Versions of the New Testament

Contents: Introduction * Anglo-Saxon * Arabic * Armenian * Coptic: Sahidic, Bohairic, Other Coptic versions * Ethiopic * Georgian * Gothic * Latin: Old Latin, Vulgate * Old Church Slavonic * Syriac: Diatessaron, Old Syriac, Peshitta, Philoxenian, Harklean, Palestinian, "Karkaphensian" * Udi (Alban, Alvan)


The New Testament was written in Greek. This was certainly the best language for it to be written in; it was flexible and widely understood.

But not universally understood. In the west, there were many who spoke only Latin. In the east, some spoke only the Syriac/Aramaic dialects. In Egypt the native language was Coptic. And beyond the borders of the Roman Empire there were peoples who spoke even stranger languages -- Armenian, Georgian, Ethiopic, Gothic, Slavonic.

In some areas it was the habit to read the scriptures in Greek whether people understood it or not. But eventually someone had the idea of translating the scriptures into local dialects (we now call these translations "versions"). This was more of an innovation than we realize today; translations of ancient literature were rare. The Septuagint translation of the Hebrew Bible was one of the very first. Despite the lack of translations in antiquity, it is effectively certain that Latin versions were in existence by the late second century, and that by the fourth there were also versions in Syriac and several of the Coptic dialects. Versions in Armenian and Georgian followed, and eventually many other languages.

The role of the versions in textual criticism has been much debated. Since they are not in the original language, some people discount them because there are variants they simply cannot convey. But others note, correctly, that these versions convey texts from a very early date. In many instances the text-types they represent survive very poorly or not at all in Greek.

It is true that the versions often have suffered corruption of their own in the centuries since their translation. But such variants usually are of a nature peculiar to the version, and so can be gotten around. When properly used, the versions are one of the best and leading tools of textual criticism.

This essay does not attempt to fully spell out the history and limitations of the versions. These points will briefly be touched on, but the emphasis is on the textual nature of the versions. Those who wish to learn more about the history of the versions are advised to consult a reference such as Bruce M. Metzger's The Early Versions of the New Testament: Their Origin, Transmission, and Limitations (Oxford, Clarendon Press, 1977).

In the list which follows, the versions are listed in alphabetical order.

An additional note: Of all the articles in this Encyclopedia, apart from those which touch on science and theology. this has been among the most controversial. I don't mean that people disagree with the results particularly; that happens everywhere. But this one seems to make people most upset. Please note that I am not setting out to belittle any particular version, and except in textual matters, I am not expert on these versions. I will stand by the statements on the textual affinities of the more important versions (Latin, Syriac, Coptic; to a lesser extent, the Armenian, Georgian, and Gothic) insofar as they are correctly incorporated into the critical apparatus. For the history and such, I am dependent upon others. If you disagree with the information here, I will try to incorporate suggestions, but there is only so much I can do to make completely contradictory claims fit together....


A name used for several translations, made independently and of very different types, used in Britain mostly before the Norman Conquest and of interest more to historians than textual scholars. But since they are important for the understanding of early English literature (they give us, among other things, important vocabulary references), it seems worthwhile to at least mention them here, while understanding that what limited text-critical value they have is mostly for Vulgate criticism.

The Lindisfarne Gospels (Wordsworth's Y -- Latin vulgate text with interlinear glosses in the Northumbrian dialect (shown in red highlight). The Latin is from the seventh century; the interlinear is from the tenth. The decorated page containing John 1:1 is shown.

Although Roman Britain was Christian, the Anglo-Saxon invasions of the late fifth century effectively wiped out Roman Christianity. And it would be centuries before Christianity completely took control of the island, because the German invaders immediately split the island into dozens of small states, of which seven survived to become the "Seven Kingdoms of Britain": Northumbria, Mercia, Wessex, Essex, Sussex, Kent, and East Anglia. To make matters worse, all these kingdoms had slightly different dialects.

It was in 563 that Saint Columba founded the religious center on Iona, bringing Celtic Christianity back to northern Britain. In 596 Pope Gregory the Great sent Augustine to Canterbury to return southern Britain to Christ. The two Christian sects were formally reconciled at the Synod of Whitby in 664. This did not make Britain Christian (and, ironically, it did not bring Ireland into line with Catholic Christianity; that island, now known for its Catholicism, was brought back into line with the Catholic church by the Anglo-Norman invaders), but the way was at last clear.

The earliest attempts at Anglo-Saxon versions probably date from this time, but they have not survived. Nor has the translation of John made by the Venerable Bede. Alfred the Great worked at a translation, but it seems never to have been completed. All that is known to have existed is a portion of the psalms, including a detailed (though often fanciful) commentary said to have been by Alfred himself. (In this connection it may be worth noting that Asser, Alfred's biographer, at several points quotes the Bible in Old Latin rather than Vulgate forms.)

Our earliest surviving Anglo-Saxon versions date from probably the tenth century. Several of these are continuous text versions; the most famous of these is probably the Hatton Gospels, now in the Bodleian; this beautifully-written manuscript is thought to be from the eleventh century. Other Old English renderings are interlinear glosses to Latin manuscripts. The interlinears are in several dialects.

In many ways the Anglo-Saxon was better suited to literal Bible translation than is modern English, since Anglo-Saxon is an inflected language with greater freedom of word order than modern English. Since, however, all Anglo-Saxon translations are taken from the Latin (unless Bede made some reference to the Greek), they are not generally cited for New Testament textual criticism. This is proper -- though they perhaps deserve more attention for Vulgate criticism; it should be recalled that the early English copies of the Vulgate were of very high value, so the translations could well derive from valuable originals.

We should note that the term "Anglo-Saxon" is now frowned upon by linguists, who much prefer the term "Old English." I have yet to see this term applied to the early English translations. The name "Anglo-Saxon" seems to be used in the same sense that "Ethiopic" is used for a version that is in a language not properly called "Ethiopic": It's a geographic/historical description.


Arabic translations of the New Testament are numerous. They are also very diverse. They are believed to have been made from, among others, Greek, Syriac, and Coptic exemplars. Other sources may be possible.

Folio 1 recto of Sinai Arabic 71 (Xth century), Matthew 23:3-15.
Thanks to Jean Valentin

Although there are hints in the records of Arabic versions made before the Islamic conquests, the earliest manuscripts seem to date from the ninth century. The oldest dated manuscript of the version (Sinai arab. 151) comes from 867 C.E. The translations probably are not more than a century or two older.

Several of the translations are reported to be very free. In any case, Arabic is a Semitic language (which, like Hebrew, has a consonantal alphabet, leaving room for interpretation of vowels) and frequently cannot transmit the more subtle nuances of Greek grammar. In addition, written Arabic was largely frozen by the Quran, while the spoken language continued to evolve and develop regional differences. This makes the Arabic versions somewhat less vernacular than other translations. This would probably tend to preserve the original readings, but may result in some rather peculiar variants.

The texts of the Arabic versions have not, to this point, been adequately studied. Some seem to be purely or primarily Byzantine, but at least some are reported to contain "Cæsarean" readings. Others are said to be Alexandrian. Still others, with something of an "Old Syriac" cast, may be "Western."

Several late manuscripts preserve an Arabic Diatessaron. The text exists in two forms, but both seem to have been influenced by the Peshitta. They are generally regarded as having little value for Diatessaric studies.

It will be obvious that the Arabic versions are overdue for a careful study and classification.


The Armenian translation of the Bible has been called "The Queen of the Versions."

The title is deserved. The Armenian is unique in that its rendering of the New Testament is clear, accurate, and literal -- and at the same time stylisticly excellent. It also has an interesting underlying text.

The origin of the Armenian version is mysterious. We have some historical documents, but these may raise more questions than they solve.

The most recent summary on the subject, that of Joseph M. Alexanian, states that the initial Armenian translation (Arm 1) was made from the Old Syriac in 406-414 C.E. This was followed by a revised translation (Arm 2) made from the Greek after the Council of Ephesus in 431. He suggests that further revisions followed.

In assessing Alexanian's claims, one should keep in mind that there are no Armenian manuscripts of this era, and the patristic citations, while abundant, have not been properly studied or catalogued.

Armenia is strongly linked with Syrian Christianity. The country turned officially Christian before Constantine, in an era when the only Christian states were a few Syriac principalities such as Edessa. One would therefore expect the earliest Armenian versions to show strong signs of Syriac influence.

The signs of Syriac influence exist (among them, manuscripts with 3 Corinthians and without Philemon) -- but so do signs of Greek influence. In addition, the text of the Armenian matches neither the extant Old Syriac nor the Peshitta. It appears to be much more closely linked with the "Cæsarean" text. In fact, the Armenian is arguably the best witness to that text.

The history of the Armenian version is closely tied in with the history of the written Armenian language. After perhaps an unsuccessful attempt by a cleric named Daniel, the Armenian alphabet is reported to have been created by Mesrop, the friend and co-worker of the Armenian church leader Sahak. The year is reported to have been 406, and the impetus for the invention is said to have been the need for a way to record the Armenian Bible. Said translation was finished in the dozen or so years after Mesrop began his work.

Despite Alexanian, the basis of the version remains in dispute. Good scholars have argued both for Syriac and for Greek. There are passages where the wording seems to argue for a Syriac original -- but others that argue equally forceably for a Greek base.

A portion of one column of the famous Armenian MS. Matenadaran 2374 (formerly Etchmiadzin 229), dated 989 C.E. Mark 16:8-9 are shown. The famous reference to the presbyter Arist(i)on is highlighted in red.

  At least three explanations are possible for this. One is that the Armenian was translated from the Greek, but that the translator was intimately familiar with a Syriac rendering. An alternate proposal is that the Armenian was translated in several stages. The earliest stage was probably a translation from one or another Old Syriac versions, or perhaps from the Syriac Diatessaron. This was then revised toward the Greek, perhaps from a "Cæsarean" witness. Further revisions may have increased the number of Byzantine readings. Finally, there may have been two separate translations (Conybeare suggests that Mesrop translated from the Greek and Sahak from the Syriac) which were eventually combined.

The Armenian "Majority Text" has been credited to Nerses of Lambron, who revised the Apocalypse, and perhaps the entire version, on the basis of the Greek in the twelfth century. This late text, however, has little value; it is noticeably more Byzantine than the early text. Fortunately, the earliest Armenian manuscripts are much older than this; a number date from the ninth century. The oldest dated manuscript comes from 887 C.E. (One manuscript claims a date of 602 C.E., but this is believed to be a forgery.)

There are a few places where the Armenian renders the Greek rather freely (usually to bring out the sense more clearly); these have been compared to the Targums, and might possibly be evidence of Syriac influence.

The link between the Armenian and the "Cæsarean" text was noticed early in the history of that type; Streeter commented on it, and even Blake (who thought the Armenian to be predominantly Byzantine) believed that it derived from a "Cæsarean" form. The existence of the "Cæsarean" text is now considered questionable, but there is no doubt that the Armenian testifies to a text which is far removed from the Byzantine, and that it contains large numbers of Alexandrian readings as well as quite a number associated with the "Western" witnesses. The earliest witnesses generally either omit "Mark 16:9-20" or have some sort of indication that it is doubtful (the manuscript shown above may credit it to the presbyter Arist(i)on, though this remark is possibly from a later hand). "John 7:53-8:11" is also absent from most early copies.

