THE FOUR GOSPELS - A Study of Origins, The Manuscript Tradition, Sources, Authorship, & Dates. By B.H. Streeter, Hon.D.D.Edin. Fellow of Queen's College, Oxford; Canon of Hereford. First published by MacMillan & Co Limited 1924. - Prepared for katapi by Paul Ingram 2004.

Chapter IV


Home | List of MSS | synopsis | the Θ family | table- illustrating Byz. Correction in fam. Θ | relation to other ancient texts | Θ & the text of Origen | the MSS. Sent to Constantine | concluding survey | Table- The MSS. & the Local Texts | (pages 77-108)


the Θ family

The new Koridethi MS. Θ has been shown by K. Lake to be the most important member of a family of MSS. of which the most important are the cursives 1 &c., 13 &c., 28, 585, 700. Accordingly the whole group may appropriately be styled fam. Θ. Each member of this family has been partially corrected to the Byzantine standard; but, since in each a different set of passages has been so corrected, we can, by the simple expedient of ignoring the Byzantine readings, approximately restore the text of the original ancestor. This illustrated by a Table. In an Appendix evidence is adduced for assigning to fam. Θ certain other less important MSS., in particular the group 1424 &c.

relation to other ancient texts

  1. The text of fam. Θ is slightly, but only slightly, nearer to the Western than to the Alexandrian type; also it has a large and clearly defined set of readings peculiar to itself.
  2. In fam. Θ are found certain striking additions to the T.R. which the Syriac shares with D and the Old Latin, beside others found only in the Syriac or Armenian.
  3. As regards, however, the longer omissions from the T.R. found in B and Syr. S., fam. Θ nearly always supports the shorter text.
  4. Fam. Θ is nearer to the Old Syriac than is any other surviving Greek text, but it is by no means identical; it is frequently supported by the Armenian against the Syriac. Most frequently of all it is supported by the oldest MSS. of the Georgian version.

Θ and the text of Origen

Griesbach discovered that Origen used two different texts of Mark; but, owing to the paucity of MS. evidence then available, he slightly misinterpreted the facts. These are as follows. In the surviving portions of the first ten books of his Commentary on John, Origen used the B א text of Mark; but in the later books of this work, in his Commentary on Matthew and his Exhortation to Martyrdom, he used a text practically identical with that of fam. Θ. The Commentary on John was begun in Alexandria but finished at Caesarea, and both the other works mentioned were written at Caesarea.

It further appears that the text of Matthew used by Origen in his Commentary on that Gospel was the fam. Θ text—a fact partly disguised in the printed editions in which the text of fam. Θ has been sporadically corrected to the Byzantine standard. Throughout the Commentary on John, Origen used an Alexandrian text of John, but in the later books he changed his text for one of the Θ type. These conclusions tested against tables drawn up by Preuschen. At a later date Origen seems to have used the Θ text for John also.

Reasons for believing that the fam. Θ text was already in possession at Caesarea when Origen arrived and was not a recension which he made himself.

the MSS. sent to constantine

The possibility that the fifty copies supplied by Eusebius to Constantine in 331 represented the old text of Caesarea. By 380 Constantinople had adopted the revised text of Lucian. This would lead to the correction of the older MSS. to the Lucianic (i.e., practically, to the Byzantine) standard. Some of these partially corrected copies would get into the provinces, and may be the parents of some existing MSS. of fam. Θ. Possibility that the Greek texts used by SS. Mesrop and Sahak to revise the Armenian were of this character.

concluding survey

Significant fact that the local texts identified above form a series corresponding to the geographical propinquity of the churches with which they are connected.

Practical bearing of these results. The textual critic, in weighing the amount of external evidence in favour of any reading, should consider primarily, not the number or age of the MSS. which support it, but the number and geographical distribution of the ancient local texts in which it can be traced.

