New Testament Manuscripts


Note: In the catalog which follows, bold type indicates a full entry. Plain type indicates a short entry, which may occur under another manuscript.

Additional note regarding the Great Uncials (especially ℵ A B C D): These manuscripts have simply been studied too fully for there to be any hope of a complete examination here, let alone complete bibliographies. The sections below attempt no more than brief summaries.

Contents: * ℵ (01) * A (02) * B (03) * C (04) * Dea (05) * Dp (06) * Dabs * Ee (07) * Ea (08) * Ep: see Dabs * Fe (09) * Fa * Fp (010) * Ge (011) * Ga: see 095 * Gb: see 0120 * Gp (012) * He (013) * Ha (014) * Hp (015) * I (016) * Ke (017) * Kap (018) * Le (019) * Lap (020) * Me (021) * Mp: see 0121 and 0243 * N (022) * O (023) * Pe (024) * Papr (025) * Q (026) * R (027) * S (028) * T (029) * Tg (Scrivener Tp): see 061 * Tk (Scrivener Tg): see 085 * U (030) * V (031) * W (032) * X (033) * Y (034) * Z (035) * Γ (Gamma, 036) * Δ (Delta, 037) * Θ (Theta, 038) * Λ (Lambda, 039) * Ξ (Xi, 040) * Π (Pi, 041) * Φ (Phi, 043) * Ψ (Psi, 044) * Ω (Omega, 045) * 046 * 047 * 048 * 049 * 050 * 053 * 054 * 055 * 056 * 061 * 065 * 066 * 067 * 068 * 069 * 071 * 076 * 085 * 095 and 0123 * 098 * 0120 * 0121 and 0243 * 0122 * 0123: see 095 and 0123 * 0142: see article on 056 * 0130 * 0145 * 0206 * 0212 * 0219 * 0243: see 0121 and 0243 * 0253 * 0254 * 0259 * 0260 * 0261 * 0262 * 0263 * 0264 * 0265 * 0266 * 0268 *

Manuscript ℵ (01)

Location/Catalog Number

The entire New Testament portion, plus part of the Old and the non-Biblical books, are in London, British Library Add. 43725. (A singularly obscure number for what is one of the most important manuscripts in the world!) A handful of Old Testament leaves are at Leipzig. Originally found at Saint Catherine's Monastery on Mount Sinai, hence the name "Codex Sinaiticus." A few stray leaves of the codex apparently remain at Sinai. ℵ is the famous Codex Sinaiticus, the great discovery of Constantine von Tischendorf, the only surviving complete copy of the New Testament written prior to the ninth century, and the only complete New Testament in uncial script.


ℵ presumably originally contained the complete Greek Bible plus at least two New Testament works now regarded as non-canonical: Barnabas and the Shepherd of Hermas. As it stands now, we have the New Testament complete (all in London; 148 leaves or 296 pages total), plus Barnabas and Hermas (to Mandate iv.3.6). Of the Old Testament, we have about 250 leaves out of an original total of some 550. Apart from the portions still at Sinai (which are too newly-found to have been included in most scholarly works), the Old Testament portion consists of portions of Gen. 23, 24, Numbers 5-7 (these first portions being cut-up fragments found in the bindings of other books), plus, more or less complete, 1 Ch. 9:27-19:17, 2 Esdras (=Ezra+Nehemiah) 9:9-end, Esther, Tobit, Judith, 1 Maccabees, 4 Maccabees (it appears that 2 and 3 Maccabees never formed part of the text), Isaiah, Jeremiah, Lament. 1:1-2:20, Joel, Obadiah, Jonah, Nahum, Habakkuk, Zephaniah, Haggai, Zechariah, Malachi, Psalms, Proverbs, Ecclesiastes, Song of Songs, Wisdom of Solomon, Sirach, Job.


Dated paleographically to the fourth century. It can hardly be earlier, as the manuscript contains the Eusebian Canons from the first hand, or at least from a scribe contemporary with the first hand. But the simplicity of the writing style makes a later dating effectively impossible.

Tischendorf was of the opinion that four scribes wrote the manuscript; he labelled them A, B, C, and D. It is now agreed that Tischendorf was wrong. The astonishing thing about these scribes is how similar their writing styles were (they almost certainly were trained in the same school), making it difficult to distinguish them. Tischendorf's mistake is based on the format of the book: The poetic books of the Old Testament are written in a different format (in two columns rather than four), so he thought that they were written by a scribe he called C. But in fact the difference is simply one of page layout; scribe C never existed. For consistency, though, the three remaining scribes are still identified by their Tischendorf letters, A, B, and D.

An interesting aspect of Sinaiticus is its severe plain-ness. Even Codex Vaticanus has occasional graphics (though a lot of them are pretty ugly) and a few instances of red ink. Sinaiticus has almost none. (This may not have been all bad. Sinaiticus is thought to have been in Palestine in the early Islamic era, and a manuscript which did not violate the Islamic ban of representations of living things perhaps had a better chance of surviving.)

Of the three, scribe D was clearly the best, having almost faultless spelling. A, despite having a hand similar to D's, was a very poor scribe; the only good thing to be said about him was that he was better than B, whose incompetence is a source of almost continual astonishment to those who examine his work.

The New Testament is almost entirely the work of scribe A; B did not contribute at all, and D supplied only a very few leaves, scattered about. It is speculated (though it is no more than speculation) that the few leaves written by D were "cancels" -- places where the original copies were so bad that it was easier to replace than correct them. (One of these cancels, interestingly, is the ending of Mark.)

It has been speculated that Sinaiticus was copied from dictation. This is because a number of its errors seem to be errors of hearing rather than of sight (including an amusing case in 1 Macc. 5:20, where the reader seems to have stumbled over the text and the copyist took it all down mechanically). Of course, the possibility cannot be absolutely ruled out that it was not Sinaiticus itself, but one of its ancestors, which was taken down from dictation. In the case of the New Testament, at least, it seems likely that it was not taken from dictation but actually copied from another manuscript.

Sinaiticus is one of the most-corrected manuscripts of all time. Tischendorf counted 14,800 corrections in what was then the Saint Petersburg portion alone!

The correctors were numerous and varied. Tischendorf groups them into five sets, denoted a, b, c, d, e, but there were actually more than this. Milne and Skeat believe "a" and "b" to have been the original scribes (though others have dated them as late as the sixth century); their corrections were relatively few, but those of "a" in particular are considered to have nearly as much value as the original text.

The busiest correctors are those collectively described as "c," though in fact there were at least three of them, seemingly active in the seventh century. When they are distinguished, it is as "c.a," "c.b," and "c.pamph." Corrector c.a was the busiest of all, making thousands of changes throughout the volume. Many of these -- though by no means all -- were in the direction of the Byzantine text. The other two correctors did rather less; c.pamph seems to have worked on only two books (2 Esdras and Esther) -- but his corrections were against a copy said to have been corrected by Pamphilius working from the Hexapla. This, if true, is very interesting -- but colophons can be faked, or transmitted from copy to copy. And in any case, the corrections apply only to two books, neither in the New Testament. There may have been as many as two others among the "c" correctors; all told, Tischendorf at one time or another refers to correctors c, ca, cb, cc, and cc*. These correctors also worked on Barnabas and Hermas, and are thought to have worked from a distinctly different exemplay.

Correctors d and e were much later (e is dated to the twelfth century, and d is from the eighth or later), and neither added particularly many changes. Indeed, no work of d's is known in the New Testament.

It is unfortunate that the Nestle-Aland edition has completely befuddled this system of corrections. In Nestle-Aland 26 and beyond, ℵa and ℵb are combined as ℵ1; the correctors ℵc are conflated as ℵ2, and (most confusing of all) ℵe becomes ℵc. It is understandable that the contemporary, nearly-indistinguishable correctors be lumped, but it would surely be clearer to cite ℵa, ℵc, and ℵe. Of course, NA26 does the same with, for instance, C and D.

(For more information about the correctors of ℵ, see the article on Correctors.)

There has been much dispute about where ℵ was written. Ceriani, who originally thought Syria a likely place, later suggested south Italy. Hort believed it was from the west, perhaps Rome. Harris preferred Cæsarea; although Ropes thinks there is no evidence for this, there are those Pamphilian correctors. Ropes and others support Egypt, very likely Alexandria. This suggestion is complicated by the fact that B and ℵ have different types, at least in Paul, and the B text goes with the Sahidic, so surely the B text is native to Egypt. On the other hand, ℵ goes with the Bohairic, so it too has a claim to Egyptian-ness. It would seem Egypt had two early text-types.

Description and Text-type

The history of Tischendorf's discovery of Codex Sinaiticus is told in almost every introduction to New Testament criticism; I will not repeat it in any detail here (especially since there is a great deal of controversy about what he did). The essential elements are these: In 1844, Tischendorf visited Saint Catherine's Monastery on Mount Sinai. (Sadly, he did not do much to investigate the many fine minuscules at Mount Sinai, such as 1241 and 1881). At one point, he noted 43 sheets of very old parchment in a waste bin, destined to be burned. Tischendorf rescued these leaves (the Leipzig portion of Sinaiticus, all from the Old Testament), and learned that many more existed. He was not able to obtain these other leaves, and saw no sign of the manuscript on a second visit in 1853.

It was not until 1859, near the end of a third visit, that Tischendorf was allowed to see the rest of the old manuscript (learning then for the first time that it contained the New Testament -- complete! -- as well as the Old). Under a complicated arrangement, Tischendorf was allowed to transcribe the manuscript, but did not have the time to examine it in full detail. Tischendorf wanted to take the manuscript to the west, where it could be examined more carefully.

It is at this point that the record becomes unclear. The monks, understandably, had no great desire to give up the greatest treasure of their monastery. Tischendorf, understandably, wanted to make the manuscript more accessible (though not necessarily safer; unlike Saint Petersburg and London, Mount Sinai has not suffered a revolution or been bombed since the discovery of ℵ). In hindsight, it seems quite clear that the monks were promised better terms than they actually received (though this may be the fault of the Tsarist government rather than Tischendorf). Still, by whatever means, the manuscript wound up in Saint Petersburg, and later was sold to the British Museum.

There is at least one interesting sidelight on this, in that Tischendorf's story of his discovery has a clear historical precedent in the discovery of the Percy Manuscript. In around 1753, Thomas Percy was visiting his friend Humphrey Pitt when he discovered the maids burning a paper folio. (A much more reasonable thing to burn than a pile of parchments, which do not burn well!) Percy was able to rescue the century-old poetic miscellany, which eventually inspired him to publish his Reliques in 1765. [Source: Nick Groom, The Making of Percy's Reliques, Oxford, 1999, p. 6.] It almost makes you wonder if Tischendorf had read Percy's account. Happily, the parallels did not extend beyond that point: Percy edited, rewrote, and generally misrepresented his manuscript -- and the owners kept it hidden for many years so that it was impossible to know just how much he had ruined its contents. By contrast, Tischendorf published Sinaiticus with great precision.

However unfair these proceedings to the monks of Sinai, they did make the Sinaiticus available to the world. Tischendorf published elaborate editions in the 1860s, Kirsopp Lake published a photographic edition before World War I, and once the manuscript arrived in the British Museum, it was subjected to detailed examination under ordinary and ultraviolet light.

The fact that ℵ is both early and complete has made it the subject of intense textual scrutiny. Tischendorf, who did not pay much attention to text-types, did not really analyse its text, but gave it more weight than any other manuscript when preparing his eighth and final critical edition. Westcott and Hort regarded it as, after B, the best and most important manuscript in existence; the two copies made up the core of their "neutral" text. Since then, nearly everyone has listed it as a primary Alexandrian witness: Von Soden listed it as a member of the H type; the Alands list it as Category I (which, in practice, means purely Alexandrian); Wisse lists it as Group B in Luke; Richards classifies it as A2 (i.e. a member of the main Alexandrian group) in the Johannine Epistles, etc. The consensus was that there were only two places where the manuscript is not Alexandrian:
* the first part of John, where it is conceded that it belongs to some other text-type, probably "Western," (Gordon D. Fee, in a study whose methodology I consider dubious -- one can hardly divide things as closely as a single verse! -- puts the dividing point at 8:38), and
* in the Apocalypse, where Schmid classifies it in its own, non-Alexandrian, type with 𝔓47.

The truth appears somewhat more complicated. Zuntz, analysing 1 Corinthians and Hebrews, came to the conclusion that ℵ and B do not belong to the same text-type. (Zuntz's terminology is confusing, as he refers to the 𝔓46/B type as "proto-Alexandrian," even though his analysis makes it clear that this is not the same type as the mainstream Alexandrian text.) The true Alexandrian text of Paul, therefore, is headed by ℵ, with allies including A C I 33 81 1175. It also appears to me that the Bohairic Coptic tends toward this group, although Zuntz classified it with 𝔓46/B (the Sahidic Coptic clearly goes with 𝔓46/B), while 1739, which Zuntz places with 𝔓46/B, appears to me to be separate from either.

This leads to the logical question of whether ℵ and B actually belong together in the other parts of the Bible. They are everywhere closer to each other than to the Byzantine text -- but that does not mean that they belong to the same type, merely similar types. There are hints that, in the Gospels as in Paul, they should be separated: B belongs to a group with 𝔓75, and this group seems to be ancestral to L. Other witnesses, notably Z, cluster around ℵ. While no one is yet prepared to say that B and ℵ belong to separate text-types in the gospels, the possibility must at least be admitted that they belong to separate sub-text-types.

In Acts, I know of no studies which would incline to separate ℵ and B, even within the same text-type. On the other hand, I know of no studies which have examined the question. It is likely that the two do both belong to the Alexandrian type, but whether they belong to the same sub-type must be left unsettled.

In Paul, Zuntz's work seems unassailable. There is no question that B and ℵ belong to different types. The only questions are, what are those types, and what is their extent? Zuntz's work is little help, but it would appear that the ℵ-type is the "true" Alexandrian text. 𝔓46 and B have only one certain ally (the Sahidic Coptic) and two doubtful ones (the Bohairic Coptic, which I believe against Zuntz to belong with ℵ, and the 1739 group, which I believe to be a separate text-type). ℵ, however, has many allies -- A, C, 33 (ℵ's closest relative except in Romans), and the fragmentary I are all almost pure examples of this type. Very many minuscules support it with some degree of mixture; 81, 1175, and 1506 are perhaps the best, but most of the manuscripts that the Alands classify as Category II or Category III in Paul probably belong here (the possible exceptions are the members of Families 365/2127, 330, and 2138). It is interesting to note that the Alexandrian is the only non-Byzantine type with a long history -- there are no 𝔓46/B manuscripts after the fourth century, and the "Western" text has only three Greek witnesses, with the last dating from the ninth century, but we have Alexandrian witnesses from the fourth century to the end of the manuscript era. Apart from certain fragmentary papyri, ℵ is the earliest and best of these.

The situation in the Catholic Epistles is complicated. The work of Richards on the Johannine Epistles, and the studies of scholars such as Amphoux, have clearly revealed that there are (at least) three distinct non-Byzantine groups here: Family 2138, Family 1739 (which here seems to include C), and the large group headed by 𝔓72, ℵ, A, B, 33, etc. Richards calls all three of these Alexandrian, but he has no definition of text-types; it seems evident that Amphoux is right: These are three text-types, not three groups within a single type.

Even within the Alexandrian group, we find distinctions. 𝔓72 and B stand together. Almost all other Alexandrian witnesses fall into a group headed by A and 33 (other members of this group include Ψ, 81, 436). ℵ stands alone; it does not seem to have any close allies. It remains to be determined whether this is textually significant or just a matter of defective copying (such things are harder to test in a short corpus like the Catholic Epistles).

As already mentioned, Schmid analysed the manuscripts of the Apocalypse and found that ℵ stood almost alone; its only ally is 𝔓47. The other non-Byzantine witnesses tend to cluster around A and C rather than ℵ. The general sense is that the A/C type is the Alexandrian text (if nothing else, it is the largest of the non-Byzantine types, which is consistently true of the Alexandrian text). Certainly the A/C type is regarded as the best; the 𝔓47/ℵ type is regarded as having many peculiar readings.

Other Symbols Used for this Manuscript

von Soden: δ2
Many critical apparatus (including those of Merk and Bover, as well as Rahlfs in the LXX) refer to ℵ using the siglum "S."


Note: As with all the major uncials, no attempt is made to compile a complete bibliography.


A full edition, with special type and intended to show the exact nature of the corrections, etc. was published by Tischendorf in 1861. This is now superseded by the photographic edition published by Kirsopp Lake (1911). And that in turn has been updated by the detailed scans at The British Library has also released high-resolution scans, at

Sample Plates:

Images are found in nearly every book on NT criticism which contains pictures.

Editions which cite:
Cited in all editions since Tischendorf.

Other Works:
See especially H. J. M. Milne and T. C. Skeat, Scribes and Correctors of Codex Sinaiticus (1938)

Manuscript A (02)

Location/Catalog Number

British Library, Royal 1 D.v-viii. Volumes v, vi, and vii (as presently bound) contain the Old Testament, volume viii the New Testament. Originally given to the English by Cyril Lucar, at various times patriarch of Alexandria and Constantinople. He had it from Alexandria, and so the manuscript came to be called "Codex Alexandrinus," but it is by no means sure that it had always been there. There is a colophon saying it was given to the Patriarchal cell in 1098, but the hand is very late (very possibly post-Lucar), so it is not considered very reliable.


A originally contained the entire Old and New Testaments, plus I and II Clement and (if the table of contents is to be believed) the Psalms of Solomon. As the manuscript stands, small portions of the Old Testament have been lost, as have Matthew 1:1-25:6, John 6:50-8:52 (though the size and number of missing leaves implies that John 7:53-8:11 were not part of the manuscript), 2 Cor. 4:13-14:6. The final leaves of the manuscript have been lost, meaning that 2 Clement ends at 12:4. Like the New Testament, the Old contains some non-canonical or marginally canonical material: 3 and 4 Maccabees, Psalm 151, Odes.


There is some slight disagreement about the date of A. A colophon attributes it to Thecla the martyr, who according to tradition was active in the time of Saint Paul (!), but this is clearly a later forgery, since; it may have been written by a Thecla, but not that Thecla. The late date of the note is shown by the fact that it is written in Arabic; Cyril Lucar himself reportedly wrote a Latin note mentioning the tradition that Thecla was the scribe -- but a fourth century Thecla, not the one from the Pauline era. (It claims that a colophon to this effect was in the lost pages.) The attribution to a fourth century scribe is barely possible. Although most experts believe the manuscript is of the fifth century, a few have held out for the late fourth. A very few have held out for later dates: Semler said seventh, and someone by the name of Oudin apparently placed it in the tenth century! (This was based on the inclusion of an alleged letter of Athanasius, which, it was claimed, must have been written in the tenth century because there were lots of forgeries written around that time. There is also another Arabic note signed Athanasius, who if it is anyone we've ever heard of, is believed to be the Patriarch Athanasius III, who was active in the early fourteenth century.) No scholar since the early nineteenth century has taken either of these claims seriously, however, and our knowledge of ancient manuscripts and their dating is vastly greater now.

Wettstein had a family story that it came to Alexandria from Mount Athos, but although the story was supposed to go back to Lucar's circle, there were several links of hearsay along the way.

The number of scribes has also been disputed; Kenyon thought there were five, but Milne and Skeat (who had better tools for comparison) suggest that there are only two, possibly three. (The uncertainty lies largely in the fact that part of the New Testament, beginning with Luke and ending with 1 Cor. 10:8, present a rather different appearance from the rest of the New Testament -- but when compared in detail, the hand appears extremely similar to the scribe who did the rest of the New Testament. Thus some think there is a change of scribes there, but others disagree. As far as I know, no one has suggested that the current binding combines parts of two different copies, perhaps written by the same scribe. The Apocalypse has been attributed to a third scribe.) Occasional letterforms are said to resemble Coptic letters, perhaps hinting at Egyptian origin, but this is not universally conceded.

A contains a significant number of corrections, both from the original scribe and by later hands, but it has not undergone the sort of major overhaul we see in ℵ or D or even B (which was retraced by a later hand). Nor do the corrections appear to belong to a particular type of text. On the whole, the corrections in the New Testament are minor.

The colophons at the ends of books are decorated (sometimes simply with borders, sometimes with very simply pen-and-ink illustrations); the borders are not very attractive and the illustrations, to my eye, worse; they don't show much evidence of real artistic skill, and certainly don't add much to the manuscript.

Description and Text-type

The story of how A reached its present location is much less involved than that of its present neighbour ℵ. A has been in England since 1627. It is first encountered in Constantinople in 1624, though it is likely that Cyril Lucar (recently translated from the Orthodox Patriarchate of Alexandria to that of Constantinople) brought it with him from Egypt. Lucar was involved in a complex struggle with the Turkish government, the Catholic church, and his own subordinates, and presented the codex to the English in gratitude for their help. The Church of Constantinople was disorderly enough that Lucar seems to have had some trouble keeping his hands on the codex, but it eventually was handed over to the English.

After arriving in Britain, it did have one brief adventure: During the English Civil War, there was threat of dispersal of the Royal Library (the core of what became first the British Museum then the British Library). When Librarian Patrick Young was allowed to retire, he took the Alexandrinus with him; it was finally returned to the Library in 1664. Given how erratic was the behavior of Cromwell's followers, that may have been just as well.

A is somewhat confounding to both the friends and enemies of the Byzantine text, as it gives some evidence to the arguments of both sides.

A is Byzantine in the gospels; there can be no question of this. It is, in fact, the oldest Byzantine manuscript in Greek. (The Peshitta Syriac is older, and is Byzantine, but it obviously is not Greek.) But it is not a "normal" Byzantine witness -- that is, it is not directly related to the Kx type which eventually became dominant. The text of A in the Gospels is, in fact, related to Family Π (Von Soden's Iκ). Yet even those who documented this connection (Silva Lake and others) note that A is not a particularly pure member of Family Π. Nor, in their opinions, was it an ancestor of Family Π; rather, it was a slightly mixed descendent. The additional elements seem to have been Alexandrian -- the obvious example being the omission of John 7:53-8:11, but A also omits, e.g., Luke 22:43-44 and (in the first hand) John 5:3. Westcott and Hort felt the combination of B and A to be strong and significant. We are nonetheless left with the question of the relationship between A and the rest of the Byzantine text. The best explanation appears to me to be that A is derived from a Byzantine text very poorly and sporadically corrected against an Alexandrian document (most likely not systematically corrected, but with occasional Byzantine readings eliminated as they were noticed in an environment where the Alexandrian text dominated). But other explanations are certainly possible.

(Ropes makes the observation that, in the Psalms, A seems to mix early and Lucianic elements, as it mixes a few early readings with Byzantine elements in the gospels. Since the Gospels and the Psalter were the most important service books of the church, they could occur together, so Ropes hints that the text of A is related in these two sections. But this depends on the relationship between "Lucian" and the Byzantine text, which is far from settled.)

The situation in the rest of the New Testament is simpler: A is Alexandrian throughout. It is not quite as pure as ℵ or B or the majority of the papyri; it has a few Byzantine readings. But the basic text is as clearly Alexandrian as the gospels are Byzantine. The Alands, for instance, list A as Category I in the entire New Testament except for the Gospels (where they list it as Category III for historical reasons). Von Soden calls it H (but Iκa in the Gospels).

In Acts, there seems to be no reason to think A is to be associated particularly with ℵ or B. It seems to be somewhat closer to 𝔓74.

In Paul, the situation changes. A clearly belongs with ℵ (and C 33 etc.) against 𝔓46 and B. This was first observed by Zuntz, and has been confirmed by others since then.

The case in the Catholic Epistles is complicated. The vast majority of the so-called Alexandrian witnesses seem to be weaker texts of a type associated with A and 33. (Manuscripts such as Ψ, 81, and 436 seem to follow these two, with Byzantine mixture.) The complication is that neither B nor ℵ seems to be part of this type. The simplest explanation is that the Alexandrian text breaks down into subtypes, but this has not been proved.

In the Apocalypse, A and ℵ once again part company. According to Schmid, ℵ forms a small group with 𝔓47, while A is the earliest and generally best of a much larger group of witnesses including C, the vulgate, and most of the non-Byzantine minuscules. In this book, the A/C text is considered much the best witness. Based on its number of supporters relative to the 𝔓47/ℵ text, one must suspect the A/C text of being the mainstream Alexandrian text, but this cannot really be considered proved -- there simply aren't enough early patristic writings to classify the witnesses with certainty.

Other Symbols Used for this Manuscript

von Soden: δ4


Note: As with all the major uncials, no attempt is made to compile a complete bibliography.


The first publication of the manuscript was as footnotes to the London Polyglot. The symbol "A" comes from Wettstein. A photographic edition (at reduced size) was published by Kenyon starting in 1909.

The British Library has now published high-quality scans of the entire New Testament volume; available at

Sample Plates:

Images are found in nearly every book on NT criticism which contains pictures.

Editions which cite:
Cited in all editions since Tischendorf (plus Wettstein, etc.)

Other Works:

Manuscript B (03)

Location/Catalog Number

Vatican Library, Greek 1209. The manuscript has been there for its entire known history; hence the title "Codex Vaticanus." The catalog made by the Vatican Librarian Platina and his assistant Demetrius Lucensis in 1475 and supplemented in 1481 does not have catalog or shelf numbers (which were not established until 1620), but it refers in the list of Greek manuscripts to a copy of the "Testamentum antiquum et novum" on parchment The 1481 report says it is in three columns. This is the only Greek manuscript listed as containing the whole Bible; it is universally agreed that the 1475 reference is to this manuscript.


B originally contained the entire Old and New Testaments, except that it never included the books of Maccabees or the Prayer of Manasseh. The manuscript now has slight defects; in the Old Testament, it omits most of Genesis (to 46:28) and portions of Psalms (lacking Psalms 105-137). In the New Testament, it is defective from Hebrews 9:14 onward (ending ΚΑΤΑ), omitting the end of Hebrews, 1 and 2 Timothy, Titus, Philemon, and the Apocalypse. It is possible that additional books might have been included at the end -- although it is also possible that the Apocalypse was not included. Indeed, it is barely possible (though this is rarely mentioned) that B originally omitted the Pastorals; this would accord with the contents of its relative 𝔓46.


Early estimates of the date of this manuscript, at a time when knowledge of paleography was limited, varied significantly -- e.g. Montfaucon suggested the fifth or sixth century, and Dupin the seventh. But Hug (who was the first to really stress the importance of this manuscript) suggested the fourth century, and this is no longer questioned; B is now universally conceded to belong to the fourth century, probably to the early part of the century. It is in many ways very primitive, having very short book titles and lacking the Eusebian apparatus. It has its own unique system of chapter identifications; that in the gospels is found elsewhere only in Ξ. It uses a continuous system of numbers in Paul, showing that (in one or another of its ancestors), Hebrews stood between Galatians and Ephesians, even though Hebrews stands after Thessalonians in B itself. There is a second system in Paul as well; we also find two sets of chapter numbers in Acts and the Catholic Epistles, save that 2 Peter is not numbered (perhaps because it was not considered canonical by the unknown person who created this chapter system). This system is also found, in part, in ℵ; Ropes considers it slightly corrupt.

A single scribe seems to have been responsible for almost all of the New Testament, though two scribes worked on the Old. (There is some dispute about this; some see four scribes, and some argue that the the scribe who wrote Psalms 79 to the end of the Old Testament also wrote Matthew 1:1-9:5; the scribe of the rest of the New Testament is sometimes said to have written the middle part of the Old Testament, from 1 Samuel 19 to Psalm 78.)

There were two primary correctors, though the dates of both are rather uncertain. The first is tentatively dated to the sixth century; the second comes from the tenth or eleventh. The second of these is much the more important, though more for damage done than for the actual readings supplied. The date of this scribe is based mostly on the handful of readings he added -- some of these are in minuscule script. This scribe, finding the manuscript somewhat faded, proceeded to re-ink the entire text (except for a few passages which he considered inauthentic). This scribe also added accents and breathings.

This re-inking had several side effects, all of them (from our standpoint) bad. First, it defaced the appearance of the letters, making it much harder to do paleographic work. Second, it rendered some of the readings of the original text impossible to reconstruct. And third (though related to the preceding), it makes it very difficult to tell if there are any original accents, breathings, punctuation, etc. Such marks will generally disappear under the re-inking. Only when such a mark has not been re-inked can we be sure it came from the original hand. Modern techniques could perhaps see through the work of this defacer, but as far as I know, no one has been willing to put up the money for this, and I doubt the Vatican would allow the manuscript to leave their library for as long as the project would need.

It is not absolutely certain when B was damaged, but it certainly happened in the manuscript era, because a supplement with the missing material was later added to the volume. This supplement is late, in a minuscule hand (manuscript 1957, dated paleographically to the fifteenth century; it is believed that the Apocalypse was copied from a manuscript belonging to Cardinal Bessarion. It has been conjectured that Bessarion contributed B to the Vatican library, but this is pure conjecture; all that is known is that the manuscript has been in the library since the compiling of the first catalog in 1475. Ropes thinks it most unlikely that Bessarion would have parted with it.)

Most think the manuscript originated in Egypt, perhaps in Alexandria. Ropes goes so far as to conjecture that it was taken from Egypt to Sicily in the aftermath of the Arab conquest of Egypt, then taken from there to Calabria.

Many have attempted to connect the manuscript to Athanasius, typically based on the books it includes. Some are quite adamant: Since it includes the books Athanasius approved, it must be based on his list. There are a number of objections to this, none of them fatal but all of them quite significant. First and foremost, it is quite possible that B predates Athanasius's career. Second, although the books in B are all in Athanasius's list of accepted books, by the time B was written, the canon was pretty well fixed; only a handful of books were still in doubt. The only really doubtful books included in B are 2 and 3 John and Jude. We do not know that B included the Apocalypse (there is no evidence either way), and we do not know that it included only the 26 or 27 approved books of the New Testament; it could have included other books such as I Clement. And, indeed, it's just possible that, like 𝔓46, it omitted the Pastorals. Yes, it is likely that B had the same canon as Athanasius, but it is beyond proof.

Tischendorf was of the opinion that the scribe who wrote the New Testament of B was also his scribe "D" of ℵ. But it should be kept in mind that Tischendorf never actually got to compare the two. This suggestion is now rejected -- and, as Ropes pointed out, any inferences made on the basis that the two were the same must also be set aside.

