GREEK SOURCES OF THE COMPLUTENSIAN POLYGLOT

Natalio Fernndez Marcos

Centro de Ciencias Humanas y Sociales. CSIC. Madrid

In the Grinfield Lectures 2003 devoted to The Study of the Septuagint in Early
Modern Europe Prof. Scott Mandelbrote deals, among other interesting issues, with the
text of the Alcal Polyglot, the earliest printed text of the Septuagint completed the 10th
July 1517. He pointed out the impact of the arrival of Codex Alexandrinus in England in
1627 and its use as one of the main authorities for the London Polyglot (16531657),
whose editor, Brian Walton, was especially critical of the text of the Complutensian
Polyglot and the precise age of the manuscripts on which it had been based.1 Indeed,
Waltons judgement is highly negative; he maintains that the Greek text of the Alcal
Polyglot is very far from the genuine Septuagint. It is a compilation of several different
texts with Hexaplaric additions and even Greek commentaries in an attempt to relate it to
the Hebrew text printed in the parallel column.2 He backs up his statement with some
examples taken from the first chapter of the book of Job.
Since then the vexed problem of the Greek manuscripts used by the
Complutensian philologists has been dealt with by different scholars, including myself.
However, I think it is worthwhile taking another look at the question in the light of new
evidence which has recently been published in the context of Septuagint textual criticism.

Scott Mandelbrote, 'The Grinfield Lectures 20032004,' BJGS 33 (Winter 20032004), 37. The Codex
Alexandrinus was presented to King Charles I as a New Years gift in 1627. B. Walton was the first who
designated the manuscript as Codex Alexandrinus. On the origin of this manuscript see Scot McKendrick,
'The Codex Alexandrinus or the Danger of Being a Named Manuscript,' in The Bible as Book. The
Transmission of the Greek Text, Edited by Scot McKendrick and Orlaith OSullivan (London: The British
Library & Oak Knoll Press, 2003), 116.
2
"Nova enim et mixta est haec Versio, partim ex Septuaginta, partim ex Origenis additamentis ex
Theodotione, partim ex Aquilae, Symmachi, aliorumque Interpretum, imo & Commentatorum Graecorum
verbis consarcinata, ut hoc modo textui Hebraeo per columnas aptius responderet," see B. Walton,
Prolegomena (De versionibus Graecis) to the Biblia Polyglotta (London, 1657), 64. Likewise in the
Praefatio B2, Walton deals with the new contributions of the London Polyglot and among them he
mentions the publication of the Septuagint according to the Sixtine edition of 1587 and Codex
Alexandrinus when it disagrees from Vaticanus, rejecta illa Complutensium, quam secuti sunt in Regiis et
Parisianis, ut omnium quae hodie extant maxime mixta et interpolata, et quae a genuina twn o omnium
longissime distat.

It is well known that cardinal Jimnez de Cisneros, in the Preface to the reader
that precedes the edition of the texts, praises the Greek manuscripts sent by Pope Leo X
from the Vatican Library and used for the Polyglot, as very old and pure (vetustissima
simul et emendatissima). He mentions, in addition, other sources: a copy from a very
correct manuscript belonging to the legacy of cardinal Bessarion, sent by the Venetian
Senate; together with other manuscripts which were the fruit of a long and costly search
for a large number of corrected codices.3
Cisneross statement must be interpreted in the context of the Renaissance,
indulging in some rhetorical bias, and the state of Biblical studies at that time. But the
basic reliability of the information transmitted cannot be put in doubt. In the 19th century,
Vercellone discovered in the Vatican Library, the proceedings of an inventory of the
library of Leo X which were drawn up in 1518; he had edited this inventory and
described the codices lent to Spain for the preparation of the Complutensian Bible,
adding that they had been returned to the library.4 These are Vaticanus Graecus 330 (=
108 of Rahlfss Catalogue) and Vaticanus Graecus 346 (= 248 of Rahlfss Catalogue).5
The first contains the Octateuch, 14 Kingdoms, 12 Paralipomena, 12 Ezra, Judith,
Esther (Septuagint and Alpha-text) and Tobit (incomplete); and the second contains
Proverbs, Ecclesiastes, Song of Songs, Job, Wisdom, Ben Sira (with the second Prolog),
12 Ezra, Esther, Tobit and Judith. It must be emphasized that ms. 330 has many
Hexaplaric notes and arabic glosses as well as some Greek scholia to 2 Kingdoms. Ms.
248 is full of Hexaplaric notes in the margins without any indication of the sigla to which
those notes should be attributed. As we shall see below, this circumstance may explain
some Complutensian singular readings at the beginning of the book of Job.
As far as the copy sent by the Venetian Senate is concerned, we have no other
information than the words of Cisneros in the aforementioned Preface. There is no
3

