Jump to ContentJump to Main Navigation
The Old Latin GospelsA Study of their Texts and Language$

Philip Burton

Print publication date: 2000

Print ISBN-13: 9780198269885

Published to Oxford Scholarship Online: November 2003

DOI: 10.1093/0198269889.001.0001

Show Summary Details
Page of

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (www.oxfordscholarship.com). (c) Copyright Oxford University Press, 2018. All Rights Reserved. Under the terms of the licence agreement, an individual user may print out a PDF of a single chapter of a monograph in OSO for personal use (for details see www.oxfordscholarship.com/page/privacy-policy). Subscriber: null; date: 16 January 2019

Lines of Enquiry

Lines of Enquiry

(p.3) 1 Lines of Enquiry
The Old Latin Gospels

Philip Burton (Contributor Webpage)

Oxford University Press

Abstract and Keywords

What is the status quaestionis in Old Latin studies? Is the term Itala helpful or confusing? What is the relative value of the manuscript evidence vis à vis the patristic citations? Within the field of textual criticism, how far is it legitimate to reconstruct unattested variants?

Keywords:   Itala, Old Latin, patristic citations, textual criticism

1.1 Introduction

There are preserved in libraries across Europe some thirty manuscripts, some very fragmentary, which contain translations of the four canonical Gospels which predate the Vulgate of Jerome. These manuscripts, of which about ten are extant at any given point, are collectively known as the ‘Old Latin Gospels’ (hereafter OLG). The textual relations and language of these manuscripts have received considerable, if uneven, attention from philologists and theologians. This present study is an attempt both to synthesize their work, and to advance it. Three main questions are addressed. How did the OLG come into being? What are the techniques employed by their translators? How far can they be used as sources for the development of post‐classical Latin?1

These questions are closely intertwined. Our enquiry into the origins of these versions will take its starting‐point from an examination of the translation of certain key words. The knowledge thus gained may cast light on their value as sources for post‐classical Latin. An evaluation of the position of the OLG within post‐classical Latin will, in turn, help us to see how far the translators are prepared to follow everyday patterns of speech and how far they felt bound by the constraints of latinitas.

At this stage it is important to delimit the areas that will not be touched upon.

First, our current concern is solely with the Gospels and not with other parts of the Scriptures. Loose references to ‘the Latin Bible’ appear to presuppose a single monolithic translation; no such homogeneity has (p.4) been demonstrated, and the term is therefore misleading. Statements made about the origins, translation technique, and language of one part of the Bible should not be generalized to the Bible as a whole.

Secondly, we are not directly concerned with the identification of the types of Greek text underlying the various Latin traditions, nor with the value of the OLG for the textual criticism of the Greek Gospels.2 However, the possibility that a particular Latin reading is due to a Greek variant will be raised from time to time, even when the putative variant does not appear in any extant manuscript. The conditions under which the possibility of such an unattested Greek reading may be raised are discussed later in this chapter.

Thirdly, we are not directly concerned with the Biblical citations given in the patristic writings.3 The volume of patristic citations of the Gospels would prohibit more than the most cursory examination of this material. In addition to this, there are unique problems with the patristic citations. On encountering a Scriptural reference in the Fathers, we do not necessarily know whether it is intended to be an exact quotation or a loose allusion or conflation of references; whether the writer is making his own version or quoting from an existing one; whether he has the text in front of him or is quoting from memory. Moreover, while there was no official policy before the sixteenth century of substituting the Vulgate readings for the original references in patristic texts, none the less it is likely that copyists familiar with the Vulgate (or other Old Latin, or liturgical) forms of a given passage would unconsciously introduce the words they knew best into their copies of the Fathers. The patristic citations are thus too complex to be dealt with in sufficient depth. However, the Oxford Old Latin editors (of whom more presently) of the last century developed the technique of comparing the readings given in the manuscript traditions with those in the patristic writings, as a means of giving a terminus post quem and perhaps a provenance for the various manuscript traditions.