Description and Text-type

This is the manuscript. The big one. The key. It is believed that every non-Byzantine edition since Westcott and Hort has been closer to B than to any other manuscript. There is general consensus about the nature of its text: Westcott and Hort called it "Neutral" (i.e. Alexandrian); Von Soden listed it as H (Alexandrian), Wisse calls it Group B (Alexandrian), the Alands place it in Category I (which in practice also means Alexandrian). No other substantial witness is as clearly a member of this text-type; B very nearly defines the Alexandrian text.

Despite the unanimity of scholars, the situation is somewhat more complicated than is implied by the statement "B is Alexandrian." The facts change from corpus to corpus.

In the Gospels, Westcott and Hort centered the "Neutral"/Alexandrian text around B and ℵ (01). At that time, they agreed more closely with each other than with anything else (except that Z had a special kinship with ℵ). Since that time, things have grown more complex. B has been shown to have a special affinity with 𝔓75 -- an affinity much greater than its affinity with ℵ, and of a different kind. The scribal problems of 𝔓66 make it harder to analyse (particularly since ℵ departs the Alexandrian text in the early chapters of John), but it also appears closer to B than ℵ. Among later manuscripts, L has suffered much Byzantine mixture, but its non-Byzantine readings stand closer to B than to ℵ. Thus it appears that we must split the Alexandrian text of the Gospels into, at the very least, two subfamilies, a B family (𝔓66, 𝔓75, B, L, probably the Sahidic Coptic) and an ℵ family (ℵ, Z, at least some of the semi-Alexandrian minuscules). This is a matter which probably deserves greater attention.

There is little to be said regarding Acts. B seems once again to be the purest Alexandrian manuscript, but I know of no study yet published which fully details the relations between the Alexandrian witnesses. It is likely that B, A, and ℵ all belong to the same text-type. We have not the data to say whether there are sub-text-types of this text.

In Paul, the matter is certainly much more complex. Hort described B, in that corpus, as being primarily Alexandrian but with "Western" elements. This was accepted for a long time, but has two fundamental flaws. First, B has many significant readings not found in either the Alexandrian (ℵ A C 33 etc.) or the "Western" (D F G latt) witnesses. Several good examples of this come from Colossians: In 2:2, B (alone of Greek witnesses known to Hort; now supported by 𝔓46 and implicitly by the members of Family 1739) has του θεου Χριστου; in 3:6, B (now supported by 𝔓46) omits επι τους υιους της απειθειας Also, B was the earliest witness known to Hort; was it proper to define its text in terms of two text-types (Western and Alexandrian) which existed only in later manuscripts?

It was not until 1946 that G. Zuntz examined this question directly; the results were published in 1953 as The Text of the Epistles: A Disquisition Upon the Corpus Paulinum. Zuntz's methods were excessively laborious, and cannot possibly be generalized to the entire tradition -- but he showed unquestionably that, first, B and 𝔓46 had a special kinship, and second, that these manuscripts were not part of the mainstream Alexandrian text. This was a major breakthrough in two respects: It marked the first attempt to distinguish the textual history of the Epistles from the textual history of the Gospels (even though there is no genuine reason to think they are similar), and it also marked the first attempt, in Paul, to break out of Griesbach's Alexandrian/Byzantine/Western model.

Zuntz called his proposed fourth text-type "proto-Alexandrian" (p. 156), and lists as its members 𝔓46 B 1739 (plus the relatives of the latter; Zuntz was aware of 6 424** M/0121 1908; to this now add 0243 1881 630 2200) sa bo Clement Origen.

It appears to me that even this classification is too simple; there are five text-types in Paul -- not just the traditional Alexandrian, Byzantine, and "Western" texts, but two others which Zuntz combined as the "Proto-Alexandrian" text. (This confusion is largely the result of Zuntz's method; since he worked basically from 𝔓46, he observed the similarities of these manuscripts to 𝔓46 but did not really analyse the places where they differ.) The Alexandrian, "Western," and Byzantine texts remain as he found them. From the "Proto-Alexandrian" witnesses, however, we must deduct Family 1739, which appears to be its own type. Family 1739 does share a number of readings with 𝔓46 and B, but it also shares special readings with the Alexandrian and "Western" texts and has a handful of readings of its own. It also appears to me that the Bohairic Coptic, which Zuntz called Alexandrian, is actually closer to the true Alexandrian text.

This leaves B with only two full-fledged allies in Paul: 𝔓46 and the Sahidic Coptic. I also think that Zuntz's title "Proto-Alexandrian" is deceptive, since the 𝔓46/B type and the Alexandrian text clearly split before the time of 𝔓46. As a result, I prefer the neutral title 𝔓46/B type (if we ever find additional substantial witnesses, we may be able to come up with a better name).

When we turn to the Catholics, the situation seems once again to be simple. Most observers have regarded B as, once again, the best of the Alexandrian witnesses -- so, e.g., Richards, who in the Johannine Epistles places it in the A2 group, which consists mostly of the Old Uncials: ℵ A B C Ψ 6.

There are several peculiar points about these results, though. First, Richards lumps together three groups as the "Alexandrian text." Broadly speaking, these groups may be described as Family 2138 (A1), the Old Uncials (A2), and Family 1739 (A3). And, no matter what one's opinion about Family 1739, no reasonable argument can make Family 2138 an Alexandrian group. What does this say about Richards's other groups?

Another oddity is the percentages of agreement. For the A2 group, Richards gives these figures for rates of agreement with the group profile (W. L. Richards, The Classification of the Greek Manuscripts of the Johannine Epistles, SBL Dissertation Series, 1977, p. 141):

ManuscriptAgreement %

This is disturbing in a number of ways. First, what is 6 doing in the group? It's far weaker than the rest of the manuscripts. Merely having a 70% agreement is not enough -- not when the group profiles are in doubt! Second, can Ψ, which has clearly suffered Byzantine mixture, really be considered the leading witness of the type? Third, can C (which was found by Amphoux to be associated with Family 1739 in the Catholics) really be the leading Old Uncial of this type? Fourth, it can be shown that most of the important Alexandrian minuscules (e.g. 33, 81, 436, none of which were examined by Richards) are closer to A than to B or ℵ. Ought not A be the defining manuscript of the type? Yet it agrees with the profile only 81% of the time!

A much more reasonable approach is to take more of the Alexandrian minuscules into account, and a rather different picture emerges. Rather than being the weakest Alexandrian uncial, A becomes (in my researches) the earliest and key witness of the true Alexandrian type, heading the group A Ψ 33 81 436 al. The clear majority of the Alexandrian witnesses in the Catholics go here, either purely (as in the case, e.g., of 33) or with Byzantine mixture (as, e.g., in 436 and its near relative 1067). In this system, both B and ℵ stand rather off to the side -- perhaps part of the same type, but not direct ancestors of anything. We might also note that B has a special kinship, at least in the Petrine epistles, with 𝔓72, the one substantial papyrus of the Catholic Epistles. Despite Richards, it appears that B and 𝔓72 form at least a sub-type of the Alexandrian text.

One thing that's worth keeping in mind as we assess B in the New Testament: Although B is generally considered the best manuscript of the Septuagint also, being frequently the best witness to the "Old Greek," there are exceptions -- places where it is kaige or mixed, such as parts of Kings and the Psalter. This does not tell us anything about the New Testament text, of course, but it is a reminder that B's text had to be assembled from multiple sources, just like every other manuscript. Most of those sources are old, but doesn't mean they are all the same.

Other Symbols Used for this Manuscript

von Soden: δ1


Note: As with all the major uncials, no attempt is made to compile a complete bibliography.


B has been published several times, including several recent photographic editions (the earliest from 1904-1907; full colour editions were published starting in 1968). It is important to note that the early non-photographic editions are not reliable. Tischendorf, of course, listed the readings of the manuscript, but this was based on a most cursory examination; the Vatican authorities went to extraordinary lengths to keep him from examining Vaticanus. Others who wished to study it, such as Tregelles, were denied even the right to see it. The first edition to be based on actual complete examination of the manuscript was done by Cardinal Mai (4 volumes; a 1 volume edition came later) -- but this was one of the most incompetently executed editions of all time. Not only is the number of errors extraordinarily high, but no attention is paid to readings of the first hand versus correctors, and there is no detailed examination of the manuscript's characteristics. Despite its advantages, it is actually less reliable than Tischendorf, and of course far inferior to recent editions. Philipp Buttmann produced a New Testament edition based largely on B, but he had B's text via Mai, which he seemingly didn't trust very much, so the resulting edition isn't much like B or anything else (except 2427, which apparently was copied from Buttmann).

Sample Plates:

Images are found in nearly every book on NT criticism which contains pictures.

Editions which cite:
Cited in all editions since Tischendorf

Other Works:
The bibliography for B is too large and varied to be covered here. The reader is particularly referred to a work already mentioned:
G Zuntz, The Text of the Epistles: A Disquisition Upon the Corpus Paulinum.
See also, e.g., S. Kubo, 𝔓72 and the Codex Vaticanus.

Manuscript C (04)

Location/Catalog Number

Paris, National Library Greek 9.


C originally contained the entire Old and New Testaments, but was erased in the twelfth century and overwritten with Syriac works of Ephraem. The first to more or less completely read the manuscript was Tischendorf, but it is likely that it will never be fully deciphered (for example, the first lines of every book were written in red or some other colour of ink, and have completely vanished). In addition, very many leaves were lost when the book was rewritten; while it is barely possible that some may yet be rediscovered, there is no serious hope of recovering the whole book.

As it now stands, C lacks the following New Testament verses in their entirety:

(and, of course, C may be illegible even on the pages which survive). We might note that we are fortunate to have even this much of the New Testament; we have significantly more than half of the NT, but much less than half of the Old Testament.


The original writing of C is dated paleographically to the fifth century, and is quite fine and clear (fortunately, given what has happened to the manuscript since). The scribes of the Old and New Testament are thought to be different; some have suggested that a third scribe wrote Acts. Before being erased, it was worked over by two significant correctors, C2 (Cb) and C3 (Cc). (The corrector C1 was the original corrector, but made very few changes. C1 is not once cited in NA27.) Corrector C2 is thought to have worked in the sixth century or thereabouts; C3 performed his task around the ninth century. (For more information about the correctors of C, see the article on Correctors.)

It was probably in the twelfth century that the manuscript was erased and overwritten; the upper writing is a Greek translation of 38 Syriac sermons by Ephraem.

The manuscript's history has been traced back to the early sixteenth century, when it was in the possession of the Florentine Cardinal Ridolfi (died 1550). There is no evidence of how it was taken from the East to Italy.

Description and Text-type

It is usually stated that C is a mixed manuscript, or an Alexandrian manuscript with much Byzantine mixture. The Alands, for instance, list it as Category II; given their classification scheme, that amounts to a statement that it is Alexandrian with Byzantine influence. Von Soden lists it among the H (Alexandrian) witnesses, but not as a leading witness of the type.

The actual situation is much more complex than that, as even the Alands' own figures reveal (they show a manuscript with a far higher percentage of Byzantine readings in the gospels than elsewhere). The above description is broadly accurate in the Gospels; it is not true at all elsewhere.

In the Gospels, the Alands' figures show a manuscript which is slightly more Byzantine than not, while Wisse lists C as mixed in his three chapters of Luke. But these are overall assessments; a detailed examination shows C to waver significantly in its adherence to the Alexandrian and Byzantine texts. While at no point entirely pure, it will in some sections be primarily Alexandrian, in others mostly Byzantine.

Gerben Kollenstaart brings to my attention the work of Mark R. Dunn in An Examination of the Textual Character of Codex Ephraemi Syri Rescriptus (C, 04) in the Four Gospels (unpublished Dissertation, Southwestern Baptist Theological Seminary 1990). Neither of us has seen this document, but we find the summary, "C is a weak Byzantine witness in Matthew, a weak Alexandrian in Mark, and a strong Alexandrian in John. In Luke C's textual relationships are unclear" (Summarized in Brooks, The New Testament Text of Gregory of Nyssa, p. 60, footnote 1). I dislike the terminology used, as it looks much too formulaic and appears to assume that C's textual affinities change precisely at the boundaries between books. (Given C's fragmentary state, this is even more unprovable than usual.) But the general conclusion seems fair enough: Matthew is the most Byzantine, John the least. In all cases, however, one suspects Byzantine and Alexandrian mixture -- probably of Byzantine readings atop an Alexandrian base. This would explain the larger number of Byzantine readings in Matthew: As is often the case, the corrector was most diligent at the beginning.

Outside the Gospels, C seems to show the same sort of shift shown by its near-contemporary, A -- though, because C possessed Alexandrian elements in the gospels, the shift is less noticeable. But it is not unfair to say that C is mixed in the Gospels and almost purely non-Byzantine elsewhere.

In short works such as Acts and the Catholic Epistles, the limited amount of text available makes precise determinations difficult. In the Acts, we can at least state definitively that C is less Byzantine than it is in the Gospels, but any conclusion beyond that is somewhat tentative. The usual statement is that C is Alexandrian, and I know of no counter-evidence. Nonetheless, given the situation in the Catholic Epistles, I believe this statement must be taken with caution.

The situation in the Catholic Epistles is purely and simply confused. The published evaluations do not agree. W. L. Richards, in his dissertation on the Johannine Epistles using the Claremont Profile Method, does a fine job of muddling the issue. He lists C as a member of the A2 text, which appears to be the mainstream Alexandrian text (it also contains ℵ, A, and B). But something funny happens when one examines C's affinities. C has a 74% agreement with A, and a 77% agreement with B, but also a 73% agreement with 1739, and a 72% agreement with 1243. This is hardly a large enough difference to classify C with the Alexandrians as against the members of Family 1739. And, indeed, Amphoux and Outtier link C with Family 1739, considering their common material possibly "Cæsarean."

My personal results seem to split the difference. If one assumes C is Alexandrian, it can be made to look Alexandrian. But if one starts with no such assumptions, then it appears that C does incline toward Family 1739. It is not a pure member of the family, in the sense that (say) 323 is; 323, after all, may be suspected of being descended (with mixture) from 1739 itself. But C must be suspected of belonging to the type from which the later Family 1739 descended. (Presumably the surviving witnesses of Family 1739 are descended from a common ancestor more recent than C, i.e. Family 1739 is a sub-text-type of the broader C/1241/1739 type.) It is possible (perhaps even likely) that C has some Alexandrian mixture, but proving this (given the very limited amount of text available) will require a very detailed examination of C.

Westcott, in his commentary on the Johannine Epistles, lists the peculiar readings of C (that is, those not shared by ℵ A B), adding that they "have no appearance of genuiness":

Westcott's statement seems to be generally true -- all the items here appear to be either singular, simple errors of omission, or minor paraphrases. But C still appears highly valuable when in company with good witnesses.

In Paul, the situation is simpler: C is a very good witness, of the Alexandrian type as found in ℵ A 33 81 1175 etc. (This as opposed to the type(s) found in 𝔓46 or B or 1739). So far as I know, this has never been disputed.

In the Apocalypse, C is linked with A in what is usually called the Alexandrian text. No matter what it is called, this type (which also includes the Vulgate and most of the better minuscules) is considered the best type. Note that this is not the sort of text found in 𝔓47 and ℵ.

Other Symbols Used for this Manuscript

von Soden: δ3


Note: As with all the major uncials, no attempt is made to compile a complete bibliography.


Various editors extracted occasional readings from the manuscript, but Tischendorf was the first to read C completely. Tischendorf is often reported to have used chemicals, but in fact it is believed that they were applied before his time -- and they have hastened the decay of the manuscript. Tischendorf, working by eye alone, naturally did a less than perfect job. Robert W. Lyon, in 1958-1959, published a series of corrections in New Testament Studies (v). But this, too, is reported to be imperfect. The best current source is the information published in the Das Neue Testament auf Papyrus series. But there is no single source which fully describes C.

Sample Plates:

Sir Frederick Kenyon & A. W. Adams, Our Bible and the Ancient Manuscripts

Editions which cite:
Cited in all editions since Tischendorf

Other Works:
Mark R. Dunn, An Examination of the Textual Character of Codex Ephraemi Syri Rescriptus (C, 04) in the Four Gospels (unpublished Dissertation, Southwestern Baptist Theological Seminary 1990)

Manuscript Dea (05)

Location/Catalog Number

Cambridge, University Library Nn. 2. 41. The well-known Codex Bezae, so-called because it was once the possession of Theodore Beza.


Greek/Latin diglot, with the Greek on the left page. The Greek currently contains the Gospels and Acts with lacunae; the manuscript lacks Matt. 1:1-20, 6:20-9:20, 27:2-12, John 1:16-3:26, Acts 8:29-10:14, 21:2-10, 16-18, 22:10-20, 29-end. In addition, Matt. 3:7-16, Mark 16:15-end, John 18:14-20:13 (a total of ten leaves) are supplements from a later hand (estimated to date from the tenth to twelfth century). The Gospels are in the "Western" order Matthew, John, Luke, Mark, though Chapman offered evidence that an ancestor had the books in the order Matthew, Mark, John, Luke.

Since the Greek and Latin are on facing pages, the losses to the Latin side are not precisely parallel; d (the symbol for the Latin of D; Beuron #5) lacks Matt. 1:1-11, 2:20-3:7, 6:8-8:27, 26:65-27:2, Mark 16:6-20, John 1:1-3:16, 18:2-20:1, Acts 8:21-10:3, 20:32-21:1, 21:8-9, 22:3-9, 22:21-end. In addition, the Latin includes 3 John 11-15.

The original contents of D are somewhat controversial. Obviously it must have contained the Gospels, Acts, and 3 John. This would seem to imply that the manuscript originally contained the Gospels, Catholic Epistles, and Acts (in that order). This, however, does not fit well with the pagination of the manuscript; Chapman theorized that the manuscript actually originally contained the Gospels, Apocalypse, 1 John, 2 John, 3 John, and Acts (in that order), and others (e.g. Ropes) have accepted the hypothesis. This makes sense, but cannot be considered certain -- particularly since it is at least possible that some works would have been included in only one language (e.g. the Catholic Epistles in Latin only, plus some other work, also in Latin only). It's even possible that there was a non-canonical work and that it was deliberately excised.


The manuscript has been variously dated, generally from the fourth to the sixth centuries (very early examiners gave even more extreme dates: Kipling, who published a facsimile in 1793, claimed a second century date, and Michaelis also considered it the earliest manuscript known, but a few guessed dates as late as the seventh century.) In the middle of the twentieth century, the tendency seemed to be to date it to the sixth century; currently the consensus seems to be swinging back toward the fifth. It is very difficult to achieve certainty, however, as the handwriting is quite unusual. The Greek and Latin are written in parallel sense lines, and the scribe uses a very similar hand for both languages -- so much so that a casual glance cannot tell the one language from the other; one must look at the actual letters and what they spell.

The unusual writing style is only one of the curiosities surrounding the scribe of D. It is not clear whether his native language was Greek or Latin; both sides of the manuscript contain many improbable errors. (Perhaps the easiest explanation is that the scribe's native language was something other than Greek or Latin.)

D's text, as will be discussed below, was far removed from the Byzantine standard (or, perhaps, from any other standard). As a result, it was corrected many times by many different scribes. Scrivener believed that no fewer than nine correctors worked on the manuscript, the first being nearly contemporary with the original scribe and the last working in the eleventh or twelfth century. Some have counted as many as twenty correctors. (Ironically, there is little evidence that the manuscript was corrected when it was written.) In general, these correctors brought the manuscript closer to the Byzantine text (as well as adding occasional marginal comments and even what appear to be magical formulae at the bottom of the pages of Mark). For more recent views on these correctors, see D. C. Parker's work on Codex Bezae; Parker redates some of the correctors (moving them back some centuries), and believes that one had an Alexandrian text. The majority of the correctors were most interested in the Greek text, although one worked mostly on the Latin.

The known history of the manuscript does not clarify the significance its text; it cannot be traced back far enough. We first hear of it at the time of the Council of Trent, when it apparently was consulted about certain Latin readings. The readings printed by Stephanus as β seem to have been taken down at this time. After this it was in the possession of the Monastery of Irenæus at Lyon (where it had probably been before being taken to Trent); it came to Beza from there in 1562, probably as a result of commotion in the town. Beza gave it to Cambridge University in 1582.

There are indications -- though they fall far short of proof -- that the codex had been in Lyon for some time before it was noticed. There are 71 notes in the margin, from a later hand than the original text but probably no later than the tenth century, that seem to have been intended for fortune-telling. These are similar to a series found in the Latin Codex Sangermanensis, also associate with Lyon. And Quentin thought that the martyrology of Ado of Lyon had certain Biblical readings associated with the Bezan text. Given that Quentin found only seven such readings, the association is suggestive but certainly not conclusive. But it might explain why the long line of Greek correctors eventually stopped working on the manuscript: there were no Greek scholars in Lyon. This would, however, hint that the manuscript originated elsewhere. This is supported by the fact that the manuscript contains indications for the Byzantine lectionary, which was not used in Lyon. Many have argued for southern Italy, but Ropes counters by saying that there were no Greek communities in southern Italy at the time Bezae was written. He prefers Sicily, which was far more Greek. I don't think this compelling; even a Latin-speaking community might have people who wanted to look at the Greek original -- and who would like a parallel version. I incline to agree with those who say the best bet for the origin is southern Italy, but honesty compels us to admit that we really don't know.

Based on multiple indications including the loss of whole lines from both the Greek and the Latin, it is believed that Bezae was copied from original that was already in bilingual sense lines (so there was evidently a specially-compiled bilingual edition of which it was a copy), but that Bezae does not follow the pagination of this original, just the lineation.

Description and Text-type

The text of D can only be described as mysterious. We don't have answers about it; we have questions. There is nothing like it in the rest of the New Testament tradition. It is, by far the earliest Greek manuscript to contain John 7:53-8:11 (though it has a form of the text quite different from that found in most Byzantine witnesses). It is the only Greek manuscript to contain (or rather, to omit) the so-called Western Non-Interpolations. In Luke 3, rather than the Lucan genealogy of Jesus, it has an inverted form of Matthew's genealogy (this is unique among Greek manuscripts). In Luke 6:5 it has a unique reading about a man working on the Sabbath. D and Φ are the only Greek manuscripts to insert a loose paraphrase of Luke 14:8-10 after Matt. 20:28. And the list could easily be multiplied; while these are among the most noteworthy of the manuscript's readings, it has a rich supply of other singular variants.

In the Acts, if anything, the manuscript is even more extreme than in the Gospels. F. G. Kenyon, in The Western Text of the Gospels and Acts, describes a comparison of the text of Westcott & Hort with that of A. C. Clark. The former is essentially the text of B, the latter approximates the text of D so far as it is extant. Kenyon lists the WH text of Acts at 18,401 words, that of Clark at 19,983 words; this makes Clark's text 8.6 percent longer -- and implies that, if D were complete, the Bezan text of Acts might well be 10% longer than the Alexandrian, and 7% to 8% longer than the Byzantine text.

This leaves us with two initial questions: What is this text, and how much authority does it have?

Matthaei referred to it as editio scurrilis, but nineteenth century scholars inclined to give the text great weight. Yes, D was unique, but in that era, with the number of known manuscripts relatively small, that objection must have seemed less important. D was made the core witness -- indeed, the key and only Greek witness -- of what was called the "Western" text.

More recently, Von Soden listed D as the first and chief witness of his Iα text; the other witnesses he includes in the type are generally those identified by Streeter as "Cæsarean" (Θ 28 565 700 etc.) The Alands list it as Category IV -- a fascinating classification, as D is the only substantial witness of the type. Wisse listed it as a divergent manuscript of Group B -- but this says more about the Claremont Profile Method than about D; the CPM is designed to split Byzantine strands, and given a sufficiently non-Byzantine manuscript, it is helpless. (Biologists have a term for this phenomenon: It's known as "long branch assimiliation." If you have a large mass of closely related entities, and two entities not related to the large mass, the two distant entities may look related just because they are way out in the middle of nowhere.)

The problem is, Bezae remains unique among Greek witnesses. Yes, there is a clear "Western" family in Paul (D F G 629 and the Latin versions.) But this cannot be identified with certainty with the Bezan text; there is no "missing link" to prove the identity. Not one Greek manuscript contains a "Western" text of both the Gospels and Paul! There are Greek witnesses which have some kinship with Bezae -- ℵ in the early chapters of John; the fragmentary papyri 𝔓29 and 𝔓38 and 𝔓48 in Acts. But none of these witnesses is complete, and none is as extreme as Bezae.

D's closest kinship is with the Latin versions, but none of them are as extreme as it is. D is, for instance, the only manuscript to substitute Matthew's genealogy of Jesus for Luke's. On the face of it, this is not a "Western" reading; it is simply a Bezan reading.

Then there is the problem of D and d. The one witness to consistently agree with Dgreek is its Latin side, d. Like D, it uses Matthew's genealogy in Luke. It has (that is, it omits) all the "Western Non-Interpolations." And, perhaps most notably, it has a number of readings which appear to be assimilations to the Greek.

Yet so, too, does D seem to have assimilations to the Latin. Many of them.

We are almost forced to the conclusion that D and d have, to some extent, been conformed to each other. The great question is, to what extent, and what did the respective Greek and Latin texts look like before this work was done?

On this point there can be no clear conclusion. Wettstein theorized that the Greek text was conformed to the Latin. Matthaei had a modified version of this in which the marginal readings of a commentary manuscript might also have been involved, and considered it deliberately edited. Hort thought that D arose more or less naturally; while he considered its text bad, he was willing to allow it special value at some points where its text is shorter than the Alexandrian. (This is the whole point of the "Western Non-Interpolations.") More recently, however, Aland has argued that D is the result of deliberate editorial work. This is unquestionably true in at least one place: The Lukan genealogy of Jesus. Is it true elsewhere? This is the great question, and one for which there is still no answer.

As noted, Bezae's closest relatives are Latin witnesses. And these exist in abundance. If we assume that these correspond to an actual Greek text-type, then Bezae is clearly a witness to this type. And we do have evidence of a Greek type corresponding to the Latins, in Paul. The witnesses D F G indicate the existence of a "Western" type. So Bezae does seem to be a witness of an actual type, both in the Gospels (where its text is relatively conservative) and in the Acts (where it is far more extravagant). (This is in opposition to the Alands, who have tended to deny the existence of the "Western" text.)

So the final question is, is Bezae a proper witness to this text which underlies the Latin versions? Here it seems to me the correct answer is probably no. To this extent, the Alands are right. Bezae has too many singular readings, too many variants which are not found in a plurality of the Latin witnesses. It probably has been edited (at least in Luke and Acts; this is where the most extreme readings occur). If this is true (and it must be admitted that the question is still open), then it has important logical consequences: It means that the Greek text of Bezae (with all its assimilations to the Latin) is not reliable as a source of readings. If D has a reading not supported by another Greek witness, the possibility cannot be excluded that it is an assimilation to the Latin, or the result of editorial work.

All that being said, there remains disagreement on just how heavily edited, just how wild, just how peculiar the text of D is. And, one suspects, that will remain true until and unless another similar Greek text shows up.

Other Symbols Used for this Manuscript

von Soden: δ5


Note: As with all the major uncials, no attempt is made to compile a complete bibliography.


The standard reference is probably still F. H. A. Scrivener, Bezae Codex Canatabrigiensis, simply because of Scrivener's detailed and careful analysis. J. Rendel Harris published a photographic reproduction in 1899. See also J. H. Ropes, The Text of Acts and A. C. Clark, The Acts of the Apostles, both of which devote considerable attention to the text of Bezae in Acts.

Sample Plates:

(Sample plates in almost all manuals of NT criticism)

Editions which cite:
Cited in all editions since Tischendorf, and most prior to that.

Other Works:
The most useful work is probably James D. Yoder's Concordance to the Distinctive Greek Text of Codex Bezae. There are dozens of specialized studies of one or another aspect of the codex, though few firm conclusions can be reached (perhaps the most significant is the conclusion of Holmes and others that Bezae has been more thoroughly reworked in Luke than in Matthew or Mark). See also the recent work by D. C. Parker, Codex Bezae.

Manuscript Dp (06)

Location/Catalog Number

Paris, National Library Greek 107, 107 AB. The famous Codex Claromontanus -- so-called because Beza reported that it had been found at Clermont. It should not be confused with the even more famous, or infamous, Codex Bezae, also designated D.


Greek/Latin diglot, with the Greek and Latin in stichometric lines on facing pages. Contains the Pauline Epistles with the slightest of lacunae: It lacks Romans 1:1-7 (though we can gain some information about the readings of D in these verses from Dabs). In addition, Romans 1:27-30 and 1 Corinthians 14:13-22 are supplements from a later hand. (Scrivener, however, notes that this hand is still "very old.") Hebrews is placed after Philemon.

The Latin side, known as d (Beuron 75) has not been supplemented in the same way as the Greek; it lacks 1 Corinthians 14:9-17, Hebrews 13:22-end. Romans 1:24-27 does come from a supplement.

Scrivener observes that the very fine vellum actually renders the manuscript rather difficult to read, as the writing on the other side often shows through. The extent of this problem varies from page to page, but often the ink from the reverse side is almost as visible as that on the side being read. This is a problem even with the photographs on the Paris Library web site. Also, the scribe apparently had rather poor pen technique; the darkness of the ink varies from letter to letter. The problem is made worse by the fact that, on many pages, the ink has flaked off. It is not a very easy manuscript to read; the online scans are often barely legible.


The first three lines of each book is written in red ink. The original ink is brown, sometimes dark, sometimes light; some of the correctors used much blacker ink. Greek and Latin hands are similar looking and elegant in a simple way.

Almost all scholars have dated D to the sixth century (some specifying the second half of that century); a few very early examiners would argued for the seventh century. The writing is simple, without accents or breathings; some of the uncial forms seem to be archaic. The Greek is more accurately written than the Latin; the scribe's first language was probably Greek. We should note certain broad classes of errors, however. The scribe very frequently confuses the verb ending -θε with -θαι; this occurs so regularly that we can only say that D is not a witness at variants of this sort.

A total of nine correctors have been detected, though not all of these are important. The first important corrector (D** or, in NA26, D1) dates probably from the seventh century; the single most active corrector (D*** or D2, who added accents and breathings and made roughly 2000 changes in the text) worked in the ninth or tenth century; the final significant corrector (D*** or Dc) probably dates from the twelfth century or later.

The corrections to the Greek side are much more numerous than those on the Latin side, which has only minimal corrections.

Description and Text-type

There is an inherent tendency, because D is a Greek/Latin diglot and because it is called "D," to equate its text with the text of Codex Bezae, making them both "Western." This is, however, an unwarranted assumption; it must be proved rather than simply asserted.