It follows that MSS. should be cited, not in alphabetical or numerical order, but in groups corresponding to the local texts, which they represent. When at least three of the leading representatives of any local text support a reading, very little is gained by citing the additional evidence of MSS. that normally support the same local text.



the Θ family

THE UNCIAL MS. to which the letter B is assigned was discovered in a remote valley in the Caucasus, where it had long been a kind of village fetish; but at a much earlier date it belonged to a monastery at Koridethi at the far end of the Black Sea just inside the old frontier between Russia and Turkey. Owing to a chapter of accidents including a disappearance for thirty years its complete text only became available to scholars in 1913.
[In the edition by G. Beerman and C. R. Gregory, Leipzig, 1913. The student should be warned that the Appendix that gives the MS. support of all variants in Θ is quite unreliable so far as its cursive supporters are concerned. As the MSS. most closely allied to Θ are all cursives, this is a serious defect. The Moscow Archeological Society, 1907; published an edition of Θ, in reduced facsimile, with Mark but this is not easily procured.]
Dr. R. P. Blake, in a joint article by himself and Prof. K. Lake in the Harvard Theological Review for July 1923, argues that the scribe was a Georgian, familiar with the Coptic script, but extremely ignorant of Greek. At any rate the ordinary tests by which the handwriting of MSS. can be dated are difficult to apply; but it probably belongs to the eighth century.

The discovery is comparable in importance to that of א or the Sinaitic Syriac—but for a different reason. The importance of א and Syr. S. depends on their early date and the relative purity of the types of text they respectively preserve. Θ is neither so old nor so pure: it has suffered considerably from Byzantine revision. Its importance lies in the fact that it supplies a missing link and enables us to see the real connection between certain cursives, the exceptional character of which has long been an enigma to the critic. In the demonstration of the relation between Θ and this group of cursives, the first and most important step was made by Lake in the brilliant article referred to above in the Harvard Theological Review.

The cursives in question are the following:

  1. Codex 1 and its allies, commonly cited as fam. 1, or 1 &c. Of this family of MSS. the only one comparable in importance to 1 is 1582 (Xcent.) recently discovered in the Vatopedi Monastery on Mt. Athos. But the inferior members occasionally preserve original readings, which have been revised out, in the two better MSS.
    [Codex 1 of the Gospels and its Allies, by K. Lake, Texts and Studies, vol. vii. (Cambridge, 1902), contains the full text of 1 collated with its inferior supporters 118—131—209, along with a very valuable Introduction. No collation of 1582 has yet been published, but it is quoted by Soden (as ε183). Soden also quotes from two others of much less importance, i.e. for Mark, Luke, John 2193 (Sod. e 1131), for Mark only 872 (Sod. ε203). Soden also includes 22 and 1278 in this family; the case of 22 is discussed by H. A. Sanders in "A New Collation of Codex 22" (Journal of Biblical Studies, xxi., pt. 2) who in general agrees. As nearly all the readings of 22 not found in 1—118—131—209 occur in other members of fam. Θ, it matters little whether it is classed with fam. 1 or as an independent member of the larger family.]
  2. The "Ferrar group" (cited as fam. 13 or 13 &c.), extended by later discovery from the four MSS. 13—69—124—346, edited by Ferrar and Abbott, to twelve, all of which are probably derived from a single lost uncial. Within this group 69, 124, and 983 are specially important as often preserving readings not found in other members.
    [A large literature has arisen round the Ferrar group (cf. Further Researches into the History of the Ferrar Group, pp. 1-8, by J. Rendel Harris, Cambridge Press, 1900). It would appear that most of the group 13—69— 124—230—346—643—788—826—828—983—1689—1709 are descended from a MS. which in the twelfth century was preserved either in some monastery in Calabria in the "heel" of Italy, or in some allied monastery in Sicily. In the classical period, S. Italy was not Italian but Greek; but by the end of the sixth century, apart from a few coast towns, it had become Latin. But in the eighth and following centuries there was an immense immigration of Greek-speaking monks refugees from the Mohammedan invasions. In the twelfth century, under Norman rule, there was an intellectual revival in the Greek monasteries of S. Italy. There is excellent evidence (cf. K. Lake, J.T.S., Jan. 1904, p. 189 ff.) that MSS. were collected at considerable expense from different parts of Greece and from Constantinople to found, or refound, libraries. The magnificent purple Codex 2, still preserved at Rossano, must have been written either in Constantinople or in Cappadocia. Accordingly, it is probable that the ancestor of the Ferrar group was brought to Italy from the East; there is no reason for connecting it with the primitive text of S. Italy, which in all probability was akin to D.]
  3. The Paris MS. 28.
  4. The "Empress Theodora's Codex" 565 (cited by Tischendorf as 2pe and by Hort as 81). [Edited by J. Belsheim, Christiania, 1885; corrections by H. S. Cronin in an Appendix to his edition of א, Texts and Studies, vol. iv. p. 106 ff. Scrivener and Hoskier cite as 473.] For Mark this is the most, for the other Gospels the least, important of the MSS. here mentioned.
    (e) The very interesting British Museum MS. 700, acquired in 1882 but not fully made known to the world till 1890. [Collation by H. C. Hoskier, as Codex 604 (Scrivener's number); with Appendix containing a collation of 1278, which Soden reckons a weak member of fam. 1 (D. Nutt, 1890).]

