666
1 AUGUSTINE'S CITATIONS AND TEXT OF THE GOSPEL ACCORDING TO JOHN by HUGH ALEXANDER GERVASE HOUGHTON A thesis submitted to The University of Birmingham for the degree of DOCTOR OF PHILOSOPHY Department of Theology and Religion School of Historical Studies The University of Birmingham July 2006

Augustine's citations and text of the Gospel according to Johnetheses.bham.ac.uk/249/1/Houghton06PhD.pdf · AUGUSTINE'S CITATIONS AND TEXT OF THE GOSPEL ACCORDING TO JOHN by HUGH

Embed Size (px)

Citation preview

1

AUGUSTINE'S CITATIONS AND TEXT OFTHE GOSPEL ACCORDING TO JOHN

by

HUGH ALEXANDER GERVASE HOUGHTON

A thesis submitted toThe University of Birminghamfor the degree ofDOCTOR OF PHILOSOPHY

Department of Theology and ReligionSchool of Historical StudiesThe University of BirminghamJuly 2006

Please note that a revised and expanded version of this thesis has been published as: H.A.G. Houghton, Augustine's Text of John. Patristic Citations and Latin Gospel Manuscripts. (Oxford Early Christian Studies). Oxford University Press, 2008. ISBN 978-0-19-954592-6. It is also available in Oxford Scholarship Online http://www.oxfordscholarship.com/oso/public/content/ religion/9780199545926/toc.html

This thesis is provided for reference only at http://etheses.bham.ac.uk/

citations should, if possible, be made to the published book.

University of Birmingham Research Archive

e-theses repository This unpublished thesis/dissertation is copyright of the author and/or third parties. The intellectual property rights of the author or third parties in respect of this work are as defined by The Copyright Designs and Patents Act 1988 or as modified by any successor legislation. Any use made of information contained in this thesis/dissertation must be in accordance with that legislation and must be properly acknowledged. Further distribution or reproduction in any format is prohibited without the permission of the copyright holder.

2

Abstract

This study assesses Augustine's worth as a witness to the text of the Bible and

evaluates his evidence for the Gospel according to John. The full collection of

citations is presented in the Appendix. In the analysis a distinction is proposed

between primary citations, which Augustine makes with reference to a scriptural

codex, for example when preaching, and secondary citations, for which this cannot

be demonstrated. The latter constitute the majority and often correspond to his

mental text, a consistent form of a verse showing characteristic alterations

attributable to memory. In polemical works, Augustine displays a different form of

text which he has normally adopted from his opponents. Such variations in the

biblical text suggest that the citations have been transmitted accurately, without

interference by copyists. Augustine's text of John demonstrates the continuity in

the Latin Bible between Old Latin versions and Jerome's Vulgate. Most of the non-

Vulgate renderings in Augustine's citations are paralleled in one or more Old Latin

witnesses, which suggests that the Old Latin texts known today are a representative

selection. Nonetheless, his primary affiliation is with the Vulgate, which even comes

to permeate citations made from memory in later works.

3

To the Diocese of Birmingham

4

Acknowledgements

Just when I thought I had finished my academic career in order to enter into ordained

ministry came the invitation to join the team working on the Vetus Latina edition of the

Gospel according to John and to undertake the research which has led to this thesis. I am

very grateful to those who made it possible to accept this call, especially Bishop Mark Santer

and Canon Marlene Parsons, and in recognition I dedicate this study to the Diocese of

Birmingham. I should also express my particular thanks to the Reverend Michael Castle and

the people of St Gabriel, Weoley Castle, for their support during the last three years.

Returning to research has been a joy, principally for the pleasure of working with such a

friendly and dedicated team of colleagues. Chief among them are Professor David Parker and

Dr Philip Burton who supervised this research. I am grateful to them not just for their

benevolent oversight, but also for their considerable generosity in offering me so many

opportunities in the academic field. They have shared their time and knowledge, lent me

hard-to-find books, and been patient with my inexperience and over-confidence. I would like

to thank the many other scholars involved in the International Greek New Testament Project

edition of the Gospel according to John for their friendship and advice, especially Dr Jon

Balserak and Dr Roderic Mullen who also lent me books and papers. Professor David Wright

kindly allowed me to borrow his microfilms of some of the manuscripts of the Tractatus in

Iohannis euangelium and gave me offprints of his articles. Others have also shared their

research or expertise, including Verity Allan, Dr Mark Goodacre, Dr Hildegund Mller and Dr

Catherine Smith. In addition, I thank the librarians whom I have encountered during the

course of this research, and acknowledge the generous financial support of the Arts and

Humanities Research Council.

My family remains a constant and reliable source of encouragement and critical advice. It is a

pleasure once more to express my gratitude to them for many years of love and support,

shown in so many ways. It is an even greater delight to be able to include among their

number my beloved wife Josephine, who has both provided a welcome distraction from this

thesis and made its completion possible.

5

Contents

Chapter One: The collection of biblical citations

1.1 Opening 11.2 Patristic citations and the Old Latin Bible: review of previous work 31.3 Methodology and principles of this collection 131.4 Presentation of citations 19

1.4.1 Additional material: manuscript readings 201.4.2 Additional material: introductions 22

1.5 Identification of citations and questions of accuracy 231.5.1 Classification of citations 27

1.6 Conclusion 36

Chapter Two: Augustine and the Bible

2.1 Introduction 382.2 Augustine's biblical education 39

2.2.1 Augustine's library and scriptorium 452.2.2 Knowledge of languages 47

2.3 Augustine, Jerome and the Latin Bible 522.3.1 Jerome and the Vulgate 542.3.2 The correspondence of Augustine and Jerome and the

Latin Old Testament 572.3.3 Augustine, the Vulgate and the Itala 602.3.4 Augustine as biblical reviser? 64

2.4 Use of the Bible in public and private 672.4.1 The Bible in liturgy 682.4.2 Stenographers in the early Church 742.4.3 Use of the Bible in public debates 782.4.4 The readership of the Bible 802.4.5 Augustine's methods of composition 82

2.5 Augustine's attitude to Scripture 842.5.1 Introductions and scriptural references 862.5.2 The canon of Scripture 912.5.3 Textual variance 932.5.4 Citation context and exposition 962.5.5 Internal parallels and biblical "chains" 100

2.6 Augustine's "mental text" of the Bible 1052.6.1 "Flattening" 1062.6.2 Conflations 109

2.7 Conclusion 111

6

Chapter Three: Characteristics of Augustine's text ofJohn in each work

3.1 Introduction 1133.1.1 Old Latin manuscripts of John 115

3.2 Primary sources (1):Tractatus in Iohannis euangelium (AUJo) 406-7, 418-422 118

3.2.1 AUJo: the text of the commentary(continuous text and sequential variants) 123

3.2.2 AUJo: non-sequential citations and titles 1333.2.3 Readings of individual tractates:

AUJo 7, 30, 42, 49, 61, 112-114. 1373.3 Primary sources (2) 152

3.3.1 De consensu euangelistarum (AUEv) 403/4 1523.3.2 Speculum quis ignorat (AUspe) 427/8 1623.3.3 Sermones 117-147A 1653.3.4 Other sermons with Johannine lections 1763.3.5 Enarrationes in Psalmos (AUPs) 394-422 183

3.4 Secondary sources 1893.4.1 Early works 190

3.4.1.1 De diuersis quaestionibus (AUq) 388/397 1913.4.1.2 De uera religione (AUrel) 390 1953.4.1.3 De sermone domini in monte (AUs dni) 392/7 1953.4.1.4 De mendacio (AUmen) 395 1963.4.1.5 Confessiones (AUcf) 397-403 1963.4.1.6 De doctrina christiana (AUdo) 397, 426 197

3.4.2 De trinitate (AUtri) 400-3, 411-422 1973.4.3 Middle period works and commentaries 203

3.4.3.1 Adnotationes in Iob (AUJb) 400/5? 2043.4.3.2 De baptismo (AUba) 404 2053.4.3.3 In epistolam Iohannis ad Parthos tractatus decem (AU1Jo) 407 2063.4.3.4 De peccatorum meritis et remissione et de baptismo

paruulorum (AUpec) 411/2 2073.4.3.5 De Genesi ad litteram (AUGn li) 404/5, 412/4 209

3.4.4 Later theological works 2093.4.4.1 De ciuitate dei (AUci) 412-426/7 2103.4.4.2 Quaestiones in Heptateuchum 419 2113.4.4.3 De gratia et libero arbitrio (AUgr) 426;

De correptione et gratia (AUcorr) 426/7 2113.4.4.4 Retractationes (AUre) 427 2123.4.4.5 De dono perseuerantiae (AUpers) and

