UC-NRLF

llllllulilllll
ill

ill liliil

ii

tab

bbfi

APPLIED PERSPECTIVE

^gft^^^gl

^H
''

'

^^^fcCwr!^f^>^-yf

'.

^^^^^^H ^^^^^^H ^^^^Bv;.'

^^^^^^^^H ^^^^^g'-

^^^SbESI^^>/;X'^;'
il;^;'.
.

^^^H ^^^H ^^^^^f ^^^^^^^^H

^iBs^g^o'
'

.-'f:^

^^^^^^MiV'.S..\^

;';;''"

^HHHj |HH|;-'':\~.
HB:;-:
,.':^.';;^

:.;'.>:'

Smm^'i
^^^:<""'-^^ "--.
,',/".

.,''

./

'.':.:'
.

^P'

.', -1
_

*/vi.^;^>:J'^;-.^_V

APPLIED PERSPECTIVE
FOR ARCHITECTS AND PAINTERS
BY

WILLIAM

P.

P.

/LONGFELLOW ^I

Peruzzi was a Painter so learned in Perspective that, studying the Proportions of ancient Columns to draw them in Perspective, he grew Serlio enamored of these Proportions, and gave himself up to Architecture.

BOSTON AND NEW YORK HOUGHTON MIFFLIN COMPANY

QL^t EtoetBtUe precis,

CamiriBp

COPYRIGHT,

I9OI,

BY WILLIAM

P.

P.

LONGFELLOW

ALL RIGHTS RESERVED

Published November, tqoi

L(,

PREFACE
The
ful

practice of perspective

depends not so much on many principles as on the

easy to acquire enough for every-day use
:

varied applications of a few.

It is

the skilfits

draughtsman

is

he who has gained an instinctive readiness of resource that

arise.

him

to

meet a variety of exigencies as they

In writing this book

have had

in

mind

both the every-day use and the skilled use.

In the

first

Part

have given what seemed

necessary to qualify the student for ordinary perspective work, and in the second have
included a series of special problems, to show what trained draughtsmen actually do.

To

these I have added discussions of

some more

theoretical topics

of oblique planes, the tangents

and axes

of circles, tri-conjugate

the perspective vanishing points

itself,

which should be of value

to

students to
to the

whom

perspective

is

a study interesting in

and should smooth the way

more complicated problems

that sometimes occur.

Feeling the paramount importance of what

faculty of perceiving perspective relations in
tion, I

may be
is

called the perspective sense, the

what

before one in nature or in imagina-

have taken pains to

set forth these relations as they

appear to the eye, and to

show how they

influence the painter's or the architect's a sense will find

it

way

of looking at things.

He

who has such

instinct

easy to acquire the various special devices and the

which he needs

for the exigencies of practice.

He who

lacks

it

will

make hard

labor of mastering any but the most obvious processes, and will always be liable to be

blocked

in his

work by an unusual problem.

of saving labor

The processes and

devices are so abundiffers

dant, offer so

many ways

and securing precision, that the adept

from the ordinary draughtsman as the skilled watchmaker from the workman who

knows only enough

to furnish

to oil a Connecticut clock.

Out

of this
to

abundance

have tried

enough
if

for the student's needs,

and to suggest

him how

to increase his

resources

he

will.
full

For students who wish for a

theoretical account of the science of perspective I

:

know no
that this

treatise in English so

good as Professor Ware's Modern Perspective

it

hope

book

will

put the application of

In the discussions
I

into serviceable shape without injustice tp

its scientific side.

have not undertaken to adhere to the rigorous

forms of theoretical mathematics, to get on with the fewest possible postulates, but
rather to assume as

much

as

seemed

safe of the
is

common
yet
I

stock of knowledge.

fair

acquaintance with ordinary school geometry

taken for granted, and the mathematical

;

demonstrations are meant to be clear and exact

have hoped that an

intelligent

458

iv

PREFACE
find profit in the book,
I

and careful reader who does not know geometry may

and learn
it

what

is

fundamental

in it

without unreasonable

effort.

have tried to give

a plan

and development

that suited this purpose.

As

for the problems, the discussions,

I

and

the methods, there are

some things

in

them which

do not see elsewhere

whether

they are really mine, and whether they have any value, other

men can

best judge.

W.
Cambridge,
20th September, 1901.

P. P. L.

CONTENTS
PART
Perspective in Nature.
I

INTRODUCTION
PAGB

The Horizon.

It rises

and sinks with Observer (Plate

I)

Effect of distance on aspect of objects

Perspective distortions

Vanishing Points of

lines

diminution and foreshortening. found by looking direction of lines

in

3
line

Parallel lines have the

same vanishing point

Vanishing Lines of planes.

Parallel planes have the same vanishing

its

(Plate II).

IIorl2ontal planes vanish in the horizon.

All lines in a plane vanish in

vanishing

line.

Projeciion.
Station Point. Picture Plane.

All objects projected by visual rays

Visual
Cone or Pyramid

Picture diminishes as picture plane approaches the eye.

Horizon Line. Postulates and

Centre. Axis. Horizon Line marks the level of the eye Perpendiculars, Ground Ground Line
definitions.
J:

lane,

.... ....
~ .
.

PRACTICE.
Perspective of Standing Screen (Figs.
3, 4)
.

9
10

DisTA.NCK Points To draw two walls with returns at right angles. To draw a horizontal square. Contiguous squares Vanishing points of their diagonals are called Distance Points. Perspective Diagonals

II

...............
far

12

Distance points are measuring points for Perpendiculars. Distance points are
in

Horizon Line, as

from Centre as Station Point

is

in space.

Pavement

of

Squares

set diagonally (Fig. 10)

Trace of a line or plane. Front Lines. Distance points of inclined lines.

All distance poinis lie in circumference about Centre. Gable in Parallel Perspective (Fig. ii). Folding Chess-Board. Inclined horizons. Compression ot distance in perspective

13

Vertical Distance Points

...

14

'....15
16
18

Circle of distance points con-tructively verified.

Problem of Diagonal Squares

Half-distance Points.

(Fig. 15)

Perspective Scales.
Craticulation.

Dividing by squares (Fig. 17) Squares the units of Scales. Scales of width, of depth. Room with Doors and Windows Placing a figure (Fig. Figures in Landscape (Figs. 19-22) Perspective and geometric division

18).

19

20
ai

The Perspective
Plinth Block
(

Plan.

Fig. 24).

Geometric and perspective plans

Problem of the Cross

(Fig. 25).

Use of scales with Half-distance Points.

vi

CONTENTS
Lines.
33

Measurement of
UUique and

Inclined Lines

Measuring Points are distance points of inclined lines. Measuring points in all horizons as far from vanishing points as S. Distance points a special case Circle of measuring points. Folding Screen (Fig. 30). To measure a line without its trace Real measurements in picture plane, proportional on front lines Real measurements by measuring pomts, proportional by vanishing points on proper horizon

24

25 26

Accidental measuring points

Essentia! that line to be measured and measuring line be in

27

same

plane.

Continuing measures on new measuring

line.

Case

of inaccessible

measuring point
at

28 acute angles of triangles

Conjugate Vanishing Points.

Relations of vanishing points of lines at right angles.

right angled at S., with hypotenuse

Station

They are VV in picture plane Point on semicircumference with VV for diameter.

29

Such vanishing points called Conjugate. Problem of Chair (Plate VII)

Perspective Chart.

called Mitre Point Vanishing point of lines at 45 Opposite distance points are conjugate

how found.

30
3'

Selection of vanishing points.

Parallel perspective.
(Plate VIII)

Problem

of

Doorway and Steps

Slope Lines of steps are parallel

Slope
line at

32

their vanishing point

and ground-projection

33

Steps constructed by slope lines (Plate IX).

measuring point of ground projection finds its vanishing point over V . has the same measuring point as its ground projection. Line to an inaccessible Its measuring point vanishing point. Terrace with Steps in three Ramps (Plate X).
.

An inclined

line

-34
35

Inclined Planes.
Descending groui d indicated by horizontals as of architecture or sea (Plate XI)
Pattern
in

....

38

Inclined Plane (Fig. 55).

Roofing of Cottage (Fig. 56)

Dormer
Circles.

in

Roof

(Fig. 57)

39 40

Circles in enclosing squares, by diagonals,

by sub-diagonals

41

Circles or arches in contact.

Circle by diagonal squares.

Circle on front diameter

Chateau of Chaumont
(Plate

42

Arch through wall

(Fig. 63).

Circles to be tested by eye.

XIV)

Horizontal and vertical

circles, inclined axes.

Broletto of Como (Plate XV).

in
St.

43

Construction of Pictures.
Effects of foreshortening

long colonnades crowding distance Teatro Olimpico Court before Peter's (Plate XVI). ... Parallel Perspective. Foreshortening of Perpendiculars. Effects of long Axis Conjugate vanishing points. Vanishing points near and Vanishing points equal distances. Too high horizons. Placing picture plane. Fixed distance between vanishing points. Distance points give longest Axis. Distance points
Tricks of perspective in architecture.
far at

45

46
47

as conjugates

' .

. -

48

Perspective Distortion

Objection that perspective distorts.

Centre.

Drawing constructed for one point of view

Picture
to be

...

49

Graphical and visual foreshortening.

everywhere visually foreshortened toward

Everything stretched out radially from Centre

50

CONTENTS
Spheres in perspective constructed as ellipse with long axes pointing to Centre. This distorts pictures seen excentrically. The shorter the Axis, the greater the distortion. These distortions avoided in sketching. Picture plane constantly changed

vii

ei

Panoramas projected on cylinder, sketches on sphere. Concessions to demands of eye. Spheres drawn as circles

figures sketched. Best remedy

is

arrangement of picture Distortions in photographs. Castle of Urbino (Plate XVIII). Dangers of Parallel Perspective of conjugates badly placed. XVIII).
in

C2

US. Treasury (Plate XIX). Cathedral of Cremona (Plate

Curvilinear Perspective.
Attempt to
der, suits
utilize

both vanishing points of a

line.

changing position of eye.

is

curved when cylinder

ures distorted.

unrolled. Objects preserve Panorama of Paris (Plate XIX)

No

Centre, no measuring

Panoramic perspective, projected on cylinpoints. Straight lines

their relative size, but rectilinear fig-

54

PART

II

Perspective Helps. Projection from plan the natural method. Convergence of parallels the most phenomenon. Vanishing points the surest and most convenient means

characteristic

Far-off vanishing points.

Vanishing points but

Centroliniads. of reach. Convergent

50

lines.

Fractional vanishing points. Square by

g-

fractional vanishing points

Perspective of Oblique Pla.ves. General maxims. Traces and horizons of planes

62

Sub-Axis, Sub-centre, Sub-distance points Conjugate vanishing points in inclined planes.

Oblique planes. Oblique horizons. Sub-centre and Sub-distance points Chart of oblique plane. Conjugate vanishing points. Slope lines and horizontals conjugate. Geometrical relations of vanishing points and measuring points. Practical construction

Perspective chart.

5,
64
.

65 ^^

Circles in Oblique Planes.

General case
Circle

... m

a given square
circles.

special cases.

...07
go

Circle on a given diameter

Concentric

Special Points a.vd Tangents.

General case
Division of circumference.

Symmetrical points.
.

.71
-,
ye

Tangent

at

any point.

Diameters and Axes.

Ellipses symmetrical.

Conjugate diameters. Parallel chords

Axes, Method of Shadows.

Special Tangents.
Vertical tangents.

Horizontal Tangents

Tri-conjugate Vanishing Points.

Rectangular object on inclined plane plane parallel to Horizon Line Box on sloping platform (Plate XXIII).
Relations of tri-conjugate points.

Three points give three horizons, forming sides of a triangle. Visual lines and visual planes meet in solid angle at Station Point Centre is meeting point of three altitudes of triangle. Only one possible Centre for three given points. Triangle always acute-angled and Centre inside it.

76

Triangles of solid angle revolved into picture plane. by two revolved Station Points

Positions of

measuring points determined

'

viu

CONTENTS
77
78

Determination of third conjugate when two are known combined construction Vanishing point of normals Monument with Tri-co.\jugate Vanishing Points (Plate Hexagons and Reciprocal Vanishing Points.

XXIV)

-79
80
.

Hexagons oblique Horizon Line Octagonal Pavements. (Figs. Horizon Line oblique to Octagons parallel
Hexagonal Pavement.
to to
it

Two

reciprocal vanishing points in every horizon (Fig. 123)

(Figs. 125, 126)
. . .

8l

127, 128)

82

Perspective from Elevations (Plate XXVI) Broach Spire (Plate XXVII) Mouldings. Base OF a Pedestal and Console Perspective of a Pediment (Plate XXIX) Entablature in Perspective (Plate XXX)

(Plate

XXVIII)

83 85 86
87

89
9'

Romanesque Arcade (Plate XXXI) Groined Vaulting (Plate XXXII) Vault with Lunettes (Plate XXXIII)

93 96

LIST OF PLATES
FACING PAGE
I.

Vessels on Horizon

II.

Two

Sides of Street

4
10 14

III.

Figs. 3-7, 10, 12.

ii,

IV. Figs,

14,

17.

V. Figs. 18-21,
VI. Fig.
22.

24, 25.

First Problems Gable, Craticulation Scales, Perspective

in

Plan, Cross

18

Figures

Landscape

20

VII. Fig. 41.

VIII.

Fig. 43.

IX. Figs.

Chair Steps and 44, 45. Steps

30

Doorway

32

34
36

X. Fig.
XI. Figs.

50.

52,

Terrace and Steps Slopes and Buildings 53.

38

XII. Figs.
XIII. Figs.

55, 56. 57, 59,

Inclined Planes 63. Dormer, Circles

40
42 44

XIV. Chaumont, Chateau

XV. CoMo, Broletto. Capital XVI. St. Peter's, Front Court

XVII. Figs. 70-77.

46
48
50
52

Towers,

Placing of Conjugate Points

XVIII. Cremona, Cathedral.

Urbino,

Palace

XIX. U.

S.

Treasury.

Panorama

of Paris

54

XX.
XXI.
XXII.

Figs. 98-101. Figs. Figs.

Circles in Inclined Planes 102-107. Circles, Concentric and Divided io8-in. Circles, Diameters and Tangents
112, 113.

66
68

70

XXIII. Figs.

Tri-conjugate

Vanishing Points
Vanishing Points

XXIV.

Fig. 122.

XXV.
XXVI.
XXVII.
XXVIII.

Monument with Tri-conjugate Figs. 125-128. Hexagons and Octagons Fig. 130. Pesrpective from Elevations Spire Fig.
132.

...

74 78

80
82

84

Figs. 133, 134.

Pedestal

XXIX.

Fig. 135. Fig. 136.

Fig. 137.

XXX.
XXXI. XXXII.
XXXIII.

Fig. 138.
Fig. 141.

Pediment Entablature Romanesque Arcade Groined Vaulting Vault with Lunettes

and Console

86
88

90
92

94

96

PART

Pu

Bi

O
b)

X
f-

o
CO
en

APPLIED PERSPECTIVE
INTRODUCTION
PERSPECTIVE IN NATURE
As we look out over sea or plain the earth seems to end in the sharp level line which we call the horizon. This horizon is in appearance just at the level of our eyes, and always remains so. It is a commonly noted phenomenon that, bt;ing on a level with our eyes, it rises with us. If we go to the house-top or to the masthead, it is still at our level. People who go up in balloons say that this appearance is so marked thai the
land or sea beneath them looks like a saucer, the middle being visible far below them,

When we see the whole horizon we see a hoop seen from the centre, of which only a small part is seen or considered at once, and every part of it looks like a straight line. Some curious constquences follow. If two persons separate, one going up and the other down, the horizon seems to the one to rise, to the other to sink, and to follow their motions, so that each may be said to have his own horizon, and to carry it with him. The position of the horizon therefore shows the observer's level he can tell exactly what part of any building or other object before him is at the height of his eye by noting where the horizon crosses it. One who stands on a field of ice covered with
while the edges look as high as themselves.
circle or

hoop about

us,

but

it is

a circle or

skaters

may

notice that

all

their heads, at
at the

whatever distance, being substantially

feet,

at the

level of his

own, form a band

height of the horizon, while their

all

owing

to the

diminishing effect of distance, appear at

heights,

sorts of levels.

This

effect, as of

people

strung on a rope passed through their eyes, while their feet dangle and kick at different

comes too near

the ludicrous for pictoriiil use, and painters instinctively avoid

representing a collection of figures at different distances at the level of the spectator. If one overlooks a plain below him, the head of any person on that plain, near or far,
the horizon, and any head that comes above it must be the head of a he looks out upon the sea, all the vessels upon it will be cut by the horizon at the same number of feet above the water (Plate I). If they are vessels of uniform
will

come below
If

giant.

height,

when he stands
not

as high as the mast-head every mast-head will touch the horizon.

to see a

Yet

it is

uncommon

marine painting
it,

in

horizon, and distant ones rise above

irrespective of iheir size.

which near vessels are below the I have seen a picture

artist.'

painted in this impossible way for the French Salon, and by a clever
1

in what is called the dip of the horizon, due to the curvature of the whereby vessels which are just disappearing over the horizon do seem to rise till their hulls come above it and then sink again behind it but ships in this position can hardly be seen without a glass, and are not subjects for pictures. The exception occurs because, the surface of the water being convex, and not a plane, the visible horizon is not exactly as high as the geometrical horizon of a plane would be this difference is one of the invisibly small things which are neglected.
is

There

a theoretical exception

earth's surface,

APPLIED PERSPECTIVE
The
horizon, being a definite line, determinable even

when

it

is

invisible, invariable

in direction,

and an index

of the relative position of the view

and the beholder,

is

urally the cardinal line or horizontal dividing axis in

pass, as

it

were, of the perspective draughtsman, for

any perspective view, a perspective drawing

the com-

is

nat-

simply a

record of the effect of relative position on the aspect of certain things.

this effect are the angle at

effect of the angle

distance.

which things are seen and the distance. would be comparatively simple if it were not complicated by that of The fact that the farther off things are the smaller they look needs no illus-

The factors of The study of the

tration, but the results of this in their aspect give

room

for study.

Not only do distant

is,

objects look smaller than

if

they were near, but their farther parts look smaller in pro-

portion than the nearer; the apparent pioportions of the objects are changed, that

they are distorted, and the study of perspective

is

mainly the study of such distortions.

windows of a building look smaller than the nearer; the farther side seems shorter than the nearer, so that in most positions the square does not look square. It is not uncommon for persons who are offended by some extreme case of this sort to speak with annoyance of the distortions of perspective, as if it were a human invention, and to be discredited by them. But perspective is in nature. It always distorts; vision, which is simply unrecorded perspective, always distorts, and the distortions, so far from being on the whole faults, are the source of most of the pleasure that we take in the forms of things that we look at. It is they that change the circle into the more graceful curve of the ellipse, and give the varying skyline to a building they give the interest and charm that an artist finds in foreshortening, and substitute variety, life, and endless change in all we see for the monotony and tameness of geometric views. But some of the distortions are becoming and some unbecoming. Extreme distortions are usually better avoided, and in objects whose proportions will
distant
of a square
:

The more

not bear alteration artists are careful to

make

the point of view so distant that the dif-

ference between the nearer and farther parts does not greatly count, and avoid such a
difficulty as

the camera

is

we sometimes see in a photograph where the hands look too large because was too close. The distinction between mere diminution from distance, and foreshortening, which diminution from obliquity of view, is not to be forgotten here. If we stand in front
its

of a square or circle,

plane being at right angles to our line of vision,

:

it

looks like a

square or circle
not change.
If

at

any distance
at
it

it

may grow

larger or smaller, but

its

proportions do

we look

obliquely, the farther parts diminish

more than the nearer,

the lines that are seen obliquely more than those that are seen squarely, and the shape
it is the most characteristic phenomenon of vision distorted. This is foreshortening and perspective. How does this foreshortening take effect, and how is it to be recorded ? Another phenomenon of perspective gives the key to this question. If we watch a bird flying straight into the distance, we see as it gets far off that it scarcely seems to change its position, but at last grows dim, and disappears without

is

apparent motion.
into viev;,
line,

If

we could

turn a telescope exactly in the direction of

it

its flight,

we

could cover beforehand the point where

would vanish, and

in

due time

it

would come

of

In the same way every straight in the axis of the telescope. we please be considered as the path of a moving object, has its place disappearance, which we call its vanishing point, and this point may be found by

and disappear
if

which may

PERSPECTIVE IN NATURE
looking into the distance
to
it.

in the direction in
if

which the

line itself runs

that

is,

parallel

It

follows naturally that

looking in the same direction we change our position,

moves with us, like a star or like the horizon, and is still before us. we look down a long straight street, we see the vanishing point of the house-cornices before us on our side of the street. If we cross the street, it follows and is on the other
the vanishing point
If

side (Plate II).

Two

other interesting consequences follow

First, if several lines are parallel,

they

have the same vanishing point, for they have but one direction, and but one straight line can lead from the eye in that direction to find a vanishing point. In fact, it is one
of the

commonly
point.
:

noticed

phenomena
in

of vision that parallel lines

seem

to

tend to the

same
street
line

VVe see this

the rails of a straight railway, in those of a long straight

we may prove by geometrical reasoning that it must be so. Second, as every common phrase two ends, that is, has two opposite directions, so it must have two opposite vanishing points, a hemisphere apart. When we look in the direction of a system of parallel lines, and see their vanishing point before us, there is
has in

another vanishing point behind us in exactly the opposite direction.

We may

distin-

guish both vanishing points sometimes in the shadows that are thrown across the sky by broken clouds at sunset, one being the sun itself, and the other at exactly the opposite point, just

below the eastern horizon.

We

shall

have

to

consider coupled vanishing

points

when we discuss

curvilinear perspective, but in ordinary problems these are

eliminated, the point of view being considered unchangeable.

It is to

so that

if

it is

be noticed how fast foreshortening increases upon a line as it retires from us, prolonged indefinitely in its distant part, miles will at last be condensed
of a dot.

into the

compass

The

effect

on a plane

is

similar,

and

it

is

this

foreshortening of the distant parts that gives character to

many

landscapes.

extreme Looking

first quarter or eighth covers most of the view, and crowded into a very narrow compass. In a fleet the distant ships seem huddled together, and the waves, which near at hand show a network of open lines, are compressed far off into close horizontal wrinkles that give the water its level look. So in an open meadow a river shows broad where it runs away from us, but where it crosses the view is reduced to a ribbon or thread, and its far-off windings are shut up like a lazy-tongs into close zigzags, or compressed into nothing for the most part, coming into sight only at the turns and so the outlines of masses of trees are condensed as if by pressure into lines almost horizontal. In painting such views the effect of space and distance, and the impression of repose that attends it depend, so far as drawing goes, on rightly seizing this compression and horizontality of lines; and it is one of the difficulties of the tyro to realize their degree and importance. So it is with the sky, which a painter not seldom conceives as a curtain hung up before him, instead of a roof under which he looks. Only the lowermost remote part of the sky is usually painted, and here, when it is clouded, we are looking into an opening, perhaps a quarter of a mile high, that extends away miles into the distance. This effect of a

out over open water we see that the

the far distance
is

retreating roof
It is

is

often striking, though

much overlooked.
except where

not only lines' that vanish: planes also have their definite visual limit beyond
line, unqualified, is

The word
thing that
is

understood to mean a straight

line,

it is

applied to some-

obviously not straight.

APPLIED PERSPECTIVE

point.

which they cannot pass, and the limit of the plane is a line, as that of the line is a The horizon is simply the vanishing line of the (approximately) level surface of
It is the far-ofif

the earth or sea.

eye,

verge of the horizontal plane that passes through the

and as we have seen, it follows the eye as does the vanishing point of a line. As parallel lines seem to meet in the distance, so do parallel planes ; and as the lines apparently meet in their vanishing point, so do parallel planes in their vanishing line. This may be proved by geometry, All parallel planes have the same vanishing line. in the case of lines I shall not stop here for the geoand verified by observation, as Everybody sees that the sea and sky .seem to meet in the horizon, metrical proof. that the walls of a street, the sides of a tunnel, the roof and floor of a long gallery tend to the same thing, and would appear to come together, if they were long enough and
:

straight enough, like the rails of a railway.

So, for instance,

all

horizontal planes

vanish in the horizon.

Naturally any lines that

lie in

a plane must vanish where the plane vanishes

other-

wise they would pass out of the plane.

lines that lie in
it,

plane

may be conceived

to consist of the
lie

just as a

woven

cloth consists of the threads that

in

it.

Where

the threads end the cloth must end, and where the lines vanish the plane must vanish.

Since planes extend indefinitely in

the horizon, extends
all

all

directions, the vanishing line of every plane, like

round the celestial sphere, and is a great circle of it; but like the horizon it appears straight, and only a small portion of it is included in a picture. As the vanishing point of a line is found by looking in its direction, that is, parallel to it, the vanishing line of a plane is found by looking parallel to it, and is the line in that is to which a visual plane parallel to it cuts what is called \he celestial sphere, say, what the sky would be if the earth could be removed and the sky left for us to see, These principles, simple as they are, are often violated in below as well as above. practice, and we may see in the same picture the horizontal lines of houses, which should vanish in the horizon, pointing higher and lower, as if the scene had been warped by an earthquake, or, worse yet, lines at the surface of still water or parallel to it, such as those of piers, are made to vanish above or below the horizon.

Pl. II

All ^H X ''^RBM^^ Mm} t^^^HP^

'*
*

^
H^^^F
r.,

P^nfl Hp9|

^HijK^^^^v

TWO

SIDES OF A STREET

PROJECTION

We see
lines,

all

objects by rays of light that

come from every

is

point of them to the pupil

of the eye.

The paths
meet

of these rays are a series of straight lines,

which we

call visual
is

and

all

practically in a point

which

the position of the eye, and

named
If

the Station Point.

When we

hold up a sheet of glass between the eye and an object,

we see against the glass what is called a projection of the object, making a picture. we could shoot needles from the eye point or Station Point through the glass at all
points, the perforations in the glass

the

limiting points of the object, the paths of the needles would be the visual lines of these

would be

their projections, defining the projection

of the object.
lines
is

If

we connected

the perforations by the proper lines on the glass, these

would outline a picture of the object as it appeared from the Station Point, and this made to do in photographing. Years ago a wanderer appeared at Harvard College who tried to introduce an apparatus and a process by which anybody could sketch without learning. The apparatus was a sheet of glass slipped into an upright frame, and a fixed eyepiece, which was
exactly what rays of light are

simply a wooden standard with a hole in

it.

The

process was to set up the apparatus

before the view, and trace the picture by a pen and ink with which he contrived to

draw
view.

lines

on the

glass,

The contrivance hejped no one

keeping the eye at the eyepiece so as to maintain the point of to draw, but it was a good illustration of the

underlying process of perspective, or of sketching, which

is drawing in perspective. For a true sketch or perspective drawing is only a copy on paper of such a tracing, or projection, as is thrown on the glass. The fixed eyepiece marks the Station Point of the artist, which must not be changed, for the view will change with it ; the glass plate

represents the surface on which the view

is

thrown, which must also be immovable, or

Every, scene or
if

the representation will alter both in shape and position on the paper.

object that

is

drawn

is

projected, that

is,

cast

upon a
is

flat

surface, as

cess as this.

When we draw something

which

before us,

we

practically look

by such a proupon it

as seen through a window, and transfer to the paper on our table, or to the sketchbook on our knee, the projection of the scene upon the glass of the window. This process of projection is in most cases unconsidered, but it is really behind what we are doing, and to recognize it adds clear-

ness to our sight as well as to our understanding. In studying Perspective

we have

to consider that

we draw

things as they are seen against an

imaginary plane set up in front of them, and that our picture only repeats what
is

thrown as by a camera

upon

this plane.

Fig. I represents an object

AB,

say a square of pasteboard, projected in this way as

APPLIED PERSPECTIVE
seen from the Station Point
the pictures of

upon the plane surface C>.

'J"he

projections

ai>

are

ylB ; the plane surfaces on which they are projected are called the PicThe farther off the picture plane from the eye, and the nearer to the object, ture Plane. the larger the picture, till when the picture plane stands against the object, the picture
is

as large as the object,

and

if

the plane passes behind

it,

the picture

is

larger than the

The picture seems to the eye at .S to cover the object in each case exactly. Again, if we should stretch threads from the Station Point to the principal points of an object, as AS, BS, etc., these threads would form the skeleton of a pyramid, or,
object.

as

it is

roughly called, a visual cone, inclosing the object (Fig.

2),

and

if

slips of

mica

or paper were fitted in between the

threads and the lines of the object, as

at
-

ABS, BCS,

etc.,

they would

make
6'

, "^

the faces of the pyramid, of which

,.----*

would be the vertex and

base.
If

ABCD
slice

the

away the farther part of the pyramid by cutting it across, as at abed, we should
have a section
in

now we should

which every point and

line in the object
If
it

line

corresponded
roller

in

its

reduced scale to the

answering point or
abed,

ABCD,

and the plane of abcd\iQ\Ad, be the picture

over the lines of the section

plane in that position.

we could pass an inked

on a sheet of paper the outlines of such a picture as would be projected on the picture plane in that position. A second nearer section at a'b'c'd' would give the smaller picture that would be projected on a picture plane in that position.

we might

print from

Thus it appears that a picture of any object is its visual projection upon a plane set between the object and the eye of the spectator, and that any point A is projected by a visual line AS which pierces the picture plane in the corresponding point a or a' of
the picture.

We may

say then, to
in

sum up

An

object represented

perspective

is

conceived to be projected upon a picture

its

plane by visual lines and planes, converging to the Station Point from

several points

and lines. These visual

lines

and planes form the perspective pyramid or cone which envelops

vertex at the Station Point.

the object, and has

its

The perspective picture is the section of this perspective cone by the picture plane. The size of the picture depends on the position of the picture plane, increasing in proportion as that is set farther from tlie Station Point. Our imaginary picture plane may be interposed at any distance, and will give a series of different projections, all
similar, but varying in size with the distances, just as parallel slices cut

from a cone

give a series of similar sections.