Quod autem ad Graecam scripturam attinet: illud te non latere volumus: non vulgaria seu temere oblata
exemplaria fuisse huic nostrae impressioni archetypa: sed vetustissima simul et emendatissima: quae
sanctissimus Dominus noster Leo Decimus Pontifex Maximus coeptis nostris aspirans ex ipsa apostolica
Bibliotheca ad nos misit: tantae integritatis: ut nisi eis plena fides adhibeatur: nulli reliqui esse videantur:
quibus merito sit adhibenda. Quibus etiam adiunximus alia non pauca: quorum partem ex Bessarionis
castigatissimo codice summa diligentia transcriptam Illustris Venetorum Senatus ad nos misit: partem ipsi
magnis laboribus et expensis undique conquisivimus: ut copia emendatorum codicum abunde supesset,
Prologus ad Lectorem, IIII of the Alcal Polyglot.
4
C. Vercellone, Dissertazioni accademiche (Rome, 1864), 409.
5
A. Rahlfs, Verzeichnis der griechischen Handschriften des Alten Testaments (Berlin, 1914).

testimony of the loan in the Marcian Library of Venice. Therefore, Delitzsch, in a visit to
the library, examined the Greek mss. that had belonged to cardinal Bessarion in order to
find out which of them could be the base of the Spanish copy. He arrived at the
conclusion that ms 68 (= Gr.V) is the best candidate, the most fitting to the adjective
castigatissimus of Cisneross Preface. He also verified that the Complutensis fills the
lacunae of 330 with the copy of ms. 68; he insists, however, that these two manuscripts
are not the only sources.6 This partial copy including Judges, Ruth, 14 Kingdoms, 12
Paralipomena, Proverbs, Ecclesiastes, Song of Songs, 12 Ezra, Esther (according to o'
and with the Greek Supplements pace Rahlfs), Wisdom, Judith, Tobit, 13 Maccabees of
ms. 68, a complete Bible7 including the New Testament, was preserved until the Spanish
civil war (19361939), in the library of the Complutensian University of Madrid, as part
of the ancient collection of the Colegio de San Ildefonso of Alcal founded by Cisneros.
It could be consulted by Samuel P. Tregelles, the editor of the New Testament, in 1860,
and Delitzsch published in 1886 its text for the Song of Debora (Judg 5) and Davids
elegy (2 Sam 1:1926) from a facsimile copy sent by the librarian Pascual de Gayangos.8
As to the fate of this manuscript, 442 of Rahlfss Catalogue, and its present state of
preservation a few words at the end of this paper seem adequate. The other manuscript
from the same collection of Alcal and used basically for the edition of the Psalms by the
Complutensian philologists is a Psalter of the 14th/15th centuries (Number 23 of Villa
Amils Catalogue = 1670 of Rahlfss Catalogue).9
For the time being, we shall leave to one side the problem of the Greek
manuscripts used for the text of the New Testament and which has still not been
answered satisfactorily and also requires further study.10 But, as far as the text of the Old
Testament is concerned, it is patent that Delitzschs studies led to a positive appreciation
of the Complutensian Polyglot as textual witness. This process of revaluation of the
Complutensian Greek text will continue in crescendo up to the present time.
6

F. Delitzsch, Fortgesetzte Studien zur Entstehungsgeschichte der Complutensischen Polyglotte (Leipzig,

1886), 26.
7
The order of the books is the same as in ms. 68. However, we can only speculate on the omission in the
copy of the Pentateuch, Joshua, the Prophets, Job, Psalms and Sira. The ms. has 13 Maccabees only like
its copy.
8
Delitzsch, Fortgesetzte Studien, 1318.
9
Delitzsch, Fortgesetzte Studien, 2829.

Closely related to the problem of the manuscripts utilized by the Complutensian

collaborators is the use they made of such witnesses. In other words, whether they
respected the readings of the manuscripts or whether they corrected them in order to
accommodate the Greek text to the Hebrew or the Vulgate printed in parallel columns.
The opinion of scholars on this issue ranges from the strong negative stance represented
by A. Masius, B. Walton and R. Simon to the more balanced opinions, concretely those
of F. Delitzsch, P. de Lagarde and J. Ziegler, who held the Complutensian Greek text in
great esteem. One thing seems clear: it is time to forget the broad generalizations and to
undertake a thorough study of the individual books in order to advance towards the
clarification of this vexed question. For example, the critical edition of the book of Job by
J. Ziegler has indirectly contributed to shed light on the Complutensian Greek text of this
book. As in the rest of the Wisdom books, the Alcal Polyglot follows basically ms. 248
lent by Leo X. But for the first ten chapters (the first chapter of Job constituted the core
of Waltons criticism!), they took the readings of Aquila, Symmachus and Theodotion
which were in the margin of this manuscript, sometimes whole sentences, because these
readings were closer to the Hebrew, I suppose. But it can be verified that in the rest of the
book, chapters 1142, only five marginal readings of ms. 248 have been introduced into
the Complutensian text.11 In defence of the editors it should be remembered that these
Hexaplaric readings of the margins lacked the sigla with the corresponding attributions.12
It can be deduced that, at times, the closeness to the Hebrew or to the Vulgate has been
utilized as a criterion for the restoration of the Greek. Certainly it is not a sound principle
according to modern textual criticism of the Septuagint, but was the kind of philology
commonly practised in the Renaissance.13 The European humanists, just as Origen and
10