De praedestinatione sanctorum (AUprae) post 429 2123.4.5 Collections 214

3.4.5.1 Sermons 2143.4.5.2 Letters 218

3.5 Conclusion 220

Appendix to Chapter Three:Distinctive Vulgate readings in John from Matzkow-Jlicher-Aland 224

7

Chapter Four: The text of John in Augustine's polemical works4.1 Introduction 232

4.1.1 Augustine and sectarian translations 2344.2 The text of John in works against the Arians 238

4.2.1 Contra sermonem Arrianorum (AUAr) 419 and theSermo Arrianorum 238

4.2.2 Collatio cum Maximino (AUMax co) andContra Maximinum (AUMax) 428/9 241

4.3 The text of John in works against the Manichees 2464.3.1 Contra Fortunatum Manichaeum (AUFo) 392 2524.3.2 Contra Adimantum (AUAd) 394 2544.3.3 Contra Faustum Manichaeum (AUFau) 400/2 2564.3.4 Contra Felicem Manichaeum (AUFel) 404 261

4.4 The text of John in works against the Donatists 2634.4.1 Contra litteras Petiliani (AUPet) 400-5 2674.4.2 Contra epistulam Parmeniani (AUPar) 403/4 2694.4.3 Contra Gaudentium donatistarum episcopum (AUGau) 418/9 271

4.5 The text of John in works against the Pelagians 2734.5.1 Contra Iulianum (AUJul) 421/2; Contra secundum Iuliani

responsionem imperfectum opus (AUJul im) 428-9 2744.6 The text of John in other authors cited by Augustine 278

4.6.1 De natura et origine animae (AUan) 419/420 2784.6.2 Contra aduersarium legis et prophetarum (AUleg) 420 2804.6.3 Catholic sources 281

4.7 Conclusion 282

Chapter Five: Augustine's readings in John: contributions to textual criticism

5.1 Introduction 2855.2 A textual commentary on selected readings of Augustine in John 287

5.2.1 Old Testament citations in John 3685.3 Augustine and textual traditions 371

5.3.1 Old Latin readings unique to Augustine 3715.3.2 Augustine as early witness to the Vulgate 3755.3.3 Diatessaronic readings 3805.3.4 Augustine's significance for the Greek text 3845.3.5 Additional material and unusual readings 3865.3.6 Characteristics of possible alteration to the biblical text

by Augustine 3895.3.7 Did Augustine revise the text of John? 394

5.4 Conclusion 401

Conclusion

Summary 404Avenues for further research 407

8

Appendix: Complete listing of all citations of Johnin the works of Augustine

List of abbreviations and editions used 410Conversion table for the Sermones Dolbeau 417Appendix 418

Bibliography

1. Reference works and editions 6572. Secondary literature 659

1

Chapter One

The collection of biblical citations

1.1 Opening

diuinae lectiones omnes ita sibi connectuntur tamquam una sit lectio, quia omnes exuno ore procedunt. multa sunt ora ministerium sermonis gerentium, sed unum est osministros implentis. (Augustine, Sermones 170.1.1).

Is there really only one reading of the Bible? What happens when, in the words of

Augustine's favourite biblical book, the word becomes flesh on either the lips of preachers or

the parchment of manuscripts? How is the biblical text treated? Does a consistent form

emerge? What is the effect of translation? How closely do the citations of ancient writers

correspond to the surviving manuscripts?

In this study of Augustine's citations and text of the Gospel according to John, I shall

investigate all these questions at a critical period in the history of the Latin Bible. During the

lifetime of Augustine (354-430), the most significant and prolific Latin Church Father, an

event took place which determined the subsequent shape of the Bible in the West for over a

millennium.1 The papal commission of a version which would become a standard, the Vulgate,

1 There are numerous biographies of Augustine, starting with that of his disciple Possidius ofCalama. Among modern contributions, Peter Brown's classic survey of 1967 has recentlybeen issued in a new edition (Brown 2000), while an English version of Serge Lancel's 1999French biography has also appeared (Lancel 2002). Shorter studies include Clark 1994 andWills 1999.

2

led to the text which continues to constitute the Latin Bible today.2 Prior to this, several

translations had been made from copies of the Scriptures in Greek by Latin-speaking

communities: the only remaining evidence for these is preserved in a handful of surviving

manuscripts and the citations of the Church Fathers. This "Old Latin" tradition, or Vetus

Latina, is an important witness to the early history of the biblical text as well as the

development of theological language and concepts in the West. The scriptural citations and

comments made by Augustine, who saw the rle of biblical exegete as one of the principal

callings of pastors and theologians, offer an unrivalled body of evidence for the state of the

Old Latin text and the early reception of the Vulgate.3

The Gospel according to John has a central place in Augustine's writings, and is probably the

book of the New Testament which he cites most frequently. It serves as a useful test case

to address both theoretical and practical issues in the collection of patristic citations and

their application to the textual criticism of the Latin Bible. How are citations identified or

defined? Is the text of each verse consistent, or is there variation over time or in different

2 On the history of the term "Vulgate", see Sutcliffe 1948; it does not appear to have beenused of Jerome's version until the Council of Trent. In the time of Augustine, uulgata editiowas used for the Septuagint and its Old Latin translation (exemplified at De ciuitate dei16.10.2).3 One of the weaknesses of the term Vetus Latina is that it has been defined to included allreadings which do not correspond to the Vulgate: "Im Neuen Testament bedeutet der Nameheute nur eine Abgrenzung gegenber der Vulgata" (Fischer 1972:4). This runs the risk oftreating them as two separate entities, as well as obscuring the variety within the Vulgatetradition: although I shall continue to use these terms for convenience, taking the editorialtext of Weber-Gryson as representative of the "Vulgate", in reality both traditions should belocated on a continuum of the Latin Bible. Hort observes that "for some purposes thedistinction is convenient and almost necessary: but it disguises the fact that there is a widerdifference between the earlier and later stages of the 'Old Latin' (in this comprehensivesense of the term) than between the later stages and the Vulgate" (Westcott-Hort1882:78). Petzer claims that "most probably not one single 'pure' Latin MS of the firstmillennium has survived" (1995:119), noting contamination in both directions betweenVulgate and Old Latin traditions. Fredouille 1985:28 suggests that Veteres Latinae would bea better description of the diverse earlier tradition (thus also Garca de la Fuente 1994:126).

3

works? Did Augustine quote from memory or use a biblical codex? How accurate is the

manuscript tradition of his works? Once the information has been gathered, the evidence of

Augustine can be set in the context of the surviving Old Latin manuscripts and also used to

shed light on his use of the Bible in his own theological development.

1.2 Patristic citations and the Old Latin Bible: review of previouswork

The importance of scriptural citations in the early Church Fathers has long been appreciated

by textual scholars.4 The provenance of biblical manuscripts is often unknown and, in the

absence of other indications, their dating relies on palaeographical evidence. The location and

dates of Christian authors, however, can be established from both internal and external

evidence with a greater degree of certainty. This means that patristic citations have often

been used to classify or identify biblical text-types. Furthermore, Church Fathers who

predate surviving biblical manuscripts provide the earliest attestation of a particular verse.

Some authors, too, may supply readings which are not attested in any other witnesses. In

the Old Latin tradition, the textual importance of citations is much greater because of the

very small number of surviving manuscripts which preserve a genuine Old Latin text.5

4 The first edition of the New Testament to include Greek patristic citations was the EditioComplutensis of 1514, which led Erasmus on to more systematic research (Prigent1972:436). For more on the history of the use of citations, see also Duplacy & Suggs 1971.Parker 1997:16 gives a concise summary of the problems of recovering a biblical text frompatristic citations.5 Hort observes that "Latin Fathers ... constitute a not less important province of Old Latinevidence than the extant MSS. ... Even in the Gospels their aid is always welcome, often ofthe highest value" (Westcott-Hort 1882:83). A comprehensive introduction to Old Latinmanuscripts of the Gospels is found in Burton 2000, which may be supplemented for otherbooks of the Bible and the Vulgate tradition by Elliott 1992.

