules about the periphery of the cyto-
plasm (figs. 2.and 21). From now on these cells grow slowly in
size, until just before delivery some of them have attained the
ee leice externa,
Rigs 2250
immense size of 30x 45y. During the first few weeks the neu-
tral fat increases until it greatly exceeds the small amount found
in the cells of the granulosa before rupture, and then grows pro-
gressively less, finally almost disappearing by the 110th day of
pregnancy. The phosphatid substance which produces the
artifacts of fixation, so frequently referred to, disappears also,
and therefore during the last third of pregnancy the cells, in
ordinary formol preparations, possess a homogeneous cyto-
plasm, strikingly different from the greatly vacuolated cell sub-
stance of the earlier stages (fig. 23). Just before delivery,
however, great globules of an osmium-staining material, pre-
sumably a fat, appear about the periphery of some of the cells.
)—
ORIGIN OF THE CORPUS LUTEUM at
Fig. 22 Cells of corpus luteum of pregnancy eighteen to twenty days old

(embryos of twenty-seven somites). Osmium tetroxide after formol fixation.
a, Diagram of a small portion of the periphery, showing a clump of theca cells
still present at the periphery at the site of a fold in the follicular wall, and in-
dicating by small rectangles the location of the cell groups in band ec. 6, Cells
from persisting theca interna. XX 1000. c, Cells from interior of corpus luteum,
showing two granulosa lutein cells and a third cell exactly resembling those of
the theca. X 1000.
Wi GEORGE W. CORNER
The cells of the other type are those described in my con-

tribution of 1915 as “additional cells of the corpus luteum,
type 1.”’. They are found throughout the corpus scattered be-
tween the larger cells or in small clumps along blood-vessels
and connective-tissue septa. They are smaller than the granu-
losa lutein cells, having diameters of 154 to 20u. In form they
are adapted to their interstitial position, being rounded, almost
rectangular, or at times compressed into polyangular shape
(figs. 23 and 24). Their cytoplasm is either finely granular or
contains regular vacuoles so closely packed as to give a foamy
appearance. Indeed, in form, size, and in intracellular char-
acteristics they present a most striking resemblance to those
cells of the theca interna which in the same preparations are
still belatedly situated at the periphery of the corpus luteum
(fig. 22). Especially in osmium preparations is the similarity
so great that one is forced to the hypothesis that we have scat-
tered throughout the organ, among the granulosa lutein cells,
the multiplied and immigrated cells of the theca interna. In
the light of the apparent origin of these cells, it would seem
well to give them the name, already established in the litera-
ture, ‘theca lutein cells,’ for though there are certain just reasons
for criticism of this term and the name ‘granulosa lutein cells’
as applied to the other great class of corpus luteum elements,
there would seem to be no better names at hand.
While the cells derived from the granulosa lose their lipoids
after the first few weeks, the smaller cells just described again
gradually increase their content of osmium-staining lipoids
during the span of gestation, and some of them come at last to
be laden with these bodies, which, however, do not altogether
resemble the lipoid granules of their earler days (fig. 24). At
the end of pregnancy the cells of this type are still present
among those derived from the granulosa, apparently having
maintained separate identity during the entire term of gestation.
Even those which remain for a while in clumps or a definite
layer about. the periphery are not found to degenerate, but
seem to be drawn in among the neighboring granulosa cells as
the corpus grows older.
)
bo
24
Fig. 23 Cells of corpus luteum of advanced pregnancy (foetuses 230 mm.
long). Formol fixation. Mallory’s connective-tissue stain. X 810. GielaGas
granulosa lutein cells; th.l.c., theca lutein cells.
Fig. 24 Same stage as in figure 22, osmium tetroxide fixation without further
staining, showing distribution of fatty inclusions in cells. SO, GrtleGrn
granulosa lutein cells; th.l.c., theca lutein cells.
173
However, when an attempt is made to classify all the elements

of the fully formed corpus luteum, the picture is complicated
by the fact that numerous cells are found which are intermediate
in size between the two classes described above, and whose
cytoplasmic vacuoles and fatty inclusions are of nature too in-
different to place them definitely with either granulosa or theca
derivatives. The evidence, therefore, is not yet conclusive as to
the exact fate of all the theca lutein cells. Either the inter-
mediate forms represent genuine transitional stages in the for-
mation of ‘lutein cells’ from theca interna cells or else they are
merely cells, actually of one line or the other, in which the
all too sight distinguishing features of the type (size, form,
lipoid inclusions and vacuoles) have not been obvious. Toward
the latter view—the intermingling of the two cell lines without
actual conversion of one into the other—the author is inclined
to lean, without more positive evidence than has already been
given.
As a digression, it may be mentioned that some few of the
theca lutein cells retain the primitive characteristics of the
theca interna, even exaggerating them at times; they are vari-
able in shape and size, they have a cytoplasm which stains deeply
with acid stains, becoming dark blue, brown, or even orange with
Mallory’s triple stain and very dark with iron haematoxylin;
the cytoplasm is usually somewhat shrunken, and contains clear
vacuoles about ly to 24 in diameter, which are quite uniform
in size, are closely packed, and which either stain intensely black
with osmium tetroxide or remain as vacuoles. The nuclei are
often very dense, even pyknotic, and sometimes stain bright
orange with Mallory’s stain. The cells are often spindle-shaped,
branched, or compressed in such a way that they give the ap-
pearance of amoeboid motion, as if they were active wandering
cells. These are the cells described in my paper of 1915 as
“additional cells of the corpus luteum, type 2” (fig. 2). Al-
though their origin is now explained, I have no more light upon
their nature or possible function than before. They appear to
be more common in the earlier half of pregnancy, but their num-
ber varies greatly from animal to animal.
ORIGIN OF THE CORPUS LUTEUM es)
The cells of the young fully formed corpus luteum are sup-
ported by a reticulum of delicate connective-tissue fibrils with
denser strands along the septa. In sections stained by Mallory’s
anilin-blue mixture and by the Bielschowsky technique as modi-
gt. lic.
Fig. 25 Corpus luteum of pregnancy (embryos 20 mm. long). Formol fixa-

tion. Bielschowsky’s silver-impregnation method, showing reticular fibrils.
< 1000. gr.l.c., granulosa lutein cells; th.l.c., theca lutein cells; cap., capillary
blood-vessel containing an erythrocyte.
fied by Ferguson (711), it is clear that neither the granulosa

nor the theca lutein cells are intimately related to the fibrils,
which form dense baskets about them, but are not found within
the cytoplasm of the ‘lutein cells’ of either type (fig. 25). In

other words, the activities of the theca interna cells have become
so far modified in the process of their differentiation from the
primitive mesoblast of the ovarian stroma that they no longer
exercise the function of fibril formation. What new activities
they may have assumed, whether they may not now share in
producing the hypothetical internal secretions of the corpus
luteum, are questions for speculation and further research.
did Dl
e
Sayr
P Exe
cae
eS
Se
SS
RRR
Se
Pee
Se
mn
rs
eS
a
ae
(ug
‘Sx
es
Arte
“3sy
=
ss
Fig. 26 Diagrams of small portions of walls of corpora lutea at different

ages, showing increasing density of connective tissue. Dianol red X after
Bouin’s fluid. xX 80. 1, young corpus luteum (ova in tubes); 2, foetuses 115
mm. long; 3, seven days after birth of young.
Having found that the theca lutein cells are not. the fibril-
producing elements, one is further surprised to find that the corpus a
luteum of the sow contains very few fibroblasts of the conven-

tional type. It is impossibleto convince oneself that there are
in the organ any cells other than the two types of lutein cells
PAR
beside the endothelial cells of the capillary wall, except here and
there along the greater vessels which run in from the periphery.
Since there are multitudinous fibrils, and no ‘fibroblasts’ to pro-
duce them, one is forced to suspect that, as in the liver, the
endothelial cells themselves lay down the reticular fibrils. This
evidence by elimination would appear to be supported by
observation of the actual fact, but the question needs fuller
investigation.*
Whatever be their source, the amount and density of the reticu-
lar fibrils of the corpus luteum increases steadily from the na-
tivity of the organ until retrogression, when the organ is entirely
replaced by scar tissue (fig. 26). It is along the large septa
that the fibers first become dense, perhaps because a few con-
nective-tissue cells of the theca externa are often drawn into the
folds produced by the collapse following follicular rupture.
RETROGRESSION OF THE CORPUS LUTEUM
Owing to conditions of the meat-packing trade, it is difficult to

obtain ovaries of sows within a few days after parturition.
Study of the two specimens from the seventh and tenth days after
littering shows that regression of the corpus luteum is as rapid as
its appearance. By the seventh day the structure, which was
a flesh-colored body 10 or 11 mm. in diameter before parturition,
is only half this size; it has already shrunken until it is almost
buried in the ovarian stroma, and it has acquired a yellow-brown
color. By the time another ovulation occurs, the former corpora
lutea are dense sear-like nodules of connective tissue rendered a
pale yellowish brown by the presence of pigment in the shrunken
cells caught in the meshes of the scar. ‘The microscopic changes
during this process are obscure to me for lack of material, and
3 The results of a study of the question made since the completion of this
paper have proved that our hypothesis was correct, and that in the corpus luteum
and a number of other organs part or all of the reticular framework is laid down
by the cells of the capillary endothelium. (See a contribution by the author to
the forthcoming volume in memory of Dr. F. P. Mall in the Publications of the
Carnegie Institute of Washington, no. 272.)
therefore until a series including every day of the first week after
parturition can be obtained, the ultimate fate of the theca interna
cells of the sow must remain unknown.
DISCUSSION
It is hardly necessary to point out that the evidence just given

completely negates the results of previous investigators in this
species, and that it therefore removes one of the main supports
of the theea-origin theory. But since this view has been held
by so many anatomists, the reader will ask more than mere
denial; stating their errors, we must also explain them. The
reply, already pointed out by Sobotta, has become emphasized
to the present. writer during the course of his own studies. It
would be a simple matter to select from this material a number
of specimens which would duplicate the figures of Clark, Jan-
kowski, or Doering, but placed in proper position in the series,
the same specimens lead to far different interpretations. The
small, pale, and inconspicuous nature of the corpora lutea at
their earliest stages has allowed them to escape the notice of in-
vestigators seeking, under preconceived notions, for brightly
haemorrhagic structures in the ovaries, and thus the all-im-
portant stages of the first four or five days have not been at hand
to explain the more difficult later conditions of development.
Insufficient data, leading to confusion with the process of atresia
and to failure to obtain the earlier stages, and insufficient num-
bers of specimens, leaving important gaps to be filled by the
imagination, are the two great sources of error in the study of this
difficult problem.
In the investigation of the human ovary the obstacles to the
acquisition of attested material are still greater than in the
mammals subject to experimental methods. With our obscure
knowledge of the reproductive cycle, the only guide to the age
of a corpus luteum is its appearance, and we have seen what a
useless aid this can be. Even the presence of a healing point
of rupture, which is so characteristic of early corpora lutea in
other animals, is not entirely trustworthy in human ovaries.
Oa
.
MeBy
oO
©ai
The specimens have nearly all been obtained at operation upon

gynecological patients, who before operation are usually sub-
jected to palpatory examinations, often none too gentle. Any
gynecologist will know that the rupture of a ‘small ovarian
cyst’ by the examiner’s hands is a not infrequent occurrence;
such cysts, were they actually large follicles, immature or in
early atresia, if removed a day or two after the artificial rup-
ture, might present the anatomist with all too convincing speci-
mens of ‘early corpora lutea.’ Other possibilities of error might
be suggested, none of which can be ruled out until the ova are
studied with the specimen. While awaiting that almost impos-
sible outcome, we shall be wise to follow those workers with the
human corpus luteum whose findings are most nearly confirmed
by the evidence of comparative anatomy.
The interpretation of the origin and morphology of the corpus

luteum given in these pages does not represent a wide diver-
gence from previous views. Nearly all observers now agree in
describing the persistence of the membrana granulosa and its
_ invasion by elements arising in the theca interna. The present
work, so far as it traces the fate of the theca cells, is in accord
with the best of recent investigations, and the author’s hy-
pothesis of the persistence of all the theca cells as distinct ele-
ments would, if finally proved, explain the remaining difficulties.
It is not so easy to align the findings in the sow’s corpus luteum
with the conceptions of Sobotta and his followers, who believe
that the theca interna cells revert to mesoblastic type and by
division give rise to a strain of fibroblasts which lay down the
connective-tissue frame of the corpus luteum. This view implies
a notion of the structure of the adult corpus which differs from
that found in swine, and will necessitate a new study of the cell
types and the relation of the reticular fibrils to fixed cells in
corpora lutea of animals such as the mouse and sheep before
the discrepancy can be understood.
In conclusion, it is a pleasure to express my thanks to Professor
Evans for his encouragement and generous provision of aid and
materials for this work.
CONCLUSIONS
1. In swine the membrana granulosa is retained intact after

the rupture of the Graafian follicle. Its cells increase in size
without division, their cytoplasm becomes laden with lipoid
substances, and they become the larger elements commonly
called ‘lutein cells’ in the fully formed corpus luteum.
2. The membrana granulosa is invaded by blood-capillaries
from the theca interna, which ramify to form an extensive vas-
cular plexus throughout the new structure.
3. The large lipoid-laden cells of the theca interna are increased
in number by mitotic division, lose many or all of their fatty
inclusions, and pass into the corpus luteum to become lodged
between the granulosa cells throughout the whole structure.
4. There is no evidence that cells of the theca interna are ever
converted into fibroblasts of the usual spindle-cell type or that
they lay down the fibrils of the close-meshed reticulum which is
present in the corpus luteum.
5. There appears to be good evidence that some of the theca
interna cells persist throughout pregnancy as distinct elements of
the corpus luteum, but the exact fate of all of them cannot be
learned by present methods because of a confusing resemblance
between some of the theca and some of the granulosa derivatives.
BIBLIOGRAPHY
AssHETON, R. 1899 The development of the pig during the first ten days.
Quarterly Journal of Microscopical Science, vol. 41, p. 329.
von Barr, K. E. 1827 De ovi mammalium et hominis genesi. Leipzig.
Quoted by Bischoff (’42).
Barry, M. 1839 Researches in embryology. Philosophical Transactions of
the Royal Society of London, 1839, pt. 2, p. 307.
Benckiser, A. 1884 Zur Entwickelungsgeschichte des Corpus luteum. Arch.
tie (Givin ORY WS), shat).
Biscuorr, TH. L. W. 1842 Entwickelungsgeschichte des Kaninchen-Hies.
Braunschweig.
1878 Uber die Zeichen der Reife der Saiugethiereier. Arch. f. Anat.
u. Entwickelungsgeschichte, 1878, 8. 43.
Bonnet, R. 1891 Grundriss der Entwickelungsgeschichte der Haussiugethiere
Berlin. Quoted by Sobotta (’96). .
Biiuter A. 1900 Entwickelungsstadien menschlicher Corpora lutea. Verh. d.
Anat. Gesellschaft, XIV. Versammlung, Anatomischer Anzeiger,
Bd. 18, 8. 150.
«a
CuarK, J.G. 1898 Ursprung, Wachstum, und Ende des Corpus luteum. Arch.
f. Anat. u. Entwickelungsgeschichte, 1898, S. 95.
Conn, F. 1903 Zur Histologie und. Histogenese des Corpus luteum und des
interstitiellen Ovarialgewebes. Arch. f. Mikr. Anat., Bd. 62, S. 745.
1909 Uber das Corpus luteum und den atretischen Follikel des
Menschen und deren cystischen Derivate. Arch. f. Gyn., Bd.87, 5.367.
Corner, G. W. 1915 The corpus luteum of pregnancy, as it is inswine. Pub-
lication no. 222 (Contributions to Embryology no. 5), of the Carnegie
Institution of Washington, p. 69.
1917 a On the lipoidal nature of structures in the corpus luteum cells
of swine. Proceedings of the American Association of Anatomists,
Anat. Ree., vol. 11, p. 345.
1917 b Variation in the amount of phosphatides in the corpus luteum
of the sow during pregnancy. Journal of Biological Chemistry,vol.
29, p. 141.
1917 ec Maturation of the ovum in swine. Anat. Rec, vol. 13, p. 109.
Corner, G. W., anp AmMspaAueaH, A. E. 1917 Oecestrus and ovulation in swine.
Anat. Rec., vol. 12, p. 287.
CristTaLui, G. 1903 Contributo alla istogenesi del corpo luteo della donna.
Arch. d. Ost. e Gin., T. 10, p. 480.
Deestre, M. 1910 Recherches sur le follicule de de Graaf et le corps jaune
de la vache. Jour. de l’Anat. et de Physiol., T. 46, p. 286.
Doérine, H. 1899 Beitrag zur Streitfrage iiber die Bildung des Corpus luteum.
KG6nigsberg, dissertation.
Ferauson, J. S. 1911 The application of the silver impregnation method of
Bielschowsky to reticular and other connective tissues. Am. Jour.
Anat., vol. 12; p. 277.
Fiemmine, W. 1885 Uber die Bildung von Richtungsfiguren in Siugethiereiern
beim Untergang Graaf’scher Follikel. Arch. f. Anat. u. Entwick.,
1885, 8. 221.
Heear, K. 1910 Studien zur Histogenese des Corpus luteum und seiner Riick-
bildungsproducte. Arch. f. Gyn., Bd. 91, S. 530.
His, W. 1865 Beobachtungen iiber den Bau des Siugethier-Hierstockes. Arch.
f. Mikr. Anat., Bd. 1, S. 151.
Honort, Cu. 1900 Sur la formation du corps jaune. Archives de Biologie,
T. 16, p. 563.
JaNkowskI, T. 1904 Beitrag zur Entstehung des Corpus luteum der Siugethiere.
Arch. f. Mikr. Anat., Bd. 64, 8. 361.
Karppett, I. 1908 Beitrige zur Anatomie und Physiologie der Ovarien von
wildlebenden Wiederkiuern und Schweinen. Bern, dissertation.
Keiser, F. 1897 Normentafeln zur Entwicklungsgeschichte des Schweines.
Jena.
Kopscn, F. 1901 Corpora lutea von Schwein und zwar vom 3, 6, und 10 Tage
nach der Befruchtung. (Priparate des Herrn Stabsartz Dr. Menzer.)
Demonstration. Verhandlungen Anat. Gesellsch., XV. Versamm-
lung, Bonn. Anatomischer Anzeiger, Bd. 19, S. 210.
Krets, L. 1899 Die Entwickelung und Riickbildung des Corpus luteum spurium
beim Menschen. Arch. f. Gyn., Bd. 58, 8. 411.
Lors, L. 1906 The formation of the corpus luteum in the guinea-pig. Journal
of the American Medical Association, vol. 46, p. 416. ;
Marsuatu, F. H. A. 1904 The oestrous cycle and the formation of the corpus
luteum in the sheep. Phil. Trans. Roy. Soc., vol. 196 B, p. 47.
Meyer, R. 1911a Uber Corpus luteum-Bildung beim Menschen. Arch. f.
Gyn., Bd. 93, S. 354.
1911-b Zur Corpus luteum-Bildung beim Menschen. Zentralbl. f.
Gyn., Bd. 35, S. 1206.
Minuer, J. W. 1911 Uber Corpus-luteum-Bildung beim Menschen. Kritische
Bemerkungen zur der im Archiy fiir Gynaekologie, Bd. 93, Hft. 2
erschienenen Arbeit Prof. Dr. Meyer’s. Zentratbl. f. Gyn., Bd. 35, S.
1089.
Niskoupina, N. 1909 Recherches sur la morphologie et la fonction du corps
jaune de la grossesse. Nancy, thesis.
Novak, E. 1916 The corpus luteum:—lIts life cycle and its rdle in the men-
strual disorders. Journal of the American Medical Association, vol.
67,, p. 1285.
O’Donoauur, C. 1912 The corpus luteum in the non-pregnant Dasyurus
and polyovular follicles in Dasyurus. Anat. Anz., Bd. 41, S. 353.
1914 Uber die Corpora lutea bei einigen Beuteltieren. Arch. f. Mikr.
Anat., Bd. 84, Abt. 2, S. 1.
1916 On _ the corpora lutea and interstitial tissue in the marsupialia.
Quarterly Journal of Microscopical Science, vol. 61, p. 433.
PatapiIno, G. 1879 Studio sulla fisiologia dell’ovaja. Struttura, genesi, e
significazione del corpo luteo. Napoli.
1880 Della emoragia e del comportarsi della granulosa allo scoppio
dei follicoli di Graaf. Napoli.
1881 Consequenze dello scoppio dei follicoli di Graaf ed in partico-
lare del corpo luteo della donna. Napoli.
1887 Ulteriori richerche sulla distruzione e rinnovamento continuo
del parenchima ovarico nei mammiferi. Napoli.
1900 Per la dibattuta questione sulla essenza del corpo luteo, Anat.
Anzeiger, Bd. 17, S. 451.
1905 Lamitose dans le corps jaune et les récentes conjectures sur la
signification de cette formation. Arch. Ital. de Biol., T 43, p. 292.
Priiicer, E. 1863 Uber die Eierstécke der Siiuger und des Menschen. Leipzig.
Quoted by Sobotta (96).
Porter, M. 1910 Contribution A l’étude anatomique, histologique et physio-
logique du corps jaune pendant la grossesse (Plus especiallement chez
la femme). Thesis, Paris.
Rast, H. 1898 Beitrag zur Histologie des Hierstocks des Menschen und der
Siugethiere, nebst Bemerkungen iiber die Bildung von Hyalin und
Pigment. Anatomische Hefte, Bd. 11, S. 109.
Ricker, G., AND DAHLMANN, A. 1912 Beitriige zur Physiologie des Weibes.
Volkmann’s Sammlung klinischer Vortriige. Neue Folge, no. 645/647.
Sanpes, I’. P. 1903 The corpus luteum of Dasyurus viverrinus, with observa-
tions on the growth and atrophy of the Graafian follicle. Proceedings
of the Linnean Society of New South Wales, vol. 28. Quoted by
O’ Pea (16).
ScuroepEr, R. 1914 Uber die zeitlichen Beziehungen der Ovulation und

Menstruation. Arch, f. Gyn., Bd. 101, 8. 1.
Sozorra, J. 1895 Uber die Bildung des Corpus luteum bei der Maus. Anatom-
ischer Anzeiger, Bd. 10, S. 482.
1896 Uber die Bildung des Corpus luteum bei der Maus. Arch. f.
Mikr. Anat., Bd. 47, S. 261.
1897 Uber die Bildung des Corpus luteum beim Kaninchen. Anat.
Hefte, Bd. 8, 8. 469.
1899 a Uber die Entstehung des Corpus luteum der Siiugetiere. Ref-
erat, Ergeb. d. Anat. u. Entwick., Bd. 8, 8. 923.
1899 b Noch einmal zur Frage der Bildung des Corpus luteum.
Arch. f. Mikr. Anat., Bd. 53, S. 546.
1902 Uber die Entstehung des Corpus luteum der Saugetiere. Ref-
erat, Ergeb. d. Anat. u. Entwick., Bd. 11, S. 946.
1907 Uber die Bildung des Corpus luteum beim Meerschweinchen.
Anat. Hefte, Bd. 32, S. 89.
SpPrEGELBERG, O. 1865 Uber die Bildung und Bedeutung des gelben Koérpers
im Hierstocke. Monatschrift f. Geburtskunde, Bd. 26, S. 7.
Straxoscu, W. 1915 Das Schicksal der Follikelsprungstelle. Arch. F. Gyn.,
Bd. 104, S. 259.
Stratz, C. 1898 Der geschlechtsreife Siugethiereierstock. The Hague.
VAN DER Srricut, O. 1901 a La rupture du follicule ovarique et ’histogenése
du corps jaune. Comptes Rendus de 1|’Association des Anatomistes,
3™° Sess., p. 33.
1901 b La ponte ovarique et l’histogenése du corps jaune. Bulletin
de l’Acad. Roy. de Medécine de Belgique, IV° Serie, T. 15, p. 216.
1908 La structure de l’oeuf de chienne et la genése du corps jaune.
Comptes Rendus de 1’Association des Anatomistes, 10° Sess. p. 1.
1912 Sur le processus de l’excrétion des glandes endocrines: le corps
jaune et la glande interstitielle de l’ovaire. Arch. de Biol., T. 27,
p. 585.
TracHer, J. H. 1908 in Teacher, Bryce, and Kerr, Contributions to the study
of the early development and imbedding of the human ovum, Glasgow,
p. 85.
Timorerev, A.I. 1918 Development of the corpus luteum of the human ovary.
_ Dissertation, Kasan. (in Russian).
Votxer, O. 1905 Uber die Histogenese des Corpus luteum beim Zeisel (Sper-
mophilus citellus). Arch. f.:Anat. u. Entwick., 1905, S. 301.
WaupryerR, W. 1870 HEierstock und Ei. Leipzig.
Wauuart, G. 1914 Uber Friihstadien und Abortivformen der Corpus luteum-
Bildung. Arch. f. Gyn., Bd. 103, S. 544.
Weysse, A. W. 1894 On the blastodermic vesicle of Sus scrofa domesticus.
Proc. Amer. Acad. of Arts and Sciences, vol. 30, p. 283.
Wiuurams, J. WuiTripcEe 1917 Obstetrics. New York. 4th ed.
Woz, E. 1912 Untersuchungen zur Morphologie der interstitiellen Eierstocks-
driise des Menschen. Arch. f. Gyn., Bd. 97, S. 131.
Zwicky, H. L. 1844 De corporum luteorum origine atque transformatione.
Zurich, Inaug. Dissertation.
af
/ yi (er ft
Af 7 p vite bY ;
4 , i> i
i i helt rs ie |ae
i " vie A. 12
ti go vie rein
ANTS ne ay . F fer i cry } Yel Ganga | ie runes vid 4

P ; . 4
. : | ¥
bh : » 6.4)
iat AL 1)
r Dae vee { PA ry teal

‘
js ¥
A rf ) l
h i Thus ; ay {
- ai :; f ; 7i be| ¥4 4 ‘ vw
16
ips aed P F ¥ ver
y | i the, with ye
‘ . ! )
ae a as : iy heehee et hie 8 ! 5 SiMe
Meta titsl Pe waPh lata ap) SUAS? GPP PARES ia
! ~ Pith Cob pe Phe Wes APR ee. bk Ny hg Kean
1 ; ru 4 t ' +f /
nt j
iy iJ Psy deal kt eae ae CATT ( he RATES | 47 ty Nate Pht
is Ven OR MC e ese
reid { ih - ake La \ een ee Ob!ie aye cf rd) re r sree Bly Aiea a ite ‘i
ion
b> UMAR i054bay Pe Sa Hd I a fh DOR nr tsvvaii
‘
i. ™ ‘ ge Od Alle Rave g o> idee ee fs Riya habth aay any io dpa a Pe a

f ‘ . i i edie | F } “a
“ 4 hs ’ ALY 1 Fi \ h, | A} aay
. f Pirie
i fe Tile ts CORUM
Pak t 1 ‘ ee MY ‘ Leute Ole aa tis Wd POA el
vy PH ely ail, a ie Avi vdfe. wate eRe
>, an %
i al pay 1's Weer di wid eh ‘

' pr di tens Pho Ala ogee} on Lie)!lit veghal sitiny Re
vey Sail ink t ‘nif jai)veeSth *) uit hioh ier VERDI i AV) paidh: awh ; ke
Ale Eats aes ae wbitulse bk! aia bdr ae Bia 1A a

Lame ' Cn Aad bh bettyki PO i cea
b9 Ui? ' ij } »
yi, Ma! aye
vid '
nl Gti aman cerlh wiht in Ne aingiPticeay shes 8) TOO
7 1 Mae ‘ An f ntVaR end! nA “he Ores!fe ety Qube: Aah ys ih axed hea
Aig, } BS HLS
ui | rm Paes Ge ; ale haby ie ssLin i of Miikeyh dei ne ait TR i Lay 1 wh, La Les ee? rt} re ;
oy Ao ere x Ti Te eee sas Js LMR Sua YAS roomed atinanhig begat {igi ad
ie ae ret Pra i, SHEA b WV)Bar i nt LS beled

ft yes BL
ine ated we uc) whan latent nikea fat ppl \grto(oilale Scam
Fab in by! dee liad, Fr Oh. ok bo ay Ls
"ea
ee Uy ne ors it Prahianealmeik
hah vc Ahi mala
“eis
an ae ated
yeaa
ih pt
a i id et a cayheat Tire “hindi
i Sines Hd:
Rohe )
fact 4 aa
NUH Bee Marri CPaNERE
ARMM, MERI Ce Ee
we pail adr bs) ithews Debi se Pas)
: (ie deena
uia
mr, a, oer ae ' PE: ee ‘a
roe
WY igs ws tu Va a wp
Uae
fa’
ye el Lat cae
a pragy hy
AUTHOR’S ABSTRACT OF THIS PAPER ISSUED BY
THE BIBLIOGRAPHIC SERVICE, SEPTEMBER 29
PH, HYPOPHYSIS. CEREBRL, OF, THE. CALIFORNIA

GROUND-SQUIRREL, CITELLUS BEECHYI
(RICHARDSON)
HAROLD J. COOPER
Division of Anatomy, Stanford University Medical School
ELEVEN FIGURES
CONTENTS
MAG ROCUCELOMA ELA eh. Pcttetcist ts wee Eee Lee oo, Seve) GE 2 ae 185
Materia leamcdsmeGhod sytecrye ek sikrs ce Gcietnlne SePeOe Da en Ghete Ok cud ye 189
GrosspieatURes tion. danlia. sissies BORNE Mae ee Ligon ono an Oatae Le Meter aes abe 190
IMINCTOSCOMICRLE VUUNES ote eye ens ae cosets so Aes oR a eens evince tthe oes oe 192
IPAarstbUCcaliga sasha ea eeben tent Ct REE any ROW ede ee her NSE keen Sh LOS
IPAS ROLISGallESve ePSiyeaeee eos Les ate ey ge eeaed wine a hia mar Baa Ss ees lous 193
RarspiUxtasncunralisn 1. Saccityicice ta daa macwaeds Wohroe sey is sls SatTS chet 198
PAGS SIMU DU ATS 55.1 cee eco ane aero eras) une ts 198
IRATE UMDETAIT Sheree he eco ee RE eS Re ere eRe OeAnne SRI RIA Te 200
RATSRMC UA SHI AEE TaN NN he oie RE EEES ih dV RA Sehr ie Utne hae 200
CLMOTO OS as Sas Sam ce ote ae Oe ere ROO See ae cers Ener Eee eee emer ne 201
SSNARIATET UPR Oe EN Boy U SON Io 5).<cla.ivetan'd cae.s,o's SS apa esAIMS sinske eTe Sitcm suagalts Erase dhche, oye 203
[ELS
G Gry Rt A aretha hel le tk MR RRR RLors hd i i SS an abe tia teat 203
_. Despite the prevailing interest in the structure and functions

of the hypophysis, but little attention has been paid to the ro-
dents, and none to the common ground-squirrel, so far as I have
been able to ascertain. Stendell (’14) gives a brief account of
the more salient features of the rodent hypophysis. Mann (’16),
working on the ductless glands of the spermophile and their
possible relation to hibernation, briefly describes the changes
which the hypophysis undergoes during these periods. ‘The
more detailed studies have seemingly been carried out on man
and the higher mammals.
Using the terminology of Tilney (’11) and (’13), the adult
mammalian hypophysis is generally recognized to-day as being
made up of a buccal and neural portion whose subdivisions may
be tabulated as follows:
185
186 HAROLD J. COOPER
I. Pars bucealis (glandularis)

1. Pars distalis
2. Pars juxta-neuralis
a) Pars infundibularis
b) Pars tuberalis
II. Pars neuralis
1. Eminentia saccularis
2. Infundibulum
3. Processus infundibuli
Stendell (713) describes with accompanying illustrations, in
many of the lower forms the relationship existing between the
different parts of the gland (glandular and neural). Sterzi and
Gentes, as cited by Stendell, called the pars distalis the chrom-
ophilic and the juxtaneural portion the chromophobic, on
account of the characteristic staining reactions. Tilney (’11)
and (’13) pictures the parts in the human; Herring (’08), in the
cat, dog, and monkey, each of which possesses its own peculiarity.
Swale Vincent (712) speaks of three types of hypophyses
among the mammalia according to the degree of investment
of the nervous lobe by epithelium and according to the extent to
which the nervous lobe is invaded by the cavity of the third ven-
tricle. The first type (cat) has a hollow nervous lobe, the cavity
communicating freely with that of the third ventricle. The
epithelial investment is very complete. The second type (dog)
has a solid nervous lobe, but a patent infundibulum. Here, too,
there is a very complete investment. In the third type (man)
both the nervous lobe and infundibulum are solid, although
traces of a lumen in the latter may exist. The epithelial
investment here is not so complete.
The classification of the cells in the pars glandularis is volumi-
nously discussed in the older papers. Flesch (’84) classified
them as chromophile and chromophobe. Dostoiewsky (’86)
endeavored to establish a relationship between the granular and
non-granular cells in regard to their distribution. In ‘the ox
and in man he found that the alveoli lying more peripherally
were made up of a greater proportion of granular cells than those
lying centrally. He also found that in these forms, the cells
ate
HYPOPHYSIS CEREBRI OF CALIFORNIA GROUND-SQUIRREL 187
lying peripherally in the alveoli themselves were more frequently

of the granular type. In the smaller animals, as a class, he found
that the more even distribution of the two cell types made any
such localization of little worth. Lothringer (’86) found a
peripheral arrangement of the chromophile elements in the horse,
dog and in man similar to that described by Dostoiewsky. Rogo-
witsch (’89) described the so-called ‘Kernhaufen’ in the pars
distalis. In this tissue the cell borders are fused and indistinct
and appear much like embryonic connective tissue. Schonemann
(92) describes the chromophobes as having ill-defined borders
and a diffusely granular cytoplasm. Scaffidi (04), using orange
G and acid fuchsin describes ‘orange G staining cells’ and ‘fuch-
sinophile cells,’ to each of which he attributes an independent
secretion. Trautmann (’09) classified them as chromophobic,
weakly chromophilic, and strongly chromophilic, the latter two
types being either acidophile or basophile.
Considerable disagreement exists among the older writers
concerning the functional individuality of the various cell types.
Benda (’00) believes them to be of one and the same cell type
in different phases of functional activity. According to him,
the chromophobic cells, by the accumulation of granular matter,
become chromophilic,-and these, after the elimination of their
secretory matter, in turn, chromophobic. St. Remy (92),
Herring (709), Stendell, and others are of the same general
opinion with slight modifications. Guerrini (’05) and Scaffidi
(04) believe in the existence of different functional types. The
sum total of evidence from a number of papers seems to leave
the exponents of the latter view quite in the minority. The
relative changes in the cell types in pathological conditions
in man and animals (hypophysectomy, thyroidectomy, acro-
megaly, gigantism, etc.) has led to much speculation as to the
- possible individuality of them. Biedl (’13) gives a general
review of these papers.
The pars infundibularis is characterized by the greater abun-
dance of basophile cells which it contains. Herring states that
there are a few deeply staining cells, but that they never contain
eosinophile granules.
The pars neuralis in the mammalia is made up chiefly of

ependyma and glia cells, among which run many fibers. Miiller
(71) regards this portion as one reduced in structure to a con-
nective-tissue appendage of the brain. Berkley (’94) describes
an elaborate arrangement of nervous elements, but his paper
stands quite without substantiation from the later writers. It
is not generally considered that there are any true nerve cells in
the nervous lobe. Nerve fibers exist in great abundance as
described by Herring and others. Clunet and Jonnesco (’10)
studied in detail the pigment of this lobe, which occurs especially
in the higher mammals.
The existence of colloid matter in the various portions of the
gland is given mention in many references, but the best accounts
are to be found in the works of Herring, Stendell, and Biedl.
Steida (90) states that the colloid in the pars distalis is found
within the chromophile cells. In the pars infundibularis it
may be in the cells or gathered into cysts. Benda believes the
colloid in the tissue to be indicative of degeneration. Guerrini
regards it as a normal product of secretion. Schonemann,
Herring, and Trautmann describe colloid masses with cell
remains and granules. The transitional part between the distal
and infundibular lobes, which goes to make up such a distinct
portion (Ubergangsteil) in many of the lower forms (Stendell),
is not present in the mammals as a strikingly different histological
region. In it, however, much colloid can usually be found, in
some forms colloid being present there when it is absent elsewhere
in the gland. Cushing (’12) calls attention to the more constant
basophilic character of the colloid in the pars juxtaneuralis,
while after passing into the pars neuralis it forms acidophile
hyaline masses. Herring states that colloid in the cleft is rare
and that it is never found very far distant from the nervous lobe.
Stendell describes secretory matter in the lumen and suggests
that it may represent a reserve supply.
Peremeschko (’67) calls attention to cilia which line the resid-
ual lumen in man. Lothringer (’86) and Stieda (’90) describe
them lining the cysts. Vanderburgh (17), in guinea-pigs,
found unmistakable patches of cilia lining the residual lumen
and in some cases being continuous with the lining of the cysts.
On the other hand, Herring states that there are none in the cat,
and Stendell regards their presence as extremely improbable.
Comparatively little attention has been paid to mitoses, in
the adult gland at least, in which, to be sure, they are rarely
seen. Jackson (717), working on albino rats during inanition,
gives an account of the disappearance and reappearance of mitoses
during and after this condition.
Numerous slight differences have been described in the nuclei,
but they seem to possess no radical variations from the usual
forms in similar organs and tissues. Stendell calls attention to
the possibility that the irregularity of outline in the nuclei,
found occasionally in the deeply chromophilic cells, may be due
to intercellular pressure.
The striking lack of information upon the Sciuridae in the

literature on the hypophysis led me to make some studies of this
abundant material, and results secured seem to warrant publica-
tion in their present form. The common ground-squirrel,
Citellus beechyi (Richardson), was used. The animals were
obtained by shooting. They were at once dissected carefully,
the hypophyses removed and placed in the fixative, care being
taken to reject any of those that showed any damage to the skull
or brain in the region of the organ. As fixatives, saturated cor-
rosive sublimate in 70 per cent alcohol, Zenkerformol, 10 per
cent formalin, strong Flemming’s solution, Bouin’s fluid, and 100
per cent alcohol were used. Delafield’s hematoxylin was used
throughout as a nuclear stain. Orange G and acid fuchsin
mixture (Scaffidi), chromotrope 2R and 2B were used as counter-
stains. Alcoholic eosin was not very satisfactory. Mallory’s
connective-tissue stain was also used, while methyl violet,
methylene blue, Wright’s blood stain, thionin, and other basic
and mixed stains were found invaluable.
Sections were cut in celloidin and paraffin, the latter in series
of 5y. The former method was employed when large portions of
brain and bone tissue were to be included in the sections.
A. GROSS FEATURES
The general topography of the ground-squirrel hypophysis is

very similar to that of other mammals (figs. 1 and 2). Using
Tilney’s terminology, it has a pars buccalis (PG), derived from
the ectoderm of the buccal cavity (but possibly also including a
portion from the pharyngeal entoderm), and a pars neuralis
(PN), derived from the nervous system. These two portions lie
B 3°37
c ro PSR
cn. ann ee
2 ‘S_7 Nee
;
. i
iNOS
epee a »)
GAS Chee Ly (7
Be wet LE ivf ya
ees ae“
eZ
OF i(SSS.
eest
eeae
"Sa LLP ee
Pa se S m4
LP. ieee oc
=
Pale
Fig. 1 Midsagittal section through the ground-squirrel hypophysis (semi-

diagrammatic). B, brain; 3rdV, third ventricle; SB, sphenoid bone; PN, pars
neuralis; RL, residual lumen; PI, PI’, PI?, pars infundibularis; PG, pars distalis;
I, infundibulum; JC, infundibular cavity; D, dura; OC, optic chiasma; PT’, pars
tuberalis. X 25.
in contact. The original cavity of the pars buccalis (residual

lumen (RL)) is maintained in the adult form, and serves to sep-
arate the pars buccalis into two divisions, structurally, and
probably also functionally distinct, namely, the pars juxta-
neuralis (PI, PI’, PI*®, PT), and the pars distalis. The pars
juxtaneuralis is divided into a pars infundibularis (PJ, PI’, PI’),
investing the infundibulum and infundibular process, and a pars
tuberalis (P7'), extending over the tuber cinereum. Since the
pars infundibularis and the pars distalis are not entirely separated,
a transition zone is found to exist at the periphery of the lumen
where the one portion merges into the other.
The hypophysis of the ground-squirrel (fig. 1) measures about
2 mm. in the anteroposterior dimension and 2.5 mm. trans-
versely. The dimensions vary somewhat in a series of measure-
ments, but the ratio of length to breadth is quite constant. The
organ lies in a shallow sella turcica, to which it is firmly bound by
Fig. 2 Section somewhat lateral to fig. 1 (semi-diagrammatic). B, brain;

SB, sphenoid bone; PN; pars neuralis; PJ, pars infundibularis; RL, residual
lumen; PG, pars distalis; D, dura. X 25.
the meninges. Considerable care has to be used to avoid break-

ing or tearing the infundibulum during removal. The infun-
dibulum lies amost horizontally in the anteroposterior direction,
and is bordered on either side, as in all the higher forms, by the
corresponding optic tract. The glandular portion is a deep blood
red and the nervous portion a glistening white in the fresh con-
dition. The glandular portion partially envelops the pars
neuralis and extends further posteriorly at its lateral margins
than elsewhere, thus forming a concave hinder surface (figs. 1 and
2). The organ belongs to the second class in Vincent’s (12)

grouping, possessing a solid nervous lobe, but an infundibular
cavity extending almost to it.
PN PG
Fig. 3 Midsagittal section of the hypophysis of a 12-mm. embryo (semi-

diagrammatic). J, infundibular cavity; PN, pars neuralis; PJ, future pars
infundibularis; R, residual lumen; PG, future pars distalis; S, epithelial stalk;
OC, optic chiasma; BC, buccal cavity. X 35.
B. MICROSCOPIC
Since I have had but little embryological material, I can give

no details of the development. The only stage I have seen is
found in a 12-mm. embryo, a midsagittal section being shown in
figure 3. The hypophyseal pouch (Rathke’s pouch) is just about
to be cut off at S. At this stage there has taken place no dif-
ferentiation of the glandular portion into distal and juxtaneural
lobes.
I. Pars buccalis
1. Pars distalis. The simple classification of the cells of this

part into chromophile and chromophobe is not very evident in
the ground-squirrel. ‘Transition forms exist in such great num-
bers that one kind is not quickly distinguishable from another.
There are, however, two extreme types, one deeply granular
and the other either slightly so or not at all.
The deeply chromophile cells are characterized by their larger
size and less regular contour. On section they may be round,
triangular, square, or polyhedral, in fact, with sufficient search
almost any shape may be encountered. The nuclei are round or
oval and contain a rich chromatin network. The cytoplasmic
granules are coarse and usually irregular in shape. The chro-
mophobe cells are characterized by a round, generally smaller,
nucleus. The cytoplasm is more or less free from granulation
and in many cases consists of only fine protoplasmic strands
with no limiting membrane. In this condition they possess no
acidophilic properties, the fibers staining a faint tinge with
hematoxylin.
The ground-squirrel hypophysis shows a number of cells vary-
ing in granulation from those with none or a few granules to those
in which the cytoplasm is practically one homogeneous acidophile
mass. The granules of the weakly chromophilic cells are usually
small and well separated, the intervening cytoplasm being very
weakly acidophilic (fig. 8, a and b). In other cells the granules
are larger but fewer in number, presumably due to the coalescence
of numerous smaller granules. In the deep acidophiles (fig. 8,
c and d) the cytoplasm may show little or no granulation, the cell
contents apparently consisting of a mass so densely packed that
the ultimate granular nature of it is obscured.
The distribution of the cells of either extreme of granulation
seems to be very irregular, no particular area being recognizable
in which either form is more common, for a cell of one extreme
may lie adjacent to one of the other (fig. 5, a and 6b), or be sur-
rounded by cells apparently intermediate between the two
extremes (fig. 7). Nor does any definite arrangement of the cells
obtain in the alveoli, they being about equally distributed. This

is just contrary to what is to be seen in the human hypophysis,
in which the elements are usually quite distinct, the chromo-
philes occupying the periphery of the alveolus, but coincides
more with Dostoiewsky’s statement that such an arrangement
is not to be observed, as arule, in small animals. The chromo-
philes may be either acidophile or basophile, and the existence
of a peripheral granular zone, described by some writers, is not
to be found in this gland. The chromophile cells in the pars
distalis are acidophilic, as a rule, when filled with granules, while
in the pars juxtaneuralis the cells are exclusively basophile.
The cells of the pars distalis bordering on the residual lumen
are somewhat pyramidal (triangular in section) and, filling up
the angular spaces between the oval and polyhedral cells imme-
diately beneath them, they present a smooth and regular free
surface. Some of the spaces referred to are filled in by irregular
cells. These have round nuclei, while those of the pyramidal
cells are oval. The chromatin in their nuclei, as seen with thionin
stain, appears to be more diffuse than in the others, in which it
is often seen massed in one or more clumps. The nuclei of the
border cells, on the whole, stain deeply and are very numerous,
appearing, especially under low magnification, as one continuous
row. The shape of these cells seems to conform to the arrange-
ment of the underlying layers. Quite often the border cells give
the appearance of a continuous layer of low cuboidal or flattened
cells (fig.4, A). The cells are not, however, of thisnature. This
appearance of the border cells is best seen in the peripheral por-
tions of the cleft. The cells stain deeply and can easily be dis-
tinguished from the paler underlying ones. They are joined end
to end, but a distinct peripheral border is discernible for each of
them. The nuclei stain deeply and the cytoplasm takes on a
dark tint in ordinary staining. In other regions along the cleft
the cells appear wedged in between the gland cells and show the
arrangement in figure 4, B.
The cells vary considerably in size, ranging from at to 6u
in diameter among the chromophobes having a measurable
periphery, to 64 to 124 among the chromophiles. As a rule, the
richer in acidophile matter the larger the cell, although this is

not accurately determinable from the fact that many of the
poorly granular cells have ill-defined margins. The nuclei of the
chromophobes usually lie centrally. . From the fact that the
nuclei of these cells seem entirely intact under practically all
conditions and show no signs of nuclear degeneration, it is prob-
able that they are not degenerated, as some have suggested, but
that they are in the early stages of accumulating granular matter.
The nuclei of the chromophiles usually lie eccentrically, being
located in that part of the cells most distant from the vascular
sinuses upon which they border.
Fig. 4 (a) From the pars distalis showing the border-cells on the lumen.
x 1200. (b) From the pars infundibularis showing the border cells with their
extensions into the gland substance. In both (a) and (b) the border cells appear
darkened. X 1200.
Figure 5 gives an example of the two extremes—one markedly

chromophilic, a, and the other chromophobiec,b. The cell outlines
of the chromophobes are not usually so distinct as shown in b.
While the two cell types here figured were adjacent in the prepa-
ration, the more common arrangement is to find various inter-
mediate forms commingled with these, figure 7 showing such a
group in one of the alveoli. It is important to bear in mind that,
while the two extremes are less sharply set apart, there are,
nevertheless, just as strongly chromophilic and chromophobic
cells in the ground-squirrel hypophysis as in other forms.
When stained withorangeG and acid fuchsin, according to Scaf-
fidi’s technic, there were seen in sections, cells staining deeply with
the orange G. No special features were observed in these except

that they were perhaps slightly smaller than the cells staining
with the acid fuchsin. In these cells, as shown in figure 6, the
ground substance of the cytoplasm stains a clear light orange and
contains a great number of irregular granules which take a deep
orange color.
Scaffidi, on the basis of this method of staining sections from
the human gland, differentiates between what he considered to
be two different functional types. Those taking up the orange G,
he states, are round, and their golden granules small, fine, and
glistening. They have a small oval nucleus with large chromatin
granules. The ‘fuchsinophile’ cells are described as larger and
more irregular and their granules irregular and coarse. The
nuclei are larger and generally rounded. In them occur small
blocks staining deeply with the fuchsin. He mentions nothing
about these blocks in detail except to say that they are never
found in the ‘orange G staining cells.’ These blocks I was unable
to find in the squirrel gland or in the human.
In order to understand the distinction made on this basis,
sections of human hypophysis and adrenal, as well as the same
tissues from the ground-squirrel, were first stained with orange
G. Certain groups of cells were located with a mechanical stage,
studied and drawn. The preparations were then decolorized
and restained with acid fuchsin. In the human hypophysis, it
was found, practically without exception, that those cells which
had at first stained with the orange G, became restained with the
fuchsin, while those remaining colorless at the first staining, con-
tinued so during the second. The same was true without excep-
tion with the squirrel hypophysis and both adrenals. Besides
this set of stains, eosin and fuchsin, eosin and chromotrope 2B,
and fuchsin and chromotrope 2B were used with the same results.
These groups of stains, when used together (in a mixture) can
be distinguished by differences in shades, and when so used, it
was seen that they, too, attacked the cells somewhat differently.
While there is a specificity shown by different cells toward one
or the other of the stains in such a mixture, it does not seem to
offer a very sound foundation alone for a division of the cells into
functional individual cell types.
the
Vascularity. The pars distalis is pervaded throughout by nu-

merous blood sinuses. They are of varying width, and owing to
their elaborate distribution they come into close relation to the
cells. By virtue of their tortuous course, no long channels are to
be seen except in very thick celloidin sections, and in the usual
paraffin preparations they appear as irregular areas containing
blood. The cells of the pars distalis le in strands or cords two
or three layers thick. The spaces separating them being occupied
by blood, it is evident that no given cell can be much more than
its own diameter distant from arich blood supply. The channels
are lined by a well-marked vascular endothelium which inter-
venes between the blood and the cells (fig. 11). The arrangement
offers strong probability of a direct secretion into the blood
channels, although it is possible that the secretion is given off
into the tissue spaces and taken up by a perivascular lymphatic
system as suggested by Edinger (’11). Thom (01) conceives of
a mixed secretion from the chromophile and chromophobe
elements as passing into interfollicular lymph spaces.
Surely, the vascularity of the organ is of considerable signifi-
cance, and it very likely plays an important part in the taking
up of the secretory products of the cells. In the sections which
I have observed in the ground-squirrel there is abundant histo-
logical evidence that the cells secrete into the blood stream. In
the region of a sinus, cells in the above-described granular stages
will be seen along its margins. When the blood sinus is cut in
cross-section the cells are seen to radiate from it in much the same
way as other gland cells are seen to be arranged about their
secreting tubules. In the deep chromophiles, which many
writers consider to be cells ready to discharge their secretion,
marked irregularities in outline may be found. These irregu-
larities are always found at that end of the cell bordering on the
sinus and separated from it only by the endothelium. In this
region of the cell the granules may be seen in clumps outside the
cell or at least exterior to the former probable limits of it (figure
11a). Under these conditions the cytoplasm at the irregular end
may be very light in granulation as compared with adjacent
heavily granular and smoothly, outlined cells. In addition to this
the basal end of the cell may be still quite filled with its acido-
philic matter, while in other cells the whole cytoplasm is free from
granulation and reduced to.a pale meshwork, simulating charac-
teristic chromophobice cells (figure 116). The chromophobe cells
seem, then, to represent those cells which have given up their
secretion, which has probably diffused through the endothelium
into the sinus. The intermediary granular cells present no
marked irregularities. The failure of observers to definitely trace
the secretion into the blood channels does not seem to me to
discredit this conception, which is held by not a few investigators.
The connective-tissue framework of this lobe is taken up by
Dostoiewsky (’86) in detail, and in the ground-squirrel there are
no particular variations from his description. The peripheral
bundles derived from the dural capsule are the thickest and form
beams which divide and subdivide, until, when they reach the
central regions of the gland, they consist of very slender fibers
forming a recticulum which encloses small groups of cells. It is
against these fibers that the cells rest, and by them and by the
sinusoids the groups of cells are outlined.
2. pars juxtaneuralis. a) Pars infundibularis. The degree to
which the pars neuralis is enclosed by the pars infundibularis
varies according to the species. In the ground-squirrel the pars
infundibularis is concave on the neural side, and, being somewhat
cup-shaped, it extends over the pars neuralis peripherally. This
envelopment is by no means complete, but at least half of the
neural portion is so enclosed (compare figs. 1 and 2).
The basophilic character of the cells of this lobe has been indi-
cated earlier in this paper, and in no cases did they show any
acidophilic granulations. They are considerably smalleras a
rule, 44 to 7u in diameter, but occasionally large giant-like cells
are observed, which, if they are pale, look like colloid cysts. The
nuclei are correspondingly smaller in the typical cells, but their
nuclear characteristics are about the same. Bordering on the
lumen, there are cells making up the free margin (border cells)
which send processes in among the glandular ones. In the pars
infundibular these processes are better seen than in the distal
lobe for the reason that the large cysts and an occasionally large
cell create larger triangular spaces between themselves and allow

for a greater inward excursion of these cell processes (fig. 4, B).
The cells are compactly arranged and no sturdy bands of sup-
porting fibers are seen. The layer is four or five cells thick and
is closely applied to the pars neuralis. Toward the center the
layer is thinner, but where it is reflected to join the pars distalis
it is again thickened with the formation of the transition zone.
This zone, in many of the lower forms, according to Stendell,
usually shows a characteristic structure, but here the transition
is so abrupt that no such modification is apparent.
The vascularity of this part is slight in comparison with that
of the pars distalis. Where in the latter lobe the tissue is richly
supplied with blood and where the sinuses are extremely numerous,
here only an occasional blood-vessel is seen, most of these being
near the marginal region, in which the tissue turns back to join
the pars distalis. As a result, the cells are not separated into
well-defined groups, but are distributed quite evenly throughout
the extent of the part.
In addition to forming a smooth surface facing the lumen, the
pars infundibularis has a fairly regular neural border. No
elaborate granular inclusions within the pars neuralis are met
with. In most cases a slight undulation of the surface is all that
is seen. Toward the periphery of the cleft, in the transitional
zone, a few small inclusions occasionally occur, but by tracing
through adjacent sections they seem to be due to slight folds in
that region, and are not patches of gland tissue at all. Sepa-
rating the pars infundibularis and pars neuralis is a series of
blood-vessels, which, as mentioned by Dandy and Goetsch (’11),
are not concerned with the pars infundibularis. This layer of
vessels is thin and connective-tissue fibers are seen on either side
of them which apparently serve to bind together the pars
infundibularis and the pars neuralis. These fibers are derived
from the dural sheath of the brain which is prolonged down the
infundibulum over the nervous lobe.
The portion of the pars infundibularis investing the infun-
dibulum is not very extensive. It is best seen on the caudal
surface of the infundibulum as a layer of cells three or four deep.
ele
as!
On the upper surface they are not very well defined and, as a
rule, only a few gland cells are scattered along its extent. The
layer is closely applied to the infundibulum and consists of small
cells (3.54 to 4.54 in diameter) packed close together. The
cytoplasm is scanty and does not stain with acidophilic stains.
The nuclei are hyperchromatic and a general basophilic character
is imparted to the cells. Just exterior to this are numerous blood-
vessels bounded still further exteriorly by a layer of the dura.
These blood-vessels are in close relation to the cells and at fre-
quent intervals they may be seen extending into the gland sub-
stance. These vessels appear to be continuous with those of
the pars tuberalis above. The part, as a result, is quite vascular,
being much more so than the portion investing the infundibular
process.
b) Pars tuberalis. The pars tuberalis extends a considerable
distance forward (fig. 1), but is not made out posteriorly except
as a few scattered cells distributed along the base of the brain in
much the same way as the cells of the upper surface of the infun-
dibulum. The cells are slightly larger (64 to 8» in diameter),
but, like those investing the infundibulum, they have a scanty
cytoplasm and hyperchromatic nuclei. The most character-
istic feature is the tendency of the cells to arrange themselves
radially about a central point like a gland follicle. Such groups
are quite numerous, and in between them the cells are distributed
quite irregularly. Numerous blood-vessels are seen which seem
to be continuous with those of the pars infundibularis below.
II. Pars neuralis
Little need be said relative to this, inasmuch as the findings

agree very closely with those of Herring for the cat. The in-
fundibulum is long and slender and is invaded by the cavity of
the third ventricle as far as the nervous lobe. The fibers of the
infundibulum run for the most part longitudinally. There are
many which run transversely from the sides of the infundibulum
toward the center where they curve either upward or downward
along the stalk. The fibers which run toward the glandular part
aS
curve outward along the periphery of the nervous lobe where

they can be traced for a short distance. Those fibers in relation
to the pars infundibularis come to lie in a thick layer adjacent to
the gland tissue, and separated from it only by the series of
vessels already mentioned. The pars neuralis is supplied by small
arterial branches which enter at its posterior end, where, in sec-
tions, there is always a break at the point of entrance. Through-
out the part, small vessels are very evenly distributed, but these
are not very numerous. Nothing was seen of nerve cells. Many
neuroglia and ependyma fibers course through the lobe.
III. The colloid
The subject of hypophyseal colloid is fully summarized by

Stendell (14) and is treated in detail in several papers, but usually
only as applied to the pars infundibularis. Here, to be sure, it
is found in great abundance, but it is also very noticeable in the
pars distalis in several of the forms which J have observed. In
the ground-squirrel hypophysis the cysts are very numerous in
the pars distalis, but usually quite small as compared with those
of the pars infundibularis, being 6 to 8u in diameter in the former
and up to 8u to 15y and even larger in the latter.
The cysts of the glandular portion (pars infundibularis and
pars distalis) can be conveniently divided into two groups accord-
ing to the character of the contents. In one type the material
is apparently made up of the remains of broken-down cells (fig.
10). The second type presents a wholly different picture
(fig. 9). The contents of these are hyaline. They never show
any particles which could in any way be associated with cell
remains. This type is not confined exclusively to any particular
area, but is, however, far more common in the pars infundibularis.
These colloidal masses appear to have a darker peripheral portion
probably due, however, to refractive differences alone. External
to this sharp border, the surrounding cells are usually arranged
radially.
In spite of close study, I have never found cysts in the pars
infundibularis showing any communication with the cavity of
the gland, nor have I found anything in the nature of cilia within
them. The residual lumen is frequently quite filled with a
granular mass not unlike the contents of many of the adjacent
cysts. The lumen contains no cilia, though they are common in
the guinea-pig, as shown by Vanderburgh (717).
The granular cysts stain very irregularly. The hyaline ones
stain more consistently with the acid dyes in the pars distalis,
but also take on a good tinge with many of the basic ones. The
granular cysts show fragments in them resembling cell particles
in their staining reaction, some of them even appearing connected
with the cells surrounding the cyst. In addition to these frag-
ments there may be observed small masses staining with basic
dyes and seemingly derived from the nuclei of broken-down cells.
The rest of the contents stain well with the acid stain. It is
quite possible, that since the cysts are seen in varying degrees of
homogeneity, that the hyaline ones represent later stages of the
granular ones. The hyaline cysts which are usually acidophilic
might then well represent an extremely fine dispersion of this
nuclear material with a possible slight reduction of its basophilic
properties during its retention in the cyst.
In the pars infundibularis the cysts are practically always
hyaline, and while they can be made to take on acid stains
they are, nevertheless, very deeply basophilic. In the pars
neuralis colloid can be found distributed exactly as Herring
(08) describes. In the pars neuralis it is always hyaline and
stains deeply acidophilic.
Stendell remarks that it is difficult to determine how much of
this material consists of concentrated secretion and how much is
cell remains, and, although it may be observed in many varying
types of appearance, composition, and staining quality, that the
colloid is the ultimate product.
I take pleasure in thanking Prof. F. M. MacFarland for his
interest and assistance which proved invaluable in the prepara-
tion of this paper.
SUMMARY
1. The ground-squirrel hypophysis has a solid nervous lobe and

a patent infundibulum, the cavity of which extends as far as the
nervous lobe.
2. The gland cells, judged from their staining reactions, are
in all probability of the same cell type in different phases of
functional maturity.
3. The colloid cysts of the pars distalis are small and very
numerous, those in the pars infundibularis being larger, but fewer
in number. Relatively, the pars infundibularis contains the
greater amount of colloid, but, absolutely, I believe the pars
distalis exceeds in this respect.
4. No continuity between the cysts and cleft was seen and in
no cases were cilia found in either.
BIBLIOGRAPHY
ANDRIEZEN, W. L. 1894 On the morphology, origin and evolution of function

of the pituitary body and its relation to the central nervous system.
Brit. Med. Journ., vol. 1, p. 54.
Berkey, H. J. 1894 The finer anatomy of the infundibular region of the
cerebrum including the pituitary body. Brain, vol. 17, p. 514.
Brenna, C. 1900 Uber den normalen Bau und einige pathologische Verinder-
ungen der menschlichen Hypophysis cerebri. Arch. f. Anat. u. Phys-
iol., Physiol. Abt., Jahrg. 1900, S. 373.
Biepu, A. 1913 Innere Sekretion. 2 Auflage, 2 Teil. Berlin.
Cuiunet, J., AND JonNEsScO, V. 1910 Le pigment du lobe postérieur de l’hypo-
physe. C. R. Soc. Biol., Paris. T.'69, p. 626.
Cusuine, H. 1909 The hypophysis cerebri. Journ. Am. Med. Assoc., vol. 53,
no. 4, p. 249.
1912 The pituitary body and its disorders. Phila., Lippincott.
Danpy, W. E. 1913 The nerve supply to the pituitary body. Am. Journ.
Anat., vol. 15, p: 333.
Danpy, W. E., anp Gortscn, E. 1911 The blood supply to the pituitary body.
Am. Journ. Anat., vol. 11, p. 137.
Dosrormwsky, A. 1886 Uber den Bau des Vorderlappens des Hirnanhanges.
Arch. f. mikr. Anat., Bd. 26, S. 592.
Epincer, L. 1911 Die Ausfiihrwege der Hypophyse. Arch. f. mikr. Anat.,
Bd. 78, S. 496.
Fiescu, M. 1884 Uber den Bau der Hypophyse. Tageblatt der 57. Natur-
forscher Versammlung zu Magdeburg, 1884, S. 195.
Guerrini, G. 1905 Uber die Funktion der Hypophyse. Centralbl. f. allgem.
Pathol. u. pathol. Anat., Bd. 16, S. 177.
Herrine, P. T. 1908 The histological appearances of the mammalian pituitary

body. Quart. Journ. Exp. Physiol., vol. 1, p. 121.
1908 The development of the mammalian pituitary body and its
morphological significance. Ibid., vol. 1, p. 161.
Jackson, C. M. 1917 Effects of inanition and refeeding upon the growth and
structure of the hypophysis in the albino rat. Am. Journ. Anat.,
vol. 21, p. 321.
Lanauey, J. N. 1884 On the structure of secretory cells and the changes
which take place in them during secretion. Internat. Monatsschr. f.
Anat. u. Physiol., vol. 1, p. 69.
Livinaston, A. E. 1916 The effect of castration on the weight of the pituitary
body on other glands of internal secretion in the rabbit. Am. Journ.
Physiol., vol. 40, p. 153.
LorTHrRINGER, S. 1886 Untersuchungen an der Hypophyse einiger Saiugetiere
und des Menschen. Arch. f. mikr. Anat., Bd. 28, S. 257.
Mann, F.C. 1916 The ductless glands and hibernation. Am. Journ. Physiol.,
vol. 41, p. 173.
Miitter, W. 1871 Uber Entwickelung und Bau der Hypophyse, u.s.w.
Jenaische Zeitschr. f. Naturwissensch., Bd. 6.
Preremescuko, P. 1867 Uber den Bau des Hirnanhanges. Virchow’s Archiv.,
Bd. 38, S. 329.
Rerztus, G. 1894 Die Neuroglia der Neuro-Hypophyse der Saiugetiere. Biol.
Untersuch. N. F. VI, S. 21.
Rocowitscu, N. 1889 Die Verinderungen der Hypophyse nach Entfernung
der Schilddriise. Ziegler’s Beitrige., Bd. 4, S. 453.
Sr. Remy, G. 1892 Contribution 4 l’histologie de ’hypophyse. Arch. de Biol.,
T. 12, p. 425.
Scarrip1, V. 1904 Uber den feineren Bau und die Funktion der Hypophyse
des Menschen. Arch. f. mikr. Anat., Bd. 64, S. 235.
ScHonEMANN, A. 1892 Hypophyse und Thyroidea. Virchow’s Archiv, Bd.
129, S. 310.
Srigpa, L. 1890 Die Hypophyse des Kaninchens nach Entfernung der Schild-
driise. Ziegler’s Beitrige, Bd. 7, S. 535.
SrenvELL, W. 1913 Zur vergleichenden Anatomie und Histologie der Hypo-
physis cerebri. Arch. f. mikr. Anat., Bd. 82, S. 289.
1914 Die Hypophysis cerebri. Oppel’s Lehrbuch d. vergl.. mikr.
Anat. d. Wirbeltiere, 8. Teil. Jena.
Tuom, W. 1901 Untersuchungen iiber die normale und pathologische Hypo-
physis cerebri des Menschen. Arch. f. mikr. Anat., Bd. 57, S. 632.
Titney, F. 1911 Contribution to the study of the hypophysis cerebri with
especial reference to its comparative histology. Memoirs Wistar
Inst. Anat. and Biol., no. 2.
1913 An analysis of the juxta-neural epithelial portion of the hypo-
physis cerebri, etc. Internat. Monatsschr. f. Anat. u. Physiol., Bd.
30, S. 258.
TrautTMann, A. 1909 Anatomie und Histologie der Hypophysis cerebri einiger
Siuger. Arch. f. mikr. Anat., Bd. 74, S. 311.
VanperBurGH, C. M. 1917 The hypophysis of the guinea-pig. Anat. Rec.,
vol. 12, p. 95.
VINCENT, SwaLE 1912 Internal secretion and the ductless glands. London.
PLATE 1
5 (a) Typical chromophile and (b) typical chromophobe from the pars dis-
talis. Shown in their natural relation. X 1800.
6 Two ‘orange G staining cells’ from the same lobe. X 1800.
7 Group of cells showing densely and sparsely granular cells in an alveolus
in the pars distalis. X 1800.
8 Four chromophiles from the pars distalis in varying degrees of granulation.
< 1800.
9 Hyaline cyst from the pars distalis. This cyst is bordering on the residual
lumen 1. X 1800.
10 Granular cyst from the pars distalis containing fragments resembling
cytoplasmic and nuclear remnants. X 1800.
11 Arrangement of the cell elements about a blood-vessel. Not all of the
cells are represented. > 1800.
206
HYPOPHYSIS CEREBRI OF CALIFORNIA GROUND-SQUIRREL PLATE 1
HAROLD J. COOPER
207
Resumen por el autor, Ralph Dougall Lillie.
La histogénesis temprana de la sangre en Bufo halophilus Baird

y Girard.
El autor ha estudiado el desarrollo temprano de las células

sanguineas en los estados 'arvarios hasta los 19 mm de longitud.
Las células sanguineas primarias se originan principa mente en
la masa celular mesodérmica ventral. Estas células dan lugar a
eritroblastos primitivos y a grandes linfocitos. Los primeros
adquieren la forma definitiva acumulando hemoglobina y multi-
plicandose por mitosis, transformandose en eritrocitos primitivos.
En los estados estudiados no ha encontrado el autor eritropoiesis
definitiva. En la masa celular ventral se originan grandes lin-
focitos, a expensas del endotelio endocdrdico y del endotelio
general, o & expensas de células mesenquimatosas no diferen-
ciadas. De los grandes linfocitos se derivan pequefios linfocitos
y los tres tipos de células granulosas, eosin6filas, bas6filas y
células especiales 0 neutr6filas.
4sA
AUTHOR'S ABSTRACT OF 7 HIS PAPER ISSUED
BY THE BIBLIOGRAPHIC SERVICE, SEPTEMBER 29
THE EARLY HISTOGENESIS OF THE BLOOD IN BUFO