(p.5) 1.2 Terminology: ‘Old Latin’ and ‘Itala’

The term ‘Old Latin’ (Vetus Latina) is now the generally accepted name for the pre‐Jerome translations. The name Itala is often found in the older secondary literature. This term derives from a passage in the second book of Augustine's de Doctrina Christiana:

(2.11) Qui enim scripturas ex bebraea lingua in graecam verterunt numerari possunt, latini antem interpretes nullo modo. Ut enim cuique primis fidei temporibus in manus venit codex graecus et aliquantulum facultatis sibi utriusque linguae babere videbatur, ausus est interpretari. . . (2.15) In ipsis autem interpretationibus Itala ceteris Praeferatur, nam est verborum tenacior cum perspicuitate sententiae.

The conflicting views on this controversial passage are admirably summarized in Schildenberger (1952),4 from which it appears that there are three main opinions.

  1. 1. The passage is corrupt. This view was first put forward by Richard Bentley, who proposed the rather banal emendation illa ceteris praeferatur quae est verborum tenacior. . . . More recent critics have attempted various emendations involving the name of Aquila, the Jewish proselyte who prepared a very literal Greek version of the Jewish scriptures around AD 130. These conjectures, however, lack textual support, and cannot always be justified as being appropriate to the argument of the passage.

  2. 2. It is a reference to the Vulgate. Augustine (according to Schildenberger) does quote Isaiah 7:9 and Isaiah 58:7 in de Doctrina Christiana 2.17 in a form similar to that of the Vulgate, but even if this is not coincidental it cannot be taken to be a wholesale endorsement of Jerome's work.5 Nor is there any other evidence for Itala as a name of the Vulgate.

  3. 3. It is a reference to an existing Old Latin tradition. This is the traditional interpretation, held by Sabatier and Jülicher; it is also upheld by Schildenberger. It is rather more plausible than the alternatives, but presents two main problems. First, it seems to take Itala to refer to a single translation of the whole canon of Scripture. In fact one‐ or two‐ volume sets of this sort (that is, pandects) are not known to have existed before the sixth century. It is known that translations of some (p.6) related books of the New Testament were circulated together; this is true of the Pauline corpus and the Catholic Epistles, and, it will be argued, of the Gospels too. But it is strange that Augustine should be able to speak without further qualification of a single Itala. Secondly, it is impossible to tie the term Itala to a single known tradition; Augustine's many biblical citations and allusions have not been identified with any one extant manuscript type.

The complications of this issue are such that, as the distinguished scholar Bonifatius Fischer observes, the term is best avoided.6 Accordingly, we shall concentrate instead on the actual texts of the manuscripts, and on what can be deduced about their relationships on purely internal grounds. The channels of communication between Latin Christian communities by which the various traditions circulated lie outside the scope of the present study; an evaluation of what the manuscript relations are must precede a study of how they came about.

1.3 Vulgate and Mixed Texts

Although we have talked in terms of a division between ‘Old Latin’ and ‘Vulgate’ translations, it should be noted that this division is in practice not such a neat one. The Vulgate Gospels were, as Jerome states, intended to be a minimal revision of the existing Old Latin versions, and do bear a strong resemblance to them.7 It has also been questioned how far modern texts of the Vulgate actually represent Jerome's work. There are two main problems: first, knowing which books Jerome actually revised; and secondly, knowing in the case of the books he did revise how far the extant manuscripts actually represent his work. As to the Gospels, it is beyond doubt that Jerome did revise them, and the manuscript evidence for them is extremely good. All the great early Vulgate manuscripts—Codex Amiatinus (C8), Codex Cavensis (C9), Codex Dublinensis (C8–9), (p.7) Codex Fuldensis (C6), Codex Mediolanensis (C6), Lindisfarne Gospels (C7), Codex Harleianus (C6–7), Codex Sangallensis (C5), Pierpoint Gospels (C10)—contain the Gospels. Modern editors are thus able to go beyond the revision associated with Alcuin, not to mention the much later Sixtine and Clementine editions.