The
please.
lines

picture

we draw
it

is

a copy on paper of the perspective picture at any scale

we

To

construct

we determine geometrically the

intersections of the visual

The most

and planes with the picture plane, by reference to certain fixed lines and points. direct and natural way is to take these intersections from plans or side views (elevations) of the objects and visual cones yet for complicated pictures shorter and simpler means have been devised, as will appear later. To reduce these considerations to practice and put on paper our copy of a scene as
;

PROJECTION
we see
it

projected on the picture plane, we must have

some

fixed line of direction,

some

starting point

by which
which

to place things.
:

natural horizontal axis of our view

ings.

it
it

The

line

is

to represent

have already seen that the horizon is the is universally used as such in perspective drawon the paper we call the Horizon Line, and we

We

draw it at a convenient height, giving room for what we have to put above and below it. This Horizon Line marks the level of our eye in the picture, and that we may have a definite point by which to fix distances to right and left, we choose the point in that line which is directly opposite the eye as we look at the view, and we mark a point for it which fixes the position of the view on the paper. This point we call the Centre. Since the picture plane and the paper which represents it are assumed to front us exactly, the Centre is the point where a perpendicular ' from our eye pierces the paper, and stands for that in which one pierces the picture plane. It is the centre of our

may or not be put in the middle of our picture, for we may choose to show one side of the view than of the other. It will be seen at once that the position of the eye must not change with respect to either the view or the picture, and that to see the picture rightly the eye must be exactly opposite the Centre, at what we have
view, but

more

of

called the

Station Point.

The

line

which joins the Station Point and the Centre,

is

necessarily perpendicular to the picture,

called the Axis.

(This special geometrical

in

use of the word

applied to

be distinguished from the general sense the Horizon Line.)

is to
is

which

it

has just been

horizontal plane

with the floor or level ground

called the

commonly assumed at the bottom of a picture, corresponding when there is such in the view, and upon this plane,
Its interis

Ground

Plane, the positions of objects are conveniently laid out. usually the bottom of the picture, and
is

section with the picture plane

called the

Ground Line, and

are determined.

it is

a useful line on which to take the measures by which positions

is

If

the spectator

conceived to stand on the floor or ground of the

Ground Line marks in the picture the Horizon Line does that of the eye. From the foregoing discussion we deduce the following fundamental postulates The vanishing point of any line is seen by looking in the direction of the line, that is, parallel to it, and so appears in the picture plane where a visual line parallel to that
view, he stands on the
level of his feet, as the

Ground

Plane, and the

line pierces thej)icture plane.

All parallel lines have the

same vanishing
is

point.

The
is,

vanishing line of a plane

it,

seen by looking in the direction of the plane, that

parallel to

and so appears

in the picture

plane where a visual plane parallel to the

given plane cuts the picture plane.

A line

in

any plane vanishes

is

in

the vanishing line of that plane.

line. It
all
is

All parallel planes have the

same vanishing

The Horizon Line

all

at the level of the observer's eye.

in
it

the vanishing line of

horizontal planes,

and

are the vanishing points of

horizontal lines.
:

The fixed lines -and points that belong to the picture are these The Horizon Line represents the visible horizon. The position of the observer's eye is called the Station Point.
1

The

extraordinary misuse of the word perpendicular which

it

is

common makes

it

proper to recall

here that

means

at right angles,

and not upright.

APPLIED PERSPECTIVE
The
point in the picture exactly opposite the Station Point
that joins the Station Point
is

called the Centre,

and

is in

the Horizon Line.

line

The

visual line perpendicular to the picture plane

and the Centre is called the Axis. It is a ; and the Centre is, therefore, the vanIn

ishing point of

all lines

that are perpendicular to that plane.

Lines that are perpendicular to the picture plane are called Perpendiculars. perspective all Perpendiculars are horizontal, and parallel to the Axis.

The

horizontal plane at the bottom of a picture

is

is

called the

Ground Plane.

Its

intersection with the picture

1

called the

Ground
The

Line,*
adjective perpendicular

The word

line, unqualified,

means a

straight line.

means

at right angles

upright lines are called vertical.

PRACTICE
For our
plane
first

blades, open at

problem let us make a perspective view of a standing screen of two any angle (Fig. 4). For convenience' sake we will assume the picture
3)1
s\va\\

PF to
its

be set against the salient angle of the screen, as shown in the plan (Fig.

so that the vertical edge dcf

it

appear

in the picture

plane in

its

real size, being, as

were,

own

perspective,

and

shall be so

shown

in

our drawing, according to the

scale which

we adopt, which

shall be that of the plan.

PPv\
edge

the plan represents the

upright picture plane, as seen from above,

Point,

and acb the plan of the screen with

HH im the Horizon Line
at

^ the
its

position of the spectator,

vertical

the Station
plane at
c.

in the picture

In the perspective (Fig. 4), we assume the level on which the screen stands for the ground plane, and taking a horizontal line for the Ground Line, draw a second horizontal

whatever distance above

to

it is

suitable.
floor,

If

we con-

ceive the spectator

and the screen

be standing on the same level

the Horizon

Line, being at the height of his eye, will be, say, five feet above the ground line, accord-

ing to the scale of the picture.

a vertical line through
it

to the

Then setting the point c conveniently on HH, we draw Ground Line zXf, and up till/^/ is equal to the height x
its

given for the screen.

place and scale.

Then_/(/ will represent the upright edge of the screen in

lines ca

proper

The

and

cb (Fig. 3) at the

bottom

of the screen, being hori-

zontals, will have vanishing points in

HH where visual lines parallel to these lines meet

the picture plane.

parallel to cb
will

From ^

then, in the plan,

to ca,

and

5 F parallel

meeting

we draw two

lines right

and

left,

SV

the picture plane in F' and F, which

be the required vanishing points. We transfer them to Fig. 4 by setting off on the Horizon Line the corresponding distances cFand cV. The lines /FandyF', drawn

to these points, represent the lower

edges of the screen prolonged

till

they disappear in

their vanishing points.

The upper
points,

edges, being respectively parallel

XofVznd/V,

and so may be drawn from d as dV and dV. It remains only to cut off these upper and lower edges at the right lengths. Now the extreme points a and b at the base of the screen in the plan will be projected by visual lines aS and bS, and will appear in the picture plane at a' and b\ at distances a'c and b'c left and right from c. Transferring these distances to the perspective drawing, we get a' and b' on each side oifd. Vertical lines from a' and b' will cut/Fand fV\ dV and at the proper distances from y and d, and will give the vertical edges of the screen. The figure agdhbf is its true perspective outline.

must have the same vanishing

dV

DISTANCE POINTS
To DRAW TWO WALLS ON LEVEL GROUND,
plane and parallel to each other (Fig.
plane, which
5).

with retums at right angles to the picture

Let the walls be assumed to be

in the

same

we take

for the picture plane, as in the plan (Fig. 5 a).

The

return walls

are of equal length.

their

The

own

perspectives.

corners of the walls, a and 6, being in the picture plane, are Suppose that the Centre C comes between the two walls, and

let ^S in

the plan be the Station Point,

SC being

the Axis

let

x represent

the height of

the walls.

GG

is

For convenience of comparison we set the plan directly over the perspective. the Ground Line, J^//, above it at a suitable distance, the Horizon Line, and we

C directly under its projection on the plan. Then the perspectives of a and l>, coming directly under them, can be projected downward upon in a and ^. Measuring off dg and t/t in the perspective equal to x, we have the corners of the walls in
put the Centre

GG

their proper height.

in

The bottom

lines will lines

be right and
right

left

like

manner,
to

will

be horizontal

drawn

and
will

left

return walls are at right angles to the picture plane, their

on GG, and the top lines, from g and A. Since the top and bottom lines, which
be Perpendiculars, and must

we assume
them
these

be

level, like

those of the front walls,

vanish in C.

We may

accordingly draw aC, bC, gC, and

kC;

it

remains only to cut

projected into
to project

off at the right points.

In the plan the extreme points

d andy are

the picture plane by the visual lines

dS

z.nd

fS

at d'

and/'.

We

have only

down upon ^ and/" (which

and give the

is

equivalent to measuring off the distances Cd' and

Cf)

in the perspective lines

aCand hC hy
It will

verticals which will cut off all the vanishing lines

at the right points,

vertical lines at the

ends of the

walls,

completing the

be seen that some of the to cross the perspective, not because they are any part of
perspective construction.

lines of the plan are allowed

it,

but only for the sake of

compactness

in

placing the figures.

When

objects are so situated, like the walls in the last problem, that the principal
it

lines either are parallel to the picture, or are perpendicular to

and therefore vanish

the simplest kind of
lines

in the Centre, they are said to

be

in Parallel Perspective.

This

is

perspective, having only

one vanishing

point, the Centre,

and allowing the

and

surfaces which are parallel to the picture plane to be represented in their true forms

and proportions, and

if

they are in that plane,

in their true size.

A HORIZONTAL SQUARE WITH ITS FRONT EDGE IN THE PICTURE PLANE. Let oa be a horizontal line in the picture plane (Fig. 6), the Ground Line or another, is the Horizon Line and C the in which lies the side ab of the required square.

To DRAW

HH

Centre.

We may

as before use a plan, which, to save space,

of the perspective.

On FF,

the projection of the picture plane,

we make on half the scale we construct the square

abed, half as large as the required square.

in

the perspective vanish to the centre

as in

may be found

and bd being Perpendiculars will The points c and d bC. the preceding problems by projecting them upon FF in the plan,
sides ac

The

C on

the lines

aC and

DISTANCE POINTS
transferring

ii

on ab, and drawing vertical lines to carry them up measure off double distances on the perspective. The to aC and bC, taking care to side cd, parallel to ab, will remain so in the perspective if the construction is correct, for But in the it is parallel to the picture plane, and so only one point c ox d\s necessary. diagonal ad passing through d cuts off the side bd at d, and if we could draw plan the this diagonal in perspective, it would give us at once the point d, through which we might draw cd parallel to ab, and finish the square without projecting the points c and d from the plan. Moreover, if we draw other squares beside the first, as bdig and gikh,

them

to the perspective

This the diagonals being parallel, one vanishing point will give them all in perspective. vanishing point, being where a visual line parallel to the diagonals pierces the picture Transferring it to parallel to ad or bi. plane, is projected on the plan by drawing

SD

the perspective,
sively bg

we

set off

CD

equal to twice

CD

in the plan,

and laying

off succes-

and gh equal to ab, draw the lines aD, bD, and gD. Then ad, passing through The other diagonals will mark the the point d, must verify the previous construction. points / and k, on the lines ^Cand hC, which we shall draw to the centre, they being the perspectives of the sides gi and hk. The farther sides cd, di, and ik will all be in the straight line drawn through the point c or d parallel to ab. By this construction the perspective might have been drawn without any plan at all
for
it was only necessary to determine C and D, besides the line ab, to have all the needed data. The first square might have been determined, like the others, by simply drawing aD, which would have cut off the side bd on the perpendicular bC. The distance CD, being the length of the axis, might have been laid off at once on HH, and

the plan dispensed with.

If

we had

several squares laid together, as in a tiled floor,

all

the diagonals would be

either parallel or continuous, all vanishing together in the vanishing point

D, and

all

the sides, perpendicular and vanishing to the centre, .would be cut off by these diago-

which would be sufficient for the whole construction. Thus in Fig. 7, we may lay on the front line ab any number of equal parts, at the points /, 2,3, 4, etc., and drawing Perpendiculars to C mark off the joints perpendicular to the picture plane. Then, having fixed D, at a distance equal to the length of the axis, we can draw as many
nals,
off

diagonals as are necessary to determine the corners of the squares, or, what
thing, the front lines through b,
c,

is

the

same

which mark the rear sides of the successive lines of squares. A very few diagonals will be enough to determine all the necessary points. At the margins of the picture. Perpendiculars which it may not be convenient to measure in front may still be determined by points and lines that have been already found, as at cD, or by counter diagonals, as at xm.
d,
etc.,

The

point Z>

is

a very important point, for by

it,

in

combination with the Centre, a

considerable number of perspective problems can be solved without other help.

two points Cand indeed fix the essential conditions of the picture, the positions of the Horizon Line, the Centre, and the Station Point; for the distance CD marks the length of the axis, and so the position of the spectator, that is, his distance from the
picture.

The

is

therefore appropriately called the Point of Distance, or

more simply

the Distance Point.

The

parallel lines

aD, cD,

etc.,

and perpendicular

to the picture plane, are necessarily at

being diagonals of squares whose sides are parallel an angle of 45 to that plane

12

APPLIED PERSPECTIVE
to the
lines,

Axis ; and all lines that are parallel to them must also vanish in D. All whether or not they appear as the diagonals of squares, are in perspective called Diagonals, and the distance point is therefore the vanishing point of Diagonals. There is of course another set of horizontal Diagonals like X7n, symmetrical with

and

such

these, running off to the horizon at the left, and there is another distance point, D', on the left in the horizon line, at the same distance from C as D, in which all these Diagonals on the left must vanish. It will be seen by looking at Fig. 7 that any Diagonal aD cuts off on the Perpendiculars

iC, 2C,jC, 4C,

etc.,

the

same perspective length

is, it

that

it

cuts off from the feet of

off j/>,

the Perpendiculars on the horizontal ax, that

on jCit cuts

equal to

ja

cut

on ax from its foot j. squares from the foot 2.

off
its

So

cuts off on
it

On 4C

cuts off

two squares, as on ax it cuts off two four squares, and on ax four squares from
foot,

2C

foot 4

and so on.

If

the sides of the squares are a foot or a yard, the distances cut

off on perpendiculars and horizontal alike represent one one yard, two yards, and three yards.

two

feet,

three feet, or

Therefore

if

we wish

to cut off a required length

on a Perpendicular, we have only draw a horizontal line through its trace,* as at a in Fig. 8, and having measured off on it the required distance to X, draw from x a Diagonal to the
to

convenient distance point D, and the

length ai cut off on
a length equal to the distance ax.

aC

must represent
which and

Note.
xd,y6,

The

geometrical reason for this construction

is

is

clear

from the plan, Fig.

9, in

the Perpendicular at a
etc., are

drawn

in its

due position

at right angles with the picture plane,

Diagonals.

The

distance's ax, ax', are repeated in the distances ab, at/, cut off

by the Diagonals
that
to

xb, x'b',

which being

parallel to the visual line

SD are parallels

make a series of triangles similar SCD, and therefore isosceles, having their sides ax and ab, ax' and at/ equal in each. Fig. 9 shows, moreover, how the Distance Point may be taken on either side of C, at D or D', and how the Diag,

onals drawn from

tances from

y and y', at equal disC with x and x', will give the
.

same points b and V

Hence we say

The Distance
^

Points in the Horizon

C, at dis-

tances from

it

equal to

Line are on each side of the length of the Axis, or as far from it as ^ is.

Any

line that vanishes in

it

them

is

a Diagonal, and will cut off on any Perpendicular

and Horizontal that

Note.
sect in space,

crosses equal distances from the intersection of these.

In measuring by Diagonals

we must be sure that the lines we represent really interand do not merely appear to cross, as they may if one is behind the other. This
'

The

point in which a line pierces the picture plane

is

called

its trace.

DISTANCE POINTS
requires that they be in the

13

same plane, which is secured by drawing them from points in the or such a line as we use in Fig. 8, the Ground Line for instance, same horizontal line, which we may call a Hue of measures. This line is the trace of the plane of the measuring lines and the lines to be measured, which all vanish in the Horizon Line.'

To DRAW

A PAVEMENT OF SQUARE TILES SET DIAGONALLY, with a border equal

of the
is

in

width

to half the diagonal of a square (Fig. 10).

and three deep toward the centre

The bay on the left side is one square wide, picture. The picture plane is for convenience
it,

so set that a small part of the floor

in front of

and

is

projected back against

it

makes no difference measures. The points a,

this

in the construction.
2,

The

line ao is taken

for the line of

j, etc., are set off each

is

the diagonal of the squares, and no plan the Perpendiculars aC, bC, etc.
ao, will divide

way from b at distances equal to needed except for explanation. We draw Diagonals drawn to D, from the points of division on
left

these Perpendiculars at the corners of successive squares, and the division

may be continued from

Horizontal _/)w//
distance point
all

other points set off to the

from

a, at

proper equal distances.

will

lines

through the division points of any of the perpendiculars

If
left of

intersect the Diagonals at the corners of the successive ranks of squares.

the

D' on the
;

C is
is

within reach,

it is

very simple to find the edges of

the squares by the division of the single front line ao, without further front lines or

Perpendiculars
test the

nevertheless

it

well to

draw one
is

set or the other, or even both, to

if

alignment of the corners of the squares, which must be exact

If

the construc-

tion is accurate.

one of the distance points

in

out of reach, the counter-diagonals

are

drawn by connecting
Diagonals

proper order the points where the Diagonals toZ? intersect

the front lines or Perpendiculars.

in the

The
at

mitre lines, as they are called, of the border

lie

at their corners, as

a and

b.

Half of ab gives the width

of the

Perpendicular border, the intersections of

bC and iC
in

with the mitre lines give the

width of the borders parallel to

It

ao.

should be noted that although we have

these last constructions taken our

lines in a horizontal

distance points in the Horizon Line, because

plane, there
is

we were dealing with

nothing in the geometry of the construction that limits these points to

If

that position.

we turn
line,

Fig. 7 about the centre Cit

may represent

lines in an inclined
if

plane, instead of a horizontal one.

The

construction will be just as correct

we

assume an inclined
so that

ax, to begin with.

Note. The only necessary condition is that we take a line parallel to ax for a new horizon, we may still represent a series of Diagonals lying in the same plane with the line of measures ax and the perpendicular aC ; if they are not in the same plane, they will not really intersect those lines (see Fig. 9). The lines are really in the same plane if they all vanish in the vanishing line of the plane, that is, DC parallel to ax, its trace. The line DC, when it
is

inclined, is
'

sometimes called a secondary horizon, serving as

is

it

does the purpose of the

The

trace of a plane, like that of a line,

is

where

it

intersects the picture plane.

The

trace of a line

is

a point, that of a plane

It is

line.

Centre, Diagonal, Perpendicular, printed with capitals, as they are used

of the

important that the student should always distinguish the different meanings of the words Axis, in this book, from the meanings
capitals.

With capitals they have the special meanings here defined, which without them they are taken in their ordinary sense. It is impracticable to discuss perspective problems without using the words in both ways.
same words without
are peculiar to Perspective
;

'

Lines parallel to the picture plane, that

is,

fronting the spectator, are called front lines.

14

APPLIED PERSPECTIVE

Horizon Line in measuring. Fig. 9, again, may be conceived to be a plan made in any inclined and D' then, may be in any and will still be just as conclusive. The distance points direction from C ; they must always be at the same distance from it, and so will all be in the circumference of a circle whose centre is at C, and its radius the length of the Axis of the Any two corresponding distance points, picture. and must lie at the opposite extremities of the same diameter, and the line of measures must be parallel to the line DCU. If we turn the paper so that DD' looks horizontal, we shall perceive the usual relations of Horizon Line, Centre, and distance points.
plane,

Therefore we

may

enlarge our maxims, and say

The

vanishing points of
is C,

all

Diago-

nals are points in the circumference of a circle

whose centre

and

its

radius the

Axis of the picture.

All these points are distance points for the

measurement of Perpendiculars by means

lie.

of lines of measures which are parallel to the diameters in which the points

Distance points vertically over or under the Centre, often called vertical distance
points, are useful in solving

many problems.

The gable

in Fig. 11 is

an example.
is

As

the adjoining elevation shows, the section of wall between

outline of the triple

and b

square, the

window is a square, so is that of the sundial, the upper lights in window are square, the pitch of the gable itself allows it to be inscribed in a square. The plane of the wall is perpendicular to the picture plane, so that all the These facts allow all the measuring to be done by horizontals vanish in the Centre. Starting from A, and making Ab in the picture plane the vertical distance point /?'. Measuring off the height Ab of the a line of heights, we draw the Diagonal AD'. square section of the front up to the base be of the gable, we draw bC, on which the diagonal cuts off the width of the base be. The height Ab repeated gives bd, that of the gable cross diagonals below give its central line, and dC fixes its peak. The line
the
:

of the window-sill
it.

is

set off at a,

and the height, divided into three equal

fix

parts,

above
the

Horizontals to

from the points of division

the top of the

window and

transom, and by their intersections with Ac, the mullions.

The top

of the dial is

measured off at bf, its distance back from b on be from e, and the diagonal eD fixes two corners by which it can be drawn. A verification may be given by measuring off the distance of the lower corner from b on a. horizontal line of measures through b by in the Horizon Line. means of the left-hand distance point at It will be noticed that the vertical distance point D' is useful in this problem

because the lines to be measured

struction
is

lie

together in a vertical plane, so that the con-

best carried out in that plane.

Each

line

separately by one of the distance points in the Horizon Line

to

might have been measured ; but it is more convenient

measure them all together by a common measuring line Ad in their own plane, and this calls for a distance point in the vertical horizon D'C of that plane. The next problem will show the convenience of measuring lines in an inclined plane by a distance point in their inclined horizon. There will be further illustration of these It is understood principles in the discussion of the measurement of inclined lines. that in these discussions horizon means a horizon in the picture, except where there is
special reason for referring to the distant horizon in the view
itself.

The folding
which
is

chess-board. Fig.

in

an inclined horizon.

shows the use of two distance points, one of The front edges of the board are parallel to the
12,

^u

JPla^U

IV

DISTANCE POINTS
picture plane.
is

IS
of the front line ab,

We

assume the centre

C,

and the position

flat,

which

and draw the inclined line be, also a front line, at the proper. angle for the tilted half. From we measure off four squares on ab toward a, and four toward c on be. From the division points and the extreme points a and b we draw perpendiculars to C. From C we mark off on HHihe length of the Axis to the distance point D, and drawing through b an inclined horizon
(5

the near edge of the squares on the half that lies

ITH'

parallel to

be,

lay off an equal distance to D'.

The Diagonals
7.

in

the plane of

the horizontal leaf vanish in

to repeat for those

D, and those

in

the inclined leaf in D':

we have only

two planes the construction of Fig.

is

is perpendicular, or normal, to the picture plane, like the Horizon Plane, and whose horizon passes through C. The horizons of all normal planes, and these only, pass through the centre, so that distance

Note.

The

inclined leaf of the chess-board

here taken in a plane that

points can be used for constructions in normal planes,

and

in

them

only.
is

The

striking compression in depth of the squares as they recede

only a particular

One does not extreme disproportion between the distance and the foreground, and it may surprise the student to learn that to an observer standing on the Ground Plane the middle line between and Hff represents a distance only just as far
case of the general
at first expect the
in the introductory chapter.

phenomenon described

GG

behind the picture plane as the Station Point

point
is

is

in front.

Thus

in Fig. 7,

although the

is

in

the remote horizon,

if

the line tnn

J

halfway from
will

GG

to

HH,

a person standing

at

be as near the picture plane as the specis

from the side view of Fig. 13, in which the right triangles CSo and oo'P being similar, if o is halfway between /'and C, .SCmust
tator.

This

clear

Fe^./J
^~^^
^"~"~~-^'

be equal to Fo'.

Here

PF represents
0.

the picture
<3

plane seen edgewise, the spectator stands at a on

the ground plane with his eye at the station point S, and
0' is

\p
the point in the ground

plane which

is

projected at

Note.
reasoning,

The application of may be verified by

distance points in the inclined horizons, justified by geometrical construction in Fig.

14.

Let

Cbe

the Centre, and

C",

D a distance
it

point in the Horizon hint

HH.

Describe a circumference about

passing through/?.-

will contain all the distance points.

Draw

the Perpendicular

aCirom any

point a, and laying

measures parallel to HH, any distance ax, join xD, which cuts off ab proDraw any inclined horizon D'CH' and on a new line of measures parallel to it lay off the same distance ax' Then a line from oc' to D' will pass through the This will be found true for every point same point b, again cutting off ab equal to ax or ax' in the circumference used with its appropriate horizon and line of measures.
off

on a

line of

spectively equal to ax.

To DRAW TWO EQUAL CONCENTRIC SQUARES SET DIAGONALLY

side ab given or constructed as a front line,
to C,

Having the (Fig. 15). and drawing the Perpendiculars be and ad

and the Diagonal bd or ae

ac
fix

to

D' below

C ox D

above,
o,

we

set out the square abed.

vertical

The Diagonals bd and

which are two angles
right triangle
e

the perspective centre

of the

and the
part

through
ob, is

it

on

and g
b'

second square.

The

oe,

equal to
ob,

a half-

diagonal of the square.

Construct

/^'

equal to mn, the real length of

by an

isosceles
oe.

mbn ;

project

on

oe

by the Perpendicular b'C and

lay off og equal to

APPLIED PERSPECTIVE
A Diagonal from e to D' below gives yj and one from g gives h. The second square is completed by joining eh and gf, or, for verification, by drawing Diagonals from g and above is in reach. This figure to Z* above, which should pass through -4 and/, if which is the overlapping portion of makes an eight-pointed star, and also an octagon

If

JPiff /J"

the two squares.

It will

be of use

in putting circles into perspective.

A
in

series of octa-

gons, such as
as

is

often seen in mosaic pavements,

may be

easily

drawn

a similar way,

by using Perpendiculars and Diagonals. It often happens that the distance point comes too far on one side, and is practically In that case it is important to find some substitute for it. inaccessible. Ordinarily if measure on the perpendicular aC, Fig. i6, we draw a front \me.ao, and markwe would ing off on it the required distance ax draw _^ y^ xD, cutting off ab, which is perspectively But if we take Df ^ halfway beequal to ax. tween C and and draw Dj^bxj^, we shall have, as the similar triangles show, ax halved like CD, That is, if we had taken axf ^ equal to half ax, and drawn xl ^Dj ^, that line would have passed through b. Hence we may take a point Dj ^ halfway from C to D, and by using half measures instead of whole, get the same division of any Perpendicular that we should get from is called a half-distance by whole measures. The point point. If CD is very long, we may use a third or a quarter, or any other fraction of it, and then of course the measures on ao are divided in the same ratio. These fractional distance points must not be confounded with the vanishing point of the Diagonals, which is always D. This device is of value, for, as a little experience will show, perspectives in which the Axis is short, and the distance point correspondingly near, look strained and unnatural, and for good pictorial effect it is desirable in most cases to use a rather distant Station Point, that is, a long Axis, and so a remote distance
shall see later,

we

point.

DISTANCE POINTS
It is

17

understood that a
will

is

the trace of

aC

otherwise ao will not be in the picture

ab,
it

plane,
length.

and though ax
This point

be perspectively equal to

will

not represent

its

true

will

be further discussed.
usually that on the opposite side of the trace a from C.

The

student will soon find by experience that of the two distance points in any horiis

zon the most convenient

PERSPECTIVE SCALES
If we turn back to Fig. 7, we shall see that the divided lines make scales of width and depth by which we might place an object in any desired position on the floor. For instance, if the divisions indicated were a foot square, an object at/ would be three feet back from the front line ax, and three to the right of aC. We may imagine any

horizontal surface divided into squares of a yard, a foot, or what not, and thus fur-

nished with a network of scales by which any things that lay or stood on
plotting the positions of the objects

it

could be
at right

put accurately into perspective position by constructing the network in perspective, and

upon

it.

An

analogous system of lines

angles has been used by painters in laying out their work, from the mural painters of
ancient Egypt to the scene painters of
in perspective,

modern

theatres

and the same system, applied

may be used

for constructing the representations of varied

floor.
is

and compli-

cated figures, such as garden-beds, or the pattern of a mosaic

This method,

which the French

call craticulation, is

illustrated in Fig. 17,

It

and
is

serviceable when-

ever irregular or freely curved lines are to be represented.

of course not indis-

pensable that the network should be of lines that are parallel or at right angles to the
picture plane, but this
It is
is

most convenient.

not necessary, however, that the horizontal plane should be actually divided

We may conceive the front line ax alone to be divided into a scale of measures parallel to the picture plane, and the Perpendicular aC into a scale of measures back from the picture plane. That is, ax may serve for a scale of width, and aC
into squares.

as a scale of depth and instead of first graduating the scales it is enough even to set off on them the particular measures that we need for placing our objects. We may add to them a vertical scale by which heights also may be fixed, and so have a means of
;

placing objects of three dimensions. In Fig. 18 the door and windows of a

the

room

are placed by this means.

:

The The

sides of Line.

room

are parallel and perpendicular to the picture plane

AB

is

the
it.

Ground

AB represents
accordingly.

the picture plane, or a plane of measures parallel to

lines of

the retreating angles of floor and ceiling are Perpendiculars, vanishing in C.

The

height and width are ^et off on

AE and

AB, and

the outline of the view completed

The

depth, 16
first

The
from

positions of the

feet, set ofif on and measured to Z), gives the corner T'". window, three feet wide and set back one foot and a half
;

AB

the door, three feet wide and set back seven feet
first,

which balances the

are fixed in the

same way.

Their heights, set

and the second window, off on A, are

projected by Perpendiculars, vanishing in C.

mm'

represents the height (six feet) of a

man whose

from the right hand wall and six feet back from the front, is determined by the same means. His height, which must be measured in the picture plane, like the rest, may be found in two ways. It may be measured on B6 and carried back on a Perpendicular 6C, which will be everywhere six feet above the floor, and the point m be projected on .5Cby the horizontal front line mn, then upon 6C by the vertical
position, four feet

i1

''Aw \\
/

\^

^"^^

\4

4./

^^

\\

(k

\^^
h
-'^
'
1

b/ y
1

1'.
-'
.