HALOPHILUS BAIRD AND GIRARD
RALPH DOUGALL LILLIE
Division of Anatomy of Stanford University Medical School
SEVEN FIGURES
CONTENTS
PU TOduetiOn ina sliters bune Nis eieess 08s B42 LS 2d) Woy Cee ok mente 209
Meapenta lean cemie th Od sates sak orca hires cparsie ety olcerroecaca seb oseen 211
em embrace lenass se este wars uctea) ss ce<yellaas eee yt ee ee Boag Stthyoyeeae Shave cs 212
eiregprimithve plood=cells.. et. kG. cote ca tlakeottdeles bates Sasser cls ee 214
The stage of differentiation of the primitive blood-cells into primitive
erythroblasts and large lymphocytes. The loss of the yolk............ 218
The further differentiation of the primitive erythrocyte stem............... 221
The hematopoietic tissue of the mesonephros............0..... 000.60. cees 222
SMALAaveaN CMCONCLUSIONS! 2:5) Saco. ate As ae eai seem nee by Settee netbee ols 228
TB OL aGaReyONY e aera Me case Oketal ia Retkece Pepe En ern” Cl? Cro ora ae ee CR acai te tee ene 230
INTRODUCTION AND LITERATURE
The histogenesis of the blood of Amphibia has occupied the

attention of but few writers. Among these the earlier ones and
Maximow (’10) in Rana hold that the primitive blood-cells give
rise only to red corpuscles, while in other vertebrates, such as
Lepidosiren (Bryce, ’04), birds and Tropidonotus (Danchakoff,
’08 a, 16 a), Chelonia (Jordan and Flippin, ’13), mammals (Max-
imow, ’09 a, and Jordan, 710), and in certain Amphibia them-
selves, (Bufo, Mietens, ’10), the same cells give rise to two lines
of descent, viz., an erythrocyte and a leucocyte stem. Further,
it has been shown that this erythrocyte stem is a transitory one,
giving place later to a definitive one which is a branch of the
leucocyte line.
This difference between the Amphibia and other vertebrates
and especially between the two Amphibian forms, suggested the
209
210 RALPH DOUGALL LILLIE
subject of the present paper: a reéxamination of the early devel-

opment of the blood elements in Bufo, and for this purpose the
California species B. halophilus Baird and Girard was found to
be the most convenient.
Bryce (’04) derives the blood and endothelium from the
splanchnic mesenchyme. The yolk-laden primitive blood-cells
quickly differentiate into a larger type with peripheral fibrils:
the erythroblasts, and a smaller lymphoid form. The larger type
pass over into yolk-free, oval, flattened cells in which hemoglobin
appears. The loss of the yolk may precede the appearance of
hemoglobin by a short time. In the next phase erythropoiesis
continues, the stem cell now being a large mononuclear cell.: The
mesenchyme gives rise to large mononuclear cells, polymorpho-
nuclear leucocytes, and granulocytes. The splanchnic mesen-
chyme, the intrahepatic mesenchyme, and the interstitial mesen-
chyme of the pro- and mesonephros are the chief sites of blood
formation. The last, at least, sets free leucocytes into the blood,
as they are more numerous in the cardinal vein than in the aorta.
Likewise, small lymphocytes appear. With the development of
the spleen further hematopoiesis is confined to it, to the lympho-
renal tissue, and to the gut-wall mesenchyme. The splenic
trabeculum cells give rise to erythrocytes through large lympho-
cytes, also to lympholeucocytes, finely and coarsely granular
eosinophils, and basophil granulocytes. In the later stages the
lymphorenal tissue is almost entirely lympholeucopoietic. Leuco-
and granulopoiesis are strictly extravascular, erythopoiesis intra-
vascular. This last conclusion is reached also by Bizzozero
(90) in birds, Danchakoff (08 a, ’16 a) in birds and reptiles,
Jordan and Flippin (’13) in reptiles, and Maximow (’09 a, ’10) in
mammals, selachians, and amphibians. And when erythrocytes
are formed outside of the vessels they soon degenerate, while
granulocytes are only formed intravascularly where the circula-
tion has been stopped up (Danchakoff, ’08 b).
In the main the differentiation of the blood-cells in the other
vertebrates follows the outline given by Bryce for Lepidosiren
(Maximow, ’09 a, 710; Danchakoff, ’08 a, ’16a; Mietens, 710;
Jordan and Flippin, ’13). e
—
HISTOGENESIS OF BLOOD IN BUFO HALOPHILUS Dia |
In birds (Danchakoff, ’08 a) and reptiles (Jordan and Flippin,

’13; Danchakoff, ’16 a) the primitive blood-cell is morphologically
a large lymphocyte and the primitive erythropoiesis gives way
gradually to the definitive. While in Selachia (Maximow, ’10 a),
Dipnoi (Bryce, ’04), Amphibia (Mietens, 710), and Mammalia
(Maximow, ’09 a) the primitive blood-cell is a distinct form and
the formation of primitive erthrocytes is separated by an interval
from that of the definitive cells.
Schridde (’07, ’08), in human embryos of 1 to 13 mm., finds
that the first blood-cells all differentiated into nucleated erythro-
cytes, whose nuclei began to fragment in the 6-mm. stage. Later,
when blood formation in the liver commences, parenchyme cells
giverise tomyeoblasts, giant-cells, and erythroblasts, lymphocytes
not appearing until much later.
Van der Stricht (’91, ’92, ’95, ’99), in mammals and birds, and
Weber (’07 a, b), in birds, find that all the first blood-cells become
erythrocytes and that no white cells occur in the blood stream
at first. .
Stockard (15 a, b) asserts that in Fundulus embryos without
circulation erythrocytes, endothelium cells, and leucocytes all
have distinct origins, no white cells being formed intravascularly
and all the first blood-cells arising from a definite anlage and
passing over into erythrocytes, and are not interrelated, no blood-
cells being formed from endothelium. He observed red cor-
puscles outside of the anlage, but accounted for them by assuming
a temporary circulation. White cells, on the other hand, arise
from the general mesenchyme. Reagan (’15b, ’17) and Reagan
and Thorington (’15) found that all the body mesenchyme of
Fundulus could furnish both blood- and endothelium cells.
My material consisted of eggs and larvae of Bufo halophilus

B. and G., from the closing of the neural groove to the beginning
of the metamorphosis. In this study only the earlier stages, up
to 19-mm. body length, are treated fully. The larvae were fixed
in Zenker-formalin for four to six hours, ten parts of formalin to
ORs RALPH DOUGALL LILLIE
one hundred of sublimate-Miiller being the proportion used.

The larvae were extracted three to six days in iodized alcohol and
preserved in 85 per cent alcohol. Imbedding was in paraffin, in
gelatin-paraffin (Weber, 714), and in celloidin-paraffin.
Serial sections 5 » in thickness of the larvae up to 9-mm. body
length were made. From the later stages the mesonephros, liver,
pancreas, and heart were dissected out and sectioned alone.
For stains Hansen’s iron hematoxylin with eosin or orange G,
-Dominici’s eosin-orange G-toluidin blue, Maximow’s eosin-azure
II, and thionin were used, the sections being cleared in xylene and
mounted in xylene-damar. Wright’s stain diluted with two parts
of water was also used, staining being sufficient in about ten
minutes, the differentiation being secured in alcohol rather than
in water. As controls and for comparison smear preparations of
adult blood and bone marrow were used.
THE VENTRAL CELL MASS
No discussion of the origin of this cell mass, whether ento-

dermal or mesodermal, will be made here, the evidence for its
mesodermal origin being deemed sufficient. In a larva of 2.5-
to 3-mm. body length it forms two plates on the ventrolateral
aspects of the yolk mass. It is distinguished from the latter by
the greater number of nuclei, the somewhat smaller size of the
yolk granules, and a line of demarkation, which, although not too
well defined, is nevertheless distinguishable. There is no strik-
ing difference in nuclear structure.
This mass is apparently syncytial in character, at least, no
cell boundaries are evident by ordinary methods. I believe that
is largely due to the pale character of the cytoplasm, which is
greatly obscured by the large rounded or elliptical yolk granules.
The same indefiniteness of cell outline obtains in the yolk mass
itself. But occasionally polygonal cells are seen outlined by
pigment, and more clearly by the blue stained (eosin-azure IT)
gelatinof the imbedding mass. These cells are rich in yolk, less
so than the yolk entoderm cells, their yolk granules being about
6 « long and half as broad. These granules stain readily with
=
a
HISTOGENESIS OF BLOOD IN BUFO HALOPHILUS Pile
eosin or orange G and very intensely and tenaciously with iron

hematoxylin. When blood elements of later stages are stained
with Heidenhain’s iron hematoxylin and eosin the granules of the
eosinophil leucocytes readily give up the black color and become
bright red. Bryce (’04) states that these granules were, in the
younger stages, either derived from, or actually were finely
divided yolk matter.
At the 3-mm. stage some few of the cells are seen to be round-
ing up and detaching themselves from the mass.
When stained with Dominici’s eosin-orange G-toluidin blue,
the nuclei of these cells take on a deeper blue than those of the
mesenchyme, showing a finely reticular structure with dark blue
nodal points. Most of the cells contain one or two purple stained,
large, smoothly rounded nucleoli. With Wright-alcohol the
nuclei show the reticular structure even better, the reticulum
seeming more irregular. Chromatin particles lie at the nodes.
The karyoplasm is basophil, with a tinge of pink in some cells,
thus suggesting the faintly oxyphil condition of the later stages.
The cytoplasm is usually colorless or nearly so, becoming in-
creasingly basophil soon after, and contains many small brownish
black pigment granules, which are present throughout all the cells
of the embryo.
The mesenchyme cell of this stage (3 mm.) is taking on a spindle
form. It has a basophil cytoplasm with fewer and smaller yolk
granules than the cells of the blood anlage. The nucleus is
clearer and poorer in chromatin, which lies in discrete, deeply
staining particles at the nodes of a loose network of lightly colored
linin threads. A deeply stained nucleolus is present.
The cell of the blood anlage and the mesenchyme cell differ in
-the reaction of the cytoplasm, in the number and size of the yolk
granules, and in the character of the nucleus. They are alike in
the possession of small dark brown pigment granules and in that
both are frequently seen in mitosis.
THE PRIMITIVE BLOOD-CELLS
The ventral cell mass breaks up into large round cells at about
3- to 3.5-mm. body length. The peripheral layer of the splanchnic
mesoderm forms an endothelial wall for the ventral sinus. The
wall next the yolk mass is incomplete, there being but few
endothelium cells on that side, so that the cavity is in part closed
in above by the yolk mass itself. This sinus seems to corre-
spond to an omphalomesenteric vein, for in later stages it can be
traced forward on each side into the sinus venosus.
Endothelium cells are seen in various parts of the body. These
are elongated cells with clear elliptical nuclei containing one or two
large nucleoli and but little chromatin. The karyoplasm is clear
and colorless. The conspicuous nucleolus is round or oval and
stains violet blue with eosin-azure II. The cytoplasm is basophil.
In these earlier stages the endothelia of the blood-vessels do
not seem complete, but present gaps here and there. Mietens
(10) finds this condition also in B. vulgaris.
In larvae of 3- to 3.5-mm. body length the ventral cell mass
becomes resolved into its constituent elements. The cells grad-
ually break apart, the mass as a whole enlarges, free space appears
within the endothelial wall, and the cells round up into spherical
elements which float free in the plasma. Not until about the
. 4mm. stage are these free rounded cells found in the cavity of
the heart or in the systemic vessels. So it would appear that the
breaking up of the blood anlage occurs before the commencement ~
of the circulation rather than after.
The primitive blood-cells (fig. 1, pbc) are large, spherical cells
about 13.5 to 22, in diameter, averaging 17 to 18u, heavily laden
with yolk granules, which are evidently solid in nature, as they
are not distorted by mutual pressure, while the nucleus is fre-
quently compressed and angular from the resistance of the yolk
granules. There are many extra nuclear pigment granules in these
and in the endothelium cells. The dark nucleus contains a con-
siderable amount of chromatin in angular or rounded masses of
varying size on a linin network. The karyoplasm is stained blue
with a reddish tinge (eosin-azure II). As a rule, it is darker and
HISTOGENESIS OF BLOOD.IN BUFO HALOPHILUS 21% |Le)
the chromatin is more abundant than in the endothelium cells,

but transition forms are found which approach the nuclear con-
stitution of the latter. The amount of chromatin does not show
any definite relation to the degree of basophilia of the cytoplasm.
The chromatin, when stained with thionin, appears as angular or
rounded blocks of an intensely dark blue color. The large round
nucleolus or nucleoli, they being frequently two in number, stains
bluish violet with eosin-azure II, and reddish violet, i.e., meta-
chromatically, with thionin. With the former stain it shows
little color difference from chromatin itself, and, as will be seen,
this difference becomes less and less apparent as development
proceeds. The cytoplasm is colorless or slightly basophil. The
basophilia appears especially around the periphery of the cell.
In some cells no basophilia is evident, though observation is
difficult on account of the large volume of yolk, but in those cells
in which the yolk is decreasing in amount considerable areas of
cytoplasm may be observed. When in mitosis these primitive
blood-cells show larger expanses of yolk-free cytoplasm, and but
little basophilia is then evident.
In these stages, 3.5- to 4.5-mm. body length, the mesenchyme
cells (fig. 2, ms) become almost free from yolk. Occasional
rounded cells (fig. 2, pbc), rich in yolk and morphologically iden-
tical with the primitive blood-cells, are seen in the loose mesen-
chyme. Two possibilities exist regarding them: that they have
lagged behind the spindle cells in yolk elimination, their nuclei
being modified so as to resemble those of the primitive blood-cells,
or that they are primitive blood-cells which have escaped from
the vessels and wandered out into the mesenchyme. The second
is not excluded, but the first seems more probable, for such cells
are found in the head mesenchyme before the breaking up of the
ventral cell mass, while the mesenchyme cells are still yolk-laden,
but are taking on spindle forms with diminishing chromatin
content of their nuclei.
The steady decrease in food yolk is more evident toward the
5-mm. stage, the larger cells becoming elongated and oval, with
a more circular cross-section, and with this diminution a centriole
sometimes becoming visible in the increasingly basophil cytoplasm.

In such larvae large cells with pale cytoplasm, a large vesicular

nucleus with little chromatin, a large nucleolus, and clear kary-
oplasm, and but little yolk are found laden with large numbers
of dark brown pigment granules. These cells circulate in the
blood stream, are seen passing through vascular walls, and are
found in the mesenchyme outside.
In the mesenchyme are also noted a few cells which are some-
what less elongated than the rest, more rounded up, and have a
more strongly basophil cytoplasm. These probably represent
the first lymphoid wander-cells. Toward the end of the phase
under description, 3 to 5 mm., the period in which the primitive
blood-cells are the only intravascular circulating elements aside
from the pigment cells just mentioned, the mesenchyme becomes
almost free from yolk, only an isolated cell here and there retain-
ing a granule or two. At the end of this phase the primitive
blood-cells have lost their spherical: form and a portion of their
yolk and are now roughly oval elements, all containing yolk,
with cytoplasm in varying grades of basophilia, and nuclei con-
taining somewhat variable amounts of chromatin, and slightly
basophil or very slightly oxyphil karyoplasm.
My results as to the primitive blood-cells coincide fairly closely
with those of Mietens(’10). He states that primitive blood-cells
resemble young erythrocytes more than they do primitive leu-
cocytes. The primitive blood-cells of Bufo vulgaris are sharply
bounded, spherical cells with dark nuclei, rich in yolk. In Lepi-
dosiren paradoxa (Bryce, ’04) the primitive blood-cells are
heavily laden with yolk, show a centrosome with aster, the
nucleus is round or oval, not distorted by the pressure of the yolk
granules as in Bufo halophilus, sometimes notched, the chromatin
is In rounded karyosomes connected by delicate processes to
form a reticulum. This, as may readily be seen, corresponds *
closely to the structure of my primitive blood-cells. Maximow
(10) characterizes the primitive blood-cells of Rana temporaria
as large spherical cells, rich in yolk, and those of Acanthias
vulgaris are hemoglobin-free, amoeboid, and basophil. In mam-
mals (Maximow, ’09 a) they are regularly spherical, smoothly
contoured cells about 10 to 11.5u in diameter, in the guinea-
ee
al
HISTOGENESIS OF BLOOD IN BUFO HALOPHILUS vA E
pig often amoeboid. The nucleus is large, surrounded by a

narrow cytoplasm. It is round or slightly indented. Within
it are pale karyosomes and one or more large distinct nucleoli,
which stain with a reddish tinge with eosin-azure II or Dominici’s
stain, lying on a linin net. The cytoplasm is very finely retic-
ular and fairly strongly basophil, more so than the cytoplasm
of endothelium and mesenchyme cells. It contains clear round
vacuoles, lying singly or in small groups. The nucleus is usually
somewhat eccentric, with its indented side toward the broader
part of the cytoplasm. On this side close against the nuclear
membrane lies a very distinct, usually hemispherical attraction
sphere, staining red with eosin-azure II or Dominici’s stain.
Around this lie many vacuoles. In iron-hematoxylin prepara-
tions a typical pair of centrosomes takes the place of the sphere.
Many mitotic figures are found in these cells. Thus it may be
seen that the nuclei are readily comparable in mammals and in
Amphibia. The difference in the degree of basophilia in these
forms may probably be attributed entirely to the presence and
gradual digestion of yolk in the cytoplasm, for when the primi-
tive blood-cells in Bufo halophilus have lost their yolk and have
differentiated into red and white cells, both types show a more
or less basophil cytoplasm (fig. 3). In birds and reptiles (Dan-
chakoff, ’08 a, ’16 a; Jordan and Flippin, ’13) the primitive blood-
cells are morphologically large lymphocytes, amoeboid, spherical,
with a more or less broad basophil cytoplasm. They have a
large clear nucleus, which may be indented, with one or two
well-marked, somewhat metachromatic nucleoli and well-marked
chromatin particles. The nucleus is excentric and a centriole
may be present. So, in the Sauropsida the primitive blood-cells
are large lymphocytes, while in Dipnoi, Amphibia, and Mammalia
‘they are a morphologically distinct form. ‘They are neither red
cells, nor are they white cells of any adult type. Their nuclei
are in B. halophilus distinct in character from those of the prim-
itive erythrocytes.
THE STAGE OF DIFFERENTIATION OF THE PRIMITIVE BLOOD-

CELLS INTO PRIMITIVE ERYTHROBLASTS AND LARGE
LYMPHOCYTES. THE LOSS OF THE YOLK
In larvae of 5- to 6-mm. body length the primitive blood-cells

lose their yolk almost completely, and differentiate into two new
types, one of which is a large, oval, flattened cell with much vacu-
olated oxyphil cytoplasm, the primitive erythroblast, the other
is morphologically identical with the large lymphocyte (fig. 3,
p.eb, L.lb, l.lc).
The yolk granules seem to disappear by intracellular solution,
not by granular fragmentation. Vacuoles appear around their
margins, enlarge and coalesce, while the yolk granule decreases
in size, finally passing into the interior of the vacuole. The
yolk granules become paler as they diminish and finally disappear
completely. The contents of the vacuoles are clear and colorless,
not staining with eosin-azure II. Bryce (’04) suggests that the
yolk granules, when they disappear, may break up into granules
which are eosinophil in nature, and that the early leucocytes
may be concerned in the distribution of yolk material. The
latter statement may be true, but as to the former, all yolk has
long since disappeared from the blood-cells before the first eosin-
ophil granulocyte appears in Bufo. We must note, however, that
Bryce lacked stages which showed the complete details of the
loss of the yolk, which might modify his interpretation. A third
alternative exists: that the yolk granules may be extruded from
the cells into the circulating blood, since one frequently sees
apparently free granules in the blood stream. Usually, however,
a small amount of basophil cytoplasm or a few pigment granules
may be seen close to them, so it may be that they lie within a
pseudopod of a cell in the next section. It seems perfectly pos-
sible, however, that the granules may be extruded from the cells -
and digested by the proteolytic enzymes of the blood.
Mietens (’10) merely states that the yolk disappears. Bryce
(04) notes that a vacuolization of the cytoplasm occurs and
seems to be associated with the disappearance of the yolk not
only in the blood-cells, but in the general tissue cells of the body
as well. I have noted such an appearance in the mesenchyme
cells of Bufo.
HISTOGENESIS OF BLOOD IN BUFO HALOPHILUS 219
Now as to the erythrocytoid differentiation of the primitive

blood-cells. The nucleus retains its round or oval shape and its
definite nuclear membrane, which may be indented by a single
remaining yolk granule. The chromatin becomes much more
abundant, condensing to form a heavy network. At this stage
the nucleolus reacts to basic stains in practically the same manner
as the chromatin of the nucleus, and as the denser network of
the latter develops, the nucleolus gradually loses its identity and
becomes merged in the common chromatin content of the nucleus.
The exact time of its disappearance is somewhat variable, in
some cases the nucleolus persisting almost up to the first
appearance of hemoglobin. The karyoplasm retains its faintly
oxyphil character as long as it is at all distinguishable in the
increasing density of the nucleus. This forms one of the distinct
differencés between the early erythroblasts and the large lympho-
cytes. As the yolk disappears from the cytoplasm, the latter
takes on an increasingly basophil reaction, varying somewhat in
intensity in different cells. With the appearance of hemoglobin
this blue color soon shows a gray tint tending either toward green-
ish or purplish, depending on whether the developing hemoglobin
takes a yellow or a pink stain with eosin-azure II. The cytoplasm
becomes much vacuolated, these vacuoles persisting some little
time after the disappearance of the yolk. The cells are oval,
but not yet flattened, with sharply defined, smooth contours
which are not necessarily perfectly uniform, but do not present
any evidence of amoeboid activity.
The second type of cell differentiating from the primitive
blood-cell is the large lymphocyte (fig. 3, J.lb, l.lc). In this
process the cytoplasm of the former becomes very strongly
basophil with the decrease of its food yolk. The cell outline
shows strongly marked evidence of amoeboid activity. The
cytoplasm of the large lymphocyte is broad, has no definite cell
membrane, and appears reticular in some cells, in others hyaline.
The nucleus is round, oval, or slightly indented, the indentation
frequently facing the remaining yolk, and central or excentric in
position. The karyoplasm becomes typically clear and colorless.
The chromatin decreases in amount and is disposed in irregularly
scattered particles and in occasional isolated strands. <A large

smoothly rounded, metachromatically basophil nucleolus persists,
showing but little change from the preceding stage. The cell is
now a typical large lymphocyte.
Such cells arise also from other sources than the primitive
blood-cells of the blood stream. Endocardial cells bulge out into
the heart cavity and become constricted from the endocardium,
at the same time becoming strongly basophil (fig. 3, en.c). These
cells contain but little yolk, otherwise being identical with those
arising from the primitive blood-cells. Endothelium cells in
various parts of the body become thickened and more basophil,
bulging out into the loose mesenchyme. Often free lymphoid
wander-cells are seen lying in contact with these endothelia.
Lymphoid wander-cells occur in many places in the mesenchyme:
in the head mesenchyme, in the interstitial spaces of the meso-
nephros, around the gut. Here they arise from the indifferent
mesenchyme cells by a withdrawal of their processes and an
increase in the basophilia of their cytoplasm, the nuclei changing
but little. It is likely that the cells resembling primitive blood-
cells which were seen in the loose mesenchyme of the earlier
stages (4 to 5 mm.) have gone over into lymphoid wander-cells
or extravascular large lymphocytes.
Although white cells do arise from these other sources, both
within and outside the vessels, they are not, as Maximow (’10)
claims they are in Rana temporaria, the sole source of white
blood-cells, for continuous series of transitions from yolk-laden
primitive blood-cells to yolk-free, and also yolk-laden, large
lymphocytes are found and these yolk-laden large lymphocytes
cannot have been derived from either the endocardium cells,
which are poor in yolk, or from the now almost yolk-free endo-
thelium and mesenchyme cells.
At .this stage, 5- to 6-mm. body length, the liver anlage is
invaded by blood-vessels. These narrow vessels contain but
few blood-cells, so the hematopoietic function of this organ is
not yet developed at this period.
HISTOGENESIS OF BLOOD IN BUFO HALOPHILUS i)iw)—
THE FURTHER DIFFERENTIATION OF THE PRIMITIVE

ERYTHROCYTE STEM
At the end of the last stage, 6 mm., the primitive erythroblast
is a large, ellipsoidal, vacuolated cell with a large, round or oval,
chromatin-rich nucleus. This cell is not at all flattened, but is
somewhat irregular in contour. With eosin-azure II its cyto-
plasm stains a light blue with no trace of pink.
Between 6- and 6.5-mm. body length hemoglobin first appears,
the cytoplasm now taking on a pink stain with triacid. With
eosin-azure II some cells may now show a greenish-gray color,
some a pink tint, but many are still pure blue. They are begin-
ning to flatten, but a considerable margin of cytoplasm still shows
on all sides of the nucleus. This body uniformly appears darker
than the cytoplasm and contains much more chromatin than
that of the lymphoid type. The erythroblasts now far out-
number the lymphocytes in the heart blood and frequently
present mitotic figures.
Between 6.5- and 7-mm. body length the erythroblasts rapidly
approach the definitive form (fig. 4, p.eb) of the adult erythrocyte,
there being now very little cytoplasm between the nucleus and
the flattened sides of the cell. With eosin-azure II the cytoplasm
is of a gray color with elements of both blue and pink visible.
The pink is more pronounced in some cells than in others, and
is determined by the increased development of hemoglobin in
their cytoplasm. Many vacuoles still remain, disappearing last
from near the nucleus. A cell containing a pu yolk granule
may still be met with occasionally.
From the 7-mm. stage on the majority of the cells, and soon
all of them, are oxyphil, hemoglobin-carrying primitive erythro-
cytes. Through all the stages studied, up to 38-mm. body
length, no definitive erythropoiesis was observed. Mietens (’10)
states that in Bufo vulgaris definitive erythropoiesis does not set
in until the time of the metamorphosis.
At the 8-mm. stage the primitive erythrocytes (fig. 5, p.ec) are
pink stained, hemoglobin-containing, oval, flattened, biconvex,
sharp-bordered discs with a spherical, flattened, or oval chrom-
atin-rich nucleus, showing an indistinct chromatin network.
papape RALPH DOUGALL LILLIE
Most of these cells have lost the vacuoles of the differentiation

stages. The dark brown pigment granules of the primitive blood-
cells are still present, but are very few in number. These cells
are still found in karyokinesis in the circulation up to 37-mm.
body length (and even in the young adult, also, according to
Freidsohn, ’10), which is as late as I have followed the larval
development.
THE HEMATOPOIETIC TISSUE OF THE MESONEPHROS
In a stage where the mesonephros is as yet represented only

by a single Wolffian duct, larvae of 8- to 9-mm. body length, one
sees mesenchyme cells multiplying in the neighborhood of this
duct, between it and the aorta, and undergoing a lymphoid
metamorphosis (fig. 6). All transitions from branched and
spindle-shaped lightly basophil mesenchyme cells (ms) to strongly
basophil, rounded, amoeboid lymphoid wander-cells (l.w.c) may
be found in a single field of the microscope. In addition, lym-
phoid cells with basophil cytoplasm and nuclei in varying degrees
of polymorphism are seen in 9-mm. larvae. All transitions from
slightly indented to horseshoe-shaped nuclei are found. The
more polymorphous of these nuclei show an increase in chromatin
and its progressive condensation about the periphery.
In this stage the first granulocytes appear. An eosinophil cell
with about twenty bluish-red (eosin-azure II) granules is shown
at e.c in figure 6. The first special cells are seen in this stage or
slightly later (fig. 7, s.c.), and have also been noted in the circu-
lating blood, but in very small numbers.
In a short time, however, the mesonephroic tubules become
convoluted and the interstitial tissue increases greatly in amount.
Much more granulopoietic differentiation is now seen. At first
only eosinophil and special cells appear, but at about 16-mm. body
length mast cells and small lymphocytes are likewise found.
Since at this stage all the later forms of the blood elements,
except spindle-cells, are represented, a description of each and its
development may be taken up at this point.
Small lymphocytes. A few typical small lymphocytes (fig. 7, s.le)

are seen in the 16-mm. stage and more in later stages. These cells
have proportionately large nuclei and a narrow rim of basophil
cytoplasm. ‘The chromatin is disposed in large angular blocks,
some typical ‘Radkern’ or wheel-like nuclei being found, but no
nucleoli may be distinguished.
Transition stages from large to small lymphocytes are found.
It is evident that the small lymphocyte is a more highly differ-
entiated cell form than the large lymphocyte, for when the latter
differentiates to a granulocyte similar changes in the nuclear
constitution occur, the chromatin becoming more abundant and
the nucleolus disappearing.
EKosinophil cells. ‘These cells, in the mature state in which they
are found intravascularly, possess horseshoe-shaped, lobed, or
segmented nuclei, cytoplasm which takes very little stain with
eosin-azure II, and many coarse round eosinophil granules of
varying size. The nuclear chromatin is larger in amount than
in the large lymphocyte, but is still quite variable, being disposed
primarily in contact with the well-defined nuclear membrane
and then in variable density over the nuclear network. In this
network no nucleolus can be distinguished. The karyoplasm is
clear and almost colorless to eosin-azure II. The larger oxyphil
granules often show a clear area in the center, staining more
deeply peripherally, while in other cells smaller, more deeply
staining, homogeneous granules occur. These granules become
very numerous in mature cells, which may occasionally show
mitotic figures. Such cells may readily be found in all phases of
diapedesis from the tissue into the vessels. They are also found
in the circulating blood. In one isolated but unmistakable case
(fig. 7, e.end) an eosinophil cell was found as a part of the endo-
thelial wall itself.
All transitions from typical large lymphocytes to mature
eosinophil leucocytes are found. First a few coarse basophil
granules, which, however, show no trace of metachromasia,
appear, usually in the concavity of the slightly indented nucleus,
soon increasing in number and changing their stain from blue to
purple red with eosin-azuré II; then the nuclear membrane
becomes evident, the chromatin at the same time increasing in

amount and in coarseness of karyosomes and the nucleolus los-
ing its metachromasia and blending with the nuclear chromatin,
and the cytoplasm becomes less and less strongly basophil as the
nucleus assumes its final polymorphous form and the granules
increase in number and come to be purely oxyphil. }
Since phagocytosis of erythrocytes is absent in this extravas-
cular tissue, it seems very improbable that these eosinophil
eranules ariseas fragmentation derivatives of ingested hemo-
globin containing cells, as maintained by Weidenreich (’11).
The interpretation of the eosinophil endothelium cell described
above is a matter of difficulty, but it seems possible that this
may be just another manifestation of the close relationship of
the endothelium cell and the blood-cell. Danchakoff (16 d) has
used differentiation into similar end types as a proof of identity
of stem cells in regard to the small cortical cells of the chick
thymus and their mother cells. So it seems here that the dif-
ferentiation of an endothelium cell into an eosinophil granu-
locyte would tend to show the identity of the endothelium cell
with the large lymphocyte, which is the mother cell of the eosin-
ophil granulocyte. According to the view of exogenous origin
of eosinophil granules, it would merely be a proof of phagocytic
activity on the part of the endothelium cell. Ordinarily phag-
ocytosis is followed by intracellular digestion, and not by the
preservation of the fragments of the ingested cell as an integral
part of the phagocyte, even when autogenous erythrocytes are
ingested by the endothelial phagocytes of the liver and spleen
(Kyes, 715). The hemosiderin granules which are formed as a
result of the intracellular digestion of red cells have no resem-
blance to true eosinophil granules.
Mast cells. This type of cell is first seen extravascularly in
the 16-mm. stage. Maximow (710) finds them in Rana only
after the metamorphosis. Mietens (710) does not find any mast
cells at all, which is probably due to mast granules being soluble
in such acid fixatives as he used. In Bufo halophilus these gran-
ules resist the action of water after fixation with Zenker formol.
The mast cells are mostly extravascular; one or two only being
found within the vessels in my preparations, and are very few in
number, not more than one occurring in each section.
The nuclei are simple, round, oval, or bean-shaped. In the
younger stages the nucleus is that of a large lymphocyte, round
and .clear, with but little chromatin in small karysomes, and a
metachromatically basophil nucleolus, the cytoplasm is broad
and usually slightly basophil, the granules are large, round, and
comparatively few in number, staining an intensely dark purplish
blue with eosin-azure IT and violet red to blue violet with thionin.
No variation in the staining quality of these mast granules could
be found in any stage of their development. Thus the possibility
of confusing the younger eosinophil granules with mast granules
isexcluded. In the older cells (fig. 7, m.c) the cytoplasm becomes
less basophil, the mast granules increase in number, the nucleus
becomes richer in chromatin, the nucleolus changes its staining
quality and becomes indistinguishable from the chromatin. In
some cells the chromatin is disposed in a typical ‘Radkern,’ being
distributed in large angular blocks lying against the nuclear
membrane and pointing toward a centrally located:one. The
karyoplasm is clear and may be slightly oxy- or basophil. Trans-
ition stages between these two types, here designated as older
and younger, are seen. One cell, manifestly an old one, was
found, in which the chromatin was decreased in amount, the
nucleus appearing vesicular, the cytoplasm was colorless and
showed ragged edges, indicating fragmentation of its periphery.
In this cell the granules were large, only about six or seven re-
maining. Some evidence of amoeboid activity may be noted in
the contour of many of the mast cells.
Special cells. Under this head are designated many cells
with purplish-pink cytoplasm and simple to polymorphous nuclei
(fig. 7, s.c). The more polymorphous of these nuclei may be
lobed and segmented, as many as five lobes being found. ‘Their
chromatin is more or less abundant and lies along the nuclear
membrane and in a chromatin net, or in small karyosomes in
faintly oxyphil or colorless karyoplasm. The cytoplasm is the
homogeneous, indistinctly granular, or, in a few cases, definitely
so. Such granules are very fine, rounded, discrete, and stain
purplish pink. In these cases the cytoplasm is colorless. The
staining reactions of these granules will be further treated later.
Transitions from large lymphocytes to special cells are readily
found. First the nucleus becomes slightly indented, then sausage
shaped, the nucleolus loses its differential stain and form and
blends with the chromatin, then a pink spot appears in the blue
cytoplasm (eosin-azure II) opposite the concavity of the nucleus.
Or the change of stain may be more diffuse and almost simul-
taneous throughout the whole cytoplasm. The chromatin
becomes more abundant and arranged along the nuclear mem-
brane, which is now more evident.
One of these cells was seen in karyorrhexis. Its eosinophil
cytoplasm was non-granular and contained a horseshoe-shaped
nucleus whose chromatin was collected into numerous round,
strongly basophil granules within the nuclear membrane, except
at the tip of one lobe, where they were escaping into the cyto-
plasm. The karyoplasm itself was clear and colorless. Mitotic
figures are not rare in the special cells.
Since it is usually stated in the literature that the special
leucocytes of the Amphibia are non-granular or have at most
only an azurophil granulation, J thought it worth while to in-
vestigate further the character of the fine granulation described
above. It may be noted here that in air-dried films of ‘adult
blood and bone marrow of Bufo halophilus all the special leu-
cocytes show a fine purplish-pink granulation on a clear karyo-
plasm when stained with Wright’s stain in the customary
manner.
In order to eliminate as far as possible variations due to age
or differences in technique, sections of the mesonephros of a
single larva of 19-mm. body length were used for the following
series of staining reactions. All the stains used were differ-
entiated in 96 per cent alcohol, except the thionin which was
differentiated in absolute alcohol and cleared in xylene. The
dilute Wright’s and Jenner’s stains mentioned were diluted to
the same color tint as eosin-azure II-1 : 10 : 1 (Maximow, ’09 c).
HISTOGENESIS OF BLOOD IN BUFO HALOPHILUS DOG
STAIN TIME RESULTS
Vennerk (Strong) au. seen eee 10 m. | Some cells show distinct pink gran-
ules.
Jenner (dilute)..................| 24 hr. | Some cells show very distinct pur-
plish-pink granules. Best granule
stain of all.
Thionin (alcoholic)..............| 30 m. | No granules in blue cytoplasm.
“TallionGhin IBS Soenont
woe eebe 10m. | No granules in blue eytoplasm.
Eosin-azure IT (1:10:1).........| 24 hr. | Some cells show purplish-pink gran-
ules.
PAT nlleg Ge 1000) Sacre nyc an: 10m. | Many cells show gray or blue-gray
granules in blue cytoplasm.
\ivieiedatn (Giakon) Racgeeoce
soda coger 10 m. | Some cells show purplish-pink gran-
ules.
\ivierelong (Cue) s coo co pose ocnesuc 24 hr. | Some cells show purplish-pink gran-
ules.
Unna’s polychrome methylene
|
SUITE 5 Beatthe oe at at ae eae 10m. | No granules in blue cytoplasm.
Hin Weh s.triaerdie. f.\. 4.052 ders 5m. | Some cells show fine red granules.
Most have red cytoplasm instead.
OSM ED oer agere ene he cielonsas q.s. No granules show.
The conclusions led to by the above table are as follows: 1)

these granules are not soluble in water after fixation, because
when stained for twenty-four hours they are no less distinct than
when they are only exposed to water for ten minutes; 2) they are
not azurophil, because they stain with Jenner’s stain and with
triacid, neither of which contains any methylene azure, and
because they do not stain pink with azure II or Unna’s poly-
chrome methylene blue, both of which contain methylene azure,
uncombined with eosin, however; 3) they are neutrophil, because
they do not stain with eosin, or with basic analines alone, but
do stain with neutral stains; 4) they have, however, a special
affinity for acid fuschin, for they stain red with triacid rather
than purplish pink.
My special cells are without doubt identical with Mietens’
(10) oxyphil wander-cells. Mietens’ failure to observe granules
in these cells can be attributed to inappropriate fixation. Max-
imow (’10) speaks of special granulocytes and of special leucocytes
with oxyphil cytoplasm without making it clear whether all the

cells were granular or not. Downey (’18) describes in Amblys-
toma special leucocytes containing many fine purplish-pink
granules (Wright’s stain) which he concludes are azurophil, dif-
fering, however, from the azurophil granules of the lymphocytes,
and being homologous with the granules of the special leucocytes
of mammals. Werzberg (711) finds in eight species of Urodela
and eight species of Anura that the special leucocytes are non-
granular, save in Salamandra maculosa and Rana esculenta in
which he describes fine azurophil granules. In these same two
species Niegelewski (’94) describes fine granules which are neu-
trophil to triacid. Griinberg (01) gets the same results as
Werzberg. Freidsohn (’10) and Weidenreich (’11, p. 75) state
that granules are lacking in the special leucocytes of Amphibia.
In Rana aurora and in Batrachoseps attenuatus I was unable to
find any granules in the special leucocytes.
In larvae of 37- to 40-mm. total length, that is, larvae which
show the beginning of hind-limb buds and measure 15 to 16 mm.
from the snout to the anus, all of the above-described cell forms
may be found in the loose connective tissue surrounding the bile
duct. Inthe mesonephros there may be seen small areas in which
all the cells are lymphocytes, while surrounding these islets all
types of white cells are mixed. As yet, few leucocytes occur in
the vessels, nor is there any definitive erythropoiesis. As in
Bufo vulgaris (Mietens, *10), there is no interstitial hemato-
poiesis in the liver of B. halophilus up to this stage.
In closing, I take pleasure in thanking Prof. F. M. MacFarland
for his aid and suggestions and for the use of his private room in
the Hopkins Marine Station during the summer of 1916.
SUMMARY AND CONCLUSIONS
1. The primitive blood-cells arise from the ventral cell mass.

Some isolated cells are found in various parts of the mesenchyme.
2. These cells lose their yolk by intracellular solution, dif-
ferentiating at the same time into primitive erythroblasts and
large lymphocytes. .
3. The primitive erythroblasts soon assume the definitive form

and acquire hemoglobin.
4, After this no heteroplastic formation of erythrocytes occurs,
up to the latest stages studied. Homoplastic erythropoiesis
continues.
5. Large lymphocytes may arise from endocardial and general
endothelial cells.
6. Mesenchyme cells all over the body may develop into large
lymphocytes or lymphoid wander-cells.
7. The large lymphocyte is the mother cell of all the other
white blood-cells. ,
8. Three types of granular leucocytes occur: eosinophils, special
cells, and mast cells.
9. The granules of the eosinophil cells are endogenous.
10. The special leucocytes of Bufo halophilus contain many fine
granules, which are not azurophil, but neutrophil.
The histogenetic relations here described may be expressed
by the following diagram, which is self-explanatory in the light
of the foregoing account.
Mesoderm cell
Primitive blood-cell Mesenchyme cell
Primitive erythroblast Large lymphocyte <—— Endothelium cell

|
\ |
| |
Primitive erythrocyte | Small lymphocyte | Eosinophil cell |

| | | |
{ { |
Mast cell Special cell
Eosinophil endothelium cell

BIBLIOGRAPHY
BapEertscHER, J. A. 1913 Muscle degeneration and its relation to the origin

of eosinophil leucocytes in Amphibia. Am. Jour. Anat., vol. 15, p. 69.
BizzozpEro, G. 1890 Neue Untersuchungen iiber den Bau des Knochenmarkes
bei den Vogeln. Arch. f. mikr. Anat., Bd. 35, S. 424.
Bryce, T. H. 1904 Histology of the blood of Lepidosiren paradoxa. I. Struc-
ture of resting and dividing corpuscles. II. Hematogenesis. Tr.
Roy. Soc. Edinburgh, vol. 41, pp. 291, 485.
DANCHAKOFF, VERA 1908 a Untersuchungen iiber die Entwicklung des Blutes
und Bindegewebes bei den Vogeln. I. Die erste Entstehung der
Blutzellen beim Hiithnerembryo und der Dottersack als blutbildendes
Organ. Anat. Hefte, Bd. 37, 8S. 471.
1908 b Untersuchungen itiber die Entwicklung von Blut und Bindege-
webe bei Végeln. III. Das lockere Bindegewebe des Hiihnchens im
fetalen Leben. Arch. f. mikr. Anat., Bd. 73, S. 117.
1916 a Uber die Entwicklung des Blutes in den Blutbildungsorganen
(Area vasculosa, Dottersackanhinge, Knochenmark, Thymus, Milz,
und lockeres Bindegewebe) bei Tropidonotus natrix. Arch. f. mikr.
Anat., Bd. 87, S. 497.
1916 b Origin of the blood cells. Development of the hematopoietic
organs and regeneration of the blood cells from the standpoint of the
monophyletic school. Anat. Rec., vol. 10, p. 397.
1916 ec Concerning the conception of the potentialities in the embry-
onic cells. Anat. Ree., vol. 10, p. 415.
1916 d The differentiation of cells as a criterion for cell identifica-
tion, considered in relation to the small cortical cells of the thymus.
Jour. Exp. Med., vol. 24, p. 87.
1916e Equivalence of different haematopoietic anlages (by the
method of stimulation of their stem cells). I. Spleen. Am. Jour.
Anat., vol. 20, p. 255.
Downey, Hat 1909 The lymphatic tissue of the kidney of Polyodon spathula.
Folia Haemat., Bd. 8, S. 415.
1913 The granules of the polymorphonuclear leucocytes of Ambly-
stoma, with a few notes on the spindle cells and erythrocytes of this
animal. Anat. Anz., Bd. 44, 8S. 309.
1914 Heteroplastic development of eosinophil leucocytes and of
haematogenous mast cells in bone marrow of guinea-pig. Anat. Rec.,
vol. 8.
1915 The origin and development of eosinophil leucocytes and of
haematogenous mast cells in the bone marrow of adult guinea-pig.
Folia Haemat., Bd. 19, S. 148.
Emmet, V. 1915 Concerning certain cellular elements in the coleomic cavities
and mesenchyma of the mammalian embryo. Am. Jour. Anat., vol.
220, p. 73.
Freipsoun, A. 1910 Zur Morphologie des Amphibienblutes. Zugleich ein
Beitrag zur Lehre von der Differenzierung der Lymphocyten. VIII.
Fortsetzung der Studien iiber das Blut und die blutbildenden und
-zerstorenden Organe. Arch. f. mikr. Anat., Bd. 75, 8. 435.
GRUNBERG, C. 1901 Beitrige zur vergleichenden Morphologie der Leucocyten.

Virch. Arch., Bd. 163, S. 303.
JORDAN, H. E. 1910 A microscopical study of the umbilical vesicle of a 13-mm.
human embryo, with especial reference to the entodermal tubules
and the blood islands. Anat. Anz., Bd. 37, S. 12, 36.
1916 Evidence of the hemogenic capacity of endothelium. Anat.
Rec., vol. 10, p. 417.
JorpDAN, H. E., anp Furppin, J. C. 1913 Haematopoiesis in Chelonia. Folia
z Haemat., Bd. 15, S. 1.
Kyers. P.1915 The physiological destruction of erythrocytes in birds. Int.
Monatschr. f. Anat. u. Phys., Bd. 31,.
LanineE, P. 1912 Des globules blancs éosinophiles dans le sang des poissons
d’eau douce. Arch. de Biol., T. 37, p. 525.
Maximow, A. 1909a Untersuchungen iiber Blut und Bindegewebe. I. Die
friihesten Entwicklungsstadien der Blut- und Bindegewebezellen
beim Saiugetierembryo, bis zum Anfang der Blutbildung in der Leber.
Arch. f. mikr. Anat., Bd. 73, S. 444.
1909 b Untersuchungen iiber Blut und Bindegewebe. II. Uber die
Histogenese der Thymus bei Siugetieren. Arch. f. mikr. Anat., Bd.
74, S. 525.
1909 c Uber zweckmiissige Methoden fiir cytologische und _histo-
genetische Untersuchungen am Wirbeltierembryo, mit spezieller
Beriicksichtigung der Celloidinschnittserien. Zeit. f. wiss. Mikr., Bd.
26, 5.177.
1909 d Der Lymphocyt als gemeinsame Stammzelle der verschiede-
nen Blutelemente in der embryonalen Entwicklung und im post-
fetalen Leben der Séugetiere. Folia Haemat., Bd. 8, S. 125.
1910 Uber embryonale Entwicklung der Blutzellen bei Selachiern
und Amphibien. Anat. Anz. Ergiinzungsh., Bd. 37, S. 64.
1912 Untersuchungen tiber Blut und Bindegewebe. IV. Uber die
Histogenese der Thymus bei Amphibien. Arch. f. mikr. Anat., Bd.
79, S. 560.
1913 Untersuchungen iiber Blut und Bindegewebe. VI. Uber Blut-
mastzellen. Arch. f. mikr. Anat., Bd. 83, S. 247.
Mietens, H. 1909 Die Entstehung des Blutes bei Bufo vulgaris. Jen. Zeit.
f. Natwschft., Bd. 45, S. 299.
1910 Entstehung der weissen Blutkérperchen und der Milz bei Bufo
vulgaris. Jen. Zeit. f. Natwschft., Bd. 46, S. 301.
_ Minor, C. 8. 1912 Die Entstehung des Angioblasts und die Entwicklung des
Blutes. Keibel u. Mall Handb. d. Entwick. d. Menschen, Bd. 2, S.
483.
NIEGELEWSKI, F. 1894 Die Ehrlich’sche Granulation der weissen Blutkér-
perchen bei einigen Tierspezies. Inaug. Diss. Miinchen, 1894.
OpreL, A. 1892 Unsere Kenntniss von der Entstehung der rothen und weissen
Blutk6rperchen. Zusammenfassendes Referat. Centbl. f. allg. Path.
u. path. Anat., Bd. 3, 8. 241, Bd. 4, 8. 193.
EW RALPH DOUGALL LILLIE
ReaGaAn, F. P. 1915 a Vascularization phenomena in fragments of embryonic

bodies completely isolated from yolk-sae blastoderm. Anat. Rec.,
vol. 9, p. 324. ;
1915 b <A further study of the origin of blood vascular tissues in
chemically treated teleost embryos, with especial reference to haema-
poiesis in the anterior mesenchyme and in the heart. Anat. Rec.,
vol. 10, p. 99.
1917 Experimental studies on the origin of vascular endothelium and
of erythrocytes. Am. Jour. Anat., vol. 21, p. 39. .
ReaGan, F. P., anp THorineton, J. M. 1915 The vascularization of the
embryonic body of hybrid teleosts without circulation. Anat. Rec.,
vol. 10, p. 79.
Ringoren, A. R. 1915 Observations on the differentiation of the granules in
the eosinophilic leucocytes of the bone marrow of the adult rabbit.
Anat. Rec., vol. 9, p. 683.
Srockarp, C. R. 1915 a The origin of blood and vascular endothelium in
embryos without a circulation of the blood and in the normal embryo.
Am. Jour. Anat., vol. 18, p. 227.
1915 b A study of the wandering mesenchymal cells on the living
yolk-sae and their developmental products; chromatophores, vas-
cular endothelium, and blood cells. Am. Jour. Anat., vol. 18, p. 525.
Scurippue, H. 1907 Die Entstehung der ersten embryonalen Blutzellen des
Menschen. Folia Haemat., Bd. 4, Sup., S. 157.
1908 Die Entstehung der ersten embryonalen Blutzellen des Men-
schen. Verh. deutsch. path. Ges., XI. Vers., Dresden, 1908, S. 360.
VAN DER Srricut, O. 1891 Le développement du sang dans le foie embryon-
naire. Arch. de Biol., T. 11, p. 19.
1892 Nouvelles recherches sur la genése des globules rouges et des
globules blancs du sang. Arch. de Biol., T. 12, p. 199.
1895 De la premiére origin du sang et des capillaires sanguins dans
Vaire vasculaire du lapin. C. R. Soc. Biol., T. 47, p. 181.
1899 L’origin des premiéres cellules sanguines et des premiers vais-
seaux sanguins dans l’aire vasculaire de chauve-souris. Bull. de
VAcad. R. de Med. Belg., Ser. 4, T. 13, p. 336.
Weser, A. 1907 a Fecrmes de transition entre les 6bauches vasculaires et les
{ ilots sanguines dans l’aire opaque des embryons de canard. C. R.
Soc: Biol,,<T...62, p. 762:
1907 b Remarques suc la développement des vaisseaux et du sang
dans l’aire vasculaire de l’embryon du canard. C. R. Soc. Anat.,
Lille, 1907, p. 18.
1914 Inclusion mixté a la gelatin et a la paraffine. Bibliog. Anat.,
T. 14, p. 146.
Werpenreicu, F. 1911 Die Leucocyten und verwandte Zellformen. Wies-
| 4@ baden, 1911. ©
Werzsera, A. Studien zur vergleichenden Haemocytologie einiger poikilo-
thermer Vertebraten. Folia Haemat., Bd. 11, S. 17.
A len
wie 4, a
a ay Pig
» WJ
ey ;
aa
a tt) :
od —
Hi
ett ie ahd 1k? PAS EN ofONa “H

'
Pee ER! bonveag abi Re i , 3
Ade deh bison ni, i

Bs ireshite itaN dprides
neh Lane bideery, nent. axitowlt 1 oo) ae

ia) Ae ate ‘pith Aapodis: Mae} T yt
BY diy Lhe iad or pay iy
Ah
one : ‘ sets od rid ae, wor: Any Ne a

ad pal ie. Kes Vina Rent ody:“bhi a ie saibyal
way pd Sarge Oe, |
Te a igo he
dal Pea
ya
Leg hevpeouns act
Onael /
:Pane, ae
rs f
Wy ide wy s
oii
‘Us i io oy ald Aver ages ai ¥ voiy
1 ae
sins asian nidghoacsOy i i ily

Py ~ a
tbemri(pat‘kyA. RG tect ac aeSie oes
cus ,
4
Des arena ie?88

PLATE 1
All the figures represent preparations stained with eosin-azure II, and are
drawn with the camera, using a B. & L. 1/12-inch immersion objective with the
1-inch eye-piece.
1 Section through the ventral sinus of a 3.5- to 4-mm. larva.