Lake made the all-important discovery that Θ and these notable cursives, taken, all together, form in reality a single family. True Θ and the five other sets of authorities mentioned do not on the face of it exhibit a single type of text; but that is because each of them has been heavily corrected to the Byzantine standard, and in each case a different set of corrections has been made. If, however, we eliminate from the text of all these manuscripts those variants that are found in the Byzantine text, we find that the residuary readings of the six different representatives of the family support one another to a quite remarkable extent. Lake illustrates this by a table analysing the variants in the first chapter of Mark.

In order to indicate the nature of his argument and at the same time to test its validity in regard to Luke and John, I have compiled similar tables (p. 83 and App. II.), only with an additional column for the readings of fam. 1424. On the left are printed the readings found in one or more MSS. of the family that differ from the Textus Receptus; on the right are the corresponding readings of the T.R. The letter f stands wherever the MS. (or group) indicated at the head of the column supports the family reading, the symbol ς when it agrees with the T.R. If any MS. supports a third reading, this is indicated in the column appropriate to that MS. by the symbol "3rd." The readings of א B and D are also given in order to show how each of them alternately supports and deserts the fam. Θ text.

From Lake's table of variants in Mk. i. it appears that there are 76 instances in which at least two members of the family agree with one another in exhibiting readings not found in the Byzantine text; while there are only 5 instances where a member of the family gives a non-Byzantine reading other than that supported by the family. The significance of these figures is made clearer when it is noted that in regard to this same set of 76 variants in Mk. i. א and B differ from one another no less than 12 times. It follows that the ancestors from which Θ and the five sets of allies were derived must have differed from one another in this chapter considerably less than א does from B. Clearly we are justified henceforth in referring to this group of MSS. by the convenient title of fam. Θ.
[When a reading is cited as occurring in fam. 13 or the like, this does not mean that it is found in all MSS. of that group, but that it occurs in at least two, and that practically all MSS. of the group which do not give it follow the Byzantine text instead.]

In the article in the Harvard Theological Review the authors confined their discussion to the text of Mark the Gospel in which, as we have seen before, the key to the history of the text of any particular MS. is usually to be found. But as I happened to have been exercising myself with the problem presented by the text of Θ, I could not rest until I had explored their solution a little further. The evidence that convinced me that Lake's conclusion holds good in regard to the other Gospels also is presented in Appendix II.

In the course of this investigation I came upon evidence that the family of which Θ is the head has numerous poor relations. That is to say, there are a large number of MSS. which appear to be ultimately descended from ancestors the same or similar to those of fam. Θ, but by lines of descent which have suffered far more correction to the Byzantine standard. For details I refer to Appendix II. Such MSS. are of interest in that they occasionally preserve apparently genuine readings of the family text, which have been revised out of the (generally speaking) better representatives. Of these MSS. the most important is the group which von Soden styles IΦ, but which by parity of nomenclature I propose to cite as fam. 1424, since the Xcent. Kosinitza MS. 1424 (Scrivener's כ) is its oldest representative.

relation to other ancient texts

But before attempting to inquire further into the origin of the text represented in fam. Θ, we must clear up its relation to other ancient texts, especially to those of B, D and the Old Syriac. This is the more necessary as von Soden has mis­represented and confused the evidence, by putting D into the same sub-family as Θ, and by making the Old Syriac another witness to the same type of text.

