H.A.G. Houghton
Search for other papers by H.A.G. Houghton in
Current site
Google Scholar
C.M. Kreinecker
Search for other papers by C.M. Kreinecker in
Current site
Google Scholar
R.F. MacLachlan
Search for other papers by R.F. MacLachlan in
Current site
Google Scholar
, and
C.J. Smith
Search for other papers by C.J. Smith in
Current site
Google Scholar
Open Access

The Old Latin Tradition of the Pauline Epistles

The evidence for the early Latin text of the Pauline Epistles is relatively sparse. Its history is similar to that of the rest of the Latin New Testament.1 An initial translation was probably made around the beginning of the third century, as witnessed by the consistent form of text in the biblical quotations of Cyprian. This was then revised in various ways in various places, sometimes with reference to a Greek text, sometimes based on internal criteria of Latin style. Although this may have resulted in a number of different traditions, over the course of the fourth century a single form of text associated with North Italy gradually achieved an ascendancy. A revision of this version at the beginning of the fifth century produced the form of text later accepted as standard in the Latin Vulgate.2 It is therefore misleading to divide the Latin tradition of the New Testament into two separate forms, Old Latin and Vulgate. The Vulgate was a revision of an existing Latin text according to a Greek form: the Gospels were the work of Jerome in 382–384, but the reviser of the rest of the New Testament is unknown. There may have been multiple early Latin translations, but the conclusion of editors of both Old and New Testament books in the Vetus Latina editions is that the surviving evidence appears to derive from a single initial version. Although the Latin tradition is best conceived of as a continuum, it is nevertheless a useful shorthand to use Old Latin (or Vetus Latina) as a catch-all designation for non-Vulgate readings, particularly those which are attested in Christian writings before the fifth century.

The reconstruction of the “text-types” of the different stages of the Old Latin tradition, based on scriptural codices and quotations in Christian authors from the first eight centuries, is the goal of the Vetus Latina edition. This is a difficult task. A combination of age and the hegemony of the Vulgate means that few manuscripts survive of the early versions; copies of biblical books made from the fifth century onwards may well be mixed texts combining Old Latin and Vulgate forms. The later form of text may also have affected the transmission of early Christian writings. It is only through the exhaustive collection and analysis of all surviving evidence that the fullest possible picture can be presented. As noted above in the Preface, the material in the present volume was assembled to give an overview of readings in the Latin tradition of the principal Pauline Epistles for the purpose of analysing the biblical text of early commentaries. It is presented here to facilitate further study of the textual history of these writings and to provide a reliable account of the most extensive early Latin evidence, replacing the entries for the selected witnesses in the Vetus Latina Database. In this way, it is hoped that it may also eventually serve as the basis for the full Vetus Latina edition of these four letters, as well as an interim point of reference for Latin sources in editions of the Greek New Testament.

Selection of Witnesses

The selection of manuscripts was made on the basis of the register of Old Latin manuscripts published by the Vetus Latina-Institut and the printed introductory fascicles of the Vetus Latina editions of Romans (Eymann) and 1 Corinthians (Fröhlich).3 The section of the register covering the Pauline Epistles comprises fifteen witnesses, with the sigla VL 75–89. Five of these are more or less complete continuous-text codices (VL 75–78, 88), while others are fragmentary (VL 79–80, 82–83, 85–86). Two are connected with commentaries: VL 81 is a partial text of Hebrews in a copy of Pelagius’ commentary, while VL 89 transmits all fourteen Pauline Epistles in full as the lemmata in an anonymous fourth-century commentary (see below). VL 84 consists of the biblical text from a list of lections in the opening pages of a Vulgate manuscript, while VL 87 is the remains of a Pauline lectionary. In addition, Eymann and Fröhlich include thirteen further manuscripts from other sections of the Vetus Latina register as witnesses to the Pauline Epistles. These comprise six complete or fragmentary New Testaments (VL 51, 54, 58, 61, 67, 135), five lectionaries (VL 31, 32, 251, 262, 271), leaves from an important codex of the Catholic and Pauline Epistles (VL 64) and a short liturgical text (VL 411). Other manuscripts in the register also contain the Pauline Epistles, such as VL 62 and 65 or the Spanish lectionary tradition represented by VL 56 and 68–73, but these are not included in the lists of Old Latin witnesses by Fröhlich and Eymann as they are considered Vulgate witnesses: they have consequently been left out of this collation. It should also be noted that the inclusion of a witness in the Vetus Latina register is not a guarantee of Old Latin affiliation in all the writings it contains or even within a single text. Of the manuscripts included in the present volume, a number are predominantly Vulgate in the Pauline Epistles, including VL 51, 54, 58, 67, 78, 84, 87, 88 and 251. We have provided these for the sake of completeness, not least as all are cited for the Pauline Epistles in the current Greek New Testament hand editions of Nestle-Aland and the United Bible Societies.4