2 Head mesenchyme of a 3.5- to 4-mm. larva.
3 Section through heart wall and cavity of a 5.6- to 6-mm. larva.
4 Erythrocytes of a 6.5- to 7-mm. larva.
5 Erythrocytes from the heart of an 8- to 8.5-mm. larva.
6 Section of the anlage of mesonephros of a 9-mm. larva.
7 Section through the mesonephros of a 16-mm. larva.
ABBREVIATIONS
e.c., eosinophile granulocyte ms., mesenchyme cell

e.end., eosinophil endothelium cell p.b.c., primitive blood-cell
en.c., endocardium cell p.eb., primitive erythroblast
end., endothelium cell p.ec., primitive erythrocyte
1.lb., large lymphoblast s.c., special cell
L.lc., large lymphocyte s.lc., small lymphocyte
l.w.c., lymphoid wandering cell sp.c., spindle cell
m.c., mast cell w., mesonephric tubule
TOGENESIS OF BLOOD IN BUFO HALOPHILUS PLATE 1
RALPH DOUGALL LILLIE
|
+
Resumen por el autor, Henry H. Donaldson.
Instituto Wistar de Anatomia y Biologia.
Estudios cuantitativos sobre el crecimiento del esqueleto en la

rata albina.
Los huesos pertenecientes a los esqueletos de 106 ratas albinas

de edad variable, desde el nacimiento a los 500 dias, fueron prep-
arados por maceracién en ‘Gold Dust’’—una lejia sélida—, y
después pesados y medidos en estado fresco, después de secos a la
temperatura del laboratorio y a la de la estufa, pudiéndose
determinar de este modo la cantidad de agua contenida en los
estados primero y segundo. El autor ha seguido el aumento de
peso de cada hueso o grupo de huesos y su relacién con el peso
del cuerpo y el peso total del esqueleto, asi como el aumento en
la longitud de los huesos largos comparada con la longitud del
cuerpo. Las proporciones del esqueleto adulto tienden a fijarse
proximamente durante la lactancia y los pesos y longitudes de
algunos de los huesos largos pronto aleanzan una relaciOn fija
con el peso de todo el cuerpo o el del esqueleto, de tal modo que
estos tltimos valores pueden inferirse por los hallados en un solo
hueso. El autor nota el cambio de forma de los huesos largos
durante el crecimiento y hace algunas comparaciones con los
datos referentes al hombre. Las tablas incluidas susministran
‘valores de referencia con los cuales pueden compararse los
correspondientes a ratas criadas bajo diversas condiciones.
AUTHOR’S ABSTRACT OF THIS PAPER ISSUED BY
THE BIBLIOGRAPHIC SERVICE, SEPTEMBER 29
QUANTITATIVE STUDIES ON THE GROWTH OF THE

SKELETON OF THE ALBINO RAT
HENRY H. DONALDSON
WITH THE ASSISTANCE OF SARA B. CONROW
The Wistar Institute of Anatomy and Biology
TWENTY-THREE CHARTS
INTRODUCTION
In the following pages we shall describe several quantitative

growth changes in the skeleton of the albino rat between birth
and maturity. This investigation is a study of bones as part
of the entire mammalian body, and not of bone as a special
tissue, and this limitation of the problem explains the method of
procedure.
For the ‘ligamentous’ skeleton of the albino rat we have the
data of Jackson and Lowrey (712) based on seven age groups.
The skeleton, as prepared by these investigators, consisted of
the bones, plus the cartilages and ligaments, which remained after
the muscles had been rapidly dissected away fron these parts in
the fresh state. Their study dealt with the entire skeleton as
one of the tissue systems. It has seemed to us, however, desir-
able to study also the skeleton as represented by the bones alone,
because this would admit of dealing with the bones in detail,
and also would give material which sooner or later could be com-
pared with the corresponding material from man.
It is the main purpose of this study, therefore, to furnish data
on the skeleton and its parts in albino rats of different body
weights and of different ages, presenting these data in such a form
that they can fairly be compared with any new set of values
taken from other rats examined by similar methods.
237
238 HENRY H. DONALDSON
MATERIAL
From albino rats between birth and old age, 106 skeletons
were prepared by Miss Conrow, the work extending from 1913
to 1917. Of these skeletons, 85 were from inbred rats taken
from Dr. King’s inbred series and the remainder were from stock
Albinos, all the animals having been reared in the colony at
The Wistar Institute under uniform food conditions. Each
skeleton contains 283 bones, including and counting the teeth.
In the preparations here examined there were, however, only
230 bones, prepared separately, and these in turn were weighed
in twenty-eight lots, in the case of each rat.
The details concerning the composition of the skeleton and
the number of bones prepared for weighing are given in Appen-
dix 1. ' All these skeletons have been stored, after being dried at
96°C. for six days.
In addition to this series of entire skeletons the long bones of
the limbs were similarly prepared from 54 young albino rats
(males 32, females 22; body weight 5 to 86 grams; age, birth to
64 days) and the data on these bones were combined in tables
8, 9, 11 to 14, 20, and 23 with those from the series of complete
skeletons, according to body weight or to age.
With this material we purpose to show in the first instance
how the entire skeleton increases in weight in relation to the
total body weight and to age. Then, the growth of its several
divisions and parts, 1) in relation to that of the entire skeleton,
and, 2) in relation to the weight of the entire body; also the
relative growth of the three divisions of the fore limbs and of
the hind limbs, respectively. At the same time the change in
the percentage of water with growth has been followed in both
the entire skeleton and its various parts, and finally, the increase
in the length of the long bones both of the fore limbs and of the
hind limbs has been determined, in relation both to the body
weight, to the body length, and to one another.
We hoped to discover that in some parts of the skeleton the
length or weight of the bones was so well correlated with the
data for the entire skeleton that it would be possible, in any
instance, to compute the body length or the weight of the entire
GROWTH OF THE SKELETON 239
skeleton from the value of a portion of it. Tables 9, 23, and 24

which follow show that this result has been attained.
Many series of computations and tables have been made and
are complete in manuscript form, but we shall use only a portion
of them for the present paper. Copies of all of the tables have
been filed, however, in the archives of The Wistar Institute and
are at the service of other investigators.
TECHNIQUE OF PREPARATION
For the details of the technique the reader is referred to Appen-

dix 2. In general the procedure was as follows: The recently
dissected parts of the skeletons were macerated by immersing
them in 50 to 200 ce. of a hot 2 per cent solution of commercial
‘Gold Dust.’ This treatment left the bones from the older rats
nearly unmodified as to their weight and water content.
The precise effect on the bones of rats of different body weights
(=ages) are given in detail in Appendix 2.
The macerated bones were weighed in closed bottles, first
fresh (i.e., immediately after maceration), then after room drying
for thirty days in the open air, and finally after six days in the
oven at 96°C. Where lengths were measured, these also were
taken in the three conditions of moisture. All the determina-
tions and computations were made separately for each rat, but
the results were later combined in body-weight groups. Between
4.3 grams and 15 grams of body weight, the groups were made at
5-gram intervals, and after that at intervals of 10 grams. This
arrangement yielded forty-two body-weight groups ranging from
birth to 513 days of age, and from 4.38 to 485 grams in body
weight.
The foregoing statements apply to all the records except
those for the lengths of the long bones of the limbs. In these
cases the data from the additional fifty-four rats were combined
with those from the main skeleton series, with the result that in
the records for bone length there were formed forty-seven, in-
stead of forty-two, body-weight groups.
The tables, charts, and discussion which follow are based on
these group values, but are presented in a modified form, as will
be explained later. The two sexes have been treated together,

since there is no clear sex difference in the relative weight of the
skeleton. All weights are given in grams and all lengths in
millimeters, and it must be repeated that all the values recorded
are those obtained after treatment with the ‘Gold Dust,’ which
reduces the fresh weights of the bones and especially those from
rats less than 100 grams in body weight or about 70 days in age,
as shown in table 34, Appendix 2.
METHOD OF WEIGHING THE BONES
The bones were weighed in closed bottles on balances sensitive

to 0.0001 gram. This procedure is simple and not subject to
serious observational errors except for the weighing of the ‘moist
or fresh bones,’ 1.e., the bones immediately after maceration and
before they are allowed to dry at all. To obtain this fresh value
for the bones the effort has been made to remove, by wiping on
filter-paper, all evident fluid adhering to the surface. This is
readily accomplished in most cases, but the moist cranium
requires special preparation before it is thus weighed.
In this case the external surfaces can be readily wiped, but it
is also necessary to remove with strips of filter-paper any accu-
mulated fluid from the nares, the bullae, and the cavity of the
cranium. With due attention, however, to this preliminary
preparation of the cranium, satisfactory weight data may be
obtained. The moment one has to deal with dry bones, the
difficulties just mentioned disappear.
ARRANGEMENT OF DATA IN THE TABLES
As the weight of the skeleton was the value to be found, the

106 rats in this series were arranged according to body weight.
It is well known, however, that the observed body weight of the.
rat fluctuates readily, and especially in old animals, tends to fall
off. In order to make the entries systematic, a fixed procedure
was followed for the establishment of the normal body weight.
The body length of each rat was taken, and, as this does not
fluctuate with the nutritional condition of the animal, it was
used as the standard,
In table 68 of ‘The Rat’ (Donaldson, ’15) a series of normal

body weights for each millimeter of body length is given for each
sex. When the observed body weight of a rat fell 5 per cent or
more below the table value given for the observed body length,
the observed value was corrected to the table value, on the as-
sumption that some unfavorable nutritional condition had reduced
the body weight temporarily.
On the other hand, we had in the case of many of the old rats
a record of their maximum body weights, reached earlier, and
often well above those observed at the time of killing, and also
above the table values, and in such instances the maximum body
weight was that entered in the table. These two procedures
eliminated from the table subnormal body-weight values.
To prevent a misuse or misinterpretation of the values given
in the tables which follow, we repeat that the absolute weights
given are those found after maceration in the ‘Gold Dust’ wash-
ing powder (Appendix 2).
This treatment reduces the weight of the bones in all cases,
but especially in those rats less than 100 grams in body weight or
70 days of age. The percentage of water is also reduced somewhat.
The lengths of the bones are but very slightly affected. The
effects of the maceration are, however, sufficiently similar among
the different bones to make determinations of relative values
based on the macerated bones applicable to the perfectly fresh
bones also.
The most evident need for the corrected values is in comparison
between the body weight and that of the skeleton or some of its
parts—since in these instances the true weights are needed to
give the correct ratios—and when such ratios are desired the
proper corrections must be made.
It should be added, however, that though the relative values of
the skeleton and its parts, as shown in table 2 and chart 2, are
somewhat modified by the use of the corrected data, yet the
modification is slight and, moreover, does not affect the interpre-
tation given to these records.
The values in our tables may be used therefore as standards
for comparison with determinations from other rats, the bones
of which have been macerated by the same or a similar process.
ON THE WEIGHT OF THE FRESH SKELETON AND OF ITS PARTS ON

BODY WEIGHT
In table 1 are given the absolute and relative fresh weights,

as observed, for the entire skeleton and its two main divisions—
the axial and appendicular skeleton (see Appendix 1 for the
grouping of the bones). In charts 1 and 2 the absolute and the
relative values, respectively, are plotted, the mean values being
indicated by smooth graphs.
In table 1 it is the observed values which are given, while in
charts 1 and 2 smoothed graphs have been so drawn that only
those observed values which lhe above or below the respective
graphs are evident on the charts. Reading the values shown by
the smoothed graphs, we obtain the data entered in table 2.
As stated earlier, the object in view is to present the data in
such shape that new observations may be compared with them,
and also to give some idea of the variability which is to be ex-
pected. After considering the several ways in which these ends
might be attained, the following plan was adopted as the most
satisfactory and usable.
All the tables appearing in this paper were first made on the
basis of the observed values, either according to, body weight, as
in table 1, or according to body length or some other standard.
The data were then charted, and on the chart a smooth graph
was drawn free hand. The values given by this smoothed graph
were then read at fixed intervals and tabulated, and these in
turn appear, as in table 2, and in all the later tables of the same
type.
In favor of this procedure are the facts that such a table,
based on the smooth graph, is useful as a standard for reference,
while the corresponding chart shows by means of the graphs the
general character of the growth change, and by means of the |
entries which lie on either side of the graph, the variability of
the records. ‘The nature of the data hardly justifies an attempt
at greater precision than this. Values between those given in
the tables may, in all cases, be obtained by interpolation.
All the principal tables which follow are arranged on the fore-
going plan, beginning with table 2, which gives the data of table.
1 in this revised form. |
TABLE 1
Absolute and relative fresh weights of the entire skeleton, axial skeleton, and
appendicular skeleton of the albino rat, as directly observed after maceration in
‘Gold Dust.’ Data entered in forty-two groups according to body weight
= 2 6 | my8 | saa
ey |25,
Be ale Ss.
ai | teak | Za
35
Serieae
a sarea peSemeeee
4a B Carl ye el =Be:
mf <- ho26
BE ues
4°R
2 = a > | fae [eae g 3 Za Ho
2
s
E
2
S
g
he | co
a
Re | Bec
a8 |geo|
tie
28 | ca
5
es | aae gaa
z >
days grams mm grams Bil grams per cent grams per cent
2 2 B 4.3 46 0.38 8.9} 0.28 6.5 0.10 2.4

+ 1 6 8.8 61 0.85} 9.6) 0.61 7.0 0.24] 2.6
2 19 16.4 83 (Su elOS ele OSs eG ORvOneAe4
4 23 24.4 95 ye Ya Je 5).7) 0.84] 3.4
1 i 25 35.6 107 DalS |) 920) 220 5.6 lee 3.4
2 1 33 45.5 119 S010) |] Getsl] Yo 4.8 1.4 3.0
2 1 37 53.9 127 4.02 | 7.5) 2.4 4.5 1.6 3.0
2 1 ol 65.5 137 NAVY etd] BI 4.7 2.0 3.0
1 46 TifeAl 140 5.00 | 6.5} 3.0 3.9 2.0 2.6
1 1 54 84.0 150 Bars) || 9°70}, Sins 4.2 2.3 2.8
2 91 95.0 157 TA08hl) (aoe Aao, 4.7 2.6 2.8
1 1 83 | 106.0} 163 7.24] 6.8} 4.4 4.1 2.8 2.7
2 153 116.0 168 8.55 | 7.4) 5.6 4.8 3.0 2.6
i 1 106 127.0 169 Selon .0) arom: 4.0 3.1 3.0
2 120 144.0 182 10.30} 7.2) 6.5 4.5 3.8 2.7
3 1 103 156.0 182 9.85 | 6.3} 6.0 3.8 3.8 2.5
2 1 109 174.0 186 MOLE) || O24] O57 3.8 4.1 2.4
2 Zi 232 | 195.0 196 IPR BY |! (574) 0) 4.1 4.3 2ril
1 1 227 | 205.0 198 12.24 | 6.0) 7.9 3.8 4.3 2.2
2 390 | 215.0 198 13243751622) Ser 4.0 4.7 2.2
1 3 283 | 223.0 | 202 | 13.46] 6.0} 8.7 3.9 4.8 2.1
1 1 328 | 233.0 | 207 TAS 6s Ont 3.9 5.1 222
2 448 | 242.0) 204 | 14.04] 5.8] 9.0 3.7 5.0 2.1
2 499 | 256.0] 210 BSE || GePA, AKO).8! 4.0 5.4 2.2
2 213 | 268.0 | 217 15.90 | 5.9) 10.1 3.8 5.8 2.1
1 1) 304 || 275.0 | 215 16.06 | 5.8} 10.2 Bad 5.9 74511
1 1 411 283.0 | 218 17.14} 6.0) 11.0 3.8 6.1 ore,
5 367 | 293.0] 218 16.74 | 5.7) 10.7 3.6 6.1 Ze
2 393 | 303.0] 219 16.85 | 5.6) 10.7 3.5 6.2 2.1
2 1 A403) +\"325).0) | 2211 18.80 | 5.8) 12.0 3.7 6.8 > 2.1
1 489 | 336.0 | 226 19.53 | 5.8) 12.5 3.7 7.0 2.1
3 1 450 | 343.0 | 225 1QE59) | Soa l2n6 3.7 7.0 2.0
1 474 | 355.0 | 228 | 19.51] 5.5] 12.5 3.5 7.0 2.0
1 513 | 366.0 | 230 19.42) 5.3] 12.5 3.4 6.9 1.9
2 469 | 375.0 | 235 TOROS eons |l2a6 3.4 7.3 1.9
1 AT BOE Onl edge 20640 | 5.1 12-8 hae 7.3 1.9
4 435 | 404.0 | 240 | 21.32] 5.3] 13.5 3.4 7:8 1.9
3 321 A420 e226) e220), 5. 113-4 3.2 7.8 1.9
3 393 | 430.0 | 2388 | 22.93] 5.3) 14.2 3.4 ites fe UY)
3 474 | 444.0} 239 | 23.48] 5.3) 15.1 3.4 8.3 1.9
2 460 | 461.0] 251 24.29 | 5.3) 15.5 3.3 8.8 2.0
1 453 | 485.0 | 246 | 23.48] 4.8) 14.6 3.0 8.9 1.8
244. HENRY H. DONALDSON
Entire Skeleton
k ae ie
9 50 00 150 200 250 300 350 400 450
Chart 1 Absolute fresh weight of entire skeleton, axial skeleton, and appen-
dicular skeleton in grams, on the body weight (albino rat). Table 2.
@ Entire skeleton. A Axial skeleton. x Appendicular skeleton.
BODY WEIGHT IN GRAMS

To
I Corr Corre CTT Tt
00 30 00 150 200 750 300 350 400
Chart 2 Relative fresh weight of entire skeleton, axial skeleton, and appen- y
dicular skeleton, on body weight (albino rat). Table 2.

@ Entire skeleton. A Axial skeleton. > Appendicular skeleton.
TABLE 2
Absolute and relative fresh weights of entire skeleton, axial skeleton, and appen-
dicular skeleton, on body weight. From smoothed graphs in charts 1 and 2
WEIGHT OF FRESH
eee ee
WEIGHT
eae
ON BODY
eee] ES |srsdeniy” |creanaice,
BODY WEIGHT| WEIGHT BODY
WEIGHT WEIGHT
grams grams per cent grams per cent grams per cent
5 0.46 9.20 0.34 6.70 0.12 2.50

10 0.99 9.90 0.69 6.87 0.3 3.03
15 1.59 10.61 0.98 6.53 0.61 4.08
20 2.06 10.30 1.24 6.20 0.82 4.10
25 2.40 9.60 1.48 5.90 0.92 3.70
30 2.70 9.01 1.68 9.09 1.02 3.42
35 3.06 8.73 1.87 5.33 1.19 3.40
40 3.39. 8.47 - 2.08 5.20 1.31 3.27
45 3.69 8.20 2.26 59.03 1.43 Bale
50 4.00 8.00 2.44 4.88 1.56 3.12
55 4.27 OU 2.62 4.77 1.65 3.00
60 4.54 Con 2.76 4.60 1.78 2.97
65 4.83 7.43 2.94 4.53 1.89 2.90
70 5.16 eal 3.15 4.50 2.01 2.87
75 5.48 7.30 3.35 4.47 2.13 2.83
80 5.82 120 3.59 4.44 2.27 2.83
85 6.12 7.20 3.75 4.4] 2.37 2.79
90 6.48 7.20 3.96 4.40 2.52 2.80
95 6.81 C17 4.18 4.40 2.63 Past
100 7.10 7.10 4.39 4.39 Dettal 2.71
110 Util (Or 4.79 4.35 2.92 2.66

120 8.33 6.94 5.16 4.30 Sralid 2.64
130 8.93 6.87 5.95 4.27 3.38 2.60
140 9.48 6.77 5.91 4.22 3.57 2.55
150 10.00 6.67 6.30 4.20 3.70 2.47
160 10.51 6.57 6.69 4.18 3.82 2.39
170 11.00 6.47 7.02 4.13 3.98 2.34
180 11.47 6.37 7.33 4.07 4.14 2.30
190 11.84 6.23 7.66 4.03 4.18 2.20
200 12.34 6.17 8.00 4.00 4.34 DENG
210 12.81 6.10 8.38 3.99 4.43 2.11
220 13.31 6.05 8.71 3.96 4.60 2.09
230 13.80 6.00 8.99 3.91 4.81 2.09
240 14.35 5.98 9.34 3.89 5.01 2.09
250 14.83 5.93 9.63 3.85 5.20 2.08
270 15.80 5.85 10.21 3.78 5.59 2.07

290 16.76 5.78 10.79 3.72 5.97 2.06
310 17.58 5.67 22 3.62 6.36 2.05
330 18.48 5.60 11.85 3.59 6.63 2.01
350 19.32 5.52 12.36 3.53 6.96 1.99
370 20.02 5.41 12.65 3.42 C30 1.99
390 20.94 5.37 13.22 3.39 ela 1.98
410 21.61 5.27 13.53 3.30 8.08 1.97
430 22.32 5.19 13.89 3.23 8.43 1.96
450 22.95 5.10 14.40 3.20 8.55 1.90
470 23.55 5.01 14.95 3.18 8.60 1.83
485 24.20 4.99 15.37 3.17 8.83 1.82
Turning now to an examination of the tables! and charts it is

seen that the absolute increase in the weight of the skeleton (chart
1) is somewhat rapid at first; indeed, the data (table 2) show
that up to 15 grams of body weight the skeleton grows more rapidly
than the rest of the body (chart 2), and as a result its relative
weight increases; but after that the growth is slower, so that the
value of the skeleton weight (observed) on the body weight falls
(table 2) from 10.6 per cent at a body weight of 15 grams to
about 5 per cent at a body weight of 485 grams. If the weights
of the skeleton are corrected for the action of the macerating
fluid (table 84, Appendix 2) then the foregoing percentages
become, respectively, 12 per cent and 5.2 per cent.
On looking at the graphs and tables for the axial and appen-
dicular skeletons, we see that the weight of the appendicular
is always smaller than that of the axial skeleton, but after a
body weight of 15 grams, the graphs for relative weights of the
two portions (chart 2) are nearly similar in their general course.
The mature relations between the two parts of the skeleton are
attained therefore at about twenty-two days, which is shortly
before the time of weaning.
The striking feature in this last comparison is the approximate
constancy in the weight relations of the two divisions of the
skeleton from an early age, although the appendicular division
grows just a trifle more slowly than the axial; falling, with shght
fluctuations, from 62 per cent of the axial skeleton at 15 grams
to 57 per cent at 485 grams, as shown by the values in Table 2.
1 Intervals used in tables.

On body length: up to 95 mm. of body length values are given at intervals of
2.5mm. From 95 mm. to 195 mm. at intervals of 5 mm. Above 195 mm. at
intervals of 10 mm.
On body weight: up to 100 grams of body weight values are given at intervals
of 5 grams. From 100 grams to 250 grams at intervals of 10 grams. Above 250
grams at intervals of 20 grams. The last interval is 15 grams.
On skeleton weight: up to 7 grams of skeleton weight values are given at
intervals of 0.25 gram. Above7 grams at intervals of 1 gram. Special intervals
are used in tables 6 and 7 giving the proportional weights of parts of the limbs.
Values falling between those entered in the tables may be obtained by simple
interpolation.
ON THE WEIGHT OF THE CRANIUM—FRESH, ROOM-DRIED, AND

OVEN-DRIED (CRANIUM = SKULL WITHOUT MANDIBLES)
The determinations have been made in this case in all three

conditions of moisture, as the weight of the cranium has signifi-
cance not only as a part of the skeleton, but also indirectly
as an index of cranial capacity and therefore of brain weight.
The data from the smoothed graphs are given in table 3 and the
graphs in chart 3.
wth of Skeleton—Albino Rat |
susseeeeeaeeeceoeessscs
Chart 3 Absolute weight of the cranium—fresh, room-dried and oven-dried—

on body weight (albino rat). Table 3.
@ Fresh. A Room dried. x Oven dried.
The form of the graphs does not call for special comment,
but a comparison of the initial values with those later in the
series shows that the cranium is increasing in weight 2+ only about
two-fifths of the rate of the entire skeleton.
ON THE WEIGHTS OF THE DIVISIONS OF THE APPENDICULAR

SKELETON
Turning now to the two divisions of the appendicular skeleton,

it is possible to compare the shoulder-girdle plus appendages
with the pelvic girdle plus appendages, in their respective rela-
tions to the body weight. This has been done in table 4, and
both the absolute and relative values on body weight are given
in charts 4 and 5. A study of these data shows plainly that

TABLE 3
Absolute weight of cranium—fresh, room-dried, and oven-dried—on body weight.

Values from the smoothed graphs in chart 3
BODY WEIGHT FRESH ROOM-DRIED OVEN-DRIED
grams 8 g grams
5 0.04
10 0.08
15 13
20 18
25 3&BrewhWd
©
Ore
Oop 22
30 20
35 . 29
40 & onwn
~I
~I Oo
45 83 OF
50 41
45
orws .49
53
7
.60
65
69
73
77
Sees
KS
ere
Ooaancoococooccoo
ee 81
—"(7)
w or
ce
cee
Ltn
ee
ce
oe
ee
ee ce
ee
cee
cee
ee
Let
ee)
eo
for)
oe)
a
ea
labrqae:
ae
aS
to
bd
WW =a
IOTNHAA
SCOMDNSAH
PWWWwWWWW
248
just at birth the pelvic girdle plus appendages is relatively poorly

developed, but it at once begins to grow more rapidly than the
shoulder-girdle and appendages, and after the first few days
forms the larger fraction of the entire body. The mature
WEIGHT IN GRAMS Soa abe et Fresh Pelvic Gird

“Fresh f Hf
+
: I i
4
3 Fresh Shoulder Gird!
2
1 T tot { a
BODY WEICHT IN GRAMS
05 50 Too 50 200 50 300 350 400 450
Chart 4 Absolute fresh weight of shoulder-girdle plus fore limbs and pelvic
girdle plus hind limbs, on body weight (albino rat). Table 4.
@ Shoulder-girdle plus limbs. A Pelvic girdle plus limbs.
PERCENTAGE
Growth of Skeleton—Albino Rat

25)
Fresh
20
Pelvic Girdl
15 |
1.0
Shoulder Gird!
0.5

5 JC 7
Q 0 50 109 150 260 250 300 350 400 450
Chart 5 Relative fresh weight of shoulder-girdle plus limbs and of pelvic

girdle plus limbs, on body weight (albino rat). Table 4.
@ Pelvic girdle plus limbs. A Shoulder-girdle plus limbs.
relations between the two divisions are established at puberty

(body weight, 100 grams).
This completes the presentation for the weight of the fresh
skeleton and its parts considered in relation to the weight of
the body.
TABLE 4
Absolute and relative fresh weights of shoulder-girdle with fore limbs and pelvic
girdle with hind limbs, on body weight. Values from the smoothed graphs in
charts 4 and 6
WEIGHT WEIGHT
Bape Watcen | seognmy stebee | xo rome uarae, |e

oe |oe
grams grams per cent grams per cent
5 0.068 1.35 0.052 1.15

10 Onlss ion 0.163 1.66
15 0.215 1.43 0.395 2.65
20 0.300 1.50 0.520 2.60
25 0.323 1.29 0.597 2.41
30 0.354 as 0.666 2.24
35 0.389 ibaa 0.801 2.23
40 0.416 1.04 0.894 2.20
45 0.437 0.97 0.993 2.21
50 0.475 0.95 1.085 2.17
55 0.490 0.89 1.160 2.11
60 0.510 0.85 1.270 2.11
65 0.540 0.83 1.350 2.07
70 0.567 0.81 1.443 2.06
75 0.593 0.79 1.537 2.04
80 0.632 0.79 1.638 2.04
85 0.646 0.76 1.724 2.03
90 0.684 0.76 1.836 2.02
95 0.713 0.75 1.917 2.02
100 0.730 0.73 1.980 1.98
110 0.792 54 2.128 1.94

120 0.852 0.71 2.318 1.93
130 0.897 0.69 2.483 1.91
140 0.966 0.69 2.604 1.86
150 1.005 0.67 2.695 1.80
160 1.040 0.65 2.780 1.74
170 1.088 0.64 2.892 1.70
180 1.134 0.63 3.006 Way
190 1.140 0.60" 3.040 1.60
200 1.180 0.59 3.160 1.58
210 1.197 0.57 3.233 1.54
220 1.232 0.56 3.368 1.53
230 1.288 0.56 3.522 1.53
240 1.344 0.56 3.666 1353
250 1.400 0.56 3.800 152
270 1.512 0.56 4.078 1.51

290 1.624 0.56 4.346 1.50
310 1.736 0.56 | 4.624 1.49
330 1.815 0.55 4.815 1.46
350 1.925 0.55 5.035 1.45
370 2.035 0.55 5.335 1.44
390 2.145 0.55 DEOL 1.43
410 2.255 0.55 5.825 1.42
430 2.365 0.55 6.065 1.41
450 2.385 0.58 6.165 1.37
470 2.397 0.51 6.208 1.34
485 2.474 0.51 6.356 1.33
bhooro
GROWTH OF SKELETON IN WEIGHT ON AGE
Although the determination of the skeleton weight according

to age was not specially considered in this study, yet the series
yields data which can be used to show this relation. If we con-
sider the body weights as they appear in table 1 to be normal for
age, the data in table 74 (Donaldson, ’15) being taken as the
standard, then between birth and 365 days the first twenty-seven
groups can be used for the males and the first twenty-two groups
for the females. The data are presented in table 5. When these
data for the weight of the fresh skeleton are plotted for each sex
on the base line for age, there appear the smooth graphs shown
in chart 6.
It is the form of these graphs which is of value, for they show
that in both the male and the female the graph for the weight
of the fresh skeleton is sinuous and very similar to the graph for
the weight of the entire body on age (Donaldson, ’06) despite the
fact already noted that after the first few days of life the relative
weight of the skeleton is continually diminishing.
While the form of the graph is trustworthy the absolute
values for the weight of the skeleton at the given ages are peculiar
to these data and not general, and therefore should be used for
comparison only with caution. The smaller absolute weight of
the skeleton in the female goes, as the tables show, with the
smaller body weight at the several ages.
Tables 1 to 5 and charts 1 to 6 show in a general way the
growth in the weight of the skeleton and its larger parts on
body weight and age. We pass now to some of the smaller
divisions of the skeleton.
ON THE GROWTH OF THE PARTS OF THE LIMBS
Our records permit us to determine the manner in which each

of the three divisions of the limbs increases in weight. ‘Table 6
shows the weight of all the bones in both fore limbs and also the
percentage values for the humerus (2), ulna plus radius (2), and
carpus plus fore foot (2) at intervals of 0.02 gram, while table 7
gives the same information for the corresponding divisions of

both hind limbs at intervals of 0.05 gram.?
Chart 7 gives the smoothed graphs for the relative weights
of the three divisions of the fore limbs on the weight of the bones
of both fore limbs, and chart 8 the smoothed graphs for the cor-
responding divisions of the hind limbs on the weight of the bone
of both hind limbs. The outstanding relations shown by these
graphs are the similar behavior of the corresponding bones within
16 f WEIGHT IN GRAMS t Soares ue pe as Male
Be
iy Seeeeee
13 i Female
12 )
i
10
9
8
7
6
5
4
3
2
1 AGE IN DAYS
9 40 80 7 160 200 240 280 320 300

Chart 6 Absolute weight of entire fresh skeleton on age—according to sex
(albino rat). Table 5.
@ Males. A Females.
each limb and the continued relative growth of the first two
segments of the limb as contrasted with the behavior of the most
distal segment, which in both cases diminishes in its relative
weight during all the later portion of the growth period—.e.,
after a limb bone weight of 0.22 gram for the fore limbs (table 6)
and 0.45 gram for the hind limbs (table 7). In the fore limb
2 The sign (2) after any paired bone indicates that the value given in the
table applies to the two bones (right and left) taken poneubeny although the
designation of the bone is in the singular number.
TABLE 5
Absolute weight of fresh skeleton, on age (males and females). Values from the
smoothed graphs in chart 6
AGE MALES FEMALES
days grams grams
0 0.35 0.35
4 0.75 0.75
8 1.10 1.10
12 1.45 1.45
16 1.80 1.80
20 2.08 2.08
24 2.34 2.34
28 2 2.57
32 2.88 2.88
36 3.20 3.20
40 3.50 3.50
+4 3.85 3.85
48 4.15 4.15
52 4.50 4.50
56 5.10 4.85
60 5.70 5.25
64 6.30 5.70
68 7.05 6.30
72 7.65 6.90
76 8.15 1.32
80 8.60 U0)
84 9.05 8.15
88 9.50 8.40
92 9.94 8.73
96 10.24 9.01
100 10.50 9.33
110 11.18 9.92

120 IL ees) 10.38
130 12.35 10.82
140 12.83 11.25.
150 Sela 11.62
160 13.65 11.89
170 13.97 12.18
180 14.25 12.45
190 14.50 PAR
200 14.77 12.92
210 15.05 13.12
220 15.30 13 225.
230 15.41 13.34
240 ; 15.51 13.40
250 15.63 13.46
270 15.87 13.56

290 16.11 13hi2
310 16.38 13.82
330 16.63 13.94
350 16.89 14.05
both the proximal and the middle segments show the same rel-
ative increase in weight, between the terminal values, while in
5 Poor a8 a Suerasnese ie
[| PERCENTAGE WEICHT ie Pen. |
4 : = a
rl
EEE
Sono
| igaeen!
cco
Eee
fe
EEE Fresh +4
‘s
| Welght of Bones of Fore fimbs {|

CECECEPEREEEEEEEEEELELELEGLET
So 8
Chart 7 Relative fresh weight of the fore-limb bones, on the weight of all the
bones composing the fore limb (albino rat). Table 6.
@ Humerus. A Ulna plus radius. x Carpus plus foot bones.
a POT rrr
HH PERCENTAGE WEIGHT
Chart 8 Relative fresh weight of the hind-limb bones, on the weight of all
of the bones composing the hind limb (albino rat). Table 7. '
@ Femur. A Tibia and fibula. »x Tarsus and foot bones.
the hind limb the greater increase is shown by the proximal seg- a
ment, the femur. i
The first column in both table 6 and table 7 gives the body ; f
weights to which the respective bone weights correspond.
GROWTH OF THE SKELETON ARS.
TABLE 6
Percentage weight of fresh fore-limb bones represented by the humerus, ulna plus
radius, and carpus plus foot bones, on the weight of all of the bones of the fore
limb. Entered at intervals of 0.02 gram. Values taken from the smoothed graphs
ain chart 7. The corresponding body weights are also given
WEIGHT OF FRESH ULNA PLUS
BODY WEIGHT ea HUMERUS (2) RADIUS (2) FORE FOOT (2)
grams grams per cent per cent per cent
6.0 0.06 46.1 30.5 23.4

8.0 0.08 45.5 30.3 24.2
9.0 0.10 44.8 30.1 25.1
EO 0.12 44.1 29.9 26.0
13.0 0.14 43.5 29.7 26.8
15.0 0.16 42.9 29.4 OH
17.0 0.18 42.3 29.2 28.5
19.0 0.20 42.1 29.0 28.9
20.0 0.22 42.1 28.8 29.1
22.0 0.24 42.2 28.9 28.9
26.0 0.26 42.4 29.0 28.6
31.0 0.28 42.7 29.2 28.1
35.0 0.30 43.0 29.4 27.6
39.0 0.32 43.3 29.6 27.1
43.1 0.34 43.7 29.9 26.4
48.0 0.36 43.9 30.0 26.1
52.0 0.38 44.3 30.2 25.5
56.0 0.40 44.6 30.4 25.0
69.0 0.46 45.6 31.0 23.4

78.0 0.50 46.3 31.3 22.4
87.0 0.54 46.7 31.5 21.8
95.0 0.58 47.2 31.8 21.0
103.1 0.62 47.6 32.0 20.4
114.0 0.66 48 .0 32.2 19.8
126.0 0.70 48.4 32.5 19.1
137.0 0.74 48.9 32.8 18.3
148.0 0.78 49.3 33.0 Ie tf
76.0 0.88 49.8 33.9 16.7

203.1 0.98 50.1 34.1 15.8
231.0 1.08 50.5 34.7 14.8
260.0 eats 50.9 34.9 14.2
294.0 1.28 51.2 35.0 13.8
328.0 1.38 51.4 35), I 13.5
361.0 1.48 51.6 35.2 13.2
394.0 1.58 LZ 35.2 13.1
428 .0 1.68 51.9 35.2 12.9
482.0 1.84 52.0 35.3 12.7
TABLE 7
Percentage weight of fresh hind-limb bones represented by the femur, tibia plus
fibula, and tarsus plus foot bones, on the weight of all of the bones of the hind
limb. Entered at intervals of 0.05 gram. Values taken from the smoothed graphs
in chart 8. The corresponding body weights are also given
WEIGHT OF FRESH
BODY WEIGHT HIND-LIMB FEMUR (2) TIBIA PLUS HIND Foot (2)
BONES (2) FIBULA (2)
grams grams per cent per cent per cent
5.0 0.05 36.6 30.0 33.4

a (0) 0.10 36.0 29.9 34.1
9.0 0.15 39.3 29.7 35.0
11.0 0.20 34.7 29.6 35.7
12.0 0.25 34.0 29.5 36.5
14.1 0.30 33.9 29.4 37.3
16.1 0.35 32.7 29.2 38.1
19.0 0.40 32.1 29.1 38.8
21.0 0.45 32.0 29.2 38.8
24.0 0.50 32.0 29.2 38.8
27.0 0.55 32.3 29.3 38.4
30.0 0.60 32.7 29.4 37.9
33.1 0.65 33.0 29.5 37.5
37.0 0.70 33.9 29.6 Sia
40.0 0.75 33.6 29.6 36.8
43.1 0.80 33.9 29.7 36.4
EVA 1.00 35.1 30.1 34.8

71.0 1.20 36.3 30.4 33.3
87.0 1.40 37.1 30.7 32.2
106.0 1.60 37.8 31.2 31.0
126.0 1.80 39.0 31.3 29.7
148.0 2.00 40.3 31.4 28.3
175.0 2.20 41.5 eS Eas) 27.0
214.0 2.50 43.4 31.6 25.0

250.0 2.80 44.5 31.8 2350
285.0 3.10 45.3 32.0 22.7
321.0 3.40 45.4 32.2 22.4
356.0 3.70 45.5 32.4 22.1
392.0 4.00 45.7 32.6 21.7
426.0 4.30 45.8 32.8 21.4
467 .0 4.65 45.9 33.0 21.1
INCREASE IN THE WEIGHT OF THE LONG BONES
A. According to skeleton weight

If we turn now to the growth of the humerus (2), the ulna plus
radius (2), and to that of the femur (2) and tibia plus fibula (2),
it seems best to follow the increase in these bones in weight first
in relation to the weight of the skeleton. Tables 8 and 9 give
the data for the absolute and relative fresh bone weights on
skeleton weight. In charts 9 and 10 the corresponding graphs
are drawn.
t + i T T a T
nn. aay
WEIGHT IN GRAMS 7 tt rate re ro
So)! SE Sh ct Ht HH Growth of Skeleton—Albino Rat
EERE EEE aaa t ECE TH Fresh a Femur
ECE det ietalsistetaeteiast eee et }
15 Hd EEE Eo |
oececovasovasaeasvezaeseserecarasoe="
SESEEEU GEEESEESESESEE! Hert eH Tibia plus Fibu
10 BE R EEE HE art
| Eee i
t t Peet Humerus:
[ + ai
0.5
a a ig ot t i
| Uina plus 'Radiu Weight of fresh Skeleton
09 1 2 3 4 5 6 7 8 9 LOT 1251S ae Selon Te SPs 20 nail 22a 24.
Chart 9 Absolute fresh weight of the limb bones, on the weight of the fresh
skeleton (albino rat). Table 8.
x Femur. o Tibia plus fibula. @ Humerus. A Ulna plus radius.
4 se
Tibia plus Fibula
oe Coo
Sieur nl el Oy min aneOnn tam ee 1300 4 7e 1) nlO Me sem 18h 19 ar20 nln 22 2324
Chart 10 Relative weight of the fresh limb bones, on the weight of the fresh
skeleton (albino rat). Table 9.
x Femur. © Tibia plus fibula. @ Humerus. <A Ulna plus radius.
TABLE 8
Absolute fresh weight of humerus, ulna plus radius, femur, and tibia plus fibula on the
weight of the fresh skeleton. Values taken from the smoothed graphs in chart 9
Mirae 2 HUMERUS (2) ULNA tO) RADIUS FEMUR (2) TIBIA Prey FIBULA
grams gra ms grams grams grams
0.50 0.024 0.016 0.019 0.015

0.75 0.035 0.023 0.031 0.027
1.00 0.047 0.037 0.045 0.043
1.25 0.058 0.043 0.061 0.069
1.50 0.070 0.057 0.088 0.088
1.75 0.084 0.055 0.120 0.110
2.00 0.096 0.064 0.137 0.126
2.25 0.108 0.072 0.151 0.143
2.50 0.121 0.079 0.170 0.160
2.75 0.132 0.086 0.200 0.176
3.00 0:142 0.092 0.220 0.194
3.25 0.150 0.100 0.249 0.214
3.50 0.155 0.105 0.270 0.233
3.75 0.159 0.110 0.290 0.251
4.00 0.166 0.117 0.314 0.270
4.25 0.176 0.123 0.338 0.289
4.50 0.186 0.129 0.362 0.308
4.75 0.196 0.133 0.386 0.328
5.00 0.204 0.139 0.406 0.348
5.25 0.213 0.144 0.426 0.368
5.50 0.222 0.150 0.448 0.387
5.75 0.232 0.156 0.469 0.403
6.00 0.241 0.162 0.490 0.419
6.25 0.251 0.168 0.511 0.435
6.50 0.260 0.174 0.532 0.450
6.75 0.269 0.180 0.553 0.466
7.00 0.279 0.185 0.574 0.482
8.00 0.315 0.211 0.659 0.541

9.00 0.349 0.238 0.745 0.598
10.00 0.388 0.264 0.832 0.652
11.00 0.427 0.290 0.920 0.704
12.00 0.466 0.317 1.008 0.754
13.00 0.504 0.343 1.097 0.796
14.00 0.543 0.370 1.186 0.851
15.00 0.582 0.396 Wea 0.917
16.00 0.622 0.424 1.370 0.982
17.00 0.661 0.451 1.462 1.049
18.00 0.700 0.477 1.555 1.116
19.00 0.741 0.504 1.649 1.184
20.00 0.780 0.532 1.744 me belay |
21.00 0.819 0.559 1.840 1.321
22.00 0.858 0.585 1.936 1.395
23.00 0.897 0.614 2.033 1.463
24.00 0.936 0.641 2.134 1.536
TABLE 9
Relative weight of fresh humerus, ulna plus radius, femur, and tibia plus fibula
on the weight of the fresh skeleton. Values taken from the smoothed graphs in
chart 10
See eee HUMERUS (2) ULNA 1 RADIUS FEMUR (2) TIBIA eo FIBULA
grams per cent per cent per cent per cent
0.50 4.76 Sole, 3.86 3.07

0.75 4.70 3.10 4.19 3.98
1.00 4.66 ; 3.70 4.50 4.30
1.25 4.60 3.40 5.10 5.52
1.50 4.68 3.80 5.86 5.86
1.75 4.77 3.16 6.35 6.28
2.00 4.80 3.22 6.55 6.30
2.25 4.82 3.18 6.71 6.35
2.50 4.84 3.14 6.80 6.39
2.75 4.81 3.12 UPA 6.40
r 3.00 4.73 3.08 71.33 6.48
3.25 4.60 3.04 7.66 6.58
3.50 4.42 3.00 Cetl 6.65
3.75 4.24 2.96 7.03 6.69
4.00 4.16 2.92 7.85 6.76
4.25 4.15 2.89 7.96 6.80
4.50 4.13 2.86 8.05 6.85
4.75 4.12 2.81 8.12 6.90
5.00 4.08 2.78 8.12 6.96
5.25 4.06 2.74 8.12 7.00
5.50 4.04 2.72 8.14 7.04
5.75 4.04 2.71 8.15 7.01
6.00 4.02 2.70 8.16 6.99
6.25 4.01 2.69 8.18 6.96
6.50 4.00 2.67 8.19 6.92
6.75 3.99 2.66 8.19 6.90
7.00 ’ 3.98 2.64 8.20 6.88
8.00 3.94 2.64 8.24 6.76

9.00 ' 3.88 2.64 8.28 6.64
10.00 3.88 2.64 8.32 6.52
11.00 3.88 2.64 8.36 6.40
12.00 3.88 2.64 8.40 6.28
13.00 3.88 2.64 8.44 6.12
14.00 3.88 2.64 8.47 6.08
15.00 3.88 2.64 8.51 6.11
16.00 | 3289 2.65 8.56 6.14
17.00 3.89 2.65 8.60 6.17
18.00 3.89 2.65 8.64 6.20
19.00 3.90 2.65 8.68 6.23
20.00 3.90 2.66 8.72 6.27
21.00 3.90 2.66 8.76 6.29
22.00 * 3.90 2.66 8.80 6.34
23.00 3.90 2.67 8.84 6.36
24.00 3.90 BAO 8.89 6.40
As can be seen by inspecting either chart 10 or table 9, after

the age which we have designated as puberty (1.e., 100 grams
body weight and 7.4 grams skeleton weight) the percentage values
for the fresh weights of the several long bones become rather
constant. Thus table 9 shows very little change in the relative
weights of the long bones of the fore hmb after a skeleton weight
of 7.4 grams. Such change as appears is in the form of an
increase. The bones of the hind limb, on the other hand, show
greater changes. The femur steadily increases in its relative
weight, while the relative weight of the tibia plus fibula decreases
up to a skeleton weight of 14 grams, after which it increases
again slightly.
TABLE 10
Constants for the relative weights of the fresh humerus and ulna plus radius, as
these appear in the last 30 entries of table 1. These entries comprise the interval
from a skeleton weight of 8 grams to that of 24 grams, as given in table 9
MEAN RELA-
TIVE FRESH
WEIGHT ON o Cave E,
SKELETON
WEIGHT
EuMienuse (2) practi weitere eee 3.88 0.127 Boe =+().285

UOlnarplus radiis:(2)S0e3. cobs... ae 2.64 0.077 2.9 +=). 253
o = standard deviation; C.V. = coefficient of variability; E, = probable error

of C.Y.
By the use of these ratios it is possible, therefore, to determine

the weight of the skeleton from the relative weights of any of
these limb bones. Some variability has to be anticipated, how-
ever, and the question also arises whether in the skeletons of
animals reared on different diets the same ratios would hold.
The determinations of these points requires further investiga-
tion, but in the meantime the ratios presented by the humerus
and the ulna plus radius seem the most trustworthy because
they show so little change over such a long range of skeleton
weight, and because the variability shown by the observed
values is distinctly low as indicated in table 10 by the standard
deviation (c), the coefficient of variability (C. V.), and the prob-
able error of C. V. (E.), which have been calculated for the
last thirty observations in the original tables (as represented

by the last thirty observations in table 1).
As will be seen, the relative values are given here (Table
9) for the fresh bone weights only. The absolute weights for
the room dried bones are, however, entered in table 11. From
this latter table the oven-dried weights may be obtained, if
needed, by using appropriate factors which appear in table 19.
WEIGHT I CRAMS H mme.3 8

Ue Growth of Skaleton—Albino Rat
i +4 Roorn dried + t1 :
1.25) ae 1 Femur
aa t Sa
At “crpt
1.00 rot
Ct Tibia plus Fibula

75) t :
= “| Humerus
50)
]
EEE
25 ’ Ulna plus Radiu {
+ | Weight of fresh Skeleton EE
0 T VPTREIS cae elma ne (Snails AED ES) en a W/ 13s S/O} 20) 21) 227 23.0024
Chart 11 Absolute weight of the room-dried limb bones, on the weight of the
fresh skeleton (albino rat). Table 11.
x Femur. 0 Tibia plus fibula. @ Humerus. A Ulna plus radius.
B. According to body weight
The foregoing tables 8 and 9 permit us to compute the prob-

able weight of the entire fresh skeleton when the fresh weight
of one or more of the limb bones is given.
While the proportions within the skeleton are probably fairly
constant in a series of individuals, it seems possible that the
proportion of the total body weight represented by the skeleton
is subject to considerable variation. In view of this fact, it is
important to know how the weights of. the several limb bones
stand in relation to the body weight in a standard series. For
this purpose the absolute weights of the fresh, the room-dried,
and the oven-dried limb bones on body weight are given in tables
12, 13, and 14. Because these relations are useful in determin-
ing, for example, the variation in the weight of the skeleton under
TABLE 11
Absolute weight of room-dried humerus, ulna plus radius, femur and tibia plus
fibula, on the weight of the fresh skeleton. Values taken from the smoothed graphs
in chart11
WEIGHT
Seana
OF FRESH
HUMERUS (2) iBneises Pr) SAMS) FEMUR (2) TIBIA PLUS FIBULA
(2)
grams grams grams grams grams
0.50 0.007 0.004 0.004 0.004
0.75 0.010 0.006 0.006 0.006
1.00 0.014 0.010 0.0138 0.013
1.25 0.018 0.015 0.020 0.022
1.50 0.022 0.020 0.027 0.031
1.75 0.027 0.024 0.034 0.039
2.00 0.034 0.030 0.048
2.20 0.041 0.035 0.051
2.50 0.047 0.0388 0.060
2.75 0.052 0.048 0.070
3.00 0.058 0.046 0.078
3.25 0.063 0.051 0.088
3.50 0.070 0.060 0.110
3.75 0.075 0.062 0.120
4.00 0.080 0.065 0)Ss
4.25 0.085 0.070 0.140
4.50 0.090 0.075 0.150
4.75 0.095 0.085 0.165
5.00 0.100 0.090 0.180
5.25 0.110 0.095 0.195
5.50 0.115 0.105 0.215
5.75 0.125 0.110 0.235
6.00 0.130 0.115 0.255
6.25 0.135 0.120 0.275
6.50 0.145 0.130 0.295
6.75 0.155 0.135 0.310
7.00 0.160 0.140 0.330
8.00 0.190 0.170 0.405

9.00 0.225 0.190 0.475
10.00 0.255 0°215 0.550
11.00 0.290 0.245 0.615
12.00 0.320 0.270 0.685
13.00 0.350 0.295 0.750
14.00 0.380 0.320 0.815
15.00 0.415 0.345 0.880
16.00 0.445 0.370 0.945
17.00 0.475 0.395 1.010
18.00 0.505 0.420 1.070
19.00 0.535 0.445 1.135
20.00 0.570 0.465 1.210
21.00 0.610 0.495 1.295
22.00 0.650 0.520 1.380
23.00 0.695 0.545 1.470
24.00 0.735 0.575 1.550
TABLE 12
Absolute weight of fresh limb bones, on body weight. Values taken from smoothed
graphs in chart 12
BODY WEIGHT HUMERUS (2) eet Pr) ReaD TUE FEMUR (2) Te 2) TAUAGREL

5 0.03 0.02 0.03 0.03
10 0.05 0.03 0.05 0.05
15 0.07 0.05 0.10 0.10
20 0.09 0.07 0.14 0.12
25 OF 0.08 0.16 0.14
30 0.12 0.09 0.19 sive
35 0.13 0.09 OF22 0.19
40 0.14 0.10 0.25 0.21
45 0.15 ORT 0.29 0.24
50 0.17 0.12 0.31 0.26
55 0.18 0.13 0.34 0.28
60 0.19 0.13 0.37 0.30
65 0.20 0.14 0.40 0.33
70 0.22 0.14 0.43 0.35
75 0.23 0.15 0.45 0.38
80 0.24 0.16 0.48 0.39
85 0.25 0.16 0.50 0.41
90 0.26 On 0.53 0.43
95 0.27 0.18 0.55 0.45
100 0.28 0.19 0.57 0.47
110 0.30 0.20 0.62 0.52

120 0.33 0.21 0.67 0.55
130 0.35 0.23 0.72 0.58
140 0.37 0.24 0.78 0.61
150 0.39 0.26 0.83 0.63
160 0.41 OF27, 0.87 0.66
170 0.43 0.28 0.92 0.69
180 0.45 0.30 0.96 0.72
190 0.47 0.31 1.00 0.75
200 0.49 0.33 1.04 0.77
210 0.50 0.34 1.08 0.80
220 0.52 0.35 iets 0.83
230 0.54 0.37 Iba t7/ 0.86
240 0.56 0.38 1 oe4 0.89
250 0.58 0.39 1.26 0.92
270 0.61 0.42 1 ats) 0.98

290 0.65 0.45 1.43 1.03
310 0.69 0.48 152 1.08
330 0.72 0.50 1.60 1.14
350 0.75 0.52 1.68 1.20
370 0.78 0.54 Se 1.25
390 0.81 0.56 1.82 1.30
410 0.84 0.57 1.88 1.36
430 0.87 0.59 1.95 1.41
450 0.91 0.61 2.02 1.47
470 0.94 0.62 2.09 ilGR:
485 0.96 0.64 2.14 1257
263

TABLE 13
Absolute weight of room-dried limb bones, on body weight. No chart
BODY WEIGHT HUMERUS (2) BONS ares) See FEMUR (2) ae me) spn
eh StS
grams grams grsms grams grams

5 0.007 0.005 0.007 0.008
10 0.013 0.010 0.012 0.014
15 0.021 0.018 0.028 0.033
20 0.030 0.028 0.044 0.043
25 0.040 0.035 0.053 0.054
30 0.047 0.041 0.067 0.070
30 0.052 0.045 0.082 0.082
40 0.060 0.052 0.098 0.095.
45 0.067 0.060 0.120 0.113
50 0.079 0.067 0.134 0.127
5) 0.088 0.076 0.153 0.142
60 0.095 0.079 0.172 0.159
65 0.102 0.088 0.192 0.178
70 0.115 0.092 0.211 0.193
75 0.124 0.101 0.229 0.214
80 0.183 0.110 0.250 0.225
85 0.141 0.112 0.268 0.240
90 0.150 0.122 0.291 0.258
95 0.160 0.131 0.310 0.276
100 0.170 0.140 0.329 0.295
110 0.188 0.153 0.370 0.337

120 0.213 0.164 0.410 0.368
130 0.233 0.182 0.452 0.399
140 0.251 0.193 0.502 0.427
150 0.267 0.211 0.540 0.446
160 0.281 0.221 0.572 0.473
170 0.297 0.230 0.613 0.500
180 0.312 0.247 0.643 0.528
190 0.328 0.257 0.674 0.553
200 0.344 0.274 0.703 0.568
210 0.353 0.284 0.733 0.591
220 0.369 0.294 0.768 0.614
230 0.384 0.311 0.799 0.638
240 0.400 0.321 0.834 0.662
250 0.416 0.330 0.861 0.686
270 0.439 0.357 0.927 0.733

290 0.469 0.383 0.987 0.774
310 0.501 0.411 1.052 0.815
330 0.523 0.431 1.112 0.864
350 0.547 0.449 1.170 0.912
370 0.571 0.469 1.224 0.952
390 0.595 0.489 1.280 0.997
410 0.620 0.499 1.325 1.047
430 0.644 0.519 1.380 1.090 ’
450 0.675 0.540 . 1.433 1.138
470 0.700 0.552 1.490 1.190
485 0.717 0.571 1.528 1.227
TABLE 14
Absolute weight of oven-dried limb bones on body weight. No chart
ULNA uD) RADIUS FEMUR (2) TIBIAae
(2) PLUS
BODY WEIGHT HUMERUS (2)
5 0.006 0.005 0.006 0.007

10 0.012 0.009 0.011 0.013
15 0.019 0.017 0.026 0.030
20 0.028 0.026 0.040 0.040
25 0.037 0.032 0.049 0.050
30 0.043 0.038 0.062 0.064
30 0.048 0.041 0.075 0.075
40 0.055 0.048 0.090 0.087
45 0.062 0.055 0.110 0.104
50 0.073 0.062 0.1238 (Oe ilily/
55 0.081 0.070 0.141 0.131
60 0.087 0.073 0.158 0.146
65 0.094 0.081 0.176 0.164
70 . 106 0.085 0.194 0.178
75 114 0.093 0.210 0.197
80 122 0.101 0.230 0.207
85 . 130 0.103 0.246 0. 221
90 .138 0.112 0.267 0.237
95 147 0.120 0.285 0.254
100 .156 0.129 0.302 0.271
110 173 0.141 0.340 0.310

120 196 0.151 0.377 0.339
130 214 0.167 0.415 0.367
140 231 0.177 0.461 0.393
150 245 0.194 0.496 0.410
160 258 0.203 0.526 0.435
170 273 0.211 0.563 0.460
180 woio)NI 0.227 0.591 0.486
190 0.236 0.619 0.509
200 0.252 0.646 0.523
210 0.261 0.674 0.544
220 0.270 0.706 0.565
230 0.286 0.734 0.587
240 0.295 0.766 0.609
250 ecooocoocaocoosocse
0.303 0.791 0.631
270 0.328 0.852 0.674

290 0.352 0.907 0.712
310 0.378 0.967 0.750
330 0.396 1.022 0.795
300 0.413 1.075 0.839
370 0.431 1.125 0.876
390 orrsa 0.449 1.176 0.917 ~
410 0.459 1.218 0.963
430 0.477 1.268 1.003
450 0.496 1.317 1.047
470 0.507 1.369 1.095
485 0.525 1.404 1.129
different diets, it has been thought proper to give the weight

values in all three conditions of moisture. Chart 12 gives the
smoothed graphs, for the fresh weights only.
eat
-~SGS8e5)
Chart 12 Absolute weight of the fresh limb bones, on body weight (albino
rat). Table 12.
x Femur. o Tibia plus fibula. @ Humerus. A Ulna plus radius.
PERCENTAGE OF WATER IN THE SKELETON AND ITS PARTS
1. According to age
As already stated, the skeleton has been dried in two stages,

first at room temperature (18°C. to 24°C., according to season)
for thirty days, and later in the water bath at 96°C. for six days.
It is generally recommended that for chemical analysis bones be
dried at 140°C., as this temperature drives off certain fractions of
water associated with the salts, but for our purpose this was not
deemed necessary, particularly as the reduction in weight_is only
about 1 per cent greater than that obtained by the method
employed. The percentages of water here given are therefore to be
interpreted in the light of the conditions under which they have
been obtained. The data for the percentage of water in the entire
skeleton after drying at 96°C. for six days, according to age, are
given under A in the left-hand portion of table 15, and the values
entered in chart 13. The percentage after oven drying was that
chosen for tabulation because it was assumed that it would be
more constant than the percentage after room drying, and in one
sense this is true.
As chart 13 shows, there is a rapid fall of about forty points
in the percentage of water up to puberty, or to about 100 days of
age, and after that a loss of about five points during the remainder
of life.
As indicated by the discussion on the effects of the macerating
fluid (Appendix 2), the data for the percentage of water are about
0.6 per cent low at birth, and this deficiency increases about 0.04
per cent for each 5 grams increase in body weight, up to 150
Chart 13 Percentage of water in the entire skeleton—oven dried—on age

and on body weight (albino rat). Table 15.
@ Age. x Body weight.
grams, after which the deficiency remains constant. The relative

values in the different parts of the skeleton are, however, not
affected by the maceration.
During early life calcification of bone is largely a function of
age, and it seemed desirable, therefore, to compare the course of
the percentage of water in the entire skeleton when the values were
entered according to the observed ages as in the left-hand por-
tion of table 15, under A, and when entered according to body
weights, as in the right-hand portion of table 15, under B. The
corresponding graphs for the entire skeleton are given in chart 13.
As the records show, after puberty there is no marked differ-
ence brought out by the two methods of presentation, which
means merely that there is little further change in the percentage
TABLE 15
A. Percentage of water in oven-dried skeleton on age in days. B. Percentage of
water in oven-dried entire skeleton and appendicular skeleton, on body weight.
Values taken from the smoothed graphs in chart 13
c] = PERCENTAGE OF
AGE = a
a VRE =AU ERT Sie
Pere nerec! weter Body. weight Eprerniere ot water LAR SKELETON
days grams
0 79.2 5 79.1 79.1

5 77.0 10 76.0 76.0
10 73.5 15 72.4 72.0
15 69.6 20 69.0 67.1
20 66.0 25 65.4 64.0
25 62.0 30 63.1 62.2
30 58.0 35 61.2 60.4
35 54.4 40 59.1 58.6
40 52.4 45 Dice 56.8
45 50.8 50 55.2 55.0
50 49.0 55 58.2 53.2
55 47.2 60 52.0 52.0
60 46.0 65 50.7 50.6
65 44.8 70 49.4 49.1
70 43.8 : 75 48.1 48.0
75 42.8 80 46.8 46.6
80 41.8 85 45.6 45.2
85 40.8 90 44.4 44.0
90 40.0 95 43.1 42.5
95 38.8 100 41.8 41.2
100 38.2 110 39.4 38.8
110 37.9 120 38.8 37.0

120 37.6 130 38.4 35.3
130 37 .4 140 38.1 34.8
140 BY(ar 150 37.9 34.2
150 36.9 160 37.6 33)7/
160 36.7 170 37.4 33.2
170 36.5 180 37.1 32.6
180 36.2 190 36.8 32.1
190 36.0 200 36.6 31.6
200 35.9 230 35.6 31:1
230 35.2 260 35.1 30.7.

260 35.1 290 34.9 30.3 (
290 34.9 320 34.7 30.1 a
820 34.7 350 34.5 29.6 é
350 34.5 380. 34.2 29.2 3
380 34.2 410 34.1 28.7 :
410 34.1 440 34.0 28.3
440 34.0 480, 33.6 28.0
480 33.6
TABLE 16
Percentage of water in oven-dried cranium and vertebrae, on body weight. Values
taken from the smoothed graphs in chart 14
BODY WEIGHT CRANIUM VERTEBRAE
grams per cent per cent
S 80.1 80.1
10 76.8 76.8
15 72.4 70.5
20 68.4 6304
25 64.0 66.8
30 60.4 65.1
35 58.2 63.4
40 56.3 61.8
45 * 64.8 60.0
50 ' 53.1 58.4
55 52.0 56.8
60 51.0 54.8
65 50.1 53.6
70 49 .2 52.1
75 48.3 50.5
80 47.4 49.4
85 46.4 48.1
90 45.6 47.1
95 44.8 46.0
100 44.0 44.8
110 43.4 42.0

120 43.2 39.6
130 42.9 38.2
140 42.7 38.0
150 42.4 38.0
160 42.2 37.9
170 42.1 37.7
180 42.0 37.5:
190 42.0 37.4
200 41.8 37.3
210 41.6 31.2
220 41.4 37.0
‘ 230 41.3 36.9
240 41.2 36.8
250: 41.2 36.7
270 41.0 36.4

290 40.8 36.1
310 40.5 36.0
330 40.2 35.6
350 40.1 35.4
370 40.0 35.1
390 39.7 34.8
410 39.5 34.5
430 39.3 34.2
450 39.1 33.8
470 38.8 33.4
485 38.7 33.1
269
of water, either on age or on body weight. The differences in the

graphs before puberty are the result of the method of construct-
ing the chart. It should be remembered, however, that age is
probably an important factor in modifying the percentage of
water, so that the use of the body weight as the basal datum is
open to question, except where the body weight is approximately
normal for the age. Such is the case, however, in our series.