My investigation of this question leads me to formulate four main conclusions:

  1. So far as minor variants are concerned and these are much the most numerous, and are of course the most significant for the study of the relationship of different texts the text of fam. Θ is almost equidistant from both the Alexandrian and the Western texts. The balance inclines slightly, but only slightly, to the Western side, while there are a very large proportion of readings found neither in D nor in the typical Alexandrian MSS. We have therefore in fam. Θ a clearly defined and distinctive text which may properly be ranked side by side with the three great texts, Alexandrian, Western and Byzantine (=Hort's "Neutral," "Western" and " Syrian ") hitherto recognised.
  2. In fam. Θ are found certain striking additions to the T.R. which the Syriac shares with D and the Old Lat., besides others found only in the Syriac or the Armenian.
  3. On the other hand, as regards the longer omissions from the T.R. which are so conspicuous a feature of the conjunction of B with Syr. S., fam. Θ nearly always supports the shorter text.
  4. Though the text of fam. Θ is nearer than any other surviving Greek text to the Old Syriac, it is by no means its exact equivalent; and it frequently goes with the Armenian against the Syriac. Further, it would appear that it is supported most frequently of all by the oldest MSS. of the Georgian version.

I proceed to summarise the evidence on which these conclusions are based. But the reader who has not previously made a study of textual criticism is advised on a first reading to skip this and pass on to the next subsection, "Θ and the Text of Origen."

  1. Lake's table shows that in Mk. i., in cases where B and D differ, B supports fam. Θ against D 16 times, while D supports the family against B 15 times, also that in 9 cases fam. Θ is supported against B D combined by one or more of the later Alexandrian group K L A W 33 579. That is to say, the text of fam. Θ, in this chapter of Mark, is somewhat more closely allied to that of Alexandria than it is to D and the Old Latin. But how far, we ask, is this proportion maintained throughout the four Gospels? To make a count of all the readings in all four Gospels is obviously impossible; but in four different ways I have been able to compile statistics that give some indication of the proportion that prevails elsewhere between the number of Egyptian and D readings.
    (a) Hoskier in his edition of 700 (p. ix) sets out all the agreements of that MS. with the great uncials against the Byzantine text. From these it appears that 700 is supported by B against D 63 times, by one or more members of the group א L C Δ against B D combined 34 times, while it joins D against B 111times,  
    (b) In the Introduction to Ferrar and Abbott's historic edition of 13—69—124—346 (p. xlviii) will be found an analysis of the variants in Mt. xix.-xx. and Mk. i.-ix.    Only those variants are counted in which the four cursives agree against the T.R. Out of 25 variants in Mt. xix.-xx., 13 &c. agree 18 times with B, 17 with D.    Out of 215 in Mk. i.-ix. they agree 88 times with B, 90 with D.    Further, it appears that in a series of selected passages from all four Gospels fam. 13 differs 376 times from א, 367 from B, 496 from D.   That is to say, while 700 is slightly nearer to D than to the Egyptian group, the Ferrar group is distinctly nearer to B א than to D.   
    (c) The statistics given below (p. 90), compiled from the lists in Lake's Codex 1 and its Allies, show the numbers of agreements of fam. 1 with the principal authorities in turn, and show that fam. 1 is only a very little nearer to B א than it is to the Old Lat. and D.    
    (d) For Θ there are no such statistics to refer to, but a study of the MS. support for variants in Mk. xiv. and xv. as set out in the Appendix of Gregory and Beerman's edition of Θ shows that for these two chapters the proportion of Alexandrian to Western readings is approximately as 3 to 4.   All these several sets of statistics, it will be observed, come to much the same thing.   It so happens that in fam. 1 and fam. 13 the Byzantine revisers have spared a slightly larger proportion of Alexandrian than Western read­ings, while  in  Θ and 700 the opposite has  occurred;  but, considered as a whole, the text of fam. Θ is not very much nearer to D than it is to B.    Thus the von Soden grouping, which puts D in the same group as Θ, 28, 565, 700, while excluding from that group fam. 1 and fam. 13, is a complete mis­apprehension of the evidence.
  2. More interesting, if not more important, is the relation of fam. Θ to the Syriac and the Armenian versions. This may be illustrated by selecting a few striking readings in which fam. Θ agrees with Syr. S., and usually Arm. also, against B.
     Ἰακὼβ δὲ ἐγέννησεν Ἰωσὴφ τὸν ἄνδρα Μαρίας, ἐξ ἧς ἐγεννήθη Ἰησοῦς is the ordinary reading. Instead of this Ἰακὼβ δὲ ἐγέννησεν, μνηστευθεῖσα παρθένος Μαριὰμ ἐγέννησεν Ἰησοῦν τὸν λεγόμενον Χριστόν is found in Θ and the Ferrar MSS. 346 — 543 — 826 — 828 (hiat. 69), Old Lat. (incl. d. hiat. DGk.). Syr. C. agrees with this, approximately. The Armenian combines both readings — a sure sign that it is a mixed text — and reads "the husband of Mariam, to whom was betrothed Mariam the virgin, from whom was born Jesus." Syr. S. has a reading, which would correspond to Ἰωσήφ μνηστευθεῖσα παρθένος Μαριὰμ ἐγέννησεν Ἰησοῦν κτλ.  To me the reading of Syr. S. looks as if it was translated from a Greek MS. of the Θ 13 &c. type in which by accident the name Ἰωσήφ had been written twice. Dittography is a very common scribal error; and seeing that every one of the preceding 39 names in the genealogy had been written twice, the repetition of this particular word would have been exceptionally easy. The reading of Syr. C. will then be explained as one among many other attempts to correct this MS. by a MS. of the D type.