I,

/-^
*//
I
<

t-+-

JItr-

.1 li

%^
Ti^.J^S
I

If

^'

GC^

i\ .^:

- .-----.

iV.i

^Tai

ri3

:^

.^4

:^'^-^*

^^--.
1,\
>~'
)

..

:'/<:'
/
;?

^_E
Fl^cte.

\^

::

i-.^v
'

4S--

21

*^/

"^

i^j "'

N,

-I
,

PERSPECTIVE SCALES
tin',

19

Then mm'n'n is a rectangle parallel to and again on mm' by the horizontal n'm'. and m7n' must be equal to nn', which is perspectively equal to B6. This is as if we had set the man in the picture plane at B6, marched him along the wall to nn', and then away from the wall and parallel to the picture plane, to his posiOr we may set off the height 4-6' from 4 on the Perpendicular .^Cby which tion mm'. the figure was placed, and drawing 6'C, cut off mm' at once. This is the simpler process, but the other method is sometimes better, keeping measuring lines at the side of the picture, where they are least obstructive. Indeed, though perspective scales may be set anywhere in the picture, it is usually convenient to put them in the margins,
the picture plane,

where they interfere

least with the things that are represented.

is

A common
of the figures

use of scales of heights

for the placing of figures in pictures.

We

establish, usually in the picture plane (Fig. 19), a vertical line

AB

that

shows the height

and the positions with reference to the Horizon Line. Horizontals, A V drawn to the same vanishing point on HH, will be everywhere six feet apart, if that is the height assumed, and a figure of that height will fit between them anywhere. they are drawn It makes no difference to what point of for if they vanish in it they must be horizontal and parallel. Therefore if we have an isolated figure to place, and know where it stands, after we have fixed our measure AB, as in Fig. 19, we have only to draw y^F through the position at m to the Horizon Line, and join BV. Then
and

H V,

HH

the vertical

mn between

A f^and

BV gvj^^ the height of the figure in

mm'
that,
till
it

its

proper place.

But
is

if

we have
is

to distribute a

number

of figures about the picture at the

same

level,

it

better to set up a scale in the margin as in Fig. 20, and then from any point
to stand

where
point

a figure

draw

a horizontal front line

meets

^ Fat

m'.

The

m'

is

projected up vertically on

BV^t

n',

and

projected again horizontally upon

the vertical at m, gives mn, the height of the figure.

tions

single scale thus gives the posilevel,

and heights of
for Fig. 18.

all

the figures that stand at the

same

by the

first

process that

we used

If

we have a

series of figures or other upright objects, like a

row

of columns or posts, as in Fig. 21, the positions of their feet being determined at 2, 3, etc., a line

may be drawn through their feet from its trace A to the Horizon Line, and the height AB being set off in the picture plane, the horizontal ^Fwill cut off all the columns at the right height.* If the figures stand at different levels, a measure must be provided at each level, as in Fig. 22 (Plate VI), where the figures on the Ground Plane are fixed by one scale, and those on the higher levels by others. If the figures vary in height, if some are children, for instance, the artist's judgment will allow for this when he has fixed the standard for figures of normal size. The proper scaling of figures and animals is important to good effect the neglect of it leads to many blunders, always irrational and often absurd. Animals and carts that could not get into their barn doors, and men who could not inhabit the houses provided for them, are among the commonest instances. But perspective scales are useful in
;

many other things than They are most commonly

1

the placing of figures, especially in architectural subjects.

set in the picture plane,

where

lines are

measured

in their

shows a convenient way to continue the spacing of such a series when the first interval has Through the middle point c of AB we draw fjf', which will bisect all the upright lines. Then a diagonal Bd through the middle d of the second will mark the point j on o K, and when the third is drawn the point 4 may be found in the same way.
Fig. 21

been

fixed.

20

APPLIED PERSPECTIVE
Nevertheless,
it

true dimensions, according to the scale of the picture.

often happens,

especially in architectural perspectives, that the scale of the data from which the draw-

ing

is

to

be made

the

plans and elevations, for

In such a case,

instance

is

too small

for

the

perspective, in which dimensions are

much reduced by
it is

foreshortening and by retiring

behind the picture plane.

often well to establish scales of widths

and heights in a plane so far behind the picture plane that dimensions that are set off For on them to the scale of the data may sufficiently increase the size of the picture. instance, in Fig. 23, b being the geometric middle of aC, and am and bn both front lines, am will be twice as long as bn, and therefore, if distances are laid off on a scale at bn, the picture will be twice as large as if they were laid off in the same size at am. In If bC is taken at a third of C, the picture will be magnified three times, and so on. Fig. 6, for instance, the scale of the sketch plan was too small for the perspective, and if the drawing had been complicated enough to require measurements from the plan, we might have had to place our scale considerably
nearer C, in order to magnify the dimensions.
other hand, to construct a picture of the same

On
size,

the

we

must use twice as long a unit

that of bn in Fig. 23.
*'
**'

for the scale of

am

as for

J
;

'v-

Note.

By

the geometric
it

middle of a
lies

line

drawn

is

on the paper. The perspective middle is the perspective of the middle point of Ti^ ^J In the same the line which the line on paper represents. way geometric division is division with reference only to the length of the line on paper; perSo we say that spective division represents in perspective the due division of the actual line. (Fig. 23) are geometrically equal, but that ao and bp are perspectively equal. ao and om
i"*

meant the middle of

as

it

^
(k

PL.

a J

THE PERSPECTIVE PLAN

In drawing objects of more complex outlines, as, for instance, the plinth block, and still more in buildings, it is often helpful to put the plan in perspective This plan, with the many lines that are needed first and construct the picture from it. points, can be set below the picture, which can be constructed to project the necessary by simply plumbing up from the perspective plan, and then determining the heights
Fig. 24,

above the Ground Plane. The difficulty that in ordinary views the foreshortening of the Ground Plane, due to its being not far below the eye, crowds the lines together, and makes their intersections very acute in the plan, may be obviated by lowering the plane of the plan. Its Ground Line may come as far below the Horizon Line as we choose, and the plan be broadened and opened out so as to be easily constructed and easily
read.

This appears

in Fig. 24.

The geometrical plan

plan.
at

of the plinth block

is

shown

at

the top, the perspective plan at a lower level, and the perspective view between, constructed by plumbing

up from the perspective

It will

be seen that the form of

It is

the picture

is

nowise changed by taking the plan

a low level.

simply as

if

the

block had been lengthened out like a chimney, and carried down to a foundation below

ground ground

the
level.

same superstructure could be

built

up from the
is

cellar

bottom or from the

to construct
it

The upper
It will

surface of the block

it

really a plan at a higher level, but so

foreshortened and crowded that

pendently.
in the lower plan, to say

would be

difficult

and uncertain
is

inde-

be easily seen that the intersections are more securely constructed

nothing of the clearness which

gained by banishing

many

constructive lines from the picture.

The

perspective plan

is

simply found by transferring the widths from the geometrical

Ground Line G'G', drawing from their traces the Perpendiculars iC, 2O, jC,' 40, and the Diagonals aD' and bD\ which vanish in the Distance Point D'. The
plan to the
4!, 5', 7', 8', from which the plan is easily conand by plumbing up to the same Perpendiculars iC, 2C, jC, 4C, in the picture, the angles of the lower bed of the block are determined. The height x being set off on a vertical 2-g in the picture plane, corresponding Perpendiculars may be drawn in the

intersections of these give the points j',

structed,

plane of the upper bed, and the corners projected up to them.

It is to

be noted that

the Horizon Line of this problem serves also for that of the next, although the two

The D, which intrudes upon the space of the other problem, is encircled in order to distinguish it. This same line is used also to represent the picture plane in the geometric plan to save space, and for the same reason the Ground Line of the
subjects have nothing to do with each other, and the Station Points are different.

distance point

picture does duty as the Horizon Line of the perspective plan.

The time-honored Problem of the cross. Fig. 25, which is given next, brings out to advantage several points in the use of the perspective plan. The geometric plan as
given at the scale of the picture
in

the last problem was awkwardly large.

at half the

In this

problem

it

and the elevation are taken

intended scale, or the scale of the

22
picture
is

APPLIED PERSPECTIVE
doubled.
is

The

cross stands on a base of three steps.

Here the Horizon

is

Line of the picture

geometric plan

used as the Horizon Line of the plan.

Since this

to double the

in scale, the

widths are measured on a scale of widths, mn, set halfway

between
shall

HII and

G'G', so that the measures, transferred to G' G' by Perpendiculars,

be doubled, as explained above.

Moreover, because the scale of the drawing

oflf,

carries the distance point inconveniently far

tuted,

the half-distance point

DJ ^

is

substi-

and the measures of depth are

set off at half the scale of the widths

on G'G'

that

is,

exactly at the scale of the geometric plan and also of the elevation.

The whole

apparatus of distance point and measuring lines has been banished from the picture to
the plan, with a considerable gain in clearness.

The

picture plane,

PP,

is

taken, here and in the last problem,

is

of the object, a condition that

often required in practice.

diagonal ao
far

is a convenient means of laying it out. enough below that of the picture to display the plan

some distance in front The plan being square, the The Ground Line G'G' is chosen
well,

and the positions of

C and a
The

are taken as

may be

convenient.

CDj ^

represents half the length of the Axis.

widths of the geometric plan are projected on PP.

in the perspective plan,

The

trace a of the diagonal,

assumed
in

projected on

mn

by the perpendicular
at 7', 2', j', etc., to

aC at

', is

taken as the

origin

from which these widths are set

in

oflE

be projected on G'G'

I, 2, J, etc.,

the doubled dimensions required for the picture.

The Diagonal ao

cannot be drawn to the distance point D, for that is inaccessible, and so the point b is determined by setting off the distance aj from PP, on G' G' at 7/3 (or, what is the

same
all

thing,

by halving

a'j

on G'G'), and so drawing ab

is

to Z'/^.
all

Its intersections with

the perpendiculars Ci, C2, Cj, etc. give two angles of

the steps and of the shaft

may be at once For the heights we use Here the heights may be got as the widths were a vertical scale at /i: in the picture. got, by setting them off directly from the elevation on a measuring line halfway from Cto 7^; or we may set them oflf at double scale directly on 7^, as is done here at p, 10,
of the cross,

and the plan

easily completed.

The

corners / and 7

plumbed up

to their respective perpendiculars in the picture.

etc.

the cross on a vertical plane

depths.

The perpendiculars /C, gC, loC, etc. mark the heights of the steps and arms JcC ; and 7C in the plan may be taken for a scale
front lines of the steps in the plan are produced
series of points
7', d, e,f, g', d, e,
till

of of

The

they cut this scale

in d,

e,

f;

and the

projected up respectively upon 7C, pC, loC, iiC, in

the picture, gives points

through which horizontals

may be drawn which

will

contain the front edges of the steps, and

by projecting up from the plan below. The horizontal lines of the arms of the cross are found in the same way, and the rest of the picture offers no difficulty. Instead of transferring our widths directly from the geometric plan and projecting
cut
oflf

maybe

at the proper points

them
is

in

full size.

double scale upon G'G', we might have at once set them off upon this line in Sometimes one method proves more convenient and sometimes the other it
;

be seen how taking the heights on the same scale jc The horizontals which transmit at one side clears the picture of obstructive lines. as well as to C, provided this these heights might vanish to any other point on were done in the plan as well as in the picture. It is not necessary that they be Perwell to be used to both.
It will

HH

pendiculars.

MEASUREMENT OF LINES
Thus
been
all

far the lines

measured have been

all

perpendiculars, and the planes used have

normal
is

to the picture planes.

It

remains to measure oblique

lines. ^

The

process

essentially the same.

The
is

distance points of Perpendiculars are set off from their vanishing point, which

and being as far from this as the Station Point is, are determined by the conditions of the picture. Distance points of inclined lines, which are called measuring points, are set off from their vanishing points, and are as far from these as the Station Point but they have to be determined for each case. For Let F(Fig. 26) be an simplicity's sake we begin as before with horizontal lines.
the Centre,
for all

once

oblique horizontal, and a

picture plane,
parallel to

its

trace in the

s^r

-y

ac' is a line of

measures

on which given distances off, and J/ a measuring point which we assume to have been found. Measuring lines bM and cMcvlX off distances ab' and b'c' which are really
ab and be are set
equal to ab and
the plan below.
tion
Point,
be.

HH,

The
will

construction

and proof are simple, as

appear from
the Staparallel to
is

Here 5 being
\x\.

SV

the plan

ae finds

V as

usual,

and
bb' is

VM
made

taken

equal to VS.
to

Then

parallel

SM,

so that ab'

is

equal to ab, and

the isosceles triangles

in

SVM
to find

z.x\A

bab' zre similar.

J/ the vanishing
is

point of bb' and ec'm perspective.

used
in

justified,

and

M we have
it

Hence The

the visual line

SM finds

construction

only to construct the triangle

we have just CVS, on the

Horizon Line

perspective drawing or elsewhere, and transfer the distance

VS
is

to the

VM.

We

say,

then, that every vanishing point of horizontals has its measuring

point in the Horizon Line, as far from

as the Station Point

in space.

But

in reality

there must be two such points, one on each side of V, just as there were two distance

points in the Horizon Line, one on each side of C.

trace a and the line of measures are taken
,

Fig. 27, in

which for variety the

above the Horizon Line, shows how the two measuring points will give the same division, but it seldom happens in practice that the two fall in equally convenient positions for use.

^^^/
ST
.

J*C'

Furthermore, as we found that distance

'

y
makes an acute angle with the
picture plane, an inclined

line

Here we note a distinction with the ground plane.

an oblique

line

24

APPLIED PERSPECTIVE
may
lie in lie in any direction from C, so any direction from V. Fig. 28 shows

points were not confined to the Horizon Line, but might the measuring points of an oblique line the use of different measuring points with the
division of the line

same vanishing

point, giving the

same

aV.

The

construction

is

obviously independent of the direction of

is

i^. ^S

^^

the Horizon Line, and

in

applicable to lines

any direction from

lie

V.

The measuring

points will

in

a circle about

as the

distance points about C, and the only condition

is,

as in the case of distance points,

that whatever

may be

the inclination of

the line VM', which as before

we may

call

a secondary horizon, the line of measures

must be

parallel to

it.

This, as in the case of distance points, insures that the meas-

uring lines and the line to be measured shall really intersect, by putting them in the

same plane, which vanishes in the horizon VM', whatever the position of that may be. So we reach the general statement Every vanishing point has a circle of measuring points about it in the picture plane,
:

is

as far from

it

as the Station Point

in space.

The

line of

measures for any measur-

ing point must be parallel to that secondary horizon in which the vanishing point and
the measuring points
lie.

This statement, when the vanishing point

is

the Centre, applies exactly to the distance

points and Diagonals, which are only particular cases of

measuring points and van-

ishing lines.

Fig. 29

shows the complete

points.

circle for different

measuring points with

All

the

same

vanishing

three

points,

M, M', and M",

give the

same

division of

aV&t

b,

and so would any other

Fig.

point in the

same circumference.
30,
illustrates

The
plane

folding screen.

the use of measuring points.

is

The

picture

set at

the near corner c of the

screen, so that the

upright edge ch

may

be measured off at once from the Ground Line at the scale of the picture, for which abed above
is

the plan and

the height.

y^'
three blades are V,

The vanishing

points of horizontals in the

V, V", determined by the visual lines SV, etc., and the corresponding measuring points M, M', M", determined as before, by taking on the plan FJ/ equal to VS, VM' to V'S, and V"M" to VS. Assuming fat a convenient point we draw cd to its proper vanishing here it is plumbed down from the plan, of GG', measuring off the width of the blade cd to d', draw a measuring line, point r", and

d'M", which cuts off the true length cd on cV". cd if hV" is drawn and the vertical at d intersecting
right-hand blade.

is
it,

the lower edge of the blade, and

The blade

eg

is

constructed in the same

we have the true outline way with the help

of the of the

MEASUREMENT OF LINES
points F' and M'.

25

To

place the corner

a,

which
its

is
e,

not in the picture plane, the

simplest
after

way

is

to prolong ab in the plan to

off ea

trace

and projecting that into GG,

drawing ^Fto measure

on

it

at

a by the measuring line aM.

This problem,
natural
visual
first

like that of the

plinth block (Fig. 24),

may be
b, c,

easily solved
ti

method upon /V and plumb down from them to their positions in the picture, The plumbfixing c upon GG, then d upon cV" and b upon cV, then a upon b V.
of visual projection.

We may

project a,

and

in

the plan

by the by

lines

lines give the vertical

edges of the screen.

This solution

is in

this case simpler than

i^V-

^^

the other.
is

But a method of finding the point a by measuring from a point

b,

which
it,

not in the line of measures eb (here the

It

Ground Line) but

is

projected upon

is

happens that we need to measure off a given distance on a line whose trace we cannot reach. Suppose, for instance, we wish to set off a length x (Fig. 31a) on the horizontal ao lying in the Ground Plane, whose trace is out of reach, but its measuring point M'xs known. Projecting a from J/ into the Ground Line at a', we measure s'l^' equal to x, and drawing b' cvX off the proper length ab. The result is the same whether, having the trace m, as in Fig. 31^, we measure off successively
impwrtant.
often

F^. 3/i
S'

ma and ab by Ma' and Alb' ; or whether, not having m but having a, we draw Maa' at once, and then, laying off a'b', draw b'M. So measures may be set off on a line from any point as well as from its trace, but the starting point must be referred back to the
same
line of

measures ma' as

if it

had been measured

off

from the trace m.

26

APPLIED PERSPECTIVE

This leads back to the principle before laid down, that all measures to the same must be conceived to be taken in the same front plane. Thus far it has usually been the picture plane, in which all measures appear in their true size ; but the measscale

among themselves if they are taken in any one plane parallel to that we have seen, all dimensions parallel to the picture plane are perspectively diminished in proportion as ihey recede from it, and all that recede equally from it, that is, are in the same plane parallel to it, are diminished in the same proportion,
ures will agree
plane.

For, as

and,

if

they are equal, remain equal in perspective,

or, if

they are not equal, they keep

their true ratios.

we would measure off lines in true dimensions, we must measure them in the picture plane; but if we have only to divide them into equal parts, or in given ratios, we may measure them in any plane parallel to the picture
Therefore,
if

remembered, that the line of measures must be which the given lines vanish, and therefore all lines of measures for that horizon, in whatever plane, must be parallel. Suppose then that we have to divide a certain length ab of a given horizontal line
plane,
parallel to the horizon in
jyj_

the original condition being

into four equal parts.

',

of

measures through
from

-'

,'

\".^

it

a,

We may draw a line and projecting b upon

at b\ divide ab^ into four equal

parts at

and j, and project these points back upon the line. Fig. 32 shows that we
/, 2,

-;';;

'-^

^7j

^^
"

should in this way get the same division as

Ground Line in same ratio. The -^'f^ ''^' difference is that in that case we should have recorded on GG, and this shows us a means of getting the exact dimensions of the parts at the real length of any perspective dimension on a line whose measuring point we know, by merely projecting that dimension back upon a line of measures in the picture
if

-j"

we projected ab upon
in

the

'*''

'"

"*'

a'(5"anddivided that

the

plane.

Note. The ab.

line of

It is

actual measuring point

for equal or proportional division of not essential to use the measuring point is the only one that will give identical lengths on ab and the
in the picture plane,

measures

because

it

is

the only one which will

make

the similar

triangles isosceles, as in Fig. 26; but lines to any point in

HH

^i\\ represent parallels

dividing

ab and

the line of

measures proportionally.
far
;

assumed that the given line ab is horizontal but the same construcIf, as in Fig. 33, the line line whose vanishing point we can use. is not horizontal, that is, if its vanishing point V\s, not in BII, or if any other reason requires, we may draw through Fand a two geometrically parallel lines Vm and ao, and laying off on ao as a line of measures the required parts to b', draw b'b produced to m
have thus
tion will serve for

We

any

Then lines to m line through V, which is the horizon that corresponds to ao. from the points of division on ab will divide ab as is required. As we used accidental distance points for the measurement of Perpendiculars in inclined normal plane.s, so we may use accidental measuring points in inclined horizons when we construct in planes
on the
that are not normal.

MEASUREMENT OF LINES
Note.
lines

27
is
still

The

essential thing here (as

was stated on page

in the

12)

that the measuring

and the
at?

line to

be measured shall be

same

plane, so

that they shall intersect.

Now

the line

(Fig. 33)

may

represent a front line

through

a,

and the

parallel

through J^the horizon

or vanishing line of the plane of ao

andrtK

Then

the measuring lines I'm, 2'm, b'm must be in the

and therefore intersect vanishing in m, and so divide the lines ao and ab proportionally. We may test our construction in Fig. 34, by drawing two lines through V. of which one or neither may be the Horizon Line, and two lines of measures parallel to them through a; if the divisions of these lines are the same, or in the same ratio, we again, it cannot be too carefully borne in mind that the division of a Kand ao is only proportional, unless a is the trace of a V, and therefore ao is in the picture plane and unless m is the actual measuring point iJ/, corresponding to V, in which case the portions cut off on the two lines are equal.

same plane
and be

of

Vm

and

ao,

ab,

in reality parallel lines

shall get an identical division of

V.

But

s:

7M,

r
'^
<

-2-

- ii'^"'^
/>-^^-^
1

It

may happen

that in measuring off succes-

"^^^^^

/'

"

s.'

sive distances
last divisions

on the same line of measures the would pass out of reach, and it is

i?^.J<^

-i:?'"

them on a new line of measures. Suppose that in Fig. 35 we have set off equal parts, the intervals of a colonnade for instance, on a^ by the measuring point j1/ and the line of measures a^', and having reached the point 4' have no room to go far,^ ^^^ "^ -"? ther. We may draw a second line of measdesirable to continue
ures

through 2 parallel to the

first.

The
first,

divisions of this line, like those of the

must be equal, and projecting the

last division

j-4 upon it from any convenient point m in VM, we get a unit j"-.^" for them. Repeating this unit at 5", d", 7", and drawing lines from these points to m, we continue the division exactly as
if

From

these examples

we had gone on with the first line ./'. we deduce the following rules
used for a line of measures
it

to to

Any
a point

front line

may be

oxi

a.

line parallel to

if the measuring lines are drawn which passes through the vanishing point of the line

be divided.
In this case the divisions
if

will

be proportional to those on the

in

line of

measures

but
cor-

they are to be equal to them, that line must be

;

the picture plane, and must pass

through the trace of the line to be divided

moreover, the point

must be M, the
can,

responding measuring point of the given

If,

line.

however,

we have determined one

division of the given line,

we

by projecting

28
it

APPLIED PERSPECTIVE
line

upon any front

It is

from a suitable measuring

point, fix a unit

by which we can con-

tinue an equal division, or which will serve for a scale of division.

obviously essential that the measuring lines should be drawn across the line that
is

to

be measured, but sometimes the condi-

Jtr

y~

tions of the drawing forbid this, as in the case

is not room on may draw any convenient line a' (^ to the same vanishing point it will then have the same measuring point Af. Having measured off the required distance a'i' from the same measuring line

71^.36

>
will

\
7>V'

\\^WC'

of a

in Fig. 36,

where there

the sheet.

In this case we

aa',

we draw

^^' parallel to aa'.

The
is

intersection
a'l)'.

l>

will

be the required point on aF, Imagine

a,

and ad

be perspectively equal to
line,

For, aa' being a front

M'

so also, and the two are parallel in space.

a measuring line drawn from the inaccessible

on aa'

at the

proper distance from

making am=a'm'. The two triangles abm and a'b'm' in space are similar, having Therefore their homologous side parallel, and having equal bases they must be equal. the parallel sides ab and a'b' must be equal.

CONJUGATE VANISHING POINTS

Thus
far

we have used only

single or unrelated vanishing points, except in the case of

Perpendiculars and Diagonals, for which we have found a special tieatment.

of the objects

But most

which are drawn

in per-

spective, buildings, furniture,

like,

and the

are either rectangular or based on

rectangular forms, and the angle being

invariable,
it

is

worth while to find

the angle in per-

-r-

some

relation

between the vanishing

fix
is

points which will

spective.
tion, of

There

a very simple relain

constant use

perspective.

If

we have a
in

horizontal square to
as
in

put

perspective,

the

plan.

Fig. 37,

when we
its sides,

lay out the vanishing

making 5 ^parallel we shall have a right angle at S; that is, 5, V, and F'will always If, then, be the angles of a right triangle whose hypoihenuse is in the picture plane. we have the position of one vanishing point given, say by the distance CV, or have the angle of one side of the square with PF, that is, the angle SVC, this fixes the triangle SVC, and we can find the other vanishing point. We need not draw the plan, or even the triangle, for we know that any right triangle VSV may be inscribed in a semicircle
points of
to ab,

and

SV to ac,

that has

for

its

diameter.

Knowing

the angle of either side of the square with the

picture plane,

point from

6'

we can construct its vanishing (Fig. 38), and then a semi-circum-

ference from

through

S,

with

its

centre in
this
is

VV\

will fix

v.

Moreover, since

purely geometrical construction for finding distances,

we may perform
it

it

anywhere and simply

transfer the distances to

It is

HH

in

the picture.

natural to lay

off

from

HH,

where

C is,

and the vanishing points are to be. The triangle really is in the Horizon Plane with 5 at
its

--^gr>^revolved about

right angle; this

diagram represents

it

as

if

FFand

laid flat in the

measuring points J/ and M\ as is evident in the figure, and the square may then be described, as is there shown. The vanishing points of two lines or series of lines at right angles to each other, in view of their close perspective relation, may be called Conjugate Vanishing Points, and will be so called here. The semicircle F.SF' with the points M, C, and M' on its diameter may be called a perspective chart (Fig. 39). It may be constructed anywhere
picture plane.

The same diagram

will furnish the

30

APPLIED PERSPECTIVE
to

and

any

scale, the

they are transferred to the picture.

points are apt to

dimensions being enlarged, or diminished in due proportion when Inasmuch as in large drawings the vanishing
far
ofT, it

come very
:

is

often convenient to construct the chart to a

smaller scale than the picture.

the rest follows necessarily
first,

If

the Station Point and one vanishing point are known,

sometimes it is convenient to fix both vanishing points and then the Station Point is confined to the circumference of which is the

VV

It is a common practice to take Fand as far apart as the paper or the drawing board allows, and adjust 5 and C to suit them. Pins are often stuck into the board at Fand V, and if a draughtsman's triangle or square is thrust between them, its right angle may be swung round to describe the semicircle and find a position for

diameter.

S, while its sides

show the

inclination of the object to the picture plane (Fig. 40).

Let us take the case of a chair, a piece of furniture that often refuses to stand firm The plan and elevation, Fig. 41, at half the .scale of the perspective, give in a picture.
dimensions and assumed position with the picture plane the perspective chart on the left is at one third the scale of the picture, and its distances are set out on accordingly. The horizontal line of measures ax is taken at the lowest .square corner The horizontal lines, vanishing right and left in V and V, are of the frame at a.
its
;

HH

measured by the corresponding measuring points J/ and M' on double the scale of the It gives clearness to the drawing to take the vertical measures directly from the elevation on a scale of half heights mn at half the distance from Vto the vertical az in the picture plane and project them on az. The spacing of the balusters under the arms, being merely the division of a horizontal line into equal parts, may be accomplished by projecting the extreme points from any convenient point, as C, in HH, upon any convenient line op parallel to HH, and dividing the intercepted section
plan and elevation.
op into the necessary
parts,
It

number

(six) of

equal

which are projected back on ee'. is often convenient to know the vanishing

point of the diagonal of a given square, or the mitre line which bisects the angle of a rectangle,

and

this is easily included in the

diagram

or chart.

vanishing

V&nd V, Fig. 42, be coordinate points, and S the Station Point, S^

Let
If

being the diagonal.

we draw the diameter

the diag-

bd perpendicular to

VV, Sd must be

onal passing through 8/ for the inscribed angle

VSd, standing on a quadrant, must be 45, and

CONJUGATE VANISHING POINTS

so half of

31

VSV. The
all lines

point 8 therefore, where

parallel to Sd.
If

ishing point of
circle

we may

bisect the arc

Sb

at a,

5^ cuts the Horizon Line, is the vanwe have not room for the lower half of the and draw the radius ao. S^ parallel to ao will give
The diagonals
of a square being at right
8,

the point h}

The

point

is

often useful, in drawing the mitres of borders or pavements,

for instance, or in constructing octagons.

angles, the vanishing point of the cross-diagonal will be conjugate to

and may be

found from
mitre points.
It will

it

by the regular construction.

The two

points

may

conveniently be called

be seen that any two opposite distance points must be conjugate, for they are

the vanishing points of lines that

make

angles of 45 with the Axis on opposite sides,

and therefore a right angle with each other. As a matter of fact the distance points in the Horizon Line are much used as conjugates, especially in architectural views. They are nearer together than any other two conjugates, and so suit a narrow sheet or drawing board, and all the construction points are symmetrically placed about the Centre the two mitre points both coincide with the centre, and all the mitre lines become perpendiculars, so that the whole construction becomes compact and simple. It has the disadvantage that in rectangular objects near the Centre the lines slope both ways at nearly equal angles, giving them a stiff and ungainly aspect which is only partially cured by setting the principal subjects as far to one side as practicable. The choice of conjugate vanishing points has much to do with the agreeable look of a picture, and the painter or draughtsman learns that things look better for being drawn Buildings and other rectangular objects are best with their vanishing points remote. shown with the lines on one side not far from parallel. One vanishing point approaches as the other recedes, and the contrast of strong foreshortening on one side when the
Painters recognize the advantage of a is broadly displayed is usually agreeable. remote vanishing point, and are apt to avoid, unless for special effects, the strongly converging lines that come with near ones. On the other hand, the steepness of horizontal lines that converge downward and vanish close at hand makes things look high, and often adds grandeur to a picture. It towers and strongholds set on high rocks or suits well with picturesque architecture, mountain tops, and the like. Foreshortening carried far enough to be conspicuous is often very efft-ctive, and tells well in picturesque architecture where there are many

other

vertical lines, as in Gothic, or in

in street

most columnar buildings.

is

It is

the essence of

efft- ct

views and long interiors of great churches or galleries, which are naturally

drawn

in parallel perspective,

where there

but one vanishing point, and the fore-

shortening and convergence are at a maximum.