In the Acts and Epistles, the Armenian continues to display a text which is not Byzantine but not purely Alexandrian either. Yet -- in Paul at least -- it is not "Western." Nor does it agree with family 1739, nor with H, both of which have been labelled (probably falsely) "Cæsarean." If the Armenian has any affinity in Paul at all, it is with family 2127 -- a late Alexandrian group with some degree of mixture. This is not really surprising, since one of the leading witnesses to the family is 256, a Greek/Armenian diglot (in fact, the Armenian text of 256 is one of the earliest witnesses to the Armenian Epistles).

Lyonnet felt that the Armenian text of the Catholic Epistles fell close to Vaticanus. In the Apocalypse, Conybeare saw an affinity to the Latin (in fact, he argued that it had been translated from the Latin and then revised -- as many as five times! -- from the Greek. This is probably needlessly complex, but the Latin ties are interesting. Jean Valentin offers the speculation that the Latin influence comes from the Crusades, when the Armenians and the Franks were in frequent contact and alliance.)

The primary edition of the Armenian, that of Zohrab, is based mostly on relatively recent manuscripts and is not really a critical edition (although some variant readings are found in the margin, their support is not listed). Until a better edition of the version becomes available -- an urgent need, given the quality of the translation -- the text of the version must be used with caution.


The language of Egypt endured for at least 3500 years before the Islamic conquest swept it aside in favour of Arabic. During that time it naturally underwent significant evolution.

There was at one time much debate over the origin of the Egyptian language; was it Semitic or not? It seemed to have Semitic influence, but not enough to really be part of the family. This seems now to have been solved; Joseph H. Greenburg in the 1960s proposed to group most of the languages of northern Africa and the Middle East in one great "Afroasiatic" superfamily. Egyptian and the Semitiic languages were two of the families within this greater group. Thus Egyptian is related to the Semitic languages, but at a rather large distance.

Coptic is the final stage of the evolution of Egyptian (the words "Copt" and "Coptic" are much-distorted versions of the name "Aigypt[os]"). Although there is no clear linguistic divide between Late Egyptian and Coptic, there is something of a literary one: Coptic is Egyptian written in an alphabet based on the Greek. It is widely stated that the Coptic alphabet (consisting of the twenty-four Greek letters plus seven letters -- give or take a few -- adopted from the Demotic) was developed because the old Egyptian Demotic alphabet was too strongly associated with paganism. This seems not to be true, however; the earliest surviving documents in the Coptic alphabet seem to have been magical texts. CopticLetters

It is at least reasonable to suppose that the Coptic alphabet was adopted because it was an alphabet -- the hieroglyphic, hieratic, and demotic styles of Egyptian are all syllabic systems with ideographic elements. And both hieratic and demotic have other problems: Hieratic is difficult to write, and demotic, while much easier to copy, is difficult to read. And neither represents vowels accurately. Some scribe, wanting a true alphabetic script, took over the Greek alphabet, adding a few demotic symbols to supply additional sounds.

Coptic finally settled down to use the 24 Greek letters plus six or seven demotic symbols. It was some time before this standard was achieved, however; early texts often use more than these few extra signs. This clearly reveals a period of experimentation.

Coptic is not a unified language; many dialects (Akhmimic, Bohairic, Fayyumic, Middle Egyptian, Sahidic) are known. The fragmentation of Coptic is probably the result of the policies of Egypt's rulers: The Romans imposed harsh controls on travel in and out of, and presumably within, Egypt; before them, the Ptolemies has rigidly regimented their subjects' lives and travels. After a few hundred years of that, it is hardly surprising that the Egyptian language ceased to be unified.

New Testament translations have been found in all five of the dialects listed; in several instances there seem to have been multiple translations. The two most important, however, are clearly Sahidic (the language of Upper Egypt) and Bohairic (used in the Lower Egyptian Delta). Where the other versions exist only in a handful of manuscripts, the Sahidic endures in dozens and the Bohairic in hundreds. The Bohairic remains the official version of the Coptic church to this day, although the language is essentially extinct in ordinary life.

The history of the Coptic versions has been separated into four stages by Wisse (modifying Kasser). For convenience, these stages are listed below, although I am not sure of their validity.

  1. The Pre-Classical Stage, 250-350 C.E. First attempts at translation, which had little influence on the later versions.
  2. The Classical Sahidic and Fayyumic Stage, 350-450 C.E. Preparation of versions for use by those who had no Greek. The Sahidic becomes the dominant version. Other versions, notably the Fayyumic, circulate but are not widespread.
  3. The Final Sahidic and Fayyumic Stage, 450-1000 C.E. The Arab conquest reduces the role and power of the Coptic church. The Sahidic begins to decline.
  4. The Bohairic Stage, after 800 C.E. The Bohairic version becomes standardized and gradually achieves dominance within the Coptic church.

A more detailed study of the various versions follows.

The Sahidic Coptic

The Sahidic is probably the earliest of the translations, and also has the greatest textual value. It came into existence no later than the third century, since a copy of 1 Peter exists in a manuscript from about the end of that century. Unlike the Bohairic version, there is little evidence of progressive revision. The manuscripts do not always agree, but they do not show the sort of process seen in the Bohairic Version.

Like all the Coptic versions, the Sahidic has an Egyptian sort of text. In the Gospels it is clearly Alexandrian, although it is sometimes considered to have "Western" variants, especially in John. (There are, in fact, occasional "Western" readings in the manuscripts, but no pattern of Western influence. Most of the so-called "Western" variants also have Alexandrian support.) As between B and Aleph, the Sahidic is clearly closer to the former -- and if anything even closer to P75. It is also close to T (a close ally of P75/B) -- as indeed one would expect, since T is a Greek/Sahidic diglot.

In Acts, the Sahidic is again regarded as basically Alexandrian, though with some minor readings associated with the "Western" text. In the "Apostolic Decree" (Acts 15:19f., etc.) it conflates the Alexandrian and "Western" forms. (One should note, however, the existence of the codex known as Berlin P. 15926. Although its language is to be Sahidic, its text differs very strongly from the common Sahidic version, and preserves a number of striking "Western" variants found also in the Middle Egyptian text G67.)

In Paul the situation is slightly different. Here again at first glance the Sahidic might seem Alexandrian with a "Western" tinge. On examination, however, it proves to be very strongly associated with B, and also somewhat associated with B's ally P46. I have argued elsewhere that P46/B form their own text-type in Paul. The Sahidic clearly goes with this type, although perhaps with some influence from the "mainstream" Alexandrian text.

In the Catholics, the Sahidic seems to have a rather generic Alexandrian text, being about equidistant from all the other witnesses. It is noteworthy that its more unusual readings are often shared with B.

The Bohairic Coptic

The Bohairic has perhaps the most complicated textual history of any of the Coptic versions. The oldest known manuscript, Papyrus Bodmer III, contains a text of the Gospel of John copied in the fourth (or perhaps fifth) century. This version is distinctly different from the later Coptic versions, however; the underlying text is distinct, the translation is different -- and even the form of the language is not quite the same as in the later Bohairic version. For this reason it has become common to refer to this early Bohairic version as the "proto-Bohairic" (pbo).From the same era comes a fragment of Philippians which may be a Sahidic text partly conformed to the idiom of Bohairic.

Other than these two minor manuscripts, our Bohairic texts all date from the ninth century or later. It is suspected that the common Bohairic translation was made in the seventh or eighth century.

It is quite possible that this version was revised, however; there are a number of places where the Bohairic manuscripts split into two groups. Where this happens, it is fairly common to find the older texts having a reading typical of the earlier Alexandrian witnesses while the more recent manuscripts often display a reading characteristic of more recent Alexandrian documents or of the Byzantine text. One can only suspect that these late readings were introduced by a systematic revision.

As already hinted, the text of the Bohairic Coptic is Alexandrian. Within its text-type, however, it tends to go with Aleph rather than B. This is most notable in Paul (where, of course, Aleph and B are most distinct). Zuntz thought that the Bohairic was a "proto-Alexandrian" witness (i.e. that it belonged with P46 B sa), but in fact it is one of Aleph's closest allies here -- despite hints of Sahidic influence, which are found in the other sections of the New Testament as well. One might theorize that the Bohairic was translated from the Greek (a manuscript with a late Alexandrian text), but with at least some Sahidic fragments used as cribs.

The Lesser Coptic Versions

The Akhmimic (Achmimic). Possibly the most fragmentary of all the versions. Fragments preserve portions of Matthew 9, Luke 12-13, 17-18, Gal. 5-6, James 5. All of these seem to be from the fourth or perhaps fifth centuries. Given their small size, very little is known of the text of the Akhmimic. Aland cites it under the symbol ac.

Related to the Akhmimic, and regarded as falling between it and the Middle Egyptian, is the Sub-Akhmimic. This exists primarily in a manuscript of John, containing portions of John 2:12-20:20 and believed to date from the fourth century. It seems to be Alexandrian, and is cited under the symbol ac2 or ach2.

The Fayyumic. Spelled Fayumic by some. Many manuscripts exist for the Gospels, and over a dozen for Paul, but almost all are fragmentary. Manuscripts of Acts and the Catholic Epistles are rare; the Apocalypse seems to be entirely lost (if, indeed, it was ever translated). Manuscripts date from about the fifth to the ninth centuries. There is also a fragment of John, from perhaps the early fourth century, which Kahle called Middle Egyptian but Husselman called Fayyumic. This mixed text is now designated the "Middle Egyptian Fayyumic (mf)" by Aland. (The Fayyumic is not cited in NA27; the abbreviation fay is used in UBS4.)

Given the fragmentary state of the Fayyumic, its text has not been given much attention. In Acts it is reported to be dependent on the Bohairic, and hence to be Alexandrian. Kahle found that an early manuscript which contained both the long and short endings of Mark.

The Middle Egyptian. The Middle Egyptian Coptic is represented primarily by three manuscripts -- one of Matthew (complete; fourth/fifth century), one of Acts (1:1-15:3; fourth century), and one of Paul (54 leaves of about 150 in the original; fifth century). The Acts manuscript, commonly cited as copG67, is perhaps the most notable, as it agrees frequently with the "Western" witnesses, including some of the more extravagant variants of the type. The Middle Egyptian is cited by Aland under the symbol mae; UBS4 uses meg.


Although the origins of many of the versions are obscure, few are as obscure as those of the Ethiopic. The legend that Christianity was carried to the land south of Egypt by the eunuch of Acts 8:26f. can be easily dismissed. So can accounts that one of the apostles worked there. Even if one or more of these stories were true, they would not explain the existence of the Ethiopic version. (The New Testament hadn't even been written at the time of the Ethiopian's conversion in Acts.)

Even the name of the version is questionable; the correct name for the official language of Ethiopia is Amharic, and the manuscripts of the "Ethiopic" version are in an old form of this language.