The manuscripts are complemented by early commentary texts and testimonia. Unlike the majority of biblical quotations in Christian authors, both these types of writing derive from scriptural codices: testimonia are collections of passages transcribed from biblical books, sometimes preserving the sequence of the original, while the lemmata preceding the portions of exegesis in a commentary may, unless there is evidence of later interference, be taken to represent the text used by the expositor. There are no fewer than six surviving Latin commentaries on the principal Pauline Epistles composed in the fourth or early fifth century: Marius Victorinus (Galatians), Ambrosiaster (all four epistles), Jerome (Galatians), Augustine (Romans and Galatians), Pelagius (all four epistles) and the anonymous commentary mentioned above (VL 89). The project supplemented Augustine’s incomplete commentaries on Romans with the biblical text from his exegetical sermons on this epistle. In addition, two Greek commentaries were translated into Latin in this period: Rufinus of Aquileia produced a Latin version of Origen’s commentary on Romans, using an Old Latin text as its biblical lemmata, while the Pauline commentary of Theodore of Mopsuestia—including its biblical text—was rendered into Latin in the circles of Julian of Eclanum, although only the epistles from Galatians onwards are preserved. The sole other original Latin commentary based on an Old Latin text of Paul is the series of brief Complexiones composed by Cassiodorus in the late sixth century. The four collections of testimonia cited in this volume are Ad Quirinum and Ad Fortunatum, both assembled by Cyprian in the third century, the anonymous Speculum or Liber de diuinis scripturis originating in Italy around the year 400, and the Speculum quis ignorat, a late work of Augustine.5

Commentaries and testimonia may preserve multiple forms of biblical text as a result of changes introduced in the course of their transmission. While much of this may be due to subconscious alteration by copyists familiar with a different version, on certain occasions this was the result of a deliberate replacement of the quotations by an editor. Both these types of writing are particularly vulnerable to this sort of substitution as they present substantial portions of biblical text in sequence.6 Augustine’s Speculum quis ignorat is the only work of his corpus in which he extensively cites the text of the Vulgate New Testament outside the Gospels, and it seems that the authorial form of the biblical text was replaced at an early point. Cyprian’s testimonia were updated using at least one other Old Latin tradition. The differing recensions of Ambrosiaster’s commentary (three in Romans, two in the other epistles) appear to be authorial, but subsequent substitution of the lemmata can be seen in some groupings of manuscripts. The discrepancy between the two principal manuscripts of Pelagius is well known: the ninth-century A (Karlsruhe, Badische Landesbibliothek, Aug. 119) has a biblical text close to the Vulgate, while the fifteenth-century B (Oxford, Balliol College, MS 157) has an Old Latin text similar to VL 61. The former is generally accepted to be authorial. The textual tradition of Jerome’s Commentary on Galatians preserves Old Latin forms not adopted in the most recent critical edition.7 For this reason, when making transcriptions of the biblical text in these writings, we have also recorded variant readings reported in the critical apparatus which affect the biblical lemmata. Further details of the manuscripts reported for each work are given in the List of Witnesses.