Mt.xxvii.16, 17.

The name of Barabbas is Jesus Barabbas,

Θ, 1 &c., Syr. S., Arm., Orig. in Mat.


After γῆς add καθὼς πέστειλεν με πατήρ, κἀγὼ ποστέλλω μᾶς

Θ, 1604, Syr. Pesh. (hiant Syr. S. and C.), Arm. (hiat. Orig.Mt.).


Before εἶπεν add πιτιμήσας

Θ, 1 &c., 13 &c., 28, 565, Syr. S., Arm.


om. ἡ ἀδελφὴ τοῦ τετελευκηκότος 

Θ, Syr. S. Arm.; Old Lat.


For Γαββαθᾶ =pavement κατφαθᾶ =arch

1 &c., 565, Arm.codd. Syr. S. and C. are both lacking; but Syr. Pesh. does not favour either Γαββαθᾶ or the reading of 1 &c. Θ has (χιφβαθα).


After διδάσκαλε add καὶ προσέδραμεν ψασθαι αὐτοῦ

Θ, 13 &c., Syr. S.; Old Lat.

Note, however, that fam. Θ gives no support to the Syriac in certain other conspicuous additions, e.g. in Lk.xi.48, Jn.iii.6, Jn.xi.39, Jn..12.   Further we note that the Armenian also deserts the Syriac here.

(3) It would appear that fam. Θ agrees with Syr. S. in a number of notable omissions wherein Syr. S. has the support of B.


"Signs of the times,"

om. 13 &c., Arm., Orig.Mt.


"This kind goeth not forth," &c.,

om.  Θ, 1604 (Arm., Orig.Mt. habent); e.


"For the son of man came," &c.,

om.  Θ, 1 &c., 13 &c., Orig.Mt. (Arm. habet).


whole verse

om. Θ, 1 &c., 28, Arm., D a e, Orig.Mt.

Mk.ix.44, 46,

"Where the worm dieth not" (1st and 2nd time),

om. 1, 28, 565, Arm.; k.


"And every sacrifice shall be salted with salt,"

om. 1, 565, 700, Arm.; k.