SSSSSeeeeeereee: ry
SSSeeeeeeeceurs 17
Chart 14 Percentage of water in the oven-dried cranium and in the vertebrae,

ae) =) body weight (albino rat) Table 16.
@ Cranium. x Vertebrae.
2. According to body weight
Since body weight is a much more common datum than age,

it was felt that graphs for the percentage of water on body
weight would be particularly useful, and body weight has there-
fore been used exclusively for the remaining charts showing the
loss in the percentage of water.
This percentage is somewhat different for the several panes
of the skeleton. Taking the body weight as a basis, we obtain
the series of values given in table 15 B, for the entire skeleton,
and the appendicular skeleton; in table 16 (chart 14) for the.
cranium and the vertebrae, and in table 17 (chart 15) for the
humerus, the ulna plus radius, the fore foot, and table 18 (chart
16) for the femur, the tibia plus fibula, and the hind foot.
If these several values are compared with one another, there
appear only slight differences between percentages of water in
GROWTH OF THE SKELETON Biel
the several parts at birth, the lowest values being found in bones
forming the middle segments of the limbs, but very evident
differences are found at puberty or somewhat earlier.
80 PERCENTACE OF WATER Cee

40 a Growth of Skeleton—Albino Rat TH a
Oven dried
60
30 Humerus
20
10 Carpus plus Foot Bones

00 50 100 150 200 250 300 350 400 450
Chart 15 Percentage of water in the oven-dried bones of fore limb, on body

weight (albino rat). Table 17.
@ Humerus. © Carpus plus foot bones. The graph for the ulna plus radius
coincides nearly with that for the carpus, and is therefore omitted.
re] PERCENTAGE OF WATER
BEC EE ECC CEC OCC ee COCO Ceo ee

100 150 200 250 300 350 400
Chart 16 Percentage of water in the oven-dried bones of hind limb, on body

weight (albino rat). Table 18.
@ Femur. x Tibia and fibula. oO Tarsus and foot bones.
Plt we compare the percentage of water in the appendicular with

that in the entire skeleton (table 15, B), it is seen that after
puberty the values for the appendicular skeleton are lower.
This implies, of course, that the corresponding values for the axial
TABLE 17
Percentage of water in oven-dried humerus, ulna plus radius, and bones of fore feet,
on body weight. Values talsen from the smoothed graphs in chart 15
BODY-WEIGHT HUMERUS (2) ULNA PLUS RADIUS (2) BOTH FORE FEET
grams per cent per cent per cent
5 79.8 73.8 80.1

10 75.8 70.0 74.0
15 72.2 66.0 68.0
20 69.1 62.8 61.2
20 66.3 60.1 58.4
30 64.5 57.4 56.0
35 62.7 54.8 53.8
40 60.8 52.4 51.6
45 58.9 50.4 49.2
50 57.0 48.2 47.0
55 55.2 46.0 44.8
60 54.0 43.9 42.4
65 52.8 41.8 40.1
70 51.6 39.5 38.0
75 50.4 37.8 36.0
80 49.2 36.7 34.6
85 48.1 35.6 33.2
90 46.8 34.3 31.8
95 45.6 33.1 30.4
100 44.4 32.0 28.8
110 42.2 29.7 26.8

120 40.6 28.3 26.0
130 39.0 27.3 25.0
140 37.6 26.3 24.1
150 37.3 25.3 23.4
160 37 .0 25.0 22.9
170 36.6 24.6 22.5
180 36.2 24.3 22.1
190 36.0 24.0 22.0
200 35.6 23.6 21.6
210 35.2 23.2 21.2
220 34.8 22.9 20.9
230 34.6 221.7 20.6
240 34.2 22).5 20.4
250 34.1 22.3 20.1
270 34.0 22.0 19.9

290 33.0 21.7 19.5
310 33.4 21.3 19.4
330 30.2 20.9 19.3
350 33.0 20.5 19.2
370 32.7 20.2 19.2
390 32.5 19.9 19.1
410 32.2 19.5 19.0
430 32.0 19.2 18.9
450 31.9 18.7 18.8
470 31.6 18.3 18.8
485 31.4 18.0 18.8
272
|P
TABLE 18
Percentage of water in oven-dried femur, tibia plus fibula, and bones of hind feet,
on body weight. Values taken from the smoothed graphs, in chart 16
BODY WEIGHT FEMUR (2) TIBIA PLUS FIBULA (2) BOTH HIND FEET

5 80.7 76.6 83.2
10 Wi2 73.4 CO.2
15 74.0 70.1 72.4
20 TAMA 66.8 67.8
25 69.2 64.4 64.4
30 67.6 62.4 60.4
35 65.8 60.6 58.1
40 64.0 58.8 56.0
45 62.2 56.8 54.0
50 60.4 55.1 51.9
55 58.6 53.2 49.8
60 57.3 51.4 47.6
65 56.0 50.2 45.0
70 54.8 49.2 43.2
75 53.4 48.1 41.2
80 52.1 47.0 39.4
85 50.9 46.0 37.6
90 49.6 44.8 35.8
95 48 .2 43.6 33.8
100 47.0 42.4 32.0
110 45.2 40.4 28.4

120 43.8 38.4 26.7
130 42.4 36.8 26.0
140 40.9 35.6 25.2
150 40.2 34.9 24.8
160 39.5 34.1 24.1
170 38.8 33.3 23.2
180 38.4 32.5 22.5
190 38.1 32.2 22.2
200 37.9 32.1 21.9
210 37.6 32.0 ZS
220 Si 4 31.9 21.1
230 3.3 31.7 20.7
240 31.2 31.6 20.4
250 37.2 31.4 20.1
270 36.9 31.2 19.6

290 36.6 30.9 19.3
310 36.4 30.6 19.2
330 36.1 30.3 19.1
350 36.0 30.1 19.0
370 30.7 29.9 18.9
390 35.4 29.5 18.8
410 35.2 29.2 18.8
430 35.0 28.9 18.6
450 34.8 28.8 18.5
470 34.5 28.4 18.4
485 34.4 28.1 18.4
273
must be higher than those for the entire skeleton, and table 16
shows that this is true for both the cranium and the vertebrae,
which form the greater part of the axial skeleton. On the
other hand, while the percentage of water in the proximal bones
of both the fore and hind limbs (humerus and femur, respectively)
is close to that in the axial skeleton, the percentages in the middle
and distal bones are much lower, being least in the distal group
(charts 15 and 16).
A consideration of these various differences renders it prob-
able that they are largely due to mechanical causes: the crevices
in the cranium, the cavities in the long bones, and possibly the
greater porosity of the vertebrae would tend to give these parts a
higher percentage of water than was found in the more solid
(distal) bones of the limbs—the radius and ulna and the bones
of the fore and hind feet, which lack a central cavity.
These water values, therefore, are anatomically useful, but no
general biological significance should be attached to the differ-
ences between them. Without question, the percentage of water
will vary in a marked way according to the nutritional condition
of the animal when this condition departs from the normal, as
when, for some reason, calcification is delayed or incomplete.
On the loss of water
Between birth and maturity the densest bones, e.g., those of

the feet, show a loss of sixty to sixty-five points in their percent-
age of water (tables 17 and 18). Just how this loss occurs we
do not know. It is possible, however, to get a general notion
of the process by a simple computation. If we assume that at
birth the bones in question are entirely unecalcified, but at
maturity have 60 per cent of their dry weight in the form of
salts (assumed in the following argument to retain no water)
and, at the same time, that the organic matter present main-
tains its initial water content of 83.2 (or 80.1) per cent, then
the mixture of 60 per cent salts and 40 per cent organic matter
would show at maturity 25 (or 24) per cent of water. We find,
however, about 18.5 per cent in the bones of both the fore and
GROWTH OF THE SKELETON BES
hind feet. This result suggests that the percentage of water in

the organic matter also diminishes with age, thus supplementing
the effect produced on the water content of the bone as a whole
by the deposition of salts in it. Moreover, as we know, the
salts carry with them some water, which they lose at 96°C.—a
fact which strengthens the foregoing argument. However, so
long as the exact percentage of salts is unknown, this conclusion
is merely suggestive.
On the percentage value of the loss in weight on passing from the

room-dried to the oven-dried condition
Although the determination of the room-dried as well as the
oven-dried weights has been made for all of the parts of the
skeleton, it has been deemed necessary to print the room-dried
values in only a few cases, as in tables 3 and 11.
A study of the full manuscript tables has, however, brought
out the fact that at all ages and in all parts of the skeleton there
is a nearly constant loss in water on passing from the room-dried
to the oven-dried condition. The data are given in table 19.
These results in table 19 are of interest.
TABLE 19
The percentage in weight lost by room-dried bones after oven drying at 96°C. Based
on the averages of the percentages as determined for each of the forty-two groups
ENTRIES ENTRIES ENTRIES ENTRIES AVERAGE

(eaans 7-12 Bopy | 13-28, Bopy | 29-42, Bopy| FORALL
eeerrso) Geer lituge, | Sores ||vie
per cent per cent. per cent per cent per cent
Entire skeleton........... 8.38 8.26 8.38 8.35 8.34

Aodaliskeleton. «<< +s. 8.24 8.62 8.23 8.38 8.37
Appendicular skeleton....| 8.64 7.79 8.45 8.26 8.28
Shoulder girdle and fore
[I-75 |ofS Ae ra A 9.74 8.10 8.35 8.13 8.58
Pelvic girdle and hind
ine SPARE peo cea toe «sty 8.67 Ute 8.55 8.40 8.35
Cranium with teeth...... 8.64 9.13 8.83 8.89 8.87
romerus: ((2).0 22.82.27. 8.33 7.54 8.33 8.19 8.10
Ulma and radius (2)....... 8.33 Wee 8.24 8.17 8.11
erie (2) hasten. Ds eocrac 7.89 7.60 8.48 8.48 8.11
Tibia and fibula (2)....... 7.65 7.75 8.55 8.17 8.00
PATAS HENRY H. DONALDSON
In general the loss of water in passing from the room-dried

to the oven-dried state is about 8.8 per cent of the room-dried
value, and the oven-dried weight is therefore 91.7 per cent of
the room-dried. Table 19 gives the several values, according to
the part of the skeleton, as determined for the four body-weight
groups which have been selected.
It is possible, therefore, from these table values to make an
approximate determination of the weight of the skeleton or its
parts in one state of dryness, if the weight in the other state is
given.
The most peculiar feature of this table, however, is the fact
that in no part of the skeleton is the percentage loss in the body-
weight groups from 4 to 50 grams essentially different from that
found in the heavier groups—despite the fact that in the first
group calcification is far from complete in any of the bones and
also tends to progress as the bones become older and heavier.
The full explanation of this result has not yet been obtained,
but some incidental tests indicate that in the room-dried state
both the salts found in bone and the collagen retain nearly a like
proportion of water, which is lost at 96°C.
LENGTHS OF BONES
It has been thought worth while to tabulate the lengths of

the long bones of the limbs during growth, in order to obtain the
relations between (A) the body weight or (B) the body length of
the rat, and the lengths of the respective limb bones, and also
to get an idea of the manner in which the lengths and weights of
the bones themselves change. For comparison with the data
from man these determinations are especially useful.
.
A. Lengths of bones on body weight
With a dial calipers measuring accurately to 0.1 of a muill-

meter, the lengths of the long bones, humerus, ulna, radius,
femur, and tibia were taken in the fresh, room-dried, and oven-
dried condition.
GROWTH OF THE SKELETON Pile
The symmetrical bones usually give similar values within the

limits of the error of observation. All the values reported are
means of the measurements of the right and left bones. As at
birth all the bones are incompletely calcified, they shrink in
drying enormously in the younger rats, the fresh bones losing
27 to 32 per cent of their length in the youngest group when
dried at 96°C.
With the increase in age and in calcification this loss on dry-
ing diminishes so that at a body weight of 100 grams (about
ninety days of age) the amount of loss is less than 2 per cent in
the case of all of the bones, and at a body weight of 140 grams
45 tH
LENGTH MILLIMETERS
40
35) ’
30
25 | re
Tibia HT
20 Femur—
reer Growth of Skeleton—Albino Rat Ui

Fresh Humerus
10
| Radius
5
T BODY WEICHT IN GRAMS
omy
0 0 50 100 150 200 250 300 350 400 450
Chart 17 Length of fresh limb bones in millimeters, on body weight (albino

rat). Table 20.
oO Tibia. x Femur. A Ulna. @ Humerus. + Radius.
(about 120 days) it has become less than 1 per cent; the stable
condition being attained a little less rapidly in the femur and
humerus than in the tibia and ulna. The lengths of the fresh
bones in millimeters are given on body weight in table 20, but
instead of printing the full tables for the room-dried and oven-
dried bones we have merely added a small table (21) of correc-
tions which apply to rats above 100 grams in body weight, and
by the aid of which one may obtain the room-dried or oven-dried
length of any of the bones when the fresh length is given, or indeed
may recover the length in any other state of moisture, provided
the length in one state is known.
TABLE 20
Absolute length of fresh limb bones in millimeters, on body weight. Values taken
from the smoothed graphs in chart 17
HUMERUS | ULNA RADIUS FEMUR TIBIA

BODY WEIGHT
grams mm, mm. mm, mm, mm.
5 7.0 | 8.0 6.4 7.4 7.0

10 10.2 11.8 9.0 11.0 12.0
15 12.0 14.0 Hit 12.9 15.6
20 12.9 15.5 12.0 14.4 17.9
25 13.7 16.5 12.6 15.7 19.3
30 14.3 17.4 13.2 16.8 20.7
35 14.8 18.2 13.7 W729 21.7
40 15.4 18.8 14.2 18.7 22.4
45 15.8 19.2 14.6 19.5 23.3
50 16.5 19.6 15.0 20.1 24.0
55 16.8 1929 15.5 20.9 24.9
60 17.5 20.3 16.0 21.5 25.5
65 17.8 20.9 16.3 22.0 26.2
70 18.3 21.2 16.7 22.7 26.8
75 18.7 21.7 17.2 23.2 27.5
80- 19.0 22.0 17.6 23.9 28.0
85 19.4 22.6 18.0 24.4 28.5
90 19.8 22.9 18.4 24.8 29.0
95 20.1 23.2 18.6 25.4 29.6
100 20.5 23.7 18.9 25.8 29.8
110 21.2 24.5 19.5 26.8 30.7

120 21.7 25.2 20.2 ost. 31.6
130 22.4 25.9 20.8 28.7 32.2
140 22.9 26.6 21.3 29.5 32.8
150 23.6 27.1 21.8 30.2 33.4
160 24.0 27.7 22.2 31.0 33.9
170 24.6 28.3 22.7 31.6 34.5
180 25.0 28.8 23.1 32.2 34.9
190 25.4 29.3 23.4 32.8 35.5
200 25.8 29.8 23.8 33.4 36.0
210 26.1 30.2 24.1 34.0 36.5
220 26.5 30.6 24.3 34.6 37.0
230 26.9 30.9 24.6 35.0 37.3
240 27.1 31.2 24.8 30.5 37.7
250 27.5 31.5 25.0 36.0 38.1
270 28.3 32.0 25.5 36.8 38.8

290 28.9 32.3 26.0 37.4 39.5
310 29.4 32.8 26.3 38.0 40.0
330 29.8 33.2 26.7 38.7 40.5
300 30.0 33.6 27.0 39.2 41.0
370 30.3 34.0 27.2 39.7 41.5
390 30.7 34.4 27.5 39.9 41.9
410 31.0 34.9 27.8 40.2 42.4
430 31.0 30.2 28.0 40.4 43.0
450 31.2 30.6 28.3 40.7 43.3
470 31.3 36.0 28.5 41.0 43.8
485 31.3 36.2 28.6 41.2 44.1
oye es Ws ee ee
278
—
Examination of table 21 shows that the shrinkage is somewhat

greater in the humerus and femur than in the ulna and tibia and
that in passing from the room-dried to the oven-dried state the
shrinkage is about half that which is found in passing from the
fresh to the room-dried state. The absolute values of the cor-
rections called for are, however, very small, since in the instance
where the correction is greatest, it amounts to only 1 per cent
(femur—from fresh to oven-dried) and the maximum absolute
value corresponding to this is 0.4 mm., as can be seen by looking at
the last entry for the length of the fresh femur in table 20.
TABLE 21
Percentage losses in the length of long bones on drying. Mean values for rats above
100 grams in body weight
AVERAGE PERCENTAGE LOSS IN LENGTH

ON DRYING
AMOUNT OF CHANGE
weil Ulna |Radius| Femur} Tibia
per cent|per cent|per cent\per cent|per cent
Hreshacomoompanied slogssss5. 55.2.0 see eoe ae 0.8 | Osée lente] Wer | On

Room! dried to oven dried: loss...............- 0.3 | O21) @.2 | Ws Oe
Hireshwcorovensdriledenlosses eee ell tO OA Oe LAOS ORG
B. Lengths of bones on body length
For a strict comparison, however, we should have the rela-

tions of these bone lengths to the body length of the rats. The
data are so given in table 22, which shows the absolute lengths
of the bones on body length. The smoothed graphs are in chart
18 and in chart 19, respectively. In table 23 the corresponding
relative values appear.
In the case of each bone the last thirteen entries in table 28
after 160 mm. (= 100 grams of body weight) are divided into
three subgroups and the averages of the relative bone lengths
on body lengths computed. For these subgroups we obtain the
values given in table 24.
These data indicate a slight tendency for the humerus and
the femur to become relatively longer, while the ulna, radius,

and tibia become relatively shorter. The changes are, however,

very slight, and by the use of the mean values here given it
should be possible to recover the body length of a rat over 100
grams in body weight, when the fresh length of one or more of
these bones is known.
| LT
940 60 80 100 120 140 160 180
Chart 18 Length of fresh fore-limb bones in millimeters, on body length

<x Ulna. @ Humerus. oO Radius.
BODY LENGTH MILLIMETE 4
945 60 30 100 120

jem
140 160
SESEESESESEEESsesesesssassse
180 200 220240
Chart 19 Length of fresh hind-limb, bones in millimeters, on body length

x Tibia. @ Femur.
TABLE 22
Absolute length of fresh limb bones in millimeters, on body length. Values taken
from the smoothed graphs in charts 18 and 19
BODY LENGTH HUMERUS ULNA RADIUS FEMUR TIBIA
mm. mm, mm. mm. mm, mm,
47.5 7.0 7.5 6.2 6.4 7.0

50.0 (ea ol 6.5 6.8 7.3
52.5 7.3 7.8 6.7 7.2 7.6
50.0 7.6 8.1 6.8 7.6 8.0
57.5 8.0 8.6 ipo 7.9 8.5
60.0 8.6 9.3 7.4 8.3 9.0
62.5 9.4 10.0 8.0 8.9 Yer
65.0 10.1 10.7 8.6 9.6 10.9
67.5 10.5 11.4 9.2 10.2 11-9
70.0 10.9 12.3 9.8 10.6 12.8
72.5 1 | 12.9 10.3 tL 13.7
75.0 11.3 13.3 10.6 11.5 14.4
77.5 11.6 13.6 10.9 11.9 15.2
80.0 Ve 14.0 1HUB 12.4 15.8
82.5 12.2 14.4 11.4 12.8 16.3
85.0 12.4 14.8 11.6 13.2 16.8
87.5 PAB 15.2 11.8 13.6 17.3
90.0 13.0 15.6 12.1 14.0 17.8
92.5 13.3 16.0 12.3 14.5 18.3
95.0 13.6 16.4 12.4 14.9 18.7
100.0 14.1 17.2 12.9 15.7 19.5

105.0 14.7 17.8 13.4 16.6 20.3
110.0 15.1 18.3 13.8 17.3 21.1
115.0 15.6 18.8 14.3 -18.2 21.8
120.0 15.9 1971 14.7 19.0 22.7
125.0 16.3 19.5 15.1 19.8 23.6
130.0 16.8 19.9 15.4 20.7 24.5
135.0 17.2 20.4 15.8 21.5 25.3
140.0 17-9 20.9 16.3 22.2 26.3
145.0 18.4 21.6 16.9 23.2 27.1
150.0 19.0 22.1 17.5 24.2 27.9
155.0 Lee) 23.0 18.2 25.1 28.8
160.0 20.6 23.9 18.8 26.1 29.7
165.0 21.2 24.8 19.5 27.0 30.5
170.0 22.0 25.3 20.1 28.0 31.4
175.0 22.8 26.1 20.8 29.0 32.2
180.0 23.3 26.9 21.4 29.9 33.1
185.0 24.0 27.8 22.0 30.8 33.9
190.0 24.9 28.6 22.8 31.8 34.7
195.0 25.8 29.5 23.6 32.8 30.6
200.0 26.3 30.3 24.2 33.7 36.4

210.0 27.6 31.5 25.4 39.6 38.0
220.0 28.9 32.8 26.3 37.5 39.6
230.0 30.2 34.0 27.3 39.2 41.2
240.0 31.3 35.0 28.2 40.5 42.6
250.0 32.2 36.1 29.1 41.8 44.2
281
TABLE 23
Relative length of fresh limb bones on the body length. No charts
BODY LENGTH HUMERUS ULNA RADIUS FEMUR TIBIA
mm, per cent per cent per cent per cent per cent
47.5 14.7 15.7 13.1 13.5 14.7

50.0 14.2 15.4 13.0 13.6 14.6
52.5 13.9 14.8 12.8 13.7 14.5
55.0 13.8 14.7 12.4 13.8 14.5
57.5 14.0 15.0 12.4 13.8 14.9
60.0 14.3 15.5 12.3 13.8 15.0
62.5 15.0 16.0 12.8 14.2 15.5
65.0 15.5 16.5 13.2 14.8 16.8
67.5 15.6 16.9 13.6 15.1 17.6
70.0 15.5 17.6 14.0 15.1 18.3
72.5 15.3 17.8 14.2 15.3 18.9
75.0 15.1 Nef 3% 14.1 15.3 19.2
dep 15.0 17.5 14.1 15.4 19.6
80.0 14.9 17.5 13.9 15.5 19.8
82.5 14.8 es 13.8 15.5 19.8
85.0 14.6 17.4 13.6 15.5 19.8
87.5 14.5 17.4 13.5 15.5 19.8
90.0 14.4 17-3 13.4 15.6 19.8
92.5 14.3 17.3 13.3 15.7 19.8
95.0 14.3 17.3 13.1 15.7 19.7
100.0 14.1 17.2 12.9 15e7 19.5

105.0 14.0 17.0 12.8 15.8 19.3
110.0 13.7 16.6 12.5 15.7 19.2
115.0 13.6 16.4 12.4 15.9 19.0
120.0 13.3 15.9 1c 15.8 18.9
125.0 13.0 15.6 12.1 15.8 18.9
130.0 12.9 15.3 11.8 15.9 18.9
135.0 12,7 15.1 ial 7 15.9. 18.7
140.0 12°78 14.9 11.6 15.9 18.8
145.0 12-7 14.9 1 7 Aa ra 16.0 18.7
150.0 12.7 14.7 . IL 7 (aye 18.6
155.0. PAR eyo" 14.8 tev 1622 18.6
160.0 12.9 14.9 11.8 16.3 18.6
165.0 12.9 15.0 11.8 16.4 | 18.5_
170.0 12.9 14.9 11.8 16.5 18.5.
175.0 13.0 14.9 11.9 16.6 18:4."
180.0 12.9 14.9 11.9 16.6 18.4.
185.0 13.0 15.0 11.9 16.6 Isis
190.0 13.1 15.1 12.0 ye7/ 18.3
195.0 13.2 15.1 12.1 16.8 © 18.3
200.0 13.2 15.2. 12.1 16955 Sea

210.0 1g. 15.0 12.1 17.0 18.1
220.0 its} 14.9 12.0 Iya Oe ifs} le
230.0 13.1 VAST as 11.9 17.0 17.9.
240 .0 13.0 146 2 11.8 16.9 re
250.0 12.9 14.4 11.6 16.7 a e
282
GROWTH OF THE LONG BONES IN TRANSVERSE DIAMETER
It is of interest for us to get an idea of the shape of the long

bones at different ages and to determine whether there is any
progressive change in shape between birth and maturity. It is
not possible, however, to do this precisely without a more elab-
orate study than we can make now, but it is possible by some
simple computations to obtain a rough idea of what is taking
place.
From the data on the lengths of the fresh bones (table 20)
and those on the weights of the fresh bones (table 12) we can
compute the weight of a running millimeter, treating the bones
TABLE24
The relative lengths of the several long bones on body length in rats over 160 mm. in
body length (i.e., over 100 grams in body weight). Averages given in three body
length groups
Group lL GROUP 2 GROUP 3 Peas
BONE
165-185 mm.| 190-210 mm.| 220-250 mm.| 165-250 mm.
per cent per cent per cent per cent
ER UNV GUSH ae ts aia ee ERE Oe eels o hee 12.94 13.05 13.02 13.00
UNIT EE Sn es Ohne elcet Aa ne eee 14.94 15.10 14.68 14.91
TRG VGH cence Me er Ue 11.86 12.08 11.82 11.92
Jiexan bie eee ay ee rs ne 16.68 16.85 16.90 16.81
GUS) ONE eee cywach ctBURN eer Be 18.42 18.22 17.82 18.15
as if they were solid cylinders having their axes equal to the bone
lengths.’
3 In making the computations which are required, the values just as they
appear in tables 12 and 20 have been used. Concerning these a word of com-
ment is necessary. The lengths used are those for the single bones—humerus,
ulna, femur, and tibia, respectively. The weights, however, are in each instance
for both bones—right plus left. Furthermore, the length for the ulna is used
as a divisor for the weight of the ulna plus radius (2), and the length for the
tibia as a divisor for the weight of the tibia plus fibula (2), so that the signifi-
cance of the values for a running millimeter is not exactly the same for the
humerus and femur as for the other bones with which they are compared. As,
however, it is the change in the weight of the running millimeter rather than its
absolute value which is here important, it has seemed best to use the table values
as they stand rather than develop new tables for this special purpose.
If the weight of a running millimeter as thus obtained, and

shown in charts 20 and 21, is accepted as the basis for compari-
son, it becomes easy by the use of this datum to follow the
changes in the dimensions of the cylinders which by assumption
represent the bones as they increase in size.
99 100 150 250 300
Chart 20 Weight in milligrams of one running millimeter of fresh fore-limb

bones, on body weight (albino rat). No table.
@ Humerus. x Ulna plus radius.
| Weight of one running millimeter

[ Toor
Chart 21 Weight in milligrams of one running millimeter of fresh hind-limb

bones, on body weight (albino rat). No table.
@ Femur. x Tibia plus fibula.
It will be sufficient for the present purpose to use the data

for four body weights only, namely, at 5 grams, or birth; 20
grams, weaning time; 100 grams, or puberty, and 485 grams, or
maturity.
Table 25 gives, however, not the values for the weights of a

running millimeter, but the ratios of the square roots of these
values. The square roots stand to one another as do the trans-
verse diameters of the respective cylindrical segments 1 mm. in
length, and it is the ratio of these transverse diameters (the
value at 5 grams being taken as unity) which is required for
comparison with the corresponding ratios of the bone lengths at
the same phases of growth.
TABLE 25
Giving the ratios of the square roots of the values for the fresh weights of a running
millimeter of the several long bones. These values are proportional to those for
the diameters of the hypothetical cylinders. Based on the division of the fresh
bone weights (table 12) by the fresh bone lengths (table 20)
BODY WEIGHT HUMERUS ULNA PLUS RADIUS FEMUR TIBIA PLUS FIBULA
grams
5 1 1 1 1
20 27 134 1.54 1.24
100 1.78 1.79 2 ae 1.92
485 2.67 2.68 3.56 2.88
TABLE 26
Giving the ratios of the lengths of the several fresh bones. Based on data in table 20
BODY WEIGHT HUMERUS ULNA FEMUR TIBIA
grams
5 1 1 1 1
20 1.84 1.93 1.89 2.56
100 2.93 2.96 3.46 4.25
485 4.47 4.52 5.56 6.30
If tables 25 and 26 are examined with care it will be seen that

in a general way about one-half the total increase in diameter and
two-fifths of the increase in length has occurred at a body weight
of 20 grams (weaning time) and two-thirds of the total increase in
both diameter and length at a body weight of 100 grams (puberty).
As the ratios in tables 25 and 26 show, the growth in length is
proceeding more rapidly than the growth in the computed diame-
ter. This of course is a continuation of a change in shape, which
has been in progress since the fetal period, at which time these
bones are characteristically short, thick, and clumsy.
In order to illustrate these changes, figure 1 is given in which
is shown at A, B, and C the left femur at three ages, as described
in the legend to the figure.
A B C D E
Fig. 1 Showing the form of the oven-dried femur at several different ages.
A, B, and C represent the relative lengths and the shapes of the bones desig-
nated, while D represents B photographically enlarged to the length of C and E
represents A similarly enlarged.
The data for the albino rats from which these oven-dried bones were taken
are as follows:
Bit
bs -
ee mm.
A... 46.1 33 16.2

Bee. 106.0 84 ADT
Cree 460.0 512 42.2
One notes the relative overdevelopment of the extremities of

the bones in the younger animals. At D is shown the bone B
enlarged to the length of C, and at E the bone A similarly en-
larged. When these enlarged younger bones are compared with
the mature femur, it is seen that while enlargement by magnifi-

cation does not reveal any marked differences in the thickness of
the shaft, yet it does give us a bone in which the extremities are
relatively very large. The actual process of growth brings
about, therefore, the changes in proportion which are indicated
by a comparison of E with C. The humerus also has been studied
in this way and undergoes a similar change in form.
It appears, then, that as these bones pass from the immature
to the mature condition, they become more nearly rod shaped
and less clumsy in form, and we note in passing that this change
in shape during growth opens a field for study which thus far
has been little cultivated.
What cannot be determined without elaborate investigation is
whether this distribution of the bony substance by weight repre-
sents the distribution as it would be brought out by compass
measurements, but the impression given by a preliminary study
of a series of bones, and by figure 1, is that between puberty and
maturity the very slight change in the relative linear dimensions
of the bones is such as to make the largest full-grown bones
relatively a trifle more slender than those which are half grown
and younger.
This result is, we think, contrary to the current view that
during postnatal growth bones become relatively thicker as they
become older. We may, however, be wrong in our impression as
to the current view concerning this change.
On the relative lengths of the long bones of the limbs when compared
among themselves
To obtain the relative length of the bones in the fore limb on

the length of the corresponding bones on the hind limb, we divide
the sum of the lengths of the humerus plus the ulna or plus the
radius, on the one hand, by the sum of the lengths of the femur
plus the tibia. The ratios are given as percentages in table 27
and the corresponding smoothed graphs in chart 22.
To compare the growth of the distal bone in each limb with that
of the proximal bone, the length of the ulna or of the radius has
been divided by the length of the humerus, and the length of the
tibia by that of the femur. The ratios are given as percentages
also in table 27 and the corresponding smoothed graphs are shown
in chart 22.
Vinally, the homologous bones in the two limbs have been
compared as to their growth in length; radius or ulna divided by
length of tibia, and the length of humerus by that of the femur;
(table 28, chart 23).
BEE .
GlEL) ») ||| PERCENTACE | }
+ $+ +4
120} Ir t qeletejate a TH
St ==
TT [CHEBaSGE ats
100 TH 1D
setts
f Het Ht
HH | 7
80}
one
R+H
TTF
60 t
Growth of Skeleton—Albino Rat eH
Fresh
40 Ratios of Bone length

A
u
H
Ff
F
20
00 50 100 150 200 250 300 350 400 460
Chart 22 Ratios of fresh bone lengths, on body weight (albino rat).

Table 27.
Radius + Humerus Ulna + Humerus
: Tibia + Femur . Tibia + Femur
Radius Ulna Tibia
% Humerus - Humerus Femur
There are two large relations which stand out clearly in this
comparison. During early life the length of the bones of the
hind limb increases more rapidly than that of the bones of the
fore limb (table 27, chart 22), so the percentage value for the
fore-limb bones diminishes from birth, at first rapidly, up to
about 50 grams of body weight (or thirty-five to forty-five days
of age), but after that period it remains practically constant.
TABLE 27
Ratios
of fresh limb-bone lengths, on body weight. Values taken from the smoothed
graphs in chart 22
eee ahr a HUMERUS +ULNA TIBIA ULNA _ RADIUS

FEMUR
+ Trsra | FEMUR + TIBIA FEMUR HUMERUS HUMERUS
grams per cent per cent per cent per cent per cent
5 94.2 106.8 104.8 104.8 83.0

10 88.1 100.0 122.0 116.4 93.9
15 82.4 93.6 124.8 119.2 95.2
20 78.0 89.1 125.9 120.8 94.4
25 75.6 87.4 126.1 121.6 91.8
30 U2 85.8 125.8 122.0 91.2
35 (2.2 84.0 124.4 122.4 91.2
40 71.6 82.8 123.0 122.0 O12
45 ale 82.0 121.4 120.8 91.3
50 70.9 81.2 120.1 119.8 91.4
55 70.7 80.8 119.4 118.9 91.5
60 70.4 80.1 118.4 118.8 91.6
65 GOR 80.0 117.8 118.7 91.8
70 70.4 80.0 116.9 118.6 91.9
75 70.6 80.0 116.5 118.4 92.0
80 70.7 80.0 116.0 118.2 92.0
85 70.8 80.0 115.4 Tei I 92.0
90 71.0 80.0 114.9 118.0 92.1
95 rile, 80.0 114.4 117.9 92.2
100 Ce 80.0 113.9 TLLZ/oS! 92.3
110 alee’ 80.0 112.9 117.5 92.5

120 (ANS) 80.0 WT eZ 92.7
130 Al 80.0 111.4 116.9 92.6
140 GES 80.0 110.6 116.6 92.6
150 71.5 80.0 110.0 116.4 92.6
160 11.5 80.0 109.6 116.2 92.6
170 (fl) 80.0 109.2 116.0 92.6
180 (1.5 80.0 108.8 115.9 92.5
190 TLRS 80.0 108.4 115.6 92.4
200 1.4 80.0 108.1 115.4 92.4
250 CUE) 80.0 106.4 114.3 92.2

300 71.2 80.0 105.4 See 91.5
350 71.2 80.0 105.4 113.1 90.6
400 fale? 80.0 105.4 112.8 90.1
450 a1 2 80.0 105.5 27 89.8
485 (2 80.0 105.6 112.6 89.8
The second large relation is between the growth in length of

the proximal and the distal bones of the limbs. In both the fore
and the hind limb the growth of the proximal bones is at first less
rapid than that of the distal bones, giving the peculiar form
of graph shown in chart 22. After a body weight of 20 to 35
grams the relative growth of all the distal bones becomes less
rapid, falling off steadily in the case of the tibia and the ulna,
and showing a slight fluctuation in the case of the radius.
Nolet - He
1] PERCENTAGE eaaee 10
A E SeeSe00008008
i HHH
90
70 C1
HH
50 cH "
Growth of Skelaton—Albino Rat i
4 =
a Fresh YU
30) eae ae Ratios

of Bone lengths
R
T
i
ryt
BODY WEICHT IN CRAMS
o0 50 100 150 200 250 300 350 400 450
Chart 23 Ratios of fresh lengths of homologous limb bones, on body weight

(albino rat) Table 28.
Humerus Ulna Radius
> rene Cs
Femur Tibia Tibia
When the ratios of the lengths of the humerus plus radius,

humerus plus ulna, and femur plus tibia are taken on the body
lengths, as in table 29, there appears rather a high degree of
constancy in all cases after a body length of 125 mm., equivalent
to about 50 grams in body weight.
TABLE 28
Ratios of fresh limb-bone lengths on body weight. Bones from corresponding seg-
ments of the two limbs. Values taken from the smoothed graphs in chart 23
ee HUMERUS ULNA RADIUS

FEMUR TIBIA TIBIA
5 109.0 108.6 89.6

10 102.0 100.0 79.2
15 95.9 91.4 70-6
20 90.8 88.1 67.6
25 87.9 86.0 65.6
30 86.4 84.7 64.4
35 85.0 83.4 63-9
40 83.6 82.2 63.6
45 82.0 81.6 63.2
50 81.3 81.4 62.9
55 80.8 81.3 62:7
60 80.0 81.1 62.4
65 80.0 80.8 62.2
70 80.0 80.8 62.3
75 79.9 80.8 62.4
80 19.8 80.9 62.6
85 79.8 81.0 62.7
90 79.3 81.0 62.8
95 79.2 81.1 62.9
100 79.1 81.2 63.2
110 78.9 81.3 63.5

120 78.6 81.5 64.0
130 78.4 81.6 64.4
140 78.1 81.6 64.8
150 17.9 81.8 65.2
160 AUS : 81.9 65.5
170 77.8 82.0 65.6
180 ; 77.9 82.0 65.9
190 77.9 82.1 66.0
200 77.9 82.2 66.1
210 17.9 82.4 66.2
220 77.8 82.6 66.3
230 17.8 82.8 66.4
240 legs 83.1 66.5
250 77.8 82.9 66.7
270 riexk 82.9 66.6

290 77.6 82.8 66.4
310 77.6 82.6 66.4
330 77.6 8225 66.2
350 77.6 82.6 66.2
370 1.9 82.6 66.2
390 RS 8227 66.2
410 77.4 82.8 66.1
430 77.5 82.8 66.1
450 TG 82.8 66.1-
470 : eo 82.8 66.1
485° Ei Wi $2.8 66.1
291
TABLE 29
Ratios of fresh limb-bone lengths on body length. A body length of 125 mm. is
normal to a body weight of 50 grams. No chart
HUMERUS + RADIUS HUMERUS + ULNA FEMUR + TIBIA

stale seat BODY LENGTH ~ BODY LENGTH — ‘BODY LENGTH
mm. per cent per cent per cent
47.5 27.8 30.4 28 .2

50.0 21.2 29.6 28.2
§2.5 26.7 28.7 28.2
55.0 26.2 28.5 28.3
57.5 26.3 29.0 28.7
60.0 26.6 29.8 28.8
62.5 27.8 31.0 29.7
65.0 28.7 32.0 31.6
67.5 29.2 32.5 32.7
70.0 29.5 33.1 33.4
72.5 29.5 33.1 34.2
75.0 29.2 32.8 34.5
MeO 29.1 32.5 35.0
80.0 28.8 32.4 35.3
82.5 28.6 32.3 35.3
85.0 28 .2 32.0 35.3
87.5 28.0 31.9 35.3
90.0 27.8 331Bae/ 35.4
92.5 27.6 31.6 35.5
95.0 27.4 31.6 35.4
100.0 27.0 31.3 35.2

105.0 26.8 ; 31.0 35.1
110.0 26.2 30.3 34.9
115.0 26.0 30.0 34.9
120.0 25.6 29).2 34.7
125.0 25.1 28.6 34.7
130.0 24.7 28.2 34.8

135.0 24.4 27.8 34.6
140.0 : 24.4 27.7 34.7
145.0 24.4 27.6 34.7
150.0 24.4 27 .4 34.7
155.0 24.5 27.6 34.8
160.0 24.7 27.8 34.9
165.0 24.7 27.9 34.9
170.0 24.7 27.8 35.0
175.0 24.9 27.9 35.0
180.0 24.8 27.8 35.0
185.0 24.9 28.0 34.9
190.0 25.1 28.2 35.0
195.0 25.3 28.3 - 35.1
200.0 25.3 28.4 35.1

210.0 25.2 28.1 35.1
220.0 25.1 28.0 35.0
230.0 25.0 27.9 34.9
240.0 24.8 27.6 34.7
250.0 24.5 27.3 34.4 -
292
SUMMARY OF THE OBSERVATIONS ON THE RAT
The entire cartilaginous skeleton increases in absolute weight

rather steadily from birth to the end of the record (chart 1,
table 2).
In relative weight it increases up to about 15 grams of body
weight, and after that decreases. With the method of macera-
tion here used, the cartilaginous skeleton represents in the
mature rat about 5 per cent of the total body weight (chart 2,
table 2).
~ On considering the two main components of the skeleton—the
axial skeleton and the appendicular—it appears that the ap-
pendicular skeleton is always smaller than the axial, but that
after a body weight of 15 grams, the relative weights of the two
parts become nearly constant, though the appendicular division
grows a trifle more slowly (chart 2, table 2).
When studied from the standpoint of relative weight, it is
seen that the early rise in relative weight is more marked in the
appendicular skeleton, indicating a rapid increase in the weight
of the limb bones just after birth (chart 2, table 2).
The cranium shows a steady increase in absolute weight, but
the rate of growth is only about two-fifths of that for the entire
skeleton, and the cranium therefore diminishes in its relative
weight more rapidly than the rest of the skeleton as the rat
increases in size (chart 3, table 3).
If the two divisions of the appendicular skeleton are examined
separately, it appears that the most marked growth before a body
weight of 15 grams occurs in the pelvic girdle and appendages,
indicating the relative immaturity of the hind limbs at birth, but
the two girdles come into their mature relations at puberty
(body weight, 100 grams) (charts 4 and 5, table 4).
When the data are plotted so as to give the growth of the
skeleton .on age, the general form of the graphs is sinuous, like
the graphs for the growth of the body as a whole, and in addition
the graphs for the two sexes diverge at about sixty days, when the
differences in body weight according to sex become clearly marked
(chart 6, table 5).
When the weight of all the bones in a limb is taken as the

basal value, then the relative weights of the bones in each of the
divisions can be computed. These relations are shown for the
fore limb in chart 7 and table 6 and for the hind limb in chart 8
and table 7.
In both limbs the relative weight of the distal segment dimin-
ishes from a body weight of about 20 grams, and in about the
same proportion. A comparison of the bones from the proximal
and middle segments, respectively, shows in the fore limb the
greater increase in the middle segment, while in the hind limb
it is the proximal segment which increases most (charts 7 and 8,
tables 6 and 7).
The inerease in the absolute weight of the long bones of the
limbs on the weight of the skeleton follows an apparently simple
course (chart 9, table 8), but when the relative weights are
determined, the relations, up to a skeleton weight of 8 grams
and a body weight of 115 grams, are somewhat complicated.
However, after that period the relative weights of the humerus
and of the ulna plus radius are nearly constant, while the per-
centage value for the femur rises, and that for the tibia plus fibula
first sinks and then rises again, but without reaching its initial
value (chart 10, table 9).
From the weight of either the fresh humerus or fresh ulna
plus radius it is possible, in rats weighing more than 115 grams,
to compute with a high degree of accuracy the weight of the
fresh skeleton.
Although it seems probable that the weight of the entire skele-
ton varies in relation to the body weight according to diet and
the physiological condition of the rat, nevertheless a series of
limb-bone weights on body weight is given in three conditions of
moisture, since the weights of these bones can be used as stand-
ards for reference (chart 12, tables 12, 18, and 14)
Percentage of water
The determinations of the loss of water by the skeleton and

its parts indicate a loss of fifty-five to sixty-five points between
birth and maturity, the dense bones, without cavities, showing

the smallest percentage at maturity. It seems probable that the
percentage of water in osseous tissue is fairly constant at ma-
turity, but in the several bones taken as units it varies according
to the presence of cavities, porosity, and other mechanical con-
ditions. A certain amount of variation is to be expected also
according to the degree of calcification and the possible age
changes in the water content of the collagen. Between the
room-dried and oven-dried condition there is in general and at
all ages a loss of 8.3 per cent of water (charts 13 to 16, tables
15 to 19).
Length of limb bones
Although it is somewhat incongruous to determine lengths in

relation to body weights, yet tests indicate that the fresh lengths
of the leg bones are closely related to the body weight of the
rat, and in cases where the rat is heavy for its body length, they
follow the body weight closely (chart 17, table 20).
After puberty the change in bone length on drying is less
than 1 per cent (table 21).
The limb-bone lengths on body length (charts 18 and 19,
table 22), when cast in the form of percentages (table 23), yield
a series of proportional values which after puberty (body length,
160 mm.) are fairly constant, notably for the humerus (table 24).
By the use of the data in table 24 it is therefore possible to
recover the body length of the rat from the lengths of its limb
bones.
Form of the limb bones
When, on the basis of the weight of a running millimeter

(charts 20 and 21), the shape of the bones at maturity is com-
pared with that at birth it is found that at maturity the limb
bones are relatively more slender than at birth (tables 25 and 26).
They are close to the mature proportions of length to diameter
at puberty. The comparison of these determinations with com-
pass measurements and a comparison by means of photographs
support this statement (fig. 1).

On the relative lengths of the limb bones
As is shown in chart 22 and table 27, the combined lengths of

the femur and tibia first increase more rapidly than those of the
humerus and radius up to a body weight of 40 grams, after which
the rate of increase is similar, and the length ratios remain nearly
constant. The same can be said when the ulna is substituted for
. radius in the comparison, only in this latter instance the constant
relations do not appear until a body weight of 65 grams.
When the increase in the length of the femur is compared with
that of the tibia, or that of the humerus with the ulna or radius,
it appears that in all instances the more distal bones grow the
more rapidly. After a body weight of 15 to 35 grams, however,
the growth of the distal bone becomes the less rapid and the
ratio of lengths falls, nearly regularly, to the end of the record
(chart 22, table 27).
Finally, when the lengths of the leg bones in relation to the
length of the body are determined (table 29), it is seen that after
a body length of 125 mm. (equivalent to a body weight of 50
grams) these ratios show a high degree of constancy. .
Taking all the data together, it can be stated that in many
cases the mature relations of weight and length among the parts
of the skeleton are attained at puberty or earlier.
DISCUSSION
In making comparisons with the results for the rat as Just

recorded we shall consider them very briefly in relation, first,
to the available records for mammals below man, and, second, in
relation to the records for man himself. So far as we know,
there is no study on any lower mammal with which our results
can be directly compared. The investigation nearest to our own
is that by Jackson and Lowrey (712) on the ligamentous skeleton
of the albino rat. This study was based on seven age groups
from 0 to 365 days. As shown in their table 4 and their chart,
(page 463, loc. cit.) the maximum relative weight of the skeleton
appears a few days after birth—in their ten-day group. Our data
show the same relation at six and nineteen days (table 1). There
is, therefore, a period of relatively rapid increase in the weight

of the skeleton shortly after birth which is shown in both studies.
At maturity the fresh skeleton as prepared by them represented
about 11 per cent of the body weight, while as prepared by us it
represents about 5 per cent.
That this difference is due to the method of preparation was
shown by the fact that when Professor Jackson kindly dissected
for us in our laboratory a skeleton which was duly weighed and
then further prepared by the macerating process, which we
employed, and again weighed, the several values obtained in
this test stood in the relation which has just been given.
In this connection we have also to consider the determinations
made by Lowrey (713) of the dry substance in the ligamentous
skeleton of the albino rat. If the percentages of water are de-
rived from the percentages of the dry substance, it appears that
Lowrey’s determinations give a higher water content than that
found by us, especially in the later age groups. The difference is
not great, however, and seems easily referable to the greater
amount of soft tissue present after his method of preparation,
combined on the other hand with the slight reduction in the
water content caused by the maceration of our material.
In a study of the weight of the room-dried cranium (Donald-
son, 712, table 4) determinations were made for five body-weight
groups of Albinos grown in The Wistar Institute colony. The
values there given are in close agreement with those presented
in our table 3. They are as follows (table 30):
TABLE 30
Weight of cranium, room dried; albino rat
BODY WEIGHT GROUP
125 grams | 175 grams | 225 grams | 275 grams | 325 grams
BromeDonaldsoni(’1278).-2--eeenee 1.05 1.41 ital 1.87 2.15

HrOmmet ableton ssc. sone we 1.04 il By 1.58 1.85 2.09
We have, however, not made any measurements on the crania

of our series that might be compared with the observations of
Hatai (07) on the linear measurements of the cranium in the

mature albino rat.
It may be noted that the records on the relative lengths of the
limb bones of the rat which appear in tables 53 and 54 of ‘The
Rat’ (Donaldson, 15) are based on some of the same data that
have been used in this present paper, but the corresponding
tables here given are more complete and are to be preferred to
those published earlier.
Although the studies of Falek (’54) on the skeleton of the
dog, of Weiske (’89) on the bones of birds, of Wildt (’72) and
Graffenberger (91) on the bones of rabbits, of Tribot (’06) on
the skeleton of the guinea-pig, and of Sedlmair (’99) on that of
the cat, all contain data which might be brought into relation
with our own, yet the comparisons which could be made are so
few, and are also subject to so many corrections for the effects
of species, age, sex, diet, technique, ete., that it does not seem
wise to attempt them.
It may be said, however, in a general way that there does not
appear in our results anything contradictory to those previously
reported. In the paper by Jackson and Lowrey (12, pp. 465
and 466) the data on the relative weights of the skeleton in a
series of vertebrates, man included, are given, and a number of
values derived from the extensive tables of Welcker and Brandt
(703) are also given in this list, but no critical analysis of the data
is there attempted. Such an analysis is, however, necessary
before one can interpret the relations between the values as
reported.
COMPARISON WITH MAN
The relative fresh (or moist) weight of the human skeleton is

reported by several authors, although the determinations are
usually deficient in the designation of race, of the relation of the.
body weight to the cause of death, as well as of the method of
preparation. Schwann (’43), Bischoff (’63), Dursy (63), v.
Liebig (74), Volkmann (’74), and Theile (’84) have furnished
data, and the general result is that the weight of the moist
skeleton is on the average about 16 per cent of the total body
weight. |
In the figures as published there is no conclusive evidence

for a difference according to sex or to age between birth and
maturity.
When the foregoing data for the rat are compared with those
for man, it appears that in man there is no phase of relatively
rapid growth of the skeleton in weight during the period just
after birth. Further, there does not appear in man a regular
diminution in the relative weight of the skeleton during the
period preceding puberty corresponding to the period in the rat
during which the relative weight falls from 10 per cent to 7 per
cent as shown in our table 2. It should be reiterated, however,
that the data for man are very incomplete.
At maturity the percentage weight of the moist skeleton in man
(16 per cent) is higher than that in the ligamentous skeleton (11
per cent) or in the cartilaginous skeleton (5 per cent) of the
albino rat. Since the form of the rat body is so different from
that of the human body in which the limbs are relatively much
longer, it is not surprising that the relative weights of the skeleton
at maturity should be dissimilar, but the apparent constancy of
the relative weight in man from birth on is a matter of some
interest. |
Theile (’84) made studies of the human skeleton according to
a plan fairly comparable with our own, but these were confined to
the first seven years of life. In so far as we can compare our
results with his, it appears that the relative growth of the parts
of the skeleton, compared with one another, takes place in man in
the same manner that it does in the rat.
The observations of Stratz (’09) in man show also the relatively
slower growth of the head, as noted for the cranium of the rat
(p. 299) as well as the relatively slower growth of the arms in
length. Indeed, at maturity Stratz gives the arm length as 80
per cent of the leg length, and this is the same ratio as is found
in the rat for the length of the humerus plus ulna divided by
that of femur plus tibia, as shown in table 27 and chart 22.
When the relative length of the humerus plus radius to femur
plus tibia is computed, it is found to range in the rat from 94
per cent at 5 grams of body weight to about 71 per cent at 40
grams, after which it varies but slightly (chart 22, table 27).
The corresponding determinations for man based on the data

given by Humphry (’58) and plotted by Duckworth (’04) show
a range from 100 per cent at birth to about 70 per cent in adults.
Thus the intermembral ratio is about the same in these two forms,
despite the very different use of the fore limbs in the two cases.
This similarity is of course a mere coincidence, as the corre-
sponding ratios among the Simiidae, zoologically closest to man,
are all much higher.
In the case of the radio-humeral index, the value in man is
about 74 per cent at birth and 72 per cent at maturity, while
for the rat the record, which is slightly sinuous, reaches a maxi-
mum of 95 per cent at a body weight of 15 grams, and then in
general falls to about 90 per cent at maturity (chart 22, table
27). The ulno-humerus ratio for the rat follows a like course
from a maximum of 122 per cent at 35 grams of body weight to a
minimum of 113 per cent at maturity (chart 22, table 27).
If we turn now to a consideration of the bone relations in the
leg, we find that the tibio-femoral index for man ranges from
about 81 per cent at birth to 80 per cent at maturity, while for
the rat the corresponding ratios are 126 per cent at 25 grams of
body weight and 106 per cent at maturity. Despite the fact,
therefore, that the intermembral ratios are rather similar, the
radius, ulna, and tibia all have a greater relative length in the
rat than in man.
The relation of the stature in man at maturity to that of the
several long bones of the limbs has been carefully worked over by
Manouvrier (’92). His tables are given by Testut (’96), and
further refinements in the application of the data have been
elaborated by Pearson (’99).
Owing to the difference in the form of man and the rat, no
really comparable measurement can be made on the rat, but if it
is desired to recover the body length (nose-anus length) of the
rat in any case from the length of one or more of the leg bones,
this can be done by the aid of table 29, which gives the ratios
of the body length in relation to the sum of the lengths of the
humerus plus radius, humerus plus ulna, or femur plus tibia.
In all these cases this ratio becomes nearly constant at a

body length of 125 mm. (body weight, 50 grams), or some time
before puberty.
In the foregoing summary of our observations on the rat, the
results to which attention is especially called have been noted.
To this summary have been added several comparisons with the
data on other mammals, including man. It remains now only
to make a few general statements.
GENERAL CONSIDERATIONS
Our records do not show just how long the growth of the skele-
ton in weight continues in the rat, but from the data at hand we
should say it was still growing at 474 days of age, which, accord-
ing to our usual computation (Donaldson, 715, p. 6) is equiva-
lent to thirty-nine years of human life. The only datum for
man with which this can be compared is the linear measurement
represented by the stature, which seems to reach its maximum
at about twenty-eight years in the human male and twenty-five
years in the female. If the increase in the weight of the human
skeleton ceases at the time when the increase in stature stops,
then it is clear that the growing period for the skeleton of the
rat is much longer continued, and this conclusion agrees with
our general impressions concerning the growth of this animal.
It is to be noted, however, that the mature relations among the
parts of the skeleton are established for the most part at puberty
or earlier, while the weight of the skeleton as a whole, relative
to the body weight, tends to decrease slowly as the rat becomes
larger.
During the period between birth and puberty there is, how-
ever, an interesting change in the form of the long bones which
we have examined in a preliminary way by the study of the
weights of a running millimeter at different ages.
When the changes in the weight-length relations during the
growth of the rat are followed by dividing the values in table 26
by the corresponding values in table 25, it appears that at
maturity the growth in length is about 1.8 times that in diameter,
while at puberty it is about 1.7 times, and at weaning about 1.5
times. We may infer, therefore, that the change in proportion

takes place during the period of rapid growth preceding puberty.
This is the period of calcification, and it seems most probable
that the change in form accompanies the change in material, as
the bones pass from a state of cartilage, a comparatively weak
material, to that of the completely ossified bone, which is remark-
ably strong, as well as possessing other mechanical properties
that are noteworthy.
The fact that the long bones at maturity are more slender than
at puberty, or earlier, is a result of interest in view of the state-
ments made here and there that in order to maintain like struc-
tural strength, when enlarged, a bone must show an increase
in diameter which is relatively greater than the increase in
length. The earliest attempt to state these relations seems to
have been made by Galileo (1638).
On page 130 of the translation of his work by Crew and
Salvio (see Galileo) is given a figure showing the computed
diameter of a larger bone three times the length of a smaller one.
The larger bone there depicted is monstrous. In this case Galileo
proceeded on the assumption that the material was the same in
both bones—that the dimensions of the smaller bone enabled it to
just resisting breaking when supported at one end—and that the
strength of the larger bone, when so-determined, was equal to
that of the smaller.
On these assumptions his conclusion is valid and the dimen-
sions shown are justified, but the selection of bones to illustrate
these relations was unfortunate and misleading because osseous
tissue has qualities quite different from those which were assumed
for the argument.
The data presented have two different uses; one, to show the
manner in which the weight and length of the various parts of
the skeleton increase, and the other, to give a series of values
to which subsequent observations can be referred for comparison.
A word of comment on this second use is in order. Despite the
fact that the bones are so dense and give so direct an impression
of immutability, yet a moment’s consideration serves to make it
plain that like the other systems of the body the osseous system
is responsive to nutritive conditions, and in reality is much

more plastic than one might at first be inclined to suppose. We
do not allude to the changes induced by recognized disease (Wolff,
’92) or starvation (Voit, ’05) or the great loss in the weight of
the bones in extreme old age in man, as, for example, in the case
of the centenarian described by Waldeyer (710), a loss which is
merely an extreme instance of the senile atrophy that is generally
recognized, nor to those atrophies which appear in the skull of
the horse (Ussow, 01), of old dogs (Schmey, 715), and in the
brain case of old mustelidae as described by Thomas (’86),
nor the modifications in the skulls of ‘park reared’ lions as
reported by Hollister (17), but to those differences in the weights
or lengths of the bones which depend on diet in the narrower
sense and which may be present during the prime of life.
It has been shown by Burnett (’08, 711) that the weight,
strength, and composition of the bones of the pig can be modified
in a very striking way by foods of different composition, and
Weiske (’95) brought about changes in the chemical composition
of the bones in rabbits by modification of the diet. Lactation in
cows changes the salt content of the bones (Forbes, 718) and in
rats which are underfed at an early age calcification can be at.
least greatly retarded. Moreover, observations by Weiske (’95))
on rabbits, by Waters (08, ’09) on cattle, and by Aron (’11)
on dogs have shown that during prolonged starvation of growing
(but not very young) animals the long bones of the limbs still
continue to increase in length, and the entire skeleton to increase
in weight, even when the animal as a whole is losing weight.
It is worth noting in this connection that the skeleton and the
central nervous system, despite the very different ways in which
they grow, are alike in their ability to increase in weight under
conditions which may bring about a loss in total body weight.
In addition to these differences there is of course some varia-
bility, which must be expected, even among animals living
under the same conditions. This sort of variability is indicated
on our charts by the relation of the several entries to the
smoothed graphs.
The postnatal asymmetries which appear to arise largely from

use in man (Gaupp, ’09) are not evident in the rat, and there
is nothing systematic in the slight deviation which we have
observed for the values of symmetrical bones. But it seems not
improbable that the weight of the skeleton is very responsive to
nutritional conditions, even when these vary only within limits
usually considered normal.
The capacity of the mammalian skeleton to respond in this
way would seem to be indicated by the changes in skull and face
measurements of the descendants of immigrants to the United
States, shown by the observations of Boas (’11) and is implied
by the use of the Bertillon system of measurements for identifi-
cation (’93)—a system founded on the idea of variability within
the race.
From all this it follows that the tables here given must be
expected to furnish standard values only for albino rats reared
under conditions very similar to those applying to the rats used
for the tables, and the skeletons of which have been prepared in a
similar manner, while for rats otherwise reared our tables furnish
merely reference values with which the new data may be com-
pared. How the feral Norway rat, living under the usual con-
ditions, will differ from the domesticated Albino here studied in
these skeleton characters we do not yet know, but one set of
observations (Donaldson 712) shows the cranium to be heavier
in the Norway than in the Albino.
Finally, it is not without interest to note that some characters
of the human body which meet with aesthetic approval, and which
depend on the proportions of the skeleton—the small head, well-
formed mandible, length of limbs, and small feet and hands—are
all of them characters which represent completed growth in both
man and the rat, and thus in approving of them in man, we
approve not only of constitutional vigor, but also of characters
which man shares with other mammals.
LITERATURE CITED
Aron, Hans 1911 Nutrition and growth: I. Philippine J. of Science, vol. 6,

no. 1, section B, Medical Sciences.
Bertitton, ALPHONSE 1893 Identification anthropométrique. Instructions
signalétiques. Imprimerie Administrative-Melu n, 148 pp.
Biscuorr, E. 1863 Einige Gewichts- und Trocken-Bestimmungen der Organe
des menschlichen Korpers. Zeitschr. fiir Rationelle Medicin, Bd. 20,
S. 75-118.
Boas, Franz 1911 Abstract of the report on changes in bodily form of de-
scendants of immigrants. The Immigration Commission, Govern-
ment Printing Office, Washington, D. C.
Burnett, E. A. 1908 Part I. Foods supplementary to corn in fattening pigs.
Part II. The effect of food on breaking strength of bones. Bull. 107
of the Agr. Exp. Station of Nebraska, vol. 21, article 1.
1911 The effect of food on the strength, size, and composition of the
bones of hogs. 24th Ann. Rep. Agric. Exper. Station, Nebraska, pp.
178-208.
Donatpson, H. H. 1906 A comparison of the white rat with man in respect
to the growth of the entire body. Boas Anniversary Volume, pp.
5-26. G. E. Stechert & Co., N. Y. .
1912 _On the weight of the crania of Norway and albino rats from
three stations in western Europe and one station in the United States.
Anat. Rec., vol. 6, pp. 53-63.
1915 The Rat. Reference tables and data for the albino rat (Mus
norvegicus albinus) and the Norway rat (Mus norvegicus). Memoirs
of The Wistar Institute of Anatomy and Biology, no. 6.
Duckxwortu, W. L. H. 1904 Morphology and anthropology.’ A handbook for
students. University Press. Cambridge.
Dursy, Emr 1863 Lehrbuch der systematischen Anatomie. Lahr.
Fatcx, C. Pu. 1854 Beitriige zur Kenntniss der Wachstumsgeschicht e des
Tierkérpers. Archiv fiir path. Anat., Bd. 7.
Forses, E. B., Hatverson, J.O. anpD Moraan, L. E. 1918 The mineral metab-
olism of the milch cow. Third paper. Ohio Agr. Exp. Station Bull.
330, pp. ‘91-134.
GauiILeEo GauILeE! Linceo 1638 Discorsi e Dimostrazioni Matematiche, in-
torno a due nuove scienze. Attenenti alla Mecanica & i Movimenti
Locali. Appresso gli Elsevirii in Leida. Translated by Henry Crew
and Alfonso de Salvio ‘‘Dialogues concerning two new sciences.’’
The Macmillan Co., N. Y. 1914. See p. 130 et seq., p. 131, fig. 27.
Gavurp, E. 1909 Die normalen Asymmetrien des menschlichen Ko6rpers.
Gustav Fischer. Jena.
GRAFFENBERGER, L. 1891 Uber die Zusammensetzung der Kaninchenknochen
im hohen Altersstufen. Landwirtschaftliche Versuchsstationen, Bd.
39, S. 115.
Harar, 8S. 1907 Studies on the variation and correlation of skull measurements
in both sexes of mature albino rats (Mus norvegicus var. albus). Am.
Jour. Anat., vol. 7, pp. 423-441.
Houuister, N. 1917 Some effects of environment and habit on captive lions.
Proc. of the U. S. Nat. Museum, vol. 53, pp. 177-193.
Hiitsen, C. 1898 Spezifisches Gewicht, Elastizitit und Festigkeit des Knoch-

engewebes. Bull. labor. Biol. St. Petersb., Bd. 1, pp. 7-37. Abstract
in Jahresberichte iiber die Fortschritte der Anat. u. Entwck., Bd. 4,
p. 146.
Hiiusen, Kart 1902 Die Druckfestigkeit der langen Knochen. (Die mechan-
ische Bedeutung der Beziehung der Linge zum Diameter der Knochen. )
Verhandl. des V. Internat. Zoologen Congresses zu Berlin, 1901.
Humpury, Grorcre Murray 1858 A treatise on the human skeleton (including
the joints). Macmillan & Co. Cambridge.
Jackson, C. M., anp Lowrey, L. C. 1912 On the relative growth of the com-
ponent parts (head, trunk and extremities) and systems (skin, skele-
ton, muscu'ature and viscera) of the albino rat. Anat. Rec., vol. 6,
pp. 449-474.
LararseT, A. 1912 Résultats expérimentaux sur l’accroissement des os en
longueur. C. R. Ass. Anat. Réun, 14, pp. 72-91.
Lizpic, G. v. 1874 Gewichtsbestimmungen der Organe des menschlichen
Korpers. Archiv fiir Anatomie, Physiologie und Wissenschaftliche
Medicin, pp. 96-117.
Lowrey, Lawson G. 1913 The growth of the dry substance in the albino rat.
Anat. Rec., vol. 7, pp. 143-168.
Manovverigr, L. 1885 Sur les proportions pondérales du squelette des mem-
bres chez homme et chez les anthropoides. Proc. .of the L’ Assoc.
France. p. l’Av. d. Se’s., Congres de Grenoble, 1885, pp. 516-521.
1892 Mémoire sur la détermination de la taille d’aprés les grands os .
des membres. Mem. de la Soe. d’Anthropol., T. 4, Paris, pp. 347-411.
Mooptigr, Roy L. 1917 Studies in paleopathology. I. General consideration
of the evidences of pathological conditions found among fossil ani-
mals. Winter number, Annals of Medical History, pp. 374-393.
Pearson, Kart 1899 Mathematical contributions to the theory of evolution.
V. On the reconstruction of the statute of prehistoric races. Philos.
Trans. of the Royal Soc. of London, vol. 192, series A, pp. 169-244.
ScHEIN, Grora 1913 Untersuchungen iiber die chemische Zusammensetzung
des Knochens an verschiedenen Kérperstellen und bei verschiedenen
Behandlungsmethoden (Halle). Leipzig, Inaug.-Diss. 64 pp.
Scumey, M. 1915 Uber die Verainderungen am Skelettsystem, insbesondere
am Kopfe, bei senilen Hunden. Archiv fiir path. Anat. Bd. 220, 8.
52-62.
Scuwann, TH. 1843 Observations des phénoménes périodiques. Mesures des
organes internes du corps humain. Nouveaux Mémoires de |’ Acad.
Royale de Bruxelles, T. 16, pp. 51-52. For plan and method refer
to Bull. de l’Acad., T. 9, no. 7, 1842.
Sepiumair, A.C. 1899 Uber die Abnahme der Organe, insbesondere der Knochen
beim Hunger. Zeitschr. fiir Biologie, Bd. 37.
Srratz, C. H. 1909 Wachstum und Proportionen des Menschen vor und nach
der Geburt. Archiv fiir Anthropologie, Bd. 36, S. 287-297.
Sue 1755 Sur les proportions du squelette de l’homme, examiné depuis | Age
le plus tendre, jusqu’A celui de vingtcing, soixante ans, et au dela.
Mémoires de Mathémat. et de Physique, Présentés a l’Acad. Royale
des Sciences, Paris, T. 2, pp. 572-585.
Testut, L. 1896 Traité d’anatomie humaine. Anatomie descriptive—His-

tologie—Développement. -T. 1, see pages 2-4. Octave Doin, Paris.
TueiLe, F. W. 1884 Gewichtsbestimmungen zur Entwickelung des Muskel-
systems und des Skelettes beim Menschen. Durch eine biographische
Notiz eingeleitet von W. His. 4° Halle, E. Blochmann u. Sohn in
Dresden.
Tuomas, OLDFIELD 1886 Notes on a striking instance of cranial variation due
to age. Proc. of the Zool. Soc. of London, pp. 125-127.
Trisot, J. 1906 Sur les chaleurs de combustion et la composition des os du
squellette, en fonction de l’Age, chez les cobayes. C. R. Acad. Se.
Paris, T. 142, pp. 906-907.
Ussow, 8S. S. 1901 Uber Alters und Wachsthumsverinderungen am Knoch-
engeriist der Haussiiuger. Arch. f. wissensch. u. prakt. Thierh. Berl.,
Bd. 27, S. 339-394.
Vinson 1904 Beitrige zur Methodik der Analyse ganzer Tierkérper. Inaug.-
Diss. G6ttingen.
Voir, Erwin 1905 Welchen Schwankungen unterliegt das Verhiltnis der
Organgewichte zum Gesamtgewicht des Tieres? Zeitschr. fur Biol.,
Bd. 46.
Votkmann, A. W. 1874 Untersuchungen iiber das Mengenverhiltniss des
Wassers und der Grundstoffe des menschlichen K6rpers. Ber. d.
Sachs. Gesell. d. Wissen., Math.-.Phys. Kl., 14 Nov., S. 202-247.
Watpryer, W. 1910 Das Skelett einer Hundertjihrigen. Sitz.-Ber. Akad.
Wiss. Berlin, S. 971-979.
Waters, H. J. 1908 Capacity of animals to grow under adverse conditions.
Proc. of the Soe. for the Promotion of Agr. Science, U.S. A.
1909 The influence of nutrition upon the animal form. Proc. of the
Soc. for the Promotion of Agr. Science, U.S. A.
Weiske, H. 1889 Untersuchungen iiber Qualitét und Quantitit der Vogel-
Knochen und Federn in verschiedenen Altersstadien. Landwirt-
schaftliche Versuchsstationen, Bd. 36, S. 81.
1895 Weitere Beitrige zur Frage iiber die Wirkung eines Futters mit
sauren Higenschaften auf den Organismus, insbesondere auf das
Skelett. Zeitschr. fiir Physiol. Chemie, Bd. 20.
WEISSENBERG, S. 1906 Die Korperproportionen des Neugeborenen. Jahrb.
Kinderheilk., Bd. 64, S. 839-847.
WELCKER, HERMANN, AND BRANDT, ALEXANDER 1903 Gewichtswerthe der
Korperorgane bei dem Menschen und den Thieren. Archiv fiir
Anthropologie, Bd. 28, S. 1-89.
We ts, H. Gipzon 1911 Calcification and ossification. Archives of Internal
Medicine, vol. 7, pp. 721-753.
Werzet, G. 1910 Volumen und Gewicht der Knochen als Massstab fiir den
phylogenetischen Entwicklungsgrad. Die Porositaét der Knochen.
Archiv fiir Ent. der Organismen, Bd. 30, 8. 507-537.
Wiupt, Eucen 1872 Zusammensetzung des Knochen des Kaninchens in ver-
schiedenen Altersstufen. Landwirtschaftliche Versuchsstationen, Bd.
15, S. 404. Inaug.-Diss.
Wotrr, Juttus 1892 Das Gesetz der Transformation der Knochen. August
Hirsehwald. Berlin. 12 plates.
APPENDIX 1
SKELETON—ALBINO RAT
The accompanying table 31 gives the parts of the skeleton fol-
lowed by the number, or groups, of bones separately weighed in
TABLE 31
Parts of skeleton
NUMBER TOTAL
wEton- we
DIVISION NAME ee eeae

wicuep |PREPARED
as al GROUP
Cranium (with 6 ear bones

Skull and 8 teeth)’. wu ens eae 1 1
Mandibles (with 8 teeth)... 1 2
: Ely Ord eee eensSues 1 1
etna Vertebrae of trunk 1 f 30
Vertebrae of tail 304
Ribs 1 26
Sternum Siero re ewe 6 ip vale a (oe 16)» 16) ete \ 6
Scapular yeyeeul. . 22 2 2
Claviclete ev ee ee 2 2
Shoulder UITNVE RUS Presa aisle Se eee 2D 2
girdle and Wilmtaee error ch a ee 2
appendages’ (|| Radiusiae..i.c 0.0. Coe ee 2 2
Carpus and bones of fore-
SCTE) 0sees pra 1 er hah 2 56°
Appendicular
skeleton Relvisi aerate. teen oe 1 2
Hemunyee etre cee eee 2 2
Poleie wicdie 2 sesamoid bones: behind
ee e ha distal end of femur....... 2, 4
ae ala Patellser- 2.85. cee 2 2
8 Mibiarand: fibulaeeeaee cea see 2, 46
Tarsus and bones of hind
ICYSL re Re Pa ge) cnrese Le Pi 545
FROtAISP aes aren eee ee ee ca ees Sa ea, ads te 28 230
‘The number of caudal vertebrae is somewhat variable (twenty-eight to

thirty), but in the mature rat there are about thirty. On the ventral aspect
of the caudal vertebrae—in the intercentral position—are a series of pairs of
the case of each skeleton, together with the total number of

separate bones in each group as these were prepared.
If we take the bones more in detail the following numbers
appear:
TABLE 32
NAME |NUMBER OF BONES
(OLN ANI a0 here ecnees in Cis REO ee To raAEA Onision rth he oti coche cecad 28
[nie [Soar ae, 5S eee RE REPRE ane acne nd cas aie Senne 6
TReeratHl ye 2 eRe SB Bc ich Oh RTE eS Wet oc Brera teatro Cag ea 16
TEXEIISTIS) acs SARA Soule Genet ck R CeeR Ne ce PRtek pen ceSiri i ora 6
Remaining bones as listed above (i.e., 230, less cranium, 1,
SDIMPIERVIS S12 aids ue Wl necty tose F~ ayes Soe hes hehehe mone re es 220
TINTS a I Pe et 8 Oe Lee en Se ere A Co Ge 283
There are twenty claws—also not weighed with the skeleton.