The sources chosen for this volume therefore present most if not all of the surviving Old Latin evidence for the continuous-text tradition of the four principal Pauline Epistles, either in the direct form of a biblical codex or the indirect form of commentaries or testimonia compiled (or revised) from a biblical manuscript. In contrast to the Gospels, but in keeping with other books of the Old and New Testament, the relative lateness of the forms of text transmitted in biblical codices means that the secondary evidence is particularly important for recovering Old Latin versions. It is therefore fortunate that so many early commentaries on the Pauline Epistles have been preserved. As noted above, not all of the sources presented in this volume are consistently Old Latin in their affiliation: we have opted to cite each witness throughout, in order to enable users to see all the available evidence, rather than making judgments about the distribution of distinctive textual forms.

There is undoubtedly further Old Latin material preserved in sources not included in this collation. The COMPAUL project was also responsible for the creation of a database of quotations of these four epistles in Latin authors based on the transcription of the digitised index cards of the Vetus Latina Database. Almost 100,000 full-text quotations are available through the portal at, which can be searched and ordered by verse. Keyword searches in electronic corpora such as the Library of Latin Texts and Patrologia Latina may identify further references not included in this collection. The importance of prefaces and series of capitula for early Latin biblical traditions should also be recognised. Even so, while the completion of a Vetus Latina edition bringing all this material together remains a future prospect, it is hoped that the present volume and its online counterparts will significantly improve access to and study of the Old Latin evidence for the Pauline Epistles.

Preparation of the Data

Full electronic transcriptions were made of each of the witnesses in plain text using the standards developed by the International Greek New Testament Project and the Verbum Project for work on the Gospel according to John.8 This involved the adjustment of a file with the Vulgate version of the epistle to match the text and layout of each manuscript. The orthography was reproduced exactly, along with details of abbreviations, punctuation, and decoration. All corrections were noted, along with alternative readings or glosses. Gaps and unclear text were also recorded. Where possible, new digital images or digitised microfilm were used for each manuscript, with reference to earlier published transcriptions only when necessary. For the commentaries and testimonia, the transcriptions were based on printed editions. We were able to use existing reconstructions of the biblical text of Marius Victorinus (Galatians), Ambrosiaster (all epistles), Rufinus (Romans) and Pelagius (Romans). In all the other cases, we went through a critical edition transcribing the lemmata or extracts as indicated by the editor. The orthography of each edition was preserved, although word and verse division were standardised. Variant readings reported in the critical apparatus which affected the biblical text were recorded in the same way as corrections in manuscripts, with details of sigla for significant witnesses, although minor spelling changes and other obvious errors in the textual tradition were usually ignored. The layout of the text in printed editions was sometimes also transcribed in order to facilitate proofreading.

Each file was proofread against the original source or, in the case of some manuscripts, a set of digital images which had subsequently become available. In the case of Pelagius’ commentaries on 1 & 2 Corinthians and Galatians, the text extracted from Souter’s edition was compared with digital images of the Karlsruhe and Balliol manuscripts to ensure that the witnesses had been correctly reported.9 The proofread files were converted to XML conforming to the TEI P5 guidelines, using scripts developed by Smith.10 Information about the sources used and the transcribers, as well as a list of subsequent alterations, was recorded in the header of each file. The final transcriptions have been published online at under a Creative Commons attribution licence in order to facilitate their reuse.