But

it

will

be noticed that

in parallel

perspective, where two sides of an object are shown, as at a street corner, the side

which fronts the spectator and shows no convergence is apt to disappoint the desires some slight convergence, and is inclined to fancy that the parallel
in contrast to the sharply

eye,

which
seen

lines,

convergent Perpendiculars, are actually divergent.

This does
lines,

not appear in interiors, where the walls that front the eye are in the remote middle of
the picture,

and are usually enclosed between two retreating walls whose

Therefore parallel perspective
is

con-

verging on both sides, balance each other.

1

peculiarly

Sd and

ao must be parallel, for the angle aob at the centre, subtending the arc ab, must be equal to

the inscribed angle Sdb subtending double the arc.

32

APPLIED PERSPECTIVE
its

showing both walls at once, But for open views which show two sides of a building, or for rectangular objects away from the Centre, it is less satisfactory. There will be more to say about certain unpleasing aspects of parallel perspective when we come to speak of perspective distortions.
convenience
in

well suited to interior views, apart from

and

it is

agreeable in most street views, which are analogous to interiors.

Fig. 43 represents a flight of outside steps.

Two

steps surround the square platform

before the arched doorway, and from this a straight ramp ascends to the right against
the wall.

On

the right are the geometric plan and elevation at half the scale of the

perspective.

We

draw

first

a perspective plan.

According

to the principles just laid

down we choose such

be presently

a view that the right-hand vanishing point shall be considerably

Assuming an Axis which, according to a rule that will long as the greatest width we mean to include in the picture, we lay out our perspective chart, fixing F, V, M, M', C, and S, in whatever place and scale are convenient, and set off the points at proper distances on the Horifarther off than the left-hand.
justified, is at least as

zon Line of the perspective plan, which

plan.

is

also that of the picture.

We

begin with this

To make
c,

the case general

a.

corner of the steps at

a, b,
ric

we take the The diagonal ad oi

line vanishing in

picture plane a

little in

front of the near

the platform, which contains the corners

8.

and

d, will

be a mitre

Producing

it

to cut

plan at A,

we

place this point on

GG,

in the perspective plan,

AB

; but it may be found by producing ea and the parallel edges of the steps to in the plan, which corresponds to A in the perspective plan, and measuring off the points /, 2, j, 6, from by M', and drawing iV,2V,jV,6V, which contain the parallel edges, and by their intersections with A^ give the corners a, b, c, and d. The

which we can measure the widths of the steps. till it cuts FP in a\ and laying off Aa' on

The corner a
at

is

FP in the geometand draw AV, on fixed by producing ea

GG

double scale

with the other corners

l>

and

GG

line en of the wall

behind the steps may be drawn to

V through
The
The counter

d.

The

lines

aV,

bV,

cV give

the corresponding front edges of the steps.

point 6 gives the right

same way, the corresponding diagonal 7^ may be drawn for verification of these corners. Successive measures for the steps of the ramp are set off on an auxiliary line of measures 6m, the intercepted interval 6-7 being repeated on it for a unit as in Fig. 35, p. 27. The width of the buttress of the ramp is set off on d F at // by projecting // from J/ upon GG (where it happens to coincide with i'). The line of the face of the wall being already drawn through e, the door opening is set off from 4 and 5, and its depth found like the width of the buttress. This
8,

front corner of the platform, 7

and

measured

in the

corners of the steps and the edges vanishing in V.

completes the perspective plan.

The

picture

is

zon Line

HM
in

a.nd a

plumbed up from the perspective plan with the help of the same Horinew Ground Line G'G'. Az is taken as a scale on which heights

are measured up to the sill of the door. From that point the heights are measured on the auxiliary scale Ah', the reduced units being transferred from Az. The arch
is

sketched

the

the base of the wall of the

method of constructing ramp is found by

circles will

setting

be given later. The line at back from g on the plan to k, and

plumbing upon gV.

the plan.

The

successive heights of the steps in the ramp, being set off on

A'z\ are transferred by horizontals vanishing

To

in /', and the corners plumbed up from determine the lines of the sloping buttress, we choose the front edge of

CONJUGATE VANISHING POINTS

33

a convenient upper step, no in the plan, and plumbing up a vertical to n in the picture, set off the edge on the height ' above by the scale A'z'. Then drawing the horizontal n'p,

plumb up the points

and/ upon

it.

The bonding

of the

masonry

offers

no

difficulty.

Examination of Fig. 43 suggests a simpler way of constructing the steps.

of a straight flight of steps are all in the

The edges

and the corners of the ramp from d to n are in the same straight line dn against the wall. If we had set out this line beforehand we could plumb up the corners upon it from the plan at once. But by the help of the slope line dn we may construct a flight of steps without a plan. Suppose we have to lay out the steps in Fig. 44, whose tread and rise are shown in the margin. There are to be five steps, the upper a square landing, and they rise to the left from A. On A^ we set off the height of five risers, and on A^' four treads, which are transferred to AVhy M. The horizontal 5 F and the vertical bb' give the upper front corner b', and the slope line /i^'must contain all the near corners. The width of the ramp is measured off on A at f, and a vertical at c intersecting i gives the corner c' ; ^Kand bV give d, and the vertical dd', cutting VV, gives d', which fixes the slope The steps can now be drawn in two ways either by dividing ib' with horiline dd'. zontals iV, 2 V, etc., for the risers, or by plumbing up the treads from AV. The front
inclined plane,

same

edges

7^', etc. will

transfer the divisions to

lines
:

c'd'.

Thus

all
it

the steps
is

may be determined
draw
also the slope

by merely dividing the upper slope

precision of the construction.

nevertheless,

well to

lines of the inner angles beneath the others,

and these

will
its

be accurate tests of the

diagonal
is

The landing being

drawn.
that, the

square,

a mitre line

vanishing
It
is

in

8,

and so

it

is

easily

clear

from the figure

upper and under slope

in

lines

once established,

the lines of the ends of the stairs

may be drawn

without farther construction than

zigzagging from one to the other with horizontals and verticals.

having placed ab and

bb',

a'b',

we may draw the

is

Thus in Fig. 45, vertical aa', then the horizontal a'b, then
on another small
line perspecwill

then
is

b'c,

and so on.

This

usually enough for free drawing or for field sketching,

is

but

dangerous where precision

It
is

required, for in piecing one part

errors are apt to accumulate, and the

draughtsman
b', c', d',

is

not likely to

come out where he

be

aims.
tively.

better to control the points

by dividing one slope

Then any
it.

error in an intersection will be isolated,

and the next point

independent of

Now going back to Fig. 44 we may see that all four slope lines are really parallel, and must have the same vanishing point if we can find this vanishing point it will be easier to place them. Moreover, the lines Ab, Ae, and i-b' are all in the same vertical plane, the plane of the ends of the .steps their vanishing points must be in the vanishing line of the plane. The vanishing line of a vertical plane is a vertical line, and since that of this plane contains the vanishing point of Ab, which is V, it must be the vertical through V. Any one of the slope lines then, if continued, would find the
: :

in its intersection with the vertical through V. The line Ab, or A V, under Ae, in the plane Abe, which is the ground plane, and is its projection on the Ground Plane we may call it the horizontal ground projection of Ae. We may say then that the vanishing point of any inclined line, any line, that is, which makes an angle with the Ground Plane, is in the vertical that passes through the vanishing

vanishing point
lies directly

34
point of
its

APPLIED PERSPECTIVE
ground projection.
line
It

remains to see how

this is to

be found.

Let a

be any inclined
projection, as

vanishing (Fig. 46) in V.

seen,
is

we have

directly

The vanishing point under Fin at the foot of

HH

of

its

ground

the vertical

W.

The lower

half of the figure represents the plan,

and the picture plane PP.

In the plan

F is
in

showing the Station Point and Axis, projected in and SV'xn SF'. The

three points S, V, and

V are
its

really at the

angles of a right triangle whose base

the horizon plane,
its

SV

is

perpendicular

FV'
is

Z'i^ y^^

in the picture plane,

and

hypothenuse
If

SV, unseen

in

the plan.

this

triangle
till it

were revolved round the vertical

VV

-^

lay in

the
to

picture

plane,

round
angle.

M, and VMV The angle VMV

would swing would be the triis

then the angle

its

VSV,
zontal

the angle between

projection, that
is
is,

aVand
its

hori-

But and

M
its

slope angle.

the measuring point of both a

horizontal projection, for MV'vs,

revolved, and

MV SV revolved.
is

SV

There-

fore the natural

into position,

way

to put an inclined line

its

when we know

its

inclination,

is

first to

place in plan

horizontal

projection with the vanishing point

V, and then from

the measuring point of that

on
the

HH draw
vertical

a line

making the proper slope angle with

HH,

which

will

intersect the
If

through the vanishing point

is
all

in

the required vanishing point V.

ground plane

not convenient, the projection of

serve our turn, for

ow any horizontal plane will such projections must have the same vanishing point and hori-

aV

zontal measuring point.

We

then arrive at the rule

An

inclined line has the same measuring point as

its

horizontal projection, and

its

vanishing point

where a line from the measuring point, making its slope angle with the Horizon Line, cuts the vertical through the
is

vanishing point of the horizontal projection.

The

relative positions of the different points

are sketched in Fig. 47, where

SHH
and

is

the

horizon

plane with

S upon
its

it,

PP

the

picture plane upright before the Station Point.

SVV
and

is

the triangle in

normal
angle

position,

MVV the
into

same

triangle revolved

round

VV
which
is

the

picture.

The

VMV,
point

the angle

VSV,

is

evidently the slope-angle of the line

MV.

The

Horizon Plane on the circular arc SM. It is sometimes important to draw a line toward an inaccessible vanishing point, and to measure distances on it. Fig. 48 shows how this may be done. Suppose we need to draw a line from a toward a point F which is out of reach, but whose distance from some point m on its horizon is known, as it may be from a chart like that in Fig.

moves

in the

"5
<^

V
s

^
1^1

vV

Nl

N
/

CONJUGATE VANISHING POINTS

44, for instance.

35

We
half,

to V, say

one

and

half of ma, or a tnird,

equal to any convenient fraction of the distance from am. We divide am in the same ratio, making mb one or whatever fraction
lay off
join

mn

mn

is

of

the

distance to the vanishing

and draw bn. Then the required line may be drawn parallel to bn, and if it were prolonged would pass through V. For suppose F in the figure, outside the drawing, to be the vanishing point, and a V The triangles mbn and maV to be drawn. are similar if mV and / are divided proportionally: therefore aFis parallel to ^//, and is the line we have constructed. The problem is a simple geometric one, and any line mn will serve, if only it passes through V if continued, and we know what part mn is of m V. The question is apt to come up with conjugate vanishing points, where one is far off. The measuring point of an inaccessible vanishing point is easily found when the
point,

conjugate vanishing point

angle

is

accessible, for the

MSM'
If

is

constant, being always 45, or

half a right angle.

Note.
triangle

we

call

MSM' gives, a = i8o-{SMM'+ SM M). Trianjjle SM'V is isosceles, therefore SM'V or SM'M = \ {r8o- V). So also SAfV or SMM' = i(/So- V). Then SMM'JrSM'M- 180 -\{ V+ V). But r + V= 90, and ^K+ V) = 43''. Hence SMM' + SM'M=i8o-45-i35, and a = 180 - 135 = 45.
Fig. 50 involves

(Fig. 49) the angle

MSM',

the

many

of the foregoing principles

shows

a flight of steps in three

is

spective

laid out

according to

and methods. The sketch plan ramps connecting terraces at four levels. The perthe sl<etch plan, the scale of feet in the margin, and

the length of Axis there shown.

picture plane so as to
at

Some

care

is

necessary to choose the angle

perspective chart
in
is

</>

of the

show

the steps well.

The

set at the top,

<^

a third the scale of the picture.

CVS,

and that again to CSV.

is

The

easiest

As appears way
to a

Fig. 51, the angle

is

equal to

to construct the chart

third of the Axis
flescribe

to set off

SC equal

and the angle CSV Xo <j>: then the semicircumference by bisecting the
or construct the
right

chord

VS,

angle

VSV.
the pic-

The corner
ture plane,

of the lower landing is in

and the Ground Line, or horizontal Ax, passes through it, being taken eight feet below the Horizon Line. The width of the picture does not take in the point V. J/ being also out of reach, we substitute M/^, a fractional measuring point at a thirl the distance \r\ the chart. By accident 't coincides from Fto Af, which corresponds directly to with M', The short ramp in the middle of the steps is the only one whose slope lines
scale

VM

36
vanish to the
left.

APPLIED PERSPECTIVE
Their vanishing point
to

F"

is

in the vertical

through F, but

M,

from which the slope angle

getting the point
off to the right

would be set oS, being

off the picture,

?', and set off FV" equal to three times Ff. from M', both above and below the Horizon Line, gives the directions of the inaccessible vanishing points V" and F"" of the slope lines of the ramps which ascend and descend to the right, these points being in the vertical which passes

we use M/^, The same angle co, set

through F.
Starting with A,

we draw

A V,

and measure

off

on

it

Ai

for the width of the landing,

using one third of the measure of length, for

To draw

AB

toward the inaccessible

to the

V
O
;

we must use the point Mj ^ we bisect AV, and draw a

for measuring.
line

from the

aO, and must point to (see Fig. 48), and the width of the landing to is measured off on the Ground Line by M'. Then F \s drawn, and 4 found by A4 which vanishes in the mitre point 8.- 1-4 is drawn and completes the landing. Points 2 and j are measured off like I, and uprights being drawn through them, the height of the corners i and c is measured off by the scale of heights at A, and the vanishing point F. The slope lines of the middle ramp can now be drawn from i and f to F". F'b marks the point 8, which gives the height of the second landing its edge 8-6 is found by bisecting 8 F,
is

middle point a

halfway point

then

drawn

parallel

to

middle point, and 8-6 parallel to this last. 8Fand 6F, by their intersections with the diagonal *8 and the upright from j, find the other corners of the second landing, whose rear edge can now be drawn. The corners of the steps might be found by measuring the widths on AFand projecting up on d8, but the construction would be crowded. It will be clearer to divide 68 or c6 perspectively.

drawing a

line to

O from

its

This

is

not a case of measurement, but of dividing into perspectively equal parts, so

line parallel to

we draw through 6 a
it

F"M,

any four equal

parts,

draw a/ through the

using the rule of Fig. 32, and laying off on last point ri^to find /on F"M, which will

giving the corners of the steps.

be the vanishing point for lines from /to the division points of d(/, that will divide 6c, 86 may be divided in the same way, and the front edges of the steps put in between, or the edges may be drawn toward F' by the process by which and 8-6 were drawn. The second method would be a little more laborious, but would be surer, for any irregularity in the double division of the slope lines would disturb the convergence of the edges. These once found, the other lines

AB

of the steps are easily found.

It

the upper,

remains to draw the ramps that ascend and descend to the right. Beginning with we set off 6-7 equal to the riser below it. F'" is out of reach, and we can-

not draw the slope 7-F'" directly: we set off from

to F',

M' on out third the distance marking the point F'jj, and draw the vertical which will cut off one third each on Af'F'" and M'F"". On M'7 we set off one third the distance from J/', and from the dividing point draw a line to F"'/J . the slope line from 7 to F'" is parallel to this. In the same way we draw 6-F'" and g-V". These slope lines are so nearly parallel
their inclined horizon
:

HH

to

M'F"'

that to divide

them

in

the ordinary

way would be

troublesome
long F8
1

'

it is

better to divide the horizontal projection 8-6 prolonged, and project

the divisions on 7- F"'.

till it

meets the scale of heights

line of

There are ten steps to be measured off on 7- F'". If we proAz in z, and draw ze parallel to Jf//, this will measures parallel to M' V" would lie so close to j-V" as to be embarrassing.

-T'

/
/
3-

/
Z'

it

t.,^._^.3Ct-:*
.^JCO^

^z^Te, :k

'O/f-

CONJUGATE VANISHING POINTS

be
in the picture plane,

37

plane of the landfrom M' at e, lay off from The e the measures of the ten treads, and project these back on 8-6 prolonged to 10'. The front edges, the lines points of division, plumbed up, divide /-lo as is desired. of the treads, and the risers are then easily drawn. The point /o at the top of the steps
line of line in the
it

and be a

measures for any

ing, in

which 8-6

lies.

We

project the lower point 6 upon

marks the level of the upper landing, and should come in the Horizon Line. It is determined by the intersection of lo'-io and 7-10, and will test the accuracy of our
construction.

The descending steps might be constructed by measuring the heights on the scale Az produced downwards, but it is easier to divide the slope lines than to draw so many
These lines are drawn like those of the it, and drawing a line from the upper division point to V""/j, and the slope line Bp parallel to this. Then by Ax we measure off the dimensions for the risers on Bm, as we did those of the upper ramp on 6-10', and plumb them down to j>.
treads toward an inaccessible vanishing point.

upper steps, joining

M',

for instance, trisecting

INCLINED PLANES
Fig. 50

shows how the sloping

lines of the

ground appear

of roadways and the

like.

V, shows the trend of those of the path. Those lines, so far as they are straight, vanish below the horizon, but higher than V"" because their slope is not so steep as that of the steps. In truth, as may be seen by comparison with Ax, they may in the drawing tend up toward the horizon, and really do so here if it were not for the contrast of the lines in the wall we could not know whether the ground at the base of the wall sloped up or down, or not at all. As a matter of fact, though lines in a landscape which have sometimes disfrom one side, that is a considerable slope and are seen obliquely play their rise or fall distinctly, the inclination of lines that slope away from the spec-

The

horizontal coursing of the terrace wall, with lines vanishing in the horizon at

tator can for the

walls, buildings, or the sea, for instance

most part be shown only by the contrast of horizontal lines, those of and the same lines of road or terrace may be
;

made

to

seem

to slope in different ways, or not to slope at

all,

according as the horizon-

tal lines

which combine with them are differently arranged. This is illustrated in Figs. 52 and 53, where the lines of the roads are the same, one being traced from the other, and only those of the surrounding buildings show in the one case that the road runs up hill,
in the other that
it is

level.

Of course

the change in the lines of the buildings indiit

cates a change in the Station Point, because

that
is

indicates a change in the horizon.

Where

clearly visible

its

level line is
;

sometimes enough
horizon
is

to indicate the direction of the

slope,

though not always

but

in pictures the

very apt to be obscured.

In any view in which the ground slopes away from the spectator the
parts appear higher in the picture than the nearer as appears from Fig. 54.
:

more

distant

Here,

6'

if they did not, being the Station Point, and /'/'the picture plane, if a is the brow of a slope or the edge

they would be hidden,

of a terrace, nothing of the

descending

ground

will

be

.seen

which sinks below the

^'^ "^^

'

-^

,^'

visual line
into view

Sam, and anything that comes beyond m will appear higher

is

than

a,

as the point b
b'

projected into

the picture at

above

'.

We
a
line or horizon of that plane,

have seen that

all lines

that lie in

plane must vanish in

the vanishing

pass through the vanishing points of

and conversely that the vanishing line of a plane must We have then a convenient all the lines in it.

means

of determining the position of lines in inclined planes.

to

draw an oblique pattern in an inclined plane, as in Fig. 55. The V and are accessible, though off the page. The slope angle of the inclined plane bed is set off at the proper measuring point M', and finds The vanthe vanishing point of be at V", also accessible, in the vertical through V.
Suppose we have
conjugate vanishing points

m
r
V
\

['

^f

./v,,,

If

=-

,--" -^C.

HL

'4'

-^

l-r'.\

/if <^^ \ i
1

/
' 1

Ij

/'

^d SSI KJ
V-'-'^

1.

^
A

'

v*"

"vl!

INCLINED PLANES
ishing line or horizon of
tfiis

'

39

plane, which contains be and cd, must pass through their

V,

two vanishing points F'" and

and

is

the line

marked

to

V and

to P"",

a part of which

appears at the top of the page.

The
than

principal dimensions on the lines that vanish at

1>F, to

Fare measured on
those on
l)t:

^F rather

avoid crowding, and thrown up to

of do/.

1>F,

but
as

are measured in their

own plane

M'

is

the measuring point of

is

6c,

well as of A/, and drawing a line of measures 6^ parallel to M' F'", be Af', and the square top of the block and its square panel are outlined.

measured by
Its

vanishing
in

diagonal
is

is

in the inclined plane,

it

and

its

vanishing point, being therefore

VV",

found by producing

to that line, at O.

The middle

points of the sides of the panel

:

are found by crossing diagonals, and the diagonal square put in

in O.

two of

its

sides vanish
;

The

pattern divides the sides of this square into three equal parts
is

and

their

actual length
line fine

not important to us, but they must be set ofl by a point

take
to

in

the vanishing

mn of it that is intercepted between the and by measuring lines to O divide the other two sides. This gives us two lines, vanishing in O, which define four points of the star the other points and the sides of the four little squares between the points follow easily. Those sides of the star which trisect the sides which vanish in O are nearly parallel to VV", in which their vanishing point would be found. This point would be conjugate to O, but
Vy'" we
trisect that part

VV". We mn parallel

for a

measuring point, and drawing any convenient measuring

sides which vanish in O,

and it is not necessary to us. shows the same principles applied to the roofing of a house. The horizontal eaves of the main house vanish to the left in V, the inclined ones to the right in F'" above V: the plane of the visible slope vanishes in its horizon FF'". In the same way the visible slopes of the L and the dormer vanish in VV". The walls and The eaves of the gables horizontal eaves are constructed by the ordinary methods. are drawn upwards toward their proper vanishing points the peaks may be found by plumbing up from cross diagonals below, or by measurement the eaves which descend toward vanishing points below F and F' may be drawn without these points. The short valley where the main roof stops against the wall of the L is perspectively parallel to the slope of the main gable, vanishing in F'". It stops and is stopped by the horizontal eaves of the L. The oblique valley which joins it lies in the planes of both roofs, being their intersection its vanishing point, which must be in both their horizons, can only be O, the intersection of these horizons. It stops and is stopped by the The valleys of the dormer are treated in the same way. The dormer ridge of the L. and chimney may be constructed with the rest from a perspective plan, or by direct measurement. The heights, measured on the vertical scale at a, may be transferred to a vertical at c by lines vanishing at F', and again to the left in the plane of the front of the dormer, by lines vanishing at F. The positions in the roof, measured on aV, are projected up the wall, and up slope of the roof by lines vanishing at F'". The chimney is constructed in the same way. In Fig. 57 is a further application of the same principles. The arrangement of the dormer in the roof and the band of slates behind it are shown in the marginal sketch. The plane of the main roof vanishes in V", its eaves and ridge in F', its slope lines in F". The plane of the dormer roof vanishes in VP' '". The near angle of the dormer is the scale of heights at A. The widths are measured on a horizontal line of
is

inaccessible,
Fig. 56

40
measures
at

APPLIED PERSPECTIVE
A, by means of M', the overhang of the gable
at its

own

level

by M.

The

valley against the side of the dormer, being a slope line of the roof, vanishes to V".

The pitch of the gable is first drawn in the plane of the window, and then projected forward to the plane of the eaves, the peak found by diagonals below, and the ridge drawn through it, vanishing at V. As in the last example, the vanishing of the oblique valley is C, the intersection of the horizons /^^'"and V'F". The ornament of the slating
is

constructed, like that of Fig. 55, by

is

construction of the gable

means of the vanishing points V" and O. The shown more clearly in Fig. 58. Only the projection at the near lower corner is measured by M, the other points following from this. The height of the main ridge is determined like the other heights *5-:j by the vertical scale at a. The ridge of the dormer stops against the valley that vanishes
at

O in
It

Fig. 57.
all

71^

-i'S

should not be forgotten that of

pitch.
all

the

'i^

lines in an inclined plane

is

one set of parallels

its

the steepest, indicating

These
the hori-

slope lines are at right angles to

zontals of the plane, and so to
its

intersection

with the ground plane and other horizontal planes, as the rafters of a roof are at right
angles to the horizontal eaves.

r"
.'

-ut-

\\.

IFi^.ss

j^
vir
.

-^-r^

>*'*
J>' -S-i

TZcU^ 32Z

CIRCLES
The
perspective of a circle
is

not a circle except

when

its
it

plane
is

is

parallel to the

picture plane, so that like any other figure in such a plane

similar to
itself,
is

projected in a figure

and then

it

may be described

with compasses.

In other positions

its

perspective

usually an ellipse, and must be determined point

by

point.

The

easiest
in

way

to find points
first.

enough

is

to enclose the circle in a square,

and put the square

circle

perspective
Fig. 59

shows two ways of

utilizing the enclosing square,

one

being

in

a verti-

cal plane, the other in a horizontal.

The

position of the square being arbitrary,

we

take one side in or parallel to the picture plane; the sides at right angles to this are
horizontal,

and vanish

in

HH

those of the horizontal square are Perpendiculars and

vanish in C, those of the vertical square are oblique and vanish in V, whose measuring
point
ures,
is M. The sides ab and cd oi the two squares being set off in the plane of measand the vanishing sides measured by J/ and respectively, in the vertical square

diagonals are drawn, as

is

seen in the smaller plan

at the side,

and two horizontals

This construction,

through the points where the diagonals intersect the circumference.

put into perspective, gives us eight points of the circumference
of the sides of the square, where the circle
is

the four middle points

tangent, and the four intersections of the

the position of the curve but

diagonals and horizontals.

its

The tangents show not only

and so are a double help. These eight points are usually enough to enable the draughtsman whose eye is well trained to draw the circle satisfactorily unless it is pretty large. This method uses few lines in the construction,
direction at the points of contact,

but has the disadvantage of requiring a plan to determine the four intersections, for
which, however, a half-plan
side ab, as in the figure.
is

sufficient,

and
is

this is conveniently set against the front

is

The

point where the diagonals cross

the perspective of the

centre of the circle and of the square, but as

is easily

not the centre of the perspective ellipse,

seen.
in the horizontal circle,

The second method, used

points

has the advantage that

all

the

maybe found

directly without the use of a plan.

The

square
is

is

divided into two

rectangles by the front diameter ef; and the front side cd

divided into quarters.

Diagonals
points of

(r<y

^and edare drawn in the to ^ and/ will intersect

The
plan,

nearer rectangle, and lines drawn from the quarter

these diagonals on the circumference, as

maybe
and

geometrically proved.
four intersections.
struction.

Here again we have

which
is

eight points, the four tangent points

is

given for illustration,

not used in the con-

The construction shown in Fig. 60 is convenient for describing a series of circles or round arches in contact. Thus, the first arch being constructed, an intermediate diagonal drawn through d and e' finds the crown c' of the second arch and the vertical c'd'; a second diagonal through b' and 0' finds a", giving the data for the second arch, and so on, the whole with very few lines. The process is analogous to that described on

"

. .

42

APPLIED PERSPECTIVE

page

19, Fig. 21, for

spacing a colonnade or other

60

series of uniform objects.

Tiie symmetrical shape of the circle allows a

square to be circumscribed wiih

direction,

its

sides in

any

and there
is

is

usually no obstacle to put-

ting one side parallel to the picture plane.

The
is

only question
inclined,

the

when

the plane of the circle

question of the vanishing point

and measuring point of the oblique sides of the square, which are in the horizon of the plane, and are determined by .the methods provided for inclined planes.

This

will

be considered

later.

We

get a surer control of the curve by combining with the method of Fig. 59 the double square of Fig. 15, as in Fig. 61, where the
8*

diagonal square gives tangents at the 45 points,

* -

^ - '

if
- . -

^
* ^
*

__/\

so that

we have
<,

eight points
its

and eight tangents.

radius at 45 fixes
is

^i^\^ /> \
1

The

auxiliary quadrant with

the point

and the diagonal square

drawn as

\
f<^

-^
i^ \

--+--.

K...\

before.

The diagonals
is

of the first square give the

tangent points of the second, and the direction of

the curve ing point

shown

at all the points.

The

vanish-

may be

the sub-centre corresponding to

in

an oblique plane
its
*

which the

circle lies,

and

CD

horizon,

being the sub-distance point, and

that

the construction will remain the same.

It

may happen

instead of having a front

tangent ab to begin with we have a front diameter at f. Then we have only to bisect the diameter at 0, draw our Perpendiculars

i^.

6z

as through f, and the Diagonals through 0, and so find a and b.

Or

the vertical distance points

out of reach (Fig. 62).

the 45 tangent points

may be Then we may find

d' in the usual

^and

way, as in Fig. 59, and draw through them the sides edh and ed'h' of the diagonal
square, which
is

then easily completed.

When we
through a
wall,

wall, the

have to draw a round arch cut curve at the back of the

in

being geometrically equal and parallel a plane parallel to the plane

to that in the face,

is

to be constructed in the
it

same way
is

of the face,

and behind
is

at a distance equal to the thickness of the wall (Fig. 63).

The

soffit

of the arch

a cylinder whose axis

horizontal

and perpendicular

to the

face of the wall, and so vanishes in

for the rear curve
is

V coordinate

with

VK

Every constructive point

to be found directly behind the corresponding point for the front

CIRCLES
curve, at a distance equal to the thickness of the wall,
ishes in V.

43

and on a horizontal that van-

In Fig. 63 the front curve

is

constructed by the method of diagonal squares

shown

in Fig. 61, abed being half of the

e',

primary square, and gjh of the diagonal square.