A legend told by Rufinus has it that Christianity reached Ethipia to stay in the fourth century. Although this is beyond verification, there are indications that Christianity did indeed reach the country at that time.

Unlike many of the languages into which the Bible was translated, Ethiopia already had developed writing at the time Christianity reached the country (the alphabet resembles the Semitic in that it uses letters for consonants and lesser symbols for vowels; however, the letter forms diverge widely from the Phoenician, and the language reads from left to right. It has been theorized that the Ethiopic alphabet is actually derived from the Old Hebrew alphabet, abandoned by the Jews themselves in the post-Exilic period. The modern "Hebrew" alphabet is actually Aramaic. Ethiopic, however, added vowel symbols at a very early date -- not as extra letters but as tags attached to letters -- in effect, a syllabary. This is further evidence of Semitic origin -- and, probably, of the absence of Greek influence).

Because written Ethiopic predates the New Testament, we cannot date the version based on the dates of the earliest written documents. Nor are the dates of the earliest manuscripts much help, since all Ethiopic manuscripts are of the eleventh century or later and the vast majority are of the fourteenth century or later. Nor did printing immediately affect the version; manuscripts continued to be copied into the seventeenth century and even beyond. Perhaps the most common theory is that the version dates from about the fifth century, when Christianity probably became widespread in Ethiopia.

It is not clear what language formed the translation base for the Ethiopic version, although Greek and Coptic are the leading candidates (the Apocalypse, in particular, contains a number of transliterations from Greek) It is possible that both were used in different books. Syriac and Arabic have also been mentioned (the version bears significant orthographic similarities to those languages), and revisions based on the latter cannot be ruled out. On the other hand, Ethiopic is not Indo-European, so many of the noteworthy features of Greek (e.g. noun declensions, word order, and many verb forms) cannot be rendered. Hints of Syriac or Arabic influence on the version may simply be because Ethiopic is closer to those languages. The problem is not simplified by the fact that the language is not well-known to scholars and the version has not been properly edited. In addition, it appears likely that different translators worked on different books (since the style ranges from the free to the stiltedly literal); it is possible that different base texts were used. It is worth noting that the Ethiopic Bible includes several works not normally considered canonical.

Based on the available information, it would appear that the Ethiopic has an Alexandrian text -- but an uncontrolled, with very primitive Alexandrian readings alternating with primarily Byzantine readings and some variants that are simply wild. Zuurmond calls it "Early Byzantine" in the Gospels, and also notes an "extreme tendency toward harmonizations." Hoskier noted that Eth had a number of unusual agreements with P46 in Paul, but undertook no detailed study. It may be that the Ethiopic is based on the sort of free text that seems to have prevailed in Egypt in the early years of Christianity: Basically similar to the Alexandrian text, with a number of very primitive readings (the latter often rather rough), but with some wild readings, others characteristic of the later text, and a number of readings that resulted simply from scribal inattentiveness. The lack of a detailed study prevents us from saying more.


If any version is most notable for our ignorance about its origin, it is the Georgian. The language is difficult and not widely know (it is neither Indo-European nor Semitic; the alphabet, known as Mkhedruli, is used only for this language. Georgian is the only language of the Kartvelian group to have a written form), the country small, and the history of the translation is obscure. Whatever its origins, however, the version is of great textual significance.

Please note that plain HTML and pure ASCII, in which this document is written, has no facilities even for transliterated Georgian; I've done my best with the technical terms, but you really need to visit a specialized site to see the correct forms of the letters.

Legend has it that the evangelist of the Georgians, a woman named Nino, came to Georgia as a slave during the reign of the Roman Emperor Constantine. Another legend has it that the Georgian alphabet was invented by Saint Mesrop some time after he had created the written form of Armenian.

Sinai Georgian 31, dated 877, folio 54 verso, Acts 8:24-29.
Thanks to Jean Valentin

Both of these legends may be questioned -- the former on historical grounds, the latter on the basis of its simple improbability. It is by no means certain that the Georgian alphabet was invented to receive a Biblical translation (if it had been, why is it so different from other alphabets?); the Georgian alphabet may well be older than the fifth century.

Given our ignorance of the history of Christianity in Georgia, we can only speculate about the history of the version. The latest possible date would appear to be the sixth century, since our earliest manuscripts (the "han-met'i fragments") are dated linguistically to that era, or perhaps even to the fifth century. The most likely date for the version is therefore the fifth century. This is supported by an account of the life of St. Shushanik, dated to the fifth century and containing many allusions to the Biblical text.

By its nature it is difficult for Georgian to express many features of Greek syntax. This makes it difficult to determine the linguistic source of the version. (Nor does it help that the language itself has evolved; the translation started in Old Georgian, but New Georgian came into existence from the twelfth century, and later manuscripts will have been influenced by the new dialect.) Greek, Armenian, and Syriac have all been proposed -- in some instances even by the same scholar! It seems clear that the version was at some time in its history revised toward the Greek -- but since manuscripts of the unrevised text are at once rather few and divergent, we probably cannot reach a certain conclusion regarding the source at this time. The current opinion seems to be that, except in the Apocalypse (clearly taken from the Greek), the base text -- what we might call the "Old Georgian," and now found primarily in geo1 and some of the fragments -- was Armenian, and that it was progressively modified by comparison with the Greek text.

The earliest Georgian manuscripts are the already alluded to han-met'i fragments of the sixth and seventh centuries, followed by the hae-met'i fragments of the next century. (The names derive from linguistic features of the Georgian which were falling into disuetitude.) These fragments are, unfortunately, so slight that (with the exception listed below) they are of little use in reconstructing the text (some 45 manuscripts contain, between them, fragments of the Gospels, Romans, and Galatians only). Recently a new han-met'i palimpsest was discovered and published, containing large portions of the Gospels, but the details of its text are not yet known; it appears broadly to go with the Adysh manuscript (geo1).

With the ninth century, fortunately, we begin to possess fuller manuscripts, of good textual quality, from which we may attempt to reconstruct the "Old Georgian" text. Many of these manuscripts, happily, are dated.

The earliest substantially complete Georgian text is the Adysh manuscript, a copy of the Gospels dating from 897 C.E. It appears to have the most primitive of all Georgian translations, and is commonly designated geo1.

From the next century come the Opiza Gospels (913), the Dzruc Gospels (936), the Parhal Gospels (973), the Tbet' Gospels (995), the Athos Praxapostols (between 959 and 969), and the Kranim Apocalypse (978), as well as assorted not-so-well-known texts. Several of these manuscripts combine to represent a second stage of the Georgian version, designated geo2. When cited separately, the Opiza gospels are geoA, the Tbet' gospels are geoB. (The Parhal Gospels are sometimes cited as geoC, but this is not as common.)

Starting in the tenth century, the Georgian version was revised, most notably by Saint Euthymius of Athos (died 1028). Unfortunately, the resulting version, while perhaps improved in form and literary merit, is less interesting textually; the changes are generally in conformity with the Byzantine text.

The text of the Georgian version, in the Gospels, is clearly "Cæsarean" (assuming, of course, that text-type exists). Indeed, the Georgian appears to be, along with the Armenian, the purest surviving monument of that text-type. Both geo1 and geo2 preserve many readings of the type, though not always the same readings. Blake thought that geo1 affiliated with Q 565 700 and geo2 with families 1 and 13.

In Acts, Birdsall links the Old Georgian to the later forms of the Alexandrian text found in minuscules such as 81 and 1175. In Paul, he notes a connection with P46, although this exists in scattered readings rather than as an overall affinity. In the Apocalypse, the text is that of the Andreas commentary.


Of all the versions regularly cited in critical apparati, the Gothic is probably the least known. This is not because it is ignored. It is because it has almost ceased to exist.

The Gothic version was apparently entirely the work of Ulfilas (Wulfilas), the Apostle to the Goths. Appointed Bishop to the Goths around 341, he spent the next forty years evangelizing and making the gospel available to his people. In the process he created the Gothic alphabet. The picture shows that it was based on Greek and Latin models, but also included some symbols from the Gothic runic alphabets.

Ulfilas translated both Old and New Testaments, from the Greek (reportedly excepting the book of Kings, because it was too militant for his flock), but only fragments of the New Testament survive. (At that, they are the almost only literary remains of Gothic, a language which is long since dead.)

The gospels are preserved primarily in the Codex Argenteus of the sixth century. Even this manuscript has lost nearly half its pages, but enough have survived to tell us that the books are in the "Western" order (Matthew, John, Luke, Mark), and that the manuscript included Mark 16:9-20 but omitted John 7:53-8:11. The image of the manuscript at right demonstrates this; the page contains John 7:52, 8:12-17.

Other than the Argenteus, all that has come to light of the gospels are a small portion of Matthew from a palimpsest and a few fragmentary verses of the Luke on a Gothic/Latin leaf destroyed during the Second World War.

According to Metzger, nothing has survived of the Acts, Catholic Epistles, and Apocalypse. Vincent Broman tells me that the Old Testament is almost all lost, though there is a fragment of Nehemiah large enough to indicate a Lucianic ancestor. Of Paul there are several manuscripts, all fragmentary and all palimpsest. Only 2 Corinthians is complete, and Hebrews is entirely lacking. (It has been speculated that Ulfilas, for theological or other reasons, did not translate Hebrews, but Broman informs me that it has been quoted in a commentary.)

Ulfilas's version is considered literal (critics have called it "severely" literal, preserving Greek word order whether it fits Gothic or not). It is very careful in translation, striving to always use the same Gothic word for each Greek word. Even so, Gothic is a Germanic language, and so cannot distinguish many variations in the Greek (e.g. of verb tense; some word order variations are also impermissible). It is also possible, though by no means certain, that Ulfilas (who was an Arian preaching to Arians) allowed some slight theological bias to creep into his translation.

In the Gospels, the basic run of the text is very strongly Byzantine, although von Soden was not able to determine what subgroup it belongs with. Burkitt found a number of readings which the Gothic shared with the Old Latin f (10), though scholars are not agreed on the significance of this. Some believe that the Old Latin influenced the Gothic; others believe the influence went the other way. Our best hint may come from Paul. Here the Gothic is again Byzantine, but less so, and it has a number of striking agreements with the "Western" witnesses. It has been theorized that Ulfilas worked with a Byzantine Greek text, but also made reference to an Old Latin version. Presumably this version was either more "Western" in the Epistles, or (perhaps more likely) Ulfilas made more reference to it there.

It is much to be regretted that the Gothic has not been better preserved. While the Gospels text is not particularly useful, a complete copy of the Epistles might prove most informative. And it is, along with the Peshitta, one of the earliest Byzantine witnesses; it might provide interesting insights into the Byzantine text.