4

The first scholar to begin a systematic assembly of Old Latin manuscripts and citations was

Petrus Sabatier with his Bibliorum sacrorum Latinae uersiones published in 1743.6 At the end

of the nineteenth century, Paul de Lagarde in Gttingen and J.W. Burgon in London

independently made collections of Augustine's citations, although in his survey of Latin

patristic citations Frede comments that, unfortunately, neither is suitable for further study.7

A more significant development was the inauguration of the Oxford Old Latin Biblical Texts

series, which featured critical editions of individual codices, comparisons of different

manuscripts and one volume of citations.8 These studies were pursued in tandem with a

critical edition of the Vulgate New Testament by Wordsworth and White: the four Gospels

appeared between 1889 and 1898. While this was in progress, Augustine's citations were

the subject of independent work by Francis Crawford Burkitt. In the second half of The Old

Latin and the Itala (1896) he contended that Augustine was the first Latin Father to adopt

Jerome's Vulgate text of the Gospels and went so far as to claim that the Itala praised by

Augustine in De doctrina christiana 2.15.22 was identical with Jerome's Vulgate. As Itala had

already become a generic term to describe Old Latin versions, this suggestion met with some

6 I have encountered references to two earlier works specifically on Augustine's citations: theBiblia augustiniana of David Lenfant in Paris 1661 (La Bonnardire 1986:445 describes it as"un index des citations bibliques de saint Augustin ... l'ouvrage est compltement prim,mais toujours utile et rimprim") and a collection of Psalm citations by Lefvre d'taples in1508 (De Bruyne 1931:522).7 "Leider wurde dieses Material nicht fr die weitere Forschung fruchtbar gemacht." (Frede1972:476). Duplacy describes Burgon's collection of 86,439 New Testament citations,collected between 1872 and 1888 as "difficiles utiliser et mettre jour" (Duplacy &Suggs 1971:196). De Lagarde's collection comprised 13,276 Old Testament and 29,540New Testament citations (De Bruyne 1931:522); Knauer (1955:26) states that he was notable to use de Lagarde's work for his study of Augustine's psalm citations.8 The first volume to appear was Wordsworth The Gospel according to St Matthew from theSt. Germain MS (OOLBT 1) Oxford 1883, followed by Wordsworth, Sanday and WhitePortions of the Gospels according to St. Mark and St. Matthew (OOLBT 2) Oxford 1886. Thecitations were from Irenaeus of Lyon, collected in Sanday and Turner Nouum TestamentumSancti Irenaei Lugdunensis (OOLBT 7) Oxford 1923. Sanday pioneered the concept ofbersetzungsfarbe to characterise different Latin traditions: see Fischer 1972:7-9 and Frede1971:86-87.

5

resistance. Burkitt re-stated his thesis in 1910, assembling more evidence and stating the

need for "a study of the character and affinities of the pre-Vulgate gospel quotations in

Augustine's earlier works" (1910:458).9

This request was taken up by C.H. Milne, in A Reconstruction of the Old-Latin Text or Texts

of the Gospels used by Saint Augustine with a Study of their Character (Cambridge 1926).

Milne assembled gospel citations from fifty works of Augustine which he dated prior to 401,

in an attempt to prove Burkitt's hypothesis that the Vulgate Gospels were adopted at Hippo

around 398. The work, however, is deeply flawed, as Burkitt demonstrated in his review.10

Milne assumed that Augustine's text would have been African in character, and used the

term "Old-Latin" to describe agreements between Augustine and the text of Codex Bobiensis

and Codex Palatinus, also referring to the citations of Cyprian.11 With the exception of Codex

Brixianus, he ignored the other surviving Old Latin manuscripts and the question of

Augustine's accuracy in citations, noting some 2400 instances in which the phraseology of

Augustine coincides neither with that of the Old-Latin, nor with that of the Vulgate

(1926:xiii). His identification of Augustine's textual affiliation relied on much smaller figures,

9 Burkitt sought to counter the position that "in the present day S. Augustine is almostinvariably considered as an 'Old Latin' authority for all parts of the Bible and in all hiswritings, and the cautions issued by Sabatier himself are generally unheeded" (1896:1-2).Vogels and Denk argued that the Vulgate nature of Augustine's citations in certain writingswas due to later corruption in the manuscript tradition (see Burkitt 1910:259). AlthoughBurkitt's claims are now generally accepted, it is still possible to find some, who, like Garcade la Fuente, treat Augustine solely as an Old Latin source: "en las obras de San Agustn ...se encuentran ms de 50.000 citas bblicas, casi todas ellas de la VL, pues el Santo Doctorprest poca atencin a la versin de su amigo Jernimo." (1994:146-147). Although Italawas initially used to designate the Latin Bible in the Thesaurus Linguae Latinae, from 1990this has been changed to Vetus Latina (Garca de la Fuente 1994:126).10 See Burkitt 1927, drawn on by Metzger who describes it as "an incompleted and ill-digested collection of data, some of which are irrelevant to the subject" (1977:361 note 1).11 Of the 1941 Gospel citations collected by Milne, 802 agree with these manuscripts againstthe Vulgate, 53 agree with Cyprian against the Vulgate, 1138 agree with the Vulgateagainst the manuscripts and 25 agree with the Vulgate against Cyprian (Milne 1926:xii).

6

and the 5% shift towards "Vulgate" readings which he observes in works written between

397 and 401 seems insignificant.12 Furthermore six of the seven latest works, including De

Baptismo from 400, have a clear majority of forms identified as "Old-Latin".13 Although this

work is often cited, its value for Augustine's text is also limited: seventeen of the works

Milne consulted were not available in a critical edition.

Another suggestion was advanced shortly afterwards by Donatien De Bruyne. Having

observed that, in the Pauline epistles, the text of Augustine's later citations corresponded

exactly with the Old Latin Freising fragments, he contended that Augustine had revised the

biblical text while writing his commentaries on Romans and Galatians.14 In his article "Saint

Augustin Reviseur de la Bible" (1931) he expanded the scope of this revision to include the

text of the Heptateuch, the Psalter, the Gospels and the other Pauline letters. For the most

part, however, De Bruyne relied on internal criteria, and attributed to Augustine a far greater

level of competence in Greek than is generally accepted. In the absence of any

incontrovertible evidence that Augustine is responsible for the renderings which he prefers, it

seems more likely that he relied on existing translations.15

12 Before 398, Milne reckons that "Vulgate" citations show a 3% majority, while worksbetween 398 and 401 (excluding De consensu euangelistarum) only display an 8% majorityover "Old-Latin" forms (1926:xiii). In order to confirm this shift, information should havebeen gathered from later works: as it is, Milne blithely asserts that "it is generally agreedthat during the remaining thirty years of his life the Bishop of Hippo regularly used therevision of the Latin Gospels made by S. Jerome" (1926:ix). Given the opposition to Burkitt'shypothesis, it would have been worthwhile demonstrating this: De Bruyne laments Milne'searly cut-off date (1931:595).13 See the table in Milne 1926:xii.14 Collectanea biblica latina 5: 1921. This is accepted by Frede 1972:459, who traces thehypothesis further back still to L. Ziegler, Die lateinischen Bibelbersetzungen vorHieronymus und die Itala des Augustinus; Munich 1879.15 De Bruyne's position is no longer supported by textual scholars: Frede describes it as"verblat und widerlegt" (1972:466; see also the articles cited in Fischer 1972:20).Nonetheless, it initially found acceptance and is occasionally revived: for a general appraisalof the thesis and an investigation of its applicability to John, see Sections 2.3.4 and 5.3.7respectively.

7

An edition of selected Vetus Latina manuscripts of the Gospels produced by Adolf Jlicher

began to appear a few years later. He divided the traditions into "Italian" and "African"

strands, offering a reconstruction of the "Italian" text found in the majority of witnesses,

while the "African" side was represented by two manuscripts: Codex Bobiensis and Codex

Palatinus. The latter, however, like many "African" sources, was contaminated with readings

from European traditions at an early stage. Furthermore, Jlicher's reconstruction of the

"Italian" text, for which he never fully articulated his editorial principles, has also been

criticised.16 He died in the same year that the first volume appeared: the rest of the series

was seen through the press by Walter Matzkow and Kurt Aland, who later produced revised

editions of the Synoptic Gospels. Commonly referred to as Matzkow-Jlicher-Aland, these

remain the principal critical editions for multiple Old Latin gospel manuscripts.

In the late nineteenth century, Joseph Denk began a collection of Old Latin patristic citations

with the intention of creating a "new Sabatier". From 1920, these were housed at the abbey

of Beuron. Here the Institut Vetus Latina was created to oversee the publication of new

critical editions of the Old Latin versions of each biblical book using all available manuscripts

16 The differences between Italian and African translations were initially thought to bedialectal, but there is no evidence to support this: they rather represent different strands ofa tradition of Latin translations (see Fischer 1972:9). Monceaux reminds us that, despite thepaucity of surviving witnesses, there was a multiplicity of African versions, of which Cyprian'stext is the best known (1901:168). Most scholars now believe that the Bible was firsttranslated into Latin in North Africa (see the summary at Elliott 1992:200-202) although itis possible that parallel developments happened independently in different places: for anaccount of technical Latin terms appearing in Christian documents from Rome in the secondcentury, see Mohrmann 1965:67-126. At any rate, convergence between different traditionssoon set in: Fischer (1972:34) lists contamination in African manuscripts, while Monceaux(1901:142) suggests that Jerome's Gospel revision introduced African readings into Italianmanuscripts and Frede (1972:464) observes that "African" renderings can also be found inEuropean Fathers through their use of earlier African works. Aland articulated the confusionsurrounding Jlicher's principles in his introduction to the 1954 volume of Luke: for critiquesof these editions, see Fischer 1972:35, Birdsall 1992:120 and Burton 2000:9.