See the interesting contribution of J. A. L. Lee, Dimitrios Doukas and the Accentuation of the New
Testament Text of the Complutensian Polyglot, NT 47 (2005), 25090.
11
OConnell thinks that the different procedures in the editing of Job are due to two different editors, one
of them more interventive than the other, not to tiredness of the editor after chapter 10 as Ziegler thought,
see S. OConnell, From Most Ancient Sources. The Nature and TextCritical Use of the Greek Old
Testament Text of the Complutensian Polyglot Bible. OBO 215 (Fribourg/Gttingen: Academic
Press/Vandenhoeck & Ruprecht, 2006), 133.
12
J. Ziegler, Septuaginta. XI, 4 Iob (Gttingen: Vandenhoeck & Ruprecht, 1982), 56-58, and Delitzsch,
Fortgesetzte Studien, 12: Die Verschmelzung solcher Marginallesarten mit dem Septuagintatext war
freilich unkritisch, entschuldigt sich aber dadurch, dass sie im codex nicht mit ASQ bezeichnet sind.
Delitzsch insists that most of the Complutensian peculiar readings are not retroversions but rely on the
witness of the manuscripts.
13
Und in der That, welchem andern Kriterium htten sie bei ihrem beschrnkten handschriftlichen
Apparat folgen sollen?, see Delitzsch, Fortgesetzte Studien, 27.

Jerome, thought that the Greek translation had been made from the same Hebrew text
transmitted by the Masoretes, a text which had been preserved unaltered throughout the
centuries. They could not even glimpse the different Hebrew texts that the Qumran
documents have patently brought to light and, consequently, the possibility of a different
Hebrew Vorlage in the hand of the translators. Consequently, in the Preface to Pope Leo
X, cardinal Cisneros points to the originals as the last criterion for authenticity in the
translated texts, ita ut librorum Veteris Testamenti synceritas ex Hebraica veritate: Novi
autem ex Graecis exemplaribus examinetur.14
But we should not think that the recourse to the original source as the arbiter of
the Greek readings was used as the supreme principle or in a systematic way. In fact, this
device has been detected in a few cases, while the editors try to draw the reading closest
to the original Hebrew from the Greek tradition, be it in the manuscripts or in their
margins. Moreover, in books like Jeremiah, the Complutensian editors are aware of the
differences which exist between the Hebrew and the Greek texts, they maintain the
Septuagint tradition as valid in its own right, and admonish, in the preface to this book,
that the Greek text must be respected without trying to correct it towards the Hebrew,
because the truth of both should be maintained.15 A surprisingly modern warning
confirmed by some of the Qumran Jeremiah fragments (4QJerb,d) that support the
Septuagint short text edition. And on the text of Reigns OConnell states: The editors
saw LXX as a separate entity and consequently the Greek text as requiring internal
criticism. This consideration, frequently in tension with that of faithfulness to their
primary source and to the Hebrew, explains the number of stylistic modifications and the
overall pragmatic approach to the text."16
It is not only the new critical editions of the Gttingen series which have helped
to illuminate the Complutensian text. Further studies, especially on the books of
Kingdoms and the Twelve Prophets have led to the conclusion that each book has to be
14

Prologus, col. 1.
Haec interpretatio hieronymi est. Si quid in ea novi erit: secundum hebraeos codices exploretur. Alia est
Septuaginta interpretum ecclesiis usitata. Quae quamvis nonnulli aliter habeat quam in hebraeis codicibus
invenitur: tamen utraque, id est secundum septuaginta: et secundum hebraeam, apostolica autoritate firmata
est: non errore neque reprehensione superiori: sed certo consilio septuaginta nonnulli aliter dixisse vel
contexuisse intelliguntur. Quod ideo praemonemus ne quisquam alteram ex altera velit emendare. Quod
singulorum in suo genere veritas observanda est, from the Preface to the Book of Jeremiah.
16
OConnell, From Most Ancient Sources, 111.
15