The Old Latin texts did not go out of use when the Vulgate appeared. The oldest extant Old Latin manuscripts date only to the end of the fourth century, that is, around the time when the Vulgate appeared. Most date from around the fifth to eighth centuries, with the latest from the thirteenth. The Old Latin texts were thus in circulation alongside the Vulgate. Inevitably, there was much cross‐fertilization between the two traditions, as Vulgate readings crept into texts that were basically Old Latin, and vice versa. In modern times it has become customary to describe as ‘mixed’ those texts which are fundamentally Old Latin in type, but with a distinctive Vulgate overlay.

It is not always easy to distinguish in any individual passage whether a manuscript should be regarded as Vulgate or Old Latin. However, given a larger portion of the text, the identification becomes easier. The most distinctive feature is the readings of the text. Jerome's major contribution to Latin Gospels was his ability as a textual critic; at many points the Vulgate differs from the Old Latin in following a text closer to that found in modern Greek editions. The renderings are also important; if a manuscript frequently agrees in its renderings with the Vulgate against the undisputed Old Latin traditions, it is likely to be a mixed text. (Occasional correspondences are non‐diagnostic; the copyist may be unconsciously recalling the Vulgate, or coincidentally altering the wording in the direction of the Vulgate, or simply copying older material which had anticipated the Vulgate.) The third feature of mixed texts is the order of the Gospels. ‘Pure’ Old Latin texts have the ‘Western’ order Matthew–John–Mark–Luke (found also in the Greek ‘Western Text’ and the Gothic version), whereas Jerome preferred the more familiar ‘Eastern Order’.

Of the thirteen main manuscripts traditionally classed as Old Latin (see Chapter 2), no fewer than six are mixed texts. The level of Vulgate admixture varies considerably; it will be argued that in John two so‐called ‘Old Latin’ manuscripts are basically Vulgate texts with Old Latin elements rather than the reverse. Nor is it always easy to identify which elements within a mixed text are Old Latin and which are Vulgate. For the purposes of this study the following principle will be observed: any reading found in a known mixed text, agreeing with the Vulgate (p.8) but not found outside the Vulgate and the other mixed texts, may be attributed to Vulgate influence.

Occasional similarities between Old Latin and Vulgate manuscripts may of course be coincidental; but in manuscripts where these resemblances frequently occur this explanation is merely otiose special pleading. Pushed to its logical extreme, it would require us to believe that two wholly identical translations of the same text (of potentially infinite length) were not genetically related to each other. The direction of the influence, however, cannot be proven absolutely. It is possible to maintain that the ‘mixed texts’ are in fact a pure Old Latin tradition which Jerome took as the basis for his revision, and that this accounts for their similarity to the Vulgate. But there is a serious chronological embarrassment; the oldest extant OLG manuscripts—roughly contemporaneous with the Vulgate, and representing traditions known from the patristic citations to be older—do not belong to the so‐called mixed‐text group, none of which is earlier than the sixth century. Again, it is possible to maintain that manuscripts of this type were indeed in circulation in the late fourth century, but no exemplars from this period have survived; but this ex silentio argument is worthless precisely because it cannot be disproved.