8

and patristic material up to the ninth century. The first volume to be published was,

appropriately, Genesis, edited by Bonifatius Fischer between 1951 and 1954. Many have

followed, and most remaining books are now in preparation.17 Denk's card index of citations

has been maintained and enlarged, and was recently made available electronically as the

Vetus Latina Database (VLD Online), with each card reproduced as a digital image.18 The

Institut also keeps the register of Old Latin manuscripts and produces an index to the most

recent critical editions of Church Fathers. The editions are supplemented by the monographs

on related topics in the series Aus der Geschichte der lateinischen Bibel. These include

Fischer's ground-breaking computer collations of test passages from more than four hundred

and fifty gospel manuscripts from the first millennium.19 Scholars associated with the Institut

continue to publish numerous important articles on the Latin Bible, and the annual

Arbeitsbericht details the current state of the edition.20

17 A fuller account of Denk's work is given in Frede 1972:477. There are several summariesof the history and work of the Institut Vetus Latina: the most recent is Petzer 1995:113-114. For a description of the Vetus Latina editions, see, amongst others, Metzger1977:320-321. The Gospel according to John is currently being undertaken by Dr PhilipBurton, while Dr Ulrich Schmid has taken on the task of editing the Synoptic Gospels (InstitutVetus Latina 2004:42).18 Denk's original index contained around 400,000 index cards: this has since doubled in size(Institut Vetus Latina 2004:9,45). For further information on the online edition, published byBrepols, see Institut Vetus Latina (2002:44) and http://www.brepolis.net. The patristicmaterial is the key to the Vetus Latina editions: see Elliott 1992:215 and North 1995:214.Frede notes that many Old Latin readings may never have appeared in manuscripts: "certes,bien des formes du texte vieux latin de la Bible, ainsi entendu au sens large, n'ont jamaisfigur dans un manuscrit biblique latin ... elles n'en ont pas moins exerc une influence sur ledveloppement du texte latin de la Bible, d'une manire indirecte, raison de l'influence del'auteur qui a coul sa citations dans une telle forme" (1995:26).19 Die lateinischen Evangelien bis zum 10. Jahrhundert. 1. Varianten zu Matthus (1988); 2.Varianten zu Markus (1989); 3. Varianten zu Lukas (1990); 4. Varianten zu Johannes(1991).20 For good summaries of the scholarly consensus on Latin biblical manuscripts and LatinChurch Fathers as established by the Institut, see Fischer 1972 and Frede 1972 respectively.The Arbeitsbericht has a list of all published editions and the volumes of Aus der Geschichteder lateinischen Bibel (see Institut Vetus Latina 2004:62ff.).

9

The Biblia Augustiniana was an independent project based on a collection of Augustine's

scriptural citations, edited by Anne-Marie La Bonnardire at the Institute des tudes

Augustiniennes in Paris. Seven volumes were produced, comprising Kings and Chronicles

(1960), The Minor Prophets (1963), 1 and 2 Thessalonians, Titus and Philemon (1964),

Deuteronomy (1967), Wisdom (1970), Jeremiah (1972) and Proverbs (1975). The format

consists of an individual verse, followed by a list of references to the works in which it is

cited by Augustine. These are arranged in chronological order, each with an orchestration

scripturaire, a list of the other biblical passages found in the vicinity of the citation. The

primary concern of this project was to assist commentators on Augustine, rather than

textual critics.21 La Bonnardire realised that Augustine's use of particular groups of

scriptural verses reflected his current theological concerns. This "orchestration scripturaire"

could therefore be used to assist in the difficult task of dating Augustine's works, especially

21 Although allusions are indicated, all references are listed under one form of words for eachverse, so the Biblia Augustiniana cannot be used for text-critical purposes (see also North1995:216). According to La Bonnardire 1965:15, the initial aim of the project was tocollect all the Psalm quotations of Augustine, a volume which never appeared: thepublications ceased due to lack of funding. An update on the project is reproduced in theChronique of the Revue des tudes Augustiniennes 9 (1963) 404, in which La Bonnardirestates that 32 of the 61 biblical books have been completed, along with 130 Psalms and halfthe work on the Gospels; she also characterises the project as predominantly patristic: "Vul'tat prsent des ditions critiques de l'uvre augustinienne, la Biblia Augustiniana ne peutprtendre donner, pour toute citation biblique de s. Augustin, le texte dfinitivement sr.Elle vise essentiellement fournir aux chercheurs une liste des rfrences bibliques et lanotation des variantes rencontres pour un mme verset. Le but de la Biblia Augustiniana estessentiellement d'ordre patristique ... la Biblia Augustiniana essaie de se prsenter commeune histoire de la rflexion de s. Augustin sur la Bible." Despite off-handed references to"toutes les citations de versets johanniques" (e.g. La Bonnardire 1965:75) there is norecord of an unpublished systematic file for the remaining books of the Bible (see Bogaert1998:33).

10

shorter pieces such as sermons.22 The recent work of Pierre-Marie Hombert is based on the

same principles, in which he proposes a re-dating of a group of works based on noeuds

scripturaires.23 La Bonnardire also edited the third volume in the series Bible de tous les

temps, Augustin et la Bible (1986). This is concerned primarily with questions of exegesis, a

preoccupation amplified in an American version produced in 1999.24

The use of patristic citations in reconstructing the text of the Greek New Testament has

been a growing preoccupation of textual critics in America over the last fifty years. The work

of Gordon Fee stands out for his theoretical and practical contributions to the discipline. One

principle on which which Fee insists is the need for a complete presentation of all the

22 She describes the orchestration scripturaire as "deux ou trois versets, parfois davantage,pris comme appuis d'un enseignement doctrinal, constituent un schme dont Augustin faitusage un certain nombre de fois, dans les cas qui prsentent entre eux une analogie relle"(La Bonnardire 1965:16). Augustine would work on several writings simultaneously, so "ilest donc normal qu'une preoccupation grave d'un moment donn apparaisse dans plusieursuvres diffrentes, qui ont entre elles le lien de la mme date" (La Bonnardire 1965:180).For example, Augustine uses one group of citations in anti-Donatist works which are notfound at all in writings against the Pelagians (La Bonnardire 1965:16). Frede 1971:93observes that the orchestration scripturaire can also be used as an indication of authenticity.23 The majority of Hombert's conclusions are adopted in Gryson 2004 and are mentioned inChapter Three below. Hombert's work is described by Dolbeau as "l'un des plus importants[ouvrages] qui aient t rcemment publis sur Augustin", although he observes that thismethod of internal criticism cannot always support the precise dating which Hombert claims(Dolbeau 2001:9-10). Hombert has collected his citations from three sources: the CorpusLitterarum Christianorum Latinum (CLCLT 3), the Augustinus Lexikon and the ThesaurusAugustinianum. This strongly suggests that La Bonnardire's unpublished files are no longerextant. Hombert also notes the difficulty of using current electronic search facilities tolocate unexpected biblical texts and allusions (2000:vii).24 Augustine and the Bible ed. Pamela Bright (1999): among the sixteen essays, she onlyincludes seven of the twenty-one in the original volume.

11

available citational evidence before the value of a particular author can be assessed.25 Fee

subsequently acted as founding editor for the Society of Biblical Literature's series The New

Testament in the Greek Fathers. Several volumes have appeared and set a benchmark with

their comprehensiveness, transparency, use of critical editions, and inclusion of variations in

the manuscript tradition.26 These combine the presentation of the data, often with a

reconstructed text, and a quantitative comparison of similarity with representative surviving

New Testament manuscripts. Although this method offers a general picture of textual

relationships, it does not always seem to be best suited to the random and fragmentary

nature of citations.27 Each survey also includes "a minimal evaluation of the Father's

citations, as to his citing habits, the reliability of his data, and the degree of certainty with

which one may use the data" (Fee as editor in Ehrman 1986:xi-xii). This places the textual

evidence into the wider context necessary for determining its significance for the history of

25 Fee 1993:304; he claims that most studies of citations in Greek Fathers prior to 1970 are"quite inaccurate and misleading" (1995:197). Initially, Fee supported "criticalreconstructions", arguing that "ambiguity will often be eliminated when a Father's biblicaltext has been critically reconstructed" (1993:303). However, he recognised the danger oftrying to reconstruct a single biblical text from works which draw on different archetypes(1993:305), and eventually admitted that "where a Father (a) cites freely, and/or (b) citesinfrequently, and/or (c) cites texts in two or more forms, the safest procedure is to list allthe various forms in which a text is cited" (1995:199). This is clearly the case withAugustine.26 For a description of the series, see Fee 1995:198-199. Authors so far covered areDidymus the Blind (Gospels: Ehrman 1986), Gregory of Nyssa (NT: Brooks 1991), Origen(John's Gospel: Ehrman, Fee & Holmes 1992; 1 Corinthians: Hannah 1997), Cyril ofJerusalem (NT: Mullen 1997), Epiphanius of Salamis (Paul: Osburn 2004) and Basil ofCaesarea (Matthew's Gospel: Racine 2004).For a list of other collections of patristic citations undertaken in the second half of thetwentieth century, see Epp 1993:30; this can be supplemented for Latin texts by Fee1993:302 (which requires some correction: Coleman's work on Lucifer of Cagliari does notseem to have included the Gospels) and North 1995:216, which includes newer works andsupplies some useful critical comments.27 A discussion of the problems of this approach is offered by Ehrman, who notes that"errors of reconstruction and systematic caution ... will tend to 'even out' the differencesamong textual witnesses" (1986:256). Fee describes some earlier attempts to establishpercentage relationships between citations and manuscripts as "almost valueless"(1993:306). See also Geer's description of Colwell's quantitative method and its subsequentrefinement (Geer 1995:255-257).