For the collation of the transcriptions, only data pertaining to the biblical text was required. Once the XML had been processed to give the desired information in JSON format, the Collation Editor was used to generate an initial collation of each verse using CollateX as the alignment software. The results were presented in the Collation Editor’s web-based interface for further editing. The first stage of this was the removal of purely orthographic variation (“regularisation”). This was undertaken through the creation of global rules, affecting all witnesses throughout that epistle, or local rules restricted to individual witnesses or occurrences. Nonsense readings or partially-extant text were allocated to the corresponding correct form when this could be identified without ambiguity. Each adjustment was recorded, with the appropriate classification to enable the different categories of change to be identified and provide a full history of editorial activity. The second stage involved the adjustment of the alignment of variant readings (“set variants”). This was necessary if the algorithm had not assigned a reading to the appropriate head word. Inconsistencies in word division were also removed and the individual readings were combined based on grammatical principles or their attestation. Where witnesses were only extant for part of the verse, they were marked as absent rather than omissive in the other units. The reviewing of all the data in this way also identified certain inconsistencies in the transcriptions, which were adjusted and recollated as necessary. At each stage, the collation interface checked that all witnesses were present and that the original sequence of words was preserved, to avoid the introduction of error. Once each verse was marked as approved, the apparatus entry was generated following a template which put the readings and witnesses in a consistent order in a standard plain-text format for publication.

Principles and Layout of the Collation

The aim of this collation is to present the evidence for the text of the Pauline Epistles in the selected witnesses as clearly and economically as possible. Particular attention is paid to variations from the Vulgate, as these are likely to represent earlier, Old Latin, forms of text.11 In order to focus on genuine textual variants, most of the incidental variation has been removed: no abbreviations (including numerals) or capitalisation are reported; orthography has been standardised on the basis of the Stuttgart Vulgate; partially-extant words and nonsense readings have, where the evidence is unambiguous, been allocated to the corresponding form. Wherever a reading includes a regularisation, however, this is indicated by a modification to the witness siglum: italics are used to show that the original reading has one or more orthographic differences, e.g. 61; a siglum enclosed in angled brackets, such as ⟨64⟩, denotes text which is incompletely preserved or only partially legible, yet may be confidently identified with the given reading; a siglum in parentheses, such as (89), indicates that the original form was judged by the editors to be a scribal error for the given reading.12 It was deemed that presenting the original form for all of these cases would have resulted in overloading the apparatus and made it more difficult for users without a high level of linguistic or palaeographic expertise. Those wishing to check the actual reading of each witness can refer to the transcriptions presented in full at The use of e-caudata (ę) was treated as a variant for ae or e, but digraphs (including ampersands) were not considered to be orthographic variants.

Outside these general categories, the aim during collation was to adjust as little of the data as possible. Wherever a word is a genuine grammatical form, it has been allowed to stand even if the orthography elsewhere in the manuscript suggests that it is likely to represent a different form. Thus both the future-tense amabit and the perfect-tense amauit are reported as written in the manuscript, regardless of the likely sequence of tenses in a verse or Greek tradition. Clearly erroneous forms such as amauo (for amabo), or amabi (for amaui) have been regularised. Within individual variation units, grammatical inconsistencies such as a plural subject with a singular verb or a masculine noun with a feminine adjective have also been regularised if they are obviously erroneous and represent simple slips or orthographic exchanges. Nevertheless, if there is a possibility that the form could be interpreted grammatically within the whole verse—however implausible the resulting sense—it has been allowed to stand, following the practice of the Novum Testamentum Graecum Editio Critica Maior. Where there is no obvious reading matching a partial or nonsensical form, or the evidence is ambivalent, the original has been preserved. In the case of proper nouns, where there is often extensive orthographical variation, variant forms are only reported if they differ in grammar or derive from a different lexical root.