submitted to a thorough analysis and the results have been nuanced according to these
new studies.17 In what follows, I do not pretend to offer a global or final solution to this
complex problem, but only to mention in passing a few recent and partial studies that
have contributed to add to the esteem of the Complutensian Greek text among the textual
critics of the Septuagint over the last sixty years.
In a study devoted to the Twelve Prophets, I pointed out that every publication of
a new witness had contributed to reduce the number of singular readings of the Alcal
Polyglot, readings too often attributed to arbitrary corrections of the editors.18 The main
problem of the Complutensian Greek text of the Twelve, die noch unerledigte Frage in
Delitzschs words19, found a partial answer in Zieglers study. He demonstrated that a
good number of readings of the Complutensian Polyglot considered to be unique were
shared by Codex Washingtonianus (W) of the 3rd century AD but only published in
192720, the marginal readings of ms. 86, the Coptic versions and, especially, the Old
Latin. However, he ended his article with ambiguous conclusions: on the one hand he
recognizes that the Complutensis transmits very ancient readings which are lacking in the
known manuscripts; the editors had at their disposal eine Vorlage, die ganz
altertmliche Lesarten berlieferte, die heute in den uns bekannten Handschriften
fehlen.21 On the other hand, he mentions frequent retroversions from the Latin into the
Greek without deciding whether these interventions had been made by the editors or
were taken from Greek manuscripts already influenced by Latin readings. Another
eminent scholar of Septuagint studies, John W. Wevers, was not satisfied with Zieglers

17

See N. Fernndez Marcos, 'El texto griego de la Biblia Polglota Complutense,' in N. Fernndez Marcos
and E. Fernndez Tejero, Biblia y Humanismo. Textos, talantes y controversias del siglo XVI espaol,
(Madrid: FUE, 1997), 221-24.
18
Todo nuevo testimonio, como se demostr con motivo de la publicacin del Codex Washingtonianus
(W) en 1927, ha contribuido a reducir el nmero de lecturas exclusivas de la Complutense, muchas de las
cuales se atribuan hasta entonces a correcciones arbitrarias de los editores, see N. Fernndez Marcos, 'El
texto griego de la Complutense en Doce Profetas,' Sefarad 39 (1979), 6. And the same could be said of the
publication of the Dodekapropheton Scroll from Nahal Hever.
19
Delitzsch, Fortgesetzte Studien, 53.
20
See Verzeichnis der griechischen Handschriften des Alten Testaments von Alfred Rahlfs. Bd. I,1 Die
berlieferung bis zum VIII. Jahrhundert, bearbeitet von Detlef Fraenkel (Gttingen: Vandenhoeck &
Ruprecht, 2004), 387-89.
21
J. Ziegler, 'Der griechische DodekaprophetonText der Complutenser Polyglotte, Bib 25 (1944), 309.

solution, and he states in a critical review that the Complutensian text of

Dodekapropheton continues to be an interesting enigma.22
With the publication of the critical editions of the Greek Pentateuch in the
Gttingen series it has been confirmed that the Complutensian editors had ms. 108, sent
from Rome, as their main source for volume 1. However, they also had recourse to other
manuscripts.
Moreover, further studies have shown that somewhere after Exod 15 and before
Exod 25 the primary source changes to a manuscript of the fgroup of Wevers edition.23
Scholars have not been aware of this change of primary source. Therefore the Gttingen
editions misrepresent Compl because they attempt to present Compl on the basis of 108.24
This verification supports Cisneross statement when he alluded, in the preface to the
reader, to other manuscripts he had purchased with much effort and large sums of money.
In the second volume a significant unevenness in editorial approach can be appreciated.
In Joshua there is a change of primary source from a ms. of the fgroup which is no
longer known (not ms. 56 as Margolis thought!) to the 108 again at the end of the book.25
Therefore, it is at least surprising that another prominent scholar, M. L. Margolis, in an
unedited study on the Complutensian text of Joshua, recently published by L.
Greenspoon, expresses a very critical judgement towards the editors of the Alcal
polyglot. He attributes to retroversion of the editors from the Latin into 'Spanish Greek'
all the readings that do not agree with mss 108 or 56 of Rahlfss catalogue, even when the
reading can be found in other known Greek manuscripts.26 His argument cannot be
sustained in the light of recent research. As OConnell states: There is only the slightest
trace of what Margolis termed 'Spanish Greek' Greek resulting from editorial
22

Ihr Text bleibt ein interessantes Rtzel ... Die Compl. Polyglotte enthlt auch eine Reihe von
nirgendswo bezeugten Sonderlesarten ... Es fragt sich, ob nicht der Text von einer anderen griech.
bersetzung beeinflusst sein knnte; denn absichtliche Revision, wie sie Ziegler vermutet, ohne die
Zugrundelegung irgendwelcher Hss. ist auch keine zufriedenstellende Lsung, see J. W. Wevers,
'Septuaginta Forschungen,' TR u 22 (1954), 105.
23
OConnell, From Most Ancient Sources, 19 and 35: The first volume of Compl is edited with
considerable care and fidelity to all the texts involved.
24
For instance, the editor or editors of Leviticus may not be accused of constructing the text. There is
some slight stylistic modification but otherwise all the readings in the Greek column are attested in the two
mss. at the editors disposition," see OConnell, From Most Ancient Sources, 28. Even the most recent
edition by U. Quast, Septuaginta. IV, 3 Ruth (Gttingen: Vandenhoeck & Ruprecht, 2006), 14, states:
"Compl. wird notiert, wenn ihre Lesart von 108 abweicht."
25
OConnell, From Most Ancient Sources, 75.
26
L. Greenspoon, 'Max L. Margolis on the Complutensian Text of Joshua,' BIOSCS 12 (1979), 4356.