There is at least one clear and fundamental difference between Bezae and Claromontanus: They have very different relationships to their parallel Latin texts. The Greek and Latin of Bezae have been harmonized; they are very nearly the same text. The same is not true of Claromontanus. It is true that D and d have similar sorts of text -- but they have not been entirely conformed to each other. The most likely explanation is that dp was translated from a Greek text similar to Dp, and the two simply placed side by side.

Claromontanus also differs from Bezae in that there are Greek manuscripts similar to the former: The close relatives Fp and Gp are also akin, more distantly, to Claromontanus. All three of these manuscripts, it should be noted, have parallel Latin versions (in the case of F, on a facing page; the Latin of G is an interlinear). All three, we might add, are related to the other Old Latin codices (a, b, m; they are rather more distant from r) which do not have Greek parallels.

Thus it seems clear that there is a text-type centred about Dp F G and the Latins. Traditionally this type has been called "Western," and there is no particular reason to change this name.

We should make several points about this Western text of Paul, though. First, it is nowhere near as wild as the text of Codex Bezae, or even the more radical Old Latin witnesses to the Gospels and Acts. Second, it cannot be demonstrated that this is the same type as is found in Bezae. Oh, it is likely enough that Bezae's text is edited from raw materials of the same type as the ancestors of D F G of Paul. But we cannot prove this! Astonishingly enough, there is not one Old Latin witness containing both the Gospels and Paul. There are a few scraps (primarily t) linking the Acts and Paul, but even these are quite minimal. Thus, even if we assume that Bezae and Claromontanus represent the same type, we cannot really describe their relative fidelity to the type (though we can make a very good assumption that Claromontanus is the purer).

We should also examine the relations between the "Western" witnesses in Paul. It is sometimes stated that F and G are descendents of D. This almost certainly not true -- certainly it is functionally untrue; if F and G derive from D, there has been so much intervening mixture that they should be regarded as independent witnesses.

Interestingly, there is a sort of a stylistic difference between D and F/G. F and G appear to have, overall, more differences from the Alexandrian and Byzantine texts, but most of these are small, idiosyncratic readings which are probably the result of minor errors in their immediate exemplars. D has far fewer of these minor variants, but has an equal proportion (perhaps even a higher proportion) of more substantial variants.

So far we have mentioned only these two uncials as relatives of D. We should note that these manuscripts were merely the leading witnesses of Von Soden's Ia1 type; with them he classified a number of minuscules: 88 181 915 917 1836 1898 1912. Several of these minuscules (e.g. 88 and 181) do appear to be somewhat related to each other, but there is no real evidence that they are akin to the key "Western" witnesses. (88*, it is true, joins the Western uncials in placing 1 Corinthians 14:34-35 after 14:40, but this is almost its only special agreement). The only minuscule to show real kinship with the Western uncials is 629. It is likely, however, that this kinship is not properly genetic; rather, 629 is a Greek/Vulgate diglot, and there are instances where the Greek seems to have been conformed to the Latin. Since the Vulgate, in Paul, has many "Western" readings, this has given 629 something of a "Western" tinge.

The case is rather different for the Latin witnesses. These clearly are related to D F G. Of the Latins, d is the closest to D, though by no means identical; b is also closely related. D, d, and b are rather more distant from a and m, and still more distant from r (the latter fragments sometimes seem to approach the Alexandrian text). The other Old Latin fragments of Paul are too short to assess properly.

The classification used by the Alands for the diglot uncials of Paul is fascinating. None of them is classified as Category IV -- in other words, the Alands do not regard them a belonging to the same type as Codex Bezae. (Of course, it should be noted have not published definitions of their categories, but that it is clear that Category IV has no definition at all; they simply placed witnesses there because they felt like it.) But the situation is curious even if we ignore Category IV. In the second edition of their Introduction, they list D, the oldest manuscript of the type, as Category III; the same description is applied to G -- but F, which is universally agreed to be a close relative of G, but inferior on the whole, is listed as Category II! The most charitable word I can think of for this is "inexplicable."

Other Symbols Used for this Manuscript

von Soden: α1026


Note: As with all the major uncials, no attempt is made to compile a complete bibliography.


Tischendorf's 1852 edition remains the standard (if it can be found); beyond that, one must turn to K. Junack, Das Neue Testament auf Papyrus, Vol. 2: Die paulinischen Briefe

Sample Plates:

Aland & Aland (1 plate); also a facsimile in Scrivener

The full manuscript can now be viewed at the Paris National Library web site,

Editions which cite:
Cited in all editions since Tischendorf, and most prior to that.

Other Works:


There are actually two manuscripts which circulate under the symbol Dabs, correctly designated Dabs1 and Dabs2. Both are Greek/Latin diglots. It is one of the curiosities of textual criticism that almost no manuscripts are known which are copies of other manuscripts. Only two uncials are known to be copies of other uncials -- and both are copies of the Pauline Codex D/06 (Claromontanus). Their descriptions are as follows:

Manuscript Ee (07)

Basel, University Library A.N. III. 12. Von Soden's ε55. Contains the Gospels almost complete; lacks Luke 3:4-15, 24:47-end. Luke 1:69-2:4, 12:58-13:12, 15:8-20 are supplements in a later, cursive hand. Dated paleographically to the eighth century (so all recent authorities; Burgon argued for the seventh; the letterforms look old, but the accents, breathings, and punctuation argue that it is relatively recent). This makes it the very first purely Byzantine uncial in any part of the Bible; it is the first Byzantine manuscript to contain not merely the small, more ordinary Byzantine readings but also the story of the Adulteress (found earlier in D, but no one will claim Bezae is Byzantine!). (In the gospels, there are earlier almost-pure Byzantine uncials: A and the Purple Uncials; elsewhere, all Greek witnesses to the Byzantine text are even later than E. Obviously the Byzantine type is much older than E. E is simply the earliest pure representative of what became the dominant type in the Middle Ages.) All examiners have agreed on E's Byzantine nature; the Alands list it as Category V; von Soden lists it as Ki; Wisse calls it Kx Cluster Ω (We might add that Kx Cluster Ω is Ki; Wisse's three chapters did not provide enough text to distinguish the two groups, but historical evidence seems to imply that Kx proper and Kx Cluster Ω are distinct although very closely related.) Certain disputed passages are marked with asterisks (Matt. 16:2-3, Luke 22:43-44, 23:34, John 8:2-11). E is well and carefully written, and probably deserves inclusion in critical apparati as the leading witness of the later Byzantine type.

Manuscript Ea (08)

Location/Catalog Number

Oxford, Bodleian Library, Laud. Greek 35. Called "Codex Laudianus" because it was donated to the Bodleian Library by William Laud (1573-1645†), the anti-Calvinist Archbishop of Canterbury under the British King Charles I.


Contains the Acts almost complete; lacks 26:29 (from παυλος) to 28:26 (resuming after λεγον). The parchment is very thin, and there is some burn-through of ink, which, combined with the light colour of some letters, occasionally makes it difficult to read. Greek/Latin diglot, with the languages in parallel columns on the same page. The Latin, which is cited as e, is on the left. The manuscript is divided into sense lines of sorts, for purposes of parallelism, but as the lines are generally no more than fifteen letters long (often consisting of a single word!), they rarely form any real sort of syntactic unit.


Dated paleographically to the sixth or seventh century, with most scholars inclining toward the sixth (a few early examiners gave fifth or eighth century dates). Early in its career, it was in Sardinia; an entry (not by the original hand) refers to an edict of a Byzantine governor of that island (which was under Byzantine rule from 534). It probably was written there, although it cannot be proved.

A very noteworthy fact is the high degree of agreement between the Latin text and that used by the Venerable Bede for his commentary Expositio Retractata. (If we assume the continued identity after the end of the surviving portion of e, Bede gives us two Latin readings now lost: 27:5, 28:2). It is generally assumed that Bede used this very manuscript for his commentary, which would guarantee that it was in existence in 716, and located at that time in what is now northern England. It has been specutated that the manuscript may have been brought to Britain by Theodore or Hadrian, respectively archbishop and abbot of Canterbury, who arrived in the country in 668; both of them were familiar with Greek as well as Latin and reportedly arrived with a significant library. Ropes, however, says that there is no record of Theodore of Tarsus bringing books to Britain; he thinks it might have been brought by Benedict Biscop and/or Ceolfrid, who came a few decades before Theodore.

I have seem some state that Bede cannot in fact have used this manuscript, despite the similarity, but I have not seen any reasons advanced. (The flip side is, of course, that E probably had an exemplar, and perhaps had offspring, and possibly Bede used one of those.)

After Bede's time it was taken to Germany, perhaps by missionaries. The best guess is that it ended up in Würzburg, and was returned to England after that city was sacked in 1631. One of Laud's buyers probably found it there.

It is hard to know what to make of the scribe. Although Metzger calls the uncials "clumsy," in fact both Greek and Latin letterforms are clearly written if large. On the other hand, the scribe had a great deal of difficulty with his pen, which ran dry every few letters. Based on this fact, it appears to me that he wrote the Latin column first, then the Greek, rather than writing across the page.

Perhaps the best explanation for his difficulties is that he was having to create his text as he went along. According to Margaret T. Gibson, The Bible in the Latin West, being volume 1 of "The Medieval Book," University of Notre Dame Press, 1993, the scribe shows clear signs of combining a Greek and a Latin text (often resulting in lines too long for the allotted space where he mis-estimated line lengths). This would imply that many of the errors are the result of trying to edit his text even while writing it. On the other hand, Ropes seems to think that there was a pre-existing edition which the scribe copied.

Lowe suggests that the scribe's first language was Greek, on the basis that many of the Latin letter forms appear to be adjustments of the way the letters are written in Greek. Ropes also thinks Greek was his first language.

Description and Text-type

The Greek of E, it is generally conceded, is more Byzantine than anything else. The manuscript is mixed, however, there are many "Western" and some Alexandrian readings. (In fact, the manuscript seems somewhat block-mixed; "Western" readings are much more common in some sections than in others.) The Latin is not the vulgate, but rather a unique version of the Old Latin.

This raises the question of whether the Greek has been conformed to the Latin or vice versa. Different scholars have answered this differently. Scrivener, for instance, reports that "the Latin... is made to correspond closely with the Greek, even in its interpolations and rarest various readings. The contrary supposition that the Greek portion of this codex Latinised, or has been altered to coincide with the Latin, is inconsistent with the facts of the case." More recent scholars such as Ropes and Clark, however, maintain that the Greek has in fact been conformed to the Latin. In this context, it is worth noting that the Latin is in the left-hand column, usually regarded as the place of honour. Ropes thinks the Latin was based on the type found in Codex gigas -- although it has been modified and made somewhat less extreme. Ropes in fact thinks that some of the Greek readings are retranslations from the Latin to supply text that was missing in the Greek.

It should be added, however, that the Latin of e seems somewhat unusual. And the arrangement of the two parts, with such short sense lines, argues that both texts may have undergone some adjustment. This is, however, only logic.... The most important point is that E has a mixed text, heavily but not purely Byzantine. It also has a number of interesting long readings, the most famous being Acts 8:37 (the Ethiopian Eunuch's acceptance of faith). By its nature, any reading in E must be taken with some hesitation and examination of its sources. This is reflected in earlier classifications of the manuscript: Von Soden listed it as Ia1 (i.e. as part of the core "Western" text), but the Alands list it as only Category II.

Ropes believes that the Greek, before it was adjusted to match the Latin, was descended from an Alexandrian text which had been heavily corrected toward the Byzantne. He considers the result so valueless that he does not even normally cite it -- a mistake, I think, because while E does not often have correct readings not found in one of the earlier types of text, it is fairly important for the history of the Byzantine text.

Other Symbols Used for this Manuscript

von Soden: α1001


Note: As with all the major uncials, no attempt is made to compile a complete bibliography.

In 2015, the Bodleian made high-resolution scans of the entire manuscript available. As of this writing, these scans are at; curiously, there is also a scan of just four leaves at


First published, with many inaccuracies, by Hearne in 1715 (Sabatier used this transcription in his Old Latin edition). Also collated by Tischendorf. Ropes and Clark also studied the manuscript in detail. Finally, if it can be found, there is a Ph.D. dissertation by O. Kenneth Walther, Codex Laudianus G 35: A Re-Examination of the Manuscript, Including a Reproduction of the Text and an Accompanying Commentary. The manuscript will also be published in the Acts volume of Das Neue Testament auf Papyrus.

Sample Plates:
Metzger, Manuscripts of the Greek Bible (1 page)
Metzger, The Text of the New Testament (1 page -- a smaller version of the above)
Sir Frederick Kenyon & A. W. Adams, Our Bible and the Ancient Manuscripts (that same page again)
Margaret T. Gibson, The Bible in the Latin West (1 page, a different one)

Editions which cite:
Cited in all editions since Tischendorf

Other Works:

Manuscript Fe (09)

Utrecht, University Library MS. 1. Contains the Gospels with significant lacunae, especially in Luke; the damage has been progressive, and some leaves have been lost since Wettstein saw it in 1730. (Between 1730 and 1830 it was in private hands, and was unbound, with the leaves becoming disordered and torn.) As it stands now, it begins with Matt. 9:1 (though in Wettstein's time it apparently started at 7:6); it also lacks Matt. 12:1-44, 13:55-14:9, 15:20-31, 20:18-21:5. In addition, SQE lists a lacuna at 24:13-15, but Scrivener says these verses are extant, and Bob Relyea points out that they can be found in the online transcription at Mark lacks 1:43-2:8, 2:23-3:5, 11:6-26, 14:54-15:5, 15:39-16:19, John 3:5-14, 4:23-38, 5:18-38, 6:39-63, 7:28-8:10, 10:32-11:3, 12:14-25, 13:34-end. Luke is in even worse shape; Scrivener reports that there are 24 different lacunae, and SQE does not even bother collating the manuscript in that book. Dated paleographically to about the ninth century (so Tischendorf, von Soden, Aland; Tregelles preferred the tenth century). It has the Ammonian sections but not the Eusebian references; otherwise it has all the features of late uncials, including accents and breathings. The text is definitely Byzantine; the Alands list it as Category V; von Soden lists it as Ki. Wisse's classification doesn't mean much in this case; he lists F as Kmix in Luke 1, but it is defective for the other two chapters. In all likelihood it is actually either Kx or Ki (what Wisse would call Kx Cluster Ω). The date of the manuscript makes it potentially important for the history of the Byzantine text, but the large number of lacunae significantly reduce its value; it would have been much better had another Byzantine manuscript (preferably one of a type other than Kx) been used in the apparatus of SQE and UBS4.

Manuscript Fa

This Symbol No Longer Used. This symbol was given by Wettstein to a manuscript of the Septuagint (M of sixth or seventh century) in which he found, in the original hand, a marginal text containing Acts 9:24-25. Uncials of the Acts were few enough that Wettstein included this as an uncial witness to Acts. Detailed examination later showed it to include several other New Testament passages. The complete list is: Matt. 5:48, 12:48, 27:25, Luke 1:42, 2:24, 23:21, John 5:35, 6:53, 55, Acts 4:33, 34, 9:24, 25, 10:13, 15, 22:22, 1 Cor. 7:39, 11:29, 2 Cor. 3:13, 9:7, 11:33, Gal. 4:21, 22, Col. 2:16, 17, Heb. 10:26. When Gregory regularized the catalog of uncials, however, he eliminated Fa on the grounds that it was not a continuous-text manuscript; it has not been cited since.

Manuscript Fp (010)

Location/Catalog Number

Cambridge, Trinity College B.17.1. (Most Biblical catalogs seem to call it B.XVII.1, but the Cambridge catalog compiled by M. R. James uses arabic numerals.) Codex Augiensis, so-called because it comes from the monastery of Augia Dives in Lake Constance. The catalog prefix B indicates that it is a theological manuscript, and the number 17 indicates that it came to the library from the Bentley collection. Most of the items in the B.17 group are printed books; this is the most important exception.


Greek/Latin diglot. The Greek lacks Romans 1:1-3:19, 1 Cor. 3:8-16, 6:7-14, Col. 2:1-8, Philem. 21-25, Hebrews. Save for the lacuna in Romans, all of these defects are supplied in the Latin. All the omissions save that in Romans are also paralleled in the sister manuscript Gp. The clear conclusion (also supported, e.g., by the pagination) is that both F and G were copied from a manuscript which omitted the passages in 1 Corinthians through Hebrews, but that the Romans passage (or most of it) was originally present in the source manuscript and has now been lost. (Note: The general run of the Latin is not the Vulgate, but Hebrews does have a Vulgate text; in addition; NA26 lists the Latin sections not paralleled in the Greek as being supplements, but this seems to be based not on the nature of the writing but on its relationship with the Greek.)

The Greek and Latin are in parallel columns on the page, with the Greek in the inner column (closer to the spine of the book) and the Latin in the outer. Where the Greek fails, the Latin occupies the full width of the page -- a curious fact which shows that the scribe knew there were defects in the Greek text. It is not obvious why these were allowed to stand.


Dated paleographically to the ninth century. Greek and Latin are both beautifully written, but the Greek quite incompetently; it is clear that the scribe was more comfortable in Latin (the most obvious example of this is word division: the exemplar clearly did not have word divisions, and while the scribe put in points to show divisions, they are very often in error. Another example is his handling of Ο and Ω; these vowels are often confused -- a trait I notice that I share as a user of the Roman alphabet). The scribe was almost certainly a native speaker of German or a related language (I have seen the Latin letterforms called "Anglo-Saxon") -- or perhaps Irish, since there were many Irish monks at St. Gall, which was possibly the original home of the manuscript. It is often said that this manuscript was written by the same scribe as Δ, and it is also claimed that this scribe wrote the Basel Psalter and a copy of Horace at Bern.

Description and Text-type

The first and most obvious point about F is that it is an immediate relative of Gp, Codex Boernianus -- a fact noticed very early on by Bentley, who acquired the manuscript in 1718 (from L. C. Meig of Heidelberg). The resemblances are both textual (they agree almost absolutely) and physical (they have many of the same lacunae).

It is generally conceded that G, although less attractive, has the better text. For this reason, many editions cite G and not F. This fact has also led to some rather absurd speculation -- notably that F is a copy of G. This is not the case. The two manuscripts are not direct descendents of one another; rather, they have a recent common ancestor. It is not impossible that they are sisters, both derived from a somewhat defective Greek/Latin diglot. Even this is by no means certain, however. It is worth noting that F and G, while they have nearly identical Greek texts, do not have identical Latin texts. The Latin of G (known as g) is a strict interlinear translation of the Greek. F, however, has a parallel Latin version, and this version is not the same as the Latin of G. Rather, the Latin of F (known as f) is a modified Vulgate. As the Latin version does not exactly match the Greek, it seems likely that it has been conformed to an Old Latin version.

It is worth noting that both G and F are written without heavy correction by the scribes. This strongly implies that both were copying texts that lay before them, rather than editing their Latin sides to match the Greek. In other words, there was probably (note the word probably; this is simply logic, and not assured!) an ancestor before the scribe of G with an interlinear Latin, and an ancestor before the scribe of F with a parallel Latin, including the lacunae in the Greek. Since the ancestor of F/G probably did not contain both an interlinear and a parallel Latin, there is presumably an intermediate manuscript in one or the other case. Continuing the logic, it appears more likely that G is copied directly from the common exemplar than that F is -- had the exemplar resembled F, it is likely that G's interlinear Latin would more nearly resembled f. Thus the highest likelihood is not that F and G are sisters, but that they are no closer than aunt and niece, and it is possible that they are merely cousins of some degree. (Thus the tendency to cite only G in the critical apparatus, ignoring F, is to be deplored; there may well be readings where F preserves the family text better than G, though it seems clear that G is overall the better and more complete witness. The only significant scholars to disagree with this assessment seem to be the Alands, who -- in what can only be labelled an inexplicable classification -- list F as Category II, but G, and D for that matter, as Category III.)

The relationship with Codex Claromontanus (D) has also been a matter of discussion. I have seen stemma implying that F and G are descended from D, and others implying a common ancestor which was the parent of D. This too is absurd; there are simply too many major differences between the three (perhaps the best single example of this is the ending of Romans: D includes 16:25-27 at the end of that book, but F and G omit altogether). No one will deny that these three manuscripts form a text-type, but they are by no means immediate kin.

For the relationship between the "Western" text of Paul (the usual name given to the text of D F G and the Latin versions) to the "Western" text of Codex Bezae, see the entry on that manuscript and the entry on Codex Claromontanus.

Other Symbols Used for this Manuscript

von Soden: α1029


Note: As with all the major uncials, no attempt is made to compile a complete bibliography.


The basic work remains F. H. A. Scrivener, An Exact Transcript of Codex Augiensis. This is now available on Google Books. One may check this against the Pauline portion of Das Neue Testament auf Papyrus.

Sample Plates:

Editions which cite:
Because of its close similarity to G, most editions cite F only intermittently. The primary exceptions are Tischendorf and NA26-NA28.

Other Works:

Manuscript Ge (011)

London, British Museum Harley 5684 (a single leaf, taken from the codex by J. C. Wolff and given to Bentley, is in Cambridge, Trinity College B.17.20; it contains Matt. 5:29-31, 39-43). This manuscript number also refers to a portion of He (013)). Called codex Wolfii A after the first important owner (though the manuscript in fact originated in the east, and was brought to the west by Andrew Erasmus Seidel), or alternately Codex Harleianus after its present location. Contains the Gospels with lacunae; lacks Matt. 1:1-6:6 (a small part of this, be it noted, being included on the Cambridge leaf), 7:25-8:9, 8:23-9:2, 28:18-Mark 1:13, Mark 14:19-25, Luke 1:1-13, 5:4-7:3, 8:46-9:5, 12:27-41, 24:41-end, John 18:5-19, 19:4-27. Portions of this damage were rectified by later hands: One scribe supplied Matt. 28:18-Mark 1:8 and John 18:5-19, another Luke 12:27-41. Earlier editors, such as Scrivener, dated the manuscript to the tenth century, but the Alands have lowered this to the ninth century. (Part of the problem may be the scribe's coarse writing, small uncials drawn with a pen much too large for the chosen size; Scrivener gives a facsimile showing irregular accents and breathings and demonstrating the ugly writing style.) There is more agreement about the text; all would agree that it is Byzantine. Von Soden classified it as Ki, and the Alands list it as Category V; Wisse describes it as Kx. There are hints of something more, though; even the Alands' figures show G as having a relatively high number of non-Byzantine, non-UBS readings (a total of 21, out of 288 readings tested; by way of comparison, E has 9 such "s" readings out of 326 readings examined, H has 7 in 265 test readings; M has 12 in 327; S has 12 in 327). It may be simply that the manuscript is carelessly written, but in working through the apparatus of SQE, I was struck by how many of the non-Byzantine readings seemed to be "Cæsarean." Great care, of course, must be taken in dealing with the "Cæsarean" text, as its very existence is questionable and the text has never been properly defined -- but this pattern of readings may imply that the handful of non-Byzantine readings, few though they are, are not errors and may have some slight value. (I repeat, however, that this is based solely on my subjective examination of the SQE critical apparatus; the matter needs to be examined in detail before this is taken as fact.)
The British Library has now released high-resolution scans, at

Manuscript Gp (012)

Location/Catalog Number

Dresden, Sächsiche Landesbibliothek A 145b. Codex Boernerianus, so-called because it was formerly owned by C. F. Börner of Leipzig.


Greek/Latin interlinear diglot, lacking Romans 1:1-4, 2:17-24, 1 Cor. 3:8-16, 6:7-14, Col. 2:1-8, Philem. 21-25, Hebrews. These defects were clearly present in the exemplar as well, as all are shared by Fp, which is universally believed to share a recent common source with G. The manuscript was damaged in World War II; the above list describes its contents before this damage, since it had been fully examined and photographed by then.

It has been argued that G and the gospel manuscript Δ were originally part of the same volume; they are are similarly written, both are interlinear diglots, and the pages are exactly the same size. We should note, though, that not all commentators are convinced by these arguments. There is at least one counter-argument, though it is textual rather than physical or paleographic: The text of Δ is Byzantine, with Alexandrian elements in Mark; the text of G is purely and simply "Western." And while there are genuine physical similarities between the manuscripts (probably because they both derive from Saint Gall), Δ appears rather finer and fancier (though this may simply be because the Gospels are usually given finer treatment).


Dated paleographically to the ninth century by all authorities. The manuscript is written without accents or breathings, but with spaces between words (sometimes misplaced), in a stiff, awkward hand; the letterforms do not much resemble other manuscripts of the period (save Δ; while the two may not be part of the same volume, they are almost certainly from the same school as they resemble each other even in small details of preparation). The Latin interlinear is written above the Greek, with the Greek lettering fairly large and the Latin extremely small. There is some slight decoration in colour, though not nearly as much as in Δ. A dot and an enlarged letter marks the beginning of phrases. It has been theorized (probably correctly) that the exemplar of G was written in some sort of sense lines, as the separate phrases and enlarged letters are almost evenly spaced.

A peculiar fact about the manuscript is that it contains (on folio 23) some verses in (archaic) Irish Gaelic referring to a pilgrimage to Rome. The writing in these verses appears similar to that of the Latin; the original scribe may have been Irish (many Irish monks settled in Saint Gall). But this point has not, as far as I know, been proved. A similar hand added six sundry notes condemning the Greek.

Another fact is that the scribe doesn't seem to have been accustomed to the type of text he copied. G (along with F and 629) omits Romans 16:25-27 -- but the scribe of G left room for the verses after 14:23. There is no sign of this in F; the simplest explanation (though by no means sure!) is that the scribe of G was more accustomed to a text containing those verses there.

Description and Text-type

In the entry on Fp, we noted the similarities between F and G. Not only are they both Greek/Latin diglots, but they have the same lacunae (with the exception of the first part of Romans, where F is defective). The similarity is further confirmed by their texts. Scrivener, who collated both, lists 1,982 differences -- but breaks them down as 578 blunders of the scribe, 967 vowel changes (including itacisms), 166 instances of interchanged consonants, and 71 grammatical or orthographic differences, 32 instances of addition or omission of the article, and 168 instances of clear variants.

Like F, the word division is sometimes peculiar, implying that the two were copied from an exemplar without word divisions. The two do not use identical word divisions, however, meaning that they can hardly have been copied from one another. That neither is a copy of the other is confirmed by much additional evidence. The key fact, perhaps, is that the two are in completely different styles: F has a facing Latin text, G an interlinear, but both are copied without major corrections by the scribes, implying that both Greek and Latin texts were present in their current forms in the exemplars. Nor do the Latin versions match closely.

Of the two, G seems to be the more accurate overall (despite the much uglier writing). One often finds G cited to the exclusion of F. This is unfortunate, since both are needed to reconstruct the exemplar, but certainly G is the one to choose if only one is to be cited.

That F and G belong to the same text-type as Dp and the Old Latin versions need not be doubted. This type is generally called "Western," though no absolutely convincing proof has been offered that this is truly the same type as found in Codex Bezae in the gospels. The relationship between D, F, and G is somewhat involved; while F and G are cousins or closer (see the discussion in the entry on F), D is much more distant -- not really kin at all, except at the text-type level. (Some manuals show D as an uncle, or even a direct ancestor, of F and G, but this is extremely unlikely -- there are too many differences; consider, for instance, their various forms of the ending of Romans.) Examination seems to show that F and G have more minor divergences from the common Alexandrian and Byzantine text than D (indeed, F and G may be the most idiosyncratic of all manuscripts in this regard, adding, changing, and omitting articles, pronouns, and other secondary words almost at random). They may actually have fewer large variants than D, however (this position was first stated by Corssen in 1889; I came to the conclusion independently). Casual inspection also seems to imply that F and G fall slightly closer to 𝔓46 and B than does D.

The Latin side of G, known as g (Beuron 77), is less interesting than the Greek. As an interlinear, it has been heavily conformed to the Greek, though there probably was an independent Latin version behind it (and used as a crib). An interesting feature of g is that it sometimes has alternate rendering. Metzger cites an example from 1 Corinthians 3:2; the Greek text reads γαλα υμας εποτεισα (NA26 γαλα υμας εποτισα). The alternate readings are for υμας, where g reads vos vel vobis. It is at least possible that some of these alternate readings are places where the Latin reference edition used to compile g disagreed with the Greek text of G (particularly as there are instances where g does not match G at all).

Most classifications of G, of course, have closely followed the classification of F -- Von Soden, e.g., lists both as Ia1, in the same group as D (and, we must note, some unrelated minuscules). The one curiosity is the Alands, who place G in Category III but F in Category II. (For further discussion, see the entry on Fp).

Other Symbols Used for this Manuscript

von Soden: α1028


Note: As with all the major uncials, no attempt is made to compile a complete bibliography.

First published by Matthei, in an edition said to be highly accurate but, of course, now nearly inaccessible. Scrivener published a detailed collation against F in F. H. A. Scrivener, An Exact Transcript of Codex Augiensis. One may check this against the Pauline portion of Das Neue Testament auf Papyrus.

Sample Plates:
Metzger, Manuscripts of the Greek Bible (1 plate)
Aland & Aland, The Text of the New Testament (1 plate)

Editions which cite:
Cited in all editions since Tischendorf, and some before.

Other Works:

Manuscript He (013)

Primarily at Hamburg, University Library, Cod. 91 in scrin.; one folio (formerly in the possession of Bentley, who never returned it to its rightful owner) is in Cambridge, Trinity College Library B.17.20; also B.17.21. (The catalog prefix B indicates that it is a theological manuscript, and the number 17 indicates that it came to the library from the Bentley collection. Most of the items in the B.17 group are printed books, but there are some manuscripts as well. Most catalogs seem to refer to the manuscripts of the Trinity B collection by Roman numerals, e.g. B.XVII.21, but the catalog of M. R. James uses arabic numerals.)
Called Codex Seidelianus II (after the man who brought it from the east) or Wolfii B after the first important owner. Contains the Gospels with major lacunae; lacks Matt. 1:1-15:30, 25:33-26:3, Mark 1:32-2:4, 15:44-16:14, Luke 5:18-32, 6:8-22, 10:2-19, John 9:30-10:25, 18:2-18, 20:12-25. It may never have been fully finished; it contains the Ammonian sections but not the Eusebian canons. Dated by all authorities to the ninth century. The text is definitely Byzantine -- though Scrivener reports that some esteemed H as having somewhat greater value than G, meaning probably that it was a little less Byzanine. This does not seem to be born out by the evidence; the Alands, naturally, list H as Category V, but also show it with a very low number of non-Byzantine readings (only 9 readings in either Category 2 or Category S; G, by contrast, has 25). My own informal experience bears this out; H has very few non-Byzantine readings. Wisse describes H as Kx. Von Soden (who designated it as ε88) listed it as Ki, a group which Wisse considers part of Kx.