12

the New Testament text. New articles and studies continue to appear, using and refining

these principles.28

Mention of recent work should also include the Biblia Patristica of the Centre d'Analyse et de

Documentation Patristique (CADP) in Strasbourg. This was one of the first projects to use

computers to create a collection of scriptural citations in both Greek and Latin. Seven

volumes and a supplement have so far appeared, although Augustine has not yet been

reached.29 In the first volume, the editors insisted that their database was compiled from a

completely new reading of each text, although they have collaborated with similar

undertakings including the Institut Vetus Latina.30 The Corpus Christianorum series of

editions of the Latin Church Fathers was to have included a separate scriptural index to the

28 In particular, Osburn (2005) takes up Fee's challenge to provide criteria to distinguishbetween the text-critical value of different types of patristic citation, while Ernest (2004)applies some of these ideas to Athanasius' biblical citations.29 The volumes are as follows: 1. Des origines Clment d'Alexandrie et Tertullien (1975), 2.Le troisime sicle, Origne except (1977), 3. Origne (1980), 4. Eusbe de Csare,Cyrille de Jrusalem, Epiphane de Salamine (1987), 5. Basile de Csare, Grgoire deNazianze, Grgoire de Nysse, Amphiloque d'Iconium (1991), 6. Hilaire de Poitiers, Ambroisede Milan, Ambrosiaster (1995), 7. Didyme d'Alexandrie (2000) and a supplement on Philo ofAlexandria (1982).30 See Biblia Patristica 1975:1-2. The database is published in print form as an index,although the description of the main database as "des fiches micrographiques qui ...permettent ... une reproduction aise de la citation et de son contexte" (1975:2) impliesthat the full text of citations is available in Strasbourg. The claim to have revisited theoriginal texts is only explicitly made in the first volume, although the principles of thecollection were laid out in a document published by the CADP in 1967 entitled "Etapes,moyens et mthode d'analyse". Each volume lists the editions consulted during itspreparation: the printed indexes will become dated as new editions appear. Other drawbacksof these indexes have been noted by textual critics: see Fee 1995:196, Mullen 1997:61 andEhrman, Fee & Holmes 1992:31.

13

whole of Augustine's works, Volume 69, to which several editors refer. However, work on

this project came to an end by the early 1980s and there is no likelihood of its publication.31

With the publication of Philip Burton's The Old Latin Gospels (2000), a reassessment of the

text and the language of the Old Latin translations, and current work towards the Vetus

Latina edition of the Gospel according to John, the need arises for a study of Augustine's

text of the Gospel conforming to critical standards comparable to those of the New

Testament in the Greek Fathers. Although its format is different, it will, I hope, also be of use

for scholars in the tradition of the Biblia Augustiniana, continuing to explore the importance

of Augustine's biblical text for matters of theology, chronology, influence and exegesis.

1.3 Methodology and principles of this collection

It should be stated at the outset that this work is not an attempt to reconstruct a single

edition of St John's Gospel which belonged to Augustine. For a start, it would be unusual if

such a prolific author's citations were sufficiently consistent as to indicate one unambiguous

31 The most recent reference to this volume appears in Mutzenbecher's 1980 edition ofQuaestionum euangeliorum (CC 44B). Luc Jocqu of Brepols informs me that the volume wasinitially advertised as a collaboration between Eligius Dekkers and Prof. J.H. Baxter of St.Andrew's University, and that the publishers have no material surviving from this work(private communication, 3.2.2003). The Italian Augustine website advertises a scripturalindex "Bibbia agostiniana: la Sacra Scrittura come letta da Agostino" (onhttp://www.augustinus.it/sussidi/index.htm), but this is still in preparation. The "Indiciscritturistici" on the same pages only has the scriptural index for De ciuitate dei, as of May2006. In 1913, De Bruyne stated that "D'Amrique et d'Allemagne on annonce des ditionsde la Bible de S. Augustin" (1913:314), but no other traces of these have come to light.

14

source, particularly in the context of a highly diverse Latin tradition.32 Secondly, the accuracy

of a Father's habits of citation and the transmission of the biblical text in his works need to

be assessed before it is appropriate to make editorial decisions about the "original" text.33

Augustine's citational practice is considered in Chapter Two, while Chapters Three and Four

analyse the textual affiliation of his citations. It is only in Chapter Five that it becomes

possible to evaluate Augustine's contributions to the history of the text of John and

distinguish different traditions within his work. The citations have therefore been recorded in

the Appendix with minimal editorial intervention in order to present all the information

preserved for the text of this Gospel in the works of Augustine. This means that each

citation can be assessed in terms of its relation to the whole, and allows for the possibility of

authorial inaccuracy in certain works or corruption in part of the manuscript tradition.34

For the purpose of this study, I have used only those writings of Augustine which are listed

as authentic in the Vetus Latina index (Frede 1995). I have also adopted the same system of

abbreviations for each work. Thus AUdo indicates Augustine's De doctrina christiana (not to

be confused with AUDo which refers to his Epistula ad catholicos de secta Donatistarum, or

32 Fee observes that "It is perhaps presumptuous to assume that any Father, writing over athirty- to forty-year period, had only one Bible; and perhaps it is folly even to assume he onlyhad one Bible at any given time" (Fee 1995:193; Fee 1993:345). Decret says that the BibliaAugustiniana "montrent assez qu'il serait vain de prtendre reconstituer toute la Bible avecdes citations scripturaires littrales de l'uvre d'Augustin" (1970:164), which Monceauxrecognised earlier: "Sa Bible latine n'a cess de se transformer, tantt par des changementsbrusques, tantt par une continue et lente volution D'aprs cela, on comprend qu'il soitchimrique d'essayer de ramener l'unit le texte sacr de saint Augustin" (1901:150).33 Compare Robert M. Grant's dictum that "Patristic citations are not citations unless theyhave been adequately analyzed" (in Parvis & Wikgren 1950:124, quoted in Osburn 2005:318and Fee 1993:340).34 These factors also underlie the minimal use of quantitative analysis in this work. Not onlydoes this method rely on a reconstructed text, but it also requires clearly defined text-types,which have not been identified for the Old Latin John. There is also a wider question as towhether quantitative analysis can meaningfully be applied to a translated tradition: see Frede1972:463.

15

AUDon which is the abbreviation for his Contra partem Donati post gesta). A full list of these

abbreviations and the editions used will be found before the complete list of citations in the

Appendix.35 Although pseudonymous works have been excluded, I have included scriptural

citations from other authors contained within the corpus of Augustine's authentic writings,

for example replies from his correspondents or sections of polemical works in which

opponents are quoted at length. These are grouped at the end of Augustine's citations for

each verse and the source is indicated in square brackets.

In assembling the citations, my normal practice was to begin with the scriptural index

provided in each modern edition. As an indication of accuracy, I also regularly checked a

selection of Johannine citations in the text to ensure that they were listed in the index.36

Needless to say, this process revealed some inaccuracies. Typographical errors such as

transposed or incorrect numerals in the chapter and verse references were usually

remediable, while incorrect paragraph and page references could only sometimes be

corrected. Citations from the Gospel according to John can occasionally be difficult to

identify, given the Evangelist's use of common vocabulary which he invests with additional

35 This list also gives an indication of the numbering system for each citation. For example,under the Confessions, "book.chap.line" indicates that the first number refers to the book,the second to the chapter (or paragraph division) and the third to the line number in theedition: AUcf 5.3.24 therefore indicates a citation in book five of the Confessions, chapterthree, line twenty-four. The significance of including the line number is to distinguishbetween multiple citations of the same verse within the same paragraph. This is not possiblein certain editions: for CSEL works, the line numbers start afresh at the beginning of eachpage, so each citation is given with the page number and line number, which is often thereference used in the scriptural index for the edition but would not identify the citations inanother edition. In many editions, paragraph and chapter or section numbers overlap:reference to the edition cited should clear up any ambiguity.36 Reading all Augustine's writings afresh would have made the project unfeasibly long, withno automatic improvement in the quality of the collection, although this was necessary forcertain works listed below. Although the Vetus Latina card index is regularly updated, thisneeds to be checked in the preparation of each edition; furthermore, I have been able toinclude introductory material and more information on the manuscript tradition.