retroversions. ... The misplaced identification [of ms 56] leads him [Margolis] to see
much more editorial activity in Compl than there actually was and to identify certain
readings as 'Spanish Greek' when in fact they were to be found in what may be termed the
ftype source. The term 'Spanish Greek' is unfortunate because it leads to
overgeneralization in the evaluation of Compl readings which are not found elsewhere.27
In a perspicuous study D. Barthlemy called the attention of scholars to a singular
phenomenon of the Greek Ezekiel: the close connection between the Complutensis and
the most ancient witness of this book, Papyrus 967, in the description of the temple. Pap
967 belongs to the end of the 2nd or the beginning of the 3rd century AD. In this part of
the book (Ez 40:4246:24) there are 215 readings of the papyrus which are not backed up
in the manuscripts of Zieglers apparat. Interestingly, 130 of these readings agree exactly
with the text of the Complutensian Polyglot. It would have been difficult to give credit to
the antiquity of these Complutensian singular readings had they not been confirmed by
the discovery and publication of new witnesses such as Pap 967. Barthlemy is convinced
that the editors of the Alcal Polyglot had a manuscript at their disposal, albeit
fragmentary, which has now disappeared but which belonged to a textual family clearly
different from that of codex Vaticanus that embraces more or less the rest of the textual
tradition. He recalls the high esteem of Lagarde towards the Complutensian text, laments
that the Alcal Polyglot had not been collated in the first volumes of the Gttingen series,
and concludes: La Complutensis se confirme donc, de plus en plus clairement, comme
un tmoin textuel de haute valeur, quoique nigmatique; et Robert Hanhart a eu
parfaitement raison de lui faire place, partir de 1960, dans la Septante de Gttingen.28
Finally in his edition of 2 Ezra R. Hanhart is more cautious with regard to the
singular text of the Complutensis. As a result of Barthlemys article, it would be
necessary, he says, to submit to a new analysis those readings in agreement with the
Masoretic text and which are lacking in the Lucianic recension represented by ms.108.
27
OConnell, From Most Ancient Sources, 75 and 78, and OConnell continues: Although there are
retroversions and errors in Compl, they are not as frequent as Margolis maintains; they are to be attributed
more to the characteristics of the source mss than to editorial intervention or negligence.
28
D. Barthlemy, 'Les relations de la Complutensis avec le papyrus 967 pour z 40,42 46,24,' in D.
Fraenkel, U. Quast and J. W. Wevers, eds, Studien zur Septuaginta Robert Hanhart zu Ehren. Aus Anlass
seines 65 Geburtstages. MSU 20 (Gttingen: Vandenhoeck & Ruprecht, 1990), 261. For more details see
D. Barthlemy, Critique textuelle de lAncien Testament. Tome 3. zchiel, Daniel et les 12 Prophtes.
OBO 50/3 (Fribourg/Gttingen: ditions Universitaires/Vandenhoeck & Ruprecht, 1992), CXXV-CXLIV.

Hanhart concludes that no sure conclusion can be reached regarding the use of
manuscripts which are now lost or the reconstruction of the text by the editors.
Notwithstanding, he observes that the eclectic use of mss. has led the editors to choose
the reading of 108 which is closer to the MT, as can be proven in the utilization of the
Lucianic doublets, taking the part of the doublet which is closer to the Hebrew.29
Another intriguing connection between the Complutensis and the extant
manuscripts was found when editing the critical edition of Exodus. It has been analyzed
by the editor J. W. Wevers and a research scholar of the SeptuagintUnternehmen, D.
Fraenkel, in separate contributions. In the Ambrosian library of Milan there is an
important uncial manuscript of the 5th century AD. (S. P. 51, former A.147 inf.), F
according to Rahlfss Catalogue.30 In the Middle Ages (12th century) this codex had been
systematically corrected in cursive script by a corrector in an attempt to bring the text
closer to the Hebrew (Fb in Weverss edition). A later corrector has completed the
omissions of the manuscript using a Vorlage that is in an order and textual form closer to
the Hebrew than the Greek Exodus. This current text of the later corrector or suppletor is
called Fh in Weverss edition. Interestingly enough, the text of this medieval Jewish
corrector agrees with the Septuagint column of the Complutensian Polyglot. As a matter
of fact Fh and Complutensis agree with the Hexaplaric witnesses in the outline and
distribution of the text but not in the textual form of the completed text, where both
witnesses go their own way against the Hexaplaric recension. One conclusion can be
drawn: that both witnesses follow a common textual tradition. From the ancient printed
editions only the Complutensian offers a text similar to that of Fh in this part of Exodus.
It means that the editor knew either the F manuscript itself with its medieval corrections
or the Jewish tradition on which these corrections are based.31 The BrookeMcLean
edition published, in an Appendix to Exodus, this text for Ex 36:839:43 (according to
29