In the case of forty skeletons the percentage of the weight of
the room-dried claws on the weight of the fresh skeleton gives
the following values:
TABLE 33
Weight of claws
MEAN PERCENTAGE
NUMBER OF CASES BODY WEIGHT—RANGE ii am tepel serous oF 20peak

SKELETON
grams grams per cen
5 4—- 13 0.4- 1.5 0.30

2 20- 27 2.0— 2.5 0.21
14 30-110 3.0- 8.0 0.15
9 110-160 8.0-11.0 0.14
10 170-413 11.5-22.3 0.13
small ossicles which represent the chevron bones. About twenty-six of these
pairs are found in the tail of the full-grown albino rat. These chevron ossicles
were not counted in enumerating the bones, nor were they weighed with the
vertebrae.
5 There are fourteen small sesamoid bones in each fore foot and fourteen small
sesamoid bones in each hind foot. These were not weighed.
6 The tibia and fibula are counted separately, but have been weighed together,
because at twenty days of age, or sometimes earlier, the fibula unites with the
tibia, and after that age it becomes increasingly difficult to separate these two
bones.
APPENDIX 2
TECHNIQUE OF PREPARATION
Theoretically, one should use as a standard for determining
the effect of different methods of preparing the skeleton, fresh
material which has been cleaned by mechanical methods only,
since the immersion of the bones in any solution at room tem-
perature or above necessarily alters their weight more or less.
It is, however, not feasible to make preparations without the
use of some solvent for the connective tissue, and in the prepa-
ration of all of the skeletons included in this study a solution
of commercial ‘Gold Dust Washing Powder’ was used. The
approximate composition of this is: Sodium carbonate, 45 per
cent; soap powder, 30 per cent; water, 25 per cent.
The rat was roughly dissected and the parts placed in 50 to
200 ce. of a hot Gold Dust solution in tap-water. For mature
rats a 2 per cent solution at 96°C. was used for an hour, more or
less, the material being watched and tested. The object to be
attained was a complete softening of the attachments of the
muscles, so that they could be readily removed with the bone
scraper, or brush used for cleaning the bones. The younger the
rat the weaker was the solution, the lower the temperature
used, the shorter the time and the more careful the watching.
When bones are immersed in the macerating fluid, which acts
mainly on the surface, it seems probable that those which are
large, dense, mature, or cubical will be less affected than those
which are small, porous, young, or flat, and our tests show this
to be the case. We have therefore chosen as typical and as
giving values which can be applied to the skeleton as a whole,
the humerus and the femur.
When the macerated humerus and femur from one side of a
rat were compared with those from the other side, freshly cleaned
by cutting and scraping, but not macerated, the method of
maceration here used causes a loss in the weights of these bones
as given in table 34.
The data in table 34 are given according to the body weights
of the rats, these latter being considered normal to the age.
i
i
GROWTH OF THE SKELETON aut
TABLE 34
Percentage of loss ‘n weight of bones after macerating in 2 per cent ‘Gold Dust Wash-
ing Powder.’ Based on the combined data for the humerus and femur. Entered
according to body weight for each sex
LOSS—MALES LOSS—FEMALES
BODY WEIGHT
Fresh Oven-dried Fresh Oven-dried
grams per cent per cent per cent per cent
5 15.85 12.85 15.70 12.70

10 14.30 11.40 14.40 11.40
15 13.40 10.45 13.45 10.45
20 12.35 9.10 12.65 9.50
25 11.70 8.50 11.95 8.75
30 11.05 8.00 11.35 8.25
35 10.65 7.60 10.85 7.80
40 10.20 (e20 10.40 7.40
45 9.85 6.90 10.00 7.00
50 9.50 6.70 9.65 6.65
55 9.30 6.35 9.30 6.40
60 9.00 6.05 8.95 6.10
65 8.70 5.85 8.65 5.80
70 8.55 5.65 8.50 5.99
75 8.35 5.40 8.25 D225
80 8.20 5.15 8.05 5.00
85 8.05 4.95 7.85 4.70
90 7.85 4.70 7.65 4.55
95 7.70 4.60 7.45 4.35
100 7.50 4.40 7.30 4.15
110 7.20 4.10° 7.00 3.79

120 7.00 3.80 6.70 3.40
130 6.70 3.50 6.40 2.95
140 6.50 3.15 5.80 2.30
150 6.20 2.60 9.35 1.85
160 5.80 2.25 4.95 1.45
170 5.40 1.90 4.55 1.15
180 5.10 1.55 4.15 0.70
190 4.70 1.30 3.90 0.50
200 4.45 1.05 3.60 0.25
210 4.20 0.75 3.45 0.08

220 4.00 0.55
230 3.80 0.35
240 3.60 0.20
250, 3.45 0.10
Ae HENRY H. DONALDSON
It appears from this table that the loss in fresh weight drops
from about 16 per cent at birth to about 7.5 per cent at a body
weight of 100 grams, and after a body weight of 200 grams the
weight of the macerated may be below that of the unmacer-
ated bones—from 4.5 to 3.5 per cent. When, however, the oven-
dried weights of these same bones are compared, it is seen that
while the losses after oven-drying follow the same course as was
followed in the case of the fresh bones, yet the values run very
consistently—about 3 to 3.4 per cent (on the average) below
those for the fresh bones.
EFFECT OF MACERATION IN ‘GOLD DUST WASHING POWDER’ ON

THE PERCENTAGE OF WATER IN THE OVEN-DRIED BONES
Maceration reduces the weight of the moist (fresh) bones, as in

table 34 under ‘fresh.’
If this treatment merely dissolved some of the bone and left
the remainder unmodified in constitution and in water content,
then the bones oven-dried after maceration would have the same
percentage of water as those mechanically cleaned. The oven-
dried bones, however, show after maceration less water than those
mechanically cleaned.
At birth the oven-dried macerated bones show 0.6 per cent
less water, and this deficiency increases at the rate of 0.04 per
cent for each 5 grams of body weight up to 150 grams. At this
body weight it amounts therefore to 1.8 per cent and the same
deficiency persists in the heavier rats.
The complete explanation of these changes in weight and
water content as the result of maceration cannot at present be
given, but it appears that in general the young bones suffer
most in the destruction of their substance by the macerating
fluid, whereas with increasing age the reduction of the percentage
of water rises to about puberty, after which it remains nearly
constant.
GROWTH OF THE SKELETON Sts
EFFECT OF THE INFLUENCE OF THE MACERATING FLUID ON THE

USE OF THE DATA
In the first place, it is evident that if we desire to obtain the

true weight of the skeleton or any of its parts—using the values
for the dissected bones as standards—it will be necessary to
correct the observed values of the macerated bones in accordance
with the percentages of loss given in table 34.
When, however, it is a question of the relative weights of the
various parts of the skeleton among themselves, then we may
safely assume that the effects of the macerating fluid are so similar
on the several parts, that the relations found would be substan-
tially those existing among the fresh bones.
Finally, if it is desired to compare the parts of the skeleton in
any rat which has been subjected to conditions that might
modify the weight or condition of the skeleton during life, then
the observed values given in our tables may be used as a basis
for this comparison, provided the skeleton of the test rat has
been prepared by the same or a similar process of maceration.
It appears therefore that, despite the very considerable modi-
fications induced by the use of the macerating fluid in the
weights and water content of the bones, the observed and uncor-
rected values are nevertheless trustworthy for showing the rela-
tions between the parts, and may be used also as standards for
comparison with other skeletons prepared in a like manner.
All the results in this paper are subject to at least the fore-
going corrections, but such corrections have not been made.
The values given in the tables are based on those which were
observed.
WEIGHING AND MEASURING
The various groups of bones were weighed immediately after

cleaning (giving the ‘fresh weight’), they were then dried for
thirty days or more at room temperature, and again weighed
(giving the ‘room-dried weight’), and finally were dried at 96°C.
for six days and weighed (giving the ‘oven-dried weight’).
Measurements of the length of the long bones were also made
with a dial calipers reading to 0.1 mm., in the three different
conditions of moisture.
All of the bones of each of the 106 skeletons have been

preserved.
It may be noted in passing that bones are somewhat hygro-
scopic and the oven-dried weights are not maintained for any
length of time under usual laboratory conditions.
As arule, the bones of the rat thus macerated are not markedly
greasy and do not show evident fat. There were a few excep-
tions to this, however (nos. 50 and 66 in this series), but no
correlated peculiarities were observed in such skeletons. When
bones were heated to 96°C. for six days the glass weighing
bottles sometimes became cloudy on the inner surface. This
cloud formed a turbid mixture with water and was dissolved by
aleohol. In no case, however, did it amount to one-tenth of 1
per cent of the weight of the bones, but nevertheless a correction
was made for it.
It may be of interest to note that several trials gave eleven
working hours as the time needed to prepare a skeleton—
including the fresh and room-dried weighings and the necessary
records. The weighings and the records for the oven-dried
material added two hours more, making a total of thirteen hours.
For the preparation of the skeleton the freshly killed animal is
to be preferred. If, however, the material needs to be pre-
served, some tests show that either saturated salt solution (brine)
or 50 to 60 per cent alcohol may be used without introducing
any marked modification in the results. Formalin is to be
avoided, as the macerating fluid acts with jareut difficulty after
formalin preservation.
BY THE BIBLIOGRAPHIC SERVICE, OCTOBER 27
THE ORIGIN, GROWTH AND FATE OF OSTEOCLASTS

AND THEIR RELATION TO BONE RESORPTION !
LESLIE B. AREY
Anatomical Laboratory of the Northwestern University Medical School
TWENTY-FOUR FIGURES (FOUR PLATES)
CONTENTS
AERO CU CLONE eerie het che oceans selec ce: A ae ee cee OL

IEITStROIRICEN Lotte Baws SUA oe arate eee aE OORT Mer Nic aan OAs oe ee 316
(UAE PRaSerect ec Bs a a LSD PSS lA 316
MMi GimMGleanlby erp micscrreak nise amiscSm cisale ie Se een te eae en ts heh 318
JDEWIDs, 8 5G cia’eS ABRs pas.Here TE PONE SEDATE Rr RO ce ag 5 ak a oe I 319
INteirtre
raloll penne Reeve eae rey Loh cieha Se Ne edly Sap ae remem ere MK Male 2 Dat 320
DORE AEATOL AS, Ses as eter ace See ROS Rone ne ee re mes Ac ateii, MoU cA aS a eo 321
DTS CUSST OPE Aeterna eee eee oe aye he eS ee ORT ak oe esas boty 328
SSSTEREO Nee ei eT a Ane A ce 334
Eiben CHT eh Liye eee eet eA wpe oo Gx. 3. cides shots aie Ne seis Aeterna cio ea ane 333
INTRODUCTION
Robin (’49) appears to have been the first to distinguish

clearly between the giant-cells of bone marrow (megakaryocytes)
and those associated with bone itself (polykaryocytes), although
it is probable that F. Bidder (’43) appreciated some such
distinction.
Due chiefly to the efforts of Kolliker (73), the multinucleate
cells of developing bone are known as ‘osteoclasts’ and are
regarded commonly as the direct agents of bone resorption.
The present communication comprises a report of certain
observations made during the last three years upon the origin,
growth, and fate of the osteoclasts, together with a critical
1 Contribution No. 64, January 25, 1918. A grant from the American Asso-
ciation for the Advancement of Science has made possible the publication of the
lithographic plate.
315
316 LESLIE B. AREY
analysis of the evidence upon which their alleged bone-resorp-

tive potentiality rests.’
The literature concerning polykaryocytes is voluminous,
many accounts dealing confusedly with those of normal and
pathological occurrence. Observations and speculations on the
giant-cells of normal bone development, of bone tumors, and on
those of other pathological origins so overlap that a complete
literature review would be both tedious and unprofitable. For
this reason the following digest of previous work will deal chiefly
with the osteoclast associated with normal bone development
and resorption.
HISTORICAL
Origin. Views as to the origin of the osteoclast are not in

accord.
Kolliker (73) observed the presence of osteoclasts on resorp-
tion surfaces; these appeared at first discontinuously in the
osteoblastic layer and increased in number, size, and multinu-
clearity concomitantly with the disappearance of the latter.
Admitting the absence of direct proof by observation, he never-
theless considered the circumstantial evidence sufficient to render
the genetic relation of osteoblasts to osteoclasts highly probable.
The latter were supposed to represent single osteoblasts which
had undergone repeated nuclear division. For the earliest
osteoclasts, and for those found during the resorption of milk
teeth, an origin from connective-tissue cells was assumed.
2 Shortly before his death Prof. C. W. Prentiss had been engaged upon a
study of the osteoclasts. Except for a brief abstract (15) of a paper on this
topic delivered before a local surgical society and for a number of unlabeled draw-
ings, there were found no notes or other data to indicate the extent or state of
completion of the project. For these reasons the perhaps tentative conclusions
reached by him are known in outline only. On assuming duties at this labora-
tory the writer became interested in the same problem and worked over the
entire field independently; some of the preparations used have been identified as
those upon which Professor Prentiss made observations. It is believed that the
scope of the investigation has been extended considerably and that certain in-
terpretations have been profoundly modified. With respect to an osteoblastic
origin our conclusions are in accord, but as to the fate and significance of the ele-
ments our opinions clearly differ. 5
ORIGIN AND FATE OF OSTEOCLASTS oe
Bassini (72), Pommer (’81), and Gegenbaur*® concurred fully

with K6lliker’s thesis; Ziegler (’78), Lewis (13), and others have
rejected it. Morrison (’73), working under the inspiration of
Kolliker, likewise reported having observed intermediates (in
the number of nuclei?) between the osteoblast and osteoclast.
An origin by osteoblastic fusion was inferred by Howell (’90).
It is, he argues, “plausible to think that the closely packed cells
might become forced to form a polykaryocyte and a number
of transitional steps [in size and number of nuclei?] can be seen
in sections Ap aca
Bredichin (’67) held the giant-cells of normal and pathological
bone resorption to be transitional stages in the transformation
of bone tissue into marrow and granulation tissue; in other words,
they are nothing more than liberated bone cells become multi-
nucleate. Murisier (’75) and Ziegler (’78) expressed a some-
what similar opinion, while Rindfleisch (72) was likewise
convinced from the study of giant-cell sarcomata that the poly-
karyocytes are bone cells‘ which have become free and gone
over into a peculiar hypertrophic state. The proliferation of
bone cells was also noted by Morrison (’73). According to
Lowe (’79), the osteoclastic nuclei arise either from bone cells
or from inwandering leucocytes.
Wegner (’72) observed a close association between poly-
karyocytes and blood-vessels in pathological bone resorption.
He also maintains that normally the osteoclasts originate as
proliferations of the vessel wall. An intimate relation to blood-
vessels, although not in every case an actual origin from them,
has been emphasized as well by Morrison (’73), Brodowski
(75), Maas (’77), Schaffer (88), Bidder (’06), and others.
According to Kaczander (’82), giant-cells (osteoclasts?) form
from enlarged, liberated cartilage cells by coalescence. These
cartilage cells may be multinucleate while within their capsules.
This interpretation is modified by Geddes (713), who considers
osteoclasts to be “hybrid syncytial masses composed of fused
cartilage cells containing osteoblasts.”
3 Cited by A. Bidder (’06).

318 LESLIE B. AREY
Ranvier (’89), Renault (’93), and Duval (’97) trace the origin
to lymphoid marrow cells, whereas Mallory (’11) insists that
osteoclasts arise unquestionably from fused, large mononuclear
leucocytes.
The results of Jackson (’04), Danchakoff (09), and Maximow
(10) agree in tracing the origin of the first osteoclasts in the early
stages of bone development to enlarged reticular cells of bone
marrow. ‘These cells possess at first but two or three nuclei
and the cytoplasm is basophilic. Later their cytoplasm appears
oxyphilic and the nuclei may become extremely numerous.
Osteoclasts are viewed by Todd (’13) as “masses of preosseus
tissue artificially separated from the fully ossified bone during
its preparation for histological examination.”’
The views of Wegner, Kaczander, Todd and Geddes, just
presented, are unusual, some of them seemingly even fanciful.
In my experience they demand no serious attention. The
remaining workers trace or infer an origin from osteoblasts,
bone cells, or marrow tissue of some sort. The relation of these
opinions to my own observations will be made clear in the pages
which follow. Briefly, I recognize all three sources of origin,
but the interpretation of the actual mode of genesis and growth
of the osteoclast, and the relative importance of each contribu-
tory element is novel.
Multinuclearity. A variance of opinion exists also as to the
manner in which the osteoclast comes to possess its numerous
nuclei.
Kolliker (’73) considered the increase in nuclei to result from
nuclear division. Adherents to this view include Bredichin
(67), Wegner (’72), Morrison (’73; by amitosis), Bohm and
Davidhofft (by mitosis), Jackson (’04 by mitosis); and Jordan
(18; by mitosis, to a limited degree).
Morrison (’73) and Danchakoff (09) speak of the confluence
of mesenchymal cells. Maximow (’10) likewise believes that
large osteoclasts arise at the expense of smaller ones; further-
more, he records having never observed nuclear division either
by mitosis or amitosis.
ORIGIN AND FATE OF OSTEOCLASTS 319
Fate. Concerning the ultimate fate of the osteoclast, there

is also no general agreement.
These giant-cells were viewed by Bredichin (’67) as transi-
tional stages in the transformation of bone tissue into marrow
and granulation tissue.
Weener (’72), observing some polykaryocytes with cavernous
recesses, was led to speculate as to whether new blood-vessels
might arise from such (compare p. 327). He also believed in
their resolution into conneetive tissue, or, perhaps, marrow
cells.
K6lliker (’73) noted that when bone deposition again succeeds
a period of resorption, the osteoclasts disappear from the resorp-
tion area and are superseded by osteoblasts. Furthermore,
where resorptive and formative areas join he found intermediate
types. The conclusion is drawn that, in such situations at
least, the osteoclasts fragment and return to osteoblasts.
K6lliker, nevertheless, emphasizes the absence of direct proof
and admits (p. 27) that: ‘Die letzten Schicksale der Osto-
klasten sind noch in grosses Dunkel gehiilt.’”’ Allowance is also
made for the degeneration of some of the giant-cells and for the
possibility of a transformation of others into connective-tissue
and marrow cells, as Wegner (’72) contended.
Gegenbaur®> and Bassini (’72) agreed with these views of
Kolliker. Pommer (’81) likewise held that osteoclasts not
only revert to osteoblasts, but also to cells of a different char-
acter and to intercellular material, whereas at the suppression
_ of sufficient nutriment they degenerate.
The removal of the stimulus to absorption was believed by
Morrison (’73) to lead to the disappearance of the osteoclasts
by ‘molecular degeneration.’
Lowe (’79) presented a curious and incredible account of the
encapsulation of osteoclasts, the fragmentation of their nuclei
and cytoplasm into discrete cells, the rupture of the capsular
wall, and the dispersal of the individual elements into the marrow.
He remarks on the similarity between these stages and those
of encystment and spore formation in protozoa.
320 LESLIE B. AREY
Jackson (’04) described and figured osteoclasts, which by the

enlargement and confluence of cytoplasmic vacuoles formed
detached cells; these remain interconnected by processes and
are indistinguishable from neighboring reticulum cells.
The view of Maximow (’10) differs from that of Jackson in
that some osteoclasts are said to be destroyed through extreme
degeneration.
Lewis (713) holds these polykaryocytes to be degenerating
cells produced by those conditions which lead to the dissolution
of bone.
Thus, these opinions pertaining to the fate of osteoclasts
either uphold their transformation into other cellular elements,
their total destruction, or admit both possibilities.
Several provisional communications by the writer on the
problem of the origin, growth, fate and significance of the giant-
cells of bone have appeared previously (’17a; ’17b; ’08).°
MATERIAL
The observations recorded in this communication have been

made on developing membrane bone of human and pig embryos.
A favorable site for study is found about the walls of the dental
6 A publication by Jordan (’18) some time after the present paper had left my
hands necessitates supplementary comment. Jordan states (p. 248) that ‘“‘The
osteoclast arises chiefly (at first exclusively) from the marrow reticulum by a fu-
sion process essentially as previously described by Maximow; in the earliest
stages the nuclei may multiply slightly by mitosis; their increase, however, is
due mainly to exogenous additions either reticular, osteoblastic, or even bone
cells. Smaller osteoclasts may fuse to form larger syncytia. These cells finally
degenerate, as evidenced chiefly by a vacuolization of their cytoplasm and a
karyorrhexis, and eventually they disintegrate. The above-described material
gives no evidences of a retransformation into marrow reticulum, as maintained
by certain workers (Jackson, Arey). . . . . Osteoclasts may arise to some
extent also from fusing osteoblasts, . . . . But the osteoblasts involved
in this process are not “‘worn out’’ as Arey maintains. On the contrary, they are
of the less differentiated types and strongly basophilic.”
As to sources of origin, method of growth, and ultimate fate these conclusions
are in harmony with my own, as set forth in former communications (717 a, 717 b;
18) and in the present contribution. In certain details, however, our opinions
diverge widely. The method of osteoclastic origin from marrow reticulum he
considers chief in importance, that from osteoblasts secondary; on the contrary,
alveoli where active bone resorption is preparing for the accom-

modation of the rapidly growing teeth. Here osteoclasts appear
in large numbers.
The material which proved most useful consisted of several
series illustrative of tooth development in the pig. In these
decalcified preparations the histological preservation was excep-
tionally good. The jaws of appropriate pig embryos were fixed
in Zenker’s fluid, decalcified in acid, embedded in celloidin,
and stained with hematoxylin and eosin or hematoxylin and
orange G. Hematoxylin and Congo red have been employed
also, but the hematoxylin-eosin combination was favored for
bringing out delicate tinctorial contrasts. Part of the material
comprised serial sections; in view of the large size of the osteo-
clasts, such series are instructive and important.
Typical stages illustrative of the history of these elements
were demonstrated before the thirty-third session of the Ameri-
can Association of Anatomists at New York.
OBSERVATIONS
The osteoclasts (named ‘Ostoclast’ by Kélliker) are large,
multinucleate cells of irregular shape and without a definite
I have maintained that, except in the youngest stages of bone development, the
reverse is true. Furthermore, those giant cells resulting from osteoblastic fu-
sion are traced by Jordan only from young, slightly differentiated cells; in my
membrane bone material these stages of osteoclast genesis have never been found,
for example, among the active osteoblasts of growing spicule tips, but only far-
ther back amid more or less ‘depleted’ cells; the term ‘depleted’ is, of course, rela-
tive, and need not necessarily signify ‘worn out’—a term not of my using. In
the first preliminary announcement of these studies (’17 a) I recorded the abun-
dance of degenerative stages and intimated a probable final disappearance, but
further added the tentative observation that ‘‘indications of a transformation
into marrow reticulum are not lacking.’’ At the time this last statement was
penned I had studied the fate of osteoclasts only partially, but had noted the re-
semblance of fragmenting osteoclasts like figure 18 to those stages held by Jackson
(’04) to depict a retransformation into marrow reticulum. In a later publication
(718) and in the present report these stages are believed to represent merely de-
generating cells undergoing disintegration (’18, p. 237): ‘“‘Neither have I seen
convincing stages of a fragmentation into reticular cells of the marrow as Jackson
and Maximow describe. The entire picture, from the early formation by the
fusion of depleted osteoblasts, seems rather to depict a progressive degeneration,
culminating in death and removal.’’
aoe LESLIE B. AREY
limiting membrane. Kolliker (’73) records their maximum size .

in the human new born as 38y x 91 yu and with as many as 50 to
60 nuclei. I have found the measurements ‘in the pig to run as
~ high as 654x105, with a nuclear count of about 125. These
figures undoubtedly are too low, for the entire cell in all
probability extended through several sections.
In shape, osteoclasts are rounded or have variably con-
spicuous processes (figs. 13 and 14). This latter configuration
is suggestive of amoeboid motility (compare Maximow, ’10),
but Kolliker (’73) and Bizzozero failed to confirm this in living
osteoclasts examined on a warm stage.
The cytoplasm is typically strongly oxyphilic and contains a
variable number of vacuoles; these Jackson (’04) believed not
to consist of fat, whereas Dubreuil (’10) is convinced of their
lipoid nature. The cytoplasm is granular, sometimes coarsely
so, and is notable for the usual absence of debris (compare p. 332).
Nuclei tend to be pyknotic, especially in the older, apparently
degenerating forms (fig. 16). Some nuclei appear shrunken or
folded (fig. 19), but convincing amitotic stages have not been
observed by me.
Certain osteoclasts exhibit a brush border along the edge in
apposition with the bone. This border stains more intensely
than the rest of the cell and may be finely striate or composed
of coarse, block-like elements (figs.8,9and17). Some also
have a fringed or toothed appearance. ‘The significance of this
condition is obscure.
In regions where bone is actively forming, the osteoblasts
are typically separate units, columnar in shape and with baso-
philic cytoplasm (fig. 2, obl.); the nuclei tend to be placed toward
the end of the cell farthest from the bone matrix. As develop-
ment proceeds the cytoplasm diminishes in amount and in older
regions the still basophilic osteoblasts flatten out and lose their
distinct cell boundaries. There are thus formed syncytial
masses of variable size (fig. 1, ocl.). That such do not result
from overstaining with basic dyes is proved by the intense
oxyphilic reaction of certain other elements in the same prepara-
tions. Close to the basophilic syneytium in figure 1, for example,
was a brilliant eosinophilic osteoclast.
ORIGIN AND FATE OF OSTEOCLASTS Soe
Whereas the osteoclasts may arise in the early stages of bone

development from the mesenchymal or reticular cells of the
marrow, as Jackson (’04), Danchakoff (09), and Maximow (’10)
contend, my observations indicate that in later stages they take
origin chiefly from the osteoblastic syncytia just described.
There were found all transitional tinctorial stages between
these syncytia with basophilic cytoplasm, staining blue with
hematoxylin, and typical oxyphilic osteoclasts, staiming red
with eosin. In figure 2, compare the intermediate purple shade
of a very large-sized intermediate with the normal basophilic
osteoblasts at the right and with a typical osteoclast such as
figure 4. Here again this characteristic coloration does not
result from basic overstaining as may be convincingly proved by
an inspection of adjacent fields.
A comparison of the nuclei found in osteoclasts and osteoblasts
can not be depended upon to furnish very reliable information
as to genetic relationship. Not only do the osteoclastic nuclei
commonly exhibit pyknosis, but the resemblance between the
unchanged vesicular nuclei and those of osteoblasts and con-
nective tissue is close; this is not surprising since the osteoblasts
themselves originate from the connective tissue. In some
instances, nevertheless, the chromatin disposition and general
nuclear structure of the osteoblasts and osteoclasts clearly
agree better than do either with the adjacent marrow reticulum.
There is evidence that the elongate nuclei in flattened osteo-
blasts are restored to the spheroidal configuration upon the
release of compression.
Not only are basophilic syncytia and syncytial masses of
intermediate stainability found, but osteoclasts may be seen
frequently continuous at one or both ends with basophilic
osteoblasts and particularly with osteoblastic syncytia. This
is represented in figure 3, especially at the right; both osteoclast
and osteoblasts have been displaced artificially from the bone
surface. Figure 4 shows on the right an osteoclast with five
nuclei continuous by a bold transition with fused basophilic
osteoblasts. Similarly in figure 11, from a human fetus, the
osteoclast lapped over the end of a bone spicule, is abruptly
324 LESLIE B. AREY
continuous at both ends, but especially at the right, with

osteoblasts.
According to these observations, therefore, the osteoclast
arises from depleted osteoblasts which have first formed a syn-
cytium before being transformed into the oxyphilic osteoclast.
As bone resorption continues, osteoblasts progressively lose
their former relation to the bone, come into association with the
advancing osteoclasts and are incorporated into them. If the
spicule in figure 11 were resorbed it is believed that the simul-
taneously advancing (thigmotactic?) osteoclast would take up
the osteoblasts in its path. In a similar manner, smaller osteo-
clasts may, by fusion, merge into larger ones.
But the osteoblasts, as such, are not the only source from
which the nuclear and cytoplasmic contents of the osteoclasts
are recruited. Bone cells, embedded in the matrix, are laid
bare by the resorptive processes and are ingested by the
oncoming osteoclasts. All intermediates may be found between
the initial and final stages of inclusion. At the left of figure 4 a
cytoplasmic process of an osteoclast is in contact with the capsule
of a bone cell which is otherwise embedded in the bone matrix.
Two succeeding stages appear in figure 5; on the right the area
of contact is extensive; at the left the bone cell is half within
and half without the osteoclast. Figure 11, from a human
fetus, and figure 10 show similar steps, as do text figures A
and B. A last stage appears in figure 12, at the right.
Furthermore, bone cells are enclosed normally within a cap-
sule which is known to be resistant, for example, to the action
of strong hydrochloric acid. Encapsulated and _ distinctly
stellate cells, which resemble bone cells identically, are
occasionally found, embedded in the osteoclastic cytoplasm
(fig. 13; compare also figs. 2,6, and 12). Such cells are inter-
preted as bone cells whose capsules have resisted cytoplasmic
digestion. From the relative infrequency with which such
persistent capsules are seen, it is probable that the enclosed
bone cells are eventually liberated. Ingested bone cells must
contribute in substantial numbers to the formation of osteoclasts.
This is perhaps especially true on flat resorption surfaces.*
Hence it appears that the degree of multinuclearity is an

index of the number of osteoblasts and bone cells entering into
the composition of the osteoclasts. Also, in general, the larger
an osteoclast, the more numerous its nuclei and the more exten-
sive its history in relation to bone resorption (compare figs. 1,
Plerands 22).
' Bsc
~<a . is
Fig. A The encapsulated bone cell, b.c., is half ingested by an osteoclast,
ocl., which lies on a spicule of bone, mtx. Only a portion of the entire osteoclast
appears in this section. Photograph. X 650.
=wm
Fig. BA stage in osteoclastic phagocytosis similar to that of fig. A. The

bone cell is half within and half without the osteoclast. Photograph. X 650.
My observations agree with those of Maximow (710) in that

nuclear division either by mitosis or amitosis has never been
observed in these older stages. Shrunken and folded nuclei
do appear, especially in cells which show other evidences of
degeneration, but convincing stages of amitosis have not been
found by me. In much of the material used mitoses were not
uncommon in the near-by germinative layer of the epithelium,
so the preparations would seem to be favorable for demon-
strating nuclear division did it occur.
326 LESLIE B. AREY
Sometimes osteoclasts are seen which are in continuity by

fine processes with the marrow connective tissue (figs. 14 and 15).
The general appearance of such cells does not indicate advanced
degeneration, hence it may be thought that they have arisen
by coalescence of connective-tissue elements. A case illus-
trative of this is seen in the rather thick section shown in figure
15. Such osteoclasts which seemingly he free in the marrow
tissue usually prove to be in contact with the bone when
recourse is made to serial sections. It is possible, however,
that even in these later stages of bone resorption some osteo-
clasts are formed from the marrow reticulum, similar to the
origin described by Jackson (’04), Danchakoff (09), and Maxi-
mow (710), and that this is an illustrative stage of such a process.
Indeed, from what is known of tooth resorption and of the
genesis of foreign-body and other giant-cells it seems reasonable
that the osteoclast has no single source of origin. An elabora-
tion of this topic will be found on pages 328 and 329 of the
discussion. :
Finally, the fate of the osteoclasts demands attention. That
these cells may be resolved ultimately into osteoblasts and
again resume bone formation seems improbable. Kolliker, who
championed this view, was unable to produce direct evidence
in its support. My preparations show nothing in favor of such
a cycle, whereas pictures of degeneration in varying degrees
are abundant. ;
A vacuolated cytoplasm is a common characteristic of many
osteoclasts. The degree of vacuolization appears to be fairly
closely correlated with the extent of degeneration (figs. 16, 17,
18, 19, and 6). In other cells granular degeneration of the cyto-
plasm occurs (figs. 7 and 20). Nuclei may show varying degrees
of pyknosis while the cytoplasm still appears to be in good con-
dition (figs.2and5); on the other hand, the nuclei of cells
otherwise exhibiting extreme degenerative changes appear to
be constantly pyknotic. Through stages of increasing vacuoli-
zation and pyknosis conditions are reached such as are depicted
in figures 6, 17 and 19. In an osteoclast like figure 6 both nuclei
and cytoplasm are far from normal; the nuclei are highly pyk-
ORIGIN AND FATE OF OSTEOCLASTS S20
notice while the cytoplasm takes the eosin poorly and is riddled
with vacuoles which at the edges produce a ragged, laced appear-
ance. In short, the optical appearance is such as one normally
associates with extreme degeneration.
As the resorption of local areas of bone is completed, the
accompanying osteoclasts become left behind, stranded in the
marrow tissue. Thus, one occasionally finds a small portion of
bone enclosed by an osteoclastic mass (fig. 17). Only when
serial sections are available is one certain that a paratangential
section of a spicule-tip had not produced a deceptive appearance
of an isolated bone fragment and enclosing osteoclast.
Left behind in regions where resorption has apparently finished
its course are sometimes found also nests of large osteoclasts
(figs. 19 and 22). Such a stage as figure 21 apparently shows
in its inception how these masses become thus isolated, for this
particular giant-cell is at the rear of an area of bone dissolution
that is almost completed locally. Some of these stranded
elements show excessive degenerative changes. Illustrations
show but poorly the pale, reticulate or fragmented cytoplasm
and the shrunken and distorted, pyknotic nuclei.
Jackson (’04) and Maximow (710), in particular, have upheld
the fragmentation of osteoclasts into detached cells which
become indistinguishable from the reticulum of the marrow
(compare p. 327). I have seen several stages which show
stellate portions of the osteoclast being cut off by vacuolization
(fig. 18). That these moieties, some of which may even lack
nuclei, persist as elements indistinguishable from the reticulum
is doubtful; on the contrary, the general appearance of such
osteoclasts seems rather to point to ultimate degeneration.
If my interpretation of the osteoclasts be correct, the entire
course from the time of osteoblastic coalescence is one of pro-
gressive decline (see foot-note 6, p. 320). To return to a healthy,
active state, the fragmentation products of such osteoclasts
would seemingly have to undergo extensive rejuvenation both
as regards nucleus and cytoplasm.
Large osteoclasts were observed within the blood-vessels of
the marrow (figs. 7 and 20). That such gain admittance and
328 LESLIE B. AREY
do not arise in situ from the endothelium is supported by their

usual retrograde appearance; vacuolization, granulation and loss
of stainability of the cytoplasm, and pyknosis of the nuclei
occur. No stages have been observed which could be interpreted
as illustrative of an origin from the blood-vessel itself; on the
contrary, the condition appears to indicate a method of final
removal. This admission into embryonic vessels does not of
itself prove or imply an amceboid activity by the osteoclast;
Meyer (718) is wrong in assuming I hold such a belief.
DISCUSSION
The conclusions of Kélhiker regarding the history and signifi-

cance of the osteoclasts have gained great prominence. It
should be kept clearly in mind, however, that his opinions were
almost wholly inferential. He neither offered direct proof of
the origin of osteoclasts nor of their fate, as has been pointed
out in the historical section. The apparent reasonableness of
these deductions, and the prestige of their originator, doubtless
account for their acceptance by numerous later investigators
and for their widespread inclusion. in texts.
Concerning the validity of many of the claims advanced by
other workers the writer can offer little except the negative
evidence of not having seen corroborative stages. Unfortu-
nately, in the past, too many dogmatic statements have been
made wholly unsupported by appropriate evidence. Only a
few have presented their claims adequately described and illus-
trated. Under these conditions it is not always easy to dis-
tinguish between surmises and conclusions drawn from actual
observations.
Moreover, it seems reasonable to believe that several mor-
phologically similar but developmentally distinct elements
masquerade under the generic term osteoclast, so that, his-
torically, the controversy has not always been over identical
structures. The careful studies of Jackson (’04), Danchakoff
(09, and Maximow (710) appear to show convincingly that, at
least in the early stages of bone development, the osteoclasts
arise from the primitive connective tissue of the marrow. At
ORIGIN AND FATE OF OSTEOCLASTS 329:
the time of resorption of the milk teeth an osteoblastic history

is likewise excluded. In the present communication are re-
corded observations of osteoclasts in continuity with the marrow
reticulum; it is entirely probable that this is not a secondary
union, but primary, and of developmental significance. My
chief results, nevertheless, point to a widespread formation of
osteoclasts in the later stages of development from osteoblasts
and bone cells.
Polykaryocytes which resemble osteoclasts morphologically
are the so-called foreign-body giant-cells produced within the
body about fat (Mallory, ’11) as a center, or developed experi-
mentally about introduced foreign bodies, such as agar, paraffin,
lycopodium spores, or bone dust (Maximow, ’02; Lambert, 712;
Mallory, ’11, et al). These are said to form from fused, wan-
dering leucocytes, and Mallory (’11) even goes so far as to derive
all osteoclasts (but on insufficient evidence) from a like source.
The foreign body origin is, nevertheless, an important concept.
It seems reasonable that on the cessation of active growth a
spicule of bone, or any quiescent portion of a spicule, becomes
essentially a foreign body and the cells in contact with it, not
necessarily all of one origin, may respond to whatever stimulus
it offers and fuse into syncytial masses. Kolliker (73) found
that ivory pegs driven into bone became eroded, the lacunae
thus formed containing polykaryocytes. Rustizky (’74) sim-
ilarly produced giant-cells, but no lacunae, by introducing pieces
of bone in the dorsal lymph saes of frogs.
The causative factors of giant-cell formation are necessarily
obscure. Kélliker considered it due to a pressure by the soft
parts underlying bone (compare also Wegner). Pommer (’81),
on the contrary, held as responsible a locally increased blood
pressure. On spongy bone, at least, it may be due to the aug-
mented mutual pressure resulting from the decreased size of the
spicules and to a concomitant loss of cell individuality as a result
of general osteoblastic degeneracy. The formation of ordinary
foreign-body giant-cells would demand a different explanation.
Presumably a definite thigmotaxis is operative in all giant-cell
formation.

330 LESLIE B. AREY
Finally comes the vexed question as to whether the osteoclasts

actually are bone destroyers. Kolliker (’73) was the first and
most persistent exponent of their bone-destroying function. He
believed that their presence in the Howship’s lacunae of resorp-
tion surfaces, on ivory pegs, and on milk teeth during resorption,
demonstrated his thesis beyond controversy. Among the many
who have subscribed to these views are Wegner (’72), Morrison
(73), Jackson (04), and Maximow (710). Mallory (11) has
even suggested that the erosion of bone may be accomplished
mechanically by the brush border. Danchakoff (’09) refers to the
dissolving action on bone, but calls attention to the relative
infrequency of osteoclasts during the erosion of calcified cartilage
and the irrationality of attributing the destruction of cartilage
solely to them. Shaffer (81) likewise held that they have a
bone-destroying action, which, however, blood-vessels chiefly
perform.
Howell (90; p. 119) took a decided stand against the probabil-
ity of the commonly accepted osteolytie function:
The function of these cells is unknown. The common view that they
are concerned in the absorption of bone (osteoclasts) seems to me to
rest upon very slight evidence. If we find them in developing bone
lying upon the cartilage trabeculae which are being absorbed, we find
them also on the partitions of sponge or pith, introduced into serous
cavities where no absorption is taking place; and the conclusion in the
first case that the absorption which is going on is due to the giant cells
(osteoclasts) is illogical. Absorption of tissues is an occurrence common
enough in the body, and it is difficult to understand why the absorption
of bone or cartilage should require the activity of a special cell, when
the absorption of other tissues does not. It would seem more probable
that this form of cell has no specific function, and that its formation
is, in fact, accidental, or, in a certain sense, pathological: that the
presence of a solid substratum leads to an abnormally rapid growth of
lymphoid cells, leucocytes, osteoblasts, as the case may be, and the
fusion of some of these to produce multinucleated giant cells.
The evidence in favor of a dissolving activity by the osteoclasts

rests upon the following relations: these elements appear at the
onset of resorption and disappear at its cessation; they are
closely applied to bone, sometimes being wrapped around eroded
spicules (fig. 10), and sometimes occupying the so-called How-
ship’s lacunae;’ their coloration with acid dyes often resem-

bles that of bone. The occasional presence of a striate border
(figs. 8 and 17) on the side in contact with bone might be thought
to indicate cellular activity (compare the brush order of the epi-
thelial cells of the intestine or kidney), but this phenomenon is
open equally well to other interpretation.
Against the view of the osteoclast as a causative resorptive
agent may be presented several objections. They are notably
searce during the resorption of calcified cartilage. Howship’s
lacunae not infrequently are seen without associated osteoclasts
(compare Rustizky, ’74); this may represent a distinct method—
“lakuniire Resorption ohne Riesenzellen’” (Kaufmann, ’07). A
‘smooth resorption’ of bone occurs where Howship’s lacunae
are entirely lacking (Busch, ’77), and they are often infrequent
where osteolysis is active. In osteomalacia, leprosy, and in
tumors and many inflammations (Ziegler, ’78), the lime salts are
removed in the absence of osteoclasts—it is significant that in
these cases typical Howship’s lacunae usually occur (Ziegler,
’78, et al). The absorption of various other tissues in the body
is accomplished without the intervention of special cells. It is
not impossible that the relation of osteoclast to Howship’s lacu-
nae may be given an inverse interpretation; if the lacunae rep-
resent softer regions in the bone, or regions for some reason more
directly subjected to the resorbing principle, perhaps these
depressions serve merely as traps in which the giant-cells collect.
If the osteoclasts arise largely from osteoblasts and bone cells,
as set forth in this communication, and if, furthermore, the sub-
sequent course is one of progressive degeneration, it seems un-
likely that such cells, which a short time before had been active
bone formers, now would take over the diametrically opposed
function of destruction. On the contrary, the concept of progres-
sive degeneration seems to militate against such a view.
7 The relation of polykaryocytes to such lacunae is demonstrated strikingly
during the resolution of deciduous teeth. At this time, the root may be com-
pletely scalloped with adjacent pits, each harboring a giant-cell. It is hoped that
a subsequent report will clear up the puzzling questions of the origin and signifi-
cance of these elements.
Soe LESLIE B. AREY
In working over this field and evaluating the total evidence the
writer has become highly skeptical concerning the potency of the
osteoblast in bone resorption. With Lewis (13, p. 86) he would
take the position that “There seems to be no satisfactory
evidence that the osteoclasts are the active causes of bone
destruction. On the contrary, they appear to be degenerating
Colleen ea
Attention has been directed in another publication (Arey, ’17 b)
to the uncritical nature of the assumption that the failure of a
cell to drink in vital dyes warrants the denying to it of phagocy-
tic potentiality. The osteoclast refuses to ‘stain’ with trypan
blue (Shipley and Macklin, 716), yet bone cells, encapsulated or
naked, laid bare by the resorptive processes, are demonstrably
engulfed by it; (compare also the observations of Wegner! and of
Rustizky (14) on ‘Kalkkorner’ within polykaryocytes, and note
the red blood corpuscles within the body of the osteoclast shown
in figure 7 of the present communication.) Furthermore, the mere
presence of cytoplasmic inclusions within an osteoclast by no
means indicates that the latter was responsible for the dissolu-
tion of the material ingested; to imply such a causal relation is
to exceed the limits of legitimate deduction. When Jordan (18,
p. 251) writes that “The osteolytie function of the giant-cells
of reticular and osteoblastic origin is proved by the presence of
globules of resorbed osseus substance [p. 262, globules of absorbed
bone] within the cytoplasm,’’ he perhaps is using ‘osteolytie’ as
a term interchangeable with ‘phagocytic;’ this is supported by
a further reference (p. 255) to the “‘phagocytie’ (osteolytic)
function” of osteoclasts. Yet in another place (p. 246) we read:
8 The presentation of these facts (’17 b) has led Meyer (’18, p. 100) to conclude
that it involved “‘apparently implying that fusion products never observed to
undergo mitosis, nevertheless may be physiologically active and continue a pro-
gressive evolution.’’ That these giant-cells are active enough to engulf bone
cells, osteoblasts, or even fragments of bone matrix is unquestionably true, but
that the osteoclasts enjoy a progressive evolution is not supported by my obser-
vations. On the contrary, both in my 1918 communication and in the present
contribution the h’story of these elements is held to be one of advancing degen-
eration, culminating in death and removal.
9 Cited by Rustizky (’74).
“the osteoblasts are constructive osseus elements; the osteoclasts

are destructive elements. The former elaborate bone; the latter
resorb it.”’
There is no direct evidence as to how bone matrix (inorganic
and organic) is resorbed. One might assume it is essentially a
double process of decalcification and of digestion of the organic
substrate; conditions such as osteomalacia, in which the lime
salts are removed leaving the organic framework, perhaps sup-
port such a view. In the absence of more appropriate evidence,
therefore, the activity of an acid (carbon dioxide or lactic acid?)
and an enzyme can be tentatively suggested.
According to Morpurgo and Satta (08 a, ’08 b), there is asso-
ciated with bone a thermolabile entity, an enzyme they believe,
responsible for calcium: removal during experimental autolysis
of bone. The provisional nature of these two communications
renders their evaluation difficult.
In his Harvey Lecture, Wells (10-11) marshals numerous
facts to support the view that bone resorption is accomplished
through the agency of carbon dioxide. It is demonstrable that
calcium is normally contained in the blood in amounts approx-
imating saturation and that this content is from two to four
times that soluble in water. This amount of calcium in the blood
is held in solution by the colloids and the carbon dioxide. ‘In
normal ossification, and in most instances of pathological calci-
fication, the deposition is probably initiated by a process of col-
loidal adsorption . . . . Reduction in the amount of
carbon dioxide in such areas, or some unknown agency, causes
a precipitation im this colloid matrix, . .-. .’’. It is well
known “that, no matter how sclerotic the walls of the veins may
become, they rarely, if ever, calcify so long as there is venous
blood rich in CO, flowing through them. As soon as they are
occluded, however, calcification occurs readily enough (e.g.,
phleboliths).” It has been long established that carbon dioxide
in solution will dissolve calcium from bone but that NaHCO,
cannot so act. Furthermore, studies on this solubility in vitro
and in vivo show “that pieces of ivory are absorbed most rapidly
in tissues whose metabolism is the most active, and where, by
334 LESLIE B. AREY
inference, there is the most carbon dioxide.’? From these and

other considerations Wells concludes: “It is indeed probable
that it is the CO. which accomplishes the resorption of dead bone
in the living body, and perhaps also the normal resorption of
bone in the various conditions in which this process takes place.”
It is, perhaps, difficult to imagine the mechanism of the local-
ization of carbon dioxide( or the stronger lactic acid?) in sufficient
concentrations to effect the selective erosion of small areas, or to
account for the frequent directional polarization of the resorptive
wave. However this may be, there is, of course, no basis for
suspecting the osteoclast of special carbon dioxide production.
On the contrary, if it is indeed a degenerating cell, its carbon di-
oxide output is presumably low. The conception of Wells adds
to the objections already enumerated against the osteoclast as
a specific osteolytic agent.
SUMMARY
The multinucleate giant-cells known as ‘osteoclasts’ probably

include several morphologically similar but developmentally
distinct elements.
In the earliest stages of bone development, and to a certain
extent in later stages, osteoclasts apparently arise from the con-
fluence of the mesenchymal cells and connective tissue of the
marrow.
The chief source of osteoclasts, however, is from old osteoblasts
and bone cells. |
Depleted, basophilic osteoblasts coalesce to form multinucleate
masses. These syncytial elements become typical osteoclasts
when their cytoplasm assumes an oxyphilic stainability. All
intermediate tinctorial stages are demonstrable.
True oxyphilic osteoclasts also exist in cytoplasmic continuity
with basophilic osteoblasts. Increase in size and nuclear con-
tent results from the engulfing of osteoblasts met in the path of
resorption and from bone cells which are ingested as the bone
matrix is resorbed.
Osteoclasts undergo retrograde changes and ultimately dis-
appear through extreme degeneration.
Only indirect and insufficient evidence points to the osteo-

clasts as the active, specific agents of bone resorption. That they
are merely degenerating, fused osteoblasts, accords better with
the known facts.
BIBLIOGRAPHY
Arzny, L. B. 1917 a On the origin and-fate of the osteoclasts. Anat. Rec.,

vol. 11, no. 6, pp. 319-322.
1917 b Phagocytosis by osteoclasts. Anat. Rec., vol. 13, no. 5, pp.
269-272, 4 figs.
1918 The origin, fate and significance of osteoclasts. Trans. Chicago
Pathol. Soc., vol. 10, no. 7, pp. 231-234.
Bassin1, E. 1872 Sul processo istologico di riassorbimento del tessuto osseo.
: Nota preventica, 1872.
BippER, A. 1906 Osteobiologie. Arch. f. mikr. Anat., Bd. 68, S. 137-214, 33
Fig.
BippEr, F. 1843 Zur Histogenese der Knochen. Arch. f. Anat., Physiol. u.
wissensch. Med., Jahrg., S. 372-395, 1 Fig.
Brepicuin, J. 1867 Uber die Bedeutung der Riesenzellen in den Knochen.
Centralbl. f. d. med. Wissensch., Jahrg. 5, S. 563-564.
Bropowsk!, W. 1875 Uber den Ursprung sogenannter Riesenzellen und iiber
Tuberkeln im allgemeinen. Arch. f. path. Anat. u. Physiol., Bd. 63,
S. 113-135, 8 Fig.
Buscu, F. 1877 Die Knochenbildung und Resorption beim wachsenden und
entziindeten Knochen. Arch. f. klin. Chirurgie, Bd. 21, S. 150-181.
Dancuakorr, W. 1909 Uber die Entwicklung des Knochenmarks bei den
Vogeln und iiber dessen Verinderungen bei Blutentziehungen und
Ernihrungsst6rungen. Arch. f. mikr. Anat., Bd. 74, S. 855-926, 18
Fig.
Dusreuin, G. 1910 Vacuoles a lipoides des ostéoblastes, des cellules osseuses
et des ostéoclasts. Compt. rend. d. 1. Soc. Biol., T. 69, pp. 189-190.
Duvat, M. 1897 Précis. d.’ histologie. Masson et Cie, Paris, 31 + 956 pp.
Geppes, A. C. 1913 The origin of the osteoblast andof the osteoclast. Jour.
Anat. and Physiol., vol. 47, pp. 157-176, 12 figs.
Howe tt, W.H. 1890 Observation upon the occurrence, structure, and function
of the giant cells of the marrow. Jour. Morph., vol. 4, no. 1, pp. 117-
130, 15 figs.
Jackson, C. M. 1904 Zur Histologie und Histogenese des Knochenmarkes.
Arch. f. Anat. u. Entw., Jahrg., S. 33-70, 16 Fig.
Jorpan, H. E. 1918 A contribution to the problems concerning the origin,
structure, genetic relationship and function of the giant-cells of hemo-
poietic and osteolytic foci. Am. Jour. Anat., vol. 24, no. 2, pp. 225-
269, 70 figs.
Kaczanver, J. 1882 Uber den Ossificationsprozess. Arch. f. path. Anat. u.
Physiol., Bd. 87, Heft 1, S. 189-193, 7 Fig.
Kavrmann, E. 1907 Lehrbuch der speziellen pathologischen Anatomie. Rei-
mer, Berlin, 10 + 1363 8., 693 Fig.
336 LESLIE B. AREY
K6uiikeER, A. 1873 Die normale Resorption des Knochengewebes und ihre

Bedeutung fiir die Entstehung der typischen Knochenformen. Vogel,
Leipzig, 6 + 8658., 95 Fig.
LamBert, R. A. 1912 The production of foreign body giant cells in vitro.
Jour. Exp. Med., vol. 15, no. 5, pp. 510-515, 7 figs.
Lewis, F. T. 1913 <A text-book of histology. Blakiston, Phila., 11 + 539 pp.,
495 figs.
Lowe, L. 1879 Kleinere histologische Mittheilungen. C. Uber die Umwand-
lungen der Osteoklasten im Knochenmark nebst Bemerkungen iiber
Knochenwachstum. Arch. f. mikr. Anat., Bd. 16, S. 618-625, 6 Fig.
Maas, H. 1877 Uber das Wachsthum und die Regeneration der Rohren-
knochen mit besondere Beriicksightigung der Callusbildung. Arch. f.
klin. Chirurgie, Bd. 20, S. 708-770, 27 Fig.
Matvory, F. B. 1911 Giant cell sarcoma. Jour. Med. Research, vol. 24, pp.
463-467, 8 figs.
Maximow, A. 1902 Experimentelle Untersuchungen tiber die entziindliche neu-
bildung von Bindegewebe. SS. Fischer, Jena, 7 + 255 S., 13 Taf.
1910 Untersuchungen iiber Blut und Bindegewebe. 3. Die embryo-
nale Histogenese des Knochenmarks der Saiugetiere. Arch. f. mikr.
Anat. u. Entw., Bd. 76, Heft 1, S. 1-113, 48 Fig.
Meyer, A. W. 1918 Some observations on megacytes in lymphatic tissues.
Amer. Jour. Anat., vol. 24, no. 1, pp. 91-107, 18 figs.
Morevurcao, B. er Satra, G. 1908 a Sur quelques particularités de l’autolyse
de l’os. Arch. Ital. de Biol., vol. 49, pp. 380-382.
1908 b Présence d’un ferment actif sur les composés de calcium dans
les os malaciques. Arch. Ital. de Biol., vol. 49, pp. 383-884.
Morrison, A. 1873 Bone absorption by means of giant cells. Edinburgh Med.
Jour., vol. 19, pp. 305-311.
Murisisr, J. 1875 Uber die Formveriinderungen, welche der lebende Knochen
unter dem Einfluss mechanischer Krifte erleidet. Arch. f. exp. Pathol.
u. Pharmakol., Bd. 3, Heft 5 u. 6, S. 325-347, 3 Fig.
Pommer, G. 1881 Uber die lacuniire Resorption in erkrankten Knochen.
Sitzungsber. d. k. Akad. d. Wissensch., math.-naturw. K1., Bd. 83,
Heft 1, Abt: 3,8. 17-120;
1883. Uber die Ostoklastentheorie. Arch. f. path. Anat. u. Physiol.,
Bd. 92, S. 296-363; 449-516.
Prentiss, C. W. 1915 The origin and fate of the osteoclasts. Surg., Gynecol.
and Obstet., vol. 20, pp. 678-679.
Ranvier, L. 1889 Traité technique d’histologie. 2d ed., Masson, Paris. 871
pp., 414 figs.
Ranaut, J. 1893 Traité d’histologie pratique. Lecrosnier et Babé, Paris,
T. 1, 39 + 973 pp., 354 fig.
Rinprvueiscu, EK. 1872 A text-book of pathological histology. Trans. by W. C.
Kloman and F. T. Miles. Lindsay and Blakiston, Phila., 695 pp., 208
figs.
Rosin, C. 1849 Sur l’existence de deux espéces nouvelles d’éléments anato-
miques qui se trouvent dans le canal médullaire des os. Compt. rend.
d. 1. Soc. Biol., Année 1, pp. 149-150.
Rustizky, J. 1874 Untersuchung iiber Knochenresorption und Riesenzellen.

Arch. f. path. Anat. u. Physiol., Bd. 59, S. 202-227, 8 Fig.
Scuarrer, J. 1888 Die Verknécherung des Unterkeifers und die Metaplasiefrage.
Arch. f. mikr. Anat., Bd. 32, S. 266-377, 26 Fig.
Surptey, P. G., anp Mackin, C.C. 1916 Some features of osteogenesis in the
light of vital staining. Amer. Jour. Physiol., vol. 42, no. 1, pp. 117-
123.
Topp, T. W. 1913 Apreliminary communication on the development and growth
of bone and the relations thereto of the several histological elements
concerned. Jour. Anat. and Physiol., vol. 47, pp. 177-188, 9 figs.
Weaner, G. 1872 Myeloplaxen und Knochenresorption. Arch. f. path. Anat.
u. Physiol., Bd. 56, Heft 4, S. 523-533.
Weis, H. G. 191011 Calcification and ossification. The Harvey Lectures.
J. B. Lippincott Co., Phila., 146 pp.
ZieautEerR, E. 1878 Uber Proliferation, Metaplasie und Resorption des Knoch-
engewebes. Arch. f. path. Anat. u. Physiol., Bd. 73, 8. 355-379, 7 Fig.
EXPLANATION OF PLATES
The figures of these plates were made with a Leitz js homogeneous immersion
objective and a no. 1 Leitz ocular, at an original magnification of 785 diameters.
With the exception of figures 8 and 11, which are from human embryos, the
illustrations represent stages from the jaws of pig embryos.
ABBREVIATIONS
b.c., bone cell obl., osteoblast

cap., capiNary ocl., osteoclast
c.t., connective tissue str., striate edge of osteoclast
mtx., matrix of bone *, junction of confluent osteoblasts and
osteoclast
PLATE 1
The preparations, from which the illustrations of this plate were made, are
from pig embryos, fixed in Zenker and stained with hematoxylin and eosin.
The reduced magnification is now 590 diameters.
1 The basophilic syncytium, ocl., illustrates the first stage in osteoclast devel-
opment by the fusion of osteoblasts.
2 The immature osteoclast, ocl., exhibits a transitional stage in stainability
between the basophilic osteoblasts, obl., and a typical oxyphilic osteoclast like
figure 4.
3 An asterisk, *, marks the continuity between an osteoclast at the left and
the osteoblastic syncytium on the right.
4 At the right is a bold osteoblast-osteoclast junction. On the left a
process from an osteoclast, in contact with the capsule of the bone cell, b.c.,
illustrates an early stage of phagocytosis.
5 Two later stages in the ingestion of bone cells by an osteoclast; the engulf-
ment of, b.c.’, is farther advanced than of, b.c.
6 Advanced osteoclastic degeneration. The nuclei are pyknotic, the cyto-
plasm highly vacuolated and palely staining.
7 An osteoclast within a blood capillary. Such usually show evidence of ©
degeneration and suggest a method of final removal.
338
ORIGIN AND FATE OF OSTEOCLASTS PLATE 1
LESLIE B. AREY
PLATE 2
Figures 8 and 11 are from human embryos, the others from pig embryos. The
reduced magnification is now 630 diameters.
8 An osteoclast showing a finely striate border in apposition with the bone.
9 A striate border of coarse, block-like composition.
10 An osteoclast wrapped around an eroded spicule and conforming closely to
its irregularities. The bone cell, -b.c., is nearly incorporated.
11 Abrupt transitions, *, between an osteoclast and adjacent osteoblasts,
The bone cell, 6.c., is confluent with the phagocytosing osteoclast.
12 On the right, the bone cell, b.c., has practically lost its identity and become
a part of the osteoclast. At the left, several encapsulated bone cells, b.c.’, lie
within the osteoclastic cytoplasm.
13 Two stellate, encapsulated bone cells, have been taken up during bone
resorption.
340
LESLIE B. AREY :
341
PLATE 3
The reduced magnification of the figures of this plate is now 630 diameters.
14. An osteoclast with irregular processes, some of which are continuous with
the marrow connective tissue.
15 An oxyphilic syncytium in the marrow, apart from the bone. Continuity
with the connective tissue suggests a possible origin from the latter.
16 An early stage in the degeneration of osteoclasts. Nuclei are pyknotic;
cytoplasm slightly vacuolated.
17 The isolated (?) spicule of bone is surrounded by a degenerating osteo-
clast which shows extreme vacuolization and cytoplasmic disintegration.
18 Vacuoles have nearly cut off that portion of the osteoclast indicated by a
cross. Such an appearance has been interpreted by some as indicative of a reso-
lution into reticulum cells.
19 A highly degenerated osteoclast apparently left behind after resorption
was locally completed.
342
ORIGIN AND FATE OF OSTEOCLASTS - PLATE 3
LESLIE B. AREY
343
PLATE 4
The reduced magnification of the figures of this plate is now 630 diameters.
20 A degenerating osteoclast within a blood capillary (compare fig. 7).
21 A large osteoclast measuring 404 x80yu. Itis at the rear of the local region
of bone dissolution and somewhat later might have become isolated like figs. 19
or 22.
22 Several huge giant-cells lie stranded in the marrow tissue in an area where
bone destruction is locally finished; possibly serial sections would have shown
these elements to interconnect. Similar nests occur in the near vicinity. The
osteoclast at the left, ocl.’, measures 65 uw x 105 uw and contains about 125 nuclei.
344
LESLIE B. AREY
a9
s\a
f
ee
y,
Are)
=e
x
45%
4S
vi
Y
Oe1
>
>
oo
7,
(2)
Os
SR>)
x
>GF
Kq °
Se
(ay 4
S
my
i
@@-\\
Rye
eee
ar
LO
Ss
=
$$$
~
=
nd
345
Resumen por Henry Mck. Knower, por el autor 8. Saguchi
Academia Médica, Kanazawa, Japén
Estudios sobre las células glandulares del pancreas de la rana.
El presente trabajo es un estudio critico y practico de la

estructura histol6gica de las células glandulares examinadas con
ayuda de varios reactivos en diferentes estados de la actividad
secretora. Un concepto verdadero de la naturaleza de los ele-
mentos nucleares depende en sumo grado de la interpretacién de
la aecién de los fijadores. El autor estudia con considerable
detalle los granulos citoplasmicos, las mitocondrias y fibrillas
con referencia al papel que desempefan con el ntcleo y nuceolo
en la formacién de la secreccién. El autor demuestra en los
cortes una continuidad estructural intima entre el nucleolo y la
red nuclear, y los filamentos mitocondriales del citoplasma. Las
observaciones sobre la formacién de grdnulos semejantes a grasa
indican su origen a expensas de las mitocondrias y su probable
paso a la sangre como parte de una secreccién interna. Los
grinulos de zimégeno se derivan de condriocontos mediante des-
integracién de estos tltimos, formdndose primero granulos de
“‘nrozimogeno,’’ muy pequenos, que creciendo se transforman en
los grdnulos de zimégeno. Las mitocondrias se estan formando
constantemente a expensas de la red nuclear para substituir a
las usadas en la formacién de los grainulos de zimégeno. El
autor estudia extensamente la red intracelular o canaliculos, que
deben considerarse como formados por materiales de deshecho
originados por una acumulacién de substancias no empleadas
durante la elaboracién de los granulos de prozimégeno. ‘Tam-
bién sigue la degeneraci6n fisiolégica de algunas de estas células
glandulares. Una secci6n del trabajo esta dedicada a Los Neben-
kerne (ntcleos accesorios) y el autor llega a la conclusién de que
estas estructuras son artefactos o partes de células degeneradas.
. BY THE BIBLIOGRAPHIC SERVICE, NOVEMBER 17
STUDIES ON THE GLANDULAR CELLS OF THE FROG’S

PANCREAS
S. SAGUCHI
Medical Academy, Kanazawa, Japan
FIVE PLATES
. CONTENTS
PRIETO CUIC UL OME -ce titi ats Pts cotecs rssche i sacl a Aor RE ee Bates area etka Siskel eisiGine 348
PRRAUNCETIUCICUSs ee ttn terae etre ccc iehctnre kre eh eee ei eae T anne tan eat 349
TTS TORIC] Barta sels Ne alas ie te Bde tte ines heat ap aCe ete tye 2 cae 349
@SerVyiAbVONS eyed Ae, cus se CANS seotacalorey olsral cacti hal eR Pee nL a crn sice 349
fee Chromatinenet works scacid 4A tone aoe oe ie onan 349
24 Nucleoli-and nucleolar corpuscles. . .2/..026 aches tc ase sino nies oy 350
Semlehevcell=bodiystsacc sci ararteee cet k hol ete whe ave eo SE A Pik et ae. 356
Atwotmeturerol-the: cytoplasm)... .:42%))/ 25 ccc fs belle ees Ces oll ee: 356
@DSerV-BTTONS a sa filo te Eco eh rin en ESS Es ee 356
I StoniealmanakCErticalr. jasc verds aren ere ae ei eoen caus 360
iE evirtochondriaisapparatus. . 62055. 5. G00 ee ee dae noe oles olees eae 364
erp lecinpubeptiony mermiys fic) (yk kOe CRM NRRL, CANIS eT 28 Cae 364
2 Shane soructire, and positiony S01). ¢ tee cud octievay cee ean oe eee 366
Sh (ESTEVE oe en tee enaR eC Fa a SRS AC SoM RO pre 369
CPP ataKepoT allcaterer. cases nays ote epee aca ee nea eiatines LUA cee tedhia 375
APES CHTITOUC MER pa caeh2 2s eS ae DAE ers cncRee, Aa nhs a AINE AMES 375
Ze SMA CLAUS OSEEION's\\h2 ae sasSedo Sherer VRE Tat oki dl ees SSE 376
COLIC SIS MEORM ie itrh 1. Unc ene RR ke 8 Site Iw 376
A SICAMMMICANCES sepsis ou «0 cis se eve SU seat Can enee Mavane cicncte Sicncuce crater 377
LDR Psat cela ern errsata.Ces a a A a Dl 379
IRRPReCHmbopLelye meres !. cs of hTAt Se ee UIE reir) 379
APS RADE BSIZe MANE POSIELON,:: «acu kontelesien (tered bin Haibecrs, Siptend eames 379
DG CIC CISPR A d. 3.c.-2 spatsraeae a eters esau nrclcuews cA eas eae 380
Ay EX CRUSTGM PEPER ook, oo eas a ot oe ee os le aaa Meee 386
E. Intracellular net- and canalicular apparatus....................... 388
ictoricals ars wee reiace ol Aes et. heagilaes dose tek ve.aortas eeeaaies 388
GIS OTIViG UIOUIS mam Rene 2 hy eee acoso edtsuey anykt A 388
PME ORECe UM ERTE MOTI. 2). tisha hee atsore aie oc ales WR AE 388
2 ACanalicwlesan rear e As ety heehee AS Fee 392
3. Significance and genesis of the intracellular net-apparatus... 393
Mee iologicd! GegenerAtlOM ke. che ec.siesieeae ccss oho pares seid emu pevew aber wlbee 396
io UUTAGISTIST yl, ogee ae en ane "8 7.00 NAB a ON eg ee 399
Geabnersp-emiled nebenkGrnez™ wean Sho! hack soe cuok ae de ya cide aes 400
LUpsugncrib G0) SAAR @E | 42:210 EA ta Nem EL eR OR RELAY bk Some el 400
SUNOT AL S SA eee enn oo! NS: 2 nese RES ORS E Otis 2 Rls 402
347