The handful of survivals are also of keen interest to linguists, as the Gothic is the earliest known member of the Germanic family of languages, predating the earliest Old English texts by a couple of centuries; it is also of significance as the only attested East Germanic language (the Germanic group is thought to have three families: The West Germanic, which includes all languages now called "German," plus English, Dutch, Frisian, and Yiddish; the North Germanic, which gave rise to Icelandig, Faroese, Swedish, Danish, and Norwegian, which are still mostly mutually intelligible and amount to hardly more than a single source; and the East Germanic, which consists solely of Gothic). Thus the Gothic is very important in reconstructing proto-Germanic -- and, indeed, Indo-European.


Of all the versions, none has as complicated a history as the Latin. There are many reasons for this, the foremost being its widespread use. The Latin Vulgate was, for millennia, the Bible of the western church, and after the fall of Constantinople it was the preeminent Bible of Christendom. There are at least eight thousand Latin Bible manuscripts known -- or at least two thousand more Latin than Greek manuscripts.

The first reference to what appears to be a Latin version dates from 180 C.E. In the Acts of the Scillitan Martyrs, one of the men on trial admits to having writings of Paul in his possession. Given the background, it is presumed that these were in a Latin version.

But which Latin version? That is indeed the problem -- for, in the period before the Vulgate, there were dozens, perhaps hundreds. Jerome, in his preface to the Vulgate gospels, commented that there were "as many [translations] as there are manuscripts." Augustine complained that anyone who had the slightest hint of Greek and Latin might undertake a translation. They seem to have been right; of our dozens of non-Vulgate Latin manuscripts, no two seem to represent exactly the same translation.

Below: An Old Latin manuscript, Codex Sarzensis (j), on purple parchment, much damaged by the gold ink used to write it. Shown in exaggerated color   Modern scholars have christened these pre-Vulgate translations, which generally originated in the second through fourth centuries, the "Old Latin." (These versions are sometimes called the "Itala," but this term is quite properly going out of use. It arose from a statement of Augustine's that the Itala was the best of the Latin versions -- but we no longer know what this statement means or which version(s) it refers to.)
j of the Old Latin

The Old Latins are traditionally broken up into three classes, the African, the European, and the Italian. Even these terms can be misleading, however, as there is no clear dividing line between the European and the Italian; the Italian generally refers to European texts of a more polished type -- and in any case these are groups of translations, not individual translations.

The Old Latin gospels generally, although by no means universally, have the books in the "Western" order (Matthew, John, Luke, Mark) -- an order found also in D and W but otherwise very rare among Greek manuscripts.

The oldest of the types is probably the African; at least, its renderings are the crudest, and Africa was the part of the Roman Empire which had the smallest Greek population and so had the greatest difficulty with a Greek Bible. In the first century, Greek was as common in Rome as was Latin; it was not until several centuries later (as the Empire became more and more divided and Greek-speaking slaves became rarer) that Italy and the west felt the need for a Latin version. Eventually the demand became so great that Pope Damasus authorized the Vulgate.

Traditionally the Old Latin witnesses were designated by a single Roman letter (e.g. a, b, e, k). As Roman letters ran out, longer names (aur) or superscripts (g1) came into use. The Beuron Latin Institute has now officially numbered the Old Latin witnesses (of which about ninety are now known), but the old letter designations are still generally used to prevent confusion with the minuscules.

The tables below show, section by section, the Old Latin witnesses available to the modern scholar. In general only those witnesses found in the NA27 or UBS4 editions are listed, although a handful of others (often Old Latin/Vulgate mixes) have been cataloged. Observant users will observe that this list omits some "Old Latin" witnesses cited in UBS4. Examples include ar c dem in Acts. The reason is that these are actually Vulgate witnesses with occasional Old Latin readings; they will be discussed under the Vulgate.

Old Latin Witnesses -- Gospels

a3IVVercellensise#Seems to be an early form of the European Latin. Closest to b ff2, but perhaps with some slightly older readings. Deluxe manuscript (silver and gold ink on purple parchment), reputed to have been written by Saint Eusebius, Bishop pf Vercelli (martyred 370/1). It has been so venerated as a relic that certain passages have been rendered unreadable by worshippers' kisses. Contains Mark 16:9-20, but on interpolated leaves; C.H. Turner believes the original did not contain these verses. Text is regarded as similar to n in the Synoptic Gospels.
a216VCuriensisLk 11#, 13#cf. n, o (both also #16)
aur15VIIAureuse#Primarily Vulgate but with many Old Latin readings.
Incidentally, combining references from several sources, it appears that this is the oldest surviving parchment manuscript with a separate title page (there seem to have been no others until shortly before the invention of printing).
b4VVeronensise#Purple codex with silver and some gold ink. Originally contained 418 leaves; 393 remain, some of which have decayed to the point of illegibility. Widely regarded as one of the very best European witnesses; almost all other witnesses of the type agree with b more than with each other. A few passages have been conformed to the Vulgate, in writing so like the original that the alterations were not noticed for many centuries.
b26VIICarinthianusLk 1-2#
Colbertinuse(apcr)Late and vulgate influenced, but apparently with some African readings (although European readings dominate; it is much closer to b ff2 than to k). The pre-vulgate readings are most common in Mark and Luke. The rest of the NT, which comes from another source, is Vulgate with scattered Old Latin readings.
Bezaee#a#c#Latin side of Codex Bezae, and almost as controversial as the Greek. It is probably based on an independent Latin version, since D and d disagree at some few points. However, they agree the vast majority of the time, even in places where they have no other Latin support. It is effectively certain that the two texts have been modified to agree more closely. The great question is, which has been modified, and to what extent?
d27IXSangallensise# Latin interlinear of D, with no real value of its own.
e2VPalatinuse#After k, the most important witness to the African Latin. (Unfortunately, the two overlap only very slightly, so it is hard to compare their texts.) Purple codex.
f10VIBrixianuse#Purple codex. The text seems to fall somewhere between the (European) Old Latin and the vulgate, and it has been conjectured that it was the sort of manuscript Jerome made his revision from. However, it has links to the Gothic (it has been conjectured that it was taken from the Latin side of a Gothic-Latin diglot), which make this less likely. It is distinctly more Byzantine and less "Western" than the average Old Latin. It is considered to be an Italian text.
ff19VIIICorbiensisMtVulgate with some Old Latin readings.
ff28VCorbiensise#European Latin, probably the best text of the type after b.
SangermanensisMt(NT)Old Latin in Matthew; rest is Vulgate (see Vulgate G)
h12VClaromontanusMt#(e)Old Latin in Matthew; rest is Vulgate.
i17V/VIVindobonensisMk#Lk#Purple codex.
j22VISarzanensis(Lk#)Jo#Purple codex. Text is described as "peculiar and valuable."
BobiensisMt#Mk#Best codex of the African Latin, unfortunately only about half complete even for the books it contains (it now consist of portions of Matt. 1:1-15:36 plus Mark 8:8-end). Noteworthy for containing only the short ending of Mark (without the long ending); it is the only known manuscript to have this form. Written in a good hand by a careless scribe -- quite possibly a non-Christian. The text seems to resemble Cyprian.
l11VIIIRehdigeranuse#"Mixed text."
Lk 16-17#
m-VMonacensisMt 9-10# The symbol m is sometimes used for the Codex Mull (35 -- e/vii), which is probably an Old Latin heavily corrected toward the Vulgate.
n16VSangallensisMt#Mk#Jo#Cf. a2, o (both also #16)
o16VIISangallensisMk#Mark 16:14-20. Cf. a2, n (both also #16).
p20VIIISangallensisJo 11#
Monacensise#Considered to have an Italian text, though perhaps with a slightly different textual base. Written in a clumsy hand by a scribe named Valerianus.
AmbrosianusJo 13#
AmbrosianusLk 17-21#
BernensiaMk 1-3#
v25VIIVindobonensisJo 19-20#

Old Latin Witnesses -- Acts

Bezaee#a#c#Latin side of Bezae (D). See comments in the section on the Gospels.
e50VILaudianusa#Latin side of Laudianus (E). The base text is considered to be European, but there is also assimilation to the parallel Greek.
gSymbol used in some editions for gig.
gig51XIIIGigas(e)a(pc)rAn immense codex containing the Bible and a number of other works. Its text in Acts is reminiscent of that of Lucifer of Cagliari, but experts cannot agree whether it belongs with the African or European Latin.
h55VFloriacensisa#c#r#Fleury palimpsest. The translation is loose and the copy careless, but the text is very close to that used by Cyprian (African).
l67VIILegionensisa#c#Palimpsest; text is vulgate with some sections of Old Latin readings (Acts 8:27-11:13, 15:6-12, 26-38). Said to be close to the Liber Comicus (t)
(m)-IV?(Speculum)eapcrSee Speculum under Fathers
p54XIIPerpinianusaOld Latin in 1:1-13:6, 28:16-end. The text is said to be similar to the fourth century writer Gregory of Elvira, and is thought to have been written in northern Spain or southern France.
ph63XIIaActs with "other material."
ro62XRodensis(e)a(pcr)Vulgate text with Old Latin readings in both text and margin in Acts.
sa60XIIIBoverianusa#Contains Acts 1:15-26.
tVII+Liber Comicusa#p#c#r#Lectionary
Wernigerodensis(e)a(p)c(r)Vulgate with Old Latin readings in Acts & Catholics.

Old Latin Witnesses -- Paul

Note: Scholars generally do not distinguish between African, European, and Italian texts in Paul (although I have seen r called both African and Italian). The reason seems to be that we have no unequivocally African texts.
(Book of Armagh)
(ea)p#(c)rGeneral run of the text is vulgate text with many Old Latin readings, but Paul (vac. 1 Cor. 14:36-39) and the Apocalypse are Old Latin with some Vulgate influence. See D of the Vulgate for full information on the history and style of this noteworthy manuscript.
pClose to d, and possibly the best Latin witness available in Paul. Most other "Western" witnesses are closer to b d than to each other.
d75VIClaromontanusp#Latin side of D. Unlike most bilinguals, the Latin and the Greek do not appear to have been conformed to each other; d seems to fall closest to b.
f78IXAugiensisp#Latin side of F. Mixed Vulgate and Old Latin (Hebrews is purely Vulgate), possibly with some assimilation to the Greek text.
g77IXBoernianusp#Latin interlinear of G. Rarely departs from the Greek text except where it offers alternate renderings.
gue79VIGuelferbytanusRom#Palimpsest, from the same manuscript as Pe Q. Contains Rom. 11:33-12:5, 12:17-13:1, 14:9-20. Merk's w.
(m)-IV?(Speculum)eapcrSee Speculum under Fathers. Not to be confused with m/mon (below)
m86Xp#The appendix of NA27 lists this as mon (the latter symbol is used in UBS), but cites it in the text as m. Not to be confused with the Codex Speculum, often cited as m. The text is said to be similar to that of Ambrose; it is noteworthy for placing the doxology of Romans after chapter 14 (so also gue; neither ms. exists for Romans 16).
monSymbol used for m in UBS4.
m82IXMonacensisHeb 7, 10#Contains Heb. 7:8-26, 10:23-39
p80VIIHeidelbergensiaRom 5-6#
Frisingensiap#Assorted small fragments, sometimes denoted r1, r2, r3. They do not come from the same manuscript, but seem to have similar texts. They have a much more Alexandrian cast than the other Old Latins, and are said to agree with Augustine. Same as q/r of the Catholics.
s87VIIIp#Lectionary fragments.
tVII+Liber Comicusa#p#c#r#Lectionary
wSymbol used in some editions for gue.
z65VIIIHarleianus(Heb#)Vulgate Bible (same codex as Z/harl); only Heb. 10:1-end is Old Latin.