Manuscript Ha (014)

Modena, Biblioteca Estense, G.196 (II.G.3), folios 9-51 (the remaining folios, which contain the Catholic and Pauline Epistles, are now designated 2125). Codex Mutinensis. The uncial portion contains Acts only, and is defective for Acts 1:1-5:28, 9:39-10:19, 13:36-14:3, 27:4-28:31. The first three lacunae have been supplied in a minsucule hand (formerly designated h), the last by an uncial hand. Overall, the manuscript is dated to the ninth century, and Burgon thought the minuscule supplements to be "scarcely later," while the uncial supplement containing 27:4-28:31 has been dated to the eleventh century. The additional material found in 2125 was dated to the twelfth century by Scrivener, but the Alands give a tenth century date. There is little to be said about the text, save that it is Byzantine; the Alands list H as Category V, while Von Soden (who gave the manuscript the symbol α6) lists it as K with some I influence.

Manuscript HP (015)

Location/Catalog Number

41 folios distributed among eight numbers in seven libraries in six cities: 8 leaves at the Great Lavra on Mount Athos; 3 leaves in Kiev (Nat.-Bibl. Petrov 26); 3 leaves in St. Petersburg (Bibl. Gr. 14); 3 leaves in Moscow (Hist. Mus. 563 and Ross. Gosud. Bibl. Gr. 166,1); 22 leaves in Paris (Bibl. Nat. Suppl. Gr. 1074 and Bibl. Nat. Coislin 202; the latter number also describing 94); 2 leaves at Turin (Bibl. Naz. A.1). Collectively known as Codex Coislinianus. Many of the surviving leaves are from bindings and are badly damaged or rubbed.


H presumably originally contained the entire Pauline corpus. At some point it was disassembled and the leaves used to bind other books (the Athos leaves were placed in the binding of a book dated 1218 by a monk named Makarius, although we cannot prove either that this was the date when H was disassembled or that the binding dates from 1218). The surviving leaves contain 1 Cor. 10:22-29, 11:9-16; 2 Cor. 4:2-7, 10:5-11:8, 11:12-12:4; Gal. 1:1-10, 2:9-17, 4:30-5:5; Col. 1:26-2:8, 2:20-3:11; 1 Thes. 2:9-13, 4:5-11; 1 Tim. 1:7-2:13, 3:7-13, 6:9-13; 2 Tim. 2:1-9; Titus 1:1-3, 1:15-2:5, 3:13-15; Hebrews 1:3-8, 2:11-16, 3:13-18, 4:12-15, 10:1-7, 10:32-38, 12:10-15, 13:24-25.


Dated paleographically to the sixth century. H is written on parchment in extremely large uncials (over 1.5 cm in height), one column per page. The text is written stichometrically. A later hand added accents and breathings to the text although not to the subscriptions of the books.

Description and Text-type

Aland and Aland list H as Category III; von Soden classifies it among the Alexandrian witnesses. From the stichometric arrangement of the lines, as well as the subscriptions to the various books (written in vermillion), H would appear to be based on the Euthalian edition of Paul -- probably the earliest example of this type.

A footnote to Titus claims that the text was corrected based on a manuscript written by Pamphilius. This is either an error or refers to the exemplar used for H; such corrections as we find in the text are almost always Byzantine (see the entry on correctors).

Overall, the text of H does appear to be Alexandrian, but with much Byzantine mixture. It is probably of more note for the history of the Euthalian text than the biblical text as a whole.

Other Symbols Used for this Manuscript

von Soden: α1022



Sample Plates:
Complete color scans of the Paris portion are now available at

Editions which cite:
Cited in all editions since Tischendorf.

Other Works:
M. H. Omont, Notice sur un très ancien manuscrit grec en onciales des Epîtres de Paul, conservé à la Bibliothèque Nationale, 1889 (a partial edition, based on materials available at the time).

Manuscript I (016)

Washington, Freer Gallery of Art, 06.275. Called Codex Freerianus or Codex Washingtonensis. Contains fragments of the Pauline Epistles. The extant fragments consists of (portions of) 1 Cor. 10:29, 11:9-10, 18-19, 26-27, 12:3-4, 27-28, 14:12-13, 22, 32-33, 15:3, 15, 27-28, 38-39, 59-50, 16:1-2, 12-13; 2 Cor. 1:1, 9, 16-17, 2:3-4, 14, 3:6-7, 16-17, 4:6-7, 16-17, 5:8-10, 17-18, 6:6-8, 16-18, 7:7-8, 13-14, 8:6-7, 14-17, 8:24-9:1, 9:7-8, 9:15-10:1, 10:8-10, 10:17-11:2, 11:9-10, 20-21, 28-29, 12:6-7, 14-15, 13:1-2, 10-11; Gal. 1:1-3, 11-13, 1:22-2:1, 2:8-9, 16-17, 3:6-8, 16-17, 24-28, 4:8-10, 20-23; Eph. 2:15-18, 3:6-8, 18-20, 4:9-11, 17-19, 28-30, 5:6-11, 20-24, 5:32-6:1, 6:10-12, 19-21; Phil. 1:1-4, 11-13, 20-23, 2:1-3, 12-14, 25-27, 3:4-6, 14-17, 4:3-6, 13-15; Col. 1:1-4, 10-12, 20-22, 27-29, 2:7-9, 16-19, 3:5-8, 15-17, 3:25-4:2, 4:11-13; 1 Thes. 1:1-2, 9-10, 2:7-9, 14-16, 3:2-5, 11-13, 4:7-10, 4:16-5:1, 5:9-12, 23-27; 2 Thes. 1:1-3, 10-11, 2:5-8, 14-17, 3:8-10; 1 Tim. 1:1-3, 10-13, 1:19-2:1, 2:9-13, 3:7-9, 4:1-3, 10-13, 5:5-9, 16-19, 6:1-2, 9-11, 17-19; 2 Tim. 1:1-3, 10-12, 2:2-5, 14-16, 22-24, 3:6-8, 3:16-4:1, 4:8-10, 18-20; Tit. 1:1-3, 10-11, 2:4-6, 14-15, 3:8-9; Philem. 1-3, 14-16; Heb. 1:1-3, 9-12, 2:4-7, 12-14, 3:4-6, 14-16, 4:3-6, 12-14, 5:5-7, 6:1-3, 10-13, 6:20-7:2, 7:7-11, 18-20, 7:27-8:1, 8:7-9, 9:1-4, 9-11, 16-19, 25-27, 10:5-8, 16-18, 26-29, 35-38, 11:6-7, 12-15, 22-24, 31-33, 11:38-12:1, 12:7-9, 16-18, 25-27, 13:7-9, 16-18, 23-25. In other words, it has at least part of every Pauline book from 1 Corinthians on. The surviving pages represent 84 leaves -- many fragmentary; often only a few lines can be read due to damage to the surviving pages. Only portions of the pages have survived, and even these were stuck together badly, adding to the damage. In general we are left with the outer portions of pages, not the central area.

Based on the numbering of the quires, it is thought that the original had either 208 or 212 leaves. Hebrews followed 2 Thessalonians. Since 210 or so leaves would have been more than sufficient for Paul, it is likely that the manuscript originally contained the Acts and Catholic Epistles also.

Wayment believes that the manuscript was copied from dictation, with the lector who read the text changing at the beginning of Ephesians. Accepting that the extremely high rate of what appear to be errors of hearing implies a text from dictation, I wonder if the data is sufficient to the argument. Wayment's argument is based on singular readings: looking at the material through Galatians, he finds just eight unique readings, compared to 131 from Ephesians on. Given the amount of text surviving in the two sections, this is not as big a difference as it sounds -- the second half has roughly twice as many singular readings per unit of text. That's not a big enough change to be sure of exactly where the change came, but it does seem significant. However, I am not sure we can conclude that it is the result of a different lector. Suppose the scribe were tired -- or perhaps hungry. (E.g. the first part might have been written before the celebration of a fast and the second during the fast.) This doesn't make Wayment wrong; it simply means we must be cautious.

Wayment also believes the scribe quite fluent in Greek (unlike the scribes of some other important manuscripts like L), because even when he produces a singular reading, it is usually makes sense and is grammatically correct.

The manuscript is generally dated to the fifth century, though a few have suggested the sixth century instead, and Kenyon suggested the seventh. There is little doubt about the nature of the text; it is clearly Alexandrian. Von Soden (who designated it as α1041) lists it as type H, while the Alands place it in Category II, ascribing it to the Egyptian text. The Alands' own numbers, however, make this dubious; of the 34 readings of I, only one is purely Byzantine, while 22 agree with UBS against the Byzantine text; six agree with neither. While this is too small a sample to allow for absolute certainty, on its face it implies that I is not Category II but Category I, and Alexandrian, not a member of the later Egyptian text. Indeed, by the numbers, I is the most Alexandrian manuscript of Paul! Wayment seems to agree, considering its value exceptionally high. And my own checking indicates that I is the closest relative of ℵ in existence (and much closer to A C 33 than it is to 𝔓46 or B or 1739). This confirms the original assessment of Henry A. Sanders, who repaired it, collated it, and placed it closer to ℵ A 33 than to B. Kenyon also agrees with this assessment. Its fragmentary nature limits its usefulness, but where it exists, I deserves to be treated with all the respect accorded to ℵ or A -- indeed, possibly more than the latter. It is truly unfortunate that it is so badly damaged.

According to Kenneth W. Clark, it was "Found in the hands of a Gizeh dealer, Sheikh Ali Abdel Hai el Arabi, in the autumn of 1906 by Grenfell and Hunt, purchased Dec. 19, 1906, along with MS 3 [i.e. W] by the late Charles Lang Freer, of Detroit." For somewhat more about this purchase, see the entry on W.

A sample image showing just how much damage it has sustained can be found on the Freer|Sackler web site at

The status of I shows, incidentally, that the use of chemicals to restore ink wasn't the only way in which earlier scholars damaged manuscripts in an attempt to read them. The pages being all stuck together, there was no way to pry them apart except with a knife, which did far more damage than modern methods would have done.

Manuscript Ke (017)

Location/Catalog Number

Paris -- Bibliothèque Nationale Gr. 63. It was taken to Paris from Cyprus in 1673, and is called Codex Cyprius.


Contains the Gospels complete, although there are many instances of water damage at the top of the page that make the text hard to read. (Some of these have been overwritten in the manuscript, most have not.) The opening and closing leaves of the codex are in very bad shape, making it possible that something has been lost.


Dated paleographically to the ninth century (though scholars up to the time of Scholz sometimes suggested the eighth, and Mill thought it tenth century or later). K is written on parchment, one column per page. The scribe was named Basil, and the manuscript was bound by one Theodulos. Wettstein thought it had latinizing tendencies, but it seems clear that this is merely because it is part of Family rather than the main run of the Byzantine text. Scrivener says of the writing, "[It has] one column of about twenty-one lines per page, but the handwriting is irregular and varies much in size. A single point being often found where sense does not require it, this codex has been thought to have been copied from an older one arranged in στιχοι.... The subscriptions, τιτλλοι, the sections, and indices of the κεφαλια of the last three gospels are believed to be the work of a later hand: the Eusebian canons are absent. The breathings and accents are primâ manu, but are often omitted or incorrectly placed. Itacisms and permutations of consonants are very frequent...." Scholz's opinion was that, although it had most of the accents expected of its date, many are incorrectly placed, and the breathings are confused (Scholz suspected it derived from an ancestor without these symbols); we also see some shuffling of vowels and consonants in words.

Description and Text-type

Recognized from a very early date as Byzantine, and still so listed today (so, e.g., Aland and Aland, who include it in Category V). Von Soden classified it as Ika, i.e. Family Π. This has been confirmed by all who have investigated the matter, most recently by Wisse (who places K in the Πa group in all three tested chapters of Luke, and calls it a core member of the group).

Wisse distinguishes two groups within Family Π -- Πa and Πb. Of these, Πa is more distinct and has more differences from the Byzantine bulk Kx. Among the more important members of this group are K itself, Π, 1079, and 1546. A (which is, of course, the earliest substantial Byzantine witness) is a diverging member of this group. The case can thus be made that K belongs to the oldest family of the Byzantine text -- and it is the oldest complete witness to this text.

Other Symbols Used for this Manuscript

von Soden: ε71



Sample Plates:

Editions which cite:

Cited by Tischendorf (who also collated it).
Cited by von Soden, Merk, and Bover.
Cited as a secondary witness in NA26 and NA27, but not in SQE13
Cited in UBS3 but not UBS4

Other Works:
All of the following pertain to Family Π, and so include information on K as well (although the works of Geerlings are sometimes guilty of dubious methodology):
Jacob Geerlings, Family Π in John, Studies & Documents 23, 1963
Jacob Geerlings, Family Π in Luke, Studies & Documents 22, 1962
Jacob Geerlings, Family Π in Matthew, Studies & Documents 24, 1964
Silva Lake, Family Π and the Codex Alexandrinus: The Text According to Mark, Studies & Documents 5, 1937

Manuscript Kap (018)

Location/Catalog Number

Moscow -- Historical Museum V.93, S.97. Originally from Mount Athos.


Contains the Catholic Epistles complete and Paul almost complete (lacks Romans 10:18-1 Corinthians 6:13; 1 Corinthians 8:8-11). Includes a marginal commentary.


Dated paleographically to the ninth century. K is written on parchment, two columns per page.

Description and Text-type

Von Soden classifies K as I1 in Paul and Aπρ1 in the Catholics. This is based, however, on the commentary (being that of John of Damascus in Paul and, according to von Soden, that of Andreas in the Catholics). The text is correctly described by Aland and Aland as Category V (i.e. purely Byzantine).

Within the Byzantine tradition, K forms a pair with 0151. The two may be sisters; certainly they are very closely related. Taking the book of Galatians as an example, we find 279 variants which can count at least two papyri or uncials on each side. K and 0151 agree on 263 of these. (In addition, K has seven singular readings and 0151 has ten.) Of these 263 agreements, seven are found only in these two manuscripts (a very high rate of subsingular agreement for Byzantine manuscripts).

Even their sixteen disagreements are suggestive:

VerseK reads0151 reads
1:22 της ταις
2:4 καταδουλωσωνται καταδουλωσονται
3:8 σοι συ
3:19 ω ο
3:22 - τα
4:4 γεννομενον εκ γενομενον εκ vid
4:6 κραζων κραζον
4:7 αλλα αλλ
5:14 σεαυτον εαυτον
5:26a γινομεθα γινωμεθα
5:26b αλληλοις αλληλους
6:4 εαυτου αυτου
6:8 εαυτου αυτου
6:9 θερισομεν θερισωμεν
6:10 εργασωμεθα εργασομεθα
6:13 καυχησωνται καυχησονται

Thus every difference between the two is trivial, usually revolving around vowel sounds. In this list there is not one instance of a reading that is clearly of genetic significance. In all likelihood these two commentary manuscripts descend from a common ancestor at a distance of no more than a handful of generations. It is unlikely, however, that one is copied from the other, since both have singular readings.

Other Symbols Used for this Manuscript

von Soden: I1 (Paul); Aπρ1 (Cath)
Matthei's g
Scholz's 102a, 117p



Sample Plates:
Metzger, The Text of the New Testament (1 page)

Editions which cite:
Cited in all editions since Tischendorf (though Nestle cites it only silently).

Other Works:

Manuscript Le (019)

Location/Catalog Number

Paris, National Library Greek 62. Codex Regius.


Contains the four Gospels with small lacunae: Now lacks Matt. 4:22-5:14, 28:17-end, Mark 10:16-30, 15:2-20, John 21:15-end. Portions of the remainder have been rendered difficult to read by damp.


Dated paleographically to the eighth century (though some early critics placed it in the ninth century); it is, by general consent, the most important manuscript of that period. The manuscript is written in a fairly firm, if clearly late, hand, but the scribe was not especially competent. Errors in the text are common; errors in externals perhaps even more common. Scrivener notes that "The breathings and accents are often deficient, often added wrongly, and placed throughout without rule or propriety. The apostrophus is common, and frequently out of place; the points for stops are quite irregular...." Spaces between words are infrequent, and iota subscriptum and postscriptum are said to be "entirely wanting." The manuscript contains many ornamentations, but they are not regarded as attractive (Scrivener calls them "in questionable taste"). In addition, the lectionary apparatus and Eusebian material is included, but the number of errors in the latter may indicate that the scribe did not understand their purpose. There are also occasional marginal comments on the text (some even stand in the text, such as that on the variant endings of Mark).

It seems likely that the scribe was an Egyptian, more used to writing Coptic than Greek.

Description and Text-type

When Hort defined his text-types, he described an "Alexandrian" text which was basically the "neutral" text with some grammatical corrections. Hort could not point to a single pure witness, but the closest he came was L.

L is fascinating because, among the late uncials of the Gospels, it is far and away the least Byzantine. If having an Alexandrian text is taken as a measure of quality, L is probably the fourth-best substantial manuscript of the Gospels, trailing only 𝔓75, B, and ℵ.

L is not without a Byzantine element; the first half of Matthew agrees almost entirely with the Majority Text. But this element fades toward the end of Matthew, and the rest is quite different. (The logical conclusion is that the ancestor of L was corrected toward the Byzantine standard, but that the corrector gave up somewhere in Matthew. This phenomenon -- revisor fatigue -- is not unusual; we see something similar in manuscripts such as 579 and 1241). From that point on, L has mostly Alexandrian readings, although there are some readings of other sorts. Some are Byzantine; others seem to be simply the sorts of readings that crept into the tradition with time. (Hort would call these readings Alexandrian, and the Alands have labelled this late phase of the Alexandrian text "Egyptian," but there is no real reason to think that this is in any sense a separate text-type. It's simply a text-type which has undergone continuous mixture and corruption. L may fairly be called a Late Alexandrian manuscript, but to call it a member of a "Late Alexandrian" or "Egyptian" text-type goes far beyond the available evidence.) As between B and ℵ, L is clearly closer to the former; L is obviously descended from a manuscript in the 𝔓75/B phase of the Alexandrian text.

The exact point at which L shifts from primarily Byzantine to primarily Alexandrian has been disputed. Some have said that all of Matthew is Byzantine; this is clearly false. My data put the change around chapter 20, but that was based on checking blocks of readings; it wasn't designed to find an exact point of change. The most detailed examination is probably Vincent Broman's; he compared L's text to the Byzantine and Alexandrian types (using Pierpont/Robinson as the standard for the former and the UBS text as a standard for the latter, while admitting the inadequacy of the latter). It is his belief that the change comes at Matthew 17:26, and is abrupt: He finds 14 straight Byzantine readings before the break, and eight straight Alexandrian readings after.

The single most significant reading in L is certainly the ending of Mark. L is the first important Greek manuscript to include both the longer ending (Mark 16:9-20) and the so-called "shorter ending." Both, of course, clearly predate L (the shorter ending is found in k, some Coptic manuscripts, and the margin of the Harklean Syriac, as well as in the uncial fragments 083 and 099; the longer ending is obviously ancient), but L is the earliest Greek witness to the shorter ending whose text-type we can exactly fix. The existence of alternate endings in this manuscript clearly indicates that the reading is not an original part of the Alexandrian text -- in other words, its omission in B and ℵ is not casual.

L has many other readings which indicate its non-Byzantine nature. It omits, for instance, Mark 7:16, Luke 11:2b, c, John 5:3b (although it includes 5:4), 7:53-8:11. These facts all combine to confirm the various classifications of the manuscript: Von Soden listed it as H (and listing it as the seventh H witness, implying that he regarded it as one of the better manuscripts of the type); Wisse lists it as a core member of Group B; the Alands list it as Category II (meaning, in effect, Alexandrian with some Byzantine mixture).

Other Symbols Used for this Manuscript

von Soden: ε56


Note: As with all the major uncials, no attempt is made to compile a complete bibliography.


Published by Tischendorf in Monumenta sacra inedita (1846). There is a strong need for a modern edition using all the current tools of scholarship.

Sample Plates:

Aland & Aland, The Text of the New Testament (1 plate)

Editions which cite:
First cited, imperfectly, by Stephanus (as η), and cited in almost every edition since.

Other Works:

Manuscript Lap (020)

Rome, Biblioteca Angelica 39. Called Codex Angelicus after the library. Von Soden's α5. Contains the Acts, Paul, and the Catholics. Acts lacks 1:1-8:10; the Catholics are complete; Paul lacks Hebrews 13:10-end. Scrivener says that is is "of a date not earlier than the middle of the ninth century," although most moderns accept a ninth century date for it. Textually, about the only thing it appears noteworthy for is its complete lack of noteworthiness. The Alands assign it to Category V (Byzantine), and I have no quarrel with that whatsoever; it appears to be among the most Byzantine of manuscripts. Von Soden also classified it as K (Byzantine), though with a few I readings. If the manuscript has any real significance, it is simply because it is among the very earliest purely Byzantine manuscripts of the books it contains.

Manuscript Me (021)

Paris, National Library Greek 48. Called Codex Campianus after Abbé François de Camps, who gave it to Louis XIV in 1707. Contains the Gospels complete; it also has the Eusebian prefatory matter and Hippolytus's chronology of the gospels. Dated paleographically to the ninth century by all authorities. Both the manuscript and the writing are small and neat (though the writing would have been more legible had a finer pen been used); some of the outer pages are badly scuffed, but the Gospels themselves are usually easy to read except for the final pages of John. The margins, however, are crowded, with lectionary, notes, Eusebian materials, and more; it appears that the edges have been significantly trimmed, since some of the marginal materials bleed into the edge. There are also many mold marks and stains. It is interesting to note the number of languages used in the margins; there are said to be Greek, Arabic, and Slavonic comments. The text, in addition to accents and breathings, has neumes for singing written in red. Some of the introductory illustrations are in a (corrosive) green, but most interior illustrations are restricted to red and blue (plus perhaps a third color that is now very hard to tell from the color of the parchment); they are not especially attractive or well-drawn.

The text of M is Byzantine but interesting; it is definitely not part of Kx. The Alands classify it (correctly, by their standards) as Category V, but the situation is more complicated than that. It was Von Soden who first tried to classify M (though earlier scholars, such as Scrivener, thought its text interesting and valuable). Soden categorized M as part of his Iφr group; other members of this group include but are not limited to 27 71 692 1194 (several of these only in certain books; these are the witnesses von Soden cited regularly; in addition, von Soden recognized subgroups within this type but did not really distinguish them in his apparatus). The Iφ groups as a whole are an interesting lot; Iφa is what Streeter calls Family 1424; Iφ;b has never received much attention; Iφc includes such noteworthy manuscripts as 945 and 1010.

This classification has, however, been heavily modified by Wisse. Wisse concedes the existence of a Byzantine sub-type including M and related manuscripts, but completely redoes the grouping. Although calling them the "M groups," M itself is listed as a diverging member of Group M27; the other M groups include M10, M106, M350, M609, and M1386, along with a variety of clusters and pairs. Wisse believes the M groups have kinship with the Π groups.

Color scans of the manuscript are now available at the Paris Library web site at

Manuscript N (022)

Location/Catalog Number

Codex Purpureus. Various libraries: Saint Petersburg, Russian National Library Gr. 537 (182 folios); Patmos, Ioannou 67 (33 folios); London, British Library Cotton Titus C. XV (4 folios); Vienna, National Library Gr. 31 (2 folios); Athens, Byz. Museum Frg. 21 (1 folio); Lerma (Spinola Collection) (1 folio); Rome, Bibl. Vat. Gr. 2305 (6 folios); New York, Pierpont Morgan Lib. 874 (1 folio); Salonika, Byz. Museum Ms. 1 (1 folio). (Total of 231 folios, representing roughly half of the original manuscript.)


Contains the Gospels with very many lacunae: Matt. 1:1-24, 2:7-20, 3:4-6:24, 7:15-8:1, 8:24-31, 10:28-11:3, 12:40-13:4, 13:33-41, 14:6-22, 15:14-31, 16:7-18:5, 18:26-19:6, 19:13-20:6, 21:19-26:57, 26:65-27:26, 26:34-end; Mark 1:1-5:20. 7:4-20, 8:32-9:1, 10:43-11:7, 12:19-24:25, 15:23-33, 15:42-16:20; Luke 1:1-2:23, 4:3-19, 4:26-35, 4:42-5:12, 5:33-9:7, 9:21-28, 9:36-58, 10:4-12, 10:35-11:14, 11:23-12:12, 12:21-29, 18:32-19:17, 20:30-21:22, 22:49-57, 23:41-24:13, 24:21-39, 24:49-end; John 1:1-21, 1:39-2:6, 3:30-4:5, 5:3-10, 5:19-26, 6:49-57, 9:33-14:2, 14:11-15:14, 15:22-16:15, 20:23-25, 20:28-30, 21:20-end. It has been thought that it was originally broken up by Crusaders (so Metzger; Scrivener says this of Φ); certainly its career was exciting (Gregory reports how the Saint Petersburg portion, when it was still in Asia Minor, was stolen -- and recovered by a crowd of angry villagers).


Although in the seventeenth century some scholars claimed that the British portion was the oldest extant Greek manuscript, that distinction can no longer be maintained. Moderns date N paleographically to the sixth century. It is written on purple parchment in (now badly faded) silver ink, with certain of the nomina sacra in gold. The letters are very large (see the reduced sample in the section on uncial script), and are very regular in form; they seem to have been stamped on the page (though there are multiple stamps for the letters, and they are not uniform in size). There are two columns per page, with the columns containing only a dozen or so letters due to the large size of the print; there are typically sixteen lines per column. Scrivener/Miller say of the manuscript, "[T]he punctuation [is] quite as simple [as in A of the fifth century], being a single point (and that usually neglected) level with the top of the letter... and there is no space between words even after stops.... It exhibits strong Alexandrian forms... and not a few such itacisms as the change of ι and ει, αι and ε."

Description and Text-type

There is general agreement that N forms a group with the other sixth century purple uncials (O Σ Φ). Cronin believed that N O Σ are in fact sisters, copied from a single exemplar (Φ he believed to have some "Western" mixture). There is less agreement about the nature of this group. Von Soden classifies it as Iπ, but this really begs the question as it is simply another of those mixed I-K groups, and has no witnesses except the purple uncials. Streeter laid claim to the group as a weak witness to the "Cæsarean" text -- but of course Streeter insisted that everything not otherwise classified was "Cæsarean." In any case, studies of the group have been hindered by the fact that O contains only Matthew, while Σ Φ contain only Matthew and Mark. Thus only N represents the type in Luke and John, and passages where all four purple uncials exist are relatively few.

In recent times, Aland and Aland have described N as Category V (Byzantine). Wisse reports that it is mixed in Luke 20; there is, of course, no text of chapter 1 and very little of chapter 10.

All of these claims are slightly imprecise. N is much more Byzantine than anything else (about 80% of its readings seem to belong to that type), but by no means purely. It omits John 7:53-8:11, for instance, as well as Luke 22:43-44. There seems to be no pattern to the non-Byzantine readings, though; certainly they are not "Cæsarean" (N agrees with the Koridethi codex in only 31 of 44 non-Byzantine readings tested, with Family 1 in 26 of 34, and with Family 13 in 23 of 36; by contrast, it agrees with A in 20 of 24, with K in 16 of 21, and with Ψ in 29 of 32). The simplest conclusion is that N is mostly Byzantine with occasional surviving readings of all types.

Other Symbols Used for this Manuscript

von Soden: ε19


Since N came to light in so many pieces, there is no complete collation. H. S. Cronin published the text as it was known in 1899 (Texts and Studies volume 4). A few additional leaves have been published in the Journal of Biblical Literature by Stanley Rypins (lxxv, 1956).

Sample Plates:

Editions which cite:
Cited in NA26 and NA27 for the Gospels.
Cited by von Soden, Merk, and Bover for the Gospels.

Other Works:
The work of Cronin cited above (and its follow-up in JTS, July 1901) discusses the relationship between the purple uncials.
B. H. Streeter, The Four Gospels: A Study of Origins, 1924, discusses on pp. 575-577 his perceived relationship between the purple uncials and the "Cæsarean" text. This discussion shows at once the strengths and weaknesses of Streeter's method; since he equates the Textus Receptus entirely with the Byzantine text, almost any manuscript -- even one purely Byzantine! -- will show "Cæsarean" readings by this method.

Manuscript O (023)

Paris, National Library MS. suppl Gr. 1286. Known as Codex Sinoponensis. Von Soden's ε21. Matthew, with some lacunae. Dated paleographically to the sixth century. Written on purple parchment with silver (often gold) ink (now quite hard to read, especially in photographs, due to the discoloration of the purple). With elaborate paintings featuring excellent handling of light and dark contrasts. The text is generally thought to go with the other purple uncials (N Σ Φ); Cronin in fact thought N O Σ to be sisters. Von Soden classified this group as Iπ, although there is little question but that the type is mostly Byzantine; the Alands place it in Category V. Streeter naturally regarded it as Cæsarean, but very weakly so; even if the Cæsarean type exists, O is so mixed as to be of little use as a witness. The least complete of the purple uncials, it is rarely cited. Given its extremely large size, it has been suggested that it contained only Matthew. This makes it almost the last, if not the last, manuscript to contain only a single gospel. However, the possibility must be admitted that it was part of a multi-volume deluxe set.

Manuscript Pe (024)

Codex Guelpherbytanus A. Wolfenbüttel, Herzog August Bibliothek, (portions of) Weissenburg 66 (loose folios). Von Soden's ε33. Palimpsest, containing small portions of all four gospels (Matthew 1:11-21, 3:13-4:19, 10-7-19, 10:42-11:1, 13:40-50, 14:15-15:3, 15:29-39; Mark 1:2-11, 3:5-17, 14:13-24, 48-61, 15:12-27; Luke 1:1-13, 2:9-20, 6:21-42, 7:32-8:2, 8:31-50, 9:26-36, 10:36-11:4, 12:34-45, 14:14-25, 15:13-16:22, 18:13-39, 20:21-21:3, 22:3-16, 23:20-33, 23:45-24:1, 24:14-37; John 1:29-40, 2:13-25, 21:1-11). Like Q (026), it is part of the under-writing of a manuscript of Isidore of Seville. It is written in two columns per page, in large uncials. Tischendorf dated it to the sixth century, and in this the Alands agree. Von Soden assigns it to the I' group. Wisse can't do much with it, since it is entirely defective for two of his three chapters; he says it is mixed in Luke 20. The Alands call it Category V, i.e. Byzantine, although with only 43 readings, their sample is too small to be confident. But I know of no particularly noteworthy readings found in it.