348 S. SAGUCHI
1. INTRODUCTION
The pancreas forms a favorable object for the study of the

activity of secretory cells in general, and has been used by vari-
ous investigators who have studied the minute structure of the
pancreas cell in relation to the secretory process, since the com-
plete processes which take place in the pancreas cell during the
production of secreting matter are to be followed with great
clearness. In looking over the literature, there can be found
three important steps in the development of the structural par-
ticularities in the cells in question. The first step was taken by
R. Heidenhain (’80). He divided the cell-body into two distinet
zones; one, a light, apparently homogeneous outer zone turned
toward the basement membrane, and the other, dark granular
inner zone near the lumen. He suggests that the expenditure
of the matter which is accompanied by the transformation of
zymogen granules into secretion is going on in the inner zone,
while the deposition of the nutritive material and the elabora-
tion of zymogen granules take place in the outer zone; thus giving
us an idea of the fundamental importance of the process going
on in the pancreatic cell. The second step was taken by the in-
vestigations of Mouret (’95, ’05), Laguesse (99), Mathews (’99),
Garnier (’00) and others, who attached considerable importance
to the fibrillar structure first observed by Pfliiger, in 1869, and
assumed the participation of the filaments in the formation of
zymogen granules. Structures of this kind have since been of-
ten described under the name of ‘Solger’s basal filaments’ or
‘Garnier’s ergastoplasm,’ the significance of which is not yet
decided definitely. Benda (’97-’03), Meves, (’07, ’08) and
many others, finally, advocate a theory which makes the granules
or filaments, named either ‘mitochondria’ or ‘chondriocontes,’ the
most important morphological constituents of the cell. These
structures were also found in glandular cells, and have formed
a problem for various investigators which is still far from a
satisfactory solution.
The questions as to whether these structures and others,
above all the intracellular net-apparatus form essential constit-
GLANDULAR CELLS OF THE FROG’S PANCREAS 349
uents of the protoplasm of the pancreatic cell, and, if this is so

what are the mutual dependence among them, the mode of their
formation, and their relation to zymogen granules, are of much
importance not only for the study of the genesis of zymogen
granules, but also for the solution of the biological significance
of these structures in general.
As material of the present investigation, the pancreas of Rana
temporaria was employed.
2. THE NUCLEUS
Historical
Ogata (’83) was the first to give fairly detailed information re-
garding the structure of the nucleus of the pancreas cell, though
R. Heidenhain (’80) previously noted the existence of the larger
nucleolus in it. Ogata worked on the amphibian pancreas, and
showed that the nucleus has a distinct, deeply basic-staining
boundary membrane with some thickenings or prolongations,
from which numerous, extremely delicate, granular threads
arise; the latter passing through the nucleus, and dividing it
into a number of fields. In these fields, independent of the
threads, there are one or more larger or smaller nucleoli, the
staining reactions of which enabled him to distinguish two kinds;
some stained with haematoxylin, and others, generally single in
number, larger and stained by eosin. He termed the former
‘karyosomes’ and the latter ‘plasmosomes.’
These observations of Ogata were on the whole confirmed by
the studies of Kosinsky, Melissinos and Nicolaides (90), Eberth
and Miiller (92), Ver Eecke (’95), Mouret (’95, ’05), and many
others in the pancreas of various vertebrates. Mouret proposed
to designate Ogata’s ‘plasmosome’ as ‘nucleolus,’ ‘karyosomes’ as
‘pseudonucleoli’ or ‘amas nucléiniens.’
Observations
1. Chromatin network. The nucleus (figs. 13, 28, 28, 37),

stained with ordinary methods, e.g., with haematoxylin-eosin,
350 S. SAGUCHI
exhibits one or two spherical nucleoli, reddish in color, and nu-

merous blue-stained chromatin granules, which, connected by
faintly stained, fine threads, form the so-called ‘chromatin net-
work.’ And the chromatin granules are, on the one hand,
closely applied to the surface of the nucleolus and are fused to-
gether so as to envelop the latter more or less completely; thus
forming what is usually called ‘chromatin shell’ (figs. 9, 12, 13,
22, 23, etc.), which seems to have been noticed by Ogata (’83),
Mouret (’95), and Arnold (’12) in the nucleus of the pancreatic
cell. Owing to the presence of this chromatin shell, it often
happens that in tangential view the nucleolus shows no char-
acteristic staining reactions, but stains blue like a chromatin
mass, as noticed by Arnold. The chromatin granules are, on
the other hand, applied upon the inner surface of the nuclear
membrane which appears, in cross-sections of the nucleus, as a
relatively thick basophilic line. The basophily of the nuclear
membrane is due to the extremely thin chromatin layer, which,
in surface view, reveals a delicate, faintly staining network.
The chromatin substance of the nucleus is not only accumu-
lated, as above described, in the form of granules, but also in the
form of irregularly curved cords. These chromatin granules and
cords are not homogeneous, but exhibit, in various fixations and
stainings, a thin, deeply staining cortex and a faintly staining
main-mass, which suggests that the two parts are perhaps dif-
ferent in composition from each other (figs. 22, 25, 30).
2. Nucleolt and nucleolar corpuscles." Most of the nuclei have
only one nucleolus; beside this there may exist several smaller
nucleoli; the larger one is then named ‘main-nucleolus’ and the
smaller ones are ‘side-nucleoli.’ There are also a number of
nucleolar corpuscles in the nucleus.
a. Main-nucleolus. The main-nucleolus is, in most cases, of
spherical form with an even contour, but sometimes oval, round-
ish triangular or polygonalinform. In the vast majority of cases
it is single in number, but it sometimes happens that two, three,
or four nucleoli of nearly equal size are seen in the nucleus (figs.
4, 23, 37); they are possibly formed by a division which is ef-
fected not by the scission, but by the stretching out of the nucleo-
lus, as seen in figure 5. The main-nucleolus is commonly located

in or near the centre of the nucleus. It can never be seen applied
to the nuclear membrane.
In the main-nucleolus one can notice three substances, which
differ in staining reactions from one another:
Main-mass: The main-mass occupies the largest part of the
main-nucleolus; it is well preserved by fixatives containing no
acetic acid, and stained by iron-haematoxylin or Altmann’s
acid fuchsin (figs. 1 to 8, 14 to 16, 21, 24, 63 to 65). The most
favorable fixatives for it are alcohol, sublimate and formalin.
In fact, acetic acid in the fixatives considerably diminishes the
affinity of the main-mass to the stains referred to, especially to
iron-haematoxylin. In preparations which are obtained from
pure sublimate material and stained by iron-haematoxylin, for
instance, we see that the main-mass of the nucleolus takes a gray
to black color, while the chromatin granules or cords are very
easily decolorized (figs. 1 to 6). When the same method of stain-
ing is applied to sections made from the sublimate-acetic mate-
rial, it will give an opposite result: the main-mass of the nucleo-
lus appears as a perfectly clear space surrounded by a deeply
staining shell (figs. 12, 80). It is probable that this aspect is
not owing tothe disappearance of the main-mass by the dissolving
effect of acetic acid, but is to be sought in the loss of its affinity
for iron-haematoxylin in consequence of the changes either of
chemical or of physical properties produced by the above reagent;
for, in sublimate preparations, the main-mass still can be made
manifest by staining with eosin.
Cortical substance. This is the second constituent of the
main-nucleolus and surrounds it in the form of a cortex (figs. 17
to 20, 55, 79, 10, 11). It always stains more heavily than the
main-mass with most of the dyes; this is especially well marked
in preparations which are obtained by the Benda staining of sec-
tions made from Champy or Meves material (figs. 17, 18, 26).
The cortex is, in reality, comparable to a crust, so that the corti-
cal substance passes gradually into the main-mass. The contour
of the cortex is not smooth, but irregularly indented or with small
granules or short rods exhibiting thé same staining reaction as
B15, S. SAGUCHI
the cortical substance. Thus the nucleolus as a whole assumes

the confect-shaped appearance. It must especially be noted
here that the cortex is question is not to be confounded with the
chromatin shell mentioned before, which is closely apposed to
the nucleolus; they are different from each other both in form
and in staining reactions, and this can be observed with ease in
such preparations as those which are made from the same fixed
material, but stained in various ways (compare figs. 10 and 11
with fig. 13).
Argentophile granules. The third substance occurs in the
form of minute granules in the main-mass, which stain a brown
to black color with the Cajal’s photographic method, using for-
malin prefixation (figs. 31 to 35). I will designate it ‘argentophile
granules,’ in order to denote its affinity for silver salts. They
can never be made manifest, so far as I could ascertain, by such
fixatives as sublimate, formalin, trichloracetic acid, alcohol,
potassium bichromate and osmic acid, with various stains, but
appear only as small clear vacuoles (figs. 1 to 4, 6, 8, 13 to 16, 29,
30). These granules are of spherical form and of variable size,
although even the largest ones do not exceed one-fifth of the
diameter of the nucleolus. They are single or several in number
and are situated, in most cases, in the middle of the nucleolus.
In case a large number of them are gathered together, it often
occurs that the centre of the nucleolus presents an alveolar ap-
pearance (fig. 29). The granules are not infrequently seen near,
or closely applied, to the periphery of the nucleolus (fig. 8).
The granules are not entirely homogeneous, but seem to con-
sist of a deeply staining cortical layer and a more slightly stain-
ing internal part, as seen in a Cajal’s preparation (figs. 31 to 35).
I have also noticed that the same granules are present outside
the nucleolus or are scattered in the nucleus; they vary in num-
ber, and are either situated near the nucleolus or are apart from
it to a greater or less extent, or even are applied to the inner
surface of the nuclear membrane. It is conceivable that the ar-
gentophile granules are passing out of the nucleolus and proceed-
ing towards the nuclear membrane; from the existence of a few
granules in the cytoplasm near the nucleus, as seen in some cases,
it would seem that they pass through the nuclear membrane.
GLANDULAR CELLS OF THE FROG’S PANCREAS B00
It is a difficult matter to dotermine definitely the significance

of the argentophile granules. They are, in all probability, prod-
ucts of the metabolism going on witirin the nucleolus, which
are to be eliminated from the latter and are to reach the cyto-
plasm in the manner above mentioned. It is quite within the
bounds of possibility that they are associated in some way with
the functional activity of the cytoplasm or they are, as waste
matters, to be cast off from the cell. Of these two possibilities
I am inclined to believe that the first one is more probable, since
the argentophile granules are often produced in considerable num-
ber in some alveolar cells of the pancreas which are undergoing
a certain change of functional activity, with which I shall have
occasion to deal at another time.
The structure of the nucleolus in the egg-cells of several inver-
tebrates and lower vertebrates has been a problem for various
investigators; some believe to have found a reticulation or vacuo-
lation of the nucleolus; others granules in it, which are known
under the name of ‘nucleolini.’ These structures can be mor-
phologically identified either with the argentophile granules or
with the negative of them, as I have seen. That the granules in
the nucleolus can be made manifest by the Cajal’s photographic
method is not novel; Lache (’06) Cajal (09), and others have
found the similar granules impregnated in the nucleolus of the
nerve cells; moreover, Lache’s figure 1, which is drawn from a
Cajal preparation, shows that the same granules are in the nuclear
sap outside the nucleolus. In the pancreas literature I find in this
connection only the paper of Calier (’96) who suggested the ex-
istence of the nucleolini in the pancreas cell of the hedgehog and
gave the name of ‘endonucleolus.’
b. Side-nucleolus. Some of the nuclei of the pancreatic cells
have, besides the main-nucleolus just spoken of, one or two side-
nucleoli (figs. 2, 5, 6, 8, 16, 21). It seems, however, that they
are lacking in many of the nuclei. The side-nucleoli are of vari-
ous sizes; but even the largest one seldom exceeding one-third
of the diameter of the main-nucleolus. They are spherical cor-
puscles with an even contour, and bear a remarkable resemblance
in their staining reactions to the main-nucleolus, though, in some
354 S. SAGUCHI
respects, differences are observed between the two. In the prep- .

arations fixed in formalin, sublimate-formalin, alcohol, bichro-
mate-formalin, Zenker’s, osmium-sublimate, ete., and stained
after Altmann, the side-nucleolus can be more easily decolorized
then the main-nucleolus. In the preparations fixed in sublimate
and stained with iron-haematoxylin, on the contrary, the former
stains more deeply than the latter and exhibits also a more
heavily stained cortical layer (fig. 2). The side-nucleoli have no
structures in their interior. They are, in most cases, situated in
the region intermediate between the main-nucleolus and the
nuclear membrane, but it is not uncommon to see side-nucleoli
closely apposed to the main-nucleolus (fig. 8), or appearing to
be constricting off of the main-nucleolus (figs. 5, 6). The latter
fact strongly suggests the derivation of the side-nucleoli from the
main-nucleolus.
c. Nucleolar corpuscles. Under this appellation I include cer-
tain small corpuscles in the nucleolus, which are stained black,
as seen in the preparations fixed in sublimate, alcohol, or subli-
mate-formalin, and stained with iron-haematoxylin (figs. 1 to 4,
7, 15, 16); in those fixed in sublimate-acetic acid, Champy’s, or
Meves’ fluid, they stain dark blue with crystal violet, and red
with acid fuchsin or safranin (fig. 20). They are either small
granules or rods of all shapes and sizes, which seem to vary with
the individual nuclei and the properties of the fixatives employed.
They are not only scattered through the nucleus, but also are in
close apposition to the nucleolus or to the inner surface of the
nuclear membrane.
It is a difficult matter definitely to determine the relation which
might exist between the position of the chromatin granules or
cords and the nuclear corpuscles; for they cannot be brought into
view by staining in one and the same preparation. ‘They, in
fact, differ from each other in staining reactions. In the prepa-
rations preserved in Champy’s or sublimate-acetic mixture, and
stained with Benda’s crystal violet, for example, the nucleolar
corpuscles take on a dark violet color, while the chromatin cor-
puscles are left unstained (figs. 10, 11, 17, 18). That the nucle-
olar corpuscles are, however, attached to the chromatin network,
may be rendered evident when we compare with each other the

preparations which are fixed in the same manner, but tinged in
different ways, e.g., the iron-haematoxylin stain with the Benda’s,
as applied to the material fixed after Meves (figs. 26 and 27),
or the haemalum stain with the Benda’s as applied to the mate-
rial fixed in sublimate-acetic mixture (figs. 10,11 and 13). From
these observations I have gained the impression that the nucleo-
lar corpuscles are imbedded in the chromatin granules or cords.
This ‘mpression was deepened by the appearance of the clear
vacuoles or canalicules in the middle of the gray- or blue-stained
chromatin granules or cords, in the preparations fixed in trichlor-
acetic acid and stained with iron- or alum-haematoxylin (figs.
29, 30). I concluded that this was, at least in part, a negative
appearance due to the nucleolar corpuscles being dissolved out
or remaining unstained.
As regards the genesis of the nucleolar corpuscles, I am fully
convinced that they are derived from the main- or side-nucleoll.
As before stated, the main-nucleolus is surrounded by a deeply
staining cortex, the outer surface of which is not only uneven, but
frequently furnished with granules or rods of various shapes and
sizes but of the same staining reaction as the cortical substance
(figs. 17 to 20, 26, 55). These granules or rods often appear to
be separating from the cortex. From these facts and from the fact
that the cortex with its appendages exhibits the same staining
reaction as the nucleolar corpuscles, the inference is warranted that
the two are the same thing, and that the latter are formed by bud-
ding and separating from the former. Now the question arises,
is the picture above mentioned not to be regarded as representing
a fusion of the nucleolar corpuscles with the nucleolus in order
to cause the increase in volume of the latter? I, however, insist
on the view above advocated by me; for I hold that the nucleolar
corpuscles are correlated with the vital activities going on in the
cytoplasm, as will be described later on, and that their separation
from the nucleolus and their progress toward the periphery are
conditions which must be fulfilled in making the above assump-
tion. It is quite conceivable, even rational, that the nucleolar
corpuscles, in passing toward the periphery, would make their
356 S. SAGUCHI
way along or through the chromatin network, which is in accord

with what has been said concerning the position of the two.
That the nucleolus is concerned in the production of the nu-
cleolar corpuscles or similar bodies has been noticed by Mont-
gomery (99). This author, in a description of the subcuticular
gland cells of piscicola, pictures the nucleolus, after having en-
larged to a certain extent, as constricting off pieces of all shapes
and sizes, which are distributed through the nucleus; and that
this is not a picture of fusion of nucleolar corpuscles with the
nucleolus is apparent from the fact that they are all cast off into
the cytoplasm. On the other hand, the fuchsinophile granules
which were seen, by Galeotti (’95), in the nuclei of the Spelerpes’
pancreas and pyrenoid corpuscles which, after Meirowsky (’08),
appear in the nuclei in the formation of pigment, are, in all
probability, structures which have been constricted off of the
nucleolus or, at least, have the same origin as this.
3. THE CELL-BODY
A. Structure of the cytoplasm
Observations. In studying the structure of the cell in general,

especially of the glandular cell, it seems to me to be most essential
to discriminate between two groups of the cytoplasmic contents;
to one group belongs the cytoplasm proper or protoplasm that
forms the ground-mass of the cell-body, and to the other, that
group formed by the vital activities of the former, or, derived
either from the nucleus or from the outside; and, as instances of
the second group, mitochondrial granules or filaments, zymogen
granules, lipoid granules, etc., may be enumerated. In the fol-
lowing I will first take up the cytoplasm proper, while the other
constituents of the cell-body will be treated of separately.
The existence of a large number of zymogen granules and mito-
chondrial apparatus often renders impossible satisfactory de-
tailed observation upon the cytoplasm proper. Under these cir-
cumstances, it will be advisable to examine a cell containing
only a few zymogen granules or such portions in which they are
generally wanting, i.e., the basal portion of the cell. Alveolar
structures of the cytoplasm brought about by the accumulation

of zymogen granules bear no relation to the proper construction
of the plasm. We know, on the one hand, from daily experience,
that the plasm exhibits extraordinary diversity of structure ac-
cording to the nature of fixatives, and that, therefore, it would
be premature to draw any conclusion from such observations;
and, on the other hand, one searches in vain, in most cases, for
minute structures of the cytoplasm as studied in the fresh condi-
tion, since the cytoplasm then appears perfectly homogeneous
except for some ill-defined granules or filaments. It would also
be rash, from this, to conclude that the cytoplasm is entirely
devoid of structure. It is, for the present, indispensable to the
study of the plasm to allow various fixatives to act upon the
cytoplasm and to compare the structural changes so produeed
with one another. In fact, some of the fixatives act in such a
way that the plasm is as well preserved as seen in the fresh state;
others cause so pronounced changes as often to lead to the forma-
tion of very fantastic structures. There are many gradations
of the action of fixatives between these two extremes.
In general, osmic or chromic acid mixtures containing a very
small amount of acetic acid or without this, such as Altmann’s,
Benda’s, or Meves,’ fix in such a manner that the protoplasm
proper of the pancreatic cell, provided the fixatives acted with
sufficient power upon the plasma, shows no structural particu-
larities (figs. 44 to 47, 55, etc.). On the contrary, the fixatives
which contain a larger amount of acetic acid, such as Zenker’s,
Carnoy’s, sublimate-acetic mixture, ete., may give rise to fibrillar
structures in the cytoplasm (figs. 36, 37, 48 to 52, 110, 118).
These fibrillae appear in the form of striation which, at the sides
of the nucleus run principally in vertical direction; while, below
it, the horizontal course predominates. The horizontal fibrillae
often pass into the vertical ones so that there occurs a concentric
fibrillation around the nucleus. In preparations taken from the
aleohol-, sublimate-formalin, or bichromate-formalin material
containing no acetic acid, the above striation can also be seen
(figs. 48 to 52).
358 S. SAGUCHI
These fibrillae are extremely fine and stain a blue or violet color
with alum-haematoxylin, as is the case with the chromatin cor-
puscles of the nucleus. That this staining reaction is not con-
ditioned by the presence of chromatin substance will be apparent
from the fact that, in iron-haematoxylin preparations made from
the sublimate or Zenker material, the fibrillae take no color or,
at the most, a gray, whereas the chromatin network stains black.
Since it is seen that the arrangement of the fibrillar structure
in question cannot be explained by the mere precipitating action
of the fixatives, we are not justified in drawing the conclusion
that it is not preexistent, although it cannot be seen in the fresh
state or in the osmium preparation. I believe that the following
explanation can be put forward to account for the genesis of the
structure: It is admitted by some investigators that the cyto-
plasm contains, besides protein substances, a large amount of
lipoid substances. In the application of such fixatives as pre-
serve lipoid substances in their natural position, e.g., of osmic
acid solution, the plasma remains homogeneous; whereas the
fixatives containing alcohol, chloroform, acetic acid, etc., seem
to dissolve out either lipoids or certain protein substances so
that the fibrillae which form a morphological constituent of the
plasm, and which in the fresh state are obscured in the plasma
sap, become differentiated both in refrangibility and in tingibility.
In other words, the plasm which holds the fibrillae in position
becomes removed, and the latter are liberated. The fibrillae
which have thus been individualized, I propose to name ‘“proto-
fibrillae.’ They are extremely fine, and it is a matter of difficulty
to perceive their individuality even in thin sections. Much
more is this the case in the fresh state where it is necessary to
make examinations in thick layers. The fibrillae which we can
recognize in fixed preparations must certainly be regarded as
having to some extent undergone definite alterations, above all
adhesion of the protofibrillae. It may be conceived that the lat-
ler, individualized in consequence of the dissolving out of a cer-
tain plasma portion as above presumed, can no longer be main-
tained in their natural position and come, as it were, to float in
the cell-body, so that there is liability to adhere to one another.
The process of adhesion may proceed to varying extents according

to the nature of the fixatives, the success or failure of fixation,
the stage of cell function, and other unknown causes, thus leading
to the formation of many enigmatic structures, as have been de-
scribed by various investigators.
As above mentioned, the fixatives such as alcohol, trichloracetic
acid, formalin, bichromate-formalin, etc., cause only slight adhe-
sion of protofibrillae, so that, in such cases, parallel striation
or only a slight sign of net formation at most is produced (figs. 48
to 52). In fixatives ike Carnoy’s, Zenker’s, sublimate-acetic,
bichromate-acetic mixture, etc., on the contrary, the changes are
much more pronounced; this is especially true of the deeper
parts of the block where the cell is ordinarily poorly preserved.
There are a great many variations in the degree of adhesion.
Either fiber-bundles, or a network situated in the basal portion
of the cell-body may be formed, as seen in figures 36 to 38, 110,
118. On the other hand comma-, crescent-, or shell-like struc-
tures (figs. 40, 84) may be formed by the adhesion of a large
number of fibrillae and then applied either to one side of the
nucleus or to the cell periphery. Finally, smaller or larger
spherical corpuscles may be formed by the agglomeration of the
fibers and show more or less pronounced concentric striation (figs.
37, 39, 84, 100, 101, 105). There are all grades of transition be-
tween mere striation and agglomeration of the fibers, and it often
happens that not only the different cells, but also one and the
same cell contains different types of structures. It must also be
here noticed that the structures formed by the adhesion of
a large number of fibers, such as thick bundles, commas, or
crescents, etc., generally stain more deeply with haematoxylin,
which is perhaps due to their compact character in consequence
of the close apposition of the constituting fibrillae.
From the above description it is evident that a large amount
of acetic acid in fixatives produces remarkable adhesion of the
fibrillae; yet the*superficial cells, even in such fixatives as bi-
chromate-acetic or sublimate-acetic mixture, are fairly well pre-
served, so that there occurs only slight adhesion (figs. 80, 81) or
none at all. This discrepancy can perhaps be explained by the
360 S. SAGUCHI
rapid action of potassium bichromate or sublimate, which restrain

acetic acid from destructive action. The diffusibility of acetic
acid, on the contrary, is great, so that it penetrates more quickly
and produces changes before other constituents of the fixative
could come to act. This is the reason why the products of adhe-
sion increase in number as we proceed toward the deeper portion
of the piece.
On the other hand, there are cases in which the formation of
fibrillar structures cannot be ascribed to the shrinking effect of
the fixative alone. Marked differences in the structures above
mentioned are often observed in individual animals, so that, pre-
served in one and the same fixative, one exhibits no remarkable
change, whereas in the other the structures are formed in so
large number that almost all the cells contain them. In the
latter case, the structures can well be seen even in the osmic
acid fixing material, where they appear indistinctly striated or
even homogeneous (fig. 88).
It happens also that certain granules, stained black by iron-
haematoxylin, are interposed among the fibrillae of the structures
in question (figs. 88, 105). If some of these granules, as often
occurs, are located in the center of the spherical body with con-
centric striation, an appearance of a centrospheré with central
corpuscles is produced. That this is, however, not the case is
obvious from the fact that the granules are situated not only in
the centre, but also in the other parts, even outside of the body
(fig. 105), and also that the number of the bodies may be two
or over; the granules in question are, rather, to be regarded
partly as zymogen or lipoid granules, partly as mere precipitates
of osmium, as will be referred to later on.
Historical and critical. Regarding the cytoplasmic structure of
the secretory cell, especially of the pancreatic cell, there is wide
divergence of opinion among various investigators.
1. Homogeneity of the protoplasm. As early as 1869 Langer-
hans described that the basal portion of the pancreas cell in
fresh state, is entirely homogeneous. Later observations of R.
Heidenhain (’80), in fresh materials, and of Mouret (’95), in pre-
served cells, confirmed this. Hoven recently (710) likewise
expressed the view that, though the inner zone of the pancreatic
cell becomes alveolar after the extrusion of zymogen granules, the
remaining portion of it lacks special structure with the exception
of chondriocontes in it. It must be borne in mind, however,
that the above investigators and others who accept the homo-
geneity of the protoplasm have made their observations either
in its fresh state or in osmium preparations.
2. Many of the investigators believe that the protoplasm con-
sists of two substances: spongioplasm or mitom and hyaloplasm.
It is a matter of course that, when the zymogen granules are
dissolved out, the inner zone of the pancreatic cell assumes an
alveolar or reticular appearance, as described by Langley (84),
Eberth and Miiller (92), Ver Eecke (95), Zimmermann (98),
and Arnold (12). It isa mistake, however, to conclude that this
is the proper structure of the plasm; for such structure is nothing
other than an appearance produced by the accumulation of zymo-
gen granules.
The fibrillar structure in the basal portion of the pancreatic
cell was first noticed by Pfliiger (’69), R. Heidenhain (’80)
ascribed the striation in question to an appearance due to the
presence of canalicules. A similar structure was observed by
many of the’subsequent investigators: Eberth and Miller (’92),
Ver Hecke (’95), Zimmermann (’98), Mathews (’99), Laguesse
(99), and others. Solger (’94, ’96) has found the same thing
in the submaxillary gland of man and given the term ‘basal fila-
ments’ to it, and, on the other hand, Garnier (’00), who took
salivary glands, pancreas, liver, etc., of different species for study,
has related the striation which he named ‘formations ergastoplas-
miques basales,’ to the secretive activity of the glandular cell.
Similar structures have been since described under the name
either of ‘Solger’s basal filaments,’ or of ‘Garnier’s ergastoplasm’
by various workers, as Mouret (’05), Bouin (’05), Prenant (710),
Champy (’11), Bensley (11), Arnold (’12), Mislawsky (713), and
many others. Their nature has formed a problem of much dis-
cussion. In looking over the literature one can classify the
structures hitherto described, according to their form and struc-
ture, as follows: a) fine filaments which course in vertical direc-
362 S. SAGUCHI
tion through the basal portion of the cell, as described and fig-
ured by Ver Eecke, Mouret (’95, ’05), Zimmermann, Mathews,
Arnold, Mislawsky and others. Zimmermann and Mislawsky
have found that these parallel striations appear in both longi-
tudinal and transverse sections, and claim that this is, in reality,
nothing but a figure of the section of a lamellar system; >) homo-
geneous rods or fibrillar bundles, such as seen in the figures of
Eberth and Miiller, Solger (94), Garnier, Prenant, and others;
c) comma- or crescent-shaped, or spherical, either homogeneous
or fibrillar bodies which are usually classified along with the so-
called ‘nebenkern,’ and which have been found by Eberth and
Miiller, Ver Eecke and Mouret (’05) in the pancreatic cells. To
my mind, these three types are not distinct structures, but be-
long to one and the same system, for it is evidently seen from the
figures of the above authors that they are connected by all grades
of transition.
The opinions of investigators are not in agreement as to the
genesis and significance of the fibrillar structures.
1. Mouret (’95, 05) Mathews, Laguesse (’99), Prenant, and
others admit that the structures under consideration are pre-
formed in living cells and take part in the formation of secreting
matter either directly by breaking down into grartules or indi-
rectly. Mouret and Laguesse have given the name ‘substance or
filament prézymogene’ to it. The assumption that the nucleus
participates in the formation of secreting matter is perhaps based
upon the facts that, first, a relation in position and in tingibility .
exists betwen the nucleus and the filaments; secondly, that the
latter run vertically from the base, where the nucleus lies, toward
the top, and lose themselves among the zymogen granules; and,
finally, that the increase and decrease of the two are reciprocal.
Moreover, Mathews and Garnier believe that the filaments are
connected by one extremity to the nucleus and receive by this
means a certain substance necessary for the elaboration of the
secreting matter from the nucleus. They explain, in this way,
the similarity of the staining reactions of the two. This argument
and others advanced by Mathews, Garnier, and their supporters
are not sufficient, at present, to warrant the unqualified accept-
ance of the above hypothesis; since the two do not necessarily

exhibit the same reaction for various stains, and also we know
that secretion granules are assumed by many recent investigators
to be derived from the other constituent of the cell. This mat-
ter will be more particularly discussed when the genesis of zymo-
gen granules is considered.
2. In more recent times, Hoven (710), Prenant (’10), Champy
(11), and Mislawsky (13) identify the basal structures with
mitochondrial filaments, assuming that the former is a modifica-
tion of the latter artificially brought about by the fixation. But,
I cannot agree with this view; for these two structures may be
easily brought into view by staining in one and the same prepara-
tion, as aready described by Regaud and Mawas (’09), Arnold
(10), and Bensley (11); they also differ markedly from each
other not only in morphology and staining reactions, but also in
their biological significance.
3. The view of Eberth and Miiller (92) and Bensley (’11) ap-
proaches that of my own. The former authors state that the
fibers and rods in question are products of the adhesion of pro-
toplasmic fibers; but how this change is brought about, in other
words, whether they are preexistent or postmortem structures,
cannot be seen in their description. On this point, Bensley ex-
presses himself more distinctly, maintaining that the basal fila-
ments are fixation artifacts produced by the precipitation effect
of acid. He says: ‘‘the real basal substance of the cells is homo-
geneous. Itis quite possible, however, that the basal filaments are
preexistent in the living cell though invisible because they are
imbedded in a substance of the same refraction index, and that
they are rendered visible in the acid fixation by contraction.”
There is, however, some difference of opinion between Bensley
and me; he considering basal filaments as preexistent and ignoring
my protofibrillae. The basal filaments which we find in fixed
preparations I believe to be nothing but my protofibrillae to some
extent adherent into bundles.

364 S. SAGUCHI
B. Mitochondrial apparatus
1. Technique. For the fixation of mitochondria I used

Zenker’s, osmium-sublimate, Flemming’s, bichromate-formalin,
bichromate-osmium-formalin, bichromate-osmium-sublimate,
Meves’-formalin (without acetic acid) and Maximow’s mixture,
especially those advised by various investigators for mitochon-
dria: Meves’, Benda’s, Altmann’s, Champy’s, Kolster’s fluid, ete.
Sections were cut from 2 » to 5 w in thickness and stained mainly
with Heidenhain’s iron-haematoxylin, Altmann’s acid fuchsin,
and Benda’s crystal-violet.
Generally speaking, the fixatives containing a great amount of
acetic acid, e.g., sublimate-acetic, Zenker’s, Flemming’s, etc., fix
the mitochondria only in cells of the superficial layer. In the
fixatives which contain a small amount of that reagent, e.g.,
Benda’s, Meves’, etc., the mitochondria in the deeper portion
of the piece are well preserved; but there is no constancy in the
results with these fixatives, their action being often limited to the
more superficial region of the piece. The mixtures which pre-
serve the mitochondria in a fairly satisfactory manner throughout
the piece are osmic acid, osmium-sublimate, Altmann’s bichro-
mate-osmium-sublimate, Maximow’s fluid, etc., all of which are
without acetic acid. Of these, the bichromate-osmium-subli-
mate mixture works much better, for it not only preserves both
cytoplasm and mitochondria in equally good manner, but also
is adapted for either of the mitochondria stains above referred to.
In the following paragraphs I will try to discuss the action of
the reagents commonly used for the preservation of mitochondria.
a. Acetic acid. That acetic acid dissolves out or destroys
mitochondrial substance is a well-known fact, but it seems that
this does not necessarily occur when employed in combination
with other reagents. Fixed in mixtures which contain a large
amount of acetic acid, such as sublimate-acetic, Zenker’s or
Flemming’s fluid, both the cytoplasm of the superficial layer of
the piece and the mitochondria in it are preserved in a satisfac-
tory manner. This is perhaps due to the rapid action of the
reagents combined with acetic acid, the fixation taking place
before the latter can exert any destructive influence. Since the
penetration of acetic acid is far greater than the other reagents,
it is conceivable that the mitochondria in the deeper portion may
easily be destroyed by acetic acid before the other reagents could
get there. Hence, the fixatives with limited acetic acid, such
as Meves’ and Benda’s, bring to view the mitochondria in deeper
parts than in the above case. Finally, the fixatives which con-
tain no acetic acid preserve mitochondria evenly throughout the
piece. In the superficial portion of the piece, however, both cyto-
plasm and mitochondria are always better preserved than else-
where, no matter what fixative we use, provided that the dura-
tion of fixing is appropriately chosen. But it is worthy of remark
that osmic acid or sublimate is essential in order to preserve the
mitochondria, even in the cells of the superficial layer, in the
fixatives containing acetic acid.
In a word, acetic acid, when a small amount of it is employed
in connection with other reagents, favors the fixation of the cell
without destroying the mitochondrial substance. On the other
hand, it can hardly be said that the fixatives which contain no
acetic acid always preserve mitochondria in a good state, since
poorly penetrating reagents cannot rapidly act upon the mito-
chondria in the deeper parts.
b. Osmic acid is the most excellent fixing se eet But this
alone, or mixtures with a great quantity of it, seems to cause
the shortening as well as thickening of mitochondrial filaments,
especially toward the centre of the piece.
c. Formalin, used singly, also well preserves mitochondria.
But it cannot be said that this fixation is adapted for all of the
mitochondria stains. Applied to material fixed in this way, the
Benda stain gives very poor results, while iron-haematoxylin has
a greater affinity for the mitochondria in the deeper portions of
the piece. On the contrary, formalin material is most adapted
for Altmann’s acid fuchsin, so that, under favorable conditions,
the most beautiful and instructive preparations can be obtained.
The affinity of formalin material for iron-haematoxylin can be
increased by treatment with osmic acid or osmium-sublimate
mixture after the formalin.
366 . S. SAGUCHI
d. Chromic acid and potassium bichromate have been fre-

quently used for the fixation of mitochondria. So far as can be
recognized from my own experience, these reagents, used singly,
could not preserve mitochondria in any satisfactory manner.
On the other hand, it can scarcely be said that the presence
of these reagents in fixing mixtures is absolutely necessary to
the fixation of the mitochondrial substance; for, even in such
fixatives as osmic acid, formalin, osmium-sublimate, ete., which
contain no chrom-salts, mitochondria are well preserved. But
there seems to be little doubt that their presence in fixatives
favors the fixing and staining of mitochondria; Benda’s ‘Post-
chromierung’ will answer this purpose.
e. Sublimate is not essential to the preservation of mitochon-
dria; since, however, this reagent fixes the cytoplasm both rapidly
and evenly, good results can be obtained when applied in combi-
nation with other reagents, e.g., bichromate-osmium-sublimate,
Maximow’s fluid, etc.
It will be seen from the above account that the reagents which,
used singly, afford fairly satisfactory results for fixing and stain-
ing mitochondria, are osmic acid and formalin, and that chromic
acid, potassium bichromate or sublimate alone do not have such
action. The combination of reagents to prepare fixatives for
mitochondria can be seen from the following diagram:
osmic acid chronic acid or
oe potassium bichromate
7
formalin sublimate
It is shown in this diagram, on the one hand, that fixatives

can be prepared by combination of two, three or four reagents,
which are connected by lines with each other; and, on the other,
that osmic acid or formalin alone can also be employed as a fix-
ative for mitochondria; acetic acid may either be added to the
fixatives or omitted.
2. Shape, structure, and position. As described by Schultze
(711), Levi (12), and others, the mitochondria in the acinus cells
of the pancreas, in fixed and stained preparations, appear as rods
or filaments, i.e., chondriocontes, when the fixation is perfect

(figs. 44 to 65, 71 to 74, 80, 81, ete.). They are of even thickness
and of smooth contour; local enlargements and the like are usu-
ally lacking. Their course may be either straight or curved,
undulatory or corkscrew-shaped. When a filament abruptly
changes its direction at a certain point, an appearance as if it
carried a grain, is often produced; in the same manner, undulatory
or corkscrew-shaped mitochondria may appear, on superficial
examination, either as filaments carrying granules or as rows of
granules. So far as could be seen by most careful observation,
however, there were no chondriocontes carrying the so-called
plastes, described by some investigators, nor rows of granules
and free granules of mitochondrial nature, except in certain lo-
cations in the cell (figs. 57 to 66, 71 to 74, 77, etc.). Hoven’s
(10, ’12) observations on such granules were perhaps made in
unsatisfactorily preserved materials. In fact, it will be seen that
similar granules on the chondriocontes as well as granules of
various sizes disseminated in the cell-body increase more and more
as we pass toward the deeper parts of the piece, where the cell
structure is usually imperfectly preserved. Mislawsky (’11,
13) and Ciaccio (718) also claim that the formation of granules
is either an artificial or a pathological process. The shape of
mitochondria varies to a more or less extent according to the
fixatives employed. Osmic acid alone or mixtures containing a
great amount of it seem to cause the thickening and shortening
of chondriocontes, which, at the same time, take a straighter
course (figs. 62 to 64).
The chondriocontes, as seen in perfectly fixed preparations,
are independent; in other words, they neither ramify nor anas-
tomose, as described by Hoven (’10, 712), Levi (12), and others.
An appearance of the net formation, as described by Mislawsky
(11, 713), is perhaps due to the interlacement or eventual adhe-
sion of chondriocontes. Maximow (’16), also, reached the same
conclusion. The occasional occurrence of anastomosis, ramifi-
cation, or rings (figs. 41-43) is probably owing to fixation arti-
facts of chondriocontes, for we never see such structures in per-
fectly preserved portions of the section. On the other hand, I
368 S. SAGUCHI
do not agree with Hoven (10, 712) that the ramification or the
close side-by-side apposition of chondriocontes are pictures show-
ing their longitudinal splitting; for it will be seen that the mul-
tiplication of the mitochondrial filament, as will be shown later
on, takes place in a quite different manner.
In addition to long filaments, relatively short ones or rods are
often met with. Whether these rods are actual structures or are
brought about in the act of cutting sections is difficult to decide,
inasmuch as the production of short rods in such thin sections
as I have made is not unusual.
A few words concerning the structure of the chondrioconte.
There are some investigators who distinguish between a cortical
and a marrow-substance. This claim is made when the filament
is shortened or thickened, or is broken down into granules, in
consequence of unsatisfactory fixation. In perfectly preserved
cells, on the contrary, the chondrioconte is always homogeneous.
The. position of mitochondrial filaments or chondriocontes is
variable according to the degree of accumulation of zymogen
granules; generally speaking, they are crowded around the nu-
cleus more than elsewhere, so that they are never lacking here ~
even when they cannot be detected in the rest of the cell-body
(figs. 45, 47, 54). Such relation in position between the nucleus
and chondriocontes was pointed out by Policard (712), who
studied liver cells. He says that this relation changes but
slightly with the different stages of secretion. The chondrio-
contes below the nucleus course mainly horizontally, while at
the sides they ascend toward the distal end of the cell, the hori-
zontal filaments often passing over directly into the longitudinal
ones (figs. 48, 56). In case the zymogen granules are few in
number or are entirely lacking, the chondriocontes situated
over the nucleus have irregular courses (figs. 55, 56, 58, 60).
When the zymogen granules are densely packed together, the
chondriocontes do not usually enter the granular zone (figs. 44,
61, 79); so, a certain region over the nucleus, which appears clear
and contains only a few zymogen granules, is commonly free from
chondriocontes (figs. 39, 44, 61). This is characteristic of such
mitochondrial preparations as Benda’s, Altmann’s, and Meves’.
The significance of this region will be discussed later on.
3. Genesis. Historical. There is considerable difference of

opinion as to the genesis of mitochondria, which may be sum-
marized as follows: 1) Benda, Meves (’08, ’10), Duesberg (08,
10), Hoven (10), and others claim that mitochondria are plas-
mic structures derived from those of the male and female sex
cell, and transmitted from generation to generation. According
to this theory they are not derived from the nucleus. 2) Mito-
chondria are mere products of differentiation of the cytoplasm,
as described by Lams (’06) and Vejdowsky (07). The former
author believes that the mitochondrial granules of the egg-cell of
Rana temporaria are elaborated under the direction of the at-
traction sphere; Vejdowsky also claims that they are formed by
the regressive transformation of the centroplasm, as products of
the formative activity of the centriole. 3) Russo (’08, ’10) and
Comes (’13) are of the opinion that mitochondria are paraplasmic
structures drawn in from the outside. The authors were led to
this assumption from observations that the mitochondrial sub-
stance in the egg-cell is increased by excessive nutrition or
by a certain food (lecithin), whereas they decrease in the
fasting. 4) There are some who accept the derivation of the
mitochondrial substance from the nucleus. Goldschmidt (’04)
believes to have found chromidia, first described by R. Hertwig,
in various somatic cells of Ascaris, and identifies them with the
mitochondria. According to this author, the chromatin sub-
stance of the nucleus consists of two components—idiochroma-
tin and trophochromatin. The former remains behind in the
nucleus, the latter passes over into the cytoplasm. To the
category of trophochromatin belongs the chromidial or mitochon-
drial substance. On the other hand, Wassilieff (07) found that
deeply staining chromatin substance passes out of the nuclei of
spermatogonia of Blatta germanica, giving rise to mitochondria;
while Jordan (’12) is of the opinion that mitochondria are de-
rived from the cast-off fragments of chromosomes in the mitosis
of spermatogonia of the bat. Very recently, Alexeieff (17) be-
lieves to have substantiated the nuclear origin of mitochondria.
The mitochondrial filaments or chondriocontes in the secret-
ing pancreas cells must be multiplied in some way or other; for,
370 S. SAGUCHI
in spite of their continual expenditure in the formation of zymo-

gen granules, as will be explained, they never disappear from the
cell. Hoven (10, ’12) believes to have found that the chon-
riocontes in the pancreas cell either multiply by longitudinal
splitting or by elongation of the remainder. So far as I have
been able to ascertain, there was no evidence of multiplication
by the splitting of the filament; the second assumption of the
author is more plausible, yet it is only a hypothesis.
Observation. In my previous study of the amphibian epi-
dermis, I observed that in the physiological degeneration of the
glandular cells a certain substance in the nucleus passes, in the
form of filaments, into the surrounding cytoplasm, and that these
filaments cannot be distinguished by form and staining reactions
from the mitochondrial filaments present. I have, further,
noticed that, in the retarded formation of Leydig’s granules, cer-
tain nuclear corpuscles stained deeply by iron-haematoxylin
may supplement the above granules. From these observations
I have been led to the view that certain constituents of the cyto-
plasm are in close relation to the nucleus, and that possibly those
which are necessary for cell function are located in mitochondria
and derived from the nucleus. In order to solve this question,
I made a large number of thin sections from variously fixed ma-
terials and searched for possible relation between mitochondria
and nuclear contents. It should be noted that this work is not
only a matter of difficulty, but also may give rise to misinterpre-
tation, so that great care should be taken in judging the findings.
In close examination of preparations taken from variously fixed
materials, such as sublimate-acetic acid, formalin, sublimate-
formalin, Zenker’s, Meves’, Benda’s, Champy’s, Rubaschkin’s,
Maximow’s, etc., I have often met with a connection between a
mitochondrial filament and the nuclear network. In these prep-
arations, also, there are many cases where the one end of the chon-
drioconte appears to be attached to the outer surface of the
nuclear membrane. It would be rash, however, from this appear-
ance, to assume the close relation between the two. An actual
connection must be, and, in fact, is, formed in such a way that
one end of the chondrioconte present near the nucleus continu-
GLANDULAR CELLS OF THE FROG’S PANCREAS av
ously passes over to a cord of the nuclear network, which, in

turn, is attached either to the nucleolar corpuscles or to the nu-
cleolus (figs. 44 to 55, 71). In addition to those shown on the
plates accompanying this paper, many other instances of such
connection could be observed. Preparations fixed in formalin
or Meves’ fluid, and stained with iron-haematoxylin are most
favorable for the study, since both the mitochondria and the
cords of the nuclear network are deeply stained andthe continu-
ity of the two can easily be followed. Since this continuity can-
not be established with certainty in asurface view of the nucleus,
we must make cross-sections. Although it cannot be said that
the connection can always be detected in such sections, it is,
however, not rare that there are seen two or three connections
in one and the same nucleus (figs. 48, 52).
From the above observations the conclusion seems Justified
that the chondriocontes of the pancreatic cell are derived from
the nucleus, by a certain nuclear substance passing out through
the nuclear membrane into the cytoplasm. The opposite condi-
tion is not conceivable, for, there can be little doubt that the
chondriocontes, as will be afterwards described, are used up in the
formation of zymogen granules. Several objections might be
raised to the preceding considerations. First, if the chondrio-
contes pass over the nucleus, an appearance of a connection of
the former with the nuclear network might be produced when
viewed from the surface. I exclude such uncertain cases from
consideration, drawing my conclusion only from the observa-
tions made upon the cross-sections of the nucleus. Secondly,
mitochondrial filaments might be brought in contact with the
nuclear membrane accidentally or be superposed upon the nu-
clear area, by the action of the knife in cutting. I do not, of
course, profess to draw the above conclusion from such an
observation; the picture of the chondriocontes passing over to
the cords of the nuclear network must be continuous in order
to prove that the connection actually exists. In addition to the
direct evidence mentioned above of the nuclear derivation of
the mitochondrial filaments, there is another, indirect: that
mitochondria are never lacking near the nucleus, even when
372 S, SAGUCHI
the remaining portion of the cytoplasm is devoid of them, as

already noted by Policard (712) in the liver cell, and as I have
also ascertained in the present investigation.
Since a certain substance in the nucleus appears to be squeezed
out through the pores of the nuclear membrane, it must be ex-
pected that the primitive shape of mitochondria, at least in the
pancreas cell, will be filamentous, and that their ramification or
anastomosis, whether it may be a normal state or not, must have
been secondarily formed. And, in order that these filaments
maintain their form, it will be most essential that the organiza-
tion of the cytoplasm remain intact. If this organization is
disturbed in consequence of some external influences, either me-
chanical or chemical, the mitochondrial filaments can no longer
exist as such, but they become thick and short; even break down
into granules or small vesicles; this is always met with in such
portions as have suffered mechanical injury or are imperfectly
fixed.
There is no especially favorable position for the passing out
of the mitochondria; it can take place on either side of the nucleus.
Now the question arises, from what nuclear substance do the
mitochondria originate? As already stated, there are three kinds
of substances which take part in the formation of the nuclear
network: one is the basichromatic substance and the others are
those forming nucleolus and nucleolar corpuscles. It has also
been concluded that the substance forming nucleolar corpuscles
and the cortex of the main-nucleolus is of the same nature. On
the other hand, the staining reactions of the mitochondria bear
a strong resemblance to those of the nucleolar corpuscles. This
strongly suggests that there may exist some genetic connection
between the two. In fact, it can often be seen that in the prepa-
rations fixed in Champy’s and sublimate-formalin mixture and
stained with iron-haematoxylin, the nucleolar corpuscles are
drawn out into filaments which penetrate the nuclear membrane
and pass continuously into the mitochondrial filaments (figs.
53, 55). These observations have led me to the belief that the
mitochondrial substance is derived from the nucleolar corpuscles.
From the staining reactions of the mitochondria, the inference

would appear justifiable that it is quite different from the chro-
matin substance of the nucleus. In alum-haematoxylin prepa-
rations, the chromatin takes on a violet color, whereas the
mitochondrial filaments remain unstained and appear as fine,
clear canalicules in a violet background (fig. 83). In the Mevey’,
Champy’s, ete., preparations, on the contrary, the nucleolar
corpuscles and the mitochondia are deeply stained, while the
chromatin corpuscles are easily decolorized. This difference in
staining reactions is inconsistent with the hypothesis of Gold-
schmidt and his pupils that the mitochondria are identical wth
the chromidia which, according to them, are nothing other than
the chromatin substance passed out of the nucleus. Since, on
the other hand, there can be no doubt that the mitochondria are
derived from the nucleus, I do not also agree with Benda, Meves,
and their supporters in regarding them as mere plasmic struc-
tures. Just as little plausible are the views held by some that
they are either products of differentiation of the plasm or bodies
drawn in from the outside.
That the nucleus is the centre of propagation and metabolism
of the cell is now fully recognized by various investigators; this
is evidenced by the experiment that the cell-body devoid of its
own nucleus cannot further continue to exhibit its vital activity.
While the chromatin is generally regarded, from its behavior in
mitosis, as forming the material basis of heredity, its relation to
the metabolic activity of the cell has been a question which re-
mains at present unanswered. Concerning the biological sig-
nificance of the nucleolar substance, there is also wide diver-
gence of opinion among different investigators. In the following
I will first consider two opposed views advanced with regard to
the relation between the nucleolus and the metabolism of the
cell, and then summarize my opinion.
According to the view of some histologists (Flemming, ’82;
Rhumbler, 93; Montgomery, ’99), the nucleolus must be con-
sidered as stored-up nutritive material for the chromatin sub-
stance or for the whole nucleus. Haecker (’93, ’95) and his sup-
porters claim, on the contrary, that the nucleolus is not nutri-
374 S. SAGUCHI
tive material, but a decomposition product which separates in

or upon the chromatic elements during the vegetative activity
of the cell and nucleus, and which, at the beginning of mitosis,
is removed from the nucleus. This conception is termed the
‘nuclear secretion theory.’
From my observations referred to above, we cannot arrive at
any satisfactory solution of the problem of the origin of the
nucleolar substance. In a previous investigation (715) I have
observed that, in the physiological degeneration of certain glan-
dular cells in the amphibian epidermis, the nucleus is characterized
by the increase of the nucleolar substance, accompanied by the
simultaneous decrease of the nuclear substance. From this I
have been led to the assumption that the former might be formed
at the expense of the latter: a view which was also held, as early
as 1888, by Hermann. Since it is obvious from my observations
that the nucleolar substance passes out of the nucleus in order
to form an essential constituent of the cytoplasm, it must be
regenerated either at the expense of the nutritive material taken
in from the cytoplasm or by the decomposition of the chromatin
substance. I find it possible rather to accept as adequate the
explanation that the nucleolus is formed by the participation of
the chromatin substance than to regard it as a mere accumulation
of nutritive material. This conception is strictly opposed to
that of Flemming, Korschelt, and others, and is rather in favor
of the view of the ‘nuclear secretion theory’ of Haecker. I re-
gard, however, the nucleolar substance not as worthless decom-
position product, but as material necessary to give rise to such
an essential substance for the cell activity as mitochondria.
The transformation of the nucleolar substance into mitochondria
must be thought of as taking place in the following manner:
the main-mass of the nucleolus undergoes change in staining
reaction at the periphery and gradually is converted into a sub-
stance that forms the cortex of the nucleolus. The latter, on
the other hand, continually emits small granules or rods, which
pass toward the periphery along the nuclear network. These
separated corpuscles are those which we have termed ‘nucleolar
corpuscles.’ They, then, pass through the nuclear membrane
GLANDULAR CELLS OF THE FROG’S PANCREAS Bh)
into the cytoplasm in the form of filaments, which are nothing

other than mitochondrial filaments or chondriocontes. These
steps of changes must be a visible manifestation of the nuclear
and cell activity; these changes are not yet completed with the
formation of the mitochondria, but the latter must undergo fur-
ther changes in order to perform the special function of the
cell. I will return to this question further on.
C. Fat-like granules
R. Heidenhain (’75, ’80) was the first to point out the presence
of fat granules in the pancreas cell. He allowed dilute alkalis to
act upon the pancreas cell, and found that the zymogen granules
grew pale, while certain granules, which he took to be of fatty
nature, are characterized by the resistance which they offer to
that reagent. The fatty granules have since been recognized by
several investigators in various glands; in the pancreas by
Mathews (’99) and Laguesse (’00), more recently by Bensley
(11), Arnold (12), Mislawsky (13), and Maximow (716).
1. Technique. In preparations fixed in pure osmic acid or
in mixtures containing osmic acid, such as Meves’, Benda’s, etc.,
the granules in question take on only a pale grayish color; in
the Mislawsky’s fixative and in osmium-sublimate with formalin
prefixation they appear as black granules; they are also impreg-
nated with the Fauré-Fremit’s and Golgi’s method. In fixatives
containing a large amount of acetic acid, they appear as clear
vacuoles which remain unstained in either of the stains employed;
in preparations fixed in Meves’, Benda’s, Altmann’s, Champy’s,
etc., fluid and stained with iron-haematoxylin, they take on a
brown to black color.
It will be seen from the above account that the granules in
question are not blackened with osmic acid alone, but stain a
grayish color which grows more and more pale in the subsequent
manipulation of the piece and sections. In order that they may
be blackened by osmic acid, the action of a reducing agent is
necessary. In the cases above mentioned, formalin and pyrogal-
lic acid serve as such. In the Mislawsy’s mixture formalin to-
gether with osmic acid act upon the piece. Formalin may also
376 S. SAGUCHI
be employed as a prefixing reagent for a piece which is to be

treated afterward with osmic acid or osmium-sublimate mixture.
From this it is evident that these granules have relatively weak
reducing power, although they eagerly take on osmic acid. They
seem also to have special affinity for chromic acid, as is seen from
the Golgi’s chromium-silver method. From these reactions and
their affinity for fat stains, such as sudan III, scarlet red, etc.,
the inference is warranted that they are of fatty nature; hence
they have been termed ‘fat-like granules.’
2. Shape and position. ‘The fat-like granules are not scattered
through the cytoplasm, but are limited between the nucleus and
basement membrane (figs. 44, 47, 57 to 70, 105). Almost all the
cells contain such granules, few lacking them. They vary greatly
in number being often numerous and accumulated in one or more
heaps in close proximity rather to the basal surface of the cell
than to the nucleus (figs. 60, 65). They also vary in size, but
such as exceed the zymogen granules in bulk cannot usually be
seen. In shape they are spherical with even contour. Confect-
shaped granules as seen in Golgi preparations (figs. 69, 70) are
perhaps due to the shrinking effect of the reagents. Small
granules appear homogeneous, while in larger ones a clear dis-
tinction can be made out between a deeply staining cortex and a
faintly staining main-mass (figs. 60, 61). In rare cases the fatty
granules can occur in the portions above the nucleus (fig, 70).
3. Genesis. From the above observation that the fat-like
granules are situated near the basal surface of the cell, the ques-
tion arises whether they are not derived from the blood- or lymph-
vessels. I will answer this in the negative, for it can clearly be
seen that they take origin here from a certain constituent of the
cell.
In preparation fixed in osmic acid or Meves’ fluid and stained
with iron-haematoxylin, we can easily find those cells which con-
tain no fat-like granules, but some tortuous mitochondrial fila-
ments running horizontally between the nucleus and the basal
surface of the cell (fig. 56). Next,we can find in the same por-
tion of other cells chondriocontes carrying small spherical enlarge-
ments in their course (figs. 57, 61, 63); the latter can be so
GLANDULAR CELLS OF THE FROG’S PANCREAS Shi
numerous that a rosary-shaped picture is produced (figs. 57, 59,

63). In still other cells, there are free small granules, the chond-
riocontes carrying granules having disappeared (figs. 58 ,62).
These three figures are connected with one another by all grades
of transition and must be regarded as different stages of one and
the same process; in other words, some chondriocontes below the
nucleus disintegrate into granules. The latter are at first small
and show the same staining reactions as the chondriocontes; in-
creasing in size, they gradually change their staining character-
istics until the fat-like granules mentioned above are formed
(figs. 60-68).
The genetic connection between the mitochondrial and the
fat or fat-like substance in the cell has received attention of some
recent investigators. Altmann (’94) describes the fat as passing
into the cell not in the form of corpuscles, but as decomposition
products; the assimilation of fat within the cell takes place
through the cell-granules; in other words, the granules, being
loaded with fat, become gradually transformed into fatty gran-
ules. The more recent investigators, Bobeau (’11), Champy
(11), Dubreuil (13), Mayer, Rathery and Schaeffer (’14),
Cowdry (716), and Scott (16), point out a striking resemblance
in the chemical behavior between mitochondria and fat or lipoid
corpuscles, and admit that the latter may be derived from the
former. Concerning the fatty granules of the pancreas cell,
Laguesse (’00) believes that they are products of disintegration
of ‘filaments baseaux’ (mitochondrial filaments?), while Bensley
(11) noticed that the fat globules are often embedded in the
mitochondrial filaments present at the basal portion of the cell;
these observations afford direct confirmation of my view. It
would be of interest to make a comparative study of this point
in various kinds of cells. .
4. Significance. Mathews (’99) relates fat-granules to the in-
ternal secretion of the pancreas and says: ‘‘Other than these
bodies there isno histological evidence of the ‘internal secretion’ of
the pancreas. It is not impossible that the substances composing
the internal secretion are components of the cytolymph.”
Laguesse (’00), who considered the behavior of fat droplets at
378 S. SAGUCHI
different stages of secretion, has found that they disappear from

the glandular cell during its activity, and that they reappear and
increase during the resting period. He regards the fat-droplets
as stored up material; after zymogen granules have been formed
from the assimilated material, the remnant of the latter is used
up in the formation of fat-droplets. At the expense of these,
zymogen granules may be formed after continued fasting. Mis-
lawsky (13) and Maximow (’16) have also found that the fat-
granules make their appearance in large numbers on pilocarpin
injection. According to the former author, their appearance
must be pathological.
I have above mentioned that the fat-like granules are produced
by the transformation of mitochondria. The pilocarpin injection
affords an indirect confirmation of this view. As Mislawsky
and Maximow have observed, and I can confirm from my own
experience, the fat corpuscles become largely increased in number
by injection of that medicament. This is probably owing to the
fact that the production and expenditure of mitochondria are
thereby accelerated, which accompanies the increased production
of the fat-like granules. This is also in accord with the observa-
tion of Laguesse that the fatty granules of the pancreas cell are
increased in the stage of exhaustion.
In studying the ultimate fate of the fat-like granules, we must
take into account that they are formed near the basal surface of
the cell, and are never seen to pass toward the lumen. It is
well within the bounds of possibility that they become used up
here or pass out of the cell. It would seem from their staining
reactions that a part of them are undergoing definite alteration
(fig. 64); though it is not certain that all of them behave in the
same manner. On the other hand, it must be noticed that gran-
ules of the same nature are often met with in the capillary blood-
vessels (fig. 67) or in the connective tissue separating the latter
from the basis of the glandular cell (figs. 66, 68). Since we have
now determined that the fat-like granules are formed within the
pancreas cell, the pictures just mentioned cannot be regarded as
evidence of their passage from the capillary blood-vessel toward
the acinus-cell, but as exhibiting their movement in the opposite
direction. If such assumption is tenable, then the question

arises as to what part they play in the blood-vessel. Are they
to be regarded as mere decomposition products, or are they to
be taken up by other kinds of cells, exerting an influence in some
way upon the activity of the latter? I will take them in the
sense of Mathews as representing an internal secretion of the
pancreas cell. This is in agreement with a recent tendency to
ascribe the process of internal secretion not only to the islet cells,
but also to the parenchyma cells of the pancreas.
D. Zymogen granules
1. Technique. Zymogen granules can be well preserved in a

variety of sublimate mixtures and in many of the fixatives for
mitochondria, and can be stained with iron-haematoxylin, acid
fuchsin, etc. In preparations fixed in bichromate acetic acid,
aleohol, Carnoy’s and Flemming’s fluid, they are either imper-
fectly stained by iron-haematoxylin or remain unstained. From
this it appears that, in the fixatives containing a large amount of
alcohol or acetic acid, even if they contain osmic or chromic
acid, the affinity of zymogen granules for iron-haematoxylin is
much diminished, and that the sublimate mixtures, even if they
contain acetic acid, will preserve zymogen granules. That
acetic acid does not totally dissolve out or destroy zymogen gran-
ules, but perhaps acts upon them in such a way that a certain
constituent of the granules which has strong affinity for iron-hae-
matoxylin is removed, is evidenced from the fact that, although
the granules fixed in bichromate acetic acid remain unstained in
iron-haematoxylin, they can yet be demonstrated by staining
with eosin. In the preparations fixed in alcohol or Carnoy’s,
on the contrary, they can no longer be stained even by eosin.
2. Shape, size and position. As is well known, the secretion
granules of the pancreas are of spherical form and are accumu-
lated in the upper half of the cell. Their increase or decrease
has, in most cases, a marked influence upon the shape and size
of the glandular cell. Generally speaking, the cells containing
a few zymogen granules are smaller than those which are full of