16

meaning through repetition: sometimes editors have identified with one verse a citation

which more closely resembles another verse of the Gospel or even the Johannine Epistles.37

One common practice is to list only the first citation in each paragraph and ignore

subsequent repetitions (even though these may still be italicised). This is understandable but

inadequate for textual purposes, so when checking references from an index I also read

through the rest of the paragraph and, if this revealed other citations, neighbouring

paragraphs as well until the next reference was reached. Nonetheless, I must record that the

scriptural indexes of St John's Gospel in editions of Augustine are for the most part highly

accurate: a tribute to the editors who cannot have imagined that they would be used for

such a purpose.38

In the case of works for which a scriptural index is not provided I had to compile my own

index of Johannine citations by reading through the entire work: a time-consuming task,

37 This is even the case in Migne's edition of the Tractatus in Epistulam Iohannis ad Parthos,e.g. AU1Jo 7.2 and 9.10. On this Johannine technique, see Burton 2000:92.38 Some indexes are more useful than others. In a class of its own, however, is the minutely-detailed index of De trinitate by Mountain and Glorie (CC50/50a) which, in its attempt to beexhaustive, is almost unusable because of the compilers' over-sensitivity to biblical allusions.For example, every occurrence of the phrase "uerbum dei" is referred to John 1:1, while theword "unigenitus" or phrase "de patre" is linked to John 1:14. In Verheijen's edition of theConfessiones (CC27), there are similar examples of ueritas and lux (used as synonyms forChrist) being referred to John 14:6 (AUcf 8.1.34) and John 1:9 (AUcf 11.23.44)respectively.It is worth recording that some editors even indicate whether citations follow a Vulgate ornon-Vulgate text. For example, Mountain and Glorie note at AUtri 5.11.13 that quoniam deusspiritus est appears in some Old Latin codices at John 3:6 as well as John 4:24. However,their indication "John 14:10 (uet.lat.)" for the phrase inseparabilis operatio does notcorrespond to any reading in Jlicher's edition of the Gospel.For the Sermones, I used the scriptural index from a complete edition in the SpanishBiblioteca de Autores Cristianos, referring to Migne for the text of the first 396 sermons andthe web-based edition at http://www.augustinus.it/latino/discorsi/index2.htm for thesupplementary sermons indicated by letters. The scriptural index is in Obras completas deSan Augustin 26: Sermones 6 ed. Pio de Luis, BAC 461 (1985). I have also includedscriptural information from the lectionary indications recorded in Migne although not fromthe individual headings for each paragraph: examples of variation between the lemma and thebody text of the sermon may be seen at AUs 129 (John 5:39), AUs 132 (John 6:56).

17

which also had a greater potential for omission.39 In most cases I was later able to check my

lists against the footnotes listing scriptural citations in the internet-based edition of

Augustine's works, at http://www.augustinus.it/latino/index.htm, a very important resource

which also allows electronic search facilities.40 I also read the whole of Augustine's sermon-

commentary on John, the Tractatus CXXIV in Iohannis euangelium. Every citation of John in

this work would be too numerous to list, so I have given the initial form of each verse in each

commentary sermon as the "continuous text" (marked by [CT]), and ignored identical

repetitions.41

The original editorial identification of the source of the citations has not always been kept:

not only have corrections been made to misprints, but I have, on occasion, reallocated

certain citations. When editors have indicated similiarity with more than one verse, I have

assigned each citation to a single verse of John but noted possible parallels at the beginning

39 These works are AUdia, AUJul im 1-3, and the works still available only in Migne AU1Jo,AUgr, AUJud, AUJul, AUJul im 4-6, AUMax, AUMax co, AUmus, AUpers, AUprae. There is alsoa problem with the numbering system for AUMax co in Migne: there are forty paragraphs,numbered 1-14 and then 1-26, so the column number must also be used to avoid ambiguity.(The augustinus.it webtext marks the second set 15.1-15.26.)40 This website has the complete Latin text of Augustine's works from Migne's PatrologiaLatina. It appears that the works were initially scanned by optical character recognition,hence mistakes such as uerbum caro iactum est at AUs 23A.3 on the website. However,silent corrections have been made to the text, over the last three years, such as thecitations of John 16:20-21 at AUs 210.5.7, where tristes has been replaced by tristi and illiby illa. The site is run by the Nuova Biblioteca Agostiniana, whose Italian translations areavailable elsewhere on the site. The site is also useful for its summary tables of dates forAugustine's sermons, and I have already mentioned the section on the scriptural text whichhas yet to appear.41 I have, however, included every variant reading and all gospel citations outside thecommentary sequence. See further Section 3.2 below.

18

of each entry to enable cross-reference.42 It should also be observed that Augustine

sometimes makes "composite citations", combining different biblical verses: within John,

these have been separated into their constituent parts. Some editors have marked citations

with a verbal correspondence to the text as allusions (and vice versa): the final classification

is mine, based on the definitions set out in Section 1.5.1 below, although this is only

intended as a rough guide in order to give a context for certain changes or inaccuracies. I

have, for the most part, kept the spelling conventions from each edition (such as assimilation

of compound verbs), but not marked such variants as alterations to the biblical text. Most

punctuation and capitalisation has been removed.43 The final collection, listed in full in the

Appendix, enables an independent comparison to be made with the Vetus Latina Database.44

42 For example, John 13:10 (et uos mundi estis sed non omnes) is linked to John 15:3 (iamuos mundi estis); John 6:57 (in me manet et ego in illo) is similar to John 15:5 (qui manet inme et ego in eo); John 1:29 has textual parallels with all three Synoptic Gospels, whoseinfluence can be seen in Augustine's narration of the baptism of Jesus. The practice of thevolumes of the New Testament in the Greek Fathers is to have an appendix with citationswhose source is ambiguous, although the same citation is occasionally printed twice underdifferent verses. The Biblia Patristica series used a marking 'E' to refer to a gospel parallelwhose referent could not be determined, allocating all such citations to Matthew (1977:10).43 This was initially a device to aid with computer collation of the text, but as punctuationwould not have been recorded in Augustine's original manuscripts, its absence is justifiable.The New Testament in the Greek Fathers series has minimal punctuation and omits accents.44 When my collection was complete, I compared the results with the Vetus Latina Databasefor selected chapters (John 7, 16 and 20). The accuracy of each collection is similar, givenslight differences in criteria: the Vetus Latina Database includes all identical repetitions (andtitles) in the Tractatus in Iohannem and permits multiple identifications of the same citation.In the 111 verses compared, my collection totalled 801, as opposed to 1100 in the VetusLatina Database; 302 of these are repetitions in the Tractatus. There were 11 verbatimcitations and 55 other references in my collection not found in the Vetus Latina Database,and 12 verbatim citations and 42 other references in this database not identified in my initialsurvey. These additional entries have subsequently been incorporated into the Appendix.Overall, this demonstrates that the Vetus Latina Database is comprehensive and rendersindependent work largely superfluous, although there are, as might be expected, a handful ofminor errors. For example, only two citations are listed for John 7:3 and Augustine is ignored;some pseudonymous material is still included under Augustine (e.g. AUs Mor 9 in John 16:8-10) and there are a number of duplicate cards. It is hardly necessary to say that inaccuraciesand oversights are bound also to occur in my own list in the Appendix.

19

1.4 Presentation of citations

There are a number of different ways in which the citations may be presented, each with its

own advantages and disadvantages. I have chosen to set out the citations verse by verse,

which allows for easy reference and textual comparison but unfortunately splits up citations

of more than one verse. I have indicated these longer citations by a reference in square

brackets with details of the preceding verse. Thus for a citation of John 7:35-37, the

citation of verse 36 would begin with [Jn 7:35] and the citation of verse 37 with [Jn 7:36].

In the case of a citation of this passage in which Augustine had omitted verse 36, running

the outside verses together, the citation of verse 37 would begin [Jn 7:35]. Other scriptural

references are also indicated in this way. Cases in which there are two citations of the same

verse in the same paragraph, once by itself and once as part of a longer citation, may lead to

difficulty in identifying the antecedent of the longer citation. This can often be resolved by

comparing line numbers, although reference can always be made to the original edition.