R. Hanhart, Septuaginta. Vol. VIII/2 Esdrae Liber II (Gttingen: Vandenhoeck & Ruprecht, 1993), 25.
Verzeichnis der griechischen Handschriften. Bd. I,1 Die berlieferung bis zum VIII. Jahrhundert,
bearbeitet von Detlef Fraenkel, 22931. As indicated in an ancient inscription of the bind folium, it was
imported from Macedonia (ex Macedonia Corcyram advecta) and bought by cardinal Borromaeus for the
Ambrosian Library.
31
See J. W. Wevers, Septuaginta. II.1 Exodus (Gttingen: Vandenhoeck & Ruprecht, 1991), 78; D.
Fraenkel, 'Die Quellen der asterisierten Zustze im zweiten Tabernakelbericht Ex 35-40*,' in D. Fraenkel,
U. Quast and J. W. Wevers. eds, Studien zur Septuaginta, 14445 and 17484. In fact, several of these
corrections also agree with the translation into modern Greek published in 1547 in Hebrew script as a
30

the Hebrew = pp. 408436 of Wevers edition), taking the text of ms. G of the Hexaplaric
recension as the basis. Fh was included in the apparatus, but under the symbol of Fb.
According to Wevers, this Second Tabernacle Account of Fh is a revised text also based
on the Greek Exodus and not a new rendering of the MasoreticText. It is a revision based
on the Hebrew, not unlike that of Origens Hexapla, but with many more corrections
towards MT. On several occasions, both the Complutensis and Fh, support the same
variant, moreover the same sentence.32 I conclude with Weverss words: Close scrutiny
reveals that Compls support is stronger for the base text than for the readings which
constitute revisions towards MT. This must mean that one of the sources of Compl must
have shared in its stemmata a parent text which also lay in the textual ancestry of Fh. That
source is not one of the extant identified sources of Compl for the Pentateuch, viz. ms
108 and some of the f mss, but one no longer extant.33
Although most of these marginal glosses are anonymous, in the notes to Gen
47:31 and Exod 16:31 the readings of Fb are attributed to to; ijoud' ( = to;
ijoudaikovn). In fact, these readings follow the tradition of the ancient Jewish
versions, particularly Aquila (cf. Exod 6:3; 15:1,11; Num 23:19; 24:4; 25:6 ), and very
often agree with the lexicon of the Greek column in the Istanbul Pentateuch (1547).34
Some comments to this chain of evidence seem appropriate. First of all, it is
recognized that the Complutensian collaborators made use of other sources no longer
extant beside those commonly accepted for the Pentateuch, ms. 108 and one of the f
group.35 These conclusions confirm the statement of cardinal Cisneros in the preface to
the reader ut copia emendatorum codicum abunde supesset. Secondly, the base text
column of the Istanbul Polyglot Pentateuch, see N. Fernndez Marcos, Introduccin a las versiones griegas
de la Biblia. 2nd edition revised and augmented (Madrid: CSIC, 1998), 18389.
32
See J. W. Wevers, Septuaginta. II, 1 Exodus (Gttingen: Vandenhoeck &Ruprecht, 1991), 40811.
33
J. W. Wevers, 'A secondary text in Codex Ambrosianus of the Greek Exodus,' in R. Gryson ,ed,
Philologia Sacra. Biblische und patristische Studien fr Hermann J. Frede und Walter Thiele zu ihrem
siebzigsten Geburtstag. Band I (Freiburg: Herder, 1993), 48. OConnell recognizes that there is a
relationship between Fh and Compl only in the texts where the main text fails. Compl borrows from Fh and
not vice versa, see OConnell, From Most Ancient Sources, 53, n. 48. There is significant unevenness in
the editing of the Second Tabernacle Account. Like Fraenkel, OConnell accepts that it is a product of the
hellenists of Alcal, but he admits in this narrative the influence of the Vulgate to the shape of the Greek
column, see OConnell, From Most Ancient Sources, 7273.
34
Fernndez Marcos, Introduccin a las versiones griegas, 18283, and N. Fernndez Marcos , 'El
Pentateuco griego de Constantinopla,' Erytheia 6 (1985), 19899. Of the 488 readings that Fb preserves in
the book of Exodus, it shares 100 with the Istanbul Pentateuch, see Wevers, Septuaginta. II, 1 Exodus, 43
44.