The
on

point

corresponding to

ee'

vanishing in V.

distance points are at

found by measuring off the thickness em of the wall Other points may be found in the same way, but if the vertical hand, may be fixed by their help as here by lines e'g', e'h', etc.
e, is

more simply and securely than by measuring

curve, which in perspective
is

off

each one independently

not exactly parallel to the front curve, the landing of the steps in Fig.
of the voussoirs seen
;

means

of them.
in

structed

this

way.

The doorway at The joints

and ihe rear traced by would be con' 43

;

is

on the face of the wall which meet

all

radiate from the centre of the arch

V,

those on the

soffit

vanish in the vanishing point

in the line 00'

and the surfaces

of the joints are in a series of planes

that joins the centres of the arch lines.

by points at interdepends much on the draughtsman's eye, for to construct points close enough together to leave nothing to his judgment would be very laborious and uncertain. It is only by careful attention that one learns the curves sufficiently well to draw them satisfactorily. Most painters and most draughtsmen, it is safe to say, do not give this attention. The ovals in which circles are projected have a great deal of beauty in their delicately graduated curvature, especially when they are much foreshortened, and the eye -^ that is not attentive to this beauty is not competent to deal with them. So it is that comparatively few painters put into their pictures an arcade or a round tower banded with string-courses at which an eye
of circles in perspective, since they are constructed
vals, necessarily

The drawing

trained to such things can look without disappoint-

ment, or even draw a bell or a

mug

void of offence.

The

relation of the successive string-courses of the

is full

round tower Chaumont).

is

The

common

to

unseen by those who paint them (Plate XIV, merge into the vertical lines by delicate tangent curves, yet it see them drawn with a corner as at A in Fig. 64, and to find the cornices
of grace which is often

rings

one above another so spread as to tilt at all sorts of angles wjth the horizontal. It is common, even among artists, to believe that the perspective ellipse of a
tical circle

ver-

stands upright, that

;

is,

that

its

longest diameter

is

vertical,

and

that of a

horizontal circle level

of the vertical circle

but this

is

true only

when the centre

^^
|

is

on the Horizon Line, and that of the

In other posi-

horizontal directly over or under the Centre.

tions the long axis
is

\\
y
.

<

neither vertical nor horizontal, but inclined

7i^

iS"

toward the Centre without pointing to it, as in Fig. 65. This inclination is often disagreeably conspicuous when the circle is far from the centre, especially in vertical circles which are enclosed between upright lines, as in a clock tower. Examples

may be

That from the Broletto of sometimes incredible to artists who have not studied such cases. It is an example out of many
seen
in

many

photographs.
is

Como

is

one

(PI.

XV), and

44
to

APPLIED PERSPECTIVE
show us
that nature
is full

do not look well in pictures. It is perhaps change slightly the inclination of the axis rather than draw what is disagreeable to most eyes ; but a better remedy is to avoid, as most painters would, a point of view which leads to such appearances of
of things that
well in sketching such views for picturesque uses to
distortion.

The curves

of perspective circles are the

more

sensitive to distortion in that they

where the curvature is sharpest, are the most troublesome parts, and it is often of advantage to get additional points and tangents But so much depends on the guidance of the eye in the curves, so at these parts. much labor may be saved in constructions that are not critical by reducing the number of points that have to be found, and so much of the beauty of a picture in which
are symmetrical.

The ends

of the ellipse,

prominent depends on giving the right character to the curves, that it is important to discipline the eye by the habit of drawing circles from nature till their character becomes familiar. There will still be enough to do in construction if we wish and no one knows this so well as he who best to get the curves into the right places
circles are
;

knows what a perspective

ends of

was just said, are the and the temptation to make the ellipse, especially if they are them swollen, as at A, Fig. 66, or pinched and sharp, as at , is more For this reason it is desirable to besetting than one might imagine. multiply points in these parts more than elsewhere, when the size of the circles gives occasion for it, and to add tangents, which show not It may be thought that only the position of the line but its direction. a fair experience in free-hand drawing would make a draughtsman
circle is like.
critical parts, as

The much

flattened,

independent of constructive methods

iarity

in

this

but a moderate familis

with perspective will convince him that this

not the case.

On

the other hand, even

when the points

trifle this

way

or that from their precise position,

may depend on
times asked
:

the thickness of

the draughtsman's sense of line.

What

is

the use of
?

sway a whose flow Here the last refinement must come from a line. The question may naturally be asked, and is somemeasuring and constructing if all this is not more
are found with extreme care they will

enough

to

mar the

curve,

exact than sketching by the eye

applies which

But the eye, which detects variations

in continuity

that are too small to measure, is yet an uncertain judge of position.

is

Here
is

the paradox

true in all fine drawing,

is

ior placing curves the eye

an inadequate
is

guide where precision

not fine enough.

necessary

for actually

drawing them measurement alone

PL.

XIV

CHAUMONT, CHATEAU

THE CONSTRUCTION OF PICTURES

We
some
have seen that the effects of perspective are due to foreshortening by oblique
projection on the picture plane, and that variations in this foreshortening are main

sources of the interest

we take
fully,

of these foreshortenings are agreeable,

possible to classify
conditions.

them

things which we depict. But and some are not ; and though it is not we can note some of their favorable and unfavorable
in the aspects of the

Perspective foreshortening affects only lines and surfaces which are oblique to the
picture,

and these unequally, the farther parts being diminished more than the near.

We
it

have seen how fast a horizontal surface like a floor or the sea shrinks together as retires into the distance. We have seen that the length of the Axis repeated behind

the picture covers half the distance to the extreme horizon, which

may be

miles away,

so that the last mile

may be shrunk

to the thickness of a line.

In like manner figures

and

trees, or the

windows and columns

and mystery

of a building, are to the eye

crowded together
is

as they retire faster than they diminish in size.

tage, lending intricacy

This may

in

some cases be an advanit

to the picture, but in others

a disadvantage,

interfering with clearness, especially in architectural subjects,

and lessening the appear-

ance of distance.
arcade, in the

We

notice often the unsatisfactory effect of a long colonnade or

gappy look of the near parts contrasted with the crowding of the distant. might be thought that this crowding of far-off objects would add to the impression It of distance, but it is not so. If we examine a photographic view down the long nave
of a church, or a pillared aisle, side

we shall find that, unless we can count the pillars, the on which they are crowded close looks shorter, and that the end on the other
off.

side looks farther

This fact helps

used
to play.

in certain tricks of

perspective that the architects of the Renaissance

is

In Palladio's Teatro Olimpico at Verona the stage

set with a street

seen endwise, lined with models of houses which diminish in scale as they recede

from the
mile.

front, so that a street forty feet effect

is

long or so seems to extend an eighth of a

it

The

greatly increased by the narrowing of the street as

retires,

by

which the buildings are brought forward and displayed instead of crowding at the far Exactly the opposite effect is shown in the colonnades which Bernini prefixed end. Here the galleries on each side which lead up to to the front of St. Peter's at Rome.
the front, enclosing an open court, spread apart as they approach the church instead

and the columns crowd together in the view much more than if their Consequently the colonnades are robbed of their apparent length, and it is difficult, as may be seen from the figure in Plate XVI, for the spectator to Viewed from believe that the court is as deep as it is wide, which is really the case.
of converging,
lines

were parallel.

the church, however, the colonnades seem lengthened out.

It is

not only the crowding of parts that mars the effect of long lines reaching into the
:

distance

not only are the farther divisions diminished by foreshortening, but the rate

46

APPLIED PERSPECTIVE
tlie

of foreshortening increases rapidly,

farth-r parts being seen

much more
it

obliquely

than the near.

the foreground.

If

an arcade or colonnade starts near the front, where

not
felt

looiis very

open, the foreshortening increases so that the distance often looks out of harmony with

This dissonance, which

is

where the

line is

broken by a marked

angle, because the eye does not expect broken lines to agree in foreshortening,

may

be

a disfigurement where the lines are continuous, whereas the abrupt change at an angle

makes an
lars
it

effect of contrast

which

is

apt to be pleasing.
is

The more

nearly parallel the

lines are to the line of vision, the

more extreme

the foreshortening.

In Perpendicu-

lines

becomes excessive as they are prolonged to .near the Centre, where the vanishing are viewed very obliquely, and that is one of the difficulties of parallel perspective.

In this kind of perspective there is some disadvantage in a long Axis, as appears in Fig. 67, where, from the Station Point S', the perpendicular distance Ax is foreshort-

ened into the length

4e
1

Ai7,

while as seen from

it

occupies

Ali.

But on the other hand, inasmuch as the Perpendicular vanishes

at C, all its length
>, 1%

beyond

x",

seen from S,
it

is

foreshortened into
the advantage

the length CV, while from S'

covers Cd.
is

Thus

>

of a near Station Point and short Axis

confined to opening

out those parts which are near the front, while the disproportionJC

ate compression of

tlie

distant parts is aggravated by

it.

The

advantage
' i\
-o

is

with the long Axis unless the important parts of the

picture are near the picture plane.

X>
yi

The opening

out of the dis-

X
s

tance of a picture
with
it

is

achieved by raising the Horizon Line, and

;

the eye, which displays the floor, so to speak

or

it

may

be opened on either side by moving the Centre to the other side, as we have seen in looking down the colonnaded aisles just mentioned.
It

may be

noticed also that in parallel perspective the impor-

'

tant lines are the Perpendiculars, for they alone S' of perspective, so that this construction
teriors
is

show the

effect
in-

used chiefly for

where these lines prevail. The lines which are at right angles with these being front lines, the parts of the picture which are constructed on them are seen in geometrical elevation, and are comparatively uninteresting. They are kept subordinate therefore, and appear best near

and

street scenes,

the Centre, where they are in the distance and the parts are small in scale, and are

moreover apt to be framed in by balanced Perpendiculars on each side. In the foreground the front lines come naturally into the margins of the picture plane where they They are therefore troubleare unbalanced, and subject to distortion, as we shall see. some, and as a rule the less they are shown the better. Indeed, it is very common to In interiors cut them off altogether in this position, and with advantage to the picture.
the depth of the view
parts which
is

usually comparatively small, so that the displaying of the near

a disadvantage.

comes with a short Axis may not be

In other forms of perspective, where the principal lines are not Perpendiculars, extreme changes of foreshortening are not less unbecoming and wheti, as someti.mes
;

happens, convenience requires us to represent apart of onr picture that stands before the Fig. 68 shows how equal lengths al> and picture plane, it may become very disturbing.

> X

6
s

o u

THE CONSTRUCTION OF PICTURES

cd

47

may be

projected from different Station Points with very different rates of fore-

shortening.

The two

projections ay and ox

from

5 differ

very much, while those from

S', ay'

and ox\ are more nearly equal.

The
71^.

variations seen from S, confined in the pic-

68

y and o, would probably look extreme; those seen from S', between _j'' and x', are likely to be more agreeable. On the whole, the longer Axis gives
ture between

almost always the more agreeable

effect.

Drawings made in Isometric Projection, which is perspective with the Station Point
at

an

infinite

distance, show, to be sure,

the

monotony

of division which an inordi-

nately long .^xis

may

give; but the practical

is safeguard enough, and the ugliness of isometric due to the equal angles of the three sets of lines, and to their lack of convergence, which disappoints the eye, still more than to monotony of foreshortening. Drawings with two or three vanishing points are apt to be built on conjugate points,

difficulty of using

a very long Axis

drawings

is

so that one point recedes as the other advances.

series of lines with equal convergence,

We

have noticed before that vanish-

ing points at equal distances from the Centre are unsatisfactory because they give two

which

is stiff.

The

stiffness is

made more

dis-

agreeable
it,

if

a building or other rectangular object is so set that the

Horizon Line bisects

fault is

giving equal slopes to lines above and below the horizon.

No

commoner

in

photographing than to take buildings from too high a point of view, especially the interiors of churches, which are often spoiled in order to humor the infirmity of a poor
lens,

by taking the camera up to the triforium, and showing half the height below the Even exterior views are constantly marred by the necessity of carrying The the camera close to a building and so bringing the vanishing points near together. recent invention of telescopic photographing relieves much difficulty when it can be used, not only because it obviates various distortions due to short axes, which will be
horizon level.
presently noticed, but because the remote vanishing points of a distant view are very

becoming

to buildings.

Whatever
is

is

true of photogiaphic views

It is well

is

true of perspective

drawings, because photography

perspective.

in pictures of buildings to

make

the most of their horizontal lines, for on these depend the breadth and repose of

their architecture.

design tempt one to emphasize the vertical lines, we

In placing the picture plane
esting side of the
effect

Even when picturesque grouping or the upward shoot of a Gothic may remember that a near vanishits

ing point and steeply converging lines rob the building of

it

height.

to the narrower

by giving the and the contrast

:

show most of the more interbuilding; when other things are equal we are apt to get the best long vanishing lines to the broader side and the foreshortened view
is

naturally desirable to

of a very considerable difference in the distances of

will do well to rid himself as early as possible and learn to use vanishing points outside it, or to employ some of the devices which are provided for drawing lines to inaccessible vanish-

the points

is

of value.

The student

of the tyranny of his sheet of paper,

48
ing points.

APPLIED PERSPECTIVE
When
he
is

constrained to keep these points within fixed limits he will find

as appears in Fig. 69 that for a given distance

VV between
is

conjugate points, since the

Station Point must be on thesemicircumference FSF', the longest

Axis

is

got by putting the Centre in the middle of

VF'

at

C;

for the

Axis SC, being the radius,

In this case

longer than any other half-chord S'C.

distance points,

Kand

V become also
There
is

CFand CF'

being equal to SC.

construction,
it.

some

practical convenience in this

and a great many architectural drawings are made by

is

The

increased length of the Axis

is

worth something, yet the

difference

usually slight, and there are losses which outweigh

the gain.

If the chief

object

is

put at the Centre, the lines slope

may be made unequal, it is true, by setting the object aside from the Centre. In Figs. 70 and 71 the same cube is shown under two conditions. The distance FF' is the same, the cube is set with its front edge in the same position between them, and at the same height on the Horizon Line, so that
equally both ways.
slopes
the lines slope alike in the two figures.

The

But the different positions

of the measuring points give quite different proportions to the two

In Fig. 71 the object, being at the Centre, is likely to be viewed from the and to look natural. In Fig. 70 it is on one side, and likely to be viewed from the wrong point ; the farther from the Centre it is, the more danger. There is a
pictures.

right point,

certain distortion in drawing the object in this position which

If
it

we

shall

have to consider.

not suffer by the change; but if it is more complicated it will probably look deformed. Figs. 72 and 73 show the same cube prolonged into the shaft of a square tower, and capped with a pyramidal roof. The effect of foreshortening is in Fig. 72 to drag the peak and the
is
it

only a cube, or other simple rectangular form, as here,

may

toward the middle, and the roof does not sit well. In Figs. and 75 the angles of the tower are cut off, turning it into an octagon. In Fig. 74 74 not only is the axis of the tower drawn toward the middle as before in Fig. 72, but the broken line of the eaves is unpleasantly distorted, seeming to be out of level. Yet the
vertical axis of the tower

tower

is

not very far from the Centre, the distance being only one third the length of

the Axis, and so quite within the limit

commonly

allowed.
will

If

we
to

fix

the eye at the

into
its

proper distance just in front of

C, the axis of the

tower

seem

draw back
but
if

proper position and the awkwardness of the eaves line

is

to disappear,

the tower

the principal object in the picture,

it is

not likely to be looked at from this position,

the universal temptation being to put the eye directly in front of the thing

In the round towers. Figs. 76 and 77, the distortion lieved by vertical lines.

is still

we look at. more unpleasant, being unre-

 D O u H C

as

H H u

0,

PERSPECTIVE DISTORTION
is brought against perspective is that an object of known form, by it, will sometimes look wrong, like the eaves of the octagonal tower in Fig. 74, and even seem offensively distorted. It is indeed inevitable that when the eye leaves the Station Point the foreshortenings, adapted to the old point of view, should not suit the new. We need to bear in mind the distinction between graphical foreshortening and visual. The first occurs when the lines, being oblique to the picture plane, are projected on it sinaller (or larger) than they would be if they faced it; the second when lines, being oblique to the line of vision, look shorter than they

An

objection that

carefully constructed

would

if

they faced the eye.

The two may occur

together, or either without the other

for a line

may be

parallel to the picture plane

e,

and oblique to the line of vision, or may front the eye and be oblique to the picture plane, or it may be oblique to both. The difficulty is in the fact that the picture, once made, is itself subject to the laws of perspective and is visually foreshortened in

,'^

'
',

"^
_'

^'

"^/^
-^
,
1

'f
-

!^
'
'

\
* ^

S
,
'

"^

'

1/
'

an oblique view.
its

'

Its
^

\'

construction provides duly for

foreshorteni
^

'*''''
>

"

'

'

ing

when

it is

seen from the Station Point, but

)'
'

'

not from any other.

the eye

Only the part which

is

directly in front of the eye, the Centre

is

when
it

"V -.'^ /,'^'

,

\'*\ [^'

^/

'

'

at the Station Point, is

exempt from

'<''
,

^'^/

.^\i

this foreshortening,
in

and

all

other parts suffer

proportion as they recede from the Centre,

because they are viewed obliquely.

than
is

This

is

to say that

from the Station Point every

other words,
is

part but the Centre looks smaller than

it

it is

drawn,
it

or, in

intended to look,

in

proportion as

recedes from the Centre

drawn larger and the

;

perspective construction provides for this.

Inevitably,

when

the eye leaves the Station

Point the parts which come opposite

radiating from the Centre.

it

are seen to be stretched out in directions

ing the Station Point at foreshortening of

In Fig. 78 the two equal horizontal lines ab and cd, frontequal distances, are not visually foreshortened, but are projected

into the picture plane in two lines

hg and fd which are quite unequal.

.S

The
if

visual

/^

corrects the inequality, so that from

they subtend equal angles

in the picture,

and look
is

of equal length.

But when once the

is

lines are

drawn

the eye

shifted to 5', opposite /', not only

the line seen to be longer than gh, but the visual

foreshortening due to the

tion
is

new

position

makes

^/^

look

still

shorter,

and

their dispropor-

aggravated, the angles which they subtend being changed, so that noticeable
is

distortion
It is

likely to follow.
itself

obvious that, the foreshortening in the picture

being everywhere

in the

'

APPLIED PERSPECTIVE
.S

direction of the Centre, to the eye at

the whole picture crowds in toward the Centre,

looking smaller than

surface
does.

it

really

is,

as every flat

The

perspective construction
in lines

meets
in the

this

by expanding everything

that radiate from the Centre, so that objects

margin of the picture are drawn larger

/'

^
'

^
^

than they look, being stretched out away from

the Centre, as
if

by a centrifugal

force.

This

^^

stretching out, seen from 5, exactly covers and

neutralizes the visual foreshortening, so that

-^^/^

the picture looks right, but as soon as the eye

leaves S^

and passes over

to

one margin, we

perceive the distortion there, and the objects

shown there look misshapen.

is

sphere, for

example, which always looks round to the eye,

projected, unless
it

comes

right in the Centre, into an ellipse, so that

its

long axis

points to the Centre.

ring of spheres centred

upon the Axis would be constructed

all

in the picture as in Fig. 79,

but to an eye at the Station Point

these radiating ellipses

would be foreshortened into circles. It is this stretching in one direction that distorts the pictures of detached objects: all directions they would only be magnified. Inasmuch as the distortion if it were in is radial, objects in the Horizon Line have no vertical stretching, and those above and below the Centre no horizontal. Equal spheres, standing in a horizontal front line, are drawn of equal height, but their pictures grow wider and wider as they recede from the centre, and so with a range of columns or a rank of soldiers: they grow fatter toward the margin of the picture. A file of figures climbing a ladder would grow taller The increase would be as they rose above the ground without getting any stouter. mote rapid as the Axis was shorter, and would be visible to an eye at the margin of
the picture.

So long

as such objects are clustered about the Centre, the distortion

:

is

not great enough to be disturbing

sufficient precaution

a long Axis and a narrow picture are in most cases

against them, but even within the ordinary limits they will some-

times give offence.

We

have seen

in Figs.

of the Centre for the sake of giving

false

74 and 75 how the moving of the tower out unequal slopes to the vanishing lines has given a
will

look to

it.

It is

impossible to hold the spectator's eye to the Station Point: the

give offence.

eye will wander, and the picture in which such things occur

The questions
use
it ?

naturally

occur:
?

How
if

are

these distortions
correct

avoided

in

pictures

sketched directly from nature

And

we cannot

them by

perspective,

why

When we

look at an object or view as a picture,

in front of

we

instinctively refer

it

to a picture
If

plane squarely
line of vision,

our eye, that

is,

at right angles to

our axis of vision.

it

we

hold up a sheet of paper to

look

test directions or

dimensions,

we hold

right across the

and change its position and direction to suit every successive object we do not think of a picture plane held rigidly in one position and everything referred to that, whether the lines of vision strike it squarely or obliquely. We look at things not simultaneously but successively, and change the picture plane every time we
at.

We

J^i^./6

^'^77

/r
Ifi

z-.^./^

;^

V/^

-^

.^
-.^"

<'
j:^

j/zox^ikee:

PERSPECTIVE DISTORTION
turn our eye.
as

51

turn from one point to another, changing the plane end not one projection, but a series of projections on different planes. A man who paints a panorama turns from point to point, sketching as he turns, till presently his back is turned where he faced before, and he keeps on turning with his picture plane until he has gone round the whole circle of the horizon and come back to his first position. The sketcher, working from nature, does His picture is a piece of a panorama. If we look high just this so far as he goes.
sketch,
is in

When we

we

we

go,

and our picture

the

up instead of turning

horizontally, the effect

it

is

similar: as
;

plane slopes toward us, overhead

is

horizontal

if

we look higher the we painted the whole

picture
sky,
it

would come down to the horizon behind us. The innumerable fragtnentary picture planes on which our picture is thrown would if put together make not a plane but a When the panorama, which is horizontal, is finished, sphere, with our eye at its centre. viewed spread out, but must be rolled up into a belt, of which we occupy it cannot be
the centre, or else unrolled before us bit by
bit.

If the

panorama instead

of a belt

would be a hemisphere, and could not be unrolled. Sketching, too, unless it is confined to the neighborhood of the horizon, is in fact drawing by spherical projection. But to construct perspective drawings by spherical projection is impracticable, and they would be impracticable for use if they were constructed. They have to be made on a table or a drawing board, and kept flat. Sketching,
took
in

the whole sky,

it

which accommodates the projection unwittingly to a flat surface, is not scientifically exact, and is discontinuous, though the dislocations which it occasions are often less conspicuous than those which perspective shows to the displaced eye.

we

be done with perspective drawings under these conditions ? Shall or put up with their shortcomings ? The treatment of isolated Where their principal dimension has been determined, their objects is not difficult. outline may be modified to suit their known proportions. A sphere is drawn as a A human figure or an animal is always drawn circle because it always looks like one.
But what
reject
is

to
?

them

A circle, with a free hand ; the only use of perspective for it is to determine its scale. which as we have seen often takes a shape that looks queer, owing to the unexpected inclination of its axes, may be modified so as to bring it into harmony with the straight lines about it: at the same time the draughtsman should accustom himself to the looks of circles in nature, so that he may not be disturbed at seeing in a picture what Some concession may be made to the common he would see in nature if he watched. Other the axis of a circle upright, but it is easy to make too much. desire to see
may be first constructed and then if necessary modified so subdued into unoffending form. Since it is the eye that protests, the eye is but naturally it needs to be a trained eye. the only guide in modifying When it comes to objects that are not isolated, but parts of larger wholes, or to subjects like architecture, which involve large combinations of lines, the questions are more difficult, for changes in a detail may throw a combination out of gear. Separate
objects of geometric shapes
as to be
;

features like capitals and bases, which

that their perspective shall be amended, as

for instance, or in

may come in unfortunate positions, often require we may see in photographs, in Plate XV The amendment perspectives constructed without that indulgence.
;

naturally it should be as slight as is a matter for knowledge, skill, and judgment can be accepted, and with care not to injure relations with the rest of the building.

'

52

APPLIED PERSPECTIVE
starved

Columns and balustrades which tend

may be

down

to their
it

vals between

them unduly

to obesity as they draw away from the Centre due proportions, and as this is likely to widen the intermay sometimes be well to shorten the whole range, if it

can be done without doing harm elsewhere.

But the most effective remedies are

in rightly

arranging the picture.

We

can restrict

the area, at least that of the parts which require perspective construction, to the neigh-

borhood of the Centre. We can attract the spectator's eye as much as possible to the Centre whether it is the middle of the picture or not, and occupy the remote parts with forms in which perspective is not important. The greatest embarrassment is found in large architectural views, as where connected buildings fill the picture. In
such views,
will
if

the visual angle of the picture

:

is

great, the relative size of the parts

be changed
if

the proportions of the ends

is

of'

buildings
if it is

may be
The

unpleasantly dis-

turbed

the picture

wide, or of the upper parts

high.

Plate
fault

XVIII,
is

of the
in

castle at

Urbino, shows the distortion of a high picture.

is

not

the

photograph, but in the point of view, which

so near that the top of the towers, being

three times as far from the Centre as their bases, are too far off for safe perspective,

and cupolas are most unpleasantly distorted. An analogous vioamounting to distortion, may be done to the proportion of a building, even when the whole is close to the Centre, if the Station Point is so near that the far-off parts are much diminished by distance, and some part which should be dominant is
and
their cornices
lence, not

reduced to insignificance, or parts that balance each other are made plainly unequal, as in the Treasury Building, Plate XIX. The remedy is in a long Axis or a change of view. Photographs taken with a telescopic lens are often a useful lesson in this
respect,

showing how the gently converging

to architecture.

lines that

go with a distant Station Point

tranquillity that is very

preserve proportions, and lend an expression of unvexed

becoming

In discussing parallel perspective, we noticed the disagreeable effect that sometimes

comes with the contrast

that the perpendiculars

of perpendiculars vanishing to the Centre,

and uniting with

front lines which are parallel.

When

the meeting

is

well to one side of the Centre, so

visibly obtuse angle with

do not converge sharply, but make a

the front lines, the effect
ant.

is

very apt to be unpleas-

The

eye, used to lines at right angles vanishis

ing both ways,

almost persuaded that the front

lines actually diverge

'
>
" ,
',

from the angle, and that both

direction.
It is better

,'

'

sets vanish in the

same
if

so

,\
>\'

'1'
1

'

to arrange the picture,

shall converge,

possible, that these lines

it

\
y;'-

'
'

'

go

*x
^
^

'*
,,
J

i'
1

i'
1

be very slightly, away which will then be no from the Perpendiculars, longer Perpendiculars ; or even, if this cannot be

though

'i

arranged, to tinker the drawing so that, while the

Perpendiculars remain, the front lines shall be

v^,

made

to converge

away from them,

if

almost insensibly.
cathedral,
set of horizontal

S
shows a
lines at
still

Plate XVIII, of the transept of

Cremona
side.

worse deformity, yet correctly photographed, where one the top of the tower on the left seems to vanish on the wrong

Here both

55

CJ

<

J <
a,

6
Z n
06

<! 06

a u I < o
<

o s H

PERSPECTIVE DISTORTION
sets of lines actually

53

do vanish

in the

same

direction, the conjugate vanishing points

is

being both on the same side of the tower; and the effect

intolerable.

The

difficulty

comes from the

tower,

fact that, given the position of the lines in the principal face of the to see the other vanishing lines

we expect
first,

on the side opposite

to the vanishing

point of the

that

is,

on the

left

instead of the right.

is

We

see round the

wrong

corner, in fact, and the acute angle on the left

gular object should be so placed that

set so far

its

This means that a rectanhorizontal lines vanish both ways it should

very ugly.
;

from the Centre, or so turned, that we see the side of it towards not be which the horizontals of the principal face converge. In Fig. 80 we see that the square A. which may represent the plan of a cube or a square tower, is so turned that from .S the retreating face of the object will be seen on the right of the principal face, and the horizontals in both faces will vanish on that side, like the Cremona tower, while in Z>,
seen differently, the sides vanish in opposite directions.

CURVILINEAR PERSPECTIVE
It is sometimes urged that in nature even straight lines look curved, tending as they do to meet their parallels in a vanishing point at each end. This is true, provided they are long enough. We see it in the rays of light that stretch across the whole sky at
sunrise or sunset.
It is

equally true of the lines of a perspective drawing.

If

they are

extended far enough, they too will look curved. But the vanishing points of a straight line are at opposite ends of the world, and cannot be put in the same plane-picture, nor
can enough of any
drawing.
If
it

line

to display the

apparent curvature ordinarily be included

in

were, the

drawn

line

would also look curved.

Yet the fact that as we

turn the eye successively toward the opposite ends of a series of parallel lines, the

cornices that border a long street for instance,

some men

to insist that they should be dr.iwn so in one

this. It is

we see them converge both ways, has led and the same picture, and to

devise ways of doing

done as

in

vertical cylinder with the Station Point in its axis, picture.

panoramas, by projecting the view as on a which cylinder is unrolled into a flat

In the plan, Fig. 8i, the two equal circles

shafts

and A' representing towers or from the same Station Point .S on the picture plane and on the cylindrical surface, the first in ab, which is practically the same on both surfaces, the second in a"b" on the plane and a'b' on the cylinder. The Centre C is here
of columns, are projected

taken as the

common
and
is
it

tangent point of plane

is

and
the

cylinder,

clear that while the

projection a"b^'

larger than ab, or even than

width of the object A', being

made

obliquely on

/V,

a'b' is

considerably

less,

less than the diameter of A' or even than ab, would be the natural appearance. A' being farther than A from the eye at S. C may for comparison be taken for a Centre on the cylindrical surface ab', but properly the Centre may be anywhere on its horizon line, or nowhere, for any radius, being normal

as

to the cylinder,

may

be called the Axis, and no point

in its

horizon line has different

properties from any other.