Manuscript Papr (025)

Location/Catalog Number

Saint Petersburg, Russian National Library Gr. 225. Called Codex Porphyrianus after its former possessor, Bishop Porphyry.


Palimpsest, originally containing the Acts, Catholic Epistles, Paul, and the Apocalypse complete. Apart from the occasional letters obliterated by the upper writing (works of Euthalius), a number of leaves have been lost, including those containing Acts 1:1-2:13, Romans 2:16-3:4, 8:32-9:10, 11:23-12:1, 1 Cor. 7:15-17, 12:23-13:5, 14:23-39, 2 Cor. 2:13-16, Col. 3:16-4:8, 1 Thes. 3:5-4:17, 1 John 3:20-5:1, Jude 4-15, Rev. 16:12-17:1, 19:21-20:9, 22:6-end. Scrivener states that, in addition, James 2:12-21, 2 Pet. 1:20-2:5 are "barely legible." Presumably modern methods have made it more possible to read these sections, but they will be poorly cited in older editions. (Scrivener notes that it also contains "a few fragments of 4 Maccabees," but given that it is palimpsest, one may wonder if these are truly part of the same volume.)


Dated paleographically to the ninth century. Considering its date, it has a rather primitive appearance; accents and breathings are fairly rare. But it does have lectionary indications in the margin. The over-writing has been dated to 1301, though the writing itself appears more typical of the thirteenth century.

Description and Text-type

The text of P varies significantly from section to section. It is quite thoroughly Byzantine in Acts; this was recognized by Hort, supported by Von Soden (who lists it as with some I influence in that book), and confirmed by the Alands (who list it as Category V in Acts). Even a fairly casual examination will confirm this point.

The Apocalypse may also be regarded as Byzantine; the Alands again list P as a member of Category V. (Von Soden lists P as H with I influence, but his classifications in the Apocalypse are now all but completely ignored.) A number of older commentators followed Von Soden as viewing P as valuable -- but this is probably due to methodological difficulties. P is a witness to the Andreas type (according to Schmid), but it lacks the Andreas commentary and differs just enough from the Andreas type of the Textus Receptus as to cause a Byzantine manuscript to appear non-Byzantine. (This just reinforces the fact that we cannot use differences from the TR as a measure of quality.) Observers were probably further biased by the fact that P is an uncial, and with only a handful of substantial uncials of the Apocalypse (ℵ A C P 046), it is natural that its importance would be exaggerated.

The matter is more complex in Paul and the Catholic Epistles. Here P is clearly a mixed manuscript. The Alands make P more Alexandrian than Byzantine in Paul; by their tables, P has 87 readings which agree with UBS against the Byzantine text, plus 31 readings which agree with neither, while it has only 82 readings which agree with the Byzantine text against UBS. My experience in working over the readings in NA26, however, made it appear that P agrees with the Byzantine text at at least two-thirds of the points of variation. Both my numbers and the Alands' agree that P is more Byzantine than anything else in the Catholics -- according to Hort, it is entirely Byzantine in 1 Peter.

In Paul and the Catholics, the Alands list P as Category III, while Von Soden assesses it as H (Alexandrian). He also places it next to Ψ in his list of manuscripts cited, implying some degree of kinship. Speaking informally, there does appear to be some truth to this; while Ψ in Paul is much more Byzantine than P, it has a significant number of non-Byzantine readings in the last few books (particularly Hebrews), and in examining the readings, I seemed to see kinship between P and Ψ. This is only an opinion, however; I have not verified this statistically.

Other Symbols Used for this Manuscript

von Soden: α3


Published by Tischendorf in volumes v and vi of Monumenta sacra inedita; the only publication based on modern methods of decipherment is in the Das Neue Testament auf Papyrus series.

Sample Plates:

Editions which cite:
Cited in all editions since Tischendorf (though some do not cite it for Acts).

Other Works:

Manuscript Q (026)

Codex Guelpherbytanus B. Wolfenbüttel, Herzog August Bibliothek, (portions of) Weissenburg 64 (folios 194-201, 299, 302, 303, 304, 311). Von Soden's ε4. Palimpsest, containing small portions of Luke and John (Luke 4:34-5:4, 6:10-26, 12:6-43, 15:14-31, 17:34-18:15, 18:34-19:11, 19:47-20:17, 20:34-21:8, 22:27-46, 23:30-49; John 12:3-20, 14:3-22, with large parts even of these verses illegible). Dated paleographically to the fifth century. Assessments of the text of Q have varied widely. Von Soden listed it as H (Alexandrian) in John and I' in Luke (I' being a large and disjoint group containing many uncial fragments -- P Q R 074 090 0116 0130 0131 -- plus the Byzantine uncials Γ 047 and a number of minuscules which generally have not been regarded as noteworthy). The Alands list Q as Category V, and regard it as the first truly Byzantine text (it should be noted, however, that Q exists for only twelve of their sample readings -- too small a number for classification). Wisse reports it as Mixed, though due to lack of text he was only able to examine chapter 20. The real truth seems to fall somewhere between these assessments. Q is much more Byzantine than anything else -- but it is no more a purely Byzantine text than is A or R. It furnishes evidence that the Byzantine type was in existence in the fifth century, but not that it had reached its final form or that it was in any way dominant. Consider the Nestle apparatus: Listing only a limited number of variants, NA27 shows Q departing from the Byzantine text 54 times (in the space of 209 verses, many of them fragmentary) in Luke, and 16 times (in 38 verses) in John. Thus Q is perhaps 80% Byzantine (though even this may be exaggerated; Q seems to be heavily given to harmonization, and some of its agreements with the Byzantine text may be coincidental). The remaining text seems to agree with the later Alexandrian witnesses (L, 33, 579) more than anything else. Physically, Q is part of a large palimpsest containing also the fragments of Pe (024) and the Gothic version; the upper writing consists of Latin treatises of Isodore of Seville. It has the Ammonian Sections, but if the Eusebian Canons were supplied, they must have been written in a coloured ink which has not survived. (This is not impossible; the manuscript seems to have had some writings in vermillion which are now illegible and barely detectable, and the Eusebian numbers were intended to be written in color.) It has a handful of breathings, though they are not applied in any systematic way.

Manuscript R (027)

Location/Catalog Number

Codex Nitriensis. London. Catalog Number: British Museum Add. 17211. Originally from Egypt; brought to England in the 1840s from the convent of S. Mary Deipara in the Nitrian Desert.


Contains palismpsest fragments of Luke: Luke 1:1-13, 1:69-2:4, 2:16-27, 4:38-5:5, 5:25-6:8, 6:18-36, 6:39, 6:49-7:22, 7:44, 7:46, 7:47, 7:50, 8:1-3, 8:5-15, 8:25-9:1, 9:12-43, 10:3-16, 11:5-27, 12:4-15, 12:40-52, 13:26-14:1, 14:12-15:1, 15:13-16:16, 17:21-18:10, 18:22-20:20, 20:33-47, 21:12-22:6, 22:8-15, 22:42-56, 22:71-23:11, 23:38-51 (the above list is approximate; in some cases the manuscript is so hard to read that we cannot tell exactly where each portion ends). A second hand adds 15:19-21, but these verses are not generally cited. The Lukan text is preceded by chapter headings. Together these occupy the first 48 folios of the manuscript; the last five folios contain a fragment of Euclid's Elements. Many of the pages are intact, but some have lost their margins, resulting in some loss of text even in the verses which nominally survive.

The British Library photographs are curiously inconsistent; in some, the under-writing is clear and easy to read, in others, almost entirely obliterated. I wonder if the different pages were not subjected to different handling when the manuscript was being prepared for re-use. Given the dramatic differences in ink colour, I suspect a multi-spectrum analysis of the manuscripts could reveal additional readings which have not so far been read.


Dated paleographically to the sixth century. R is written on parchment, two columns per page. The hand is very large and clear, though Scrivener calls the letters "somewhat irregular and straggling," and notes that "the punctuation is effected by a single point almost level with the tops of the letters, as in Cod. N. The pseudo-Ammonian sections are there without the Eusebian canons." In the eighth or ninth century the manuscript was overwritten with a Syriac text of Severus of Antioch against Johannes Grammaticus. (R was not the only manuscript demolished to hold Severus's text; a manuscript of the Iliad was used as well as the fragment of Euclid mentioned above.) The upper writing was copied by Simeon, a recluse the monastery of Mar Simeon of Kartamin.
The Euclid is not in the same hand as Luke; the writing is smaller, and has a distinct slope to the right, as opposed to the almost entirely vertical writing of Luke. Looking at it, I was immediately reminded of W, although I have not done a comparison of the letterforms.

Description and Text-type

Assessments of R over the years have varied. Hort says of it (§209, p. 153) that it is mixed, but has "a large proportion of Pre-Syrian [i.e. non-Byzantine] readings." Von Soden assigns it to I' (which tells us very little, since this is one of the catchall groups, containing both mixed and purely Byzantine manuscripts). Wisse, based on the fragments available to him, lists it as Kx in Luke 1, Kx in Luke 10, and mixed in Luke 20. The Alands list it as Category V (Byzantine).

Of all these assessments, the most accurate appears to be Hort's. The Alands, in particular, base their opinion on a mere nineteen readings -- too small a sample to tell us anything.

A much more detailed assessment can be made by examining the apparatus of NA26. The table below classifies readings in the Nestle apparatus into six categories: Those where R agrees with the Majority text against B, those where R agrees with B against the Majority Text, those where R agrees with both 𝔐 and B but where at least two important witnesses have a different reading, readings where R disagrees with both 𝔐 and B, and those where the majority text is split but R either agrees or disagrees with B. The numbers given below are slightly approximate (due mostly to the readings where the apparatus only cites evidence for one reading), but these generally affect the third category, which is the least significant for our purposes.

R with 𝔐
against B
R with B
against 𝔐
R with 𝔐
and B
R against 𝔐
and B
R with B
against pm
R with pm
against B
Luke 1-313313120
Luke 4-632816340
Luke 7-9511329262
Luke 10-1225620303
Luke 13-1512209820
Luke 16-18331311410
Luke 19-21561319602
Luke 22-242869411
Readings: 5132508212631168

Thus we see that, no matter where we look, about 20-25% of R's readings are non-Byzantine, and that the manuscript is not Byzantine at all in about chapters 13-16. Although it is by no means a primary witness, R should not be completely ignored.

Other Symbols Used for this Manuscript

von Soden: ε22



Sample Plates:

Editions which cite:

Cited by Tischendorf, who also collated it.
Cited by von Soden, Merk, and Bover.
Cited in NA26 but deleted in NA27

Other Works:

The entire manuscript is now available in high resolution scans at

Manuscript S (028)

Codex Guelpherbytanus B. Rome, Vatican Library Gr. 354. Von Soden's ε1027. Contains the four gospels complete. Dated by its colophon to 949. This makes S the only dated uncial (other than Γ, which has a partial date which we cannot interpret with certainty). It is also one of the four oldest dated New Testament manuscripts (the oldest being the minuscule 461, from the year 835; this is followed by 2500, from 891, then by S and the minuscule 1582, both from the year 949). Textually, it is entirely Byzantine. Von Soden classified it as K1 (along with such other Byzantine uncials as V and Ω); Wisse has made the minor correction of listing S as Kx Cluster Ω. (The other members of this group include E V Ω and some thirty-three minuscules.) The Alands corroborate this by listing S as Category V. The writing is large and compressed (see the sample in the Table of Scripts Used in Various Uncials), and appears Slavic. Scrivener notes that it "contains many later corrections... and marginal notes" (both patristic and textual, e.g. one of them obelizes John 5:4) as well as the Eusebian apparatus. It also includes neumes. The scribe was a monk named Michael. Note: The symbol S is also used in some apparatus for ℵ. (These apparatus will usually use 028 as a symbol for the real S.) One should always be aware of which symbol is used for which manuscript.
Sample plates in Metzger, Manuscripts of the Greek Bible; Edward Maunde Thompson, An Introduction to Greek and Latin Paleography (plate 50).

Manuscript T (029)

Location/Catalog Number

Codex Borgianis. Catalog Number: Rome, Vatican Library Borg. Copt. 109, Borg Copt. 109; New York, Pierpont Morgan Library M. 664A; Paris, National Library Copt. 129.7, 129.8, 129.9, 129.10. The various fragments, when discovered, were designated T (029), 0113, 0125, 0139.


Contains fragments of the gospels of Luke and John, in Greek and Sahidic (Sahidic on the verso), with the Greek containing Luke 6:18-26, 18:2-9, 10-16, 18:32-19:8, 21:33-22:3, 22:20-23:20, 24:25-27, 29-31; John 1:24-32, 3:10-17, 4:52-5:7, 6:28-67, 7:6-8:31 (with some of these leaves being fragmentary; also, there is damage to some of the upper outside corners). The following list shows how the various portions are designated:

The Pierpont Morgan portion, at least, came from Deir Amba Shenoudah, the White Monastery in Upper Egypt, and was purchased for the Pierpont Morgan library in 1912.


Usually dated paleographically to the fifth century, though Giorgi, who first published portions of it, prefers the fourth, and Clark suggested sixth or seventh. T is written on parchment, two columns per page -- but, curiously, the Greek and Sahidic are not in facing columns but on facing pages. Tischendorf thought the scribe was a Copt, as the letters often show Coptic forms. Clark sees confirmation in the errors in teh Greek text. It has a handful of breathings, but they are not supplied consistently. As far as the punctuation goes, Scrivener notes that "a single point indicates a break in the sense, but there are no other divisions."

Description and Text-type

That T stands close to B has been widely observed -- e.g. by Hort; von Soden classified all four parts as H, and the Alands place it in Category II. (Wisse was unable to classify it, as no text exists in his sample chapters.) But few seem to have realized how close the two are. The following tables show the relations between T and thirteen other witnesses in Luke and John. The readings are the variants in NA27 which are supported by at least two of the witnesses cited.

Affinities of T -- Luke

MS.𝔓75 ABDKLT Γ Θf1f13

Affinities of T -- John

MS.𝔓75 ABDKLT Γ Θf1f13

Examining these numbers tells us that T is not simply close to B in Luke; it is immediate kin -- as close to B as is 𝔓75. Indeed, T agrees with these two more than they agree with each other. The difference is not statistically significant given the size of the sample, but if this is in fact the case, it would imply that T is actually closer to the group archetype than either 𝔓75 or B. In any case, it deserves to be on a footing equal to theirs.

The matter is not quite as clear in John. T is still very close to 𝔓75 B, but not as close as in Luke. In first examining the data, it appeared to me that T had acquired some Byzantine mixture. Full examination of the data, however, makes it appear that instead it had been infected with late Alexandrian readings -- of the sort we find, e.g., in L. Thus in Luke T is a manuscript of the first magnitude, though in John its value is slightly less.

Other Symbols Used for this Manuscript

von Soden: ε5 (=T), ε50 (=0113), ε99 (=0125), ε1002 (=0139)


As this manuscript was recovered in sections, there has been no comprehensive publication. The first edition, by Giorgi in 1789, includes only the portions of John then known.

Sample Plates:

Editions which cite:

Cited by Tischendorf as far as known.
Cited by von Soden, Merk, and Bover as far as known.
Cited in NA26 and UBS3 (under four sigla) and in NA27 and UBS4 under the combined symbol T.

Other Works:
Kenneth W. Clark, Greek New Testament Manuscripts in America, pp. 161-162, discussed the Pierpont Morgan portion and gives a bibliography.

Note: The symbol T was used by Tischendorf and Scrivener for certain other manuscripts: Tb = 083; Tc = 084; Tg = 061; Tk = 085; Twoi = 070.

Manuscript U (030)

Venice, Biblioteca San Marco 1397 (I.8). Called Codex Nanianus after a former owner. Von Soden's ε90. Contains the four gospels complete. Dated by modern sources to the ninth century, though Scrivener, based on Tregelles, writes that it dates "scarcely before the tenth century, although the 'letters are in general an imitation of those used before the introduction of compressed uncials; but they do not belong to the age when full and round writing was customary or natural, so that the stiffness and want of ease is manifest.'" It is an ornate codex, with full marginalia, as well as pictures and golden ornaments. Textually, it appears Byzantine; the Alands place it in Category V (though their statistics for the manuscript are manifestly wrong; a complete copy of the Gospels will have many more than the 155 readings they list!). Wisse calls it Kmix/Kx/Kmix, with some similarity to 974 and 1006. This not-quite-pure Byzantine-ness may explain why Von Soden lists the manuscript as Io; Io contains a number of manuscripts strongly but not entirely Byzantine (e.g. X and 1071), though there is no real reason to think they are related.

Manuscript V (031)

Moscow, Historical Museum V.9, S.399. Called Codex Mosquensis. Von Soden's ε75. Contains the four gospels almost complete; Mark and Luke are intact; Matthew lacks 5:44-6:12, 9:18-10:1, 22:44-23:35; John is lacking from 12:10-25. The latter two lacunae are recent; they apparently developed between 1779, when it was first collated, and 1783, when next examined.
Although we say the manuscript runs through John 12:10, the uncial portion breaks off with John 7:39, the rest being in minuscules, dated to the thirteenth century. Matthei, who first examined V, dated the uncial portion to the eighth century, but moderns tend to prefer the ninth. It is written on parchment, one column per page. Textually, it is universally agreed that it is Byzantine; von Soden classified it as K1; the Alands place it in Category V; Wisse calls it Kx Cluster Ω. Thus it is a very typical Byzantine manuscript; it has been suggested that it would be a good manuscript to use as a standard for the Kx type.

Manuscript W (032)

Location/Catalog Number

Washington, D.C., Freer Gallery of Art 06.274 (Smithsonian Institution). Called Codex Washingtonensis for its location, or the Freer Gospels for its purchaser. Freer apparently found it and several other manuscripts in the stock of an antiquities dealer, Ali Arabi, in Giza, and purchased the lot on December 19, 1906 (for the absurdly low price of £1600).


Originally contained the four gospels complete; now lacks Mark 15:13-38, John 14:27-16:7. In addition, John 1:1-5:11 are a supplement, probably to replace a quire that was lost (but see below). Gospels are in the "Western" order: Matthew, John, Luke, Mark. However, as Schmid has pointed out, the number of leaves in the quires were adjusted such that each gospel begins and ends on a quire boundary. To emphasize: the individual gospels do not merely end at the end of a page, they end at the end of a quire. This means that the order of the books was determined by the binder, not the scribe -- it could even have been changed with rebindings. Thus we should not draw too many inferences from the book order we now have, since it may not be original.


Now generally dated to the fifth century, though from its first discovery some have preferred a date in the late fourth century. The supplemental leaves are usually dated from about the seventh century. The cover may also date from this time; it is wax painting on wooden boards, showing the four Evangelists, and survives fairly well. On the other hand, the chains that were added with the covers kept the book from being properly opened for reading! This very likely protected the volume -- but rendered it largely useless.

Although the date above is that usually accepted, Ulrich Schmidt has now suggested that W should be dated to the sixth century -- and while his argument is based on only a handful of witnesses, it seems logically sound; I think we must at least allow for a later date. Which, as James Dowden points out to me, means that the original text and the supplement are of almost the same age, which, in turn, means that the supplement may not be a replacement for lost leaves but rather (just possibly) a cancel. (Sanders argued strongly against this, based on the fact that the scribe compressed his text on some pages and then had to try to make it longer on the last page of the supplement, but as Schmid has shown, Sanders's hypothesis is far too complex an explanation. A cancel would also show require the scribe to make the text fit the available leaves.) If it is a cancel, it would explain why the supplement's text seems to be quite similar to the text of the rest of John in the original hand. (The differences between the two seem to have been exaggerated by those who want to make them seem more distinct.)

The handwritings of the main text and supplement are, nonetheless, very different. The original hand has a strong rightward slant; it is very angular, with few curves -- even the sigmas look more like ⊏ than C. The hand of the supplement is more vertical, more narrow, and more curved; it does not have the tall, serif-y look of a ninth century uncial, but (for lack of a better way to say it) it looks like it wants to.

It might perhaps be added that, although we think of the Chester Beatty Papyri as being the first great find coming from Egypt, in fact the Freer collection probably deserves that title. In addition to W, which is substantially intact, it included I (016), a badly damaged but very good copy of Paul, plus two important LXX manuscripts, the Freer Psalms and the Freer copy of Deuteronomy and Joshua, both of which were edited by Henry A. Sanders.

Unlike most of the old uncials, W has an indication of pre-Renaissance ownership. At the end of Mark (i.e. at the end of the codex), there is an inscription which appears to have been rewritten twice, with the final owner being listed as Timothy (τιμοθεου). That's obviously not much to go on, but it might indicate that the manuscript was in private hands. (Most scholars seem to think that this Timothy was a monk, but I don't see why that has to be so.) Sanders thought that W, and the other Freer codices which he thinks probably represented a multi-volume Bible even though the books are not the same size, came to be hidden or discarded after the Arab conquest of Egypt. Sanders, on this basis, concluded that the books belonged to the church, or monastery, of St. Timothy in Egypt. I must confess that this strikes me as a lot of speculation built on very little evidence, and it has never been universally accepted.

We do not know the precise place where the codex was discovered, but the best available evidence, from the man who sold the codex to Freer and his associates, seemed to point to Dimai, in the Fayyum, which in ancient times was on or near the shore of Lake Moeris (although it is now far from the shore because the lake has receded). This location is somewhat troubling for the dating of the codex, because Dimai was abandoned around 336 C.E. -- in other words, early fourth century. How can a codex dated late fourth century or later be from a place that was abandoned before that? Furthermore, extensive archaeological work at Dimai has failed to find any sign of a Christian community likely to have used the books. So while the books may well be from the Fayyum, I doubt they are from Dimai, and certainly we cannot base any conclusions on hypotheses about where the volumes originated.

Description and Text-type

W is textually a curiosity, as the nature of the text varies wildly. The usual statement (found, e.g., in Kenyon/Adams, p. 215) is that Matthew is Byzantine, Mark chapters 1-5 (possibly 1:1-5:30) are "Western," Mark chapters 6-16 are "Cæsarean," Luke 1:1-8:12 are Alexandrian, Luke 8:13-end are Byzantine, John 5:13-end are Alexandrian. (The supplement in John 1:1-5:12 is variously assessed; in my experience, it is Alexandrian, though perhaps not quite as pure as the original text of John. Based simply on the text, it is not impossible that the replacement quire was actually copied, at least in part, from the quire that it replaced. Or, as Dowden suggested, it might be a cancel but still copied from the same manuscript as the rest of John.) The boundaries between types are, of course, impossibly precise; one cannot determine a text-type boundary to the nearest sentence. But that there are shifts at about these points seems true enough.

For a long time, surprisingly little was done to test the assignments described above, even though the examinations were anything but rigorous. My own experience indicates that the Byzantine designation in Matthew and the Alexandrian designation in John are correct -- and, as far as I know, these have never been seriously questioned.

Jean-François Racine did do a study of the text of Matthew (published in Hurtado's The Freer Biblical Manuscripts), breaking the boo down into four-chapter blocks and comparing it against witnesses of various text-types. This was extraordinarily flawed, methodologically, in multiple particulars: first, Racine took readings represented in Basil of Cæsarea and called this a random sample, which it is not; second, Racine compared manuscripts to W, but did not compare those manuscripts to each other, so there is no baseline; third, Racine gives only percentage agreements and "error correction" -- which I think is supposed to be an estimate of sampling error, but he doesn't explain his method of calculation and he can't make that computation for a non-random sample. Yet again, the work should not have been published. Still, the data offered gives hints of a classification picture. I observe that W's agreements with the Byzantine witnesses are generally in the range of 78%-88% -- much greater than its agreement with any non-Byzantine manuscripts, but short of the Colwell-Tune 90% threshold. I emphasize that this means nothing given Racine's methodological errors. Racine's data does not prove that W is purely Byzantine in Matthew. But I will concede that it affirms the conclusion that W is more Byzantine than anything else.

Racine does try a second interesting analysis, of W's "cohesion" -- that is (to oversimplify) how well the narrative flows. Cohesion, or lack thereof, could be a sign of editing (that is, the text might be edited to flow better, or its cohesion might be damaged by an incompetent editor); it might also be the result of natural processes (scribes might smooth out a rough passage, or an error might create one). So cohesion by itself is neutral but interesting. Unfortunately, Racine compares only two texts, B and W. This limits the conclusions we can draw. If W is more (or less) cohesive than B, is it because W has been altered in a way that alters its cohesiveness -- or is it because W is Byzantine and it was the Byzantine text that was altered? Still, Racine's study is interesting: It shows that, where the two differ, W is much more likely to add a proper name (typically ΙΗΣΟΥΣ) than is B, that W is much more likely to have an article before names, that it adds many more conjunctions (usually και, δε, or οτι; the use of other conjunctions hardly seems affected). Racine is quick to point out that the differences in usage are few compared to the amount of text the two have in common -- but the nature of the changes is quite dramatic.

Things are a not quite as clear in Luke as in Matthew. Here, Wisse assesses W as Group B (Alexandrian) in Luke 1, as expected. In Luke 10, he lists it as Kx, while in Luke 20 it is mixed. The classification in Luke 10 is, in a sense, what we expect, since it was generally felt that W is Byzantine in that part of Luke. But the finding that it is Kx is extraordinary; this makes W the earliest Kx manuscript by at least three centuries. The "Mixed" assessment in chapter 20 is also somewhat surprising, as it indicates that W is less Byzantine than expected in that chapter. It's worth noting, though, that all these assessments are based on single chapters; assessments of larger sections of text might produce a slightly different view. The assessment that W's text of Luke is Alexandrian in the early chapters and Byzantine in the final two-thirds is probably essentially accurate.

In John, Dennis Haugh studied the singular readings of W and found them much rarer than in Mark. He considers this evidence (as I read it) of less scribal freedom. Unfortunately, this does not follow. If W is Alexandrian in John, but Western-or-something in Mark, then it will have more allies in John than in Mark. Consider: If W's only ally in Mark is D, then every scribal error in D in a specifically "Western" reading will produce a singular error in W. But in John, W has allies including B ℵ L 892 etc. There cannot be a case where an error in any one other manuscript produces a singular reading in W. So Haugh's method of computation is not particularly helpful. (This is not to deny that there appear to be deliberate changes in W in John -- the obvious example being in 21:17, where W has Jesus ask Peter for the third time, αγαπας με rather than φιλεις με as in all other witnesses. The omission of part of John 21:4, which implies that the Apostles are too dim-eyed to recognize Jesus, is also a hint of deliberate rewrite. It's just that the system Haugh uses is not mathematically trustworthy. Also, Haugh spends several pages examining historical trends in the usage of verb tenses. This is an interesting idea that probably deserves more examination when applied to the whole textual stream -- but is essentially meaningless when applied to a single manuscript. The rate at which historical trends are adopted varies from region to region and person to person. To take an obvious English example: The English subjunctive is rapidly failing. It is not unusual today to hear someone say "If I was to use a subjunctive...." Yet I still say, "If I were to use a subjunctive..." -- or even "Were I to use a subjunctive," which is a formation almost extinct. If you went by my usage of subjunctives, I lived in the nineteenth century or earlier -- but I don't.)

The question of Mark is much more complicated. Sanders, who first edited the manuscript, linked 1:1 to 5:30 to the Old Latin (claiming even to see Latin influence in the text). The rest of Mark he recognized as non-Byzantine and non-Alexandrian, but he thought it was not "Western" either; he linked it to manuscripts such as 1 and 28.

At this point Streeter entered the picture. Streeter claimed the last ten chapters of Mark as "Cæsarean," basing this mostly on a comparison against the Textus Receptus. Unfortunately for Streeter's case, this method is now known to be completely faulty (as he should have known himself). Streeter's "proof" in fact proved nothing -- although we must remember that his method was merely faulty, not necessarily that it produced inaccurate results. His contention may be true; he simply didn't prove it.

There things sat for half a century, while the "Cæsarean" text was sliced, diced, added to, subdivided, and finally slowly dissolved under scrutiny. Finally Larry W. Hurtado published Text-Critical Methodology and the Pre-Caesarean Text: Codex W in the Gospel of Mark (1981). This study compared W, chapter by chapter, against some of the leading witnesses of the various text-types.

Unfortunately, Hurtado's study has its own defects. The analysis is rather rigidly defined by chapters, and several vital witnesses are ignored. The key defect, however, is the fact that it simply counts readings without weighing them. This is fine for detecting immediate kinship, but less effective for dealing with mixed manuscripts -- and even Streeter admitted that all "Cæsarean" witnesses, except W itself, are mixed. But if all other "Cæsarean" witnesses are mixed, then none of them will show strong agreement with W at all points.

Hurtado found about what one would expect: W, in Mark 1-4, is indeed "Western" (note that he moved the dividing line toward the beginning of the book somewhat). Starting with chapter 5, it is something else, and that something does not match any of the other witnesses precisely. It is assuredly not Byzantine or Alexandrian. But neither does it agree particularly closely with the so-called "Cæsarean" witnesses.

Hurtado's study has been viewed, quite inaccurately, as dissolving the "Cæsarean" text. In fact it does no such thing, in that Hurtado nowhere so much as addresses Streeter's definition (which finds the "Cæsarean" text in the non-Byzantine readings of the "Cæsarean" witnesses. Since Hurtado did not classify readings, he could not study the type as defined by Streeter). Nonetheless, Hurtado did a reasonable job of demolishing Streeter's claim that W is a pure "Cæsarean" witness in the latter portions of Mark. The fact that the "Cæsarean" witnesses do not agree with each other is not relevant (the effect of random mixture is to make the mixed witnesses diverge very rapidly). The fact that they do not agree with W, however, is significant. W can hardly be part of the type from which the surviving "Cæsarean" witnesses descended. This still does not, however, prove that it is not "Cæsarean" -- merely that it does not spring from the sources which gave rise to Θ, 565, and Family 13. Further conclusions must be left for a study which addresses Streeter's text-type according to Streeter's definitions. (For what it is worth, my statistical analysis does seem to imply that the "Cæsarean" type exists -- but the sample size is not enough to allow certainty about W's relationship to it.) Hurtado found that W had a special relationship with 𝔓45, and this is by no means improbable. Hurtado also hypothesized that W in the final chapters of Mark was still "Western," but with mixture. This too is possible, and given Streeter's sloppy methods, it might explain why Streeter associated W with the "Cæsarean" type. But Hurtado's method cannot prove the matter.