1.4 Editions

For many parts of the Scriptures the most accessible edition is still Sabatier's monumental Bibliorum sacrorum Latinae versiones (1743). However, in respect of the Gospels it has been superseded, mostly through the efforts of the editors of the Oxford Old Latin Biblical Text series (OOLBT), J. Wordsworth, J. Sanday, H.J. White, and E. Buchanan, between 1882 and 1911, and the studies of Heinrich Vogels between 1913 and 1953. The unique contribution of these scholars was to examine each manuscript minutely and individually, with a view to recovering not only its text but its relations with other manuscripts, and the various stages in its prehistory that could be discerned. This contrasts on the one hand with Sabatier's view (derived from his reading of the Itala passage in Augustine's de Doctrina Christiana) of a single unified Old Latin Bible, and on the other hand with the account given by Ziegler (1879), who had argued in favour of a multiplicity of translations, also on the basis of the statements in the Fathers rather than on an examination of the extant texts. The findings of Vogels and (p.9) the Oxford editors are summarized in Chapter 2. Among the other scholars, particular mention may be made of F. C. Burkitt, who first established connections between the Gothic Bible and a branch of the Old Latin. The speculations of J. Rendel Harris (notably on Codex Bezae Cantabrigiensis) were less temperate, though it is unfortunate that many of the questions raised by him (especially on the position of the Old Latin vis‐à‐vis the Old Syriac) have yet to be answered. The work of J. Belsheim in editing several of the manuscripts should also be mentioned; unfortunately, his texts have been found unreliable in many places and his introductions failed to take into account the methods being applied by the contemporary Oxford series.

Since 1949 a major series of volumes has been published by the Stiftung (now Institut) Vetus Latina at Beuron in South Germany. These aim to provide a ‘new Sabatier’, presenting not only all the manuscript evidence but also all the relevant patristic material. This splendid series has, however, not yet been extended to cover the Gospels. For the individual Gospel manuscripts at least, the best texts are still usually the Oxford editions.8

1.5 Matzkow–Jülicher–Aland

But excellent as the Oxford texts are, it is inconvenient to the point of impracticality to rely on separate editions of each individual manuscript. For this reason the edition that will be used here is the Itala of Adolf Jülicher, revised by Walther Matzkow and Kurt Aland (hereafter MJA). Churlish as it may seem, we should still consider the limitations of this invaluable edition before proceeding further.

Jülicher adopted the format of an upper line of text giving a version he designated ‘Itala’, and a lower line giving one designated ‘Afra’.9 This assumes both a neat division between traditions and a degree of homogeneity within them that Jülicher did not attempt to demonstrate. The problem is aggravated by the fact that in the top line of the ‘Itala’ section Jülicher did not follow any one manuscript (as he did in the ‘Afra’), but attempted a composite reconstruction from various manuscripts. No rationale was given for this reconstruction, and the revisors (p.10) retained it only out of respect for the original editor.10 It will be argued that for the Synoptic Gospels Jülicher's assumption is, in fact, broadly valid, whereas for the Gospel of John it is misleading.

In order to compress so many manuscripts into this format Jülicher was obliged to omit details of columniation of texts, of capital letters, and of lectionary notes, which often give information about the relationship between manuscripts. Less pardonably, he was inconsistent in giving variants of orthography or of abbreviations; occasionally the individual editions of manuscripts give non‐standard spellings where MJA has the standard forms. In these cases the non‐standard spellings are more likely to be correct.

The MJA volumes are now out of date in one important respect. The text of the Vulgate cited was the Oxford edition of J. Wordsworth et al. (1898). This has now been superseded by the so‐called ‘Stuttgart Vulgate’ of R. Weber et al. (1969).

1.6 ‘Typical’ Renderings

Citations from MJA in this study will often give the reading of only one manuscript, or (less frequently) of Jülicher's reconstructed ‘Itala’ line, noting that this is the ‘typical’ rendering, or that a ‘similar’ rendering is found in a range of manuscripts. Some gloss is needed. In such cases there may be considerable differences between the extant texts; a reading is regarded as ‘typical’ of a larger group when it agrees with them on the specific point under discussion. Thus for example in Appendix 1 the translation practices of the Old Latin and the Vulgate are compared and contrasted. It is noted that in the Parable of the Tenants in the Vineyard the Old Latin translators usually render ὁγεωργóς by the more specific colonus (‘tenant’), whereas the Vulgate has the more general agricola. Thus it might be said that at Matthew 21: 35 the Old Latin texts designated a b d ff2 h q r1 e typically have the rendering et coloni adprehensis servis unum ceciderunt; they all agree in reading coloni, against agricolae in the Vulgate and mixed texts aur f g1 l. In fact, there is some divergence between them; ff2 r1 have coloni autem, e has the spellings adpraehensis and caeciderunt, d (here as often the most idiosyncratic) has accipientes servos coloni eius quem quidem ceciderunt. This is an extreme case, and most of the variations thus passed over by this (p.11) method are of the minor kind: et for autem, small differences in word order or orthography. These may be matters of some importance for critics concerned with tracing the affiliations and prehistory of individual manuscripts. But in the context of a comparison of Old Latin and Vulgate translation technique they are less relevant, and may be omitted. The point of citing texts in their ‘typical’ form is to include all the facts germane to the issue at stake, while passing over those that are not.