10

followed in the Second Tabernacle Account of Exodus, forms part of a Jewish tradition
connected with the ancient Jewish versions of the Bible, especially Aquila, and which is
also present in the Greek column of the Jewish Istanbul Pentateuch of 1547. We lack the
proof for the actual link used by the Complutensian editors. But all the traces lead to the
conclusion that the contacts of cardinal Cisneros with these manuscripts from North Italy,
corrected by Jewish hands, was not only possible but quite plausible. The migration of
Greek manuscripts from Constantinople to Italy even before 1453 is well attested.36 The
main collection of the Marcian Library in Venice came from the legacy of cardinal
Bessarion. Cultural contacts between Spain and Venice, Milan, Rome and other cities of
Italy were frequent in the 15th16th centuries. Some of the Complutensian collaborators
such as Alfonso de Zamora or Pablo Coronel were conversos who had access to
Hebrew manuscripts and Jewish exegetical traditions. Moreover, Demetrio Ducas, the
Cretan, was called by Cisneros to collaborate in the edition of the Greek text of the
Complutensis when he was in Italy working on the publication of Greek texts for Aldus
Manuzius in Venice.37 Hernn Nez de Guzmn (also known as Pinciano and
Comendador griego), the other collaborator in the edition of the Greek text was twice in
Italy (probably in Bologna before and after 1499), where he purchased several Greek
manuscripts.38 Nebrija itself had studied ten years in Italy.
The philological criteria of the Renaissance for the edition of translated texts were
well established in the Preface of the Complutensian Polyglot: when the witnesses
disagree the truth relies on the originals: the Hebrew text for the Old Testament and the
Greek text for the New. It is not surprising that, given these criteria of textual criticism,
the Complutensian philologists gave priority to the manuscripts and readings which
supported the Masoretic Text. To this aim they sometimes preferred even the marginal
readings, whether they belonged to the Hexapla as in the ten first chapters of Job or to
35

The following mss. belong to this group in the Pentateuch: 53 56 129 246 and 664.
J. Signes Codoer, 'Translatio studiorum: la emigracin bizantina a Europa Occidental en las dcadas
finales del Imperio (13531453),' in P. Bdenas de la Pea and I. Prez Martn (eds.), Constantinopla 1453.
Mitos y realidades. Nueva Roma 19 (Madrid: CSIC, 2003), 189-246. In relation with the Concile of
Florence (1438/39) Signes ('Translatio studiorum,' 232) states: "Muchos eruditos griegos trajeron desde
Bizancio manuscritos de autores antiguos que difundieron entre sus colegas humanistas, presentes en las
sesiones conciliares."
37
J. Lpez Rueda, Helenistas espaoles del siglo XVI (Madrid: CSIC, 1973), 20.
36

11

para-Hexaplaric notes and texts as in the late part of Exodus. As Delitzsch stated, which
other criterium could they have followed at the time? They thought that the ancient Greek
translators had before them the same textus receptus which had remained unaltered since
the time of the translation.
The last study of the Complutensian Greek text by S. OConnell reveals that one
can hardly speak of one text of the Alcal polyglot: The Greek column is an eclectic text
constructed by a number of editors who worked semiindependently."39 These editors
apply different procedures. They usually follow a main manuscript as the base which they
correct with readings of another manuscript. But the manuscript chosen as the base may
change from book to book, from one part of the Bible to another. This may explain in
some way the enigma of the Complutensis.
OConnell slightly modifies Fraenkels conclusions on Exodus 3540. The Greek
column bears traces of two hands, one non-interventive and the other quite interventive.
The incorporation of marginal glosses is a significant feature of the Complutensian text,
but there is no convincing evidence of spontaneous retroversion from MT or Vulgate.
The editor made an effort to find textual sources for all the readings in the Second
Tabernacle Account.40
Concerning the work of the hellenists of Alcal in the second volume OConnell
concludes that they executed their task with noteworthy seriousness: the evaluation of
their work must strive to avoid facile and simplistic solutions."41
As far as Ezekiel is concerned Barthlemys conclusions are confirmed. The
sounding of other chapters shows that there is a relationship, albeit hidden, with Pap 967
for some passages where 967 is still extant. There is an increase of unique Complutensian
readings in the first ten chapters of Ezekiel, a section for which 967 is not extant. These

38

J. Signes Codoer, C. Codoer Merino and A. Domingo Malvadi, Biblioteca y epistolario de Hernn
Nez de Guzmn (El Pinciano). Una aproximacin al humanismo espaol del siglo XVI. Nueva Roma 14
(Madrid: CSIC, 2001), 7-12.
39
OConnell, From Most Ancient Sources, 166. However, the number of editors should not be exaggerated.
We know only three or at most four editors for the Greek column: Demetrio Ducas, Hernn Nez (or
Pinciano, disciple of Nebrija)) and Lpez de Ziga. Perhaps they were helped by Juan de Vergara.
40
OConnell, From Most Ancient Sources, 3773.
41
OConnell, From Most Ancient Sources, 103.