There has been much discussion of why W is so strongly block mixed. Sanders thought that it was compiled from bits and pieces of other manuscripts. Streeter counter-argued that an exemplar was heavily corrected from several different manuscripts, each manuscript being used to correct only part of the exemplar. Neither theory can be proved; they have different strengths and weaknesses (Sanders's theory explains the abrupt textual shifts, but is it really probable that any church would have so many fragments and no complete books? Streeter's theory eliminates this objection, but does very little to explain why the text does not show more mixture of types of text in each section. W is block mixed, but the text is generally pure in each part -- at least, the Alexandrian and Byzantine parts are quite typical of those types; nowhere is there an Alexandrian/Byzantine mix.)

The most noteworthy reading of W is the so-called "Freer Logion" (so-called because it occurs only in W; Jerome quotes a portion of it). This passage, inserted after Mark 16:14, is quoted in most textual criticism manuals and will not be repeated here.

In addition to its block-mixed nature, several scholars (e.g. Haugh) have suggested that W, at least in John, rewrites the text "to avoid embarrassing references to Jesus, his disciples, and his family."

There is little else to say about the text of W. The Alands list it as Category III, but of course this is an overall assessment; they do not assess it part by part (if they did, the assessment would probably range from Category II in the Alexandrian portions to Category V in the Byzantine). Von Soden's classification is more complex (Iα -- i.e. mainstream "Western"/"Cæsarean" -- in Mark, H in Luke and John), but this tells us little that we did not already know.

Sanders thought that three correctors worked on the codex in addition to the original scribe, and Haugh agrees although he changes the attributions of a few corrections; the original scribe, according to Haugh, corrected his own text 84 times, the διορθωτης changed 69 readings, and the later correctors made ten and four changes, respectively; there were two changes that could not be assigned with certainty. The supplement has eleven corrections, from its original scribe and two other hands. Sanders thought the changes by the original scribe were often instances where the scribe was choosing between variants that were before him, and then changed his mind, but Haugh doubts this, and so do I; the scribe was simply correcting places where he accidentally wrote the wrong text. There are a few places where the scribe's exemplar probably had a reading and correction of its own, and the scribe didn't notice, but these are the exception (Haugh suggests Luke 11:6 as the clearest example of this). Haugh believes that the second corrector, the διορθωτης, corrected W against its immediate examplar; although the number of corrections is significant, they would have been far more had the manuscript been corrected against any normal manuscript that was not block mixed! The third and fourth correctors do not appear to have been working from another manuscript; their handful of changes were simply adjustments of obvious slips by the first scribe. Thus, while W's ancestry is block-mixed, there is no sign that it represents more than one exemplar.

Other Symbols Used for this Manuscript

von Soden: ε014


Note: As with all the major uncials, no attempt is made to compile a complete bibliography.

The basic edition for a century was Henry A. Sanders, Facsimile of the Washington Manuscript of the Four Gospels in the Freer Collection, plus (again by Sanders) The New Testament Manuscripts in the Freer Collection, Part I: The Washington Manuscript of the Four Gospels. In the mid-2000s, Prior and Brown hoped to produce a new edition, but if it ever appeared, I can't find a reference to it.

Sample Plates:
Almost every handbook has a photo, but it's almost always the same page (the Freer Logion in Mark 16). The book In the Beginning: Bibles Before the Year 1000 has a large color version of that page and the facing page (pp. 152-153); on pp. 151-152 it shows the cover artwork of the four evangelists. Finegan has a plate of the supplement in John 1. Ulrich Schmid's article in The Freer Biblical Manuscripts: Fresh Studies of an American Treasure Trove on pp. 228-229 has a very useful photo allowing comparison of the supplement with the main text.

The manuscript has now been digitized, in color, at

Editions which cite:
Cited in all editions since Von Soden

Other Works:
See most recently and most notably Larry W. Hurtado, Text-Critical Methodology and the Pre-Caesarean Text: Codex W in the Gospel of Mark. This is largely a reaction to Streeter; for Streeter's opinions concerning W, see Appendix V to The Four Gospels: A Study of Origins. Hurtado also edited The Freer Biblical Manuscripts: Fresh Studies of an American Treasure Trove. This book is not exclusively about W, but many of the articles discuss it.

Manuscript X (033)

Location/Catalog Number

Codex Monacensis. Munich. Catalog Number: University Library fol. 30. It arrived in Munich in 1827; prior to that it had been in Landshut (from 1803), still earlier in Ingoldstadt; its earliest known home was Rome.


Contains the Gospels in the "Western" order Matthew, John, Luke, Mark (as presently bound, there are actually leaves of Matthew at both beginning and end of the codex, and Scrivener implies that the original order was John, Luke, Mark, Matthew, but this is probably a binding error). It has suffered some damage, and now contains Matt. (5:45 in commentary only), 6:6, 10, 11, 7:1-9:20, 9:34-11:24, 12:9-16:28, 17:14-18:25, 19:22-21:13, 21:28-22:22, 23:27-24:2, 24:23-35, 25:1-30, 26:69-27:12, John 1:1-13:5 (2:23-6:71 lost but added in a later hand), 13:20-15:25, 16:23-end, Luke 1:1-37, 2:19-3:38, 4:21-10:37, 11:1-18:43, 20:46-end, Mark 6:46-end (with portions of chapters 14-16 illegible and 16:6-8 completely lost). Text with commentary; most of the marginal material is from Chrysostom. The commentary is very full in Matthew and in John; that in Luke contains references to the previous sections as well as new material; Mark has no commentary at all. (Which to me suggests that the use of the "Western" order is probably original and might hint that the text and commentary were assembled in an ancestor of our present volume.) The commentary is written in minuscules and is contemporary with the uncial text.


Dated paleographically to the tenth or possibly ninth century. X is written on parchment, two columns per page. The hand is described as "very elegant"; Scrivener quotes Tregelles's work to the effect that the letters are "small and upright; though some of them are compressed, they seem as if they were partial imitations of those used in the very earliest copies." The text has, apart from the commentary, relatively few guides for the user; there are no lectionary notes or κεφαλαια.

Description and Text-type

The most recent assessment of this manuscript, that of the Alands, is stark: they place is in Category V as purely Byzantine. This is, however, much too simple. While it is certainly true that the manuscript is more Byzantine than anything else, it has a number of noteworthy readings not of that type. Wisse, for instance, finds it to be mixed insofar as it exists, with "some relationship to Group B."

Von Soden isn't much help in this matter; he classified X as Io. However, the members of this group, according to Wisse, are a very mixed lot: U (Kmix/Kx; close to 977 1006), 213 (Mix), 443 (M159), 1071 (Mix; "some relationship to Group B"), 1321(part) 1574 (Mix) 2145 (M1195/Kx). Still, a handful of striking readings will show that X is at least occasionally linked with the Alexandrian text, especially with the B branch:

It appears that the largest fraction of X's Alexandrian readings is in John; this may explain why the Alands (who did not examine John) classified it as Byzantine.

James Dowden informs me that Wieland Willker and Bruce Morrill independently found a significant link between X and 865, a commentary manuscript on John which is dated to the fifteenth century. 865 is in Rome, which was where X was first found; this hints that there was a commentary text that was used by both. Von Soden classified both X and 865 as having the Antiochene Commentary (X is his A3 and 865 his A502); it might be worth looking at how X relates to the text of Chrysostom (who, as mentioned, is said to have produced the commentaries on Matthew and John). Checking the six special readings listed above, there is no obvious link to Chrysostom; according to the readings of Chrysostom cited in the UBS4 apparatus, Chrysostom and X disagree at Matt. 16:2-3, John 14:4; they agree at John 7:53-8:11, 12:1; Chrysostom is not cited for Luke 15:21, John 13:32.

Other Symbols Used for this Manuscript

von Soden: A3



Sample Plates:

Editions which cite:

Cited by Tischendorf, who also collated it.
Cited by von Soden, Merk, and Bover.
Cited in UBS3 but deleted in UBS4

Other Works:

Manuscript Y (034)

Cambridge, University Library Additional MS. 6594. Called Codex Macedonensis. Von Soden's ε073. Contains the four gospels fairly complete. Matthew 1-8 are lacking and 9-11 are somewhat mutilated. Mark is intact. There are small lacunae in chapters 15-16 and 23 of Luke and the last two chapters of John.
There is one column per page. Most authorities agree that it dates to the ninth century. Textually, it is universally agreed that it is Byzantine; von Soden classified it as Iκ. which is Family Π, and Wisse concurs, placing it in group Π171. The Alands place it in Category V, which is their usual classification for the uncial (although not the minuscule) members of Family Π.ß

Palimpsest Z, Trinity College, Sixth Century

Manuscript Z (035)

Codex Dublinensis Rescriptus. Dublin, Trinity College K.3.4. Von Soden's ε26. Palimpsest, containing portions of Matthew (Matt. 1:17-2:6, 2:13-20, 4:4-13, 5:45-6:15, 7:16-8:6, 10:40-11:18, 12:43-13:11, 13:57-14:19, 15:13-23, 17:9-17, 17:26-18:6, 19:4-12, 21-28, 20:7-21:8, 21:23-30, 22:16-25, 22:37-23:3, 23:15-23, 24:15-25, 25:1-11, 26:21-29, 62-71). The upper writing is a cursive, no earlier than the tenth century, consisting of works of various church fathers, Chrysostom contributing the largest share.

Of the original 120 or so leaves, fourteen double leaves and four half-leaves survive -- in other words, a bit less than 15% of the original text of Matthew.

Dated paleographically to the sixth or possibly fifth century. Written in a large, attractive, and very precise uncial, with the Ammonian Sections but seemingly no Eusebian canons. It has spaces at key points, but very little punctuation, and no breathings or accents. Quotations are indicated with the > symbol. Scrivener notes that it displays unusual forms of many letters, and gives evidence from Abbott that it is relatively free of scribal errors.

Assessments of its text have universally rated it highly; Von Soden lists it as H (Alexandrian) and the Alands show it as Category III. The text is in fact very close to ℵ, and may be regarded as that manuscript's closest ally. On the Aland scale, it probably ought in fact to be Category II, perhaps even Category I, except that there is too little text to be useful for much.

The illustration at right was taken from a very bad photograph, and has been heavily manipulated to try to bring out as much as possible. The colours are false. Note that this is the correct orientation to read the Greek text; the over-writing is at right angles to the lower text.

The manuscript, which by then was already "miserably discoloured" (so Scrivener), was exposed to chemicals by Tregelles in 1853, which yielded little new text but has of course not helped its legibility.

Γ (Gamma, 036)

Oxford, Bodleian Library, Auct. T. infr. 2.2 and Saint Petersburg, Russian National Library Greek 33. Codex Tischendorfianus IV. Von Soden's ε70. The date of this manuscript is a mystery: It gives a date by indiction, and also mentions that (in modern terms) November 27 was a Thursday. Tischendorf calculated that the only date between 800 and 950 which fits this description is 844, but Gardthausen calculated that 979 also meets the conditions. Paleographers tend to date the manuscript to the tenth century. It is rather sloppily written, with lines ruled badly and irregularly. A later corrector added additional accents and breathings to those supplied by the original scribe; Scrivener calls these additions "very careless" and describes the later scribe as a "scrawler." It also has neumes; I do not think these are by the original scribe, but they may be contemporary. Γ is a copy of the gospels, with many lacunae in Matthew (lacks Matt. 5:31-6:16, 6:30-7:26, 8:27-9:6, 21:19-22:25) and one in Mark (lacks Mark 3:34-6:21); Luke and John are complete (there is some damage from damp to the end of Luke, but this does not render the manuscript illegible). Γ was found by Tischendorf in "an eastern monastery" and divided into rather odd portions: England has part of Matthew, all that survives of Mark, all of Luke, and a few leaves of John; Russia has the rest of Matthew and the larger portion of John.
Assessments of the text of Γ have varied a great deal. Scrivener, without being able to examine it fully, remarked that "Some of its peculiar readings are very notable, and few uncials of its date deserve more careful study." Von Soden also saw some value in it, as he classified it as I' (in other words, among the miscellaneous members of the I group. Most I' members seem in fact to be mostly Byzantine). But one has to suspect that this classification is actually based on only a single reading: Γ is one of the several manuscripts to exclude Matthew 16:2-3 (others which do so include ℵ B X f13 157 579 and many of the early versions). Recent assessments of Γ have been much less kind. The Alands classify it as Category V (with only one non-Byzantine reading in 286 test passages, though it also has 12 readings which agree neither with the Byzantine nor the UBS text). Wisse lists it as Kx in all his test chapters. On the face of it, it would appear Γ, rather than being an unusually distinguished manuscript for its date, is in fact a perfectly typical Byzantine manuscript with more than its share of singular errors, the work of a somewhat inept scribe. Scans of three of the leaves at the Bodleian can be found at the Bodleian Library web site,

Manuscript Δ (Delta, 037)

Location/Catalog Number

Saint Gall, where it has been as long as it has been known (hence the title Codex Sangallensis). Catalog number: Stiftsbibliothek 48.


Contains the gospels almost complete; it lacks John 19:17-35. The Greek is accompanied by an interlinear Latin translation (designated δ). It has been argued that Δ was originally part of the same volume as Gp; for the arguments for and against this (e.g. their similar appearance and identical size, as well as the fact that both are from St. Gall), see the entry on that manuscript.

Curiously, after the end of Matthew (p. 129), there is an inventory of illustrations, partly in Greek and partly in Latin -- but no such illustrations appear in the text! (It has been suggested that this might be a guide for someone who would copy Δ and illuminate it -- by why, then, put it after Matthew rather than at the beginning or end of the volume?) An illustration of this page can be seen on p. 81 of Bernice M. Kaczynski, Greek in the Carolingian Age: The St. Gall Manuscripts (Medieval Academy of America, 1988).


Usually dated paleographically to the ninth century. (It can hardly be earlier, as marginal reference is made to the (heretical) opinions of Godeschalk at Luke 13:24, John 12:40. These references appear to be in the original hand, and Godeschalk died in 866.) A few sources prefer a tenth century date.

The hand is quite awkward and stiff, resembling Gp in this as in many other ways. The Latin is written above the Greek, and the scribe seems to have been more comfortable with that than with Greek. (There are many reasons for believing this; one of the more noteworthy is his regular confusion of certain Greek letters. Also, the Greek letterforms are not consistently drawn.) It has been widely suggested that his native language was (Irish) Gaelic -- Ireland had a strong link with St. Gall. Harris declares, "Our St Gall scribe is an ignorant person, as mechanical as most of his tribe in his own day"; he was not competent to decide word breaks in Greek.

The form of the manuscript again reminds us of G: It is written in continuous lines, but appears to have been made from a manuscript written in sense lines of some sort; there are enlarged, decorated letters in almost every line. (Though the decorations are very inartistic; Gregory suggests that "[t]he larger letters are rather smeared over than painted with different colours.") The enlarged letters do not really correspond with sentences, but rather are quite evenly spaced. Spaces are supplied between words, but these are very inaccurate (more evidence of the scribe's weakness in Greek). There are only a few accents and breathings, not always accurate. Gregory notes that "[t]he titles for the chapters often stand in the middle of the text."

Rettig believed that several scribes worked on the manuscript. This is a difficult question to say the least. The style of the manuscript is very similar throughout. At first glance -- indeed, at any number of glances -- it appears that the scribe is the same throughout. But this is because the hand is so peculiar. The evidence of G indicates that this was more or less the normal style at Saint Gall. So it is possible that there were several scribes -- but the matter really needs to be investigated with modern resources. It has been claimed that one of the scribes of Δ, as well as writing G, also wrote the Basel Psalter and a copy of Horace at Bern.

Description and Text-type

For once there is almost universal agreement: Δ is block-mixed. The usual assessment is that Matthew, Luke, and John are Byzantine, while Mark is Alexandrian. (Indeed, Δ was the single most important prop in Streeter's argument that manuscripts should be examined first in Mark.) Interestingly, most formal investigations have not precisely confirmed this division into parts; von Soden listed Δ as H, and the Alands list it as Category III. Even Wisse does not find it to be purely Byzantine in Luke 1; his assessment is that it is Mixed in Luke 1 and Kx in Luke 10 and 20.

It should be noted, however, that both the Aland and von Soden were listing text-types for the gospels as a whole; they are not book-by-book assessments. (The Alands, at least, did not so much as examine John.) An examination of the actual readings of the manuscript shows that conventional wisdom is correct at least in general: Δ is Byzantine in Matthew, Luke, and John, and is Alexandrian in Mark. We should however add that it is not purely Alexandrian even in Mark; nowhere does it approach the quality of B, or even of L. It is a late Alexandrian/Byzantine mix. It is also my personal impression that Δ has rather more Alexandrian readings in the early part of Mark, and that the Byzantine component increases somewhat in the final chapters -- but I have not formally verified this.

NA28 says that Δ*vid had John 7:53-8:11, which the corrector omitted, but the actual situation is that Δ has a gap there, as if the scribe expected the pericope but did not find it in his exemplar.

The interlinear Latin version is sometimes listed as an Old Latin version; hence the designation δ. This is probably at least technically a misnomer; the Latin version was probably prepared after the translation of the Vulgate. But since it has been made to correspond to the text of Δ, it is not a pure vulgate text. Still, it has no real critical value.

Other Symbols Used for this Manuscript

von Soden: ε76


H. Ch. M. Rettig's edition of 1836 remains the only full-fledged edition. Fortunately, this edition is said to be highly accurate (Gregory calls it the best edition of a manuscript prior to Tischendorf).

Sample Plates:
Metzger, The Text of the New Testament (1 page)

Editions which cite:
Cited in all editions since Tischendorf

Other Works:
J. Rendel Harris: The Codex Sangallensis: A Study in the Text of the Old Latin Gospels, 1891 (reprinted 2015 by Wipf & Stock) offers a detailed examination of the Latin text δ.

Manuscript Θ (Theta, 038)

Location/Catalog Number

Tbilisi, Georgia (the former Soviet republic), Inst. rukop. Gr. 28 Known as the Koridethi Codex or Codex Koridethianus (after its earliest known location).


Contains the four gospels almost complete; lacks Matthew 1:1-9, 1:21-4:4, 4:17-5:4.


The writing of Θ is unique -- see the sample letterforms in the article on uncial script; note in particular the delta (well on its way to becoming a Cyrillic letterform), kappa, lambda, mu, and the horizontal shape of chi. The odd letterforms make the manuscript impossible to date; extreme estimates range from the seventh to the tenth century. A late date is all but assured, however, by the generally narrow letterforms and the strong serifs. The most common estimate is the ninth century, and later seems more likely than much earlier.

The scribe of Θ was, to put it mildly, not comfortable in Greek; there are strange errors of spelling and grammar on every page. In addition, the scribe does not seem to have been trained to write Greek; he has been accused of drawing rather than writing his letters. Certainly they vary significantly in size and in their relationship to the line. If the scribe knew Greek at all, it was probably as a spoken language.

Gregory and Beermann gave this information about the codex (thanks to Wieland Willker for making this available to me): "In the year 1853 a certain Bartholomeé visited a long abandoned monastery in Kala, a little village in the Caucasian mountains near the Georgian/Russian border... he discovered the MS. The MS rested there probably for several hundred years (Beermann: ca. 1300 - 1869).... Before this time the MS was in a town called Koridethi. This was a village near the Black Sea, near today's Bat'umi in Georgia. There should still be some ruins of a monastery. Notes in the Gospel indicate dates from ca. 965 CE on. At around this time, according to a note, the book has been rebound. The book was there until around 1300 CE."

The most likely explanation is that the scribe was a Georgian, or possibly (as Beermann argued) an Armenian. Not only is the manuscript from the Caucasus, but it has a Georgian inscription on the back cover. In addition, the text appears to have affinity with the Georgian and Armenian versions.

Description and Text-type

Other than Codex Bezae, perhaps no other manuscript has been so enshrouded in scholarly controversy as the Koridethi Codex. The common statement in the manuals (e.g. Metzger, Kenyon) that it is Byzantine in Matthew, Luke, and John, while having a different text in Mark is simply false; it is based on a misreading (I am tempted to say a perverse misreading) of Streeter. Whatever Θ is (and we must defer this question a bit), it is the same throughout the Gospels: It is a mixture of readings of the Byzantine type and something else.

The key question, though, is What is the something else? That the manuscript was interesting was obvious from the very start. When it was first published, it was obvious that some of the non-Byzantine readings were typical of the Alexandrian text, others of the "Western" type.

It was Kirsopp Lake who first looked at those other readings, and perceived a kinship. It appeared to him that these readings were similar to the non-Byzantine readings of manuscripts such as 1, 13, and 565.

At this point, B. H. Streeter entered the picture. He found, in these readings, a kinship to the text which Origen used while in Cæsarea. He therefore declared this type to be the "Cæsarean" text. Within this type, he included the non-Byzantine readings of a large number of manuscripts, notably (family) 1, (family) 13, 28, 565, and 700 -- but also such things as the purple uncials (N etc.) and family 1424.

But note the key phrase: the non-Byzantine readings of these manuscripts. This proved to be a real sticking point. It has two problems. One is methodological: Streeter assumed that the Textus Receptus is identically equal to the Byzantine text. This is simply not the case; while the TR is Byzantine, it is not a normal Byzantine text. To make matters worse, the chief non-Byzantine influence on the TR is none other than 1. This means that the TR itself has "Cæsarean" readings -- and that, in turn, means that a reading which is purely Byzantine might be classified as "Cæsarean" under Streeter's system.

So does the "Cæsarean" text exist? This is an extremely vexed issue. Streeter described the text as having a mixture of Alexandrian and "Western" readings. Here, again, the description muddies the picture. If the "Cæsarean" type is real, it has only "Cæsarean" readings; it's just that it shares some with the Alexandrian text, and it shares a different set with the "Western" text. (This is to be expected; the majority of variants are binary -- that is, have two and only two readings -- so it follows, if the Alexandrian and "Western" texts disagree, that the "Cæsarean" text will agree with one of them.) But this leads to a problem: If all "Cæsarean" readings are shared with either the Alexandrian or Byzantine or "Western" texts, how do we tell a "Cæsarean" witness from an Alexandrian/Western mixed text? (To add to the uncertainty, we have to decide what is the "Western" text; the fact that Codex Bezae is our only Greek witness, and it in many ways peculiar, makes this a very difficult question.)

There are two partial answers to the question of how to tell a "Cæsarean" from a mixed manuscript: One is that the "Cæsarean" text does have some unique readings. A famous example is Matt. 27:16-17, where Θ f1 700* arm geo2 call the criminal released instead of Jesus "Jesus Barabbas," while all other Greek witnesses read simply "Barabbas."

The other is the pattern of agreements. If you create two manuscripts which arbitrarily mix Alexandrian and "Western" readings, they will only agree on half the readings where the two types separate. If two manuscripts have a percentage of agreements which is significantly higher than this, then they are kin.

This was more or less Streeter's argument. But Streeter had a problem: All his "Cæsarean" witnesses were mixed -- they had definite Byzantine overlays. That meant that he could only assess the nature of the underlying text where the manuscripts were non-Byzantine. This was a real difficulty, and made worse by the fact that Streeter (because he used the Textus Receptus to represent the Byzantine text) did not know what the Byzantine text actually read!

Streeter, in examining the non-Byzantine readings of his sundry witnesses, found agreement rates usually in the 70% to 90% range. This is a weak point. Allowing for a 50% inherent agreement rate, and 10% readings where the TR is not Byzantine (making agreements against the TR actually Byzantine), and 10% for coincidental agreement (e.g. harmonizations which could occur independently), and the expected rate of agreement in non-Byzantine readings is on the order of 70%. (I have verified this in testing a number of manuscripts. Unrelated manuscripts usually agree in 60% to 70% of non-Byzantine readings.) Certainly 70% agreement in non-Byzantine readings doesn't prove much.

The result was some decades of confusion. Streeter, by his faulty method, managed to make nearly everything a "Cæsarean" witness, and many scholars followed him. For some decades, there was a hunt for "Cæsarean" witnesses. This more or less culminated in the declaration that 𝔓45 was "Cæsarean."

At this point, the whole edifice started to crumble of its own weight. Everything not nailed down had been declared "Cæsarean," often on no stronger basis than the fact that it wasn't in pure agreement with the Textus Receptus. People started wondering about the "Cæsarean" text.

These doubts began to surface as early as the 1940s, but the single strongest blow was not struck until the 1980s, when Larry W. Hurtado published Text-Critical Methodology and the Pre-Caesarean Text: Codex W in the Gospel of Mark. This dissertation attempted to re-examine the entire "Cæsarean" question.

Great claims have been made about the results of this study (not least by its author), but in fact it was a limited work. Still, it accomplished two things: First, it demonstrated (as was already known) that the members of the "Cæsarean" text were not immediate kin, and second, it showed that 𝔓45 and W, often treated as the earliest and key "Cæsarean" witnesses, were not "Cæsarean" at all. (That 𝔓45 was simply a mixed witness had already been shown by Colwell, who found it to be a freely edited manuscript, but Hurtado generally confirmed Colwell's findings.)

But Hurtado's study had severe flaws of its own. One Hurtado has admitted in internet correspondence: The study did not examine all of the leading "Cæsarean" witnesses. The other is more fundamental: He refuses to acknowledge Streeter's definition of the "Cæsarean" type. Streeter defined the type in terms of non-Byzantine readings. Hurtado dealt with all readings. While he did some classification, it was not Streeter's method of classification. The two are talking past each other. Thus the final word on the "Cæsarean" text remains to be spoken. (As is shown by the fact that many modern scholars firmly believe in the "Cæsarean" text, while others are equally vehement in denying its existence.)

We, unfortunately, cannot prove the matter. The nature of the case, however, is that we must look at the matter using multiple statistical measures -- because only by this means can we reconcile the contradictory results of the single-statistic studies. Those who dismiss the "Cæsarean" text use Hurtado's method of overall agreements. Streeter defined it in terms of non-Byzantine agreements. And those who believe in the type today tend to point to the unique readings of the type, such as the "Jesus Barabbas" reading noted above.

There is, in fact, no fundamental reason why all three methods cannot be used. I have attempted this myself (see the article on Text-Types). The results are interesting: Θ and the other "Cæsarean" witnesses do not show unusually high degrees of overall kinship (except that Θ and 565 are quite close in Mark). They show high degrees of agreement in non-Byzantine readings -- but not so close a degree of kinship that we can automatically say it is statistically significant. In near-singular readings, however, there does appear to be kinship.

Does this settle the matter? No. Since we don't have a mathematical definition of a text-type, we can't just state that the numbers tell us this or that. It appears to me that Streeter's definition is sound, and that Θ is the best surviving witness of a small group (Θ family 1 family 13 565 700; I am less certain of 28, and I find no others) which have a text-type kinship but have been heavily mixed. Streeter's claim that these are a family (i.e. a group of closely related manuscripts, close than a text-type) is, however, thoroughly untrue. A final answer, however, must await better definitions of our terms.

This has not, of course, kept people from classifying Θ. Von Soden, who was the first to really examine the manuscript (and who worked before Streeter) listed it as Iα, i.e. as a member of the main "Western" group. (We should note that Streeter took all the Iα witnesses, save D, and declared them "Cæsarean.") Wisse classified the manuscript as "mixed" in Luke (a result which should have told him something about his method, but didn't. Certainly Θ is mixed -- but we don't want to know if it's mixed; we want to know what elements compose the mixture! Wisse could detect a weak Group B manuscript, because manuscripts like B and ℵ gave him a clear Group B profile -- a profile so clear, in fact, that he could include D in the type! But there is no pure witness to the "Cæsarean" text; meaning that Wisse could not have hoped to identify a "Cæsarean" type if one exists). The Alands, who do not classify by text-types, simply list Θ as Category II.

Other Symbols Used for this Manuscript

von Soden: ε050
(and note that the symbol Θ was used for assorted small uncial fragments until Gregory reorganized the manuscript list)


Note: A true bibliography about Θ is impossible, since every work about the "Cæsarean" text is largely about Θ. The following list includes only a selection of key works.

A Russian facsimile edition of Mark is extremely hard to find. Gustav Beerman and Caspar René Gregory published the complete text in Die Koridethi Evangelien Θ 038 (1913). Streeter, however, warns that the secondary collations in this book (comparing Θ against other manuscripts) are highly inaccurate, at least for the minuscules.

Sample Plates:
Aland & Aland, The Text of the New Testament (1 plate)

Metzger, Manuscripts of the Greek Bible (1 plate)

Editions which cite:
Cited in all editions since Von Soden

Other Works:
Kirsopp Lake and Robert P. Blake, "The Text of the Gospels and the Koridethi Codex" (Harvard Theological Review, xvi, 1923) is the first major work on what came to be called the "Cæsarean" text.
B. H. Streeter, The Four Gospels: A Study of Origins, (1924) is the basic definition of the "Cæsarean" text.
Larry W. Hurtado, Text-Critical Methodology and the Pre-Caesarean Text: Codex W in the Gospel of Mark is the most recent major study of the "Cæsarean" text.