It should be possible to reconstruct the text of each manuscript from the collation, along with the editorial text of each patristic witness. Given the partial nature of many witnesses (see the Summary of Material below), those present are listed at the beginning of each verse. If a witness is only present for part of a verse, it is listed as lacunose (in the case of continuous-text manuscripts) or absent (for extracts) in other variation units. Glosses or alternative readings are presented on an ad hoc basis (apart from VL 77, where most if not all are part of the text written by the first hand): indications are also given of differences where a verse appears more than once in the same witness. The collation itself is given in the form of a negative apparatus, with the text of the Stuttgart Vulgate as the lemma, in bold, for each variation unit. If no variant readings are given, all witnesses agree with the Vulgate. Witnesses which are not reported in support of a variant reading agree with the Vulgate. If this agreement is the result of regularisation, the sigla are included following the lemma text with the modifications described above. If a first-hand reading is reported for a variant and a correction is not given, the correction is to the Vulgate text. In the case of commentaries and testimonia, the editorial text (ed) is always specified when a variant is present unless it agrees with the Vulgate and has not been regularised. Specific manuscripts of Ambrosiaster are indicated when a variant is peculiar to them (see the List of Witnesses). As no editorial text is available for Pelagius outside Romans, manuscripts A and B are reported throughout the three latter epistles: if only one manuscript is indicated, then the other agrees with the Vulgate; the agreement of the two manuscripts is only explicitly stated when a variant is present elsewhere in the textual tradition and A and B do not match the Vulgate.

The length of variation units has been determined by a combination of attestation and practicality. Grammatical units or phrases are treated as single variants so long as this does not give rise to too many different readings: if the number of resulting readings is more than seven or so, then they are divided into their constituents. The same is also true of transpositions: it is normally possible to present the whole phrase for shorter transpositions, but longer transpositions (particularly with intervening variants) may be reported as omissions and additions to the standard flow of text. Smaller variation units have often been combined in order to reduce the size of the printed volume. For ease of presentation, it was decided that each reading should only be presented once even though the collation software had the option of treating transpositions as overlapping variants in addition to any lexical differences within the phrase. The same sequence of witnesses is used throughout: manuscripts are listed in numerical order and then the other two classes of evidence are listed in chronological order within each type, giving the sequence MAR AMst HI AU PEL RUF THr CAr for the commentaries and tes for spm spe for the testimonia. Variant readings are given in the order of the first witness in which they are attested. Only manuscripts are cited for the incipit and explicit of each epistle.

Lines of Research

The full analysis of this data has yet to be undertaken.13 The oldest textual stream is that found in Cyprian’s testimonia (tes and for), followed by the adjustments to match differing Old Latin forms as preserved in their manuscript tradition.14 The lemmata of Marius Victorinus and Ambrosiaster represent an Italian text of the early fourth century. A slightly later version of this is the earliest form attested in biblical codices, found in the fifth-century VL 75, its descendants VL 76 and VL 83, and VL 89. The Pseudo-Augustine Speculum (spm) stands in a similar position, as does the lemma text provided by Rufinus for his translation of Origen (RUF). Jerome’s lemma text of Galatians (HI) has both Old Latin and Vulgate readings: the latter are likely to derive from editorial intervention early in its transmission.15 Augustine (AU), VL 64, VL 80 and the Old Latin part of VL 88 witness to a separate early branch from the fourth century. The most distinctive other non-Vulgate text is that found in the ninth-century VL 61 and the lemmata of manuscript B of Pelagius (PELB), although the date of its origin has not yet been established. The other witnesses, by and large, are Vulgate texts with occasional Old Latin readings.

Two witnesses stand apart from mainstream tradition: the interlinear glosses of VL 77 and the ad hoc translation of the lemmata of Theodore of Mopsuestia (THr). The inclusion of both in the Vetus Latina registers provides sufficient reason to report their readings here, even though they both seem to be one-off productions which had no influence on subsequent tradition (apart, perhaps, from the glosses in VL 78). In fact, the variety of renderings in these two independent versions serves to highlight the overall consistency of the rest of the material, suggesting that the extant Latin texts of the Pauline Epistles also derive from a single initial translation.


For a fuller treatment of the whole corpus as well as specific observations on the Pauline Epistles, see H.A.G. Houghton, The Latin New Testament. A Guide to its Early History, Texts, and Manuscripts (Oxford: OUP, 2016).


A summary of scholarship on the origin of the Vulgate version of the Pauline Epistles is given in the contribution of Anna Persig to the forthcoming Oxford Handbook of the Latin Bible.