I have arranged the citations for each verse according to the alphabetical order of

Augustine's works. Although a case could be made for a chronological sequence, which might

display a change over time, there is no universally-agreed scheme for dating Augustine's

works, and dates are frequently reassessed.45 This sequence would also involve the separate

placing of each sermon (several of which are not assigned dates) and the division of works

composed over a period of time, which would raise more problems in locating a particular

citation than it is likely to provide illumination of any textual development. Chronological

45 For the purposes of this study, I follow the dating of the Institut Vetus Latina publicationsunless specified otherwise (Frede 1995, Gryson 1999a, Gryson 2004). It is worth notingthat Gryson 2004 has a number of changes in the dating of Augustine's works, incorporatingsome of the conclusions of Hombert 2000. Not even Milne presents the citations accordingto his chronological order. On the dating of Augustine's works, see La Bonnardire (1965)and Hombert (2000).

20

order, however, features in the analysis of Augustine's text of John in Chapter Three, in

order to complement the listing of citations.

At the head of each verse is given its text from the current critical edition of the Vulgate,

the Stuttgart Vulgate (Weber-Gryson 1994). Where the citations differ from this text

(including omissions), this is marked in bold type, partly for ease of recognising differences

and partly as a double-check when transcribing in order to ensure careful comparison

between both texts.46 I considered using the continuous text from Augustine's Tractatus in

Iohannis euangelium as a base text. Although this would have the merit of presenting solely

Augustinian material and not giving undue weight to the Vulgate tradition, there are too

many arguments against this: different tractatus follow different biblical texts, there are

some groups of verses and numerous half-lines which do not feature at all in the tractatus,

and the only modern edition falls short of critical standards in many ways. The Old Latin

manuscripts are too divergent to provide an obvious base text.

1.4.1 Additional material: manuscript readings

One of the principal concerns with the use of patristic evidence is whether a Father's biblical

text has been faithfully transmitted. It was observed by F.J.A. Hort that manuscripts of

46 Milne 1926 uses a complicated system of bold and/or italic type to indicate agreementwith different text-types. See also Fee's suggestion of a typographic system for indicatingconfidence in the text of citations (1993:351-354). Note that the Vulgate has a slightlydifferent numeration from most editions at the end of John 6, splitting John 6:51 into twoseparate verses: this means that the list of citations does not coincide with Matzkow-Jlicher-Aland or Nestle-Aland in these verses.

21

Church Fathers were especially prone to corruption in scriptural citations.47 This potential

problem is especially pertinent in Old Latin studies, given the widespread acceptance of the

Vulgate. Such harmonisation may be either conscious or unconscious, but once introduced it

is unlikely to be corrected. The only way to approach the issue is to include all the evidence

of manuscript variation within Augustine's own textual tradition, to see if any "Vulgatising"

tendencies can be detected. I have therefore included all variants in the gospel citations

which are listed in the critical apparatus of the editions of Augustine's works. The majority of

Augustine manuscripts date from the ninth and tenth centuries, over five hundred years

after his death, so it is often not possible to trace the whole transmission history of a work,

and corruption may have crept in at an early stage.

There is also the possibility that genuine Old Latin variants are present in the manuscript

tradition but do not feature in the editorial text. Older editors, too, have been accused of

harmonising biblical citations to later versions.48 In the Appendix, manuscript variants have

47 Westcott-Hort 1882:202-203. Similar observations are made by Fee (1993:335), Burton(2000:4) and Frede, who observes "Nicht selten wurden die Bibelzitate in den patristischenHss von Lesern und Korrektoren nach der ihnen gelufigen Fassung abgendert" (1972:469)and offers examples on the following page. The extent of the problem is difficult to assess:Ehrman's claim that corrections are present in "virtually all the church Fathers" (1986:6)contrasts with Mullen's more sober account of "Byzantinizing modifications" in only onemanuscript of Cyril of Jerusalem (1997:338), Suggs' observation that "there is littleevidence of systematic revision of New Testament citations except in translated works"(1958:140) and Monceaux's statement that "ces substitutions sont relativement rares dansles ouvrages des premiers Pres, surtout quand il s'agit de citations courtes." (1901:100).48 Erroneous readings in the editorial text of citations in Augustine's works edited by Zycha,Cyprian by Hartel and Pelagius by Souter are noted by Frede 1972:470. On Zycha's text, seealso the comments of Solignac, who emphasises "la ncessit de respecter la traditionmanuscrite quand il s'agit d'un texte biblique. Sans doute mme vaut-il mieux dans ce caspcher par excs plutt que par dfaut" (1987:552). Fahey (1971:23) lists the errors inHartel's Cyprian and provides a completely new scriptural index. On the Greek side, Ehrmanclaims that Migne's Patrologia Graeca is "of practically no value for establishing the originalwording of the NT" (1986:6), while in contrast Suggs reckons that the overall picture ofvariants in Migne offers "a fairly accurate picture" (1958:141). This corresponds to myexperience of using the Patrologia Latina for Augustine's sermons, which feature a number ofnon-Vulgate readings.

22

been indicated following the citation, identified by the editorial sigla from each edition.49

There are comparatively few significant variants for Augustine's text of John in the

manuscript tradition of his works: those of interest, I have considered in the discussion of

individual works in Chapter Three.

1.4.2 Additional material: introductions

When transcribing citations, it soon became apparent that the context of each citation,

especially its introduction, was of importance in assessing its textual significance and forming

a picture of Augustine's citing habits.50 Some verses might be explicitly introduced as gospel

citations, while others might be more loosely indicated. On occasion, the citation is prompted

by the occurrence of a particular word in Augustine's argument or another citation, which is

then glossed by the words from John: key words from the citation may also be anticipated in

Augustine's introduction or recur afterwards. I have therefore included in italics such extra

material as seemed relevant from before the citation and also any interjections in the biblical

text, for example the addition of dixit after direct speech or quid hoc? preceding the

evangelist's explanation, using the sign "..." to indicate words omitted.51

49 I have also included some manuscript readings from other editions. These are:1) AUs: in certain sermons, following the edition of Poque 1966. (If the editorial text differsin this edition from Migne, I have preserved both in my list, as at John 13:37.)2) AUdo: the CSEL edition (Green, CSEL 80: 1963) includes a number of manuscripts notrecorded in Martin (CC 32: 1962): I have added any variants in brackets after the CCapparatus: there are no differences in the editorial text of the citations of John between thetwo editions. It should be noted that some editors include earlier editions or the Vulgateamong their textual witnesses, marking Vulgate forms with a 'v' in the critical apparatus.50 Compare Prigent: "le contexte patristique est capital pour l'apprciation de la ralit dutexte allgu" (1972:441).51 Brooks 1991 uses "..." to indicate both omission of non-scriptural text and identity ofcitation with his base text, which is confusing. I have recorded all scriptural text in full.

23

Other scriptural citations in the neighbourhood of each example are represented by a

reference in square brackets, as already seen in the method of indicating citations which

span a number of verses. For example, Augustine follows a citation of John 11:44 with

Matthew 18:18 in Sermo 139A.2:

Christus dicit: Soluite eum, et sinite illum. Soluite, soluite: quae solueritis in terra,soluta erunt in caelo. (AUs 139A.2)

In the listing in the Appendix, the text of non-Johannine citations is not normally included,

but only the reference. Intervening words are marked in italics. Thus the entry for this

citation reads:

christus dicit soluite eum et sinite illum. soluite soluite [Matt 18:18] AUs 139A.2

The inclusion of this information follows the principle of the orchestration scripturaire in the

Biblia Augustiniana and is useful for purposes of exegesis as well as this method of dating. I

have not been able to find any explanation of the criteria for the orchestration scripturaire as

featured in the Biblia Augustiniana, so I have only included references directly adjoining the

citation, and have made no attempt to be exhaustive. In commentaries such as De sermone

domini and the Enarrationes in Psalmos, I have also endeavoured to include the reference to

the text under consideration.

1.5 Identification of citations and questions of accuracy

A number of ways of identifying citations have been proposed, together with different

categories for classification. A scriptural citation may be defined a s the deliberate

employment of a sequence of words corresponding to one found in the

Bible.52 As all Christian discourse, and particularly the Latin of the Church Fathers, is

52 I will use the term "citation" to indicate biblical references in this broadest sensethroughout my study, unless I specifically indicate that the more narrow definition of"citation" in Section 1.5.1 is intended.