Λ (Lambda, 039)

Oxford, Bodleian Library, Auct. T. infr. 1.1. Codex Tischendorfianus III. Von Soden's ε77. Dated paleographically to the ninth century by most authorities, although Scrivener allows the bare possibility of the eighth century and the Bodleian Library suggests mid-tenth. It is a curious manuscript, containing only Luke and John in uncials. The gospels of Matthew and Mark were written in minuscules; this half of the manuscript is numbered 566 (von Soden still calls it ε77) and located in Saint Petersburg. It has the Eusebian apparatus and a few comments in the margins. It is also noteworthy for having the "Jerusalem Colophon" after all four gospels. Textually, Von Soden listed it as Ir; other members of this group include 262 (which also has the colophon) 545(part) 1187 1555 1573. Wisse lists it as a member of his Group Λ (though with some "surplus"); this is his equivalent of Soden's Ir. Other members of the group, according to Wisse, include 161 164 166 173(part) 174 199 211 230 262 709 710(part) 899 1187 1205 1301(part) 1502(part) 1555 1573(part) 2465 2585(part) 2586 2725(part). Wisse notes that the group is fairly close to Kx, falling between Group 1216 and Kx. That it is close to the majority text is confirmed by the Alands, who place Λ in Category V (Byzantine). There are scans of the first and last pages of Luke and John at the Bodleian web site,

Ξ (Xi, 040)

Cambridge, University Library, British and Foreign Bible Society MS. 24. Codex Zacynthius. Von Soden's A1. Palimpsest, with the upper writing being the lectionary 𝓁299 (thirteenth century). Presumably originally contained the entire Gospel of Luke with a catena (probably the oldest catena left to us, and the only one with both text and commentary in uncial script; nine Fathers are thought to have been quoted.), but the surviving leaves contain only Luke 1:1-9, 19-23, 27-28, 30-32, 36-60, 1:77-2:19, 2:21-22, 2:33-3, 3:5-8, 11-20, 4:1-2, 6-20, 32-43, 5:17-36, 6:21-7:6, 7:11-37, 39-47, 8:4-21, 25-35, 43-50, 9:1-28, 32-33, 35, 9:41-10:18, 10:21-40, 11:1-4, 24-33 (86 full leaves and three partial leaves, originally quite large in size). Dated by W. P. Hatch and the Alands to the sixth century, but Scrivener argues that the writing in the catena (which is interwoven with the text, and clearly contemporary, in a hand so small as to be all but illegible since its erasure) belongs to the eight century, and other authorities such as Greenlee have tended toward the later rather than the earlier date (though the absence of accents and breathings inclines us against too late a date). Textually, Ξ clearly has Alexandrian influence, probably of a late sort (indeed, it appears to be closer to L than any other manuscript). Wisse lists it as being Kx in Luke 1 and Group B (Alexandrian) in Luke 10, but this probably does not indicate block mixture so much as sporadic Byzantine correction. Since it is a catena manuscript, Von Soden does not really indicate a text-type (listing it simply as one of the witnesses to Titus of Bostra's commentary), but the Alands assign it to Category III. Perhaps even more interesting than the text, however, is the system of chapter division, for Ξ uses the unusual scheme of divisions found in Codex Vaticanus (B), though it also has the usual system of τιτλοι. This serves as additional reason to believe that the text is basically Alexandrian. First edited by Tregelles in 1861, the text has been re-edited as recently as 1957 (by Greenlee), but probably is due for another examination with the most modern technology.

Π (Pi, 041)

Saint Petersburg, Russian National Library Greek 34. Codex Petropolitanus. Formerly owned by Parodus of Smyrna, who was persuaded by Tischendorf to give it to the Tsar of Russia. Von Soden's ε73. Dated paleographically to ninth century. Contains the four gospels with minor lacunae: Matt 3:12-4:18, 19:12-20:3, John 8:6-39. In addition, Scrivener reports that John 21:22-25 are from a later hand.
When Π was first discovered, it was observed that it generally agreed with the Byzantine text, but with certain unusual readings, most of which agreed with Ke. This kinship was later formalized by Von Soden, who declared Π (along with K Y and a number of minuscules such as 265 489 1219 1346) to be members of his Ika group. Von Soden felt this group to be a mix of I and K (Byzantine) texts, with K heavily predominating.
Speedy confirmation of Von Soden's results followed, though the studies (primarily by Lake, New, and Geerlings) were subject to severe methodological flaws. It should be noted, however, that the type was first discovered long before their time; Scholz called K and its relatives a text-type, the Cyprianic. This is exaggerated, since it's much more Byzantine than anything else, but it stands well apart from the Byzantine mass and can be seen even using relatively small samples. Ik, now generally known as Family Π, is a genuine and highly recognizable Byzantine subgroup. The most recent work, that of Wisse, finds Family Π to involve in excess of 100 manuscripts, and breaks it down into two basic groups (Πa and Πb) plus ten sundry clusters. Of these, Πa is the largest (65 members) and most significant, containing the two uncials K and Π (both of which Wisse calls core members of the group) and many minuscules. (The other Family Πa uncial, Y, Wisse places in the group Πa171.) Wisse also places A in the Πa group (an opinion first stated by the Lake/Geerlings studies), but admits it is a diverging member.
The Πa group is clearly distinct from the "mainstream" Byzantine text of Kx; in his three chapters of Luke, Wisse notes some three dozen places where Πa and Kx diverge (apart from passages where neither group formed a fixed reading), out of 196 passages tested. If one takes the readings noted in the footnotes of UBS/GNT, the number is somewhat smaller (on the order of 10-12% of the readings), but still large enough to allow easy recognition of Famiy Πa readings. The type is Byzantine, but few Byzantine groups differ so sharply from the Byzantine norm.
The other interesting point is that it is old. A is not a perfect member of this group, but it isn't a perfect member of the Byzantine text, either. Still, A attests to the existence of the Byzantine text in the fifth century -- and to the existence of Family Π in the same century. The earliest witnesses to the Kx/Ki/K1 group, by contrast, is E of the eighth century. Although Family Π did not prove to be the dominant Byzantine group (Kx certainly provides more manuscripts, and Kr probably does as well), the possibility must be considered that this is the earliest form of the Byzantine text.
About Π itself there is relatively little to add. The Alands, naturally, list it as Category V. Interestingly, however, it has obeli by John 5:4 and 8:3-6 (omitting the earlier portion of the pericope); we also note that Mark 16:8-20, while present and not marked doubtful, are not as fully annotated as the rest of the manuscript.

Φ (Phi, 043)

Tirana, Staatsarchiv Nr. 1. Formerly at Berat, hence the name Codex Beratinus. Von Soden's ε17. Dated paleographically to fifth (Scrivener) or sixth (Aland) century (Scrivener reports that it "may probably be placed at the end of the fifth century, a little before the Dioscorides (506 A.D.), and before the Codex Rossanensis." No supporting evidence is offered for this.) Purple parchment. Contains portions of the gospels of Matthew and Mark (the loss of Luke and John may be traced to "the Franks of Champagne."). Matt. 1:1-6:3, 7:26-8:7, 18:23-19:3, and Mark 14:62-end are lacking. Textually, Von Soden classified Φ as Iπ, that is, as part of the group which also contains N O Σ This assessment has been all but universally accepted, though assessments of the text of the group itself have varied. The Alands place all four manuscripts of the group (the Purple Uncials) in Category V, and it is certain that they are more Byzantine than anything else. Streeter, however, felt that the group had a "Cæsarean" element (for discussion, see the entry on N), which accords with Von Soden's view that they were members of the I text. Samples do not indicate a clear affiliation with any text other than the Byzantine (it should be noted, however, that their defects have kept the profile method from being applied to any of these manuscripts). Of the four, Φ is generally regarded as being the most unusual -- though this may be based primarily on a single reading, the "Western" addition in Matthew 20:28 about seeking what is greater (shared by D a b c d e ff1 ff2 hubmarg ox theo cur harkmarg?). Scrivener describes the writing as follows: " The pages have the κεφαλαια marked at the top, and the sections and canons in writing of the eighth century at the side. The letters are in silver, very regular, and clearly written. None are in gold, except the title and the first line in St. Mark, and the words Πατηρ, Ιησους, and some others in the first six folios. There is no ornamentation, but the first letters of the paragraphs are twice as large as the other letters. The letters have no decoration, except a cross in the middle of the initials O's. The writing is continuous in full line without stichometry. Quotations from the Old Testament are marked with a kind of inverted comma. There are no breathings.... Punctuation is made only with the single comma or double comma... or else with a vacant space, or by passing to the next line.... Abbreviations are of the most ancient kind." Edited by P. Batiffol in 1887.

Manuscript Ψ (Psi, 044)

Location/Catalog Number

Mount Athos, where it has been as long as it has been known. Catalog number: Athos Laura B' 52


Ψ originally contained the entire New Testament except the Apocalypse. All of Matthew, as well as Mark 1:1-9:5, have been lost; in addition, the leaf containing Hebrews 8:11-9:19 is lost. The Catholic Epistles have 1 and 2 Peter before James. Ψ is written on parchment, 1 column per page. It has been furnished with neumes -- one of the oldest manuscripts to have musical markings.


Usually dated paleographically to the eighth/ninth centuries; the latest editions (e.g. NA27) date it to the ninth/tenth centuries.

Description and Text-type

Ψ has an unusually mixed text. Aland and Aland list it as Category III in the Gospels, Acts, and Paul, and Category II in the Catholic Epistles. Von Soden lists it as generally Alexandrian.

In fact the situation is even more complicated than this. In Mark the manuscript is distinctly Alexandrian, of the sort of late, mixed cast we see, e.g., in L; like L, it has the double Markan ending. In Luke the manuscript loses almost all traces of Alexandrian influence and becomes predominantly Byzantine. In John the manuscript is mixed -- more Byzantine than anything else, but with significant numbers of Alexandrian readings.

In Acts Ψ is largely Byzantine.

In Paul Ψ is more Byzantine than anything else (it is perhaps the earliest substantial witness to that type), although there are certain Alexandrian readings (which seem to bear a certain similarity to those of P). The Alexandrian element seems to be slightly greater in the later books.

In the Catholics Ψ is again mostly Alexandrian, though with Byzantine influence. The text seems to be of the type found in A 33 81 436.

The distinction between the text of Mark and the other gospels has been known since the time of Lake, who published his collation in 1903. Lake compared the text of Mark against Westcott and Hort's text, and found that, in 480 or so places where the Textus Receptus disagrees with the WH text, Ψ agrees with the Textus Receptus in only 42 -- in other words, it has very few Byzantine readings. I count 17 readings which Lake lists as going with D and the Old Latins but not with WH, the TR, or the Old Syriac. He has seven which go with the Old Syriac against the WH, TR, and D; eight with agree with D, OL, and the Old Syriac against TR and WH; 16 or 17 which we would probably call "late Alexandrian" (agreeing with L or a similar manuscript without agreeing with B); 18 with B and no more than two other uncials; 25 singular or subsingular readings that do not appear to be errors, plus four pretty clear errors (several of which led Lake to conclude that, for Mark at least, Ψ's exemplar had roughly 19 to 22 letters per line. Which makes it rather likely, although by no means certain, that it had two columns per page).

Lake's collations of the other books show the shifting nature of Ψ's text. In Colossians, there are only 39 differences from the TR, some of which are probably readings where Ψ agrees with at least part of the Byzantine text and the TR doesn't. The collation of Luke is a little more than four pages long -- short, considering the length of the book. The collation of John is a bit less than four pages -- meaning that it has slightly more variations per unit of text.

Other Symbols Used for this Manuscript

von Soden: δ6


Kirsopp Lake, "Texts from Mount Athos," Studia Biblica et Ecclesiastica V (Oxford, 1903; there are now several low-quality print-on-demand reprints), pp. 105-122 prints the text of Mark in full; pp. 123-131 give collations of Luke, John, and Colossians

Sample Plates:
Metzger, Manuscripts of the Greek Bible (1 page)

Editions which cite:
Cited in all editions since von Soden.

Other Works:
Kirsopp Lake, "Texts from Mount Athos," Studia Biblica et Ecclesiastica V (Oxford, 1903; there are now several low-quality print-on-demand reprints), pp. 95-104 discusses this manuscript in some depth

Ω (Omega, 045)

Athos, Dionysiou (10) 55. Von Soden's ε61. Dated paleographically to the ninth century (Scrivener says eighth or ninth). Contains the four gospels complete almost complete; part of Luke 1 is missing. Textually, Von Soden classified Ω as K1, which Wisse modifies to Kx Cluster Ω (which is, however, just another name for the same thing). It has in fact been suggested that Ω is a good standard for Kx. The Alands generally confirm this assessment by placing Ω in Category V.

Manuscript 046

Rome, Vatican Library Greek 2066. Soden's α1070; Tischendorf/Scrivener B(r). Contains the Apocalypse complete, along with much other non-Biblical matter (the Biblical text occupies folios 259-278) including homilies of Basil the Great and Gregory of Nyssa. It has been variously dated; Scrivener favours the eighth century, Aland the tenth. The text is of the Byzantine type (so von Soden, who listed it as K, and all experts since); 046 is the earliest manuscript of the main Byzantine group ("a"). The Alands therefore classify it as Category V, though early manuscripts of the Apocalypse are so rare that even a Byzantine uncial deserves special attention. Scrivener describes the writing thus: "the uncials being of a peculiar kind, leaning a little to the right; they hold a sort of middle place between square and oblong characters.... The breathings and accents are primâ manu, and pretty correct..." while the punctuation is fairly well evolved.

Manuscript 047

Princeton, New Jersey, University Library Med. and Ren. Mss. Garrett 1. Soden's ε95; original Gregory ב. Contains the Gospels with some mutilations (in Matt. 2-3, 28, Mark 5-6, 8-9, John 12, 14, and breaking off in John 17). Dated paleographically to the around the ninth century (Clark says eighth; von Soden and Lake the ninth, Gregory and Scrivener the ninth or tenth. Apparently originally from Mount Athos (Gregory and Lake both saw it there); what seems to be the earliest library stamp reads, in Russian, "Russian Athos Library of the Monastery of Saint Andrew." It was brought to the United States in 1925. Textually of no great interest; von Soden places it in I' (with such diverse manuscripts as P Q R Γ 064 074 079 090 0106 0116 0130 0131 4 162 251 273 440 472 485 495 660 998 1047 1093 1295 1355 1396 1604 2430), but the Alands simply list it as Category V (Byzantine), and Wisse corroborates this by placing it in Kx throughout. Legg cited it, but the newer editions generally do not. What interest 047 has is, therefore, derived from its format, for the manuscript is written in the form of a cross (photo in Aland & Aland and in Metzger, Manuscripts of the Greek Bible). It is believed that this is the only continuous-text cruciform manuscript (the lectionaries 𝓁233 and 𝓁1635 are also cruciform, and 𝓁2135 has some cruciform pages. So too the Latin codex Cavensis among others). This format has many drawbacks; it is very wasteful of writing materials (047 has about 37-38 lines per page; of these typically ten reach the full width of the page, with about twelve lines above and fifteen below being slightly less than half the available width. Thus about three-eights of the usable area of the page is blank), and the format makes it harder to use the marginalia. These are no doubt among the reason the format is so rarely encountered. The manuscript has some marginal corrections (including, e.g., one obelizing John 5:4).

Manuscript 048

Rome, Vatican Library Greek 2061. Soden's α1; Tischendorf/Scrivener ב(ap). Double palimpsest (i.e. the biblical text has been overwritten twice), resulting in a manuscript very difficult to read even on the leaves which survive (and the leaves which survive are few -- only 21 of what are believed to have been originally 316 folios. The leaves that make up 048 constitute folios 198-199, 221-222, 229-230, 293-303, 305-308 of Vatican Gr. 2061). These surviving leaves contain (according to NA27; other sources give slightly different contents, no doubt based partly on the illegibility of the manuscript) Acts 26:6-27:4, 28:3-31; James 4:14-5:20; 1 Pet. 1:1-12; 2 Pet. 2:4-8, 2:13-3:15; 1 John 4:6-5:13, 5:17-18, 5:21; 2 John; 3 John; Romans 13:4-15:9; 1 Cor. 2:1-3:11, 3:22, 4:4-6, 5:5-11, 6:3-11, 12:23-15:17, 15:20-27; 2 Cor. 4:7-6:8, 8:9-18, 8:21-10:6; Eph. 5:8-end; Phil. 1:8-23, 2:1-4, 2:6-8; Col. 1:2-2:8, 2:11-14, 22-23, 3:7-8, 3:12-4:18; 1 Th. 1:1, 5-6, 1 Tim. 5:6-6:17, 6:20-21, 2 Tim. 1:4-6, 1:8, 2:2-25; Titus 3:13-end; Philemon; Heb. 11:32-13:4. The hand is dated paleographically to the fifth century. The manuscript is one of the very few to be written with three columns per page. Due to the small amount of text, the manuscript's type has not been clearly identified. The Alands classify it as Category II, which is probably about right, but this is on the basis of a mere 44 readings in Paul. Von Soden did not classify it at all. Observation shows that it is clearly not Byzantine; the strongest element is probably Alexandrian, though some of the readings may be "Western."

Manuscript 049

Athos, Codex Lavra A' 88. Tischendorf Sap. Von Soden's α2. Contains the Acts and Catholic Epistles complete, and Paul with some lacunae (lacking 1 Corinthians 5:9-13:7; parts of Ephesians 1-2; parts of Philippians 2-3; and breaks off in 1 Thessalonians 4, lacking all of 2 Thessalonians, 1 Timothy, 2 Timothy, Titus, Philemon, and Hebrews. So the Alands, anyway; Scrivener has a somewhat different list of lacunae). Dated by modern sources to the ninth century; Scrivener says eighth or ninth. The text, based on the Aland statistics, appears extraordinarily dull; they count only 13 non-Byzantine readings in the whole manuscript, and place it in Category V. Von Soden too called it primarily Byzantine, although he would allow a little I ("Western") influence in Acts.

Manuscript 050

In four parts: Athens, National Library 1371 (2 folios); Athos, Dionysiu (71) 2 (7 folios); Moscow, Historical Museum V. 29, S. 119 (7 folios); Oxford, Christ Church College Wake 2 (3 folios). Tischendorf Oe (the Moscow portion), We (the Oxford portion, John 4:7-14). Von Soden's Cι1. Contains John 1:1-4, 2:17-3:8, 3:12-13, 3:20-22, 4:7-14, 20:10-13, 20:15-17. With commentary. Dated to the ninth century by the Alands. With so little text, it is hard to say much about its contents, but the Alands place it in Category III. Von Soden merely lumped it with the manuscripts containing the "anonymous catena."

Manuscript 053

Munich, Bavarian state library Greek 208, folios 235-248. Tischendorf Xb. Von Soden's A4. Contains Luke 1:1-2:40, with a commentary, the text and scholia reported by the Alands to alternate. Von Soden lists it among the manuscripts with Titus of Bostra's commentary. Dated to the ninth century by the Alands. With so little text, it is hard to say much about its contents; the Alands place it in Category V (Byzantine) but with a question mark. Wisse classifies it as Kx, which supports this belief. NA27 does not bother to cite it fully.

Manuscript 054

Rome, Vatican Library Barberini Gr. 521 (Scrivener says Barberini 225). Tischendorf Ye; Von Soden's ε54. Contains John 16:3-19:41. The original hand is believed to be from about the eighth century. According to Scrivener, it has been retraced "coarsely." The accents and breathings are relatively few and often incorrect; it has a peculiar system of punctuation with up to four points. It has the Ammonian sections but no Eusebian numbers. The Alands put it in Category V (Byzantine) with a question mark, but Von Soden wasn't sure whether it belonged with K or I (although he inclined toward the former), and Scrivener says the text is "midway" between that of A and B, which would imply a manuscript that belongs to about Category III.

Manuscript 055

Paris, National Library Gr. 201. Tischendorf/Scrivener 309e. Dated paleographically to the very end of the uncial period (e.g. Aland lists XI; Scrivener says X-XII). Despite being numbered among the uncials, it is not a true New Testament manuscript, containing rather a commentary with partial text (Chrysostom on Matthew and John, Victor on Mark, Titus of Bostra on Luke). Thus it has not been subjected to textual analysis; Von Soden did not even include it in his catalog (despite listing manuscripts of his A type with even less text), the Alands did not place it in a Category, and Wisse did not profile it. Such minimal evidence as is available indicates, however, that the text is Byzantine. The writing itself, as might be expected of a semi-uncial manuscript variously listed as an uncial and a minuscule, is reported as "very peculiar in its style and beautifully written."

Manuscript 056

Paris, National Library Coislin Gr. 26. Soden's O7; Tischendorf/Scrivener 16a, 19p. Contains the Acts and Epistles complete, although it appears some pages of prefatory material is lost. Hebrews follows Philemon. Dated paleographically to the tenth century or even after (Scrivener lists the eleventh century). There are some later verses added at the end. Commentary manuscript; the main text is in small red uncial, the commentary in minuscule. The commentary is described by Scrivener as "like" that of (the pseudo-)Oecumenius, and of course Soden lists 056 among the Oecumenius manuscripts. The Paris Library web site also lists the commentary as by Oecumenius, but that it has been expanded. Formerly from the Great Lavra on Athos. The manuscript also includes, according to Scrivener, "a catena of various fathers [and] a life of St. Longinus on two leaves [ix]." Textually, 056 has been little studied; Soden simply listed it as having the Oecumenius text. The Alands correctly place it in Category V (Byzantine). This is elaborated somewhat by Wachtel, who lists it among the manuscripts which are 10-20% non-Byzantine in the Catholic Epistles, pairing it with 0142 (also an Oecumenius manuscript, Soden's O6) and 1066 (another Oecumenius text, though this one exists only in the Acts and Catholic Epistles; Soden's Oπρ21). That 056 also goes with 0142 in Paul and the Acts is easily demonstrated; indeed, they seem to be closer than we would expect even of Oecumenius texts, and probably go back to a recent common exemplar. In Acts, for instance, the two agree in 184 of 189 test readings (the test readings being of UBS3 for which both exist, including a subsingular reading in Acts 28:14, επι, found in 056 0142 pesh). For comparison, 056 agrees with other Byzantine witnesses as follows: L, 127 of 141; P, 172 of 183; 049, 174 of 190, 1241, 170 of 187. The five differences between 056 and 0142 in the test readings in Acts are as follows:

ReadingText and Supporters of 056Text and Supporters of 0142
Acts 5:16 εις Ιερουσαλημ D E P Byz Ιερουσαλημ 𝔓74 ℵ A B 0189 a gig vg
Acts 10:5 ος; ℵ E P 33 Byz τινα ος A B C 81 1739 a vg
Acts 11:9 απεκριθη δε φωνη εκ δευτερου εκ του ουρονου 𝔓45 𝔓74 ℵ A 049 81 1739 gig vg απεκριθη δε μοι φωνη εκ του ουρονου (singular reading, probably a parablepsis for the reading απεκριθη δε μοι φωνη εκ δευτερου εκ του ουρονου of P Byz)
Acts 13:42 παρκαλουν τα εθνη εις το P Byz παρκαλουν εις το 𝔓74 ℵ A C (D) 33 81 1739 al
Acts 27:5 καθηλθομεν 𝔓74 ℵ A B P 33 81 1739 Byz gig καθηλθομεν δι εμερων δεκαπεντε (singular reading, probably derived from the καθηλθομεν δι δεκαπεντε εμερων 614 1518 2138 2147 2412 a h hark**)

Thus it would appear that, if anything, 0142 is the ancestor of 056, but examination of the data in Hebrews makes it appear more likely they are derived from a common exemplar, with 0142 perhaps copied slightly earlier. A notable peculiarity of both manuscripts is the use of extra iotas at the end of words. Most of these (perhaps all of them) are instances where an iota would normally be found subscripted, but neither manuscript is consistent in this usage.

Black-and-white scans of 056 (from a microfilm, and not always easy to read) are at;2.

Manuscript 061

Paris, Louvre MS. E 7332. Tischendorf's Tg; Scrivener's T or Tp; Von Soden's α1035. Contains a small fragment of 1 Timothy, 3:15-16, 4:1-3, 6:2-4, 5-8, on two leaves, both damaged. Dated paleographically to the fifth century by most authorities; Scrivener says IV or V. Based on this date, it is very surprising to find the Alands classifying it as Category V -- and even more surprising to find them calling it Category V with singular readings (!). They do not make it easy to check the point, however, as 061 is not cited in the Nestle-Aland edition. Nor did Von Soden classify the manuscript. It must be regarded as a small question mark in the manuscript lists.

Manuscript 065

St. Petersburg, Russian National Library Gr. 6 I, folios 1-3. Tischendorf's I(1); Von Soden's ε1. Contains John 11:50-12:9, 15:12-16:2, 19:11-24 on three leaves. Palimpsest; the upper text is a Georgian calendar. This 28-leaf document contains palimpsest fragments of seven different Biblical manuscripts. Dated paleographically to the sixth century by the Alands; Scrivener seems to argue for the fifth. Tischendorf found it in 1853 in St. Saba. It is said to be very hard to read, which perhaps helps to explain why the Nestle-Aland edition does not cite it. Scrivener's data seems to imply a Byzantine text, but Von Soden classified it as H. The Alands, however, relegate it to Category V.

Manuscript 066

St. Petersburg, Russian National Library Gr. 6 II, folio 4. Tischendorf's I(2); Von Soden's α1000. Contains Acts 28:8-17. Palimpsest; the upper text is a Georgian calendar. This 28-leaf document contains palimpsest fragments of seven different Biblical manuscripts. Dated paleographically to the sixth century by the Alands. Von Soden classified it as Ib; the Alands place it in Category III. The Nestle-Aland edition does not often cite it, but in five of the six instances where it is cited, it goes against the Byzantine text.

Manuscript 067

St. Petersburg, Russian National Library Gr. 6 III, folios 7-12. Tischendorf's I(3); Von Soden's ε2. Contains Matthew 14:13-16, 19-23, 24:37-25:1, 25:32-45, 26:31-45, Mark 9:14-22, 14:58-70 (with some defects in the Markan leaves). Palimpsest; the upper text is a Georgian calendar. This 28-leaf document contains palimpsest fragments of seven different Biblical manuscripts. Dated paleographically to the sixth century by the Alands. Von Soden classified it as I' in Matthew, Iα in Mark; the Alands place it in Category III "influenced by V" (whatever that means).

Manuscript 068

London, British Library MS. Additional 17136, folios 117 and 126. Tischendorf's Ib, Scrivener's Nb; Von Soden's ε3. Contains John 13:16-17, 19-20, 23-24, 26-27, 16:7-9, 12-13, 15-16, 18-19. Double palimpsest; the upper texts are Syriac, with the top text being hymns of Severus. Dated paleographically to the fifth century by the Alands; Tischendorf could not decide whether it was fourth or fifth century. Von Soden classified it as H; the Alands place it in Category III. This seems about right; based on the Nestle apparatus, it has some Alexandrian readings, but seems a little more Byzantine than anything else. Its most noteworthy reading may be in John 13:24, where it agrees with B C L 33 892 in reading και λεγει αυτω ειπε τις εστιν against 𝔓66 ℵ A (D) W Byz UBS3.

Manuscript 069

University of Chicago, Oriental Institute 2057. Formerly Haskell Oriental Museum, Oxyrhynchus Papyrus 3. Recovered by Grenfell and Hunt and granted to the University of Chicago in 1904. Originally designated Tg; von Soden's ε12. Dated paleographically to the fifth (so the KListe) or sixth century. A single folio, mutilated, with seven lines remaining out of probably about 25 per page as originally written. Contains portions of Mark 10:50, 51, 11:11, 12. Von Soden classified it as K; the Alands, somewhat surprisingly, make it Category III. This even though Clark reports "No variant from ς," which would imply a purely Byzantine text -- and Clark's assessment appears correct. The only reading for which it is cited in NA26 and later editions is Mark 11:11, where it reads οψιας along with A B D K N W Γ Δ Θ Ψ Byz. UBS4 lists it as a witness but does not list any variants in the part of Mark for which it is extant. Legg lists it as reading αναστας in Mark 10:50 with A C W Byz against αναπηδησας of the modern editions, omitting λεγει αυτω ο Ιησους with A W Byz in 10:51, and omitting θελεις ποιησω σοι with A D W Byz in 10:51.

Manuscript 071

Harvard University, Semitic Museum Papyrus 3735. Oxyrhynchus Papyrus 401. Recovered by Grenfell and Hunt and granted to Harvard in 1904. Von Soden's ε015. Dated paleographically to the fifth or sixth century. A single folio, one column, 13 lines per column. Contains portions of Matthew 1:21-24 and 1:25-2:2. Von Soden classified it as H; Clark says it goes with ℵ. the Alands assign it to Category II, although it doesn't have any of their sample readings. It is cited only three times in the Nestle apparatus, agreeing with ℵ B Z in all three readings.

Manuscript 076

Pierpont Morgan Library (New York), Pap. G. 8 (Amherst VIII). It had a complex history before it reached its present home. Recovered by Grenfell and Hunt, it went into the collection of Lord Amherst of Norfolk; after his death in 1909, Morgan acquired it in 1913. Von Soden's α1008; also briefly designated יa. Dated paleographically to the fifth or sixth century. A single folio, of medium size, two columns, 23 lines per column. Contains portions of Acts 2:11-22. It reportedly has suffered more damage since being found. Von Soden classified it as H. the Alands assign it to Category II. The Nestle apparatus cites it only twice (Acts 2:12, 17). In the first reading, it agrees with ℵ A B; in the latter, only with B. Grenfell and Hunt, however, thought they read a reading that agreed with D against B, so perhaps it should be given a more detailed examination.

Manuscript 085

Currenty Saint Petersburg, Russian National Library Greek 714 (formerly at Cairo). Tischendorf's Tk; Scrivener's Tg; Von Soden's ε23. Contains a small fragment of Matthew, 20:3-32, 22:3-16. Dated paleographically to the sixth century by most authorities, though Scrivener allows the possibility of a seventh century date. He notes that the letters resemble Coptic. Textually, it is regarded as Alexandrian; Von Soden lists it as H, while the Alands place it in Category II. A spot check seems to show a mixed manuscript; looking at the places in Matthew 22 where NA27 cites 085 explicitly, we find 32 cited readings of 085, of which 16 agree with ℵ, 16 with B, 19 with D, 22 with L, 17 with Θ, and 12 with the majority text; a couple of readings are subsingular.

Manuscript 095 and 0123

Saint Petersburg, Russian National Library. 095 is MS. Greek 17; 0123 is the first folios of MS. Greek 49. 095 is Von Soden's α1002 and Tischendorf/Scrivener Ga; 0123 is Von Soden's α1014 (Gregory 70apl, possibly Scrivener 72apl, though Scrivener's list gives a ninth century date and gives the contents incorrectly. Scrivener also lists it as a palimpsest, but the Alands simply list it as fragments; one must assume that portions of this manuscript, so fragmented as to be mistaken for a lectionary, are partly palimpsest). Scrivener dates it to the seventh century, but the Alands describe it as being from the eighth. Scrivener reports that the portion known as 095 (his G or Ga) was "written in thick uncials without accents, torn from the wooden cover of a Syriac book." 095 contains Acts 2:45-3:8. The portion known as 0123 consists of fragments with parts of Acts 2:22-28. It is difficult to assess the manuscript's type because of its small size. Von Soden listed 095 as H (Alexandrian), and the Alands list it as Category III, while Scrivener admits "a few rare and valuable readings." If we examine the apparatus of Nestle-Aland27, we find the manuscript cited explicitly only six times; in these, it agrees with A and C five times (and with 𝔓74 in all four readings for which both are extant); with E, Ψ, 33, and 1739 four times; with the Majority Text three times; and with ℵ, B, and D twice. If such a small sample means anything at all, it would seem to imply a late Alexandrian witness.