1.7 Greek Variants

It has been mentioned above that manuscripts may differ not only in their rendering of the Greek, but also in the reading of their underlying Greek text. Such differences are easy to identify in cases where the Latin reflects an attested Greek variant.11 To take a simple example, in the Transfiguration story at Matthew 17: 2 Jesus' clothes become ‘white as light’,                       Lines of Enquiry. In all the Old Latin traditions save Codex Monacensis q, and in the Vulgate, this appears as candida (or alba) sicut nix. This must reflect the variant ὡς χιών found in the Greek half of Codex Bezae (D),12 supported by the Curetonian Syriac and the Bohairic Coptic; it is perverse to imagine that it could have arisen in so many places independently. However, sometimes the Latin text does not correspond exactly to any attested form of the Greek; it may be hard to discern whether the Latin translators are making a free rendering of an attested text, or following a lost Greek tradition. Appeals to lost readings cannot by definition be verified and so must be made with caution. In this study such appeals will be made only when one or more of the following circumstances obtains:

  1. 1. If the Latin text cannot reflect any attested Greek reading. For example, at Luke 4: 19 most Latin texts have praedicare annum acceptum Domini et diem retributionis or similar; the Greek has simply                       Lines of Enquiry. The last three words of the Latin (or others (p.12) corresponding to them) are missing only from Codex Bezae d; the passage is a citation from Isaiah 61: 2 and has clearly been added to the Greek from the Septuagint, though there is no direct attestation of this.

  2. 2. If the Latin text corresponds not to the attested Greek text but to one which could have arisen from a plausible palaeographical variation. Thus at Luke 5: 10 most Old Latin texts have ex hoc iam eris homines vivificans for the Greek                       Lines of Enquiry; the Latin would suggest an unattested variant                       Lines of Enquiry or                       Lines of Enquiry.

  3. 3. If the Greek variant is not attested at the place in question, but variants of the same type are attested elsewhere. This is particularly frequent in the case of near‐synonyms, where often the less common term is displaced by the more common. Thus at Matthew 27: 15, 20, 24 the word populus is found in almost all the Latin manuscripts where modern editors read ὁ ὄχλoς. Now populus does not usually translate ὁ ὄχλoς but ὁ λαός, which is in fact attested as a variant to ὁ ὄχλoς in a few Greek codices at Matthew 27: 24, though not at verses 15, 20. Conversely ὁ ὄχλoς is sometimes found as a variant for ὁ λαός (for example, at Mark 11: 32). Clearly, the two terms are to some extent interchangeable in the Greek; it is, therefore, reasonable to suggest that at Matthew 27: 15, 20 the reading ὁ λαός underlay the rendering populus, even though it is not directly attested.

  4. 4. (In the Synoptic Gospels) If the Latin corresponds not to the accepted form of the Greek text at that point but to the form of the same passage found in the Synoptic parallels. For instance at Mark 4: 19 the word ἡ ἀπάτη (or ἀπάται; in the phrase                       Lines of Enquiry) in modern editions appears in various Old Latin traditions as delectationes (c ff2), inlecebrae (f), or oblectationes (e). These words may be regarded as interpretative glosses on αἱ ἀπάται, but are more likely to reflect the reading                       Lines of Enquiry in the parallel passage Luke 8: 14, though no such reading is attested at Mark 4: 19.