24

developed from and heavily influenced by the biblical register, it is necessary to demonstrate

that reference to Scripture is intended rather than coincidental. For example, to categorise

every occurrence of agnus dei as a citation of John 1:29 is excessive: this image is

widespread elsewhere in the New Testament and in Christian liturgy. Length is therefore also

a factor in identifying citations: unless there is compelling evidence that a citation is

intended, I have ignored references of only two or three words.53 It should be possible to

show pragmatically that the reference is intentional, either by means of an introductory

formula or some other functional signification, such as reference to the scriptural context of

the verse. Citation formulae can vary from the general scriptum est ("it is written") or

dominus dixit ("the Lord said") to explicit indications of book and passage. These are clear

indications that a Church Father is making a deliberate reference to Scripture, although they

do not in themselves guarantee the accuracy of the biblical text which follows.54

An alternative and complementary approach is to identify citations on the basis of their

correspondence to known forms of the biblical text. This is necessary for citations which are

not introduced as such, but a degree of latitude must be allowed: if only those texts are

admitted which are identical to surviving biblical manuscripts, then no new variant readings

53 A minimum length of "at least three words" is approved by Osburn 2005:319. In theAppendix, I have sometimes marked these shorter phrases as allusions. An exception mayoccur when words or phrases are repeated from a passage already cited in full, for example aword by word commentary on a verse. If the words correspond to the Vulgate text, I haveignored repetitions, but if they feature non-Vulgate terms or phrasing, I have included them(for example, the reading latenter in John 7:10).54 For indications of intentionality in a citation, see the examples quoted in Osburn2005:319-323. Frede constructs a system of classification based on the introductorymaterial: "Fehlt eine Einfhrungsformel die das Zitat kenntlich macht, ist dieWahrscheinlichkeit gering, da er wrtlich zitiert; wird ausdrcklich das Buch namhaftgemacht, dem die Stelle entnommen ist, oder handelt es sich um eine lngere Anfhrung,kann mann gewhnlich damit rechnen da ein wrtliches Zitat vorliegt." (Frede 1972:460).Ehrman, however, is pessimistic, warning that (for Didymus the Blind, at any rate): "citationformulae can just as easily precede paraphrases as citations" (1986:12).

25

or alternative material will be found in the Church Fathers! It is therefore appropriate to

invoke a "corresponding sequence of words" rather than exact repetition: if a Father was

using a biblical exemplar, it may no longer be extant. In an extreme case, there may be no

verbal correspondence with known forms of the biblical text. This is especially likely among

the variety of Old Latin versions, where the same Greek text may be rendered in a number of

ways. For example, in De Baptismo, Augustine cites John 20:23 in the following form:

si cui dimiseritis dimittentur si cui tenueritis tenebuntur. (AUba 5.21.29)

Although this broadly translates the Greek, the form of text is completely different in the

Vulgate:

quorum remiseritis peccata remittuntur eis quorum retinueritis detenta sunt. 55

It is therefore important to be alert to the variety of renderings which may be found in Latin

works.56

It is commonly agreed that citation from memory was the norm in the early Church, and in

the majority of cases it is impossible to prove that an author relied on a codex for his biblical

text.57 Even so, citations made from memory are not without value: the importance of

55 There is limited support for Augustine's reading in the surviving Old Latin tradition. Theclosest parallel is in Codex Palatinus: si cui remiserites peccata remittentur illi si cuiusdetinueritis detenta erunt.56 Frede notes a further complicating factor: as well as deriving from Latin translations of theBible which were already in circulation, there is always the possibility that an author mayhave gone back to a Greek Bible to produce his own form of text (as may be the case withTertullian) or translated a citation found in the writing of a Greek Father on which he wasrelying (1972:462, 467). Neither of these seem likely for Augustine, whose knowledge ofGreek appears to have been limited (see Section 2.2.2 below).57 Compare the preference of Papias for oral rather than written tradition (Parker 1992:280;Gamble 1995:30). Ehrman claims that "with the exception of lengthy citations, quotationswere normally drawn from memory without consulting a biblical manuscript" (1986:5),although the practice is likely to have been different for certain genres of writing. Monceaux1901:109 claims that Tertullian was more likely to cite from memory in apologetic works,while Hoek (1996:224-225) and Mullen (1997:19) give examples of Clement of Alexandriaand Cyprian respectively relying on notes.

26

memorisation in ancient education resulted in a correspondingly greater level of accurate

recall, and the textual form of citations made from memory reflects versions which were in

circulation.58 There remain errors associated with memory lapse, such as the substitution of

words or the conflation of different text forms. Nonetheless, it can be difficult to distinguish

these from other variations in the biblical text or even deliberate alteration. Given that

scriptural quotation is frequently used to support a particular argument, this rhetorical

concern could easily lead to distortion of the original.59

There is no guarantee that Church Fathers intended every citation to be a verbatim

reproduction of Scripture. Indeed, once a phrase has been introduced with a citation formula,

the actual form of text may become less important because the audience has been made

aware that a scriptural reference is intended. There were traditionally six ways in which

quoted material could be varied in antiquity: addition or omission, substitution, changes in

word order, a combination of the preceding, transposition of sections and major rewriting.60

In the first Christian centuries, it was considered a mark of style to incorporate citations

58 Augustine's ability would have been honed by his rhetorical training, as Knauer observes:"seine Ausbildung als Rhetor, durch die er sicher mit mnemnotechnischen Hilfsmittelnvertraut war" (1955:192). Burkitt notes that "there is no doubt that in the kind of quotationwhere it is least likely that a writer would look the passage up in his Codex he used the typeof text current in his younger days." (1896:57; see also Comeau 1930:56). Nonetheless,Gryson warns that "la mmoire des anciens tait plus exerce que la ntre, mais elle ne leurvitait pas de prendre ce genre de libert avec la lettre du texte biblique" (1978:64). Inspite of this, Frede remarks on the similarity observable in some authors between verbatimcitations and those made from memory (1972:460).59 Compare Osburn's examples of different types of adaptation, where he notes that "attimes a word or phrase more in line with the author's intended point replaces the reading inthe text" and that "changes in word order appear to be intended to emphasise a particularword or phrase that is crucial to the writer's use of the text" (2005:331). Alterationsassociated with memory are examined at Section 2.6 below.60 Thus Fee 1993:195. A similar list can be found in Mller 2003:22. Suggs enumerates fourareas in which motivated variants may occur, based on the study of Leon E. Wright: Ethicaland practical, Explanatory, Stylistic and Dogmatic (1958:142).

27

grammatically into the flow of an argument.61 On the other hand, the development of a

canon of Scripture and the growing emphasis on the authority of the text meant that later

authors were often more concerned to preserve the integrity of the original. Furthermore,

increasing familiarity with the Bible in the wider Church meant that authors and preachers

were able to allude to pericopae or scriptural images without needing to quote directly from

a text.

1.5.1 Classification of citations

In order to distinguish between different types of biblical references, a number of schemes of

varying complexity have been proposed for the classification of citations. A balance must be

struck between too few categories, which fail to differentiate the material adequately, and

excessively detailed criteria, which introduce anachronistic distinctions and rely overly on

subjective editorial allocation. The most basic division, used by the Biblia Patristica and many

editorial indices, is between verbatim citation and non-verbatim allusion. For text-critical

purposes, however, at least one intermediate stage is desirable, to indicate citations which

have undergone some form of alteration but are still sufficiently close to biblical witnesses to

be of textual value. This is embodied in the scheme developed by Gordon Fee for the New

Testament in the Greek Fathers, which involves a tripartite division between citations,

adaptations and allusions. Although this seems to be broadly adequate, the reliance on

61 See, for example, the observations of Clement of Alexandria and Rufinus mentioned inOsburn 2005:330. Frede terms this "die Einheitlichkeit Stilprinzip" and notes "Er arbeitetnicht mit dem wrtlichen Zitat, sondern mit Anklang und Anspielung, mit Worten undWendungen, die in den eigenen Sprachstil in freier Weise umgesetzt werden" (1972:457).Gryson too notes in Latin Church Fathers a tendency "qui les porte, suivant les canons de laprose classique, conformer leurs citations au discours dans lequel elles s'insrent"(1978:49). We should also bear in mind Van den Hoek's observation that "in ancientrhetorical traditions, citing by name was not customary or even polite, because the educatedaudience was supposed to know their classics" (1996:229).

28

textual correspondence to define these three categories is also a drawback to their wider

use. Two broader types of reference have therefore been added from patristic studies,

reminiscences and locutions, which gives a five-fold classification.62

Citations normally comprise the majority of scriptural references in any collection and

criteria for further differentiation would be an advantage. This is demonstrated by Osburn's

lengthy treatment of different types of "Accurate Explicit Citations", including no fewer than

nine different sub-groups (2005:319-330). Although some of these are defined by the

presence of introductory formulae or the length of the citation, the key distinction is that of

the certainty with which a Father's text may be treated as representative of a biblical

manuscript known to him. Fee gives four instances of "absolute certainty":

1. When in his subsequent discussion the Father makes a point of the very wordsused by the biblical author.

2. When in a commentary or homily the subsequent discussion confirms the wordingof a citation.

3. When the Father actually cites a known variation to his own text.

4. When in a commentary, homily, or controversial treatise, the Father repeats thetext in the sa