Old Latin Witnesses -- Catholics

Bezaee#a#c#Latin side of D (Bezae). Greek does not exist for the Catholics, and of the Latin we have only 3 John 11-15.
ff66IXCorbeiensisJamesSouter describes it having "some readings unique (almost freakish) in their character...." Overall, it seems to have a mixed text, not affiliated with anything in particular.
h55VFloriacensisa#c#r#Fleury palimpsest. Contains 1 Pet. 4:17-2 Pet 2:7, 1 John 1:8-3:20. The translation is loose and the copy careless, but the text is very close to that used by Cyprian (African).
l67VIILegionensisa#c#Palimpsest; small sections exist of all books of the Catholics except Jude. Said to be close to the Liber Comicus (t)
(m)-IV?(Speculum)eapcrSee Speculum under Fathers
qSymbol used for r in UBS4.
Monacensisc#Same as r of Paul. Denoted q in UBS4.
s53VIBobiensisc#Palimpsest. Old Latin in 1 Pet. 1:1-18, 2:4-10
tVII+Liber Comicusa#p#c#r#Lectionary
w32VIGuelferbitanusc#Palimpsest lectionary, Vulgate with sections in Old Latin.
z65VIIIHarleianus(c#)Vulgate Bible (same codex as Z/harl); only 1 Pet. 2:9-4:15, 1 John 1:1-3:15 are Old Latin.

Old Latin Witnesses -- Revelation

(Book of Armagh)
(ea)p#(c)rVulgate text with many Old Latin readings; Paul and the Apocalypse are Old Latin with some Vulgate influence. See D of the Vulgate.
gSymbol used in some editions for gig.
gig51XIIIGigas(e)a(pc)rAn immense codex containing the Bible and a number of other works. Its text in the Apocalypse is Old Latin but seems to be a late form of the European type, approaching the Vulgate.
h55VFloriacensisa#c#r#Fleury palimpsest. The translation is loose and the copy careless, but the text is very close to that used by Cyprian (African).
sin74Xa#r#Contains Rev. 20:11-21:7.
tVII+Liber Comicusa#p#c#r#Lectionary

When discussing the Old Latin, of course, the great question regards the so-called "Western" text. The standard witnesses to this type are the great bilingual uncials (D/05 D/06 F/010 G/012; E/07 is bilingual but is not particularly "Western" and 629 has some "Western" readings but its Latin side is Vulgate). That there is kinship between the Latins and the "Western" witnesses is undeniable -- but it is also noteworthy that many of the most extravagant readings of Codex Bezae (e.g. its use of Matthew's genealogy of Jesus in Luke 3:23f.; its insertion of Mark 1:45f. after Luke 5:14) have no Latin support except d. Even the "Western Non-interpolations" at the end of Luke rarely command more than a bare majority of the Old Latins (usually a b e r1; occasionally ff2; rarely aur c f q).

It is the author's opinion that the Old Latins, not Codex Bezae, should be treated as the basis of the "Western" text, as they are more numerous and show fewer signs of editorial action. But this discussion properly belongs in the article on text-types.

Some Latin witnesses

Three Latin versions. Left: The final page of k (Codex Bobiensis), showing the "shorter ending" of Mark. Middle: Portion of one column of Codex Amiatinus (A or am). Shown are Luke 5:1-3. Right: The famous and fabulously decorated Book of Kells (Wordsworth's Q). The lower portion of the page is shown, with the beginning of Luke's genealogy of Jesus (Luke 3:23-26).

The Vulgate

As the tables above show, the number of Old Latin translations was very large. And the quality was very low. What is more, they were a diverse lot; it must have been hard to preach when one didn't even know what the week's scripture said!

It was in 382 that Pope Damasus (366-384) called upon Jerome (Sophronius Eusebius Hieronymus) to remedy the situation. Jerome was the greatest scholar of his generation, and the Pope asked him to make an official Latin version -- both to remedy the poor quality of the existing translations and to give one standard reference for future copies. Damasus also called upon Jerome to use the best possible Greek texts -- even while giving him the contradictory command to stay as close to the existing versions as possible.

Jerome agreed to take on the project, somewhat reluctantly, but he never truly finished his work. By about 384, he had prepared a revision of the Gospels, which simultaneously improved their Latin and reduced the number of "Western" readings. But if he ever worked on the rest of the New Testament, his revisions were very hasty. The Vulgate of the Acts and Epistles is not far from the Old Latin. Jerome had become fascinated with Hebrew, and spent the rest of his translational life working on the Latin Old Testament.

Even so, the Vulgate eventually became the official Bible of the Catholic Church -- and, despite numerous errors in the process of transmission, it remained recognizably Jerome's work. Although many greeted the new version with horror, its clear superiority eventually swept the Old Latins from the field.

Vulgate criticism is a field in itself, and -- considering that it was for long the official version of the Catholic church -- a very large one. Sadly, the official promulgation of the Sixtine Vulgate in 1590 (soon replaced by the Clementine Vulgate of 1592) meant that attempts to reconstruct the original form of the version were hampered; there is still a great deal which must be done to use the version to full advantage.

Scholars cannot even agree on the text-type of the original Vulgate. In the gospels, some have called it Alexandrian and some Byzantine. In fact it has readings of both types, as well as a number of "Western" readings which are probably holdovers from the Old Latin. The strongest single strand, however, seems to be Byzantine; in 870 test passages, I found it to agree with the Byzantine manuscripts 60-70% of the time and with Aleph and B only about 45% of the time.

The situation is somewhat clearer in the Epistles; the Byzantine element is reduced and the "Western" is increased. Still, it should be noted that the Vulgate Epistles are much more Alexandrian than the Old Latin versions of the same books.

In the Apocalypse the Vulgate preserves a very good text, closer to A and C than to any of the other groups.

These comments apply, of course, to the old forms of the Vulgate, as found, e.g., in the Wordsworth-White edition. The later forms, such as the Clementine Vulgate, were somewhat more Byzantine, and have more readings which do not occur in any Greek manuscripts.

With that firmly in mind, let us turn to the various types of Vulgate text which evolved over the centuries. As with the Greek manuscripts, the various parts of Christendom developed their own "local" text.

The best "local" text is considered to be the Italian type, as represented e.g. by am and ful. This text also endured for a long time in England (indeed, Wordsworth and White call this group "Northumbrian"). It has formed the basis for most recent Vulgate revisions.

Believed to be as old as the Italian, but less reputable, is the Spanish text-type, represented by cav and tol. Jerome himself is said to have supervised the work of the first Spanish scribes to copy the Vulgate (398), but by the time of our earliest manuscripts the type had developed many peculiarities (some of them perhaps under the influence of the Priscillians, who for instance produced the "three heavenly witnesses" text of 1 John 5:7-8).

The Irish text is marked by beautiful manuscripts (the Book of Kells and the Lichfield Gospels, both beautiful illuminated manuscripts, are of this type, and even unilliminated manuscripts such as the Rushworth Gospels and the Book of Armagh are beautiful examples of calligraphy). Sadly, these manuscripts are often marred by conflations and inversions of word order. Some of the manuscripts are thought to have been corrected from the Greek -- though the number of Greek scholars in the Celtic church must have been few indeed. Lemuel J. Hopkins-James, editor of The Celtic Gospels (essentially a critical edition of codex Lichfeldensis) offers another theory: that this sort of text (which he calls "Celtic" rather than Irish) is descended not from a pure Vulgate manuscript but from an Old Latin source corrected against a Vulgate. (It should be noted, however, that Hopkins-James tries to use statistical comparisons to support this result, and the best word I can think of for his method is "ludicrous.")

The "French" text has been described as a mixture of Spanish and Irish readings. The text of Gaul (France) has been called "unquestionably" the worst of the local texts.

The wide variety of Vulgate readings in Charlemagne's time caused that monarch to order Alcuin to attempt to create a uniform version (the exact date is unknown, but he was working on it in 800). Unfortunately, Alcuin had no critical sense, and the result was not a particularly good text. Still, his revision was issued in the form of many beautiful codices.

Another scholar who tried to improve the Vulgate was Theodulf, who also undertook his task near the beginning of the ninth century. Some have accused Theodulf of contaminating the French Vulgate with Spanish readings, but it appears that Theodulf really was a better scholar than Alcuin, and produced a better edition than Alcuin's which also included information about the sources of variant readings. Unfortunately, such a revision is hard to copy, and it seems to have degraded and disappeared quickly (though manuscripts such as theo, which are effectively contemporary with the edition, preserve it fairly well).

Other revisions were undertaken in the following centuries, but they really accomplished little; even if someone took notice of the revisors' efforts, the results were not particularly good. When it finally came time to produce an official Vulgate (which the Council of Trent declared an urgent need), the number of texts in circulation was high, but few were of any quality. The result was that the "official" Vulgate editions (the Sixtine of 1590, and its replacement the Clementine of 1592) were very bad. Although good manuscripts such as Amiatinus were consulted, they made little impression on the editors. The Clementine edition shows an amazing ability to combine all the faults of the earlier texts. Unfortunately, it was to be nearly three centuries before John Wordsworth undertook a truly critical edition of the Vulgate, and another century after that before the Catholic Church finally accepted the need for revised texts.

Despite all that has been said, the Vulgate remains an important version for criticism, and both its "true" text and the variants can help us understand the history of the text. We need merely keep in mind the personalities of our witnesses. The table below is intended to help with that task as much as possible.

Note that there is no official list -- let alone set of symbols -- for Vulgate manuscripts. Single letters are used by Merk and by Wordsworth/White; the symbols such as am and ful are typical of editions of the Greek text such as Tischendorf. All manuscripts cited in these editions are listed. The quoted comments are primarily from Scrivener; the textual descriptions from Metzger and others.