These are the conditions under which it has been thought legitimate to propose an unattested variant in the Greek. It should be noted that we are not here concerned with the reconstruction of the Vorlage (underlying Greek text) for its own sake, but only where it casts light upon some difficulty in the Latin. When a Latin text or texts does not give an obvious rendering of the Greek, some criteria are needed for deciding whether the translators are following a variant reading or adopting a freer technique of translation. It should be added that while the principles listed above are quite straightforward, their application is (p.13) less so. In the last example, for instance, it might be argued that ‘the delights’ (delectationes, inlecebrae, oblectationes) of wealth was an interpretative gloss on ‘the deceits’ of wealth, and were renderings of αἱ ἁπάται. However, on a balance of probabilities it seems more likely that the translations delectationes, inlecebrae, oblectationes are literal renderings of an underlying variant, though this is impossible to prove.


(1) An essential introduction to the Old Latin New Testament is given by Fischer (1972). Good surveys of the issues in Old Latin studies are given in Schäfer (1957), Metzger (1977: 285–330), Reichmann (1980), Bogaert (1988), and Elliot (1992). For an exhaustive bibliography of Old Latin and cognate studies in recent years see Bogaert (1974, 1995) and Gribomont (1991). These works between them may be taken as representing the scholarly consensus.

(2) Aland's view (in Nestle et al. 1985: 54) that ‘the early versions . . . are frequently unwarrantedly overrated’ may stand (despite its tautology) as typical of current critical opinion. Contra Gryson (1988) argues that the Old Latin is a ‘témoin privilégié du texte du Nouveau Testament’; his argument is cogent for the passage (Matthew 13: 13–15) on which he concentrates, but it is invalid to extrapolate from this to the rest of the Gospels, let alone the NT (New Testament) as a whole. Certainly the OLG may be used to reconstruct Greek readings from the third and fourth centuries; these readings should then be subject to the normal rules of textual criticism.

(3) On the general problem of using patristic citations see Frede (1972).

(4) See also summary in Metzger (1977: 290–3).

(5) ‘Augustine appears to have used Jerome's gospels regularly since about the year 400, yet nowhere does he betray the slightest knowledge of a version by Jerome of any other book in the New Testament . . . nor does he ever appear to quote the Vulgate beyond the gospels’ (Sparks 1970: 519).

(6) ‘Heute wird es besser vermieden, weil er unklar ist, hauptsächlich wegen der verschiedenartigen Bedeutungen, die ihm beigelegt werden’ (Fischer 1972: 5).

(7) In the Epistula ad Damasum prefatory to the Vulgate Gospels Jerome writes ‘ . . .ita calamo imperavimus (or temperavimus) ut his tantum quae sensum videbantur mutare correctis, reliqua manere pateremur ut fuerant’; the nature of Jerome's linguistic revision is examined in Appendix 1. The Greek and Latin bases of the Vulgate Gospels are analysed by Vogels (1928a). On the general background to Jerome's work see Sparks (1970), Kelly (1975). On the early textual history of the Vulgate see Berger (1893).

(8) The most thorough collection of Latin Gospel readings is Fischer (1989). However, with an average of four Latin words per page it cannot well be read as continuous text.

(9) For a discussion of the origin and value of this term see Ch. 2.

(10) ‘Von der ursprünglichen Arbeit Jülichers is nur die Leitzeile geblieben und zwar als Akt der Pietät’ (Aland in intro. to MJA vol. iii).

(11) A reading is counted as attested if it is listed in the text or apparatus of Nestle et al. (1985) or (in the case of Matthew and Mark) of Legg (1937 and 1940). I much regret that the continuation of this work by the American and British Committee of the International Greek New Testament Project came to my attention too late to allow me to make full use of it.

(12) Not in itself good authority, as this is a Greek—Latin bilingual codex, and the reading may be a back‐translation from the Latin half of the codex.