380 S. SAGUCHI
them; still there are cases where this does not hold. Zymogen
granules, even when they are densely packed together in the cell-
body so that the nucleus is compressed against the base, can
never be seen between the nucleus and the basement membrane
(figs. 39, 44, 61). Above the nucleus there is usually a light
area in which zymogen granules either are wanting or are pres-
ent in small numbers; in other cases, it contains smaller granules,
yet it is well marked off from the remaining portion by a clear
zone (figs. 61, 75, 76, 78, 81). It can often be seen that the
area is subdivided by clear stripes into a number of parts, so that
it exhibits a reticular appearance. This area is usually situated
immediately above the nucleus, or more or less apart from the
latter; it is very seldom situated nearer the cell-periphery than
the nucleus. I shall return to this area a little later when the
genesis of zymogen granules is considered.
Zymogen granules exhibit a little variation in size, which indi-
cates that they have a definite limit of growth and are well indi-
vidualized so that they do not flow together as such. On the
other hand, we cannot find such small granules as coincide in
diameter with the mitochondrial filament, at least not in mito-
chondria preparations (figs. 39, 44, 55) in which some believe to
have found small granules, which according to them, must grow
up to zymogen granules.
Macallum (’91) admits that zymogen granules present next to
the lumen are larger than those near the nucleus, since they in-
crease in size during their passage toward the lumen by the depo-
sition of a certain substance derived from the protoplasmic area
of the cell; Mathews (’99), on the contrary, describes the gran-
ules next to the nucleus as always larger. So far as can be seen
from my preparations, there are no such relations between the
sizes of zymogen granules and their positions. I have also no-
ticed that the periphery of the granules is more deeply stained
than the center, but not to such an extent as to lead to the forma-
tion of ‘crescent-shaped corpuscles,’ as pointed out by M. Heiden-
hain (’90) in the pelvic gland of the triton.
3. Genesis. Historical. Regarding the origin of secretion
granules in general, there is full discussion in the papers of
Laguesse, Hoven ('12), Mislawsky (13), and others, to which

the reader is referred for details. In the following the views
of various investigators concerning the origin of zymogen
granules will be summarized.
a. Cytoplasmic origin. Langley (’84) and Carlier (’96) be-
lieve that zymogen granules are derived from the hyaline sub-
stance or hyaloplasm; an opinion which seems to have secured
few adherents among histologists.
That the filaments present in the basal portion of the cell take
part in the formation of zymogen granules was first suggested by
R. Heidenhain as early as 1880. The similar filaments have since
been described by various investigators under the name ‘Solger’s
basal filaments’ or ‘Garnier’s ergastoplasm,’ and there has been
much discussion as to their significance, especially as to their
bearing upon the origin of zymogen granules. Mouret (’95, ’05),
Laguesse (’99), and others believe that they are formed by disin-
tegration of the basal filament; hence the authors gave the term
‘une substance prezymogéne’ to the filaments. Garnier (’00),
on the other hand, has made a study of various glands, inclusive
of pancreas, and has come to the conclusion that the filaments
are indirectly concerned in the formation of zymogen granules.
According to him, the process is as follows: the nucleus, first
of all, increases in volume, accompanied by the enlargement of
the plasmic nucleolus and by the diffusion of the chromatic sub-
stance through the nuclear sap. In the next stage the basal
filaments come in contact with the nucleus and then is acquired
the basophilic property. Garnier explains this phenomenon
hypothetically by assuming that the nucleus furnishes the
chromatic substance by osmosis to the filaments; he gave the
term ‘excretion nucléaire’ to the process. Next, the basal
filaments leave the nucleus, momentarily, and distribute the
loaded chromatic substance throughout the cytoplasmic retic-
ulum. Then, there appear basophilic granules at the nodes
of the reticulum, while the chromaticity of the basal filaments
decreases. Finally, zymogen granules make their appearance in
the meshes of the network. The conception of Mathews (’99)
includes the two views above mentioned: the cell-threads (basal
382 S. SAGUCHI
filaments) which, according to him, are nothing other than the

chromatin passed out, give rise, by decomposition, to zymogen
granules.
On the other hand, there are many investigators who hold
that zymogen granules are derived from special cytoplasmic
granules or filaments. Altmann (’94), who made an exten-
sive study of the secreting process, came to the conclusion
that the secretion granules, whatever kinds of glandular cells
we may take, are formed by the enlargement of the ‘primary
granules,’ which multiply by means of the formation and divi-
sion of the ‘vegetative threads.’ Thus, he advanced the theory
that ‘‘die Sekretion ist ein granulirer Process.’’ The filaments
found by Michaelis (00) in vital stained preparations of the
pancreas and salivary glands of different species are in all
probability to be identified with the vegetative filaments of
Altmann; it was further noticed by Michaelis, who used janus
green combined with neutral red for vital staining, that, in
addition to the filaments stained green and zymogen granules
stained red, there are in the basal portion of the pancreas minute
granules which stain partly with janus green, partly with neutral
red, and which are to be regarded as young zymogen granules.
From these observations the author was led to the assumption
that zymogen granules are derived from the filaments in ques-
tion. Since Benda’s investigations, the. specific granules and
filaments which have been included under the term ‘mitochon-
dria’ have been recognized by several investigators in different
kinds of cells, inclusive of the glandular cells. That they take
part in the formation of secretion granules is assumed by Bouin
(05), Regaud and Mawas (’09), and others in the salivary
glands, by Hoven (710, 712), O. Schultze (11), Champy (11),
Arnold, (712), Maximow (’16), and others in the pancreas.
Concerning the mode of participation, there is some difference of
opinion: according to Regaud and Mawas, the mitochondrial
filaments can fix the substance extracted from the blood, which
becomes accumulated at one or several portions of the filament
in the shape of spherules; the authors regard the latter as the
foundation of secretion granules, assuming that they gradually
increase in volume and then become separated from the filaments,

which remain behind. On the contrary, Hoven, O. Schultze,
Arnold, and Maximow believe that the mitochondrial filaments
break down into granules, from which secretion granules are
to be derived. On the other hand, there are some (Levi, 712;
Mislawsky ’11, ’13; Ciaccio, 713) who definitely reject the
possibility that mitochondria participate, whether direct or
indirect, in the formation of secretion granules, regarding the
granular decomposition either as pathological (Ciaccio) or
as an artificial process.
b. Nuclear origin. That the nucleus of the salivary and
pancreatic glandular cell undergoes changes in shape, position,
volume, and chromaticity during the stages of secretion was
first noticed by R. Heidenhain (’68,’75,’80); in the resting
condition of the cell, the nucleus exhibits a shrunken appearance,
whereas during activity it is spherical and contains distinct
nucleoli. Similar changes have been subsequently observed
by various investigators (Schmidt, ’82; Hermann, ’88; Carlier,
96, M. Heidenhain, ’90; Steinhaus, ’90) and assumed by some
to be due not to the mere accumulation and disappearance of
secretion granules, but to an active process associated with the
genesis of secretion, without definitely involving a decision as 2
its mode of participation.
Some investigators (Macallum, 791; Mathews, 99; Garnier,
00; Carlier, ’99, 07; Maziarski, ’10) are inclined to assign to
the chromatin substance of the nucleus a part in the genesis of
secretion, assuming that it passes into the cytoplasm either in
a corpuscular or an amorphous state. That Garnier is of opinion
that the basal filaments are impregnated by the chromatin
substance passed out and that Mathews even regards the fila-
ments as the chromatin itself, has been stated before.
Finally, the nucleolus must be taken into consideration. It was
maintained by many investigators, as Ogata (’83), Platner (’89),
Melissinos and Nicolaides (’90), Ver Eecke (’95), Laguesse
(99, 00), and others, that the accessory nuclei (nebenkerne,
plasmosomes, corps paranuclaires) often found in the pancreatic
cell are nothing other than the nucleoli passed out, and that
384 S. SAGUCHI
they take part in the formation of zymogen granules after

undergoing certain changes in the cytoplasm. Galeotti (’95)
believes that both nucleoli and fuchsinophile granules pass out
of the nucleus in order to form secretion granules.
Observations. In the previous section we have found that
fat-like corpuscles are formed by the disintegration of mitochon-
drial filaments; we see here again that zymogen granules are
derived from the chondriocontes.
I have before pointed out that there is an area between the
nucleus and the free end of the cell, which either appears clear
in consequence of the absence of secretion granules or is dis-
tinctly marked off from the adjacent cytoplasm (figs. 39, 61,
75, 76, 78). Since the elaboration of zymogen granules takes
place in this area, it can well be termed ‘secretogenous area.’
The process is as follows: the mitochondrial filaments or chon-
driocontes first converge to this area; some of them being passed
into it (fig. 71). In other preparations there can be seen small
granules, the diameter of which is as large as, or a little larger
than, that of the chondriocontes (fig. 73). To see how they
have been formed, we must take another preparation, in which
there can be observed chondriocontes carrying spherical swell-
ings (figs. 72, 77). These different pictures must be regarded
as one and the same process going on in such a manner that
the chondriocontes disintegrate into granules in this area. These
small granules, for which it will be better to reserve the name
‘prozymogen granules,’ because of the fact that they give rise
to zymogen granules, are at first few in number, but they become
gradually augmented in consequence of the breaking down of
chondriocontes entering the area in succession (fig. 73). It
must be observed especially, in this connection, that the chon-
driocontes of the pancreas cell, except those which are located
close to the cell basis and in the secretogenous area, can neither
be seen carrying spherical swellings nor breaking down into
granules; an observation upon which I lay considerable stress as
differing from those investigators who believe in the break-
ing down of chondriocontes into granules everywhere in the
cell-body.
It is necessary to say here a few words with regard to the

fixing and staining characteristics of prozymogen granules. In
mitochondria preparations, such as Altmann’s, Benda’s, and
Meves’, they cannot be brought to light. The most distinct
pictures have only been obtained in preparations fixed in Miiller-
formalin and stained with iron-haematoxylin; they can also be
exhibited, though not distinctly, by Zenker-formalin (without
acetic acid) fixation and iron-haematoxylin staining (fig. 77).
From these reactions it follows that prozymogen granules are
preserved, in a satisfactory manner, by the prolonged action
of potassium bichromate, and that they are destroyed by the
presence of acetic acid, even in a very small amount, in the
fixatives. This is a characteristic property of the granules in
question, by virtue of which they can be distinguished from
mitochondria and zymogen granules; and this might be also
the chief reason why investigators who employed fixatives
containing more or less acetic acid for exhibiting the mito-
chondria, did not notice the prozymogen granules within the
secretogenous area.
The first-formed prozymogen granules are very small (fig. 73) ;
they gradually increase in volume (figs. 74, 75, 76, 78), probably
owing to their growth, but not to fusion with each other.
With increase in volume, they undergo changes in their stain-
ing reactions, thus becoming gradually transformed into typi-
cal zymogen granules. The young zymogen granules thus
formed can persist for some time within the secretogenous area
(figs. 75, 76, 78); they sooner or later leave it, however, to pass
into the surrounding plasma portion. Most of them pass
upward and lateralwards; but there are some which proceed
toward the nucleus (fig. 76). This indicates that there is no
uniformity as to the size of zymogen granules with respect to
their position; the occurrence of the larger ones near the nucleus
thus being explained. If the passing out of the granules of the
area is, for some reason or other, retarded, they will continue
to increase in volume, so as often to lead to the formation of
larger zymogen granules within the area (figs. 75,78). From
the above description, it is evident that there is a marked dif-
386 S. SAGUCHI
ference between the observations of some investigators (Regaud

and Mawas, Hoven, O. Schultze), who believe in the derivation
of zymogen granules from the mitochondria, and my own find-
ings. They believe to have found the disintegration of the chon-
dricontes and the increase in volume of the granules thus formed
up to the zymogen granules, in those preparations which are
usually employed to exhibit the mitochondria, such as Benda’s,
Mevey’, etc. I have, however, never been able to follow such a
direct transition between the two in the above preparations; but
there are granules intermediate between the mitochondria and
the definitive zymogen granules. These granules, prozymogen
granules are, therefore, different in behavior to fixing and
staining reagents from the two other constituents of the cell.
Secondly, it seems from the descriptions and drawings of the
above investigators that there is no locality of predilection for
the disintegration of the chondriocontes; whereas, in my cases,
the process takes place in the secretogenous area only.
So far as I have been able to ascertain, there was no note-
worthy relation between the shape, structure, and chromaticity
of the nucleus and the elaboration of secretion granules; in
spite of the great accumulation of the latter, there could not
be seen any considerable flattening of the nucleus, although
it is often pressed down toward the basis of the cell. The
mitochondrial filaments are usually more numerous in empty
cells (fig. 55); this would seem to indicate that their passing
out of the nucleus mainly takes place in such cells. In fact,
the connection between the chondriocontes and the nuclear
network can easily be seen in such empty cells; but it is not
rare, that it can be found in cells heavily laden with secreting
granules. These facts are in essential harmony with the find-
ings above mentioned that the nucleus exhibits no visible changes
in the structure and staining reactions according to the stages
of secretion; in other words, the mitochondria are continually
supplied from the nucleus as they are used up in the formation
of zymogen granules.
4. Extrusion. Concerning the question as to whether zymogen
granules are eliminated as such or after undergoing certain
changes, there is some difference of opinion. Some investi-

gators (Mouret, 795, 05; Galeotti, ’95) believe to have found
zymogen granules in the lumen of alveoli and ducts, and con-
clude from this that the dissolution of at least a part of them is
accomplished outside the cell-body. The opinion of Steinhaus
(90), Carlier (96), Garnier (’00), and others is adverse to this;
they maintaining that the granules are dissolved by the time
of elimination; while Miiller (’96, ’98), and Babkin, Rubaschkin
and Ssawitsch (09) claim that the granules liquefy within
the cytoplasm, forming the so-called ‘secretion vacuoles’.
Altmann (94), on the other hand, considers that the lique-
faction of zymogen granules takes place either within or outside
the cell-body, according to the kinds of glands.
So far as can be seen from my preparations, there is no indica-
tion of the extrusion of granules as such; it seems, rather, that
they are eliminated after undergoing certain changes, but
without having formed vacuoles with liquid. In preparations
treated according to Marchi and stained with iron-haematoxylin,
in which the zymogen granules remain unstained, so that the
upper half of the cell-body exhibits a reticular appearance,
there can be discerned in close proximity to the lumen some
granules or masses, which are irregular in shape and stained a
grayish color, and which show a marked tendency to fuse
together (fig. 84); they can never be found elsewhere. While
in many preparations the contents of the lumen appear clear
and homogeneous, in the Marchi preparation there can often be
seen irregularly shaped masses, the staining reaction of which
agrees with that of the masses present at the free end of the cell.
Under favorable conditions, even the continuity of the two
with each other can be discerned. From these observations, the
inference would appear justifiable that zymogen granules, after
having reached the distal end, undergo changes in physical and
chemical properties and form by fusion the grayish staining
masses above mentioned. ‘These pass out of the cell through
the pores into the lumen, a mode of secretion, which is in full
accord with that found by me (’15) in the dorsal glandular
cells of Hynobius larvae.
388 S. SAGUCHI
EK. Intracellular net and canalicular apparatus
Historical. The existence of canaliculi or vacuoles in the

glandular cell has often been noticed by various investigators.
Laserstein (’94) describes clear paths in the granular zone
of the pancreatic cell as canaliculi running through the cyto-
plasm. E. Miiller’s investigation (98) of the gastric fundus
gland has drawn attention to these structures. These cells con-
tain, according to this author, a network of clear distinct bands
which are connected with the intercellular secretion capillary,
and which are nothing other than the secreting matter formed
within the cell; he identified the network with a structure pre-
viously found by him by means of Golgi’s method and named
‘Korbkapillaren.’ These observations have been confirmed by
Zimmermann (’98). The latter describes the intracellular secre-
tion capillaries as limited to the gastric fundus gland, the sweat
gland, and the liver, in contradistinction to the view of Krause
(95), who had found canaliculi both in serous and mucous
cells of the salivary gland.
The existence of a typical network in glandular and epithe-
lial cells, on the other hand, has been recorded at some length
by Negri (00). He has demonstrated the network of the
pancreas cell by means of Velatti’s mixture and found it inde-
pendent of the intercellular secretion capillaries impregnated
at the same time. Similar networks or canaliculi have since been
described by Holmgren (’02), Bergen (’04), Babkin, Rubasch-
kin and Ssawitsch (09), Kolster (’13), Bensley (’11), and
others in the pancreatic cells of various species.
Observations. So far as I have been able to determine, there
can be made manifest, in the pancreatic cells of Rana temporaria,
either networks consisting of solid cords or canaliculi, according
to the technique employed. Although both pictures seem to
belong to one and the same structure, as will be afterward
explained, they are conveniently treated here separately.
1. Intracellular network. a. Technique. By ‘intracellular
network’ we mean the apparatus which can be made manifest by
various methods, such as Kopsch’s, and Weigl’s osmium method,
Sjovall’s formalin-water-osmium method, Cajal’s uranic nitrate-

silver method, and Golgi’s arsenic method. Of these I have
obtained excellent results with Kopsch’s, Weigl’s, Sjévall’s and
Cajal’s methods, whereas Golgi’s method afforded no satis-
factory results in spite of repeated trial. In addition, the net-
work may occasionally be brought into view by the Flemming
fixation and iron-haematoxylin staining. The network can also
be exhibited, though in an ill-defined manner, in preparations
fixed in Zenker’s fluid and stained by iron-haematoxylin (figs.
80, 81).
b. Shape and position. The intracellular network (figs.
88 to 96) is situated above the nucleus; it is occasionally placed
near the distal end of the cell rather than the nucleus, and
spreads to a greater or less extent basalward along the sides of the
nucleus. The area which is occupied by the network has all
sorts of indefinite shapes and varies greatly in extent. The net-
work itself consists of thick or thin cords which ramify and
anastomose in various ways. Sometimes the net is not com-
plete, so that long or short cords or fragments of the net are scat-
tered through the granular zone. In spite of these varieties,
there can be seen no intimate relation between the network and
the nucleus; they are never in close apposition to, or in direct
connection with, each other.
The network, as seen in a Kopsch preparation (fig. 88), is of
compact character, the cords being relatively thick and those
projecting into the surrounding plasm being usually short or
often lacking. In the remainder of the cell, especially in the
basal portion, there can be seen fine filaments or granules stained
black and often attached to the network. Iam unable to deter-
mine whether these structures are to be regarded as consisting
of the same substance as the network or as mere precipitation
of osmium.
In preparations made according to Weigl’s (figs. 92 to 94),
or especially Sjévall’s (figs. 89 to 91) and Cajal’s methods (figs.
95 to 98), the shape of the network, as a rule, is very irregular,
extending over a polygonal or elongated area. ‘The most striking
picture obtained by these manipulations is that the network
390 S. SAGUCHI
emits shorter or longer prolongations which either consist of

solid cords or still exhibit a reticular structure, and which can
be traced through the granular zone to varying distances; even,
under favorable conditions, as far as the distal (figs. 91, 93, 95,
96) or lateral margin of the cell (figs. 89, 94). These long pro-
longations, if present, are usually one to three in number, and,
in either case, they can never be followed toward the cell basis.
Further, in the Cajal preparation, spherical corpuscles stained
the same color as the cords may often be seen suspended on the
network (fig. 98). If such corpuscles are present in large num-
bers, the typical network is no longer visible, the cords of the net
being represented by rows of granules or droplets, which are
arranged in such a manner as to form an ill-defined network
(fig. 97). The same corpuscles are also seen scattering sparsely
through the granular zone, some of them being attached to the
distal or lateral border of the cell (figs. 95, 97, 98); even some-
times lying in the lumen.
From the above observations it seems most probable that the
substance constituting the network is being discharged either
by means of projections or in such a way that the cords disinte-
_ grate into granules or droplets, both projections and droplets
passing through the granular zone to reach the cell border.
It is also probable that, of these projections and granules, those
which are directed toward the distal margin must be extruded
into the lumen, while those which proceed toward the lateral
margin reach the intercellular canaliculi. In fact, the lumen
and especially the intercellular canaliculi are often found
filled with substance which shows the same staining reaction,
the droplets or projections in the cell-body being in direct con-
tinuity with the mass outside the cell (fig. 98). As these pic-
tures are by no means artifacts, nor can be interpreted as repre-
senting the passage of a substance from the outside into the cell,
it is, I think, impossible to avoid the conclusion that the net-
work is composed of a secreting material which is formed within
the cell and which is to be discharged into the lumen or into the »
intercellular canaliculi.
No mutual relation could be found between the extent of the

network and the amount of zymogen granules. There are
cases in which the former is well developed, while the latter are
few in number, and vice versa. The zymogen granules are,
in general, not found in large numbers between the meshes of
the intracellular network.
A similar network to that described above can be made mani-
fest by Flemming fixation and iron-haematoxylin staining
(figs. 100 to 105). It can scarcely be said that this method is
adapted to any animal, for the conditions under which the net-
work is fixed by Flemming’s fluid seem to depend either upon
the nutritive state of the animal, or the functional activity of
the cell, or some other unknown causes. I could demonstrate it
by means of the above method in sections of the pancreas taken
from an animal captured in the month of September. The
structure is located above the nucleus and consists of thick
or thin tortuous cords which form a network by interlacement
and anastomosis with each other. The cords are stained black
by iron-haematoxylin, and are, in most cases, solid, but some-
times canalized to a greater or less extent; the canaliculus is
bordered by two black lines, which are the optical section of
its wall. Just as has been observed in other preparations, the
network also emits longer or shorter cords or canaliculi which
pass through the granular zone; the longer ones often extending
to the cell border. It can further be seen that the darkly stain-
ing granules or droplets are either suspended on the cords or
canaliculi, or scattered through the granular zone, or even
attached to the distal or the lateral margin of the cell (figs. 101
to 104). These pictures are in full accord with those obtained
by means of the methods of Kopsch, Weigl, Sjévall, and Cajal,
with the sole difference that, in the Flemming preparation, the
cords are often canalized, which is to be regarded as due to the
dissolution of their substance, rather than to the real existence
of canaliculi.
It must be noticed in this connection that the granules or
droplets seen in Cajal and Flemming preparations are per-
fectly distinct from those in Marchi preparations, although
392 S. SAGUCHI
they bear striking resemblance in shape to each other. The

granules or clumps which are produced by the alteration of
zymogen granules appear invariably at the distal border of the
cell, close to the lumen, while those derived from the intracel-
lular network seem to be extruded into the lumen as well as into
the intercellular capillary.
2. Canaliculi. In preparations fixed in Miiller-formalin,
bichromate-formalin, Regaud’s fluid, ete., and stained with
iron- or alum-haematoxylin, the cytoplasm takes on a more or
less dark color, while certain nets or canaliculi remain totally
unstained. All these structures are not the same thing, but
they are grouped into two classes.
a. I have described in the previous section that there is a
relatively well-defined area between the nucleus and the distal
cell border, and that the zymogen granules are being elaborated
in this area: ‘secretogenous area.’ On examining this area care-
fully in the above preparations, especially in those cells which
contain a small number of zymogen granules, we find that there
are present canaliculi with clear lumen, which intertwine or
anastomose with one another so as to form a sort of network
from which similar canaliculi project into the surrounding
cytoplasm (figs. 75, 82, 83). The meshes of the network are,
in most cases, occupied by small granules, which are nothing
other than prozymogen granules or zymogen granules in the
earlier stages of development; in other words, the formation of
the zymogen granules is taking place in the meshes of the canalic-
ular network.
I am of opinion that the canalicular network is to be identified
with the intracellular network in the sense of Golgi; the former
merely being the negative of the latter, as already shown by
Bergen (’04), Bensley (’11), and others. The substance which
constitutes the intracellular network cannot be made manifest
by the fixing and staining methods above mentioned, while
the surrounding cytoplasm is stained to a more or less extent.
A possible transition between the canalicular and Golgi’s net-
work is brought to light by Flemming fixation and iron-haema-
toxylin staining, and is formed partly of solid, partly of canalized
cords. In conclusion, it must be noticed that the three states

of cords, that is to say, one, solid, and the two others canalic-
ular with or without the wall, are probably produced according
as they are perfectly preserved or are partly or entirely dissolved
out.
b. There can be seen another system of canaliculi in the
pancreas cell, which seems to be different both morphologically
and topographically from those above mentioned. ‘They are
tortuous, fine, clear canaliculi which ascend vertically along
the sides of the nucleus and lose themselves in the granular
zone or in the secretogenous area (fig. 83). These canaliculi
are, to my mind, to be regarded as the negative of the mitochon-
drial filaments, which, in the preparations we are dealing with
here, remain unstained, while the cytoplasm takes on a more or
less dark color.
3. Significance and genesis of the intracellular net apparatus.
It has been assumed that there are in the pancreas cell two kinds
of canalicular systems, one of which is located above the nucleus
and is to be identified with Golgi’s intracellular network. ‘The
question as to whether the intracellular network or canalicular
apparatus communicates with the exterior or not has been .
answered in the affirmative, in contradistinction with the view
of Negri (’00), Bergen (’04), and Bensley (11). We have also
been led to the conclusion that the intracellular network con-
sists of secreting material which is to be extruded either directly
into the lumen or indirectly into the intercellular capillary.
The connection which may exist between the network and the
intercellular capillary is expected from the other side of observa-
tion. As has already been noticed by Langerhans (’69), Laser-
stein (’94), and others, the intercellular capillary of the pancreas
is a canaliculus which passes between the cells downward and
terminates with the blind end before reaching the basal mem-
brane. This is made evident in an injection preparation, in
which we see that the injected intercellular canaliculus carries
on its surface spinous or club-shaped projections, although the
intracellular network cannot be injected from the duct (fig. 99).
But the existence of these projections would indicate that the
394 S. SAGUCHI
injection mass is passed into the spaces within the cell. The
latter, in part at least, correspond to the granules or droplets
lying at the lateral cell margin and connected with the inter-
cellular capillary, as seen in the Cajal and Flemming prepara-
tion (figs. 97, 98, 101 to 104). That the button-like secre-
tion vacuoles communicate with the intercellular canaliculi
was noticed by Kupffer, as early as 1873, in an injection prepara-
tion of the liver; Retzius (’92), using the Golgi method, found,
in salivary glands, drop-like appendages on the intercellular
canaliculus, which must be, according to him, the same thing
as Kupffer’s secretion vacuoles. Inthe pancreas, similar vacuoles
were also described by Miller (’98) as lying near the lumen or
the intercellular canaliculi. Furthermore, it must be taken
into consideration that the attachment of the projections or
droplets derived from the network to the lateral cell margin is
limited to the upper half or two-thirds of the height of the cell,
where the intercellular canaliculus will just be found. This.
also strongly suggests that the two are in close relation to each
other.
From the preceding observations and considerations, I have
been led to the conclusion that the substance composing the
intracellular network is formed in some way within the cell-
body and is to be eliminated from the cell. This view is
strikingly at variance with that of Holmgren (’02), who claims
that the intracellular network or his ‘trophospongium’ in the
pancreatic cell, just as in other kinds of cells, especially in the
nerve cell, is of exogenous origin. In spite of a more careful
study, I have never been able to find out the connection between
the cords of the network and the basket- or centroacinal cells,
as has been described and figured by this author.
Now arises the question as to the genesis of the intracellular
network. It is exceedingly difficult to determine whether the
network is perfectly impregnated or not. The examination of
those cells which contain no zymogen granules or only a small
number of them showed that the network is well developed in
some cells, while in others it is totally lacking; there being
intermediate conditions between these two extremes. In the
cells destitute of the network, there can be seen minute granules

or fine filaments scattered through the cytoplasm. In other
cells small clumps or rods are present above the nucleus (fig. 85).
Examination of still other cells shows that there are all grades
of transition between these clumps or rods and the typical net-
work (figs. 86,87), which is formed by an anastomosis or a
fusion of the former. Whether these gradations are to be
regarded as-developmental stages of the network or as a retro-
gression is a question which is difficult to answer. At any rate,
it is worthy of remark that the network is fixed in a definite
position above the nucleus which exactly corresponds to the
secretogenous area. We have also noticed that the prozymogen
granules occupy the meshes of the network; in other words, the
secretogenous area is traversed by the intracellular network.
It can further be seen in the preparations that the extent of the
secretogenous area, which varies according as the prozymogen
granules are increased or decreased, is in keeping with that of
the network. Under these circumstances, it seems probable
that the genesis of the zymogen granules and that of the net-
work must be in some way correlated with each other. How
this is accomplished, must for the present remain an open ques-
tion. It is conceivable, however, that the formation and growth
of prozymogen granules accompany an import and export of
substances, and that the substance given off deposits in the
secretogenous area and flows together, so as to form a network.
However this may be, it is quite certain that the secretion
process of the pancreatic cell is carried on in two ways: one
which is marked by the formation of granules and the other
in which the secreting material is deposited as unformed masses
from the beginning. In the former process the granules, from
the first formed prozymogen up to the mature zymogen granules,
preserve their individuality; they take up substances and per-
haps give them off, too. During growth, they gradually change
their properties and finally lose their individuality, fusing together
in order to form the secreting mass. The other process of secre-
tion, on the contrary, is characterized by the formation of secre-
tion material which is to be eliminated without undergoing
THE AMERICAN JOURNAL OF ANATOMY, VOL, 26, NO. 3

396 S. SAGUCHI
any remarkable change in character, and which must be, from

its behavior in the secretion process, regarded as a fluid mass.
Secreting matter of this kind, in all probability, forms the greater
part of the pancreas fluid and serves as a vehicle for dissolving
the more viscous mass derived from the zymogen granules.
Altmann’s theory, that ‘die Sekretion ist ein granulirer Proc-
ess,’ cannot be applied as such to this case; the secretion proc-
ess at least of the pancreas is both granular and fluid; the specific
ferments are perhaps furnished from the zymogen granules, al-
though the substance of the intracellular network cannot be
looked upon as consisting of waste matters only.
4, PHYSIOLOGICAL DEGENERATION
The secretory cells, as is the case with other kinds of cells,

undergo degeneration under some unknown physiological con-
ditions. The changes found in this process occurring in the
pancreas are in full accord with those which I have observed in
the epidermal glandular cells of amphibian larvae and described
in the previous paper (’15), to which the reader is referred for
details. The following is only a brief account of the process.
The first changes consist in the chromatic and nucleolar hyper-
chromasy, followed by chromatic separation. The latter proc-
ess is characterized by a flowing together of chromatic cor-
puscles, which leads to the formation of more or less large granules
or thick cords (figs. 110, 111). These further fuse together or
anastomose with one another so that at last a network or even
a capsule is formed at the periphery of the nucleus (fig. 112), a
state which is designated by pathologists as ‘hyperchromasy of
the nuclear wall.’ The area surrounded by the above capsule
is filled with nuclear sap and contains a somewhat enlarged
nucleolus. At successive periods, the nucleolar hyperchromasy
of the nuclear sap gradually diminishes. Meanwhile the cell-
body decreases in volume in consequence of repeated frag-
mentation. This process is either limited to the cytoplasm
or continues so that the division of the nucleus is followed by
that of the cell-body (fig. 113). The fragments thus formed are
usually spherical in shape, but vary much in size, and especially

in structure. Some are without nuclear fragments but full
of zymogen granules; others contain one or more small chromatic
corpuscles; in still other cases the nuclear portion, as compared
with the plasma portion, is so large that the fragment is almost
entirely filled up with it, an appearance of a free nuclear frag-
ment being thus produced. These fragmented corpuscles are,
sooner or later, taken up by neighboring normal glandular ce'ls.
This occurs either simultaneously with the process of fragmenta-
tion in such a way that the fragment presses against the neighbor-
ing cell so as to lead to the formation of a cavity which, being
more and more deepened, finally encloses the fragment (fig. 113),
or the fragments formed persist for some time in the intercellular
space, causing depressions upon the adjacent cells. The ultimate
fate of these corpuscles is the same as in the former case;
that is to say, they are all embraced by the surrounding plasma
of the normal glandular cells and finally taken up by them
(figs. 114, 118). There is no constancy in the position in the
cell nor in the size of fragments thus taken up; if they are large,
the nucleus is often distorted by their pressure. The frag-
ments gradually diminish in volume, partly by repeated frag-
mentation, accompanied by simultaneous changes in structure.
The nuclear fragments come to stain more and more faintly with
chromatin dyes (figs. 115 to 117), or their substance diffuses
through the plasma portion, while the zymogen granules seems
to fuse together. Thus, corpuscles of various sizes and struc-
tures are formed, which vary not only: in individual cases, but
also according to the stages of degeneration, so that no definite
light can be thrown from merely morphological study upon
the significance of the corpuscles. They seem to be digested
and absorbed by the surrounding cytoplasm, finally leaving
there clear vacuoles, which gradually decrease in volume and
disappear.
The mitochondrial filaments in degenerating cells appear at
first to be increased in number. In successive periods they
thicken and gradually become faintly tinged (fig. 112), and at
last disintegrate into granules, while the intracellular network is
398 S. SAGUCHI ‘
broken down into irregular clumps or granules and disappears

(figs. 119, 120). |
From these observations and those which I have made in the
study of the epidermal glandular cells of Amphibia, the con-
clusion is warranted that the physiological degeneration of
these cells is characterized by nuclear and nucleolar hyper-
chromasy and by chromatic separation, succeeded by the con-
striction of the nucleus and the cell-body; and, secondly, that
the formed fragments are invariably taken up by the neighbor-
ing normal glandular cells, but are never eliminated into the
lumen nor, which appears very strange, absorbed by the
phagocytes.
That the pancreatic cell undergoes physiological degenera-
tion, has already been noticed by Gaule (’80) and Nussbaum
(’82); Platner (’89), however, was the first to describe the nuclear
changes brought about by that process. The author has found
pictures corresponding to one which is designated by Flemming
as ‘chromatolysis’ in the pancreas of various species. In con-
sequence of the chromatolysis, there are produced smaller or
larger spherical corpuscles, which at first stain with safranin a
red color, but later refuse to take the stam. The structures are
delineated in his figures 11 to 13. Of these, the large corpuscle
in the cell a, figure 12, corresponds to a nucleus which has already
undergone chromatic separation, while those seen in the cells,
figure 12, b, figures 11 and 13 cannot be interpreted as anything
else than fragments constricted off. In addition to these,
Platner found such corpuscles in the normal cell with an intact
nucleus, and assumed that they are produced by the partial
chromatolysis of the nucleus. In my opinion, these corpuscles
are in all probability fragments taken up by the normal cell, as
is clear from a comparison of his figures 11 to 13 and those of
my own (figs. 114, 118). Furthermore, he does not express
himself regarding the process of his partial chromatolysis.
Melissinos and Nicolaides (’90), Macallum (91), and Garnier
(00) also describe the occurrence of karyolysis in the pancreas.
According to Macallum, the product of karyolysis and cytolysis
consists partly of protoplasm, partly of eosinophile substance;
the chromatic substance being limited to crescent-shaped or an-

nular corpuscles, which are situated upon or in the protoplasmic
substratum of the body. Those structures, which are described
under the name of ‘nebenkern’ by various investigators, Ogata,
Eberth and Miiller, and others, belong, in part at least, to this
eategory. I will return to this question further on.
5. MITOSIS
Since the pancreatic cells undergo physiological degeneration,

it is a matter of course that they must be supplemented in some
way; this is effected by mitosis, which varies more or less accord-
ing to individual cases. The nucleus preparing to divide by
mitosis migrates upward and comes to lie about midway between
the base and the distal end of the cell (fig. 106); the cytop!asm
becomes stained more or less pale, the position and arrange-
ment of zymogen granules and mitochondrial filaments differ-
ing in no way from those of resting cells (figs. 106, 107). The
plane of cleavage being vertical (fig. 108), the daughter cells
acquire both zymogen granules and chondriocontes (fig. 109).
I have never been able to observe that the latter are increased
in number by division or that an equal amount of them passes
into each of the daughter cells, as assumed by some. Sometimes
there are seen glandular cells with two nuclei; whether they are
produced by mitosis or amitosis I was unable definitely to
determine.
The mitotic figures were observed, in the pancreas, by Gaule
(80), Nussbaum (’82), and by many of the subsequent investi-
gators (Bizzozero and Vassale, ’87, Platner, ’89, Steinhaus, ’90).
Podwyssozki (’87) believes to have found mitosis in young
animals, while Eberth and Miller (92) deny its occurrence in
any case. The most curious view concerning the regeneration
of the pancreatic cell is advanced by Ogata (’83). This author
believes that the plasmosome of the nucleus passes out into the
cytoplasm and there forms the so-called ‘nebenkern,’ which is
transformed through gradual increase into a new cell, while
the old one undergoes degeneration. The result of this process
400 S. SAGUCHI
is not, therefore, the production of two cells from one, but the
formation of a new cell from an old one, which process he has
designated ‘renewal of the cells.’ Such a mode of cell genesis
can no longer be assumed in the present state of our knowledge
regarding cell multiplication, just as it is little probable that
the ‘nebenkerne,’ are passed out nucleoli. A comparison of his
figures with those of my own shows that his ‘nebenkerne’ are
nothing other than the products of degeneration and that what
he mentions as stages of renewal of the cell is but a reversal of
the process of degeneration.
6. THE SO-CALLED ‘NEBENKERNE’
Historical
a. Shape. Nussbaum (’81,’82) and Gaule (’81), nearly

about the same time, but independently of each other, have
found certain structures in the pancreatic and other kinds of
secretory cells, and have given the name ‘nebenkerne’ to them.
According to the former author, the structures are either solitary
or multiple, solid oval or spirally coiled, even curly twisted.
A few days after feeding they are found in every cell, while
they are rarely met with in the fasting condition. Their biolog-
ical significance remained unsolved, although the author brought
them into the same category as the ‘yolk-nucleus’ of the egg-
cell found by Wittich, as the ‘nebenkern’ of the spermatocyte
discovered by La Valette St. George, and finally as the struc-
tures noticed by Leydig in the epidermal cells of Pelobates
larvae. The same or similar structures have since often attracted
attention of those investigators who have made a study of the
pancreas of the higher and lower vertebrates. And a review of
the literature shows that what is described and figured under
the name of ‘nebenkern,’ ‘paranucleus,’ etc., is not necessarily
concerned with one and the same structure, but can be classed
into at least two types, as already done by Eberth and Miller
(792).
The first type includes those corpuscles which are irregular
in shape, and have generally a fibrillar structure. They are
either filamentous or spindle-shaped, crescent- or comma-formed

or even spherical. In some cases the structures have been
described as being homogeneous. This type is described and
figured by Nussbaum (’82), Steinhaus (90), Macallum (91),
Eberth and Miiller (92), Ver Eecke (95), Mouret (795, ’05),
Carlier (96), Mathews (99), Champy (11), and others.
To the second type belong those corpuscles which are spherical
in shape and of various sizes. They are either homogeneous
or granular, or contain spherical or crescent-shaped bodies.
This type is found by Nussbaum (’82), Ogata (83), Platner
(’89), Melissinos and Nicolaides (90), Macallum (91), Eberth
and Miiller (92), Ver Eecke (’95), Galeotti (95), Mouret (795,
05), Mathews (’99), Garnier (’00), Babkin, Rubaschkin and
Ssawitsch (’09), and others.
b. Origin. Various views advanced concerning the origin of
the ‘nebenkern’ can be summarized as follows:
Nuclear origin. Ogata, Galeotti, and Platner believe that the
‘nebenkern’ is derived from the plasmosome passed out of the
nucleus; Macallum, Ver Eecke, Melissinos and Nicolaides, and
Garnier maintain a nuclear. origin for a part of the structures,
while another part is regarded as made up of products of
chromatolysis.
Plasmic origin. Mouret identifies the ‘nebenkern’ with the
matrix of the cell or his ‘prézymogenes;’ Carlier also describes
the accessory nucleus as indistinguishable from the neighboring
dense granular spongioplasm. Eberth and Miiller, Mathews,
Melissinos and Nicolaides, and Garnier ascribe to a part of the
‘nebenkerne’ a plasmic origin; according to Eberth and Miller,
they are produced by the adhesion of protoplasmic filaments;
Mathews likewise holds that they are nothing other than coiled
or contorted cell-threads. However, the question, how they
are formed, remains unanswered. In addition, Champy identi-
fies the ‘nebenkern’ with the sphere of the egg-cell, and
claims to have found central corpuscles in its centre. Babkin,
Rubaschkin and Ssawitsch, finally, advance the view that the
‘nebenkern’ is derived from zymogen granules. In their experi-
mental study, they believe to have found that a plasmic portion
402 S. SAGUCHI
along with some of the zymogen granules becomes sharply

limited from the surrounding plasma, and forms a mass which
comes to lie in a vacuole. This corpuscle gradually changes
its staining reactions and finally is cast off into the lumen.
Exogenous structures. The accessory nuclei, in part at least,
have been regarded either as parasites (Stemhaus, Mathews,
Macallum) or as leucocytes passed into the cell (Melissinos and
Nicolaides).
c. Significance. There is also much diversity of opinion with
regard to the significance and the final fate of the ‘nebenkern.’
Ogata, Platner, Laguesse (99), Ver Eecke, Galeotti, Mouret,
Mathews, Garnier, Babkin, Rubaschkin and Ssawitsch, and many
others believe that the ‘nebenkern’ disintegrates into zymogen
granules, while Eberth and Miiller believe in its indirect par-
ticipation in the elaboration of secretion granules. On the other
hand, Steinhaus and Macallum deny such a participation,
whether direct or indirect.
A few investigators advance the very curious view that the
‘nebenkern’ is a plasmosome (nucleolus) passed out of the nucleus
which becomes converted into a new nucleus (Ver Hecke), even
into a new cell (Ogata).
Nussbaum, and Melissinos and Nicolaides reached no definite
conclusion as regards the significance of the ‘nebenkern.’
Critical
In looking over the literature, one is confronted with an ap-

palling mass of conflicting observations. The corpuscles which
are described under the name of ‘nebenkern’ show much varia-
tion not only as regards shape and structure, but also in ori-
gin and significance. These variations would indicate that the
so-called ‘nebenkern’ either includes things which vary in origin,
or, if of the same origin, must be a body which exhibits an extra-
ordinary diversity of shape and structure. Under these circum-
stances, the inference would appear justifiable that the ‘neben-
kerne’ cannot be integral constituents of the pancreatic cell,
but must be considered products either of artificial or of physio-
logical or pathological changes. I am of the opinion that they

must be derived from two sources totally differing in character.
a. We have seen in the section on ‘structure of the cytoplasm’
that the protofibrillae which may be regarded as preéxistent in
the cytoplasm, adhere to one another in consequence either of
the action of fixatives or of the subsequent manipulation, so
that there are formed various structures according to the degrees
of adhesion, such as thin or thick bundles, crescent- or shell-
formed, spirally coiled, or even irregular spherical corpuscles.
They are characterized, first, by fibrillar structure. It often hap-
pens, however, that they appear homogeneous, when relatively
well preserved or examined with a lower magnifying power.
Secondly, they vary greatly in shape. Finally, the contour of
these corpuscles is very irregular, as if it had been gnawed, or
fibers arise from them and continuously pass into the cytoplasmic
filaments. ;
It will be seen when my figures are compared with those of
various investigators above mentioned, that these artificial prod-
ucts agree in character with those ‘nebenkerne’ which are de-
scribed as being of irregular shape or as exhibiting fibrillar struc-
ture. Of the views advanced concerning the origin and signifi-
cance, that of Eberth and Miller and of Mathews is nearly in
agreement with that of my own. The former workers say: “‘sie
sind umgewandelte Protoplasmafiden, welche, indem sie mit
ihren Nachbarn verschmelzen, zu spindelf6rmigen, sichelfér-
migen, kommadhnlichen Kérpern werden, die vielleicht voriiber-
gehend, vielleicht dauernd ihre fibrillére Zusammensetzung noch
mehr oder weniger bewahren oder dieselbe ganz verlieren und
dann gliinzende homogene Kérper darstellen.”’ Mathews also is of
opinion that the structures in question are derived from the cell-
threads; he says:‘‘ The threads are not, however, perfectly straight,
but in all cases are more or less twisted and bent gts
and often contorted or even spirally coiled on themselves. These
spiral twists, or coils, form the so-called ‘Nebenkern’ of the pan-
creas.’ How these corpuscles are formed, that is to say, the
question whether they are produced under physiological condi-
tions or by the action of fixatives, remains unanswered.
404 S. SAGUCHI
As already mentioned, these fibrillar structures when pre-

served in alcohol, sublimate mixtures, ete., stain with alum-hae-
matoxylin a blue color, as is the case with the chromatin. From
this staining reaction alone, however, it cannot be inferred that
they are derived from the nucleus, as assumed by some; for, if
the stains, such as safranin, iron-haematoxylin, ete., are em-
ployed, they are easily decolorized, whereas the chromatin is
heavily stained.
b. I have before mentioned that the physiological degenera-
tion of the cell is characterized by chromatic separation followed
by constriction of the nucleus and the cell-body, and that the
spherical fragments produced by the latter process and taken up
by the neighboring glandular cells exhibit every variety of struc-
ture. The nuclear portion in the fragment, when it exists, is
variable, and is either spherical or crescentic, or ring-formed
according to its size and the phase of the nuclear change; while
the cytoplasmic portion, small or large in amount, either contains
unaltered zymogen granules or may be entirely devoid of them.
There are also many corpuscles which contain no nuclear
fragments.
These corpuscles are to be identified with those ‘nebenkerne’
which I have before summarized under the second type. The
‘nebenkerne’ which belong to the second type are spherical in
shape and have an even contour; they are surrounded with a
clear halo; in other words, they lie in vacuoles of the cytoplasm.
This fact shows that these corpuscles are not closely related to
the cytoplasm containing them; it is therefore not without reason
that they have been supposed by some to be parasites passed
into the cell. That these ‘nebenkerne’ cannot be interpreted as
anything else than corpuscles produced by the fragmentation of
the degenerated cell will be evidenced by considering their struc-
ture. In the following I will try to criticise the observations of
various investigators, comparing the various types of ‘nebenkerne’
with our fragments.
First, in figures 4, a, d, e, and 5, a, accompanying Ogata’s
paper, it can clearly be seen that these corpuscles are nothing
other than nuclei, which have already suffered chromatic
separation, so that the condensed chromatin is accumulated in

the periphery while the nuclear sap and the nucleolus are en-
closed by this chromatic capsule. On the other hand, Platner’s
figure 12 and Ogata’s figure 7, a, correspond to the glandular
cells containing these changed nuclei. Ogata’s figure 7, a, es-
pecially shows an elongated cell, which perhaps will be frag-
mented in the next stage.
Secondly, Ogata’s figure 5, b, c, d, Platner’s figures 11 and 13,
Melissinos and Nicolaides’ figures 9 to 15, Eberth and Miiller’s
figure 16, Ver Eecke’s figures 17 to 24, and Babkin, Rubaschkin
and Ssawitsch’s figure 10 (the cell to the left) show cells which
contain spherical bodies with either ring- or crescent-shaped,
larger or smaller spherical chromatic corpuscles; these ‘bodies
are to be regarded as constricted off fragments of degenerated
cells, which have been taken up by the normal glandular cells;
the chromatic corpuscles correspond to nuclear fragments. On
the other hand, there can be seen in Ogata’s figure 6, 6, and in
Babkin, Rubaschkin and Ssawitsch’s figure 10 (the cell to the
right), seventeen bodies which contain no chromatic corpuscles,
and which are, in all probability, cytoplasmic fragments with or
without zymogen granules.
Finally, Ogata’s figure 3, a, b, Galeotti’s figures 28 and 29,
Garnier’s figures 34 and 35, etc., show that spherical, ring or
crescent-shaped chromatic corpuscles are contained in the nor-
mal glandular cells; they are regarded by these authors as
nucleoli passed out of the nucleus. In my opinion, they are
nuclear fragments with a very small amount of the cytoplasmic
portion, which have been reduced in size by repeated fragmenta-
tion. These chromatic corpuscles gradually grow pale and then
stain only with plasma dyes. Under these circumstances, it is a
difficult matter to decide whether the spherical corpuscles stained
.by plasma dyes are derived from the fragments with or without
nuclear portions.
It is evident from the above comparison that the spherical
corpuscles which have been regarded as belonging to the so-called
‘nebenkern’ are the same as the fragments produced by the
fragmentation of the degenerated glandular cell. That the sig-
406 S. SAGUCHI
nificance of these corpuscles remains as yet undecided, in spite

of the fact that they have attracted much attention of those
investigators who were engaged in the study of the pancreas, is
due essentially to the fact that their genesis was not fully followed
out. In fact, there are some who regard the ‘nebenkerne,’ which
are spherical in shape and marked off sharply from the surround-
ing cytoplasm in which they are embedded, as products of the
chromatolysis (Melissinos and Nicolaides, Macallum, and Gar-
nier). Nevertheless, these investigators have failed to throw
light upon the process of their formation. The fact that these
corpuscles are found beside the nucleus within the normal cell
tends to contradict the view that they are products of the chro-
matolysis. This is perhaps the reason why Platner has taken
them to be produced by the partial degeneration of the nucleus,
and has named the process ‘partial chromatolysis.’
Kanazawa, Japan, June 15, 1918.
BIBLIOGRAPHY
ALEXEIEFF, A. 1917 Mitochondries et réle morphologene du noyau. Compt.

rend. Soe. biol., T. 80.
ALTMANN, R. 1894 Die Elementarorganismen und ihre Bezeihung zu den
Zellen. 2. Auflage, Leipzig.
Arnotp, G. 1912 The réle of chondriosome in the cells of the guinea-pig’s pan-
creas. Arch. f. Zellforschung, Bd. 8.
BaBkKIN, B. P., RupascHKin, W. J., unp Ssawitsca, W. W. 1909 Uber die
morphologischen Verainderungen der Pankreaszellen unter der Ein-
wirkung verschiedenartiger Reize. Arch. f. mikr. Anat., Bd. 74.
BENSLEY, R. R. 1911 Studies on the pancreas of the guinea-pig. Am. Jour.
‘ Anat., vol. 12.
Bercen, F. v. 1904 Zur Kenntnis gewisser Strukturbilder (Netzapparate,
Saftkanilchen, Trophospongium) im Protoplasma verschiedener Zel-
lenarten. Arch. f. mikr. Anat., Bd. 64.
BizzozERo, G., ET VASSALE, G. 1887 Sur le tissu des glandes excrétantes.
Arch. ital. de biol., T. 9.
Bopeav, G. 1911 Mitochondries et lipoides dans les glandules parathyroides
du cheval. Compt. rend. Assoc. Anat., 13 Réun., Paris.
Bourn, P. 1905 Ergastoplasme et mitochondria dans les cellules glandulaires
sereuses. Compt. rend. Soc. biol., T. 59.
Casa, 8S. Ramén 1909-11 Histologie du systéme nerveux de l’Homme et des
vertébrés. Paris.
Car iER, E. W. 1896 On the pancreas of the hedgehog during hibernation.
Journ. of Anat., vol. 30, 1899 Changes that occur in some cells of
the newt’s stomach during digestion. A cell study. La cellule, T.
16 (quoted from Maziarski, ’10).
1907 De certains changements qui peuvent étre observé dans les cel-
lules du foie pendant la digestion et de leurs relations avec la sécrétion
hépatique. Compt. rend. Assoc. Anat., IX. Réun., Lille (quoted
from Maziarski, ’10).
CuHampy, C. 1911 Recherches sur |’absorption intestinale et le rdle des mito-
chondries dans l’absorption et la sécrétion. Arch. d’anat. micr., T.
12.
Craccro, C. 1913 Zur Physiopathologie der Zelle. I. Entartungsbilder der
Plastosomen. Centralb. f. allgem. Pathol. u. path. Anat., Bd. 24.
Comes, 8. 1913 Apparato reticolare o condrioma? Condriocinesi o dittocinesi?
Anat. Anz., Bd. 48.
Cowpry, E. V. 1916 The general functional significance of mitochondria. Am.
Jour. Anat., vol. 19.
DuesBeErG, J. 1908 Sur l’existence de mitochondries dans l’oeuf et l’embryon
d’Apis mellifica. Anat. Anz., Bd. 32.
1911 Les chondriosomes des cellules embryonnaires du poulet, et leur
role dans la genése des myofibrilles, avec quelques observations sur le
développement des fibres musculaires striés. Arch. f. Zellforsch., Bd.
4,
408 S. SAGUCHI
Esertu, C. J., unp Mituer, K. 1892 Untersuchungen iiber das Pankreas.

Zeitsch. f. wiss. Zool., Bd. 53, Supplement.
FLemMMING, W. 1882 Zellsubstanz, Kern- und Zelltheilung. Leipzig.
Gaxurorti, G. 1895 Uber die Granulationen in den Zellen. Internat. Monatsch.
f. Anat. u. Physiol., Bd. 12.
Garnier, C. 1900 Contribution 4 l’étude de la structure et du fonctionnement
des cellules grandulaires séreuses. Du réle de l’ergastoplasme dans la
séerétion. Journ. de l’Anat. et de la Physiol., 36 année.
GaAuLE, J. 1880 Kerntheilungen im Pankreas des Hundes. Arch. f. Anat. u.
Physiol., Physiol. Abth.
1881 Die Beziehungen der Cytozoen (Wiirmchen) zu den Zellkernen.
Arch. f. Anat. u. Physiol., Physiol. Abth.
1881 Kerne, Nebenkerne und Cytozoen, Centralb. f. med. Wiss., Nr.
31.
GoutpscumMipT, R. 1904 Der Chromidialapparat lebhaft funktionierender
Gewebszellen. Zool. Jahrb., Bd. 21.
Harcker, V. 1893 Das Keimblischen, seine Elemente und Lageverinderung.
Arch. f. mikr. Anat., Bd. 41 and 42.
1895 Die Vorstadien der Eireifung. Zusammenfassende Untersuch-
ungen iiber die Bildung der Vierergruppen und das Verhalten der Keim-
blaschen-Nucleolen. Arch. f. mikr. Anat., Bd. 45.
HerpENHAIN, R. 1868 Beitrige zur Lehre von der Speichelabsonderung, Stu-
dien d. physiol. Instit., Breslau (quoted from Maziarski, 10).
1875 Beitrige zur Kenntniss des Pankreas. Arch. f. ges. Physiol.,
Bd. 10.
1880 Physiologie der Absonderungsvorginge. Handbuch der Physi-
ologie von L. Hermann, Bd. 5.
Heipennain, M. 1890 Beitriige zur Kenntniss der Topographie und Histologie
der Cloake und ihrer driisigen Adnexa bei den einheimischen Tritonen.
Arch. f. mikr. Anat., Bd. 35.
1907 Plasmaund Zelle. Jena.
Hermann, F. 1888 Uber regressive Metamorphosen des Zellkernes. Anat.
Anz., Bd. 3.
Hoimaren, E. 1902 Weiteres tiber das ‘Trophospongium’ der Nervenzellen und
der Driisenzellen des Salamander-Pankreas. Arch. f. mikr. Anat.,
Bd. 60.
Hoven, H. 1910 and 1912 Contribution 4 l’étude du fonctionnement des cel-
lules glandulaires. Du réle du chondriome dans la secrétion. Anat.
Anz., Bd. 37 (Communication préliminaire) and Arch. f. Zellforsch.,
Bd. 8.
Jorpan, H. E. 1912 Notes on the Spermatogenesis of the Bat. Anat. Anz.,
Bd. 40.
Koster, R. 1913 Uber die durch Golgi’s Arsenik- und Cajal’s Urannitrat-
silbermethode dargestellten Zellstructuren. Verh. d. anat. Gesch.
Korscueut, E. 1889 Beitriige zur Morphologie und Physiologie des Zellkerns.
Zool. Jahrb. Abt. f. Anat., Bd. 4. .
Krausg, R. 1895 Zur Histologie der Speicheldriisen. Die Speicheldriisen des
Igels. Arch. f. mikr. Anat., Bd. 45.
Lacue, J. G. 1906 L’aspect du noyau de la cellule nerveuse dans la méthode

a largent réduit. Anat. Anz., Bd. 28.
Lacursse, E. 1899 Corpuscles paranucléaires (parasomes), filaments basauz,
et zymogéne dans les cellules sécrétantes (pancréas, sous-maxillaire).
Compt. rend. Soe. biol., volume jubilaire.
1900 Sur les variation de la graisse dans les cellules sécrétantes
séreuses (pancréas). Compt. rend. Soc. biol., T. 52.
Lams, H. 1906 Le corps vitellin de Balbiani et la masse vitellogéne dans
Voocyte de Rana temporaria. Verh. d. anat. Gesch.
LANGERHANS, P. 1869 Beitrige zur mikroskopischen Anatomie der Bauch-
speicheldriisen. Inaug.-Diss. Berlin (quoted from Laserstein).
LANGLEY, J. N. 1884 On the structure of secretory cells and on the changes
which take place in them during secretion. Internat. Monatschr. f.
Anat. u. Physiol., Bd. 1.
LasERSTEIN, 8S. Uber die Anfiinge der Absonderungswege in den Speicheldriisen
und im Pankreas. Arch. f. ges. Physiol., Bd. 55.
Levi, G. 1912 I condriosomi nelle cellule secernenti. Anat. Anz., Bd. 42.
Macautum, A. B. 1891 Contributions to the morphology and physiology of
the cell. Transact. of the Canadian Institute, vol. 1 (quoted from
Mathews and Eberth and Miller.)
Matuews, A. 1899 The changes in structure of the pancreas cells. A con-
sideration of some aspects of cell metabolism. Jour. Morph., vol. 15,
supplement.
Maximow, A. 1916 Sur la structure des chondriosomes. Compt. rend. Soc.
biol., T. 79.
Mayer, A., RatHery, F., ET SCHAEFFER, G. 1914 Sur les variations experi-
mentales du chondriome hépatique. Parallélisme entre la composi-
tion chimique du tissu et ses aspects cytologiques. Compt. rend. Soc.
biol., T. 76.
Maziarski, 8. 1910 Sur les changements morphologiques de la structure
nucléaire dans les cellules glandulaires. Contribution 4 l’étude du
noyau cellulaire. Arch. f. Zellforsch., Bd. 4.
Metrowsky, E. 1908 Uber den Ursprung des melanotischen Pigments der
Haut und des Auges. Leipzig.
MELIsSINos, C., UND NicouaipEs, R. 1890 Untersuchungen iiber einige intra-
und extranucleare Gebilde im Pankreas der Saiugetiere und ihre Bezie-
hung zu der Sekretion. Arch. f. Anat. und Physiol., Physiol. Abth.
Meves, F. 1907 Uber Mitochondrien bezw. Chondriokonten in den Zellen
junger Embryonen. Anat. Anz., Bd. 31.
1908 Die Chondriosomen als Trager erblicher Anlagen. Cyto-
logische Studien am Hiihnerembryo. Arch. f. mikr. Anat., Bd. 72.
1910 Uber Aussat minnlicher Mitochondrien im Ei bei der Befruch-
tung. Anat. Anz., Bd. 36. :
Micuae.is, L. 1900 Die vitale Farbung, eine Darstellungsmethode der Zell-
granula. Arch. f. mikr. Anat., Bd. 55.
Misuawsxky, A. N. 1911 Beitrige zur Morphologie der Driisenzelle. Uber
das Chondriom der Pankreaszelle einiger Nager. Vorl. Mitt., Anat.
Anz., Bd. 39.
410 S. SAGUCHI
1913 Uber das Chondriom der Pankreaszellen. Arch. f. mikr. Anat.

Bd., 81.
Montaomery, T. H. 1899 Comparative cytological studies with especial
regard to the morphology of the nucleolus. Jour. Morph., vol. 15.
Movret, J. 1895 Des modifications subis par la cellule pancréatique pendant
la sécretion. Compt. rend. Soc. biol., Année 46, 47.
1905 Contribution 4 l’étude des cellules glandulaires (pancréas).
Journ. de |’Anat. et de la Physiol., Année 31.
Mituer, E. 1896 Driisenstudien. Arch. f. Anat. u. Physiol.
1898 Driisenstudien II. Zeitschr. f. wiss. Zool., Bd. 64.
Necri, A. 1900 Uber die feinere Struktur der Zellen mancher Driisen bei den
Sdugetieren. Verh. d. Anat. Ges.
Nusspaum, M. 1881 Einige Beobachtungen, den Nebenkern der Zelle anlan-
gend. Verh. d. Naturhist. Vereins d. preuss. Rhein. u. West f., Jahrg.
33 (quoted from Oppel).
1882 Uber den Bau und die Thitigkeit der Driisen. Arch. f. mikr.
Anat. Bd., 21.
Oaata, M. 1883 Die Verinderung der Pankreaszellen bei der Sekretion. Arch.
f. Anat. u. Physiol., Physiol. Abth.
Oppet, A. 1900 Lehrbuch der vergleichenden mikroskopischen Anatomie
der Wirbeltiere. Jena.
Priiicer, E. 1869 Die Endigung der Absonderungsnerven in den Pankreas.
PuLaTNER, G. 1889 Beitraige zur Kenntniss der Zelle und tiber ihrer Teilung.
Popwyssozk1, W. 1887 Les lois de la régénération des cellules glandulaires 4
état normal et pathologique. Bulletins de la Soc. Anat., de Paris.
Année 70.
Pouicarp, A. 1912 Attitudes fonctionelles du chondriome de la cellule hépa-
tique. Rapports des chondrisomes et du noyau. Compt. rend. Soc.
biol: 272:
PRENANT, A. 1910 Les mitochondries et l’ergastoplasme. Journ. de |’Anat.
et de la Physiol., Année 46.
Reeaup, C. 1909 Attribution aux ‘formations mitochondriales’ de la fonction
générale d’ ‘extraction et de fixation électives’ exercée par les cellules
vivantes sur les substances dissoutes dans le milieu ambiant. Compt.
rend. Soe. biol., T. 66.
Reaaup, C., pT Mawas, J. 1909 Ergastoplasme et mitochondries dans _ les
cellules de la glande sous-maxillaire de homme. Compt. rend. Soc.
biol., T. 66.
Sur la structure du protoplasme (ergastoplasme, mitochondries, grains
de ségrégation) dans les cellules sérezymogénes des acini et dans les
cellules des canaux excréteurs de quelques glandes salivaires de Mam-
miféres. Compt. rend. Assoc. Anat. (quoted from Hoven, ’10).
Rerzius, G. 1892 Uber die Anfinge der Driisengiinge und die Nervenendigun-
gen in den Speicheldriisen des Mundes. Biol. Untersuch., Neue
Folge. III.
Ruvumster, L. 1893 Uber Entstehung und Bedeutung der in den Kernen

vieler Protozoen und in Keimblaschen von Metazoen vorkommenden
Binnenkérper (Nukleolen). Eine Theorie zur Erklirung der ver-
schiedenartigen Gestalt dieser Gebilde. Zeitsch. f. wiss. Zool., Bd. 56.
Russo, A. 1908 Sull ’origine e sulla funzione dell’apparato mitochondriale
nelle cellule sessuali dei mammiferi. Boll. Accad. Gioenia, Se. nat.
Catania., Ser. 2, Fasc. 2 (quoted from Luna, 713).
1910 Ancora sui Mitoconri dell’oocite di Coniglia, sulloro aumento
e sulla loro funzione. Anat. Anz., Bd. 37.
Saacucut, Sakaye 1915 Uber Sekretionserscheinungen an den Epidermis-
zellen yon Amphibienlarven nebst Beitrigen Zur Frage nach der
physiologischen Degeneration der Zellen. Mittteil. der mediz. Fakul-
tit der kais. Universitit zu Tokyo, Bd. 14.
Scuutrzr, O. 1911 Uber die Genese der Granula in den Driisenzellen. Anat.
Anz., Bd. 38.
Souter, B. 1894 Zur Kenntniss der secernierenden Zellen der Glandula sub-
maxillaris des Menschen. Anat.-Anz., Bd. 9.
1896 Uber den feineren Bauder Glandula submaxillaris des Menschen
mit besonderer Beriicksichtigung der Driisengranula. Festschr. f. C.
Gegenbaur., Bd. 2.
Srernuaus, J. 1890 Uber parasitiire Einschliisse in den Pankreaszellen der
Amphibien. Ziegler’s Beitrige, Bd. 7.
Vespowsky, F. 1907 Neue Untersuchungen iiber die Reifung und Befruchtung.
Verh. k. bdhm. Ges. Wiss. Prag (quoted from Luna, 713).
Ver Eecxe, A. 1895 Modifications de la cellule pancreatique pendant |’ activ-
ité sécrétoire. Arch. de Biol., T. 13.
WassitieFF, A. 1907 Die Spermatogenese von Blatta germanica. Arch. f.
mikr. Anat., Bd. 70.
ZIMMERMANN, K. W. 1898 Beitriige zur Kenntniss einiger Driisen und Epithe-
lien. Arch. f. mikr. Anat., Bd. 52.

EXPLANATION OF PLATES
All figures are drawn with Zeiss camera under Zeiss ;’s oil-immersion objective
and Zeiss compensating ocular 18, at the distance of 250 mm.; tube length, 160
mm. By reproduction, figures 36 to 120 were reduced to four-fifths of the original.
PLATE 1
1 to 35 Nuclei of the pancreatic cells of Rana temporaria.

1 to6 Sublimate; iron-haematoxylin.
7 Alcohol; iron-haematoxylin.
8and9 Formalin; figure 8, Altmann stain; figure 9, 1ron-haematoxylin.
10 to 13 Sublimate-acetic acid; figures 10, 11, Benda stain; figure 12, iron-
haematoxylin; figure 13, alum-haematoxylin.
14 to 16 Sublimate-formalin; figure 14, Altmann stain; figures 15 and 16,
iron-haematoxylin.
17 to 20 Champy’s method for mitochondria; figures 17 and 18, Benda stain;
figure 19, iron-haematoxylin; figure 20, safranin.
21 Potassium bichromate-formalin; Altmann stain.
22 and 23. Regaud’s bichromate-formalin-acetic mixture; figure 22, iron-
haematoxylin; figure 23, alum-haematoxylin.
24 Osmium-sublimate; Altmann stain.
25 Fixed and stained after Benda.
26 and 27. Meves’ fluid; figure 26, Benda stain; figure 28, iron-haematoxylin.
28 Zenker’s fluid; alum-haematoxylin.
29 and 30 Trichloracetic acid; figure 29, Altmann stain; figure 30, iron-
haematoxylin.
31 to 35 Cajal’s photographic method.
GLANDULAR CELLS OF THE FROG’S PANCREAS PLATE 1
8. SAGUCHI
415
PLATE 2
36 Carnoy’s mixture; iron-haematoxylin.