Catalog of Vulgate Manuscripts

AamAmiatinusc. 700OT+NTConsidered to be the best Vulgate manuscript in existence. Copied in England, but with an Italian text. Written in cola et commata, with two columns per page, in a beautiful calligraphic hand. 1209 leaves total. Believed to be the oldest surviving complete Bible in Latin (or, perhaps, any language).
--andSt. Andrew?eFormerly at Avignon, but lost by Scrivener's time.
arsee under D
e#"Probably written in France, but both the text and the calligraphy show traces of Irish influence."
BbamBambergensisIX(e)apc "One of the finest examples of the Alcuinian recension, and a typical specimen of the second period of Caroline writing and ornamentation."
Be or
e"[W]ritten in a fine revived uncial hand" in cola et commata. Berger describes the text as having the sort of mix of Spanish and Irish readings which underlie the French text.
bodlsee under O
CcavCavensisIXTO+NTAlong with tol, the leading representative of the Spanish text. Among the earliest witnesses for the three witnesses in 1 John 5:7-8, which it possesses in modified form. The scribe, named Danila, wrote it with a Visigothic hand.
ccolbColbertinusXII(e)apcrSame as the Old Latin c of the Gospels. Often cited as Old Latin elsewhere, but the text is vulgate. The two sections are separately bound and in different hands. The Vulgate portion of the text is considered to be French.
cantabsee under X
Dar or
(Book of Armagh)
ea(p)c(r)Paul and Revelation are Old Latin (#61, cited as a or ar). Famous Irish codex -- the only (nearly) complete New Testament regarded as being from an Irish source. Also unusual in that we know a good deal about the scribe: It was written by one Ferdomnach. The dating is somewhat uncertain. We know from the Annals of Ulster that Ferdomnach died in 845/6, after a long career. The book is not itself dated, but there are hints (somewhat confusing) in the colophon. Ferdomnach is said to have worked under the direction of abbott Torbach of Armagh, who held that post from 807-808 -- but we also see a reference to Ferdomnach as "the heir of Patrick," i.e. Abbot or Bishop of Armagh), which post he held from 812-813. Thus various scholars have dated the work to 807 or to 812. If we must choose between the two dates, I would incline to 807, since the higher title might have been inserted later. But it is at least possible that the book took four or so years to complete; it is a major production, consisting not just of the New Testament but an introductory section, in Latin and Gaelic, of documents regarding Saint Patrick, followed by the New Testament, and then a life of Saint Martin of Tours. Brian Boru, the most famous early King of Ireland, would later add his name to it.
The hand is a small cursive and has been described as "beautiful," though to me it looks rather crabbed. Like most Irish manuscripts, it is handsomely illustrated with figures of animals and the like incorporated into the initial letters, though the only separate drawings are of the four creatures which represent the evangelists. As it currently stands, it consists of 442 pages, mostly in two columns.
The Vulgate portions reportedly have an Irish text. The Gospels are said to show signs of correction from Family 13. It includes the Epistle to the Laodiceans. Lacks Matthew chapters 14-19.
--dem or
DemidovianusXIIOT+NTLost; our knowledge is based on Matthei's collation (which included only the Acts, Epistles, and Revelation). Appears to have been Vulgate with many Old Latin readings in the Acts and Epistles.
--durmachDurmachensisVI/VIIeBook of Durrow. Illuminated manuscript. Colophon (probably copied from its exemplar) states that it was executed by Saint Columba himself. Reportedly close to Amiatinus. The images in this book are a curious mix; the image of Matthew is said to have Anglo-Saxon and Syriac elements, the Markan lion is Germannic and Pictish; the calf symbolizing Luke is again Pictish. The images are not very clear, though they are surrounded by the beautiful swirls and figures of Celtic art.
Ddunelm DunelmensisVII/
e#Said, probably falsely, to have been written by Bede; it may have come from the Jarrow monastery. Related to Amiatinus.
e#Despite having been discovered in France, the text is considered Irish. Many mutilations, especially in Mark.
--emSt. Emmeram's870e"[W]ritten in golden uncials on fine white vellum, a good deal of purple being employed in the earlier pages; there are splendid illuminations before each gospel."
or EP
ept EpternacensisVIII/IXeFrom Echternach (Luxembourg), but now in Paris. A colophon associates it with Saint Willibrord (or, perhaps, with a manuscript he owned). Irish hand, and the basic run of the text is said by some to be Irish, but with corrections reported to be of another type (perhaps of the Amiatinus type). Further investigation is probably warranted. The colophon claims a date of 558 C.E., but all agree that it must be at least two centuries later.
Ffu, ful
or fuld
Fuldensis546eapcrConsidered, after Amiatinus, the best Vulgate manuscript. Copied for and corrected by Victor of Capua. Italian text. The Gospels are in the form of a harmony (probably based on an Old Latin original, and with scattered Old Latin readings). Includes the Epistle to the Laodiceans.
forsee under J
--fossSt. Maur des FossésIXe
GSangermanensisIXOT#+NTOld Latin in Matthew (where it is designated g1). French text with some Old Latin elements. Order of sections is eacrp.
--gatVII-IXeReferred to Saint Gatian of Tours. Said to resemble Egertonensis (E) in text, and to have many Old Latin readings. There are many variant readings in the text, usually vulgate and old Latin, written between the lines.
--gigGigas HolmiensisXIIIe(a)pc(r)Same as gig of the Old Latin. Rarely cited as a Vulgate witness, as the Vulgate text is late.
gue lectsee gue among the Old Latin witnesses in Paul
OT+NT#Original text may have been Italian (close to Amiatinus); it has been corrected (often by erasure) toward Theodulf's revision. Three columns per page. The text breaks off at 1 Pet. 4:3. The hand is said to "strongly resembl[e]" that of Q.
harlsee under Z
IingIngolstadiensisVIIe#Many mutilations, especially in Matthew (only 22:39-24:19, 25:14-end remain of that book).
e#Italian text. A legend, obviously false, has it that the portion of this manuscript at Prague was part of the original the Gospel of Mark! Distributed across several libraries. The Markan portion is often illegible, and the final chapters of John are fragmentary. Portions of Mark (at Prague) cited by Tischendorf as prag.
KkarKarolinusIXOT+NTAlcuin's revision. Called "Charlemagne's Bible."
Formerly designated Landavensis. Illuminated manuscript with an Irish text. (The writing is describes as "Irish half-uncial.") Contains Matt. 1:1-Luke 3:9. Legend attributes it to St. Chad.
LVIIIpWritten in a Lombard hand.
LLemovicensisIXc"Mixed" text, containing a part of 1 John 5:7.
MmedMediolanensisVIe#Italian text, considered by Wordsworth & White to rank with Amiatinus and Fuldensis. Assorted lacunae (Matt. 1:1-6, 1:25-3:12, 23:25-25:41; Mark 6:10-8:12) and a few small supplements (Mark 14:35-48; John 19:12-23). Has "interesting lectionary notes in the margins."
MMonacensisIXacr"Good text, but rather mixed, especially in the Acts, where there are strange conjunctions of good and bad readings." Written in "large rough Caroline minuscules."
mac-regolsee under R
martsee under MT (MT)
mmsee under E
MT or
mt or
e"[G]old letters, interesting text."
NVe#Palimpsest. Text is regarded as very valuable.
Obodl or
Bodleianus or
VIIe#Legend has it that this was given by Gregory the Great to Augustine of Canterbury. "British" (i.e. Italian?) text.
a#Described as "most valuable." Lacks 14:26-15:32.
OBodleianusIXp#"Irish hand." Colossians follows Thessalonians. Hebrews breaks off at 11:34. Has been heavily corrected by three different hands. The text of the first hand may have been Old Latin (designated x).
Ppe or
PerusinusVILuke#Luke 1:1-12:7, mutilated. Purple manuscript.
pragsee under J
eBook of Kells (now in Dublin). Generally considered to be most beautiful illuminated manuscript in existence; there is at least some colour on all but two of its 680 pages. Irish text, said by Metzger to have "a peculiar fondness for conflate readings." (An extreme example comes in Matt. 21:31, where, when asked which of the sons did the will of the father, some vulgate texts say "the first," others, "the last"; Kells reads "the first and the last"!)
eRushworth Gospels (so called for the seventeenth century owner who donated it to the Bodleian Library), written by a scribe named Mac Regol who reportedly died in 820 C.E. (Hopkins-James, however, says Mac Regol or "MACREGUIL" died in 800 and was Bishop of Birr; Hopkins-James doubts he was the actual scribe.) Has an interlinear Anglo-Saxon gloss (Matthew in Mercian, Mark-John in Northumbrian; they are listed as the work of scribes named Farman of Harewood and Owun). Skeat declared it to be close to the Lindisfarne Gospels, but Hopkins-James disagrees strongly and says it has a Celtic (Irish) text. Reported to show many alterations in word order.
RReginae SueciaeVII/
pItalian text -- one of the best in Paul.
--regVIII?e#54 leaves of Matthew and Mark, containing less than half of each. Gold uncials, purple parchment. Many old readings.
S or
sanSangallensisVe#Oldest surviving manuscript of the Vulgate Gospels; only about half the leaves have been recovered from manuscript bindings. Italian text, of "remarkable" value.
SstonStoneyhurstensisVIIJohnReportedly found in the coffin of Saint Cuthbert. "A minute but exquisitely written uncial MS. with a text closely resembling A[miatinus]."
SSangallensisVIIIar"Text interesting but mixed." Written by a monk named Winithar. Contains extra-biblical matters as well as the Bible text.
--sanVIe#Matt. 6:21-John 17-18, sometimes fragmentary. The scribe claims to have compiled it from two Latin manuscripts with occasional reference to the Greek.
--sanVIp#Palimpsest (lower text Latin martyrology). Contains Eph. 6:2-1 Tim. 2:5
--theo or
TheotiscaVIIIe#Matthew 8:33-end, mutilated. Old German text on facing pages.
TtolToletanusVIIIOT+NTAlong with cav, the leading representative of the Spanish text. Among the earliest witnesses for "1 John 5:7-8," which it possesses in modified form. Written in a Visigothic hand, it was not new when it was given to the see of Seville in 988.
Th or
theodTheodulfianusIXOT+NTTheodulf's revision, possibly prepared under the supervision of Theodulf himself. The Gospels and Psalms are on purple parchment.
UUltrarajectinaVIMt#Jo#Matt. 1:1-3:4 and John 1:1-21, bound with a Psalter and written in an "Anglian hand" resembling Amiatinus.
UUlmensisIXapcr"Caroline minuscule" hand. Includes Laodiceans. Now in the British Museum.
VvalVallicellanusIXOT+NTAlcuin's revision, written in Caroline minuscules. Considered the best example of this type.
WWillelmi1254OT+NTWritten by William of Hales for Thomas de la Wile. Cited by Wordsworth as typical of the late mediaeval text.
XcantabCantabrigiensisVIIeSaid to have been corrected toward a text such as Amiatinus. Like O, legend has it that Gregory the Great sent it to Augustine of Canterbury.
YlindLindisfarnensisVIIIeIlluminated manuscript with interlinear Anglo-Saxon gloss (old Northumbrian dialect). Second only to the Book of Kells in the quality of its illuminations (some would esteem it higher, since it uses less garish colors). Italian text, very close to Amiatinus. Written by scribes directed by Eadfrith, bishop of Lindisfarne (fl. 698-721 C.E.) in honour of St. Cuthbert.
eItalian text, "in [a] small but very beautiful hand, and with an extremely valuable text."
ZharlHarleianusVIIIpcr#"Written in a French hand, but showing traces of Irish influence in its initials and ornamentation; the text is much mixed with Old Latin readings; it has been corrected throughout, and the first hand so carefully erased in places as to be quite illegible." The base text is late Vulgate, but there are many early readings. The Old Latin portions are designated z. Rev. 14:16-end have been lost.