12

unique readings are to be attributed more to the characteristics of the lost primary source
than to the interventions of a particular editor.42
If we pay attention to the shape of the text it must be said that the Vulgate was
used as a guide in establishing the text. Vulgate does not always influence the
Complutensian text, but it is a constant factor in the editing. The Masoretic text is of
lesser influence. It is significant in longer texts that it is Vulgate which acts as a guide in
interpreting the Hebrew. In Ezekiel, a very clear influence of Jeromes commentary is
evidenced in the final chapters. OConnell concludes his important monograph with the
following assertions: It is possible to distinguish contrasting techniques and tendencies
throughout the Greek column: in the degree of Vg use; in the use or non-use of MT; in
the incorporation of marginal glosses; in the use of Jeromes commentary; in the style of
combining source mss and in differing stylistic and orthographic criteria. It must,
however, be stated that the stylistic modification is quite restrained. There is no question
of the editors changing to have a more elegant text. Their modification is aimed at
restoring what had become corrupt or was lost. They were people of their time and
Compl testifies to their expertise within their time.
The practical effect of this is to show the editorial activity of the Complutensian
Hellenists in a much calmer light, especially that the amount of borrowing from the
secondary source is comparatively limited as is the number of readings attributable to
spontaneous editorial intervention."43
Behind all the efforts of the editors lies the most important component of all: the
manuscripts which the editors used. It is the manuscripts which give the fundamental
colour to the Greek column. It is still valid to maintain that the text critical value of this
text lies in the value of its sources.
As we have shown above, for the first volume, the Pentateuch, and the second,
Joshua2 Chronicles, the main sources of the Greek column were ms 108 and a second
manuscript of the fgroup. But for Judges and the following books, the editors had at
their disposal a third manuscript, a partial copy of ms. 68 sent to Alcal by the Venetian
42

OConnell. From Most Ancient Sources, 163: The most significant result of the investigation of Ezek
42c was that many of the unique Compl readings, rather than being supposed retroversions were in fact preHexaplaric readings of very high quality... The specific Compl readings show that the editor had access to a
ms no longer extant.
43
OConnell, From Most Ancient Sources, 168.

13

Senate. It is the ms. 116Z36 (22 of VillaAmil, 442 of Rahlfs).44 A printed text of this
manuscript for Judges 5, Deboras song, and 2 Samuel 1:1926, Davids elegy over Saul
and Jonathan, was published by Delitzsch according to a copy sent by Pascual de
Gayangos, the librarian of the Madrid University.45 OConnell describes the use of this
manuscript made by the Complutensian philologists with the following words:
However, with the appearance of ms 442 for Judges, the editors had available to them, a
ms which was much closer to ms 108. The sudden swich to ms 442 then becomes more
understandable: in Judges the close relationship between mss 108 and 442, while not
members of the same ms group, offered the editors a more stable path in establishing the
Greek column and provided a sense of security as to where the LXX lay.46
My interest for this manuscript goes back to my postgraduate studies at the
Complutensian University of Madrid. A document signed by the director of the
Complutensian Library of 5th June 1973 informed that ms 22 ( = 442) with other Hebrew
and Latin Bibles were not accessible to the researchers since they had to be restored. My
interest increased when in the Gttingen edition of 1 Ezra, in the description of the
available manuscripts R. Hanhart stated on 442: Die Hs. ist laut Mitteilung der
Bibliothek vom 24.10.1969 im spanischen Brgerkrieg verbrannt.47
My joy was still greater when I could verify two years ago that ms. 22 was not
completely lost but only damaged, that it had been partially restored, even digitized, and
that I could have in my hands the extant folios and fragments of the presumed burnt
manuscript. In a separate publication I have made known and described the contents and
remnants of the restored manuscript.48

44

J. VillaAmil y Castro, Catlogo de los mss. Existentes en la Biblioteca del noviciado de la Universidad
Central (procedentes de la antigua de Alcal). Parte I: Cdices, Madrid, 1878, 56.
45
Delitzsch, Fortgesetzte Studien, 1317.
46
OConnell, From Most Ancient Sources, 86. Actually, in Judges ms. 108 belongs to the Hexaplaric group
and ms. 442 to the group MNhyb2 of BrookeMcLean, although ms. 442 was not collated by Brooke
McLean.
47
R. Hanhart, Septuaginta. Vol. VIII/1 Esdrae Liber I (Gttingen: Vandenhoeck & Ruprecht, 1974), 14.
Information repeated in the specialists publications until now, see U. Quast, Septuaginta IV, 3 Ruth, 10,
and OConnell, From Most Ancient Sources, 82: While the ms has been rendered unreadable by the
ravages of the Spanish Civil War ...
48
N. Fernndez Marcos, Un manuscrito complutense redivivo: Ms. griego 442 = VillaAmil 22,
Sefarad 65 (2005), 65-83.

14

15