37 Sublimate-acetic mixture, alum-haematoxylin.
38 to 40 Zenker’s fluid; iron-haematoxylin.
41 and 42 Meves’ fluid; iron-haematoxylin.
43 Formalin; iron-haematoxylin.
44 to 47 Meves’ fluid; iron-haematoxylin.
48 to 52. Formalin; iron-haematoxylin.
53 Sublimate-formalin; iron-haematoxylin.
54 Zenker’s fluid; iron-haematoxylin.
55 Champy’s mixture; Benda stain.
414
8. SAGUCHI
36
Ss9
"= e
@>
s*
@
Es.
ons
4
os Sy gga
¢¥
PLATE 3
56 to 61 Meves’ fluid; iron-haematoxylin.
62 to 64. Two per cent osmic acid; iron-haematoxylin.
65 Fauré-Fremiet’s method for mitochondria.
66 and 67 Mislawsky’s fixing method for mitochondria; Benda stain.
68 Meves-formalin without acetic acid; iron-haematoxylin.
69 and 70 Golgi’s chromium-silver method.
71 to 76 Miiller-formalin, and succeeding treatment with Miiller’s mixture;
iron-haematoxylin.
77 and 78 Helly’s mixture; iron-haematoxylin.
79 Champy’s method; Benda stain.
416
8. SAGUCHI
417
PLATE 4
80 and 81 Zenker’s fluid; iron-haematoxylin. The intracellular network

is scarcely visible.
82 Kopsch’s fixing method; iron-haematoxylin.
83 Miiller-formalin and succeeding treatment with Miiller’s fluid; iron-
haematoxylin.
84 Marchi’s fixing method; iron-haematoxylin.
85 to 88 Kopsch’s osmium method for the Golgi’s network.
89 to 91 Sjévall’s formalin-water-osmium method for the Golgi’s network.
92 to 94 Weigl’s method for the Golgi’s network.
95 to 98 Cajal’s uranic nitrate-silver method for the Golgi’s network.
418
»
GLANDULAR CELLS OF THE u FROG’S PANCREAS PLATE 4
8. SAGUCHI
419
PLATE 5
99 From an injection preparation.
100 to 105 Flemming’s fluid; iron-haematoxylin.
106 to 109 Bichromate-formalin; iron-haematoxylin.
110 Sublimate-acetic acid; alum-haematoxylin.
111 to 117 Meves’ fluid, iron-haematoxylin. Figure 113 shows a cell which
is just undergoing constriction. The fragments are to some extent invaginated
the neighboring cells, which are not delineated in the figure. Figures 115 to 117,
constricted fragments which still contain nuclear fragments and zymogen
granules.
118 Sublimate-acetic acid; alum-haematoxylin. Figures 114 and 118 show
normal cells containing constricted fragments.. Some of the latter contain
nuclear portion, others none and appear homogeneous, and still others contain
zymogen granules.
119 and 120 Cajal’s uranic nitrate-silver method.
420
GLANDULAR CELLS OF THE FROG’S PANCREAS , PLATE 5
S. SAGUCHI
He) I) IG
qe, 120. ei 7
»* ry : : 7é ¢ é <a
421
Resumen por el autor, Frederic T. Lewis.
Escuela Médica Harvard, Boston.
Il trayecto de los tubulos wolffianos en los embriones de los

mamiferos.
Después del estado bien conocido de tubos en forma de §, los

tuibulos wolffianos del hombre, cerdo y otros mamiferos adquieren
la forma de doble espiral descrita por Milhakovies (figuras 1—4).
Esta doble espiral es divisible en tres segmentos que, comen-
zando por la edapsula glomerular, pueden designarse con las letras
C, U y Z, respectivamente. El ultimo segmento desemboca en
el conducto wolffiano (fig. 8). En los embriones humanos aunque
el tubo se alarga desde 0.8 mm en un embrién de 7.5 mm hasta
1.5mm en un embri6n de 22.8 mm, la disposicién no se hace mas
compleja. Al contrario, mediante el alargamiento de los seg-
mentos C y Z (el segmento U no aumentando de longitud) el
tiibulo forma una sola asa. (figs. 5-7). En el cerdo, los tubulos
al principio tienen la misma forma de doble espiral que en el
hombre. Pero rapidamente se hacen mucho mas largos (crecen
desde 1.9 mm en un embri6n de 6 mm hasta 13.5 mm en un em-
brién de 20 mm) y son por consiguiente mas complejos. Entre
los segmentos Y y Z se forma un asi V, proyectandose en la con-
cavidad de la U. El aspecto asi producido se ha representado
en las figuras 10-13, que demuestran una sorprendente regulari-
dad en la disposicién de las vueltas de espiral.
Translation by José ’. Nonidez
BY THE BIBLIOGRAPHIC SERVICE, OCTOBER 27
THE COURSE OF THE WOLFFIAN TUBULES IN

MAMMALIAN EMBRYOS
FREDERIC T. LEWIS
Harvard Medical School, Boston, Massachusetts
THIRTEEN FIGURES
In early stages, when the Wolffian tubules are so short that

they may be cut throughout in single sections, their course from
glomerular capsule to Wolffian duct has been thoroughly studied.
Kolliker, Mihalkovics, Meyer, Schreiner, and others have de-
scribed their S-shaped form. But in older embryos the length
and sinuosities of the tubules are such as to require reconstruc-
tion. ‘‘Nevertheless,’ as Nicolas wrote in 1891, ‘it can be
recognized without too great difficulty that their general orienta-
tion has remained unchanged—they have merely elongated and
folded capriciously in different planes.”’ The idea of capricious
folds is well illustrated in the single reconstruction of a tubule
from a human embryo of 10.2 mm., published by Kollmann,
and many times reproduced. But Felix records that he has
“studied two hundred models of mesonephric tubules from the
most different stages of development,” and in none of them has
he found ‘‘any coiling of the tubule.’”’ Consequently he infers
that Kollmann’s model must represent ‘a very exceptional rarity’
—a conservative and accurate conclusion. The only change ob-
served by Felix after the tubule has become S-shaped occurs near
the junction of the middle and distal segments of the S (distal
meaning toward the Wolffian duct), and consists in the formation
of “a loop directed either cranially or caudally.”’ Unfortunately
Felix has not figured in detail any of the late stages included
among the two hundred tubules modeled, presumably from
human embryos, so that Kollmann’s drawing appears to occupy
the field alone.
423

424 FREDERIC T. LEWIS
In the pig the tubules are more highly developed than in man.
‘They have been carefully studied by MacCallum, and are shown
in two published reconstructions which seem to reveal more than
is brought out in the accompanying text. Taken in connection
with two models now in the Harvard Laboratory, they indicate
that there is a pattern in the Wolffian tubule of the pig, quite
as definite and interesting as that of the metanephric or renal
tubule. Although a larger number of tubules should be modeled
to establish this conclusion and to show the range of variation,
the following interpretation of the somewhat laborious work °
already done may be of value. A simple pattern for the human
tubules, and a more complex one for those of the pig, will be
presented, showing how both may be derived from a common
origin, one by simplification and the other by elaboration.
Omitting from consideration the transformation of the pri-
mary vesicle, we may begin with the S-shaped stage which was
well described, and perhaps for the first time, by K6lliker in
1879. Referring to a rabbit embryo somewhat older than that
here shown in figure 1, he writes:
“From the Wolffian duct there arises first a very slender tubule which
passes medially along the dorsal side of the Wolffian body, clear across
the organ; then, making a loop, it bends upon itself and retraces its
course to the lateral side; finally, after a third coil, it ends in the Mal-
pighian corpuscle, medially placed on the ventral side.” He disposes
of the later stages as follows: ‘“These three chief coils become compli-
cated by the formation, at the places where the bends occur, of acces-
sory coils in different planes, so that finally the course of a single tubule
becomes so complicated that it cannot be unravelled in sections.”
That the young tubules have the form of a letter S was further
established by Mihalkovics in 1885, from studies of the lizard,
duck, chick, and sheep. Beginning as a detached vesicle, the
tubule becomes cupped or crescentic, with the glomerulus develop-
ing within its coneavity. It then takes the form of a ‘ladle’ or
‘sickle,’ as a short handle is marked off, joining the Wolffian duct.
This stage, as Mihalkovies found, gives place to the 8. He con-
sidered that the distal limb of the S could be set apart as the
tubulus collectivus and that at the region of the distal bend it
became dilated and coiled, forming a tubulus secretorius, but
WOLFFIAN TUBULES IN MAMMALIAN EMBRYOS 425
to these designations Nicolas and Von Winiwarter have taken

exceptions.
Of greater interest perhaps, and not open to question, is the
finding by Mihalkovics that in the sheep the middle or transverse
limb of the S elongates and becomes itself S-shaped, though the
long axis of the second § is horizontal, being at right angles with
that of the first. ‘‘Here in a narrow space,” according to Mihal-
kovies, ‘‘a winding-up process takes place,’’ which leads to a
double or reversed spiral, as shown in figures 3 and 4. Mihal-
kovies did not follow the further development of this pattern.
SO eOe,
Fig.2.
: o
G
Fi9.3.
Fig).
Fig. 1 Section of an S-shaped Wolffian tubule. Rabbit embryo, 103 days,

5.4mm. Harvard Embryological Collection, series 560, sections 158-161. X 200.
Fig. 2 Diagram of the S-shaped tubule.
Fig. 3. Diagram of the double-spiral tubule.
Fig. 4 Section of a double-spiral tubule. Pig embryo, 4.5 mm. Harvard
Embryological Collection, series 1404, sections 114-115. X 200.
Gl., glomerular capsule; W.d., Wolffian duct.
In 1890 Meyer described the simple S-stage in man, and in 1902,

with a wealth of lithographs, Schreiner repeated the observations
of Mihalkovies on the early stages in the lizard, duck, and chick,
with the rabbit in place of the sheep. He refers to the double
spiral pattern as a ‘much-coiled §,’ but there is nothing to suggest
that he regarded it as a particularly significant form, essentially
different from the simple 8. Minot, in 1892, in an account based
upon Mihalkovies, notes that the tubules of amniota ‘‘retain for
some time their simple S-shape, although the curves of the 5
become more and more exaggerated.”” ‘These appear to be the
Fig 4.
Figs. 5 to 7 Models of Wolffian tubules in human embryos. X 175 (approxi-

mately).
principal references to the double spiral which follows the simple

S and which is here regarded as the common origin of the later
patterns and the key to their interpretation.
A suitable designation for the form which we have called a double
spiral (fig. 3) is unfortunately not available. It is a shape which occurs
at the apex of the coil of large intestine in the pig, where some of its
interesting properties have been discussed by Lineback, and it may
perhaps be found elsewhere. Impressed with Thompson’s application
of mathematics to biological problems, the aid of a mathematician was
sought in describing the form in figures 3 and 4, and in determining
what factors of growth might account for the regular transformation
of the simple § into the double spiral. There are, however, so many
unknown factors in the nature of the surrounding medium, which may
keep the growing coil within a circular area, and in the physical nature
and method of expansion of the tube-wall itself, that a mathematical
interpretation was not forthcoming, and the problem was returned as
‘biological.’ It may be noted that mitotic figures are rather evenly
distributed, and that no evidence is found to support the idea of Mihal-
kovies that a special elongation of the middle segment in cramped
quarters produces the double spiral. In fact, if the middle portion
remains short and the two ends of the simple § elongate, continuing
their respective curves, the double spiral will result, though the axis
of the figure rotates. In the shifting position of the Wolffian duct in
later stages, there is clear evidence that such a curved elongation of the
distal segment actually takes place.
In human embryos the simple S may be expected in 6-mm.

specimens and in those somewhat smaller. At 7.5 mm. the
double spiral has formed, as shown in the model, figure 5. The
tubule chosen was one in which the essential bends fell within
the plane of section, thus lessening the chance of distortion in
the process of modeling. The liberty has been taken of showing
the tubule reversed in the figure, as if it belonged to the left
Wolffian body, to facilitate comparison with other models. Ex-
Fig. 5 Embryo of 7.5 mm. Harvard Embryological Collection, series 256,

sections 471-485.
Fig. 6 16.0 mm. Harvard Embryological Collection, series 2044, sections
1045-1067.
Fig. 7 22.8 mm. Harvard Embryological Collection, series 871, sections
887-908.
Fig. 8 Diagram of the double-spiral stage, shown in figure 5, with its sub-
division into segments C, U, and Z.
Fig.9 A corresponding diagram of the stage shown in figure 7.
cept for reversal, no difference in shape could be observed

between the tubules of the opposite sides. The embryo shows
thirty-four tubules entering the right Wolffian duct, which is the
number credited to this specimen in Doctor Bremer’s earlier
study. On the left there are thirty-seven tubules, including
four of somewhat retarded development at the caudal end. Be-
ginning the count anteriorly, the tubule modeled is the twenty-
seventh. As calculated from the model, which was made at
an enlargement of 500 diameters, the length of the tubule from
glomerular capsule to Wolffian duct is 0.3 mm.
For purposes of description this double-spiral tubule may be
divided into three parts, as shown in the diagram, figure 8.
Beginning with the capsule, there is a portion which may be
designated C, this letter suggesting its shape. It curves laterally
and dorsally and passes into the lateral arm of a portion named
U, the letter again suggesting the form. Finally, the medial
limb of the part U ends in another simple C-shaped curve, which,
as it is the terminal portion, may be designated Z. This swings
laterally to end in the Wolffian duct which at this early stage is
both lateral and dorsal. The rather cumbersome subdivision of
the tube into C, U, and Z portions will prove especially service-
able in the more complicated later stages in the pig embryo.
The second model of a human tubule (fig. 6) is from an embryo
of 16 mm., excellently preserved and stained with iron haema-
toxylin. The tubule selected is from the left Wolffian body, in
which there are thirty tubules entering the Wolffian duct, this
one being the eighteenth. Presumably it has passed through a
stage similar to that in figure 5, but has increased in length from
0.3 mm. to 0.8 mm. Although cytological differentiation has
advanced, the coiling of the tubule has undergone a slight
retrograde movement. Beginning at the glomerular capsule,
there is a somewhat constricted neck, as others have frequently
noted, and portion C is recognized as the tubulus secretorius
of Mihalkovies or tubulus postglomerularis of Nicolas. It is
capacious, with wide lumen, and thick walls composed of some-
what elongated cells distinctly marked off from one another.
These cells have conspicuous terminal bars, a frayed brush-
border, a superficial granular zone not deeply stained, and a still

paler basal zone containing the oval nuclei. In a typical manner
they show the histological features of secretory activity which
cannot be reconciled with Gadow’s opinion that “‘the Wolffian
body in birds and mammals appears never to function, not even
in the embryo.’
In the 16-mm. embryo, the segment U may be considered to
begin where the tubule becomes abruptly narrower, with a corre-
sponding radical change in the histological nature of its walls.
The lumen is narrower, and the cells which are longer, with less
cytoplasm, stain deeply with iron haematoxylin. The portion
U is therefore a subdivision of the tubulus collectivus of Mihal-
kovies or tubulus terminalis of Von Winiwarter. Whatever its
function, it must be very different from that of the preceding
part. With certain minor swellings and constrictions, the portion
U makes the bend and passes gradually into Z. As the Wolffian
duct shifts ventrally, this terminal part of the tube has become
longer and more curved. Sometimes, as noted by Felix, it does
not remain in the same horizontal plane with the rest of the
tubule, but may bend either anteriorly, as in the figure, or
posteriorly.
In an older embryo—22.8 mm., figure 7—certain tubules are
much longer than before, though degeneration of the Wolffian
body has become active. Felix found that from the stage of
21 mm. onward, the Wolffian tubules are ‘almost all broken in
one or several places,” notably at the neck of the glomerular
capsule, at the junction of the secreting and collecting portions,
and at the outlet into the Wolffian duct, and that ‘‘all degenerat-
ing tubules show a tendency to lose to a greater or less extent
their S-shape.”’ The tubule here modeled appeared to be intact
throughout, though at this stage they are so difficult to follow
that errors are quite possible, either in introducing a disconti-
nuity or in filling an actual gap. The tubule chosen for modeling
1 This statement is found on page 829 of Bronn’s Klassen und Ordnungen des
Thier-Reichs, Bd. 6, Abth. 4, Végel, von H. Gadow and E. Selenka. It appears
to be in Gadow’s portion of the work and is wrongly credited to Selenka by Weber.
who among others, accepts it. See Weber, In Schwalbe’s Archiv, 1897, page 621.
is the twenty-eighth to enter the Wolffian duct on the left side;

three more occur below it. Of these thirty-one tubules, the
anterior are extensively degenerated, and lend no support to
MacCallum’s statement that “the degeneration of tubules pro-
gresses from the posterior end of the organ forward.” On the
contrary, a posterior tubule was chosen as probably typical of
those most highly developed. Its actual length, as calculated
from the model, is 1.5 mm.—nearly twice that of the preceding
stage. The portion C, with its glomerular end bent upward or
anteriorly, lies within the concavity of portion Z, which also
extends upward to enter the Wolffian duct. This disposition of
parts would produce the result observed by MacCallum when
injecting the tubules of the pig. The fluid could be seen, through
the thin overlying tissue, to pass from the Wolffian duct into
the tubules and to flow through them around the lateral surface
of the gland to the dorsal border. At a certain place, in tubules
just beneath the superficial ones, the fluid could be seen to run
in the opposite direction.
Although the tubule at this stage is of a simpler pattern than
the double spiral, and consists of two rather than three segments,
the remains of the middle or U-portion can be identified. A
narrowing of the tubule marks the place where the U begins,
and it proceeds around the bend into Z, where it ends without
definite boundary. An interpretation of the model in terms of
the double spiral is shown in.figure 9, and the group of figures
(5 to 9) presents the evolution of a human tubule from a length
of 0.3 mm. to 1.5 mm. without the formation of capricious folds,
but according to a very simple law.
That the Wolffian body of the pig is more ee developed
than in man is shown in several ways, but very clearly in the
greater length of its tubules. The length in a 6-mm. pig, as
calculated from a single model, is 1.9 mm., and at 20 mm. it has
increased to 13.5 mm.; that is, the tubules of the pig are found
to be from six to eight times longer than those of man, and more
extensive coiling may be anticipated. The typical double spiral
is formed early in the pig, as already shown in figure 4, from an
embryo of 4.5 mm. A somewhat later stage was modeled in
1909 by Dr. F. T. Krusen, in connection with the undergraduate

course in embryology given by Professor Minot and assistants.
This model, duly catalogued and placed inthe Harvard Collection,
has apparently never before been utilized. Its essential features
are shown in figure 10, in which such slight displacements have
been made as are necessary for following the coils easily. In
this tubule, from a 6-mm. pig, the segment U is directed toward
the concavity of C, and a new loop, V, intervening between U
and Z, has appeared, pointed toward the concavity of the U;
Z is without special features. This pattern might be regarded
as a casual form were it not duplicated in MacCallum’s figure,
here reproduced reversed as figure 11.
MacCallum described his figure as a “diagrammatic recon-
struction of Wolffian tubules” from a pig embryo of 8 mm., and
he presented it as applying to ‘‘a number of tubules.”” Referring
to an older embryo, he writes: “Special names might be given
to the different parts of the tubule, but until their significance
is more definitely known this could be of little value. There is,
however, a very distinct division into a secretory and a conduct-
ing part.’ But without discussing relative values of morpho-
logical and physiological subdivisions, it is clear that we stand
on firmer ground in describing the course of the tubule than in
assigning a functional significance to its various bends. It is
unfortunately true that the limits of the secretory portion in
the pig remain unknown and MacCallum did not succeed in
defining them. He put together the upper part of C (in fig. 11)
and the adjacent limb of the U as the ‘secretory loop,’ and this
loopforms the only subdivision of the tubule which he recognized.
But judged by its large diameter in figure 10, the remainder of
the U is also secretory; whereas in man none of the part U was
of the same nature as C, and the latter was certainly secretory.
Whether the subdivisions here imposed upon MacCallum’s figure
are profitable or not may be questioned, but they enable one to
sketch quickly the course of the Wolffian tubule not only in the
young embryos now considered, but also in older ones where
previously no description was attempted.
For permission to make and publish figure 12, the writer is

under special obligation to Dr. Frank H. Rose, who made the
model, and to Prof. Franklin P. Johnson, who directed Doctor
Rose’s study in the University of Missouri. This model of a
tubule in a 20-mm. pig is incidental to a comprehensive study of
that embryo, being made by Professor Johnson and his pupils,
and later to be published in detail. The simplified sketch, show-
ing all the essential bends and loops in this long and involved
tubule, may readily be compared with the figures of the earlier
stage already discussed. From the glomerular capsule, with
initial kinks, the portion C sweeps in a well-rounded curve to
the dorsal border, and forms a U in the concavity of the C and
a V in the concavity of the U, with a final sweeping curve enclos-
ing them all—the portion Z.
This model again accords, in surprising detail, with Mac-
Callum’s figure of the tubule from a much larger pig, measuring
80mm. Neither the length of the tubule nor the magnification
of the figure is stated, but the tubule, here shown reversed in
figure 13, is evidently much longer than in the preceding stage.
A secondary coil has appeared along the portion U and another
in connection with V. The coil in the descending limb of the
U chances to be comparable with a kink in figure 12, but the con-
stancy of this feature remains to be determined. The new loop
in segment V, which may or may not correspond with a slight
bend in the tubule at 20 mm., is not directed toward the apex
of the V; if it should become so, it would carry the evolution of
the characteristic pattern one step further than has yet been
observed.
Finally, it may be repeated that the total number of tubules
studied is small for drawing general deductions. But the close
agreement in the findings of independent observers and the
absence of a single aberrant form from the group carefully mod-
eled are evidence in favor of the interpretation presented. In
these models an easily recognized pattern exists, much more
complicated than the double-spiral form from which it springs.
WOLFFIAN TUBULES IN MAMMALIAN EMBRYOS 42 ww
e
Fig. pee Fig. 13.
Figs. 10 to 13 Wolffian tubules of pig embryos. Small irregularities in the

models have been omitted, and the coils have been slightly spread apart to dis-
play the course of the tubules. Colors have been used to mark out the segments
C, U, V, and Z, thus demonstrating more clearly the type pattern.
Fig. 10 Model by F. T. Krusen. Pigembryoof6mm. Harvard Embryologi-
cal Collection, series 918.
Fig. 11 Reversed drawing of MacCallum’s ‘diagrammatic reconstruction.’
Pigembryoof8mm. Am. Jour. Anat., vol. 1, p. 251, fig. 7.
Fig. 12 Model by F. H. Rose. Pig embryo of 20mm. University of Missouri
Embryological Collection, no. 1.28.
Fig. 13 Reversed drawing of MacCallum’s ‘wax reconstruction.’ Pig embryo
of 80mm. Am. Jour. Anat., vol. 1, p. 252, fig. 8.
LITERATURE CITED
Bremer, J.L. 1911 Morphology of the tubules of the human testis and epididy-
mis. Am. Jour. Anat., vol. 11, p. 404.
Feitx, W. 1912 The mesonephros. Human embryology, edited by Keibel and
Mall, vol. 2, p. 807.
Gapow, H. 1891 Végel. Bronn’s Klassen und Ordnungen des Thier-Reichs.
Bd. 6, Abth. 4, Th. 1, S. 829.
Kouircer, A. 1879 Entwicklungsgeschichte des Menschen. Zweite Auflage,
S. 944.
Kouimann, J. 1898 Lehrbuch der Entwickelungsgeschichte des Menschen,
S. 398-399.
LineBAcK, P. E. 1916 The development of the spiral coil in the large intestine
of the pig. Am. Jour. Anat., vol. 20, pp. 488-489.
MacCauium, J. B. 1902 Notes on the Wolffian body of higher mammals. Am.
Jour. Anat., vol. 1, pp. 250-254.
Meyer, H. 1890 Die Entwickelung der Urnieren beim Menschen. Arch. f.
mikr. Anat., Bd. 36, 8. 161.
Minaxxovics, G. V. von 1885 Untersuchungen iiber die Entwickelung des
Harn- und Geschlechtsapparates der Amnioten. Internat. Monatschr.
f. Anat. u. Physiol., Bd. 2, 8. 76-79.
Minot, C.S8. 1892 Human embryology, p. 239.
Nicouas, A. 1891 Contribution 4 l’étude des cellules glandulaires. I. Les
éléments des canalicules du rein primitif chez les mammiféres. Ex-
trait du Journ. internat. d’Anat. et de Physiol., T. 8, p. 94 (of the
reprint )—‘‘Conclusions.”’
Scurerner, K. E. 1902 Uber die Entwicklung der Amniotenniere. Zeitschr.
f. wiss. Zool., Bd. 71, 8. 96.
Tuompson, D. W. 1917 On growth and form. Cambridge University Press.
Weser, 8. 1897 Zur Entwicklungsgeschichte des uropoetischen Apparates bei
Saugern. Morphol. Arbeiten. Bd. 70, S. 621.
~ Wintwarter, H. von 1910 La constitution et l’involution du corps de Wolff.
Arch. de. Biol., T. 25, p. 184.
ADDENDUM
The foregoing account pertains only to mammals but it may be of interest

to compare the models of the Wolffian tubules in pig embryos with those of the
corresponding structures in adult amphibians. For this purpose two figures are
available, by Nussbaum and Huber respectively. Nussbaum (Arch. f. mikr.
Anat., 1886, Bd. 27, Taf. 23, Fig. 28) shows a mesonephric tubule isolated by
maceration from the kidney of Rana esculenta. Any subdivision into C, U, V
and Z portions seems impossible. Huber (Anat. Rec., 1917, vol. 13, p. 310) pre-
sents a wax reconstruction of a tubule from Rana catesbiana, which has ‘many
points of similarity’ to Nussbaum’s figure, and yet, notwithstanding many
secondary convolutions, may readily be resolved into the pattern in question.
Further study is needed to show whether this interpretation of the model is of
any real significance.
Resumen por el autor, William H. F. Addison.
Universidad de Pennsylvania.
Estudio histolégico del bazo del conejo durante el aumento de

la actividad fagocitica.
E] autor ha estudiado la capacidad fagocitica de los espleno-

citos del bazo del conejo después de inyectarle corptisculos san-
guineos lavados de paloma. La hemolisis ordinaria de los cor-
pusculos de la paloma sigue a la inyeccién y, casi como un efecto
inmediato, en un cierto ntiimero de casos siguid una liberacién
de considerable niimero de células sanguineas maduras y no ma-
duras por la médula de los huesos. La sangre hemolizada de la
paloma y las células de la médula é6sea del conejo, al llegar al
bazo transportadas por la sangre, permanecen dentro de los
canales sanguineos cavernosos de la pulpa y comienzan pronto a
ser ingeridas por las células esplénicas. Los esplenocitos crecen
aumentando su contenido hasta que al cabo de 16 horas adquieren
considerable tamano. Asi, por ejemplo, un esplenocito que media
55 x 23.3 yw presentaba 20 inclusiones celulares en un solo corte
de 4 uw de espesor. Los esplenocitos que contienen de 5 a 10
células son de menor tamafo. A medida que procede la diges-
tidn de las células ingeridas, los esplenocitos disminuyen de
tamano y al cabo de 48 horas muchos de ellos no son mayores que
los que se encuentran en el bazo normal. De los productos de la
hemolisis de la sangre de la paloma después de la inyeccién poco
queda de un modo permanente en las 6rganos del cuerpo; la
mayor parte se eliminan por los rifones durante las 16 horas.
En el bazo los esplenocitos presentan un aumento de substancia
férrica, mientras que las células endoteliales no presentan, en
su mayor parte, un aumento de tal substancia. En este pro-
cedimiento experimental dondequiera que células o fragmentos
celulares sirven de estfmulo para la fagocitosis, los esplenocitos
son los primeros en actuar y continuan actuando como los prin-
cipales agentes fagociticos.
BY THE BIBLIOGRAPHIC SERVICE, NOVEMBER 17
HISTOLOGICAL STUDY OF THE SPLEEN OF THE

RABBIT UNDER HEIGHTENED PHAGOCYTIC
ACTIVITY
WILLIAM H. F. ADDISON
Anatomical Laboratory and the Department of Research Medicine, University of
Pennsylvania, Philadelphia
ONE PLATE (SIX FIGURES)
The phagocytic reaction in the mammalian spleen has been

studied under various experimental conditions and from different
points of view. Histologically, the accumulated observations
show that the same cell types are not the active phagocytic
agents in all cases. Thus in some conditions it is found that
fixed cells, such as reticular and endothelial cells, are most active
(Heinz, ’01; Cary, ’15) while in other conditions free cells, such
as splenocytes, are the predominant phagocytic cell type (Muir,
02). In other conditions, again, cells of both types take part
(Paton and Goodall, ’03). These findings are to be correlated
with the well-recognized fact that phagocytic cells have a selective
action, some types reacting to one stimulus, some to another.
With the purpose of examining some of the conditions under
which the phagocytic cells of the spleen act, we have studied
this organ in one definite experimental procedure, paying special
attention to the histological character of the cells involved in
the reaction.
The stimulus used has been the injection of washed pigeon
corpuscles into the circulation of the rabbit. Rapid destruction
of the foreign corpuscles ensues and, as an almost immediate
effect, is followed by the release of great numbers of mature and
immature blood-cells from the bone-marrow. Both the hemolyz-
ing blood of the pigeon and the bone-marrow cells of the rabbit,
being brought by the circulating blood to the spleen, are delayed
within the cavernous blood-channels of the pulp, and are there
exposed to the phagocytic action of the splenic cells.
437

438 WILLIAM H. F. ADDISON
EXPERIMENTAL PROCEDURE
The pigeon blood was prepared for injection in the following

manner. The pigeon was bled from the jugular veins into 2
per cent sodium citrate in 0.85 per cent sodium chloride solution.
Usually 10 to 12 cc. of blood was available, and about 25 ee. of
the citrate solution was used to receive the blood. This was
centrifuged for fifteen minutes, and the supernatant fluid pipetted
off. The corpuscles were next shaken up with 2 ce. of the citrate
solution, and then the centrifuge tube filled with 0.95 per cent
NaCl solution. The corpuscles were washed twice again, each time
first shaking up the corpuscles with a small amount of the citrate
solution, and the tube then filled with the 0.95 per cent NaCl
solution. Finally, the volume of the washed corpuscles was
made up to 10 to 12 cc. with the 0.95 per cent NaCl. Under
this treatment there was no visible hemolysis and under the
microscope there was seen no clumping of the red blood-cells.
Usually 10 ee. of this was injected into a superficial ear vein of
the rabbit. The rabbits were taken at varying intervals after
injection, from one hour to forty-eight hours after injection.
After etherizing the rabbit, the animal was bled through the
inferior vena cava and normal saline transfused through the
thoracic aorta. After the spleen was partly washed out, as
shown by its paler color, the fixing fluid was run in, and attempts
were made to distend the spleen as much as possible. The fixing
fluid used was a mixture of commercial formalin, made neutral
by previously adding magnesium carbonate in sufficient quantity
to form a thin layer at the bottom of the stock bottle, and 3 per
cent aqueous solution of potassium bichromate. The two fluids
were mixed, just before using, in the proportion of one part of
neutral formalin to nine parts of the bichromate. Small pieces
of the organ were subsequently fixed for a period of twenty-four
hours in the same fluid, before washing, dehydrating, etc., pre-
paratory to embedding in paraffin. Hematoxylin and eosin or
acid fuchsin were used for routine staining. Tests for the pres-
ence of iron were made by the potassium ferrocyanide and
hydrochloric acid method.
PHAGOCYTIC ACTIVITY OF THE SPLEEN 439
NORMAL SPLEEN
In the spleen of the normal rabbit can always be found phago-

cytic cells with inclusions. The detection of these cells is greatly
facilitated by spreading apart the constituent tissues of the spleen
pulp by intravascular injection of fluids. Although part of the
blood content is driven out by this procedure, one can never force
out all the blood, and the great advantage is that the resultant
preparation 1s much more easily studied.
The largest of these phagocytic cell types are the free-swimming
mononuclear cells, called splenocytes. These vary in size accord-
ing to the bulk of their contents. The nucleus tends to be
rounded or oval in shape, and is usually eccentrically placed.
The nuclear membrane is very well defined, and the chromatin
is distributed in one or two larger dots and in a network of fine
lines. In general the nuclear contents are pale-staining, as com-
pared with the nuclear membrane. The most evident character-
istic in the cytoplasm is the presence of inclusions. These inclu-
sions are sometimes single, but are usually multiple. They are
most frequently in the form of light yellow spots embedded in
the cytoplasm. Indeed, a useful method of detecting the spleno-
cytes is to search over the microscopic field for little clusters of
golden-yellow dots. If such a cluster is in a closely compacted
mass of cells, diagnosis of the containing cell is very difficult.
But if the cell lies free within a pulpar blood space, then the
configuration of the cell is readily seen, as shown in figure 1.
These contained spots are of varying size, and may not all be
of exactly the same color. Sometimes they are closely packed,
making a nearly homogeneous mass. In other cells the dots are
farther apart, so that the intervening cytoplasm is seen between
them. All these appearances are regarded as products of the
intracellular digestion of entire erythrocytes, or fragments of
them, in the form of hemosiderin. In testing for iron in such
cells, these clusters of dots give a strongly positive reaction.
Occasionally one sees what appears to be an entire red blood
corpuscle still intact within the splenocyte, but final judgment
is often difficult. Or again, the inclusions may appear as hemo-
globin-bearing fragments of erythrocytes, as indicated by color

comparison with adjacent red blood corpuscles. Not infre-
quently, too, a splenocyte is seen containing a polymorphonuclear
leucocyte.
These polymorphonuclear leucocytes in the rabbit are con-
spicuous, for-in addition to the characteristic shape of nucleus,
their numerous fine cytoplasmic granules stain a vivid red color
with stains containing eosin. This coloring may lead to some
confusion at first in distinguishing them from true eosinophiles.
However, the fact that the granules in the latter are larger than
in the ordinary polymorphonuclears, and also that their color
after eosin staining is more yellowish, helps one to distinguish
between them.
SPLEENS OF EXPERIMENTAL ANIMALS
After injection of the pigeon blood, prepared in the manner

above described, the corpuscles retain their individuality for
but a short time. At the end of one hour they are found to be
swollen, rounded, and the nucleus has lost its characteristic
staining capacity. Most of the corpuscles are single, but some
have begun to agglutinate into little masses. In some elements,
the nucleus is still distinguishable by its refractivity. Brown-
Sequard, in 1857 (Hunter, ’01, p. 120), noted that while dog or
rabbit corpuscles injected into the circulation of the domestic
fowl could be found at the end of a month, the corpuscles of the
fowl injected into the dog or rabbit were no longer recognizable
at the end of one hour. Our preparations, in the main, confirm
the latter observation.
At this stage, the spleen has grossly a purplish-black color,
and its channels are congested with the masses of intermingled
rabbit and pigeon corpuscles. It was found that perfusion was
even less successful than usual in washing out the blood. A few
splenocytes containing inclusions recognizable as entire pigeon
corpuscles are seen, but splenocytes with inclusions of amorphous
brownish masses are more often to be found, but not in every
field, as viewed with the ;5 oil-immersion lens. Polymorphonu-
clears are seen more frequently than in the normal spleen, and
PHAGOCYTIC ACTIVITY OF THE SPLEEN 44]
are readily detectable by reason of their red staining granules.

No color was yet visible in the urine.
Two hours. At the two-hour period the injected pigeon cor-
puscles are no longer recognizable as such, having lost their
original form and finally become drawn out and subdivided into
smaller and smaller globoid and irregular fragments. That the
amount of hemoglobin, either free or still in combination with
minute portions of the substance of the corpuscles, is overtaxing
the capacity of the organism is shown by the very evident hemo-
globinuria. The cells of the liver show many minute brown
granules in their cytoplasm. The number of pseudo-eosinophilic
polymorphonuclears has increased, often a dozen being seen in
one field. A new and larger type of cell, containing red-staining
granules, now appears in small numbers, at most two or three in
afield. These are cells with a single eccentrically placed nucleus,
which has a very well-defined nuclear membrane. Within the
nucleus are one or two larger dots of chromatin, but otherwise
the chromatin is very finely divided. The shape of the nucleus
usually tends to be spherical, but may have one side flattened or
slightly concave. An examination of the bone-marrow shows
that similar-appearing cells are there in great numbers. They
are both in the blood-channels and in the framework tissue, and
most of them, though not all, are packed with the red-staining
granules. This comparison leads to the conclusion that the new
type of cell now appearing in the spleen has come from the bone-
marrow and is of the myelocyte type.
The splenocytes are not greatly different in number or size
from before. Some are conspicuous by their multiple brownish-
colored inclusions. One of these cells measured 20.5 » X 12.3 u,
which is no larger than splenocytes, containing several inclusions,
from the normal spleen. Occasionally a splenocyte is seen con-
taining a pseudo-eosinophile, in addition to remains of red blood
corpuscles.
Three hours. At three hours after injection, the pigeon cor-
puscles have been reduced to brown-colored masses and clumps
of fine particles. Even more intense hemoglobinuria is found
than at the preceding stage. Polymorphonuclears and myelo-
cytes are present in about the same numbers as before. The

splenocytes contain brown masses derived from the ingested
particles of blood. These may exist as a single large mass in a
cell or as many small ones. The average size of the splenocytes
has not increased to any extent.
Four and one-half hours. Fragments of the injected pigeon
corpuscles are still abundant, intermingled with the rabbit cor-
puscles. The polymorphonuclears have increased in number and
myelocytes are frequent. The splenocytes may contain a single
brownish mass, or more frequently many inclusions, usually
brownish globules and often leucocytes. One cell was seen
containing seven polymorphonuclears, as well as many fine brown
dots. The average size of five splenocytes was 16.7 x 10.8 u,
with nucleus 6.8 xX 5.9 uw. Measurements of five myelocytes
showed average size of cell-body of 10 x 8.7 u, with nucleus
7X 5u. The average size of the pseudo-eosinophilic polymor-
phonuclears, as seen in sections, is 6.5 X 6 uy.
Six hours. After six hours, the appearance of the sections is
very striking from the presence of numerous myelocytes, with
red-staining granules. What appears to be mitotic division in
these cells is seen not infrequently (fig. 2, My:). Whether this
may be considered a normal mitotic division has been questioned.
The polymorphonuclears (fig. 3) are even more numerous than
the myelocytes, but the latter on account of their larger individ-
ual size form the most conspicuous elements in many fields.
Splenocytes are scattered here and there and have inclusions
similar to those in the preceding stage. There are occasional
extracellular masses of yellowish hemosiderin. Several mega-
cytes were seen, and these are apparently cells which have re-
mained in the spleen from fetal life, for in rodents’ spleens
such cells are found not infrequently. One of these is shown
in figure 3 (M), and its size is useful for comparison with the
size of the splenocytes in their several stages of phagocytic
activity. The hemoglobinuria still persists. Within the endo-
thelial cells of the blood-channels very fine yellowish granules
could be seen. Tests for the presence of iron showed this to
be present within these cells and also very strikingly within the
splenocytes.
Eight and one-half hours. The appearance at this stage is

similar to the preceding. There is an abundance of myelocytes
and of pseudo-eosinophiles. The splenocytes usually contain
numerous small inclusions of hemosiderin and are not very
conspicuous.
Twelve hours. The splenocytes are now beginning to show
inclusions of nucleated cells, in addition to the blood pigment
derived from the pigeon red blood corpuscles. This is the last
stage at which hemoglobinuria was noted, and apparently the
pigeon blood has for the most part been disposed of. The
phagocytosis of the bone-marrow cells by the splenocytes is now
commencing, and shows most strikingly at the next stage.
Siateen hours. At this period, one can readily discern the
splenocytes with a low power of the microscope, by reason of
their increased size. In figure 4 is shown a characteristic field.
Every splenocyte contains a number of nucleated cells, both
polymorphonuclears and myelocytes with their granules still
staining characteristically. An average size is 25 x 20 yu, in
contrast to their previous size of 16 X 10 ». A large one, shown
in figure 5, measures 55 X 23.4 uw, and in single 4 » section, twenty
cells can be counted within its cytoplasm. This period evidently
shows the splenocytes at the height of their phagocytic activity
toward the animal’s own bone-marrow cells. There are scattered
extracellular masses of hemosiderin. The polymorphonuclears
are still abundant, but the myelocytes are apparently not so
numerous, and this is probably due to the activity of the spleno-
cytes. At this period there was no discernible hemoglobinuria.
Tests for iron showed this to be present in many endothelial cells,
but in much greater amounts within splenocytes.
Twenty-one hours. The splenocytes are now smaller in size,
and accordingly less conspicuous than at the preceding period.
Their diminution in size is to be associated with the shrinkage
of their cellular inclusions. While in some of the splenocytes,
the ingested cells show both nuclei and granules stained character-
istically, in others, only the nuclei of the ingested cells remain
recognizable, and apparently the red-staining substance has been
altered, probably by digestive ferments within the cell. Some
splenocytes also contain small yellowish or light brown inclusions

derived from the original pigeon blood. Occasionally a spleno-
cyte is seen containing one single larger rounded mass of hemo-
siderin. While polymorphonuclears are still scattered about, no
distinct myelocytes were found.
Forty-eight hours. The splenocytes are again easily detecta-
ble in the microscopic field, because of the deep staining of the:
remains of the cells which they had ingested. These cellular
inclusions, for the most part, no longer resemble the original
cells. The nuclei show as shrunken round little masses, which
stain intensely with the hematoxylin, and no remains of their
cytoplasmic granules are seen. The size of the splenocytes has
much diminished, and has returned nearly to the size seen at
the early stages. A characteristic one, with a dozen little dark-
stained dots within it, measured 18 K 15 uw. Others, in which
the digestion of the cells has not advanced so far, are of larger
size. In some cells, in addition to the shrunken remains of the
nucleated cells, are seen yellowish granules, derived from the red
blood corpuscles of the pigeon. The polymorphonuclears are
few in number, and no myelocytes remain. |
DISCUSSION OF OBSERVATIONS
In considering the results of these experiments there are several

variables, which have to be taken into account. First, in the
washing out of the spleen, the perfusion is more successful in
removing the contents of the vascular channels in some cases
than in others. However, as mentioned before, the result is
never complete washing-out, and by studying various areas in
the sections one can reach definite conclusions as to the relative
number of the different types of cells present. A second and
even more important point is the difference in amount of reaction
which the bone-marrow gives. In some animals apparently
great numbers of cells are released and in others relatively few.
This may depend upon the age of the rabbit, but many other
factors must be considered before a solution can be arrived at.
The outstanding features of the experiments are: 1) the definite

cycle which can be followed in the phagocytic activity of the
splenocytes, and, 2) the bone-marrow crisis occurring as a reaction
to the hemolysis of the pigeon corpuscles.
In following the phagocytic history of the splenocytes, one
notes that the hemolysis of the pigeon blood is so rapid that few
of these cells are taken up intact. It is true that some are taken
up whole, but as dissolution proceeds, it is more and more the
fragments which are ingested by the splenocytes. At the one-
hour stage, the reaction has already begun, some of the inclusions
being recognizable as entire red blood-cells, others as amorphous
masses derived from the pigeon corpuscles. However, the
splenocytes are not greatly enlarged, and even by the end of
six hours, after an interval of several hours during which the
splenocytes have been surrounded by the disintegrating blood-
cells, their size is not conspicuously increased. Either the out-
put from the cells keeps pace with the intake or the phagocytic
process requires a number of hours for its completion. That
time is a necessary factor in the engulfment stage of phagocytosis
is also indicated in the action of the splenocytes toward the bone-
marrow cells. These first appear in the spleen at the two-hour
stage, and are abundant by the six-hour stage, but it is not until
the twelve-hour stage that any conspicuous number of these
cells are seen within the splenocytes. Thereafter the engulfing
process continues rapidly, and at sixteen hours, practically all
splenocytes are laden with these cells.
As the result of the ingestion of myelocytes and polymorphonu-
clear cells, the size of the splenocytes greatly increases. An
average size is 25 & 20 uw, but many are larger than this, and the
example shown in figure 5 measures 55 xX 23.4 uw. In this cell
at least twenty bone-marrow cells can be counted within its
cytoplasm in a single 4 u section, so that one can judge that its
total capacity must be several times this number.
After the stage of engulfment comes the digestion of the cells.
First the pseudo-eosinophilic granules lose their characteristic
staining capacity, although the nuclei still show their original
form. Gradually, however, the nuclei begin to shrink, and
apparently the polymorphous nuclei separate into small pieces.

These shrunken nuclear remnants stain more deeply than do
the original nuclei. The twenty-one-hour material shows spleno-
cytes containing some cells which still show red-stained granules,
and other cells in which the granules are no longer apparent.
At the forty-eight-hour stage most of the inclusions have been
reduced to the condition of fine dark-staining granules, while
others (fig. 6) may still retain the original form of nucleus. Those
splenocytes (fig. 6) which contain cells which have been altered
but little, are usually larger than those in which the digestion
of the cells has advanced farther. Indeed, many of the spleno-
cytes are now but little larger than some found in the normal
spleen. They are, however, quite conspicuous, on account of
the intense staining of the nuclear remains within their cyto-
plasm. The digestion of the cells by the enzymes of the spleno-
cytes is evidently a process which requires considerable time.
At the twenty-one-hour stage no free myelocytes were found,
and relatively few polymorphonuclears, and it appears that the
splenocytes even at the sixteen-hour stage, were already well
filled. So it would seem that the cells ingested at sixteen to
twenty-one hours had not been completely digested by the
forty-eight-hour stage.
These experiments indicate that the splenocytes can ingest
material from a very small size to cells 16 x 10 » in dimension.
They also readily take up the ultramicroscopic aggregates of
trypan blue. So itis apparent that they have a very wide phago-
cytic capacity as regards size of object to be ingested.
Viewing this in relation to their activity toward effete red
blood corpuscles of their own circulation, one must conclude
that they are capable of taking up either entire red blood-cells
or fragments of them. In a recent paper, Rous and Robertson
(17) have found that phagocytosis of entire red cells is frequent
in normal dogs, rats, and guinea-pigs, slight in man, the rhesus
monkey, and many rabbits. This phenomenon is practically
absent in cats and in some dogs and some rabbits. In these latter
they find that fragmentation of the red cells occurs, and they
question how the fragments are disposed of. Judging by the
range of activity of the splenocytes, one must conclude that the

splenocytes are just as capable of ingesting schizocytes as of
taking up the entire red blood-cells and that phagocytosis must
still be regarded as a sufficient general explanation of normal
blood destruction.
Cary (715) studied the method of disposal of bovine corpuscles
when injected into the blood stream of the rabbit. His method
of study was by applying the reaction for iron. He found the
injected corpuscles to be taken up by hemophages, which he
indicates are fixed cells. In contrast to these results, we have
found free-swimming cells to be the most active phagocytic
agents in the spleen. Muir (’02), studying the rabbit’s spleen
in infections, found non-granular free cells containing as many
as twenty or more red corpuscles and a half-dozen leucocytes in
various stages of disintegration. This is a reaction comparable
in many ways to that which we have been following.
Muir also observed myelocytes in the splenic pulp, in infectious
conditions, especially when severe. The greatest numbers he
found in cases of variola. Once or twice he observed them
undergoing mitosis. The bone-marrow crisis is evidently the
result of the action of the toxic material set free in the hemolysis
of the pigeon blood. That it begins between one and two hours
after the injection is indicated by the myelocytes being seen in
the spleen at the two-hour stage. This time coincides with that
at which the crisis of white cells takes place after acute hemor-
rhage, and it is to be noted how promptly the spleen reflects the
bone-marrow crises. By four or five hours after injection of
the foreign blood, great numbers of the bone-marrow cells have
been caught within the spleen. These are detained within the
cavernous blood-spaces, and finally ingested by the splenocytes.
SUMMARY
The cycle of changes associated with the phagocytic activity

of the splenocytes of the rabbit has here been followed.
When washed pigeon corpuscles are injected intravenously,
they are rapidly hemolyzed. ‘The hemolysis of the pigeon blood
results in the liberation of great numbers of bone-marrow cells,
mature and immature. These are caught within the spleen, and
quickly begin to be ingested by the splenocytes. The spleno-
cytes grow with their increased contents, until at the sixteen-hour
stage they reach a very large size. As many as twenty cells are
visible in a 4 uw section of a splenocyte, measuring 55 X 23.4 u.
As digestion proceeds, the splenocytes become smaller, and at
the forty-eight-hour stage they are much reduced in size, some
not much larger than normal.
Of the products of hemolysis of the pigeon blood after a single
injection, little remains within the organs of the body and most
is excreted through the kidneys within sixteen hours. In the
spleen, the splenocytes have an increased amount of iron-con-
taining substances, while the endothelial cells show, for the most
part, very little. The comparatively small results seen within
the phagocytic cells follow from the rapid reduction of the foreign
blood corpuscles to particles of a very small size, and from the
short time in which these fragments remain within the blood
stream.
In this special experimental procedure, where cells and cell
fragments are the stimulus to phagocytosis, the splenocytes are
the first to act, and they continue to act as the main phagocytic
agents.
LITERATURE CITED
Cary, W. E. 1915 The fate of foreign erythrocytes introduced into the blood-
stream of the rabbit. Journal of Infectious Diseases, vol. 17, pp. 432-6.
Heinz, R. 1901 Uber Blutdegeneration and Regeneration. Ziegler’s Beitrige
zur Path. Anat. u. Allgem. Path., Bd. 29, S. 299-404.
Hunter, W. 1901 Pernicious anaemia. London, Griffin & Co., 1901.
Moir, R. 1902 The reaction of the bone marrow and other leucocyte-forming
tissues in infections. Transactions of Pathological Society of London,
vol. 53, pp. 379-405.
Paton,D. N., AND Goopatt, A. 1903 The spleen in relationship to the process
of hemolysis. Journal of Physiology, vol. 29, pp. 411-439.
Rous, P., anp Ropertson,O.H. 1917 Thenormal fate oferythrocytes. Journal
of Experimental Medicine, vol. 25, pp. 651-666.
PLATE 1
The outlines were first drawn with the aid of a camera lucida at a magnification
of 770 diameters. In the reproducing of the drawings the illustrations were
reduced to 575 diameters. The sections from which the drawings were made
were 4u thick and stained in hematoxylin and eosin.
1 Normal spleen of rabbit, showing several splenocytes with inclusions. Spy,
splenocyte with single inclusion, colored with blood pigment. Sps, splenocyte
with several vacuole-like areas in its cytoplasm. Sp3, splenocyte with inclu-
sions of polymorphonuclears.
2 Spleen of rabbit, six hours after injection of pigeon corpuscles, showing
myelocytes in the vascular channels. Sp, splenocyte with inclusion colored with
blood pigment. B, extracellular mass of blood pigment. My, myelocyte show-
ing stage of mitotic division, with chromosome-like masses of chromatin. My,
myelocytes with pseudo-eosinophilic granules.
3 Spleen of rabbit, six hours after injection of pigeon corpuscles, showing
myelocytes (My), pseudo-eosinophilic polymorphonuclear leucocytes (P), mega-
cyte (M), splenocyte (Sp), and blood pigment, both intracellular and extracellular.
4 Spleen of rabbit, sixteen hours after injection of pigeon corpuscles, showing
three splenocytes with cellular inclusions. N.Sp., nucleus of splenocytes.
5 Spleen of same rabbit shown in figure 4. Shows a splenocyte greatly en-
larged by reason of its numerous cell inclusions. N.Sp., nucleus of splenocyte;
R, reticulo-endothelial cells.
6 Spleen of rabbit, forty-eight hours after injection of pigeon corpuscles,
showing two splenocytes with inclusions. The splenocytes (Sp) are now becom-
ing smaller, as the digestion of the phagocytosed cells approaches the end of the
process.
450
PHAGOCYTIC ACTIVITY OF THE SPLEEN PLATE 1
WILLIAM H. F. ADDISON
XQ
451
SUBJECT AND AUTHOR INDEX
CTIVITY. Histological study of the Cee cells of the frog’s pancreas. :
spleen of the rabbit under heightened Studies on the
phagocytic Grandulosa and theca interna. On the origin
Activity of its extensively vascularized con- of the corpus luteum of the sow from
nective tissue. The histology of the OGTR error, Se ee
umbilical cord of the pig, with special ref- Groove and origin of the interventricular
erence to the vasculogenic and hemo- septum. The development of the cardiac
poietic loop in the rabbit, with especial reference
Appison, Witu1AM H. F. Histological study to the bulvoventricular...................
of the spleen of the rabbit under height- Growth, and fate of osteoclasts and their rela-
ened phagocytic activity tion to bone resorption. The origin
Albino rat. Quantitative studies on the Growth of the skeleton of the albino rat.
growth of the skeleton of the... 237 Quantitative studies on the............... 237
Argy, Lesutig B. The origin, growth, and
fate of osteoclasts and their relation to |S nbs FRANK Bratr. The ontogeny
bone resorption and phylogeny of the sternum 41
LOOD in Bufo halophilus Baird and Hemopoietie activity of its extensively vas-
Girard. The early histogenesis of cularized connective tissue. The his-
tology of the umbilical cord of the pig,
Bone resorption. The origin, growth, and wath special reference to the vasculogenic
fate of osteoclasts and their relation to....
Bufo halophilus Baird and Girard. The Bastowencas of the blood in Bufo halophilus
early histogenesis of the bloodin.......... 209 Baird and Girard. The early..
Bulboventricular groove and origin of the in- Hypophysis cerebri of the California ground-
terventricular septum. The _ develop- pauarel: Citellus- beechyi (Richardson).
ment of the cardiac loop in the rabbit, Ceasers HOR EA ates ieee eee 185
with especial reference to the 29
ct
Agee On the origin of the corpus
RDIAC loop in the rabbit, with especial luteum of the sow from both granulosa
reference to the bulboventricular groove andithecases was lees tees e oe eee 117
and origin of the interventricular sep- Interventricular septum. The development
tum. The development of the of the cardiac loop in the rabbit, with
Cells of the frog’s pancreas. Studies on the especial reference to the bulboventricular
fedEroVo
hu Wen ey EE NGA cn art, eis Bee Rene grooveland originiof the. ...+-sacsseecceee 29
Cerebri of-the California ground-squirrel,
Citellus beechyi (Richardson). The ORDAN, H. E. The histology of the
DY POD MY Sisiancan cicarte SENET elas evevers sb" 185 umbilical cord of the pig, with special
Connective tissue. The histology of the reference to the vasculogenic and hemo-
umbilical cord of the pig, with special poietic activity of its extensively vas-
reference to the vasculogenic and hemo- cularized connective tissue.............-..
poietic activity of its extensively vas-
cularized EWIS, Frepreric T. The course of the
Cooper, Harotp J. The hypophysis pecebn Wolffian tubules in mammalian em-
of the California ground-squirrel, Citellus bryos
beechywGkichardson))ceeeeeeee ieee... 185 Lintiz RatpH Dovueaty. The early histo-
Cord of the pig, with special reference to the genesis of the blood in Bufo halophilus aa
vasculogenic and hemopoietic activity Baird and Girard
of its extensively vascularized connective Loop in the rabbit, with especial reference
tissue. The histology of the umbilical.... to the bulboventricular groove and origin
Corner, GeorcGe W. On the origin of the of the interventricular septum. The
corpus luteum of the sow from both development of the cardiac...............
granulosa and theca interna............... Luteum of the sow from both granulosa and
Corpus luteum of the sow from both granulosa theca interna. On the origin of the
and thecainterna. On the origin of the.. COEDUS ese tte are ciaoaee caeieleoins 117
ONALDSON, Henry H. Quantitative AMMALIAN embryos. The course of

studies on the growth of the skeleton the) Woltianitubules im’. o- sc... cc
of the albino rat Murray, Henry A., Jr. The development
of the cardiac loop in the rabbit, with
MBRYOS. The course of the Wolffian especial reference to the bulboventric-
tubules in mammalian ular groove and origin of the interventric-
UI AISe DUI Meets eeicc niemioni erin a cine 29
ATE of osteoclasts and their relation to
bone resorption. The origin, growth, NTOGENY and phylogeny of the
CRG lacsvein ROC Rote CE oe ao ENE sternum. The
Frog’s pancreas. Studies on the glandular Origin, growth, and fate of osteoclasts and
Cells otghe Meron vs asictaleata oetcns cia crave siete their relation to bone resorption. The.... 315
453
454 INEDX
Origin of the interventricular. septum. The Skeleton of the albino rat. Quantitative
development of the cardiac loop in the studies on the growth of the ............. 237
rabbit, with especial reference to the Sow from both granulosa and theca interna.
bulboventricular groove and.............. 29 On the origin of the corpus luteum of
Osteoclasts and their relation to bone resorp- the
tion. The origin, growth, and fate of..... 315 Spleen of the rabbit under heightened phago-
cytic activity. Histological study of the. 437
ANCREAS. Studies on the glandular Squirrel, Citel us beechyi (Richardson).
cellsiofutheuiroristeeeaeey.c
>ce cece ies 347 The hypophysis cerebri of the California
Phagocytic activity. Histological study of VOUT Ga). othe tteiian tae rr ae cee a See 185
ake spleen of the rabbit under height- Sternum. The ontogeny and phylogeny of
NEC eee Mees one ewwiore ciejajoamneniae 437 PHO re crear BA Naetee seh oe co ae eee 41
Phy lege of the sternum. The ontogeny
Nyce Pe ee sD ACIne oT eee Soren 41 HECA interna. On the origin of the
Pig, with speéial reference to the vasculogenic corpus luteum of the sow from both
and hemopoietic activity of its exten- gstanulosa and js. jiiesarenssiets ate cee 117
sively vascularized connective tissue. Tissue. The histology of the umbilical cord
phe histology of the umbilical cord of i of the pig, with special reference to the
LANG OEE le AW bic no RR SA DEEa GOD obra ooccis vasculogenic and hemopoietic activity
of its extensively vascularized connec- '
ABBIT under heightened phagocytic VO ssinisse cen nace sachsen eicee se ties AEE
activity. Histological study of the Tubules in mammalian
Bpleeniol CHE es <5. n:0.0 5 erersaisisiare
piste nisletaiejec 437 course of the Wolfian-:..--...0sne cee ene.. 423
Rabbit, with especial reference to the bulbo-
ventricular groove and origin of the inter- Winseeson cord of the pig, with special
ventricular septum. The development reference to the vasculogenic and hemo-
of the cardiac loop in the................. poietic activity of its extensively vas-
Rat. Quantitative studies on the growth of cularized connective tissue. The _his-
the skeleton of the albino.. 237 tologyolthe sects: sce eee cone one 1
Resorption. The origin, growth, ‘and fate
of osteoclasts and their relation to bone... 315 ASCULOGENIEC and hemopoietic activ-
ity of its extensively vascularized con-
AGUCHI, 8. Studies on the glandular nective tissue. The histology of the
cells of the frog’s pancreas............... 347 umbilical cord of the pig, with special
Septum. The development of the cardiac reference to'the/::0), 0% sh cee pores Aas:
loop in the rabbit, with especial reference
to the bulboventricular groove and origin OLFFIAN tubules in mammalian
oftheanterventricular...-.. 0... cease oe 29 embryos. The course ofthe ......... 423
hie
Ses on
i) fi
Dh wit Bn
ia
NAAM
beyts
Ay
Arert
iH
‘sis
ilI Wan
HO ie Seria
5 WHSE 02078
can
4)
wy
=
tert
feet
ttrert
crete
iigsetoaees
ener
tieiyeds
Le
U
giyer
rie
OOO
ise
iu
PITTS TS
SSCS ee
SSeS
Sharemepens
ee
ee
eS
=
9
=.
SS
Se
‘ SOE: ES SSSSSe sss S

ES SEEESEE
aot
SSS = : 3
SRESSSasSSs==
SELL IIa
i
li
f
mee
=
i it
= ae
ES=s2 : Sse
eee ISS =
=SSSs-SossSs
SSE SESE s ey:
SESE SS =
Soe See S mh
: =
Say Ses SSS
cs
esbewStatoetuncru sta
tat
=5aes
SSSS=SS255S53
SS ie Se eS Sees
SSSEee
SESS
SSS SSE:
SREP
EE eee ES
SESe
Besse saseSse
LES ESTES
Rest ceSe
Sapna
Seek eat erasae
ea tere ae
COSAESCEEENS

American Journal o 26 Wist

Uploaded by

Copyright:

Available Formats

American Journal o 26 Wist

Uploaded by

Document Information

Original Description:

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

American Journal o 26 Wist

Uploaded by

Copyright:

Available Formats

Se

Cuarues R. BARDEEN G. Cart Huser J. Puayrain McMurrica

Henry H. DoNALDSON GEORGE 8S. HUNTINGTON ’ GrorGe A. PIERSOL

Simon H. GaaGeE Henry Mck. KNowir, SECRETARY

THE WISTAR INSTITUTE OF ANATOMY AND BIOLOGY

AY nis aed J Wieepiagali ae ae

Pa tO CRD ets py Whe? a C

No. 2. NOVEMBER, 1919

No. 3. JANUARY, 1920

o) Sel, ihre cular,wit

BY THE BIBLIOGRAPHIC SERVICE, JULY 7

THE HISTOLOGY OF THE UMBILICAL CORD OF THE

with a brief description of the comparative histology of this

THE UMBILICAL ARTERIES AND VEIN

an internal elastic membrane, but several muscle layers beyond

The covering ectoderm constitutes a stratified epithelium of

The yolk-stalk remnant of one of the two practically full-term

twelve cells occupy the diameter of each of the two elements.

The allantoic duct (fig. 1, All.) is shown highly magnified in

THE CONNECTIVE TISSUE

The character of the connective tissue varies in different

themselves in line with the capillaries and ultimately become

In the description of the mode of vasculogenesis illustrated

tiated into a hemoblast, and subsequently differentiated an

cell, young endothelial cell, hemoblast, erythroblast (‘megalo-

f may be regarded as a later stage in the intracellular eryth-

From the foregoing description it will be clear that the con-

the latter interpretation would seem to be the correct one; that

The method of erythrocytogenesis here described for this

The matter may be summed up with figures 15, a, b, and e,

is only when such a cell becomes inclosed by endothelium that

THE AMERICAN JOURNAL OF ANATOMY, VOL. 26, NO. 1

thrown into the surrounding mesenchyme, as in the allantoic

The results of this study of the umbilical cord of the pig,

DaANcCHAKOFF, VERA 1918 Cell potentialities and differential factors con-

2 Photomicrograph of the ectodermal covering of the cord in the region,

4 Photomicrograph of allantoic duct. The lining epithelium is thrown into

< a ego OLE re

El desarrollo del asa cardiaca en el conejo, con especial mencion

E] autor ha estudiado varios embriones de conejo para deter-

THE DEVELOPMENT OF THE CARDIAC LOOP IN THE

The process of union of the two lateral cardiac vessels to form

some conditions are to be recommended. For the finer details,

‘The valuable study of the development of the pericardium in ferrets by

THE AMERICAN JOURNAL OF ANATOMY, VOL. 26, No. l

with a right and left shoulder; d) the restricted portion between

MIDDLE CARDIAC PLATE AND INTERVENTRICULAR SEPTUM

In figure 3 observe the middle cardiac plate connecting the

? As Mall says, it is more correct to speak of the downward growth of the

growth in thickness of the ventricular wall is first manifest on

THE BULBOVENTRICULAR GROOVES

grooves are horizontal. The left groove is already assuming a

The processes in the early development of the rabbit’s heart

Bremer, J. L. 1912 Development of the aorta and aortic arches in rabbits.

La ontogenia y filogenia del esternon.

Existen en la literatura tres teorfas diferentes sobre el orfgen

THE ONTOGENY AND PHYLOGENY OF THE STERNUM

TWELVE PLATES (FORTY-NINE FIGURES)

The shoulder-girdle complex presents some of the most fasci-