The following tables facilitate conversion between Wordsworth-White and Tischendorf symbols.
Tischendorf to WW
TischendorfWW TischendorfWW TischendorfWW

WW to Tischendorf
WWTischendorf WWTischendorf WWTischendorf
BF-- L3--Ttol
D-- MTmtU2--
EP-- O2--W--

Old Church Slavonic

Some versions of the New Testament are all but lost. (The Gothic is an example.) Others, such as the Armenian, have survived very well. But few other than the Latin Vulgate have achieved canonical status in their own right. The Old Church Slavonic is an exception.

In the case of the Vulgate, the canonization is perhaps understandable; it is fairly old as versions go, and it was prepared by the greatest scholar of its generation.

A late text of the Slavonic version:
The Gospels of Tsar Ivan Alexander, Tsar of Bulgaria 1331-1371
British Library Add. 39627, comissioned 1355
A copy of the Bulgarian recension.
Shown is folio 88, the beginning of the Gospel of Mark
Gospels of Tsar Ivan Alexander

The case of the Slavonic version is somewhat different from the Latin. It is much newer than the Vulgate, and its translators, while venerated, were not the tremendous scholars that Jerome was. This has meant that the Old Church Slavonic, although it is the Bible to most Slavic Orthodox churches, has received little critical attention -- though rather more attention from linguists, since Old Church Slavonic is the earliest Slavic language with any literary remains.

The history of Christianity among the Slavs is uncertain. One report claims that the Byzantine Emperor Heraclius made an attempt to evangelize the Slavs around the beginning of the seventh century. This account, however, is so littered with contradictions that it cannot be treated as history.

More solid are the accounts of a ninth century mission led by the brothers Methodius and Constantine. Around 860 the two were sent among the Slavs. (There are reports that they found Christians there, and that they were possessed of a partial Bible translation, but we are simply unable to determine the truth, or the details, of this.) In 863 the two went to Moravia and began teaching the locals. From there on the story becomes complicated (if it wasn't before), with local and church politics playing a large role. Leaving aside these details, we are told that Constantine (who eventually took the name Cyril) devised a Slavic alphabet and prepared the Slavic translation.

Here again we run into trouble, because there are two Old Slavonic alphabets, the Glagolitic and the Cyrillic. The Glagolitic is a geometric alphabet, made up of circles and squares and other simple shapes and not evidently related to any other form of writing. The Cyrillic is clearly derived from Greek letter forms.

Despite its name, most experts feel that the Cyrillic alphabet is not the work of Cyril/Constantine (some have credited it to Kliment, a pupil of Constantine and Methodius who worked in Bulgaria). Had the Cyrillic been older, it is hard to see how the Glagolitic could have arisen. The oldest manuscripts of the Old Church Slavonic, which appear to date are from the tenth century, are usually Glagolitic, but the Cyrillic appears not long afterward. Even these early manuscripts show signs of dialectial variations (many of which later became separate languages), so they are probably somewhat removed from the original translation. These also developed minor textual differences, so that we might speak of Bohemian, Bulgarian, Croatian, Russian, and Serbian "recensions" of the Slavonic.

The Old Church Slavonic was translated primarily for liturgical use, so it should not be surprising that lectionary manuscripts are common, and that manuscripts of the Apocalypse (which is not used in the lectionary) are rare.

Research on the Slavonic text has been limited, both because of the difficulty of the language (Old Church Slavonic is, of course, Indo-European, but of the Slavic branch of the family, which is not well known to Western scholars) and because of the lateness of the translation. Slavonic generally renders Greek well (except in matters of verb tense and specific vocabulary), but the text seems to be late. Its Byzantine cast is clear, although there do appear to be some early readings . Vajs considers the basic text in the Gospels to belong with Family P, but with significant "Western" influence.

In Paul, the text is again largely Byzantine, though with some interesting and unusual readings. These do not appear to align with any known text-type.

One can only hope that the future will bring more information to light about this widely revered but little-studied version.


Most versions of the New Testament exist in several recensions. Sometimes these recensions can be very different textually. But usually each successive recension is a revision of those which have gone before -- generally intended to bring the version into closer conformity with the Greek original and the Byzantine Text.

Not so with the Syriac version. Here there was at least one "fresh start," and possibly as many as three. (This is not to say that the newer versions were not influenced by the older; merely that they were not actual revisions of the older.)

The Diatessaron

The history of the Syriac versions probably begins with the Diatessaron, the gospel harmony which Tatian compiled (in Greek or Syriac) in the second half of the second century.

Although the Diatessaron was compiled by an editor who had been in Rome (Tatian was expelled from that city in 172), and although it existed more or less from the start in both Greek and Syriac, it was only in the Syriac church that it is believed to have been regarded as "official." Perhaps it was that Tatian's heretical attitudes fit better with the mood of the church there.

The problems of the Diatessaron are deep and complex; they cannot be dealt with here. No Syriac manuscripts of the version survive, and we have no more than a small fragment of the Greek (in the Dura parchment 0212, a gospel harmony thought by some to be Diatessaric, though the most recent editors think otherwise). But the mass of quotations in Ephraem and others, as well as the number of Diatessaric harmonies in other languages, show its depth of influence.

Eventually, however, the Syriac church felt compelled to set the Diatessaron aside. We have reports of bishops ordering churches to replace their copies of Tatian's document with copies of the Four Gospels. The effectiveness of their efforts is shown by the absence of Diatessaric manuscripts in Syriac. The change was not immediate (writers continued to use the Diatessaron for some centuries), but was eventually complete.

We note incidentally that the Diatessaron, and its suppression, has much to tell us about what can happen to a text. Certain scholars, especially Byzantine prioritists, make a great deal of noise about "normal" transmission -- transmission without interference by external factors. Which is all well and good, but there is no reason to believe that transmission is "normal." If it were, we would have many manuscripts of the Diatessaron, because it would have continued to be copied. Instead, we have no substantial copies of the Diatessaron. Its transmission was not "normal" -- and, given the great range of historical accidents that can happen, the onus is on those who which to claim that transmission is "normal."

The Old Syriac

Competing against the Diatessaron was the Old Syriac. This version (or more correctly, this series of versions) is of uncertain date (some have placed it as early as the second century, others as late as the fourth), and may even be earlier than the Diatessaron, but it was initially far less successful.

The Old Syriac survives in only two manuscripts: The Sinaitic Syriac palimpsest of the late fourth century and the Curetonian Syriac of the late fifth century.

The Sinaitic Syriac (sin or sys), which first came to light in the 1890s, is in many ways the more interesting of the two. Despite the difficulty of reading the text (which was overwritten in the eight century), it is the more complete of the two manuscripts (142 of 166 leaves survive; including Matt. 1:1-6:10, 7:3-12:4, 12:6-25, 12:29-16:15, 18:11-20:24, 21:20-25:15, 25:17-20, 25:25-26, 25:32-28:7, Mark 1:12-44, 2:21-4:17, 5:1-26, 6:5-16:8 (without either the long or the short ending), Luke 1:36-5:28, 6:12-24:52, John 1:25-47, 2:16-4:37, 5:6-25, 5:46-18:31, 19:40-end). Its text is often regarded as more primitive than the Curetonian, with rougher renderings. The text is usually considered "Western," although it is considerably less wild than the text of D.

The Curetonian Syriac (cur or syc) shows most of the peculiarities of the Sinaitic, but perhaps to a lesser degree. Recovered in 1842 and published over the next several decades, it contains about half the Gospels (in the order Matthew, Mark, John, Luke). Specifically, it contains Matt. 1:1-8:22, 10:32-23:25; Mark 16:17-20; John 1:1-42, 3:6-7:37, 14:10-29 (mutilated); Luke 2:48-3:16, 7:33-15:21, 17:24-24:44.

It has been supposed that the Curetonian version is a revision of the Sinaitic translation, probably in the direction of the developing Byzantine text. The Sinaitic, for instance, omits Mark 16:9-20, while the Curetonian contains the verses (16:17-20 being the only parts of Mark to survive in the Curetonian). This should not be considered absolutely certain, however (just as we should not be entirely sure of the relative dates or relationships of the translations). The Sinaitic seems to have stronger affinities to the Alexandrian text, and could conceivably be a revision of the Curetonian text (presumably more Antiochene in the geographical sense, and perhaps with more "Tatianisms") toward the text of Egypt.

The Old Syriac is often regarded as "Western." Certainly the text is quite distinct from the Alexandrian text, and it has many of the hallmarks of the "Western" text -- e.g. paraphrases (in Matt. 1:16, for instance, the Sinaitic has the rather amazing reading "...Jacob fathered Joseph; Joseph, to whom Mary the virgin was engaged, fathered Jesus who is called the Christ") and free insertions and deletions. Certain of these are shared with D and the Old Latins, but many are not -- for instance, of the seven "Western Non-interpolations" in Luke 24, the Old Syriac agrees with D it in 24:40, 52 (cur is defective for 52). However, the manuscripts disagree with D etc. in 24:3, 6, 12, 36, 51 (cur is defective for 51) and have a peculiar omission of their own in 24:32.

And we cannot avoid the fact that the two manuscripts -- especially sin -- have a number of clear agreements with the Alexandrian text. Notable among these is the omission, already alluded to, of Mark 16:9-20 in sin. Both sin and cur join Aleph B X f13 in omitting Matt. 16:2-3. Both join Aleph* B Q 33 579 892* in omitting Matt. 17:21. Sin omits Matt. 18:21 along with Aleph B L* Q f1 f13 33 892*.

Finally, we might note several agreements with the so-called "Cæsarean" witnesses. An obvious example is Matt. 27:16-17, where sin (hiat cur) reads "Jesus Barabbas" with Q f1 700* arm geo2.

The Old Syriac also has a large store of unique readings, some of which may come from local tradition. Thus in Matt. 10:3 sin (hiat cur) lists neither Thaddeus nor Lebbaeus as the apostle, but "Judas of James."

These examples could easily be multiplied. While a handful of examples cannot prove the text-type of the Old Syriac, it is clear that it is not identical to that of D. Some have suggested that the Old Syriac deserves it own text-type (perhaps reasonable, but it would be nice to see a Greek example first...). Streeter's geographical theory place it between the "Cæsarean" and "Western" texts. Others still view the type as "Western," though most would place it in a different subgroup from D.

There are no manuscripts of the Old Syriac outside the gospels. The version certainly existed, but it can only be reconstructed from quotations and commentaries. Based on the materials available, the Old Syriac epistles (which may well be older than the Gospels, since the Diatessaron served as "the gospel" for so long) have a textual complexion similar to the gospels.

The Peshitta

The Peshitta is the oldest Syriac version to survive in its entirety. On that there is general agreement. That is about all that can be stated with certainty.

The date of the Peshitta is perhaps somewhat open to doubt. This question, as we shall see, is of some significance for the history of the text.