There being no Centre, there are no Diagonals, and no

phenomena of and not of the picture, the whole apparatus of measuring is done away with, and the picture must be constructed by conical projection in the manner described above as the natural system, the horizontal projections being taken from a plan, as in Fig. 8i, and the heights from a side projection or elevation. Straight lines projected on the cylinder look straight from the Station Point, of course, but are really hoops centred on that point, as the horizon is in space and all but the Horizon Line are ellipses, being sections of the cylinder by oblique visual planes, while
vision,
;

distance points, and though there are vanishing points, for these are

PL.

XIX

U.

S.

TREASURY

PANORAMA OF PARIS

CURVILINEAR PERSPECTIVE
opposite sides of the cylinder.
these sections

SS

the vanishing points are the points where these ellipses intersect the Horizon Line on

When

the cylinder
all

is

developed,

unrolled, that
is,

become elongated

curves, like

oblique plane sections of a cylinder,

except the Horizon Line, which, being a right section, becomes a straight line. The extreme possible length of such a picture unrolled, taking in the whole horizon and returning into itself, is only the circumference of the cylinder, and the distance between
vanishing points of any straight line
picture
it

is

half the circumference, corresponding to the

theoretical but impossible projection of half the horizon on a plane surface, so that the
is

assumes to take

narrowed within convenient limits, instead of being extended enoimously when in a wide area. Fig. 82 shows the projections of Fig. 81 on the pic-

ture surfaces, at

the regular plane perspective

cylinder developed.
It will

construction, at

D the

be seen how the projections widen in the first as they withdraw from the Centre, and how any limit nar-

j>'

rows the width of the cylindrical picture when it is unrolled. This is a considerable advantage, the
only resource,
in

c
8"

i^.<y^

truth,

when one takes

all

in a very

wide area; but the fact that

but vertical lines

it

come
street,

out conspicuously curved

makes

impossible

J>

^
>
c^

to represent the straightness of a long building or

_
o
a.

bringing distortions that no change of posiwill cure,

a'

tion

by the spectator
fact,

This

with the labor and difficulty of

but the verticals

perspective,
to find

and that often are as offensive as those that are amended. making large drawings in which all the lines are curves to be constructed point by point, like circles in ordinary
too laborious, and after
all

makes curvilinear perspective much use in ordinary pictures.

too unsatisfying,

Plate XIX shows part of a panoramic view of Paris, photographed on a cylinder with a revolving camera, and then unrolled, so that it is in true cylindrical perspective. It
will

be seen that the only straight horizontal

line in

it is

the Horizon Line.

The

lines

of the principal building, the Hotel de Ville, which are really as straight as they can be
built, are

conspicuously curved in the picture, and those of the adjoining buildings,

parallel to them, are

which are

warped

into a crescent.

The main

proportions of the

building are preserved as they would not have been in a plane photograph, but at the
cost of very unflattering distortion.
If,

however, the picture were rolled up into a

cylinder of the right diameter, and viewed from the axis of the cylinder at the height of

the Horizon Line, the curves would straighten to the eye, and the distortion disappear.

PART

II

PERSPECTIVE HELPS
out of ihe geometric process of vision.
of

Perspective construction by direct projection from a plan is the most direct following It is in many simple problems the easiest method

making

a perspective picture,
first

and

is

made

useful in the preliminaries of

many more.

problems were solved by it. Yet the most conspicuous phenomenon of any perspective view is the convergence of lines that in nature are parallel. The use of vanishing points not only keeps the artist's attention fixed on this phenomenon, the most characteristic, and therefore to him the most essential, but it is the surest
or two of our

One

method
painter

for,

the vanishing points once placed, the lines that are drawn to them are
to

compelled into their true relation

each other.

It is

an education to the eye, for a

who does not

feel that certain

separated lines converge to a certain point will

will

never draw them quite right, and his work

sensitive eye than his own.

not satisfy a more instructed or a more

simplest way, for

It is also usually the

left

when we have
in

the

vanishing point there

place any
line.

is

only one point

to

determine instead of two,

order to

have to place a close series of parallel lines, as in drawing a it is extremely difficult to get their true convergence, on whicii the right look of the drawing depends, if both ends of each line have to be fixed The inconvenience point by point: it is very laborious, and seldom quite successful. is so much felt that draughtsmen are constantly tempted to mar their drawings by using
cornice or a group of mouldings,

When we

vanishing points that are too near rather than do without them, because those that are
far off are

awkward.

Perspective drawings, especially large ones, look so

much

better with far-off vanish-

ing points that various contrivances have been devised under the
for drawing lines toward an inaccessible point of convergence.

name of Centrolineads

Perhaps the simplest and pinned to the drawing board, against which is a metal plate, cut to a circular arc traverses a T-square by two pins set in the head, as is shown in Fig. 83. The centre of the arc is the vanishing point, and as the two pins traverse the arc the line aa that p// The working edge of joins them is a chord. ^.
the T-square, which must be
lar to the

made perpendicuwill

chord

at its

middle point,
is,

always

<

point to the centre, that

point.

to the vanishing

The two edges of The

device

the plate can be cut to

for V.

and so give two different distances is simple and practical, but its convenience is narrowed because it can only be set to fixed distances. Another centrolinead, which obviates this difficulty, is a T-square with isosceles triangular head traversing between two fixed pins, as in Fig. 84. Here again the line aa may be conceived as the chord of a circle, whose arc will be described by the point b If the working edge is set to bisect the angle b, it will at the vertex of the triangle.
different arcs

6o

APPLIED PERSPECTIVE
always point to the farther extremity

Foi

that diameter of the circle which

bisects the chord aa.

According as
of

the pins are set nearer or farther apart,

the chord
circle will

and the diameter

the

diminish or increase, and

reversing the blade of the
it

the vanishing point approach or recede.

By

square, so that

shall point

toward

F
in

instead of from

it,

the instrument

may

be accommodated to short distances.

This may be accomplished by making the head a lozenge, as shown by dotted lines
Fig. 84.

Note. The geometrical theory of this centrolinead is simple. Let a and a (Fig. 84) be the pins, and 6^ the line of the edge of the T-square prolonged to K The angle d traverses the arc aca, and Fis at the extremity of the diameter c^J^ perpendicular to aa. Since the angle
two
ada
is

bisected, the

on equal

The

size
is

which

Being equal inscribed angles, they must stand and Fis always the middle point of the arc aP^a. of the circle depends both on the angle of the instrument aia, and on the distance aa, a matter of adjustment; but the ratio of aa to cKis constant for anyone instrument.
arcs: therefore the arcs

two angles ad l^ are equal.

aFare

equal,

system of perspective, however, which depends solely on vanishing points and

is

direct projection

incomplete.

It

continually happens in large drawings, or in scene

some needed vanishing point is out of reach. occasion in the problem of the steps and others (Figs. 50, 113, 122) to construct lines for an inaccessible vanishing point.
painting, or in pictures on canvas, that

There

is

Suppose we have two

lines al>

and

i:d

running to the same inaccessible point (Fig. 85), and wish to draw another line through m to the same point. We may join the lines by any two parallels ac and it/, one passing through m : the problem is equivalent to dividing ac at n in the ratio in which is divided at m. We join 6c. Through m we draw mo parallel to crt', and through the intersection o draw on parallel to ai, cutting

ac in .

Then mn

will

be the required

line.

bm : md bo : oc, and in bm : md= an : nc. That

one
point.

like
is,

Note. Since om is parallel to cd, we have manner since c is parallel to ab, an : nc = bo : oc. Hence ac and bd are divided proportionally, and ab, mn, and cd meet in

The

following method, already used in Fig. 50,

are several lines to be

given point C,

is often more convenient when there drawn to the same inaccessible point V, whose distance from a If we wish to draw a line to such a point from say the Centre, is known.

a point a(V\g. 86)

we may set

off

on the line joining

Fand C

the horizon,

if it is

that

any convenient fraction, say a

third, of that distance

from Cat Vj ^

in the direction of V.

PERSPECTIVE HELPS
Then, joining a and
from
C,

6i

C,

we

set off

on
to

that,

Ca' equal to one third of Ca.

We
it,

join a' F/g,

and a

V must be parallel
a!

for

the triangles a
lar triangles.

VC and

K/3C must be

simi-

Suppose again that we wish to construct any figure, a square for instance, on a line ab whose vanishing point is out of reach, and even unknown, though we know the horizon in which it vanishes (Fig. 87). We join one end of the line, a, with C, and dividing aC in a convenient ratio, set off one part of it, say a third, from C at a'. Through a' we draw a parallel to ab, meeting the horizon in Vj ^ ; and taking this as a
vanishing point, we find
its
if

conjugate in

the usual way at V'j ^, and

we are draw-

r:'

c^

9-

Jk.

ing a square their mitre point 8/3. These three points will be severally one third as
far from

as the

unknown

V,

V, and

8.

Joining bC, we cut off a'b' on a' F/^ and

^^.

draw
if

y'l^ and (7;c parallel to it, which, prolonged, would pass through the
a'

vanishing point

V and

is

perspectively

perpendicular to ab.
a'S/3,

If

now we draw
it

and

b'

V'j ^, a'8/3 will be as

were
a'b'd'e' at

a mitre line cutting off b'd' perspectively

equal to

a'b',

and

a'ei'

as

it

were the diagonal of a square

one third the

size

of the required one.

The
last

required square

may now be drawn by drawing CV'and Cd'

in the
e'd' and bd same point d on Cd', and ad promaking C8 equal to three times C8/3.
e,

sufficiently prolonged, then

drawing from b and

thus found, ed parallel to

parallel to b'd'.

These

two

lines will

meet
8,

duced For it

will
is

meet

CV in
in

the true mitre point

clear that

we have here two systems

the smaller.

of similar triangles, constructed

on lines

radiating from C, and that every line in the larger system will be three times as great as

the homologous line

larger one will not be parts of the

It must be noted that the small square and the same perspective picture, for they are similar figures

constructed to different scales, the smaller being only a geometrical reduction of the
larger, auxiliary to its construction.

PERSPECTIVE OF OBLIQUE PLANES

definitions

Before considering farther the perspective relations of planes it is well to recall some and maxims that have been given, and to state some geometrical conseline of a plane is

quences whose proof may be taken for granted.

The vanishing
perspective
is
is

seen by looking in the direction of the plane.

it

Its

the line in which a visual plane parallel to

cuts the picture plane, and

called

its

horizon.
is

The

trace of a plane

the line in which

it

cuts the picture plane

its

trace

and

horizon are always parallel.

Parallel planes have parallel traces,

and the same horizon.

:

The

trace

and horizon
If
it.

of a vertical plane are always vertical

is

those of a horizontal

plane horizontal.
also parallel to

a plane

parallel to the

Horizon Line,

its

horizon and trace are

Such a plane may be

inclined, but not oblique.

The

trace

and

horizon of an oblique plane are neither vertical nor horizontal.

The ground trace of a plane is its intersection with the ground plane. As planes slope in all directions, so their horizons incline in all directions.
horizons of
all

The

which are normal, that

is

perpendicular to the picture plane, pass through

the Centre, because the visual planes which are parallel to them are axial planes, that

pass through the Axis. The Horizon Line is the one of these horizons by which drawings are commonly constructed, not because it has any geometric properties which the others have not, but because land and sea are horizontal, and men, animals, and
is,

things stand upright on them, and positions and directions are estimated by them.
All the constructions of perspective can be carried on
:

by

inclined horizons as well as

by the Horizon Line if the unaccustomed position causes embarrassment, we need only cheat the eye by turning the paper till our horizon is horizontal, and proceed in the usual way, remembering that the trace of the plane in which we are working, which This is the only will be parallel to its horizon, will take the place of the ground line. difference to be regarded in working in normal planes. The problem of the inclined
chess board. Fig. 12, has already given us an example.
Inclined planes which are not normal but oblique, whose horizons do not pass through
the Centre, require

somewhat

different construction.

Inclined planes which are parallel to the Horizon Line, sloping directly away from
the spectator

up or down, have

their horizons

and traces

parallel to that line.

The

laws of vanishing and measuring are applied as

tion of vanishing points

in horizontal planes,

but the determinaSince such a

and measuring points requires special study.

its

plane does not pass through the Axis, nor

horizon through the Centre, the Axis and

it.

Centre are not directly used

line

in

constructing in

Instead of the Axis we use a visual

which
line

is

perpendicular to
:

its

horizon, and so measures the distance from the Station

Point to this horizon

instead of the Centre, the point where this line meets this horizon.

The

may be

called a sub-axis,

and the point a sub-centre.

In Fig. 88

is

a view of

PERSPECTIVE OF OBLIQUE PLANES

the picture plane
Station

63

Point

S,

FP, horizon plane SHH, Axis SC, and horizon

H'H'

of

an inclined visual plane

H'H'EF

^' j-i^.ss

parallel to

the Axis

HH. A vertical plane through CS will be perpendicular to the

and therefore
to

picture plane,

HH and
:

H'H'.
lar to

Its intersection

SC

with the in-

clined plane will therefore be perpendicu-

and will be the sub-axis will be perpendicular to and H'H'. The triangle be the sub-centre, and is rightangled at C, its base SC being the Axis, and its hypothenuse the To apply this to the ordinary construction for a sub-axis, obviously longer than SC.
will

H'H

CC

HH

sec

SC

distance point (Fig. 89) length of the Axis.

triangle
If

let

D be the distance point,

CC
perpendicular to
the triangle

and

we draw

HH

DC therefore equal
z.nA

to

C.S',

the

H'H', and

join

DC, the ^'

CDC
Fig.

will represent

CSC
will

of

88 revolved about

CC

into

the picture plane,

and

its

hypothenuse
revolved.

DC
CD,

be the sub-axis

SC

equal to

CS,

laid off

on H'H',

will give

the sub-distance points

D'

belonging to the
deter-

horizon

H'H.
and sub-centre, once
in this

The
mined

sub-axis

way,

may be used

for fixing

measuring points and coordinate vanishing

points of lines in the inclined plane, just as
is

done

in horizontal planes.

It will

be seen that
in

if

two vanishing points in

for,

HH are
far-

coordinate, the points directly over them

H'H'

cannot be,

the sub-axis being

it

longer than the Axis, the coordinate vanishing points corresponding to

ther apart, as

must be

CD

in Fig.

89

is

longer than

CD ;

but the sub-distance points are

conjugate, as are the distance points (p. 31).

The

position of the Station Point

is

the perspective chart

laid

down

as simply as before.

and the length of the sub-axis being determined, If in Fig. go V is one vanish^

ing point in

H'H',

to find its coordinate point

perpendicular to we need only draw H'H' and equal to CD, and passing a semicircumference through V and 6' in the usual way fix the point V. The measuring points Let it be noticed that the follow regularly. represents the angle between the angle

CS

Ju.^

<5><3

>s

CDC

S' {/ -7SZ

c
\.
c

^ Xj

2r'

r"'

oblique plane and the horizon plane.

It

may

be called the slope angle or pitch of the plane. If we take the more general case of an
oblique plane, that
is,

\.,'

:^
is

one that

is

neither parallel to

HH nor normal, the construction

not quite so simple, yet not complicated.

being

still

taken perpendicular to H'H',

is

The plane of the triangle SCC now oblique to HH: when it is

(Fig. 91)

revolved

64

APPLIED PERSPECTIVE
round
to

CC, CS

will fall

at right angles to

CC

on Co, a line and so parallel

to

H'H'.

Take CS
;

to represent the

Axis revolved
will

then

CS' equal

CS
sub-

be the sub-axis,
left

C being the
andD' on

centre,

and

and this, measured from on H'H',

off right
will give

the distance points Z>'

horizon.'

that

If ^is a vanishing point on H'H', we may lay off CS' on CC' produced,

equal to

C'D\ and

describing a semi-

circumference which passes through

V and
make

S', with its

diameter in
chart

H'H,
in

the

perspective

the

usual way, and find

V, and

then

M
C

and M'.

The plane
being in
its

of

the triangle

SCC,

H'H', will contain both a slope line and a normal of the visual plane which vanishes in H'H'. SC will be the slope line, and the vanishing point of all slope lines. A perpendicular to SC at ^, shown in its revolved position as Sn, will be the normal, and n, in produced, the vanishing point of all
natural position normal to

CC

normals.

The most important conjugate

vanishing points for lines in an inclined plane are

those of
as
its

horizontals and

its

slope lines, which,

we know,

are at right angles.

its

The

first is

the

point where
for

horizon meets the Horizon Line,

all its lines its

must vanish
is,

in its

own

horizon,

and
in

horizontals vanish also in

HH,
is

therefore

both,

that

in

their intersection.

The

vanishing point of the slope lines

the

found by

method

of p. 34, Fig. 46.

V is

the vanishing point of horizontals in the plane which vanishes

in

Thus in Fig. 92 H'H', and F'

that of horizontals which are perpendicular to these

and therefore \.oSV.

The

vertical

V y",
and
gate,

therefore, will find the vanishing point of slope lines

at its intersection

V" with

H'H.

nSF"
v.

are both right angles:

Fand

(See page 34.) F' then are conju-

VSV

V and

V"

are

also conjugate,

and V"

is

vertically

over

Jf>?.^sS These facts are important in architectural perspective. If two walls of a gabled house make a right angle, as at Fig. 93, the angle cab at the eaves is also a right angle, and the vanishing points of the horizontal and the sloping
1

The

loci of all the distance

points of planes parallel to the Horizon Line are the branches of an

is

equilateral hvperbnla

whose centre

C and

its

vertices

D and D.

The same

thing

is

true for any series

of planes parallel to an oblique horizon 00 passing through C.

PERSPECTIVE OF OBLIQUE PLANES

65

eaves are conjugate, that of ac being vertically over that of ad. This is the condition we had in Figs. 56 and 57, the vanishing points in each slope of the roof being conjugate in both those problems.
process,
lines.

and

lines in either direction in either slope

Their measuring points may then be found by the usual be measured and divided like oiher

The geometrical
are often useful.

relations of these several vanishing points

are represented as before in Fig. 94, in the Horimeasuring points of Fand

They

and measuring points and J/' being the

zon Plane, and

series of points,

the oblique plane.

M and M" of Fand F" in V is common to both

and
its

two measuring points

it,

iJ/and
its

M are

equally distant from

as all

be found also that the other two measuring points are equally distant from V'\ so that
It will

measuring points must be.

V"M'i= F"M'.
will

The

angle

V'M'V",

it

be seen,

is

equal to the pitch of the

oblique plane, as

CDC

was that of the

is

plane in Fig. 90.

The

practical construction

then exceedthe vanishits

ingly simple.

Given

F and

V,

ing points of the horizontals of the given plane and of the horizontal projection of
slope lines, and the pitch of the plane

say

the

T"
71^.

vanishing points of the courses in the walls of a

building and the pitch of
its

roof

out a chart thus

at

(Fig. 95).

The

M'

gives

Fig. 46).

The

V" on the vertical arc M' M" from the

from the centre

and we may lay slope angle from V (page

laid off

f^
yiT"

34,

centre

M",

the arc
all.

MM

F gives

V" gives M, and

fy^ ^/^"\/
:^

Y-

this is

Note.

That

this construction is true

may be proved by comparing

the perspective diagram,

Fig. 96, with Fig. 94.

For,

first,

the plane of

the triangle
lar to

CCS has
that

VV", so

SC
is

been taken perpendicuand S'C\ which are

equal, are also perpendicular to

VV".

Then
:

the

triangle

VS' V"

equal to

VSV"

it

is
:

that triangle revolved up into the picture plane

so

VS= VS', and V"S= VS.

Second, by conso that

all

struction

VM= VS',
V" and M'

and

VM= VS,
M'
is a

four of these are equal, and

if

VM VM.
V"
(cf.

Also,

we

join

since

measuring
p. 34),

point of lines that vanish in

Rule,

-r^ f ^

V"M'

and

V"M"

must be equaL

66

APPLIED PERSPECTIVE
No
convenient relation appears between the
S'

mitre points 8 and

(Fig. 97), so that

first
;

if

3' is

needed, S' must be found

simply done.
\-r'

but this
0'

is

very

The middle point

the middle point of
0'

of

VVJ',
is

centre of the semicircumference V'S'V",

tically

ver-

/ -\ Wi^^-^i y ^ y\U o

over

0,

VV.

An

arc
0'

described from
will

as a centre with a radius

'^>T'

by

its

intersection with the arc

SMM
is

de-

termine S', from which the line Sh'

to

be

drawn, making an angle of 45 with S'V. use of this process

will

The
of

sometimes save trouble,

is

and

if

the pitch of

VV"

high, a

good deal

space.

CIRCLES IN OBLIQUE PLANES

Fig. 98 shows the construction in an oblique plane the sub-centre.

whose horizon

is

If 'IT',

Z? is a distance point,

and

CD = CS the

C being
:

length of the Axis

the

tri-

angle

CCS,

right-angled at C, gives the sub-distance points D', and the diagonal ab

of the square, vanishing in >',

measures the length of the vanishing

side.

The

rest of

the construction

is

as in Fig. 59.

To
line,

INSCRIBE A CIRCLE IN A GIVEN SQUARE.

is

If

One side of the square

is

is

a front

the problem

already solved.

But we may have a square which

already fixed,

and oblique to the picture plane. In Fig. 99 let VV'' be the horizon of any plane, V conjugate vanishing points of the sides of the given square abed in it, and 8 the and mitre point. Two sides of the diagonal square and the diagonals of the given square must vanish in S. Draw any convenient front line af, and produce two opposite sides ad and be of abed till they meet it. On the included portion ag describe a semicircumference, and draw the tangents at 45, meeting a/" inland h. Join these points and the centre e to V, and draw through the centre of abed' the diameter that vanishes in V, prolonging it till it meets F^and V'h in/ and s. Now/>4 and ps are two lines which in space are crossed by a series of parallels, the lines that vanish in V, and so are divided proportionally in perspective by these parallels. Hence Ve passes through But ^is equal to the half diagonal 0, the perspective centre oi ps, and oq : op =.eg : ef. of a square whose half side is eg ; therefore, since oq is half the side of abed, op is equal

to iis half diagonal ob,

is

and/

is

the opposite angle.

the square as before,

/S then and the

finds

one angle of the coordinate square, as m on eV, and sh finds , and stn
circle

in Fig. 98,

and

.f

?inA

pn complete

may be

inscribed as usual.

The

line_/% has for

convenience been passed through

a, but the auxiliary semicircumference

may be drawn

anywhere across the

position reduces
sections.
If
it

lines that vanish in

V, on
is

the front Vm&f'h' for instance, yet this

in finding the inter-

its size

so

much

that there

danger of inaccuracy

happens that one vanishing point Voi the given square

it.

is

out of reach, the

projected on

diameter /J cannot be drawn to

In that case the 45 points

may be

fh

and IV to the diagonals ae and bd Sii i, 2,j, and transferred by and 4. Then the sides / and sn may be drawn through i and j, and the second square be completed as before. This case would have occurred if mpns had been the given square, and 8 being interchanged. nor 8 is in reach, as in Fig. 100, so that we can draw It may happen that neither Then, having deneither the diameter /.f nor the diagonals to their vanishing point. scribed our semicircumference with its tangents on af, by projecting the 45 points on i, 2, j, and 4. We produce the diagonals ae and bd we get four points of the circle these diagonals till they meet Vf and V^ in tn, n, q, and r, joining m to n and q to r, which gives the points e and g. Last, drawing ei and g2, produced till they meet in/,
at

k and

/,

kV

68

APPLIED PERSPECTIVE

J, 2, J,

and 4 and gj meeting in s, we complete the second square, givdng the tangents at and 4. The diagonal ps fixes the points of contact of the sides ad and be. The
it

construction,

will

be seen, requires great nicety.

given, as in Fig. loi, abed being the

But

if

only the vanishing point of a diagonal

is

given square, we
as a hypothenuse

may draw any

it

front Wn&fh, here

lines

drawn through
it

a,

and project the

Upon_/7i

corners of the square upon

by

from

S.-

ah must as before bisecty/;.

we

erect an isosceles right triangle, inscribing in

a semicircle whose

centre will be
points k and
lines to
8.
/,

a,

and project the tangent points by perpendiculars onfk.

lines kh

thus found, and from the extremities of the semicircumference,

From the we draw

/, 2,

The

and hh

will cut the sides of the

square in the four points

The lines i-j and 2-4 joining these J, and 4, where they are tangent to the circle. will be diagonals of the second square and will find its corners,/, g, s, and e, on points
the lines that join 8 with the extremities of the semicircumference.

Connecting these

corners completes the second square

its

points of tangency are found by drawing ac

and

bd.
all

In

these cases the position of the plane of the circle

:

makes no

difference in the

construction
plane.

yV or IfJI may

be the Horizon Line, or the horizon of any oblique

To CONSTRUCT A
the given diameter

PERSPECTIVE CIRCLE ON A GIVEN DIAMETER IN A GIVEN PLANE. If is a front line ab, and the plane is normal, the problem is already
If the

solved in Fig. 61.

plane of the circle

is

oblique,

its

own

sub-centre and sub-

distance point must be substituted for

C and

D.

When

the given diameter ab

it.

is

by bisecting
the centre
0.

Through one extremity a draw a

an oblique line (Fig. 102), the centre cannot be found front line of measures, and upon this

project b at n from the measuring point

drawn to the middle of an will find on which the lines aJlf and wJ/" will intercept Through draw also the Perpendicular oC, and through ^ and the front diameter go/>. p Perpendiculars C^, and Cp, producing them. Diagonals from / and ^ to the proper distance point will fix the extremities of the Perpendicular diameter de, and front lines

M.

Mm

Draw

a front line through

through these extremities will give the perspective square y^M in which the be inscribed, six points being already determined.

circle

may

The

construction

is

justified

by proving that

aM and

.^ intercept

the diameter ^op.

In the plan. Fig. 103, the triangle

SVM

is

isosceles (compare Fig. 26)

and

is

the

middle of an.

and qp to
op are

isosceles and equal ; oa, oq, ob, same circle, and ab and qp are its diameters. This problem applies when a round tower has to be added to a building or an apse
all radii

The measuring lines aq, mo, and nb VM. Then the triangles aoq and bop are
of the

are parallel; ab

is

parallel to

SV, and

to a church.

Concentric circles do not

in perspective

make

parallel ellipses or concentric ellipses,

or even similar ellipses, nor have they parallel axes except in special cases.

The axes

look parallel, and are convergent in perspective.

helps which
point.
it is

we look
the

for in concentric figures,

none of the but each has to be constructed point by

curves
offer, then,
its

The

When
if

first circle

has been put into perspective with

enclosing square,

easy,

we know

the diameter of a concentric circle, to describe a corresponding

CIRCLES IN OBLIQUE PLANES

square within the
to

69

first or outside it on the same diagonals and cardinal diameters, and draw the second curve through the corresponding intersections (Fig. 104). The method of the double square is convenient here as before. The corresponding points are found, evidently on the same diameters of the circles, for instance, / and i on oa, 2 and 2 on ob, and so on. Thus the figures on a clock dial or the joints of a round arch

converge to the original centre of the

already constructed.
^, ^'

circles, that is the centre of the

enclosing squares.

This suggests a more general way of drawing a second

is

circle concentric with

one that

We

see in Fig. 105 that

if

we have two

pairs of corresponding

points

and/, /' in two concentric ciicumfcrences, the chords which connect the /' and q\ will be parallel, and in perspective will have the same vanish^, If in Fig. 106 we have the circle mpqn constructed about the centre 0, and ing point. know the extremity/' of any radius of the second circle, we may connect the corresponding point / of the first circle with any other point q of it. Then drawing the chord pq, and producing it to its vanishing point /' in the horizon of the circle's plane, draw a chord from p^ of the second circle to the same vanishing point this will Having intersect the radius oq in q', which will be the point that corresponds to q. found q\ we may draw any secant qr to its vanishing point ', and connecting f' with that vanishing point find the point r' on r, and so on indefinitely.
points/ and
:

SPECIAL POINTS AND TANGENTS

It
is

easy to put special points on the circumference of a circle into perspective by

the following process.

inscribed,
lines.

C being
is

the Centre, or the proper sub-centre,

Let 1-2-J-4 (Fig. 107) be the square in which a circle is to be and 1-2 and j-4. being front

If

the centre of the circle, ae will be the front diameter, and

oC will

contain

the perpendicular diameter.

of which
will

On
:

ae as a diameter describe the semicircumference ampe,

will

be the centre

it

represent half the required circle revolved about

ae

till it is

parallel to the picture plane.

Let

be any point
ae.

in

which

in perspective.

Draw

the half chord

mb

perpendicular to

Now

it is to be found any chord at right

into the plan of the perspective circle become a Perpenand the chord mb will fall on bC. It may be measured off on bC by laying off bj equal to bm and drawing a measuring line to the distance point by laying off the length both ways from d and measuring, both extremities of the chord m'm' may be found. This process may be used to divide a circumference into any required parts. Sup-

angles to ae will

when revolved

dicular vanishing in C,

pose, for instance, that

we have

to divide

it

into twelve equal parts.

We

divide the

and dropping perpendiculars on ae at b and c, This may serve for an arch, a clock, or an architecrepeat the construction as before. tural decoration. The chord /';' once determined, the corresponding one on the other side of oC may be found by setting off od equal to ob, and drawing the perpendicular chord bC and the front line m'm". It is worth noting that if the quadrant as (Fig. 107) is divided into equal parts, any two symmetrical points n and m, which are at equal distances from a and s, being joined by a chord, that chord will be at an angle of 45 with ae and os. When the
thirds, at

quadrant anms into

and

n,

semicircle

is

pendicular, the chord, being

ing in

D
mn

falls on oC, becoming a Peran angle of 45 with it, becomes a Diagonal vanishthe very Diagonal that measures the position of both n' and m'. If we

revolved into the plane of the. circle, and so

still

at

prolong

till

it

meets

ae,
r.

the axis of rotation, in

It

r,

the chord

when revolved
is

will

continue to pass through

may
it

then be drawn through r at once, serving for the

measuring

line of

m' and

',

and

will

be found that the distance rb

equal to

bm and

re to en, as they should be.