Manuscript 098

Grottaferrata, Biblioteca della Badia Z' α' 24. Tischendorf Rp. Von Soden's α1025. Palimpsest, containing only 2 Corinthians 11:9-19. The upper writing is the Iliad. Dated by the Alands to the seventh century; Scrivener (who gives the catalog number as Z' β1) says seventh or eighth. With such a small amount of text, it is hard to assess the manuscript, but von Soden considered it to be non-Byzantine, although he couldn't decide whether it was H or I. The Alands give it the amazing status of a Category I manuscript. This is quite surprising given that they cite it only twice, and neither reading is particularly unusual although neither one is Byzantine.

Manuscript 0120

Rome, Vatican Library Greek 2302. Tischendorf/Scrivener Gb; Von Soden's α1005. Palimpsest, six folios (only five of which had been discovered by Scrivener's time), consisting of pp. 65-66, 69-72, 75-76, 79-94 of the upper manuscript (a menaeon). The manuscript consists of Acts 16:30-17:17, 17:27-29, 31-34, 18:8-26. The date of the manuscript is most uncertain; the date listed in Scrivener (apparently from Gregory) is fourth century (with a question mark); the Alands date the manuscript to the ninth century! (In favour of the later date is the fact that the Alands will have examined the manuscript using more modern methods.)
0120 is rarely cited; it is not, e.g., a "constant witness" in the Nestle-Aland text. We are, as a result, largely dependent on the classifications of others. The Alands list 0120 as Category III. Von Soden listed it as Ib1, which (if accurate) is very interesting; Ib1 is the group containing witnesses such as 206 429 522. In other words, in Acts, this is a weaker branch of Family 1739. Unfortunately, we must remind ourselves that Von Soden's results are anything but reliable, particularly for fragments. Clearly a more thorough examination is called for.

Manuscripts 0121 and 0243

Location/Catalog Number

0121: London. British Museum Harley 5613.
0243 (Corinthian portion): Venice. San Marco Library 983 (II 181)
0243 (Hebrews portion): Hamburg. Univ. Libr. Cod. 50 in scrin.


As currently designated (the designations have changed over time):
0121 contains 1 Cor. 15:42-end, 2 Cor. 1:1-15, 10:13-12:5
0243 contains 1 Cor. 13:4-end and all of 2 Cor.; Heb. 1:2-4:3, 12:20-end.


Henke, who first published the Hebrews portion (the first to be found), believed it to be the two outer leaves of a six-leaf quire. He noted that it uses only a single point as punctuation, except that there is a single instance of an interrogative mark. It has accents and breathings, but no iota adscript. This portion was known as Codex Uffenbachianus, after M. von Uffenbach, its first known possessor, who suggested a seventh or eighth century date. Wettstein gave an eleventh century date. Henke moved this back to the ninth century. Since then, we've reached a sort of compromise: both 0121 and 0243 generally dated to the tenth century (so, e.g. NA27).

(Facsimile of 0121)
Facsimile of 2 Cor. 1:3-5 in 0121 (after Scrivener). Colors are exaggerated and manuscript is enlarged. The unaccented text reads

G. Zuntz, however (The Text of the Epistles, London, 1953, pp. 74, 286-287) states that 0121 "is by no means an 'uncial': its letters are the kind of majuscule which scribes of the tenth and later centuries often used to distinguish marginal scholia from the text. In M [=0121] these majuscules contain a significant admixture of minuscule forms.... I should ascribe M to the twelfth century." (See facsimile above right.)

Both 0121 and 0243 are written in red ink on parchment, two columns per page, which explains why their contents were confused for a time. Although they are not the same manuscript, they were probably the product of the same school of design, and perhaps the same scriptorium.

Description and Text-type

Before we can describe these manuscripts, we must describe their recent history, since it affects how they are cited in the various apparatus. When the first two portions of the manuscripts (what we now call 0121 and the Hebrews portion of 0243) were discovered, it was observed that both were of about the same date, that both were in red ink, that they had similar texts, and that both were in two columns on parchment. It was naturally assumed that they were the same. In Tischendorf, the fragments were referred to as M. In the Gregory catalog, this became 0121. Then Birdsall observed that the two were in distinct hands. So the Corinthian portion became 0121a and the Hebrews portion 0121b. They were cited in this way in NA26.

Of course, many manuscripts are written by more than one scribe. This was not really proof that they were originally separate. But more was forthcoming.

At about the same time Birdsall discovered that the two were by different scribes, the larger (Corinthian) portion of 0243 came to light. Some time later, it was realized that this was part of the same manuscript as 0121b. This finally proved that 0121a and 0121b were separate (since 0121a and 0243 overlap). So 0121b was renumbered 0243 and 0121a became 0121. This is how things are cited in NA27 and after..

If this is confusing, maybe this table will help:

ContentsTischendorf SymbolGregory Symbol NA26 symbolNA27 symbol
1 Cor. 15:42-end, 2 Cor. 1:1-15, 10:13-12:5 M01210121a0121
Heb. 1:2-4:3, 12:20-end M01210121b0243
1 Cor. 13:4-end and all of 2 Cor. --02430243

In all this shuffling, one thing remains certain: Both manuscripts are closely affiliated with 1739. 0243 is a probably a first cousin (perhaps even a sister); 0121 is a cousin or descendant.

Several striking examples of agreements between 0243 and 1739 may be cited. Perhaps the most noteworthy is Hebrews 2:9, where 0243, 1739*, and perhaps 424**, alone among Greek manuscripts, read ΧΩΡΙΣ ΘΕΟΥ instead of the majority reading ΧΑΡΙΤΙ ΘΕΟΥ.

The reader who wishes further details, including a comparison of the readings of 0121 and 1739, is referred to the entry on 1739 and family 1739.

Von Soden lists 0121 as H. Aland and Aland list 0121a as Category III and the Corinthian portion of 0243 as Category II (its sister 1739 is, however, a Category I). 0121b is still in their list, and is Category III (!).

Other Symbols Used for this Manuscript

For 0121: von Soden: α1031. Tischendorf: Mp

J.N. Birdsall, A Study of MS. 1739 and its Relationship to MSS. 6, 424, 1908, and M (unpublished Ph.D. dissertation, 1959)

A collation of the Hebrews portion of 0243 is available here.

Sample Plates:
The British Library portion has now been scanned:*&index=885.

Editions which cite:
Cited in NA26 as 0121a, 0121b, and 0243.
Cited in UBS3 as 0121a, 0121b, and 0243.
Cited in NA27 and higher as 0121 and 0243.
Cited in UBS4 and higher as 0121 and 0243.
Von Soden, Merk, and Bover cite the "M" portions.

Other Works:

Manuscript 0122

Saint Petersburg, Russian Pubic Library Greek 32. Soden's α1030; Tischendorf/Scrivener N(p); Hort's Od. Two folios containing small fragments of Paul: Gal. 5:12-6:4, Heb. 5:8-6:10. Dated paleographically to the ninth century. Textually, the Alands have assigned it to Category III, but Von Soden listed it as K (purely Byzantine), and the latter assessment seems to be correct. An examination of its readings in Galatians reveals the following departures from the Byzantine text:

VerseByzantine reading0122(*) readscomment
5:12αναστατουντες 0122* ανασταντουντες singular; probable copying error at some stage
5:14εαυτον 0122* σεαυτον Byzantine text divided
5:17 α αν 0122* α εαν also found in ℵ A pc
5:22δε 0122c omitssingular reading
5:23εγκρατεια 0122c εγκρατεεια υπομονη singular reading
5:24Ιησου Χριστου 01221 Χριστου Ιησου 0122*, 01222 with the Byzantine text
6:1προληφθη 0122c προσληφθη singular reading
6:3τι 0122c omitssubsingular, found also in B* 075c

It will thus be observed that all deviations from the Byzantine text are relatively trivial and generally poorly supported. I have not examined the portion in Hebrews in detail, but the Nestle apparatus makes it appear that 0122 is equally Byzantine there. It will be observed that the manuscript has been fairly heavily corrected (observe the double correction in Gal. 5:24), but the corrections have no more significance than the original text; indeed, in this admittedly tiny sample they seem simply to be more idiosyncratic.

Manuscript 0130

St. Gall, Stiftsbibliothek 18 (pp. 143-146); Stiftsbibliothek 45 (pp. 1-2); Zürich, Central Library MS. C.57 (folios 5, 74, 93, 135). Mark 1:31-2:16; Luke 1:20-31, 64-79, 2:24-48. Von Soden's ε80; Tischendorf's Wc. Dated paleographically to the ninth century. Palimpsest (except for part of St. Gall 18 p. 146 which was not erased); the upper text of St. Gall 18 is a twelfth century Vulgate psalter; St. Gall 45 is a thirteenth century Vulgate; the Zürich portion also has a thirteenth century upper text. Von Soden classified it as I', which in a fragment so short probably means only that it is not purely Byzantine. The Alands list it as Category III "influenced" by V (in other words, mostly but not entirely Byzantine) but offer no statistics about its text. Legg cited it only for Mark 2:8-16. Very likely originated in St. Gall, where Greek was still used (if poorly understood) long after it had been forgotten in the rest of the western church; St. Gall 18 also includes (p. 4) sample Greek uncial and minuscule alphabets, presumably to help scribes read or learn them.

Manuscript 0145

Current location unknown. Formerly at Damascus, Kubbet et Chazne. Von Soden's ε013. Fragment of a commentary manuscript, with the only surviving portion being John 6:26-31. Dated to the seventh century. Von Soden tentatively classified it as a K (Byzantine) manuscript but with a possibility that it was H (Alexandrian). The Alands list it as Category III but offer no statistics about its text. With so little text to work with, we cannot really hope to offer a judgment -- but we observe that it has the Byzantine reading in the only place in NA27 where it is cited (John 6:29). There are three readings in SQE13; two are Byzantine and one is subsingular. Probably we should simply call the manuscript "unclassifiable."

Manuscript 0206

This manuscript may have had a complex history. Found by Grenfell and Hunt, and numbered Oxyrhynchus Papyrus 1352, it was granted to Bonebrake Theological Seminary in Dayton, Ohio, in 1924, but was not to be found there when Clark looked for it in the 1930s. But it's now in the United Theological Seminary in Dayton, the successor of Bonebrake. Also, apparently someone wanted to number it 𝔓25 before it was noticed that it wasn't on papyrus! Contains 1 Peter 5:5-13. Dated to the fourth century. The writing is said to be large and somewhat similar to ℵ. The surviving page is described as damaged and worm-eaten by Clark, but most of the original page survives. The Alands list it as Category III but do not cite it in the Nestle-Aland edition. It does have at least one subsingular reading, although we don't know exactly what it is; in 1 Peter 5:7, for the επιριψαντες of ℵ A B* and επιρριψαντες of B* 𝔐, Das Neue Testament auf Papyrus reports it has [????]ψατε. In 5:8 it has καταπειν with ℵ*, but this might well be orthographic. In 5:9 it omits τω before κοσμω along with A and all other witnesses except 𝔓72 ℵ* B; later in the verse, it agrees with ℵ A B* K 33 1611 2344 in reading επιτελεισθε. In 5:10 it has ημας for υμας with 442 2492 vg (but that may not be genetic); it appears to omit Ιησου with ℵ (B) 1611 hark and to omit θεμελιωσει with A B Ψ. In 5:11 it seems to omit any of the variants which include the word δοξα but has the long reading αιωνας των αιωνων with ℵ A 𝔐 against 𝔓72 B 307 bo. In 5:12 it omits του before θεου with 𝔓72 Ψ. It will be evident that the text is interesting; unfortunately, with so little text to work with, we can't really be sure of its affinities.

Manuscript 0212

New Haven, Yale University Library P. Dura 10. 0212 is not technically a New Testament manuscript; rather, it is a fragment of a gospel harmony. It was discovered in the ruins of Dura Europus in 1933. Since Dura was a Roman fortress town sacked by Shapur I of Persia in 256/7 C.E., the assumption is that the manuscript was written in the first half of the third century, though an earlier date cannot be excluded. The fragment was found in an earth embankment believed to have been built for the final defense of the town. It was fairly close to a small Christian chapel, but far enough away that it may have come from some other source. Physically, the surviving fragment (usually regarded as only a portion of a leaf, though the edges are sharp and some seem to have been cut with a knife) measures 10.5 cm by 9.5 cm. It is written on only one side, and may well have come from a scroll. (The most recent study of the manuscript, D. C. Parker, D. G. K. Taylor, M. S. Goodacre, "The Dura-Europos Gospel Harmony," published in Taylor, Studies in the Early Text of the Gospels and Acts, concludes that it is definitely a scroll, not a codex, based on observations of holes along one edge which seemingly correspond to stitches. I think this very likely, but I would not consider it definitive; the holes do appear to be there, but they might arise from, say, pricking lines to align the text, and the manuscript would naturally tear where pricked.) The surviving column originally contained about 30-35 letters per line (with the first five or more letters lost, and with additional damage to certain of the lines). Portions of fourteen lines survive, although the last two lines, especially the very last one, are almost gone, and the five lines above that all have damage; depending on who is reading it and how confident they are in their eyesight, there are probably fewer than 250 letters all told (I personally counted 171 that I thought legible in the photograph). As noted, it is a gospel harmony, containing phrases seemingly from Matt. 27:56-57, Mark 15:40, 42, Luke 23:49, 50, 51, John 19:38. (So Kraeling, who first edited the manuscript; for this transcription, see e.g. Metzger, Manuscripts of the Greek Bible, p. 66. The reconstruction of Parker, Taylor, and Goodacre, found in the article cited above, differs in only a few particulars, though some of the differences are significant). The manuscript has some unusual orthographic features, including the Nominum Sacrum στα -- an abbreviation found nowhere else, with uncertain meaning.
0212 has generally been regarded as a fragment of Tatian's Diatessaron, though the small size of the fragment meant that this was never certain. Parker, Taylor, and Goodacre, upon detailed examination and comparison with recent studies, are convinced that the fragment is not Tatianic, but is a fragment of a separate Gospel harmony (perhaps devoted solely to the passion narrative), compiled in Greek from Greek sources.
Since 0212 is not a New Testament fragment, the Alands did not analyse it, and it is too recent to have been analysed by Von Soden. It appears to contain a unique reading in Luke 23:49, referring to the wives of Jesus's disciples. This text is, however, only partly legible.
High-resolution scans are available on the Yale University site at

Manuscript 0219

Vienna, Austrian National Library, P. Vindobonensis 26083 and 36113. As the catalog numbers indicate, it consists of two fragments, which were found separately. The first one found contained Romans 2:21-23; the second, Romans 3:23-25, 27-30. The first fragment has portions of eight lines, the second has portions of ten; it is thought the original had two columns per page, with 26 lines per column. Found in the Fayyum. Dated paleographically to the fifth century by Treu; the Alands say fourth or fifth. The full text is printed in in G. H. R. Horsley, New Documents Illustrating Early Christianity, 2, 1982, p. 134. The Alands place it in Category III. It is cited for only two readings in NA28, and even those are uncertain; it appears in Romans 3:8 to add εφ ημας after ελθη along with 81 a and the Bohairic (the actual surviving text is εφ η; the rest is missing), and it appears to omit της before πιστεως in Romans 3:25 along with ℵ C* D* F G 365 1505 1506 1739 1881. This reading is based solely on space considerations; the manuscript has the first two letters of the preceding δια but nothing else. Horsley also thinks that, in 3:30, it has επειπερ along with ℵc D* F G L P 33 1175 1505 1881 Byz rather than ειπερ of ℵ A B C 6 81 365 1506 1739, but this argument is based entirely upon space considerations; NA28 refuses to cite it for this variant, and given that there are only two letters per line at that point in the parchment, I think this correct. If these readings are correct, though, what we have is a mixed text -- one reading found elsewhere mostly in late Alexandrian texts, one that is perhaps originally "Western," one Byzantine.
Based on codicological considerations, Treu believed that this manuscript contained only the Pauline Epistles, but Horsley has shown that this was a miscalculation. All that is certain (or nearly) is that Romans began at the top of a page, very likely the first page.

Manuscript 0243

See: 0121 and 0243

Manuscript 0253

Fromerly at Damascus, Kubbet el Chazne, but it had vanished from sight even before the troubles in Syria. Fortunately, it had been carefully examined before it was lost. Tiny fragment containing Luke 10:19-22. A very narrow column of parchment; the full height of the page (31 cm.) has been preserved, but only about 9 cm. of a page width estimated at 25 cm. The Greek text is of the sixth century; it was overwritten (without being erased) by an Arabic religious text. The hand is quite large, there are 14 lines per page. Only about seven letters per line have been preserved. The line lengths (based on the full text printed in G. H. R. Horsley, New Documents Illustrating Early Christianity, 2, 1982, p. 128) are very irregular; excluding one short line, those on the recto range from one line with twelve letters to two with seventeen letters, with the average being about fourteen letters per line. Thus we have about half the text of this one leaf. Based on the surviving portion, the use of the nomina sacra is very inconsistent, although it may simply be that the scribe had a different list. Treu, who edited the piece, estimated that it would have required about 250 leaves just to hold the Gospel of Luke, so it is almost certainly a copy of Luke alone, not of the four gospels. The Alands list it as Category V. The manuscript exists for only a handful of variants:

It will be evident that 0253 does agree primarily with the Byzantine text. On the other hand, it does have one unusual reading, and this is a very short sample. On the gripping hand, it does not attest to any particularly significant or rare readings. The Aland declaration that it is Byzantine may not be entirely true. But it is certainly true that 0253 doesn't tell us much that is useful.

Manuscript 0254

Lost. Formerly Damascus, Qubbat al-ḫazna (Kubbet el Chazne). Fortunately, B. Violet, who found it, photographed one side of it in 1901. Palimpsest, with Arabic upper writing. A single leaf with portions of Gal. 5:13-17. In the photograph, there are twenty lines, of which eighteen are legible. Dated paleographically to the fifth century. The surviving text is printed in G. H. R. Horsley, New Documents Illustrating Early Christianity, 2, 1982, p. 137 (although it incorrectly states that it contains Gal. 3:13-17). The Alands list it as Category I but offer no statistics -- and NA28 cites it for only one reading, Gal. 5:14, where it reads πεπληρωται with 𝔓46 ℵ A B C 0278 33 81 104 326 330 1175 1739 against πληρουται of D F G K L P Ψ 056 075 6 1505 1611 1881 2495 Byz vg and ανακεφαλαιουται of 365 1319 1573 (i.e. Family 1319/2127). But there are many variants in these verses, and it appears 0254 agrees with the NA/UBS text in all of them (and with B everywhere except in 5:14, where B* has αγαπησις instead of αγαπησεις, a reading corrected in Bc). Little surprise, then, that the Alands like it a lot. It is so short, however, that it really adds very little to our knowledge. It would be fascinating if it could be found and its other side read.

Manuscript 0259

Berlin, Staatliche Museen, P. 3605. Thought to be from the Fayum. A strange almost-square leaf, with very small pages (about 8 cm. high by 9 cm. wide), and one page blank; it has been suggested that it was a school practice exercise (although I wonder if it wasn't a leftover scrap that someone was using to test an ink or something). The text contains 1 Timothy 1:4-5, 6-7. The surviving text is printed in G. H. R. Horsley, New Documents Illustrating Early Christianity, 2, 1982, p. 137. Dated paleographically to the sixth century. The Alands list it as Category III but do not cite it in NA28. The manuscript has an interesting and somewhat unusual mark, ``, to mark sense breaks. The only variant in NA28 to which it attests is
1:4 -- οικονομιαν ℵ A F G K L P Ψ 33 81 365 1505 1739 1881 Byz versus οικοδομην D(2) a b d vg.
0259 does have one unique reading in 1:6, where it has εξετραπτησαν for εξετραπησαν (which might just be a scribal error).
There are a few other variants in these verses where one uncial, or the related uncials FG, deviate from the reading of NA28, but the uncials ℵ A L Ψ 056 075 0142 and the minuscules 223 876 1739 1022 1960 all agree with 0259 for every reading it has, and D agrees with it for every reading except the variant in 1:4. Thus any classification of the manuscript is really pretty meaningless; agreement about the reading of these verses is too close to universal.

Manuscript 0260

Berlin, Staatliche Museen, P. 5542. Tiny fragment containing a few letters of John 1:30-32 on one side, and a Fayyumic Coptic text of John 1:16-18 on the other. The full Greek text, but not the Coptic, is printed in G. H. R. Horsley, New Documents Illustrating Early Christianity, 2, 1982, p. 131. It begins with [γεγο]νεν [οτι] πρω[το]ς and ends with τυρησεν Ιςαννη[ς] Portions of just sixteen lines survive, with no more than seven extant letters per line; four or five is more typical, and some have fewer. It seems likely that it was originally in two columns, but this cannot be certain -- the extant columns are very narrow, typically six or seven letters per column, so there might well have been many columns. (Indeed, were it not for the fact that both Greek and Coptic are from near the beginning of John, I might have speculated that this was a scroll with Greek on one side and Coptic on the other.) Dated to the sixth century. The Alands tentatively list it as Category III but offer no statistics. The only common variant to which it appears to testify is in John 1:30; it omits τω before υδατι with 𝔓45vid 𝔓66 𝔓75 ℵ B C G L P T Wsupp Θ Ψ 0233 f1 33 69 579 788 892 1241 1424; A E F H K N Γ Δ Ω 13 346 543 565 700 1006 1342 1506 pm add the article. Although most Byzantine manuscripts include τω, enough omit it that this single variant really tells us nothing at all about the manuscript's type.

Manuscript 0261

Berlin, Staatliche Museen, P. 6791, 6792, 14043. Three fragments containing parts of Galatians 1:9-12, 19-22, 4:25-27, 28-31. The two fragments containing chapter 1 are from a single leaf, one fragment containing eleven lines, the other seven, from the middle and lower portions of the leaf; the portion from chapter 4 contains twenty lines and is from the upper part of the leaf. It appears that there were two columns per page and probably 25 lines per column. On this basis Turner believed that the fragments are from the first and seventh leaves of a single quaternion, but I don't think this can be regarded as proved. The full Greek text is printed in G. H. R. Horsley, New Documents Illustrating Early Christianity, 2, 1982, p. 136; most of it can also be reconstructed from Reuben Swanson's New Testament Greek Manuscripts: Galatians. Dated paleographically to the fifth century. Horsley calls it Alexandrian, but the Alands list it as a mere Category III. Its use of the nomina sacra is curiously inconsistent, and the line lengths are curiously irregular; the columns seem to have been narrow, from a minimum of about nine to a maximum of about thirteen letters; eleven or twelve letters is typical. This wide variation makes it difficult to judge add/omit variants on the basis of space (e.g. in Gal. 1:10, for the variant add/omit γαρ between ει and ετι, adding the conjunction gives us a line of 14 letters, omitting gives us 11. In 1:11, δε gives us twelve or thirteen letters in the line, γαρ or ουν thirteen or fourteen. The shorter readings are more probable but by no means sure).
The only readings for which it is cited in NA28 are:

In addition, in 1:19, where ℵ A B 33 1739 etc. read ουκ ειδον, D* F G read ειδον ουδενα, and 𝔓46 has ουκ ειδον ουδενα, all that survives in 0261 is the final ν, but this plus the space available makes it extremely likely that it read ουκ ειδον.
In 4:26, it cannot have the reading (η) συστοιχουσα of (D*) F G, although we cannot be entirely sure that it has the reading συστοιχει δε of the other manuscripts; the surviving text is [..]οιχει δε.
In 4:30, it cannot have the reading μου Ισαακ of D* F G;, although we cannot be sure that it has the reading της ελευθερας of the other witnesses; the surviving text is ελευ[.....].
Thus 0261 is pretty definitely not "Western" or Byzantine.

Manuscript 0262

Berlin, Staatliche Museen, P. 13977. A small scrap of parchment, 7.5 cm. tall by 12 cm. wide, with two columns of text with six lines in each column. The writing is on the flesh side only; the hair side is blank. It has been suggested that it was part of an amulet. The text contains 1 Timothy 1:15-16, preceded by the mark of a cross. The surviving text is printed in G. H. R. Horsley, New Documents Illustrating Early Christianity, 2, 1982, p. 138. Dated paleographically to about the seventh century. Horsley says the orthography is "very poor," and the use of the nomina sacra inconsistent. I would consider "very poor" to be a kind description of the orthography; as an extreme example, the first words of 1:15 are [πισ]τος ω λλοκος instead of πιστος ο λογος. Also it has ομ for ων and ιμιν for ειμι; in verse 16, it spells αλλα as αλα, τουτο as τατο, and so forth. My strong suspicion is that the copyist was copying from dictation -- and very likely was not entirely fluent in Greek. The Alands list it as Category III, but NA28 cites it for only one reading, in 1 Timothy 1:16. The manuscript in that verse reads ΧϹ[..]ΤΗΝ, which NA28 interprets as Χριστος Ιησους την, the reading of A D* H Ψ 33 81 104 326 365 629 1175, as opposed to the Ιησους Χριστος την of ℵ D2 K L P 1505 Byz or Ιησους την of F G 1022 1739 1881. But given 0262's inconsistent use of the nomina sacra, this reading must be regarded as uncertain (and indeed NA28 marks it as vid). There is one other unique reading in 1:15, where 0262 has αποτοχης for αποδοχης, but this might be another mis-hearing.
This really isn't much to go on; despite the Alands' categorization, we can't say much about the textual affiliations of 0262.

Manuscript 0263

Berlin, Staatliche Museen, P. 14045. Tiny fragment containing a few letters of Mark 5:26-27, 31 (based on the full text printed in G. H. R. Horsley, New Documents Illustrating Early Christianity, 2, 1982, p. 127, there are just 40 extant letters, and a few of which are uncertain). It is part of a single leaf, probably of one column with about 17 or 18 lines per page. Dated to the sixth century. The Alands do not even attempt to classify it. In Mark 5:26, it has the reading εις το [χειρον], as in ℵ A B C L W Δ 33 892 Byz, as against επι το... in D Θ 565, and in 5:31 it has [συνλιθ]βοντα with all major witnesses as against the συνπνιγοντα of 565(=Luke 8:42), but these readings tell us little except that the text is probably not "Western" or "Cæsarean;" we cannot tell if it is Alexandrian, Byzantine, or something else.

Manuscript 0264

Berlin, Staatliche Museen, P. 14049. Fragment containing John 8:19-20, 23-24 (based on the full text printed in G. H. R. Horsley, New Documents Illustrating Early Christianity, 2, 1982, p. 133). A tiny scrap of a single sheet of what appears to have been quite a small codex; seven lines survive of what are estimated to have been eighteen lines per sheet. The surviving text comes from the inner edge of the leaf. The lines appear to have averaged about twenty letters (although there is a lot of variation), but in no case do as many as ten letters survive on an individual line. I count 88 surviving letters, some of them not fully legible. It is dated paleographically to the fifth century. In 8:24, it has ουν, which is omitted by 𝔓66 ℵ 543 a e but found in B etc. and it appears to agree with B etc. in omitting μοι, which is added by ℵ D Θ f13 e. There is also a peculiar reading in verse 23, which Horsley says is an accidental omission of υμεις εκ τουτου του κοσμου, but there are variations in word order here; I do not think we can form a firm conclusion as to its reading. Horsley concludes that it "clearly sides with the Alexandrian [text]," but the Alands describe it as too short to place in any Category, and in this instance, I think they are right.

Manuscript 0265

Berlin, Staatliche Museen, P. 16994. Tiny fragment containing a few letters of Luke 7:20-21, 34-35 (based on the full text printed in G. H. R. Horsley, New Documents Illustrating Early Christianity, 2, 1982, p. 127, there are just 53 extant letters, and many even of those are uncertain). It is the bottom of a single leaf, probably in two columns with about 24 lines per page; the surviving text is from the bottom five lines. Dated to the sixth century. The Alands tentatively list it as Category V but offer no statistics. It is extant for only one reading cited in NA28, in Luke 7:21, where it has [αuτη δ]ε τη ωρα -- which, if the reconstruction is right, agrees with A D K Γ Δ Θ Byz against the Alexandrian reading εκεινη τα ωρα of 𝔓75 ℵ B L W 892. (There is also some text of a variant in Luke 7:35, but not enough to be sure of the original reading). On this basis it has been said that the manuscript is not Byzantine, but of course it's just one reading. We really don't have enough information to say anything useful about it.

Manuscript 0266

Berlin, Staatliche Museen, P. 17034. Very long, narrow fragment containing bits of Luke 20:19-25, 30-39. Dated paleographically to the sixth century. There are two fragments which contain fragments of twenty lines, out of an estimated 33-34 lines per page. The surviving scraps are from the upper part of the leaf. The reconstruction by Treu (reprinted in G. H. R. Horsley, New Documents Illustrating Early Christianity, 2, 1982, p. 129) shows about 30 letters per line, but on many lines only three or four letters survive, making it very hard to know just where the line breaks actually fell. The Alands list it as Category III but offer no statistics. In fact it testifies to the following variants

This is a small sample, and many of the readings are very uncertain, but it is clear that 0266 is not Byzantine, and it does not agree at all with D. We note that, apart from some probably singular readings, the only place where 0266 disagrees with B is in 20:24, ad even there, it is close to ℵ. On the evidence, the Alands' decision to place it in Category III is too low; it should surely be Category II or even Category I. However, given the limited amount of text and the high degree of uncertainty of the readings, it must be admitted that 0266 is of very little use to us.

Manuscript 0268

Berlin, Staatliche Museen, P. 6790. Tiny fragment containing a few letters of John 1:30-33. The full text is printed in G. H. R. Horsley, New Documents Illustrating Early Christianity, 2, 1982, pp. 131-132. This fragment contains parts of 17 lines (seven of them so damaged that no text at all can be reliably read; the Alands therefore list it as having only ten lines). It appears the text originally averaged 11-12 letters per line, but I count only 59 surviving letters, many of them only partially legible; some lines are completely obliterated, and none appear to have more than eight surviving letters. Treu, who edited it, dates it to the sixth or seventh century; the Alands say the seventh. It has been suggested that it was part of a miniature copy of the Gospel of John. The surviving text contains an illustration of an embellished cross; it has been suggested that this is some sort of place marker, but I wonder if it might not be some sort of illumination. The Alands freely admit that there is too little text to classify. The only readings to which it appears to offer any testimony at all is are:

Clearly there isn't much to be learned from this fragment.