Roger Gryson (ed.), Altlateinische Handschriften/Manuscrits vieux latins. Répertoire descriptif. Mss 1–275. (Vetus Latina 1/2A). Freiburg: Herder, 1999; Hugo Eymann (ed.), Vetus Latina. Band 21. Epistula ad Romanos. Einleitung. Freiburg: Herder, 1996; Uwe Fröhlich (ed.), Vetus Latina. Band 22. Epistula ad Corinthios I. Einleitung. Freiburg: Herder, 1995–1998.


Holger Strutwolf et al. (ed.), Nestle-Aland Novum Testamentum Graece. 28th revised edition. Stuttgart: Deutsche Bibelgesellschaft, 2012; United Bible Societies, Greek New Testament. 5th revised edition. Stuttgart: Deutsche Bibelgesellschaft, 2014. For a concordance of sigla used for these manuscripts in different editions, see the Appendix.


See further Houghton, The Latin New Testament, 9, 38–39, 170–172.


For an illustrated examination of this question with particular reference to the writings included in the present volume, see H.A.G. Houghton, “The Layout of Early Latin Commentaries on the Pauline Epistles and their Oldest Manuscripts” in Studia Patristica vol. XCI. Papers presented at the Seventeenth International Patristics Conference, ed. M. Vinzent (Leuven: Peeters, 2017), 71–112.


H.A.G. Houghton, “The Biblical Text of Jerome’s Commentary on Galatians”. Journal of Theological Studies ns 65.1 (2014) 1–24.


See further H.A.G. Houghton and C.J. Smith, “Digital Editing and the Greek New Testament” in Claire Clivaz, Paul Dilley and David Hamidović (edd.), Ancient Worlds in Digital Culture. Digital Biblical Studies 1 (Leiden: Brill, 2016), 110–127.


Errors in Souter’s apparatus have been indicated as notes in the transcriptions.


For a description of the XML encoding, see H.A.G. Houghton, “The Electronic Scriptorium: Markup for New Testament Manuscripts”, in Claire Clivaz, Andrew Gregory and David Hamidovic (edd.), Digital Humanities in Biblical, Early Jewish and Early Christian Studies. Leiden: Brill, 2014, as well as the protocol at


The Stuttgart Vulgate has been taken as the standard: R. Weber, R. Gryson et al., Biblia Sacra iuxta Vulgatam versionem (fifth edition). Stuttgart: Deutsche Bibelgesellschaft, 2007.


Where a later hand corrected this form but the resultant reading is the same, this is indicated by the addition of the corrector siglum outside the parentheses, e.g. (89*)C.


Fröhlich provides a comprehensive introduction to 1 Corinthians, which may be supplemented by J.J. Kloha, “A Textual Commentary on Paul’s First Epistle to the Corinthians” (PhD thesis, University of Leeds, 2008); earlier works such as Friedrich Zimmer, Der Galaterbrief im altlateinischen Text als Grundlage für einen textkritischen Apparat der Vetus Latina (Königsberg: Hartung, 1887) and H. Zimmermann, Untersuchungen zur Geschichte der altlateinischen Überlieferung des Zweiten Korintherbriefes (Bonn: Hanstein, 1960), may still be useful. An overview and discussion of key readings is given in Houghton, The Latin New Testament (esp. 169–176), and from the perspective of Greek tradition, B.M. Metzger, A Textual Commentary on the Greek New Testament. Second edition. (Stuttgart: Deutsche Bibelgesellschaft, 1994) remains the fullest treatment.


See further R. Weber, Sancti Cypriani Episcopi Opera pars I: Ad Quirinum, Ad Fortunatum. Corpus Christianorum series latina 3 (Turnhout: Brepols, 1972), pp. li–lx.


See Houghton, “The Biblical Text of Jerome’s Commentary on Galatians”.

  • Collapse
  • Expand