The draughtsman soon

discovers that the direction of a curve at any point greatly

it.

influences the curvature on both sides of

There

is

great advantage, therefore, in

defining the curvature of the perspective circle by tangents.

points have been already provided by the enclosing squares.
last

The

tangents at the 45

method provides a tangent

the semicircle
j

at

any point.

semicircle, at right angles to the radius op,

shows that the draw the tangent at a point/ of the and produce it till it meets the axis ae in /.
Fig. 107

We

When

is

revolved as before, the point

/,

being
to
/.

in the axis of rotation,

remains fixed

the tangent

may

therefore be

drawn from /'

\
1

\ \

DIAMETERS AND AXES

The
ellipses into

which

circles are projected in perspective

being symmetrical curves,

some advantage may be taken of this quality in drawing them. It is convenient to know the axes, the longest and shortest diameter, which are at right angles ; it takes, however, some trouble to determine them we shall see how later. But every diameter has its conjugate diameter, and each of the pair bisects all chords which are parallel to the other, as we know by geometry. Now in Fig. io8, cd, being a Perpendicular, repre-

sents a diameter of the circle which

is

at right angles to the front diameter ab.

It

therefore bisects the chord ab of the ellipse,

and
is

all

chords parallel to

it,

for these chords,

It is

being front

lines, are

divided by
its

it

in perspective in their true proportion.

then

a diameter of the ellipse, and

conjugate

that one of these chords which passes

is

through the centre of the

cally,
its

ellipse.

This centre, o\
it

found by bisecting cd geometriin

cd.

and a front chord

all

ej

drawn through
its

will

be the conjugate diameter, and will

turn bisect
will

chords that are geometrically parallel to

conjugate point
n,
if

Any

point

already

found
cd,

then determine

we draw through

a line parallel to

and

lay off

on

it

tions will give a

pn new

equal io pm.

Thus every

point found by the previous construc-

point (not the corresponding point m'), and the symmetrical

character of the parallel chords will assist the eye in drawing the curve.

The

determi-

nation of the ends of the diameter

^ will be best shown

in

in

another problem (Fig. no).

Here again there

is

convenience

taking two sides of the enclosing square parallel

to the picture plane, for that provides a pair of

conjugate diameters, one parallel to

all the front chords are give a corresponding point '

these sides, and the other a Perpendicular.

bisected by
cd,

Moreover, since
it,

any point m, found on one side of

will

on the other side by simply repeating the distance from cd on a front chord. Conjugate diameters help considerably in defining the curve of the circle in perThis is not spective when still greater precision is needed it is well to find the axes. a complicated process when we have once determined, as we have just done, a pair of
:

conjugate diameters.

ab being the tangent

regarding for the

Let cydx (Fig. 109) be the perspective ellipse, and xy and cd the conjugate diameters, Disat one extremity of the longer diameter, and parallel to xy.
<:

moment

the question of perspective,

let

us consider the ellipse as the

section of a circular cylinder

and imagine By an ingenious device called the Method of Shadows the ellipse is conceived of as the solar shadow of The this circular section, the rays of light being parallel to the axis of the cylinder. centre 0' will be the shadow of the centre of the circle, and the shadow of any diameter

whose

axis is oblique to the plane of the paper,

a right section of the cylinder,

made by

a plane passing through ab.

be a diameter of the ellipse. That diameter of the circle which is parallel must be parallel to the plane of the ellipse, and therefore parallel and equal to its shadow on that plane. This shadow must be xy, the only diameter which is parallel to
of the circle will to ab
ab.

The

circle,

revolved into the picture plane about ab, will be the circle here drawn.

72

APPLIED PERSPECTIVE

with radius co equal to half xy and perpendicular to ab, o being the revolved position of
its centre. Diameters of the circle, prolonged, must intersect ab, and must meet their shadows at the point of intersection then if we draw any diameter ob, we can draw its shadow (^b. Parallel lines give parallel shadows therefore any pair of diameters
:

in the circle

which are

at right

angles give conjugate shadows, for the tangents at the

extremities of each are parallel to the other, and this will be true of their shadows.

Now
its

the only pair of conjugate diameters in the ellipse which are at right angles are

axes.

Hence the problem reduces

itself to

finding the two diameters of the circle

whose shadows shall also be at right angles, that is, to find two pairs of radial lines oa, ob, and o'a, o'b meeting on ab, and each pair at right angles. These right angles may be inscribed in the halves of a circle whose diameter is ab, and which Then oo' is a chord of this circle, and the centre must be found passes through o and o'. It must also be found on ab, and must be the on a perpendicular at the middle of oo' intersection k. The circle described from k and passing through o and o' fixes the points a and b, as yet undetermined o'a and o'b if produced will then contain the axes. To determine their extremities we have only to find the shadows of the extremities of the corresponding diameters of the circle, by drawing mp and nq parallel to oo', which represents the direction of the rays of light. When / and q are found, p' and q' are measured off accordingly, or are found by extending ao and no into diameters and casting the shadows of their farther extremities the two methods will verify each other.
at right angles

SPECIAL TANGENTS
Points where the tangents are of special interest are the ends of horizontal circles, where the tangents become vertical in perspective, as at the base or top of an upright cylinder, a tower for instance or the upper and lower points in a vertical circle, such as an arch or a clock, where the perspective tangent is geometrically horizontal. These for the tangent at the crown of last points are not the crowns of the arches or circles,
;

an arch vanishes
picture.

in the horizon,

but

the points where the arches are highest in the

Fig. 1 16 shows a method of getting the vertical tangents of a horizontal circle, taken above the Horizon Line. The plan of the circle is set over the perspective for the construction's sake, against the front line PP, which in the plan stands for the picture plane, the position of the Station Point ^ being marked on the vertical T'Cat the proper distance from PP. Imagine the circle to be the section of a vertical cylinder, and two

and These planes will be projected in the plan into the lines Sc and Sd, tangent to the circle at c and d, and will cut the picture plane in vertical lines at c' and d', which will be the required tangents. To draw the tangents in plan, and find the points c and d, describe an arc from the middle point of So, passing through o, and cutting the circumference of the circle in two points which will be the required c and d. For if we join Sd and od, the angle Sdo is really inscribed in a semicircle whose diameter is So. It is then a right angle, and Sd, being perpendicular to a radius at its extremity, is the tangent at d ; and so with c. It remains to put the circle in perspective, and find its tangent points. The diameter^// is projected in ab, which is our line of measures, the enclosing square and the determining points for the curve being found as usual. The vertical lines from c' and d', as has been said, are the vertical tangents. We have then to find the points of contact by drawing the chord den. This may be done by projecting both the points on PP, as c at k, measuring off the distance ck on the Perpendicular kC, by the distance point ; or by finding one of them, c, and producing the chord cd in the plan to its trace , and drawvertical visual planes

drawn tangent

to the cylinder, therefore tangent to the circle,

,5'.

meeting

in

a vertical line that passes through

ing en through to
spective as to

d in

the picture.

The

point

is

apt to come so near h in the per;

make

its

independent determination troublesome

precise.
:

the use of n

is

often

more convenient and more

The chord cd
if

looks like the transverse axis of the ellipse

is,

it is it

in fact that of the

visual ellipse, that

of the ellipse as
cd,

it

appears to the eye, and as

would be on paper

the picture plane were parallel to

so that cd was a front

line, in

which case o would

be vertically over C, and cd parallel to HH. It is not the axis, for the tangents at the extremities of an axis must be perpendicular to it, and here they are oblique but these tangents are parallel, and therefore cd is a diameter. The centre o' of the ellipse is the
:

intersection of c^ with
in the circle

st,

for st also is a diameter, as in Fig. io8.

in the plan at the intersection of the

Its eccentric position

may be seen

corresponding

lines,

74
where
it

APPLIED PERSPECTIVE
is

not even in the middle of cd. The diameter conjugate to cd must pass and be parallel to the tangents, that is, vertical it must therefore be the line Now if we pass a vertical visual plane through o\ its trace on the picture plane xy. must be this vertical diameter xy, and the chord of the circle to which this corresponds will be the chord xy in the plan ; and if we wish to find the extremities of xy we maydo it by determining x and y as we determined c and d. We may draw the perpendicuThe lar yy', and putting it into perspective measure its length by the distance point. point X here comes, and is likely to come, so near the line ab that there is no room to measure, and we may trust to the course of the curve to find it.

through

o'

The

highest point in perspective of a circle or arch

is

is

evidently that where the per-

spective curve

horizontal,

and

is

determined

in the

same way.

To

see the application

of this last construction to that case

will

become
and c
is

vertical

and

CP the
is

we have only to turn Fig. Horizon Line. The trace of

cj.

no

round 90, when

is

PP

the tangent plane

then

cS,

the tangent point of the horizontal tangent

In this case the circle

cylinder,
is

assumed
vertical,

to

be normal, and

is

a right section of the assumed

whose

axis

is

taken parallel to the picture plane.

still
it

not normal, though

still

But if the plane of the circle no longer gives a right section of the cylinder,

which being
the

parallel to the picture plane

circle still lies in the

same tangent

visual plane,

becomes elliptical. The tangent to the and pierces the horizontal plane not in
which is a horizontal shows these conditions in plan. ebS", eb being the diameter, and SS"

same point

S,

but in another

in the trace of the visual plane,

line passing

through

5 and

parallel to ac.

Fig.

in

The plane
is

of the circle appears in plan as the line

the horizontal trace of the tangent plane.

The tangent

to the circle then pierces the

horizontal plane in S".

The

semicircle and tangent are

shown revolved
c the

into the hori-

zontal plane,

being the Centre, eb the diameter, and

tangent point revolved.

S"c

is

then the revolved position of the tangent, and transferring the distance

oS"

to

the picture at the proper scale, instead of oS,

we proceed
in

as before.

Turning the

circle

away from the normal

position,

and lengthening

consequence the subtangent ^^.S raises

the tangent point higher on the shoulder of the circle.

^ CP^

//ay

TIo-^aJLML

y^

TRI-CONJUGATE VANISHING POINTS

As most plane figures which are put into perspective are based on the right angle and constructed by conjugate vanishing points, so most solid objects also are based on rectangular forms, and constructed on three sets of lines at right angles to each other, like the axes of a cube. So long as such objects stand on a horizontal plane, their plans are composed of lines which vanish in the Horizon Line, and the other lines are vertical
lines

which have no vanishing point

is

therefore they are constructed by one set of con-

jugate vanishing points in the Horizon Line.

one edge

parallel to the picture plane,

Even when such an object is tilted, if we have still but two vanishing points, which
is

are conjugate,

and

their horizon passes through the Centre, while the lines of the third

system are front

lines.

Then

the picture

constructed exactly as

if

this horizon

were

the Horizon Line, and by turning the paper through the proper angle the construction

may be brought
is

to the customary aspect.

The box on

the sloping platform in Fig. 112

constructed by the conjugate vanishing points

vertical horizon

V and

V and

their

measuring points

on the

way that will be easily understood by inspection. In this case the planes of the top and bottom of the box vanish in the sub-horizon H'H' : and so do the retiring lines in them, which vanish in F, and the lines of the lozenge on the cover, which vanish in D', the sub-distance point. The other lines in
in

VCV,

the upright edges vanish in F', in VC these planes are front lines, and do not vanish produced below, that point being the conjugate of V. The dimensions of the box, set If the figure is turned 90 the vertical out below it, are measured oil in the usual way. horizon VCF' takes the position of the Horizon Line, and the box is seen to be drawn
:

like

an ordinary parallelopiped standing upright.

the Rectangular object
is

When
lines,

so placed that none of

its

edges

is vertical,

and none

parallel to the picture, so that all are vanishing lines,

we have

three systems of parallel

vanishing in three points which, taken two by two, are conjugate.

We may call these

lie in

points tri-conjugate vanishing points.

The

three series of lines, moreover,

all

three

planes or series of parallel planes which are

ishing points
Fig. 113,

oblique and inclined, so that the van-

lie two by two, in three inclined horizons, as we see by the same box in where no side is parallel to horizon plane or picture plane. Here the parallel lines a^, de^fg, vanish in V; the other system of parallels ac, df, eg, in its conjugate V. The planes of the top and bottom of the box, which contain these lines, vanish in the

horizon

parallels

which joins these vanishing points. The da and y? vanish in V" ; the plane of the right side, which contains these lines and ac and df, vanishes in the horizon V", and so on.

W,

Let
If

V,

V, and

be three tri-conjugate vanishing

points making the three vertices of a triangle (Fig. 114).

will

is the Centre, the sub-centres on the three horizons be found by dropping perpendiculars from C on FF',

76
V'V",a.i\d

APPLIED PERSPECTIVE

W". Now

the three visual planes which vanish in these horizons, and

the three visual lines which vanish in the three vanishing points, meet in S, and form

a solid angle, whose edges are the visual lines SV,

lines

SV,
is

and SV".

But these visual

of three right
is

and planes, being

parallel to the sides

and edges

of a rectangular object, are

themselves

at right angles,

and the solid angle at

therefore

angles (Fig. 115).

Therefore by the theory of projections,

graphic projection of
will

made up FC", which

the ortho-

SV
is,

be perpendicular to

on the picture plane, V'V", the horizon of

will coin-

the plane

SV V", that
CC",

cide with

or the perpendicular CC" prois

FC prolonged

duced will pass through F. In other words, the line which joins any one of three tri-conjugate
JTt^./rs
vanishing points with the centre
passes through

perpendicular

to the horizon which contains the other two,

its

and

sub-centre.

That VC is perpendicular to V" may be proved geometrically thus SC, being normal to the plane V", which is the picture plane. Join FC, Fig. 1 5, produce it till it meets V" in jr, and draw Sx. The plane VSx, containing the Axis SC, must be normal to the picture plane. It is also perpendicular to the plane V'SV", because it contains SV, which is perpendicular to that plane. It is therefore perpendicular to the intersection F'F", and any line Vxm it must also be perpendicular to VV". But CC", connecting the V", is also perpendicular to Centre with the sib-centre C"' for the horizon j therefore

Note.

the Axis,

is

VV

VV"

the two perpendiculars must coincide, and

x must

be C".

It follows that the

Centre

is

always the point where the three altitudes of the triangle

Consequently, when the Station Point

VV'V"
is

meet, and therefore that for any system of tri-conjugate vanishing points there
is

but one possible position for the Centre.

fixed,

and therefore the Centre,

It will
is

follows of course.

two of the tri-conjugate points are fixed the third be seen presently that even if the horizon which contains the
if

two points

determined, the third point follows of course, which

a rectangular solid
is

is

only saying that

if

set

on a fixed inclined plane the

direction of the lines that are perpendicular to that plane

is

also fixed, or finally, as

is

the horizon of a plane

normal lines

we already know, that when known the vanishing point of its maybe determined. The solid angle .Swith
as the corner sliced off by the picture

the three vanishing points and the intercepted planes

may be regarded

plane from a great parallelopiped whose edges are parallel

to those of the object,

of the

box

in

Fig. 113, for in-

stance.

It follows also that the

Centre must always be inside the triangle

VV'V",
5' re-

which means that the triangle must be acute-angled (Fig. 116). In Fig. 117, we have all three of the triangles that make up the solid angle
volved into the picture plane,

S', S", and S would be carried with them. VS' and VS" represent the same edge of the solid angle, and are therefore equal, and so for the same reason are VS' and VS'", V"S" and

S'" being the several positions into which

TRI-CONJUGATE VANISHING POINTS

V"S"'.
usual way.

77

The measuring points are set off in the VS' and VS", for instance, give the

measuring points and yy", which

M and M

in the

horizons

yV

are, as they

should be, at the

same distance from V, all the measuring points for F being, as we know, in a circumference about
it,

y, then,

and so with the others. Lines that vanish in may be measured either in the planes
in

that vanish in the horizon yy', or in those that

vanish

yy",

as the condiiions of the problem

may

require,

using
lines

the

appropriate measuring
parallel

points,

and

of

measures

to

the

appropriate horizons.

But inasmuch as each pair of measuring points and M, for instance, by an arc from S' or ^9" it will be found, as appears in Fig. 1 18, that only two of these revolved Staof the revolved Station Point

is

given by either

isf

two positions

tion Points are

needed

to fix all six

meas-

uring points.

Thus an
yS' gives

arc struck from

y with radius
from

y with
and J/".

M and M;

radius y'S' gives

M'

one and

7l/',one from F"with radius

F"5" gives
y"S',
the arc

M"
ence

Here, the semicircumferbeing

C,
S'.

yS'y

described,

drawn through

or perpendicular to
If

yy,

determines

we describe

intersection with the perpendicular from F' to

S'MMS" trom Fwith a radius FS', its yy" will determine S", and the other

measuring points
It is a

will follow.

simple matter, when the centre and two conjugate vanishing points are fixed,
In Fig. 119
if

to determine the tri-conjugate point.

we

make yy the diameter we may draw the chords

of a

semicircle enclosing C,

yCC"
Then

and y'CC" through

the
lines

C and

they will intersect the semicircumference in the

sub-centres

C" and C".

yC" and
F".

^. //^
-P~"

y'C'" must For we know

intersect in the

tri-conjugate point

that these last two lines are perpendicu-

lar to the altitudes

yC"

yC" y
It will

and y'C"' F,

and y'C", and the angles being right angles, must inscribe in a semicircle of which

yy

is

the diameter.

be seen that vanishing point

F"

is

independent of changes of position of

on the same horizon J/' If'. This will appear experimentally if we use other pairs of points on H'H', taking care that they are really conjugate, with the same Centre and Axis, for the construction will always give the same point F". This is what we should expect, for F" is only the

and

y so long as these are con jugate -and remain

78

APPLIED PERSPECTIVE
vanishing point of lines which are normal to inclined planes whose horizon is H'H' (cf. p. 64);

any

line

which

is

rectangularly coordinate

normal to a given plane will be with any two lines in

is,

that plane which pass through

at right angles, that

its foot and are have conjugate vanishing

points.
Fig. 120

shows how the process of

p.

64 for
to

finding

the vanishing point of normals

the

planes that vanish in

H'H'

will

determine F".

T"
the sub-centre on that horizon.
as in
Fig. 91,
If

H'H'

being the horizon which contains

it,

V
is

and
only

V, and V"C being perpendicular to we revolve the triangle SCC into the

picture plane

SC

will

represent the sub-axis, and the right angle

this construction will

CSV",

inscribed

in the semicircle, will

determine the required vanishing point V".

Fig. 119,

Fig. 121

shows how

combine with that of and may prove experimentally that both processes give the same result. This method is useful in problems of reflection and some others where the normal to a plane
is

required.

The

relations of

symmetry that

exist be-

tween tri-conjugate vanishing points and their measuring points do not appear between the mitre points, and these are to be
found singly by the ordinary process when they are needed.
Let us revert now to Fig. 113.
is
is

naccessible.

The planes

the horizon of the

V\% the only vanishing point on the sheet, and V" and bottom of the box vanish in H'H', which platform, and is taken parallel to the Horizon Line. The upright
of the top
in

edges are then normal to planes that vanish

is

H'H'. The vanishing

point of normals

indicated as

shown on page

64,

by laying

off the

length of the sub-axis from

to

^on HH, and

C'C
it

in

drawing a perpendicular to OS, which would meet the prime vertical the vanishing point of normals, that is, in V". But since this point is inacces
is

sible,

a point Sj ^

taken at a third the distance from

C to

S,

and a

line

drawn from

parallel to

JF" finds
V" by

V"j^,

its

substitute.

The

upright edges are drawn towards the

inaccessible

the process that was used in Fig. 50, and will be used again in the

problem next to

follow, Fig. 122.

^-f

MONUMENT WITH TRI-CONJUGATE VANISHING POINTS

on a terrace at the head of a flight of steps the lid is tilted up against the terrace wall, none of its lines being vertical, but all oblique boih Here the conjugate points Fand V, the vanto the picture and to the horizon plane. and the Centre being ishing points of the sides of the bed of the stone, are assumed F, at C, the positions of the other points are worked out by the chart at half scale. though off the paper, is within reach ; V" comes out of reach, and V"/^ at one half the distance, is substituted for it on the horizon V'V", and the triangle VVI^Vj^ furnishes the chart. A figured plan and section are given at the scale of the chart. gives the sub-centre on VF' ; half of it set off from The perpendicular gives C/^, representing C on the gives C/2. and o" a perpendicular at C'/g half A semicircumference being described on chart. VJ ^, and V'B and VJ ^E being drawn through C/g, the chords ^1 ^-B and V'E produced will with C'j ^Cj ^ all meet in and V" I ^, completing the conjugate triangle of the chart. The measuring points are found from the chart M' being off the sheet, it is more convenient to use M'l ^. The Ground Line being drawn at the proper height, and the Centre projected on it at c, the near corner A of the lid is found from the plan. A V and A will contain the sides Ab and ae of the bed of the stone. The line of measures b'Ae' is drawn parallel to the horizon : Ae', taken at the scale of the plan, gives by M'j^ the corner e, and Ab' These two fix the bed Abde of the stone. at double the scale gives by jtfthe corner b. for The upright edges vanish in V". This being out of reach, we do as in Fig. 50, the edge at A, for instance, we take the middle point/' of AV, and joining it with F"/j draw ^parallel \.of'V"j^. For the edge at b we take h', the middle point of bV\ and so on. The line of measures for these uprights is Am, parallel to the horizon V", and the measuring point is M". The rest of the construction follows natuIn
Fig. 122 a sarcophagus
is

set

CC

CC

M"

VV

rally.

It

remains to put
its lid.

in

the terrace and steps, and the

leans against the wall,
its

body

of the

awaits
that

The stone
it is

upper edge

in contact,

monument, which and so tilted

of the wall,

no

line in

horizontal.

Since the edge de

lies in

the plane of the face of the

wall, its

ground projection

lies in kt, the

base line of the wall.

The plane
its

then, vanishes in a vertical horizon passing through V,

V^,

and

horizontals vanish in

tals

on the Horizon Line vertically under V. The arrises of the steps and the horizonon the long sides of the sarcophagus also vanish in V^. The horizontals of the buttress-walls and of the ends of the sarcophagus vanish in F^'^, conjugate to V^, which
is

ing SV^, and

found by drawing C'6' vertical and equal to the Axis (given by the chart) and drawSV"^ at right angles to it. The base line kl of the buttress-wall of the steps pierces the picture plane in the Ground Line at /, which is found from the plan,
which we set up a scale of heights. The measuring points M^ andM"^ enable off the dimensions in the ground plane, and to draw there the plan op^r of the sarcophagus, which is projected up to its proper level. The pitch of the steps, laid off

and

at

us to lay

at

M"^,

gives F'" for the vanishing point of the slope lines of the steps and buttresses.

HEXAGONS AND RECIPROCAL VANISHING POINTS

is

the measuring point of the other

In every horizon there are two vanishing points which have the peculiarity that each These are the they may be called Reciprocal.
:

vanishing points of lines which make angles of 30 on each side of the Axis, and are This is easily seen in Fig. 123, where ^Fand SV, at therefore at 60 with each other.

30 on each side of SC, are the sides of an equilateral triangle KS'F', and are obviously
at

equal distances from SC.

The symmetry

of

the triangle
at

makes

it

isosceles, the angle of 60

5 makes it equilateral, and therefore VV'-==. VS=^V'S, so that each vanishing point, being as

from the other as the Station Point, is the measuring point of that other. All the reciprocal vanishing points stand in a circumference about
far

:^i^./i-i

the Centre, as do the distance points, there being

only two in any possible horizon.

iarity is not of general in the construction of

Their peculis

importance, but

available

hexagons and forms based

side ab being laid
oflf

on the hexagon.
Fig. 124 applies this property in

drawing a hexagon.

The

on

the line of measures, the vanishing lines

"

Fand
ax and

bV
bx,

give the directions of the sides

ac and bd.

v-^

71^.

/X.4l>

By laying off the same length at and drawing measuring lines to Fand V, we measure off ac and bd reciprocally, the measuring lines being also the vanishing lines, which are cut off by bVa.nA a v. These last are diagonals of the hexa-

gon, fixing the points e and_/J and so determining the whole figure.

culty.

This method enables us to draw a continuous pavement of hexagons without diffiWhere, as in Fig. 124, a side of the hexagons is parallel to the Horizon Line,

the lines aV,

like

aV,

bV, and b

V are repeated

as often as there

is

need, and front lines

^are drawn through

their intersections to contain the sides that are parallel to

If. We have only to pick out carefully those parts of these three systems of parallels which are the sides of the hexagons. If, as in Fig. 125, the hexagons present their angles symmetrically to the Horizon

H.

Line, one set of sides

out from the line ab,

becomes Perpendiculars, which becomes a front line.

parallel to bd, so that the figure is laid

Then

the reciprocal lines af 3.x\dbf no

longer contain the sides, but

make a

series of inscribed equilateral triangles like aft,

on

which the hexagons may be constructed.

points draw
lines to

We

lay off then on our line of measures the

is

side ab of an inscribed triangle, repeated as often as

necessary, and from the division

a/b, etc.

^'and V, describing the triangles

From

the point a

we

1^

HEXAGONS AND RECIPROCAL VANISHING POINTS

81

draw a line through the vertex of the adjoining triangle at g which will cut off the side bd oi the first hexagon on ihe Perpendicular bC, and a front line through d cuts off all A few diagonals like a;/ give the corresponding sides of the first row of hexagons. intersections enough to find all the angles that correspond to d, and the vertices of the It will be noticed that the sides of triangles _/; ^, etc. are the other angles required. the hexagon are every third section in a series of oblique lines. If, finally, the hexagons are oblique to the Horizon Line, reciprocal vanishing points The vanishing points of the are of no use, and the problem is solved without them.
three series of sides are not

commonly

all

accessible at once, but having two of them,

we may
point
is

substitute for the third that of the perpendiculars to one side

whose vanishing
its

known, as

in Fig. 126.

Here the most

its

available series of lines are ab and

parallels, vanishing in
a/"

V; ad and

parallels, vanishing in the conjugate point

and

and

its parallels,

vanishing in V".
the

the plan, or

V,

being conjugate to V,
lie in

angles of each hexagon

first

The three vanishing points are determined from may be fixed by a chart. Four sides and all the and third series. The right intersections being
will

picked out, the other sides

series of parallels,

may be drawn, and

is

be alternating sections of a fourth

is

whose vanishing point

perhaps inaccessible,- and

not necessary.

OCTAGONAL PAVEMENTS
The
use of diagonal squares
If
tlie
is

illustrated in the perspective of

lie

pavements which are

based on octagons.
Fig. 127, they are in

the octagons

with sides parallel to the picture plane, as in

position of the octagon formed

Fig. 15.

The

point

being chosen on the front line

points

iD

J, are set off from the plan, the centre of the first square gives
I, 2,
;

by the overlapping squares in the point c and the repeated and Perpendiculars drawn to C. The Diagonal a front line dd will give the points of contact of

AB,

and by the help of Diagonals to both distance points, or of set of Diagonals and of Perpendiculars, all the points of contact can be found and one all the squares drawn in. If no sides of the octagons are parallel to the picture plane, we have two pairs of conone set of cojugate vanishing points,
the front rank of squares
;

^^'

"^f

,''

ordinate lines making angles of 45 with

the other set.
that
all

It will scarcely

happen

four vanishing points are within

if

reach, but

one pair

is

determined by

the ordinary perspective chart, as in Fig.

128, the mitre point of this pair will

be

one of the other

pair,

and the pavement

may be
point.

constructed without the fourth

Here

the

pavement

is

laid out as a series of larger squares with smaller squares

interpolated, cutting off their corners,

leading lines those that vanish in F, and those in

and turning them into octagons. We choose for 8, and add those which vanish in F'

conjugate to
etc.

F".

Starting from A, according to the plan,

8 contain

we

set off the points /, j, 5,

Lines drawn from these points to

two sides each of a series of squares

parallel to that of
2, 4, 6, etc. will

which Al>

is

one

side.

Next, lines to

Firom
mark

the intermediate points

contain diagonals of these squares, and

their corners,

which are
is

the centres of the small squares, so that the remaining sides, whose vanishing point
inaccessible, can be

These centres being found, a series of lines to Fat proper distances each side oi 2 V, 4V, etc. will cut the sides of the large squares at the corners of the small ones, whose sides may then be filled in, completing the pavement. Fig. 129 shows the process on a larger scale. It will be noticed that a pavement of octagons, even more than one of squares, looks that is, unless one series of its lines tilted if it is not symmetrical to the Ground Line therefore will prefer to arrange his pavement symmetrically The artist is of front lines.

drawn

in.

with the Ground Line unless the exigencies of the rest of his picture forbid.

V.
'^^

PERSPECTIVE FROM ELEVATIONS

In Fig. 130 is such a problem as is constantly occurring in an architect's office. Sketches of elevations have been made to scale, and a perspective view is to be constructed from them.'

allowed

here twenty-eight

The vanishing
inches.

points were set as far apart as the drawing board

will

siderably

The picture more foreshortened than the other, and

look best

if

one elevation

is

con-

the side will foreshorten to the best

The Centre then is put twenty inches from the vanishing point Fon the and only eight inches from on the right. From the elevations so much of a perspective outline plan as is necessary can be constructed, and the vertical lines are plumbed up from it. The near angle A of the porch is put in the picture plane at
advantage.
left,

the Centre

through

it

are

scale of heights in the picture.

drawn the horizontal scale in the perspective plan and the The measuring points may be determined in full scale

on the board, or as here by a perspective chart at a fourth the scale, to be transferred to the drawing by multiplication. The two elevations and perspective are drawn to the

same

scale.