The Peshitta can absolutely be dated to the fourth century or earlier. This is implied by the oldest manuscripts (since several are believed to date from the fifth century). Burkitt also points out that it is used by all branches of the Syriac church (which were well and truly sundered by the fifth century -- eventually they even came to develop different versions of the script, so that one can tell by the writing style which Syriac church used a particular manuscript). This implies (though it does not quite prove) that the version was in use before the date of the schism. Folio 154 verso of Sinai Syriac 2
(Peshitta translation, V/VI century), John 17:7-17.
Thanks to Jean Valentin

But if the latest possible date is the late fourth century, what is the earliest? A very early date was once assumed; in the nineteenth century, many scholars would have dated it to the second century. In the twentieth century, this view has largely been abandoned -- not because of any specific evidence, but simply because the earliest Syriac authors (Ephraem in particular) do not quote the Peshitta. We note in addition that the translation includes James, which was not strongly canonical in the second century. In addition, it is generally thought that the Peshitta is dependent on the Old Syriac, which obviously makes it later than the earliest Syriac versions -- though, since the dates of those are disputed, it again fails to prove much. All in all, it's a combination of guesswork and an argument from silence (i.e. it's flatly not proof), but in the absence of anything better, the fourth century date seems to have swept the field.

Whatever its date, the Peshitta is well preserved. Manuscripts from the sixth century are common, meaning that we have substantial early witnesses. Moreover, the manuscripts are considered to agree very closely; with the exception of Vööbus, most scholars believe that we have the version in very nearly the exact form in which it left the translators' hands. (It should be noted, however, that many Peshitta manuscripts, including some of the very oldest, have not been properly examined.)

The style of the Peshitta differs noticeably from the Old Syriac. It is more fluent and more natural than the other Syriac versions. Most scholars therefore believe that it was a substantially new translation rather than a revision. There are readings which remind us of the older Syriac versions, but these may be simple reminiscences rather than actual cases of dependency.

The text of the Peshitta is somewhat mixed. Various studies, mostly in the gospels, have attempted to tie it to all three text-types, but on the whole the Gospels text appears distinctly Byzantine (which is why the date of the Peshitta is so important. Whatever its date, it is the earliest Byzantine witness -- but if it is of the second century, that witness is of much greater significance than if it is of the fourth). This is not to say that the Peshitta is purely Byzantine, or shows the peculiarities of the Textus Receptus. It does not. It omits John 7:53-8:11, for instance. But it includes Matt. 16:2-3, Mark 16:9-20, Luke 22:43-44, 23:34, etc. (most of which are omitted by the Old Syriac). Such non-Byzantine readings as it includes are probably survivals of an older Greek exemplar which has been heavily corrected toward the Byzantine standard.

In the rest of the New Testament the situation is rather different. While the Byzantine text remains the strongest single element, in Acts and Paul the Peshitta includes significant elements of other types. In my estimation, these constitute about 30-40% of the whole. These readings do not, however, seem to belong to any particular text-type; they are neither overwhelmingly "Western" nor Alexandrian. I would guess that the text of the Peshitta here retains hints of the same sort of text we find in the Old Syriac, with some Byzantine overlay. It does not agree with the later (Harklean) Syriac version.

The Peshitta does not contain the Apocalypse, and among the Catholic Epistles it has only James, 1 Peter, and 1 John. Little has been done on its text in the Catholics, except to establish that it is not purely Byzantine. Here again, kinship with the Harklean is slight.

The Philoxenian

The Philoxenian is perhaps the most mysterious of the Syriac versions, because what survives of it is so slight. The only thing we can positively identify as the Philoxenian are certain translations of the books not found in the Peshitta: 2 Peter, 2 John, 3 John, Jude, and the Apocalypse. Such short fragments are not enough to tell us much textually.

Historically, the data are equally confusing, because it is difficult on the face of it to tell the Philoxenian from the Harklean. The Philosenian, we are told, was made in 507-508 C.E. at the instigation of Philoxenus of Mabbûg. It was probably prepared by Polycarp, chorepiscopos of Mabbûg, and was designed as a revision of the Peshitta and intended to render the Greek more precisely as well as supplying the missing five books (and, perhaps, John 7:53-8:11).

Given the uncertainty about this version, there is very little else to be said about it. In the Epistles and the Apocalypse, it is clearly not purely Byzantine -- but the work done on identifying its text beyond that is so out-of-date that it is best ignored.

The Harklean

Of the history of the Harklean version we know little except that it was intended to be a scholarly revision of the Philoxenian. It was undertaken by the Syriac scholar Thomas of Harkel (later Bishop of Mabbûg), and completed in 616.

Given the poor state of preservation of the Philoxenian version, it is hard to be sure to what extent the Harklean is a revision and to what extent it is a new translation. On the basis of the books preserved in both, however, it would appear that the Harklean is substantially new. Whereas the Philoxenian strives for good Syriac style, the Harklean is possibly the most literal translation ever attempted in any language. Even prepositions and particles are translated with wooden consistency, and word order precisely (often slavishly) retained, whether the result is good Syriac or not. The Harklean is completely unsuitable for public use. On the other hand, it is eminently suitable for text-critical work.

Perhaps even more interesting than the Harklean's very literal text is the fact that it is a critical edition. Throughout the New Testament, Thomas used several manuscripts (at least two and perhaps three in all areas), and regularly noted their differences. In the text we find many readings enclosed in obeli, and in the margin we find variant readings in both Greek and Syriac.

This immensely complicated apparatus is one of the chief problems of the Harklean. It is difficult for scribes to copy, and so copies are often imperfect. Before we can reconstruct Thomas's exemplars, we must reconstruct his text, and even that is a major task. Fortunately we have a fair number of manuscripts from the eighth century, and a handful from earlier, so at least we have good materials for reconstructing the version (though critical editions are only now starting to appear).

Even so, we can reach some clear conclusions by studying the Harklean text. In the Gospels, it would seem that all the manuscripts consulted were Byzantine. At least, it has almost all of the longer Byzantine readings (Matt. 16:2-3, Mark 9:44, 46, Luke 22:43-44, 23:34, as well as the full form of the Lord's Prayer in chapter 11, and it has all of the "Western Non-Interpolations" in Luke 24). We do find the shorter ending of Mark in the margin (the long ending in the text); John 5:4 is in asterisks, and the best manuscripts omit John 7:53-8:11.

In the Acts and Epistles, the Harklean is much more interesting; here the manuscripts consulted in preparing the version came from several different families.

In Acts, the Harklean margin was long considered an ally of the "Western" text. It now appears more likely that the Harklean was derived from a Byzantine manuscript and a manuscript of family 2138. Some of the wilder marginal readings may come from a true "Western" source, but most of them are probably of the 2138-type.

This affiliation with family 2138 continues in Paul and the Catholics. In Paul, the Harklean is clearly affiliated with 1505 1611 2495; in the Catholics it goes with the large family 614 630 1505 1611 1799 2138 2412 2495 etc. Of course, it is dependent on a Byzantine source also.

With this information, we are at last in a position to begin reconstructing the translation method of the Thomas of Harkel. Based on the data in the Catholic Epistles, it appears to me that Thomas wanted to preserve the full text of both his exemplars. So, wherever they were variants, he noted them. If the variant was an addition/omission, he put the longer reading in the text but enclosed it in obeli. Where the variants involved substitution, one went in the text and one in the margin. There appears to be no pattern as to which one went in the text; Byzantine and family 2138 readings are found in both text and margin. Presumably there was a critical principle involved, but it was not evident to me.

Little research seems to have been done, to date, on the Harklean version of the Apocalypse.

With the Harklean version, the history of the Bible in Edessene Syriac/Aramaic comes to a close. The Arab Conquest seriously weakened the Christian church, and the demand for new translations probably declined. It also led to an evolution of the Aramaic language. With the call for new renderings so muted, the Peshitta and the Harklean were able to hold the field until modern times. Other Syriac versions exist, but they are in different dialects and completely unrelated. The one verified version in the alternate Palestinian dialect is known, logically enough, as

The Palestinian Version

If the other Syriac versions are like a tree growing out of each other, the Palestinian Syriac (also known as the Jerusalem Syriac or the Christian-Palestinian-Aramaic) may be regarded as from another forest entirely. Dialect, text, and history are all entirely different -- and generally less well-known.

The other Syriac versions are written in the dialect of Edessa, which is properly called Syriac. The Palestinian Syriac is written in a similar script, but the language is that of Palestine (it would be better if it were simply called Aramaic rather than Syriac).

The history of the Palestinian Syriac is largely unknown. No account of its origin has survived. All that can be said with certainty is that the earliest manuscripts appear to date from the sixth century. Most scholars would assign it a date in the fifth or sixth centuries.

The Palestinian Syriac survives primarily in lectionaries. The most important manuscripts of the version are three substantial lectionaries -- one in the Vatican and dated to 1030 C.E. and two at Sinai and dated to 1104 and 1118 C.E. (Ironically, by this time Palestinian Aramaic was evolving into more modern forms, and the copyists had some difficulty with the language.) In addition, there are fragments of the Gospels, Acts, James, 2 Peter, and most of Paul in continuous text manuscripts.

The Palestinian Syriac was clearly made from the Greek. The basis of the version has been the subject of debate. It is clearly not Byzantine, but neither does it appear purely Alexandrian nor "Western." Many have seen it as "Cæsarean," and this seems reasonable on the face of it. More we can hardly say at this time.

The "Karkaphensian" Version

This version will occasionally be referred to in the older manuals. It is not, however, an actual version. Its name was given before the version was properly known, based on a comment of Gregory Bar-Hebraeus, who listed a "Karkaphensian" Syriac version.

The version that passes by this name is not, however, a continuous translation. Rather, it is a collection of passages calling for some sort of scholarly annotation. Sometimes it explains odd words; sometimes it demonstrates the correct orthography of an unusual word. It has therefore been compared to a Syriac "Massorah" such as accompanies the Massoretic Text of the Hebrew Old Testament.

This apparatus seems to exist in two forms -- one Nestorian, one Jacobite. Almost all of the handful of copies are Jacobite, and date from the ninth to the twelfth centuries. Since it is based on the other Syriac versions (especially the Peshitta), it has no proper place in a critical apparatus.

Udi (Alban, Alvan)

This language is of no real interest to the textual critic, since the Udi version no longer exists, but it has some historical significance.

Udi (commonly known as Alban, or more correctly Alvan) is an East Caucasian language, sometimes (rather confusingly) called Caucasian Albanian. It is not Indo-European, but is considered part of the Nakh-Caucasian family, which also includes such tongues as Avar, Chechen, Lezgian, Tabassaran, Lak, and Dargwa. There are only a few thousand speakers left today, most of whom speak at least one other language; it has no literature except perhaps some oral poetry, and that likely to fade soon. Most of the remaining speakers live in Azerbaijan. But Azerbaijan, in the first millenium of the Christian Era, was known as the Kingdom of Alba/Albania. This nation is reported to have been Christianized. Indeed, it is reported that Mesrop, who worked on the Armenian Version, also created an Alban alphabet and an Alban translation.

No traces of this version survive; indeed, no ancient Alban literature is known. We have a few isolated samples of the alphabet on ostraca, just enough to show what it looks like. It cannot even be proved that modern Udi is descended from ancient Alban. But if an Alban New Testament should emerge, it would be among the earliest versions still surviving.