Ay and AV in the plan

On them
bays are
laid off

show the

positions of the two visible walls of the house.

the dimensions of the front and side are

measured
off.

off

by .^and M', and the

the projections of the

widths of the porches, bays, and openings are set

When

on

AV produced
is

of the perspective plan as

Ground Line, the outlines of so much needed may be put in. By plumbing up from the plan
in front of the

and measuring the heights on Ax in the picture with horizontals vanishing to V znd V, the house is blocked up to the roof. A house looks better, as well as more natural, when the Horizon Line is put rather low, not much higher than the eyes of a man and we choose the transom of the bay-windows for the horizon standing against it, level. The placing of the mullions of the windows and the arches of the porches is most easily done without measuring, by bisection with the help of diagonals. The roofs, gables, eaves, and chimneys require special constructions. The gables are laid out first in the plane of the wall, and the overhang is then added. The pitches are all 45, and if the proper vanishing points are within reach, the construction is simple. Assuming that they are not, we begin with the side gable, measure the horizontal distance of the peak from the corner at c, and plumb up to the picture the vertical cc. The height, measured on Ax at a, is transferred by the horizontal a to cc at c in the picture. A horizontal ^Fgives the ridge of the gable, and is prolonged in front of the wall to cover the overhang. The height of the point b, where the vertical line of the corner pierces the roof plane, is measured directly from the elevations, and the horizontal bV, representing the intersection of the planes of the front and the roof, is prolonged like cViox the sake of the overhang. This overhang may be measured in the plan on the prolongation of VA, as the projection of the bays was measured, and the line b'

The

elevations are copied as I find

them printed

in the

American

Architect,

and some small

details

are omitted for the sake of clearness.

84

APPLIED PERSPECTIVE
The plan of the horizontal eaves in front is manner, and their intersection b' is plumbed up to b' in the picture. Then tending to V", accessible or inaccessible, represents the eaves-slope of the gable, and
in like
c'

represents the line of eaves in the plan.

found
b'c',

its

intersection
is is

with cV'is

its

peak, to be transferred to the picture.

rest of the eaves in like

of the gable of the ridge

easily found,

and the

manner.

The other side The farther end

lines being here

hidden behind the front gable.

will find

This front gable we

by a different process, the measuring

is

somewhat crowded.
point which
its real
is
is

The

ridge of the gable produced will pierce the picture plane in a

found in the picture by plumbing up to hV^ the peak k got from the plan after determining the overhang as for the other gable. The face
height above

projected in the plan at h, and

HH

ai

and on the

indefinite line of the ridge

of the gable projects as far as the front of the bays, while the eaves

have a farther over-

corner/ is fixed like h by measuring its height in the picture plane. The farther end of the ridge, /, is the point where the ridge pierces the vertical plane that contains the cross ridge cV, and the half of an assumed rear gable may be drawn to determine the bit of its slope, In, which rises above the cross ridge. The lines of the verge boards and mouldings of the gables can be drawn to their respective vanishing points above and below /^and V, where these are accessible, or can be determined by one of the processes already given for such cases, or finally, when as here the scale is small and the lines close together, they may be considered as sensibly parallel, and drawn in accordingly. The valley eo is determined thus the point e is plumbed up on bVixom. the plan, and eo, yrr ., ^

hang

the

being parallel to
if

be,

vanishes in V".

But

V"

is

not in reach, o

may be found by

the horizontal

rVin

the plane of the roof,

whose intersection with the return wall plumbed up from the plan is o. The simplest way to fix the hip sf is to draw the ridge tF' which unites it to the main roof
(seen in the elevation but hidden in the
perspective) from the trace
of the front gable.
/'

in

which

it

pierces the picture, determined like the ridge

Then

the point /

is

plumbed up from the plan upon f'F'.

like that of the ridge

The

chimneys are

first

placed on the plan.

The

height of the main chimney, which bestrides

;

the ridge of the front gable,

may be determined
wall, is
;

that of the other,

which stands flush upon the side

on the scale of heights Ax. The arches on this small scale are sketched in the rest of the detail needs no comment. One window in the side elevation, and the dormer which would flank the front gable on the right, are omitted to avoid crowding the picture. They present no difficulty. Fig. 131 shows on an enlarged scale the construction of the overhanging corner of the
measured
directly

eaves.

j^;7><?^^,

:?rr

XX V//

A BROACH SPIRE
The
broach spire in Fig. 132
is,

is

tower of the same diameter.

cutting off ihe corners, that

Its

an octagonal pyramid, set in the usual way on a square plan is a regular octagon inscribed in a square by

the octagon

made by

the overlapping of two concentric

filled out by four triangular pyramids or engaged pinnacles which lean against the adjacent sides of the spire, as shown in the lower half of the plan. The spire is a high pyramid whose vertex is in the axis of the tower, and the salient edges of the four small pyramids would meet in a lower point

squares.

The

corners of the square of the tower are

of the axis, for they are symmetrically set, and

may be regarded
and
is in
its

as the visible corners

of a single square pyramid which caps the tower,

spire.

great part covered up by the

plan, are constructed on the

The

lines of the eaves of the tower,

which indicate

diagonal squares used before for the octagon and circle.

The

perspective square abed

being set out by

its

conjugate vanishing points

Fand V,

the angles of the octagon

are found as in Fig. 15 by the

method
e'

of overlapping squares.

The

half-diagonal e'f of

the square, set off both ways from

on ax, thence measured on the side ab prolonged, and transferred to the diameter^, gives the diagonal of the second square. Lines ^8 andyS, intersecting the other diameter, fix the corners h and k, and drawing also gk

we determine all the visible points at the base of the spire. Its edges may be drawn when its vertex has been measured off, which is done here by a line m'h perspectively
parallel to the diagonal ac, at the proper height in the scale of heights ani\ intersecting

the axis of the spire in m.

edges meeting
arris is

in ,

determined

The corner pyramids are drawn in like manner, the salient like tn. The nearer one only is completely seen its
:

stopped against the middle

line

pm

of the near face of the spire, or the height

since the edge ap
is in

of

its

vertex

may be measured from/' by

am' and

8,

the vertical plane

of measures that passes through

otn.

The

lower dormers, which centre on the

middle lines em, etc. of the cardinal faces of the spire, are best placed by drawing lines am, bm, dm, which indicate the pyramid of the spire filled out to the square of the top

pyramid with squares at the height of the bases, eaves, recession of these dormers from the face of the tower The width is calls for a special construction to determine their width and projection. plumbed up from the eaves of the spire at i' and /', on verticals on which the height The same height being projected of the bases is projected from the scale of heights.
of the tower,
this

and encircling

and ridges

of the dormers.

The

for the

encompassing square, the width is projected back upon this encompassing in / and /, and straight lines drawn upward from i' and /' through * and / will represent parallels in the face of the spire which will contain the valleys between the dormers and the spire. A similar construction gives the positions of the upper lines of dormers, which are based at the height of the point n.
square from the verticals

MOULDINGS BASE OF A PEDESTAL AND CONSOLE

The
drawing of architectural mouldings
is

in

perspective requires peculiar care

the
of

representations are of no value to the architect unless they are precise.

Any want

accuracy

apt to change their effect greatly, and mislead the designer, or misrepresent

the object.

The

profiling

is

seen chiefly at the corners of mitres, and

in perspective

many mouldings
is

miss their intended effect from not having been studied from this point of view.
Fig. 133

shows the base mouldings of a pedestal

below
at

a profile of

the base at the scale of the perspective,

scale.
in

and a plan and elevation

one fourih that

The

mitre profiles give the best opportunity of constructing the perspective, for
sets of horizontal lines meet.

them two

They

are

front corner, where the vertical

Ax

is

the scale of heights.

and the
ings at

lines
c,

AB

and

AD, we

lay off

from

on

Ax

drawn first, beginning with the Having placed the corner A the heights of the different mould8 will

e,f, g, h.

Lines drawn from these points to

be horizontals

in the mitre

plane, on which are to be found the points of the mitre profile.

The

horizontal

Aa

on

the profile shows the projections on the mitre line.

If the

measuring point

M" for the

mitre lines that vanish in 8

is

within reach, these projections can be measured off at once

<:8,

on Aa, from which verticals intersecting

profile in perspective.

^8, etc. will

give the points of the mitre

Aa

to

Fand

Lines drawn from these points and from the division points of give the perspectives and the plan of the base and mouldings, and if

Bb and Dd could and the points of the mitre profiles projected up from them. But it is out of reach, and the mitre line Dd is found by completing the perspective square ADEF, and drawing its diagonal FD, which is produced in Dd. The diagonal of a
the vanishing point conjugate to 8 were within reach, the mitre lines
it,

be drawn toward

gives Bb in like manner. It may be, however, that the measuring point for 8 is not at hand. Then instead of measuring the projections of the mouldings directly on Aa, they must be first measured on a right section and transferred Here they are measured on a line Dk, vanishing in V, which is the horizontal to Aa.
similar square constructed on
trace of a right section plane.

AB

from the measuring point

M.

transferred to the mitre lines

The point D is referred to the front line of measures at The projections m-i, 1-2, 2-j, measured on Dk, and Aa and Dd, give the same points that were found, or
first

would have been found, by the

construction.

In Fig. 134 are two elevations and the perspective of a console to the same scale. The console is rectangular in plan the conjugate vanishing points are within reach,
:

though

off the

sheet; the measuring points and 8 are only indicated by their directions;

the angle between the wall and the console in the perspective serves for a scale of
heights,

the ordinates

ures ab,

The projections, shown by off directly from the elevations. and the widths on the face are measured on the line of measthe measuring lines being mostly erased to avoid crowding the drawing, which
which are ruled
i, 2,

j, etc.

sufficiently explains the construction.

7i^.

/33

A
C
z

'^

Z-'i<or.^.

mrpTTT

PERSPECTIVE OF A PEDIMENT
Fig. 13s is the perspective of a pediment, in which the difference between the raking cornices and the horizontal somewhat complicates the problem. The figure gives the

showing both cornices at ihe scale of the perspective. The Aa in the extreme front corner, and the perspective plan The members of the horizontal and raking cornices meet in the in dotted lines above. We might measure off the projections of the mitre profile for the mitre plane at ab. plan at ab' by the measuring point M', on ad' vanishing at V, and then transfer each measurement to ab' by lines which vanished in ; but we shall save transferring if,
front elevation of one half,

scale of heights

is

taken on

after finding the

produce

it

extreme point 7 on ad', and transferring it to b' on rtS, we draw jb' and to the accidental measuring point m, and then draw from the other points
ab',
8,

The points thus found on J, 4, 5, 6 directly to m. drawn at the proper heights on and vanishing in

projected

down on

horizontals

AC

give the points of the perspective

upward raking cornice converge to the vanishing point y", vertically over F' and determined as usual by a^slope line from the measuring point M'. Then the lines of the two upper fillets of its crown moulding and the corona can be drawn in at once. The rise of the pediment being laid off at Ac, and the half width
profile

Ab.

The

lines of the

at

trc',

the vertex

is

found

at d.

The plane

of the ridge profile passes through d, its

horizontals vanish at V, and the ridge appears only in plan

the

line d'

its

per-

spective being hidden by the pediment.

The

projections of the ridge profile are found

profile ab' to

on d' V\n the plan, projected by

the scale of heights at
cf.

lines

from the angle

i, 2, /, etc.

V, and

its

heights on

The

points

taken from the elevation, are prolines vanishing in

jected on dd' from cf {dA their true height on

Ad) by

V.

It

must be

borne

in

mind, however, that the heights of the crown mouldings are not the same as
vertical section at ac in the elevation, while the other

those of their horizontal counterparts, which they intersect obliquely, and are to be taken

from the

intersect the horizontal ones, are free to correspond to them.

The

members, which do not lines of the upward

rake can be drawn through the angles of the ridge profile from the vanishing point V", and those of the two fillets should coincide with those which are already drawn. If
the point

V" had been

out of reach,

it

meHiate

profile, parallel to the ridge profile.

would have been necessary to construct an interThe lines on the soffit of the upward
cornice might be drawn directly from the ridge
that

raking corona are out of sight.

The

lines of the

downward raking
if

profile to their

vanishing point V",

were accessible.

In this case

it

is

out of

we construct the mitre profile at the farther corner e, which is measured off on cV like ^and plumbed down on AV. The mitre line for the profile at e" in the But the mitre plan cannot be drawn to its vanishing point, for that also is out of reach. lines at a and e" in the plan, being perspectively symmetrical about the line r/'F(which is in the same horizontal plane with them, and being in the vertical plane of the ridge
reach, and

88

APPLIED PERSPECTIVE

is at right angles with ae") must intersect on it. Therefore we extend ab' till it meets d'Vin h, and draw ^'h, which must contain the mitre line at e". Upon this mitre line we project the dividing lines of the profiles at ab' and hd' ; then the points of division are projected vertically on the lines of the horizontal cornice already drawn toward V,

making the points of the

profile in the picture.

>

^'

ENTABLATURE IN PERSPECTIVE
Fig. 136
is

the corner of a modillioned entablature, with the plan and elevation at

The perspective could be made from geometric drawings on a smaller scale, but in constructing architectural detail it is best to make the necessary enlargement beforehand in the geometric drawings, for the chance of accumulated errors is so great as to make perspective enlargement hazardous. For a problem as complex as this it is well to make a detached perspective plan, and The scale to show it from beneath, with the arrangement of the modillions and dentils. of heights aab is taken at the projecting angle a of the upper moulding, where it interferes less with the picture than it would at the angle of the frieze, though it slightly
the scale of the perspective.

reduces the scale of the picture.

also the extreme height

V, V, M, and 8 are assumed to be determined, and above the Horizon Line. The perspective plan is displayed by setting its angle a well above the picture. The corners of the frieze in the plan is found on the mitre line at by measuring off the overhang at (/from the horizontal scale ax by means of M, and drawing adV; cV 2mA. cV are the lines of the frieze. The overhangs of the members of the cornice are found by the method of p. 25 and Fig. 31, setting off ae on the scale ax for the full overhang, joining ec, and prolonging it to the Horizon

Line at m, which
I, 2, J,

will serve for

an accidental measuring point.

The

principal points

4,5

are

first

found, representing the overhangs of the corona, the modillions,

modillion-band, dentils, and dentil-band; and lines drawn to

Kand

V represent these
are
8,

members
.

in the

plan on the two faces of the entablature.

2', etc.,

The corresponding heights

being drawn to

set off on ab in the picture at 7',

/, 2, etc.

and mitre

lines

the points

are projected

members

in the picture,

on them from the plan. This fixes the corners of the principal from which lines are drawn, representing their edges, to

Fand
The

V'.

it will be That on the left at the angle clearer and safer to construct them first stands in line with the band to the right ; its front side is found on the plan by producing V3 to (> ; and 6, projected back from ^upon the measuring line ax at 6\ is used The lines of the modilas a starting point for measuring off the rest at/', 8', g', etc. lions drawn in by means of these points are projected down directly upon the picture. For those on the right face the measuring point M' is out of reach. They are found by

modillions and dentils might be constructed directly on the picture, but

on the plan.

projection from those on the

left,

continuing the sides

till

they intersect aS in

k,

/,

etc.,

and then projecting these points from f^upon j V. For want of room to continue on the left the measures beyond 8 are transferred to a second line lower down as in Fig. 35. The dentils are laid out as is shown by the scale on the right of a, one centring with each modillion, and three between but to avoid confusion of lines we start from the corner 5 of the dentil band, projecting 5' on HI/, and use the accidental measuring
;

point

n.

mitres of the moulding which crowns

Mitre lines to 8 through the front corners of the modillions, as in them and the band between them, and

/t,

give the

will cut off

90

APPLIED PERSPECTIVE

squares at the ends of them by which counter diagonals can be drawn to give the mitres at the alternate corners. In the same way the mouldings of the panels between the
modillions can be drawn with their mitres.
picture
is

The

transference of these details to the

a simple matter of vertical projection.

Eggs and darts and the leaves of

may be sketched in, one egg or one leaf over or under each dentil their positions may be determined by carrying profile lines down the face of the cornice without referring to the perspective plan. The drawing of the intermediate lines of the mouldings requires no new construction.
the bed-mould at the base of the cornice
;

A ROMANESQUE ARCADE
In Fig. 137 a Romanesque arcade is shown in perspective. The geometric plan is on the scale of the picture, the partial elevation on half that scale, the perspective chart
fifth. The arches are semicircles recessed in two orders, showing a main arch and a sub-arch carried by piers and columns in alternation, and the piers are subdivided to match the imposts of the arches. It is these imposts that are constructed in the perspective plan, for by them are determined the positions of the arches, which are the most important elements of the problem. The section of the arches is shown in con-

on one

nection with the perspective plan.

The

abaci and the plinths repeat the outline of the

imposts

in the piers,

but not

in

the columns, whose abaci and plinths are square.

The
is

picture plane passes through the axis of the shaft attached to the front of the

first pier,

which axis

is

in the

plane of the wall, as appears at

in

the geometric plan, and

is

the scale of heights in the picture.

in the elevation.
in the

The

angle of

shown at the picture plane and the wall plane is shown by

is

The

height of the Horizon Line

HH
PP

geometric plan, and the Centre

off the sheet.

opposite the centre of the

first

arch.

Fand

F' are both

The means

perspective plan being constructed in the usual way, the arches are drawn by
of tangents at 45, with the help of the auxiliary quadrants be,cf,

and

dg.

They

are lifted on low stilts

whose upright edges, plumbed up from the

centre
in

plan, are vertical tan-

gents to the arch curves at the springing, the impost being on the horizontal

aV.

The
/'

middle point
vertical ^/'on

in the plan gives the perspective

of the first arch,

and the central

which we project, by lines that vanish

c'.

V,

the tangent intersections

and

s'

and the crown points d' and

",

The tangents

are fixed

sections/ and n with the vertical at /"upon that at/' in/' and

and when /and k are also projected in /' may be drawn. The face of the sub-arch recedes from that of the wall, displaying the under surface or soffit of the main arch, and making a recess whose section is the broken line k"k"'h". The outer curve of the sub-arch, in the angle of the recess, is geometrically equal and parallel to the inner curve of the main arch, answering exactly to the rear curve of the arch in Fig. 63, and constructed in the same way. The point 0', plumbed up from the plan, is the centre of the front curves of the sub-arch, and the upright o's" contains its vertical diameter. Then s", r", c", and b", projected from s', r', c', and b', upon this vertical give corresponding points by which the curves c"k"' and b"h", etc. are fixed. A further projection backward gives r'", b'", h'", etc., to determine the curve at the back of the sub-arch. It is to be remembered that the broken line l"k"k"'h"h'", like the joints in the arch
outer curves of the main arch
of Fig. 63, lies in a plane that passes through the

in/" and

by projecting their inter', and upon that at _/" and /", ^'and /J", the inner and

common
0,

axis of the soffits,

V,

and

its

successive portions, prolonged, pass alternately through

and

0'.

Therefore when

the tangents are fixed and the point /" found, this line can be continued round the

92

APPLIED PERSPECTIVE
Nevertheless, in a matter so
better to use this last
it

arches without projecting the successive points from At.

delicate as the deiermination of the arch curves,
is

safer to construct the points inde-

pet)dently than to follow round from one to another,

for a verificiit'on.
It will

and

method

be seen, too, that the tangents, being 45 lines in the plane of

the wall face or parallel vertical pl.ines, will be parallels which vanish in the vertical

sub distance points above and below V, the vertical that passes through
horizon of
all

V being

the

such planes.

The

rest of the construction offers

nothing new.

I-

-i-^^J!

'

GROINED VAULTING
The groined vault shown in Fig. 138 illustrates some of the devices which special problems suggest, as well as the use of tangents for perspective curves. The vaults are of uniform semicircular section, covering two series of aisles at right angles, and card id on columns. The arrangement of the plan and elevation are given at one third the
scale of the perspective.

The vaults

spring from cubical

stilt

blocks, which are as broad

as the plinths of the bases, their diameter being one fourth the span of the vaults. picture plane
is

The

passed through the axis of column

.(4,

for

convenience of measurement,

and

is

tangent to the near corner of the plinth B.

C is

the Centre, and

SC

the Axis,

as usual.

By

the angle

EBA

the perspective chart

may be

constructed, say at the

of distances to the

scale of the plan,

and the points transferred with due enlargement

Horizon Line.

below HH, and draw AA the axis of column it, so that the plan may open well, A. The columns stand at the corners of adjoining squares, and laying out these squares we find the axes of as many columns as will appear in the picture. The square plan of the first base being constructed in the usual way at A, and its vanishing sides being continued indefinitely toward F'and F', two corners of each of the other bases are determined on these lines by drawing through their centres diagonals to the mitre point A wall runs behind the third range of columns 8, and the plans of all are easily found.

We

begin with a perspective plan.

Taking GG we project A upon

at a convenient distance
'

against the plinths of the bases, touching also the

free.

stilt

blocks, while the shafts stand

The counter diagonals

of the bays should be drawn, to determine their centres,

for the intersections of the groins

positions by drawing an axial line

come over these centres. It is well to verify their down the middle of each aisle, which must pass through the centres of all the bays. The axes in both directions will prove useful. Having fixed a suitable height for the springing of the vault, we next construct the
stilt

blocks,

plumbing up

their vertical

necessary lines to
requires

Fand much care. The

X^
*^g

V for their upper and lower edges.

upright edges of the
stilt

edges from the plan below and drawing the The drawing of the groins
blocks provide vertical tangents at
the springing points yj /J etc.; the horizontal tangents at the crowns are the

..

-^
V
''

'

'

|N.
'

diagonals of a series of squares corre-

'*.

_l^X--_

sponding to those

in the plan.
let

^'V ^^9

!j---r-^''^^~~~-S\r^"^N.
I

To

provide intermediate tangents,

^-1?^\^, / ^\_
'

-"-"^

'

_-^^^^^Tx
Nn^v

"^ imagine a tangent plane to the cylinder of one vault ; it will contain all the

/
,

V^J^
'

tangents

to

sections

of

the

cylinder
in Fig.

L__.

--!---;

rVr'^"

IJ^-"-^-''^
'
'

'

'

along the line of contact.

Thus

*"--.. Lr--*'''''T*

^39 ""^ represents a right section of the vault and adcd a tangent plane at, say,

94
45.
If

APPLIED PERSPECTIVE
mn
its
is

cylinder at
d^nd

some point fp

the line of contact, any oblique tangent bk in the plane will touch the / of this line. Let ihe vertical plane of this tangent be egbk,

fm/i

section of the cylinder, of which

will

gh

is

a vertical diameter,

li

fmk

is

groin, the vertical

be the edge of a stilt block, and gh the vertical through the centre of a bay, the axis of the bay we may call it. The plane of the springing of the vault is bae, which will contain the points / and g. There will be four groin tangents for each bay, corresponding to bk, intersecting in k, and forming the edges of a square pyramid whose base will be in a horizontal plane, the plane of the springing
appears in Fig. 140.

bae, as

The

four tangent points will

all

be at the same height in

a horizontal plane at the level of mn, as here

shown.

We

have

to

construct such a pyramid

will give

over each bay, which

us tangents to each

of the groins, while the point h will give the intersection of the groins, where the tangents are
horizontal.

In Fig. 138 at

a,

the level of the springing, in

axis of the

column A, we
'*'

set

up against the

column

the auxiliary quadrant afh', representing half the

section of the vault, and draw the 45 tangent bk'. Then h' marks the height of the crown of the vault, k' of the vertex k, and m' of the tangent point m. The heights of h' and k' may be measured off on the axis of the first bay, plumbed up from^, by diagonals vanishing in the mitre point 8, which will give k and h. In constructing the pyramid
of tangents
it

will

be convenient to use, instead of the feet of the tangents, b for instance,

the points where they intersect the axes of the columns, because these points are each

common
the
falls
stilt

to four tangents in adjoining bays.

Now

the vault springs from the face of

in

block, represented in the quadrant

;

by/' (which
c'

the drawing accidentally

on one of the lines of the vaulting)

a'c'
a/i'

fa'

represents half the thickness of the block,

by the tangent. Then c' horizontally c. Drawing ck, therefore, we have the tangent to that groin of the first bay which springs from column A at/ behind the impost. In the same way m" projected on ak', gives the height m' of the tangent point, to be again projected on the tangent at w by a mitre line to 8. The point c m:iy be projected on the other three axes by horizontal lines which will form a square add directly over the square of the bay in the plan, giving the points from which the tangents that complete the pyramid may be drawn to k. The tangent point m may be repeated in the same way on the square mnmn. The intersection h of the groins at the crown of the vault is already determined the tangents here may be found by repeating the horizontal square at the level of h and drawing its diagonals, of which hh', already drawn, is one. We have now five points and five tangents for each of the groin curves of the first bay, by which the gxo\ns/mhm/ and enhne may be drawn in. It is a simple matter to extend the construction to all the other bays by horizontal projecting lines. The construction of the arches at the back wall, which are right sections of the vaults, is the usual one, all the lines being in one plane. The scale of heights ach'k' is projected back on the plane of the wall on the vertical from/, the centre line of the back of the plinth of column F, making the scale a"c"h"k", and the heights are carried along
the axis of the column, intersected in

and

projected on the real axis

gives the point of intersection

AD

GROINED VAULTING
the wall in horizontals vanishing at V.

95

The meeting
from

points of the tangents and the

the centre lines of the bays of
etc., in line

crowns of the arches are

the wall
;

in the verticals

q, q', etc.,

the lower ends of the tangents are in the verticals from/,/',

,

with

the axes of the columns, at the height of c".

The

plan of a base at the scale of the perspective

is

the shaft

half the diagonal of the plinth or stilt

shown at K. The diameter block, and being determined for

is

of

may be

projected from column to column, or in a drawing to so small a scale as

lines,

this,

where the half the diagonals are almost front

each base with less danger of error than
constructed by means already known,
if it

can be projected.

may be measured geometrically on The capitals may be

it,

the scale of the drawing calls for

or they

may be sketched, as may be more clearly

is

in front of the

The bases are omitted in order that the perspective plan displayed. The column E is included in the picture, although it picture plane, to show the intersection of the groins on the left of A,
here.

the choice of the limits of an interior view being always arbitrary, and determined
the artist's idea of a desirable effect.
It is

by

sometimes convenient to place a column, or

in the plan,

even a series of columns, by visual projection, as at E'

this gives at

where, as

is

obvious,

once the position and the diameter of the

shaft.

A VAULT WITH LUNETTES

Fig. 141
section.
is

the perspective of a vault with lunettes,

vault
is

all

of

which are of semicircular

;

The main

a cylinder of which

is

the radius

is

that of the lunettes,

whose vaults, springing at the same level, intersect it at right angles. J^f^is the axis of the main vault, and JSVihe line of its springing, both vanishing at F. is the conjugate point in which vanish ON, and the other axes of the lunette vaults. The plane of the lunettes is somewhat recessed from tlie springing line, to give relief to the pendents of the groins. Assuming the point a on the springing line .^F for the foot of one groin, we construct aA, the depth of the recess, taking A for the angle at the base of the lunette, and At for the scale of heights, AA' or AB being the radius of the lunette, and BA' the auxiliary quadrant with its 45 tangent tj. To construct the curve of the

lunette with precision, intermediate points i, 2, 4, 5, are taken on the quadrant A'B. They are projected, like the tangent point j, both horizontally and vertically, and being

constructed in their proper perspective position, the curve of the lunette through them.

is

drawn

The

principal point in the

problem

is

the construction of the groins.

If the vaults of

the lunettes were cylindrical, like the

main

vault, a series of horizontal planes

through
all

i-r, 2-2,3-3, etc. would contain corresponding horizontal right-lined elements of

the vaults

the intersections of corresponding elements would be points of the groins,

easily fix their curves.

But these vaults are cones with horizontal axes at main vault, vanishing, as we saw, at V. The vertex of the cone of the first lunette is TV; on O V. NB', drawn to the crown of the lunette, will be the highest element of its vault, and will penetrate the main vault in F', the crowning point of the groin. All the vertices N, N', etc. lie in a horizontal line which vanishes in V. A plane passing through this line NN' and touching B would be tangent to every lunette at the crown, and contain the highest elements of all the cones. It would cut the main vault in a line VF' which would contain the crown points of all the groins and vanish in ?^ A plane that passes through NN' and through the points 3, 3 on the curve of the lunette, cutting the plane of the lunette in the horizontal j-j, will contain two elements jA^ jA^ which will pierce the main vault in two points j',j' of one horizontal element, and these points will be points of the groin. The same will hold for the other pairs of points, i-t, 2-2, etc. To find the element j'-j' of the main vault, we construct a right section F'P of the vault made by a vertical plane passing through Oi', which shall contain the axis and be perpendicular to the lunette. This plane is cut by the plane N3-3 in a line c'cN which intersects 3-3 in c. If c' is the intersection of Nc with the section-curve, the horizontal c'Fwill be an element of the main vault lying in the plane N3-3, and the points j',j' in which it cuts the lines j-j', 3-3' will be points in both main vault and lunette vault, and therefore in their intersection, the groin. The other pairs of points, /'-/', 2'-2', etc. may be found by the same process, and the line of the groin drawn through them. The entablature from which the vault springs is constructed as in a previous problem by the help of the mitre-point. Here, as elsewhere, some numbers which would duplicate others symmetrically placed are left off the drawing for the sake of clearness.
right angles to the axis of the

and would

ON

CAMBRIDGE MASSACHUSETTS U S A

14 DAY USE RETURN TODESK FROM WHICH BORROWED

LOAN
is

DEPT.
last
I

This book

date stamped below, or on the date to which renewed. Renewed books are subjea to immediate recall.

due on the

^^

iv.^ Tpr

^r

1974

"

RCC^D LD
C C2

JA^iil

1957

2'63-4P M

DUdJ
JUN 2 2 1966 2
Ac-

21 '668
1968

APR 24

2^^^

A^

X :^t^

H ecei

^Ten

(B9311BKPPB

Si

rsity

of California

Berkeley

(^^^

^if^uItJr"