That this was originally absent from fam. Θ may be inferred from the scholion to ἐφοβοῦντο γάρ, Mk.xvi.8, in certain members of the family. In the newly discovered Vatopedi MS. 1582—the oldest MS. of fam. 1—there is a concluding ornamentation after ἐφοβοῦντο γάρ, Mk.xvi.8, followed by a scholion: [Cf. Gregory, Textkritik, iii. p. 1160.]  "In some copies the Gospel ends here, up to which point also Eusebius Pamphili made his canons, but in many (copies) there is also found this." Then follows xvi.9-16. An identical scholion occurs in 1, in the margin; but Dr. Blake informs me that in 1582, which he has photographed, this note is written right across the page in uncial letters as a colophon. In 22 the word τέλος is written after ἐφοβοῦντο γάρ and the same scholion, only with the allusion to Eusebius omitted, follows. In three of the oldest Armenian MSS. the Gospel ends at this point. So also does the oldest (Adysh) MS. of the Georgian version.


"Ye know not of what spirit ye are," &c.,

om. 28, 1424 &c. (Arm. hab.).


The angel and the Bloody Sweat,

om. N 1071. In fam. 13 it is omitted here but inserted after Mt.xxvi.39, where it occurs in Greek Lectionaries as a Good Friday Lesson. This is explicable only if it was originally absent from the text of the family in Luke, and was inserted in Matthew by a scribe who supposed the Lectionary to represent the true reading of that Gospel. Some MSS. of Arm. omit.


"Father, forgive them,"

om. Θ (Arm. hab.); D a b.


Pericope Adulterae,

om. Θ, 22, 2193, 565, 1424 &c., Arm.; a f q ; in 1 and 1582 at the end of the Gospel—with a note that it is found in some copies but not commented upon by the holy Fathers Chrysostom, Cyril Alex., and Theodore of Mopsuestia; inserted by 13 &c. after Lk.xxi.38. It is absent from all old Georgian MSS., having been introduced by George the Athonite in his revision, c. 1045.

In view of this concurrence between B, Syr. S. and fam. Θ, in the omission of conspicuous passages, three points require notice,

  1. There is no evidence that fam. Θ omitted Lk. xi. 38, "in letters of Greek and Latin and Hebrew," with B, Syr. S. and C. (Arm. hab.); or Jn. v. 4 (the moving of the waters) with B, Syr. C. (Mat. Syr. S.), Arm.codd. Though, of course, the words may have been inserted in all MSS. of the family by Byzantine revisers,
  2. Fam. Θ agrees with Syr. S. in certain conspicuous insertions, which are found also in D. By reference to any good Apparatus Criticus the student may verify this under the references Mt. v. 22, Mt. x. 23, Mt. xxv. 1, Mk. x. 24.
  3. Fam. Θ seems to support B against both Syr. S. and D in omissions in Mt. iv. 10, Mt. xx. 16, Lk. xx. 34.
  1. t is clear that the Greek text from which the Old Syriac was translated is more closely related to that of fam. Θ than to any other extant Greek MSS.; but it would be a great mistake to suppose that it is in any sense the same text. Indeed a notable feature of the fam. Θ text is the number of its agreements with B against the Syriac. It is also noteworthy that the fam. Θ is frequently supported by the Armenian against the Old Syriac. The lists of readings in Lake's edition of Codex 1 provide materials on which a rough estimate may be based.
    From these lists I have compiled the following statistics:

Variants quoted in which fam. 1 differs from T.R.


Of these, number peculiar to fam. 1




Readings of fam. 1 found in Syr. S. or C. but not in Arm.


Readings of fam. 1 found in Syr. S. or C. supported by Arm.




Readings of fam. 1 found in Arm. but not in Old Syr.


Readings of fam. 1 found in D or Old Lat. but not in א, B or L.


Readings of fam. 1 found in א or B but not in D or Old Lat.


In considering these statistics it should be remembered that many variants in the Greek cannot be represented in Syriac or Armenian, and therefore the proportion of agreements with these versions as contrasted with B, א D or L, etc., is necessarily understated. Nevertheless they show clearly
( α) that fam. 1 (which previous statistics have shown is a typical representation of Jam. &) does not by any means stand to the Old Syriac in the same relation as does D to the Old Latin.
(β) That its affinities with the Armenian are almost as numerous (95 as against 103) as those with the Old Syriac.

When this chapter was already in slip proof Dr. R. P. Blake, who is working on the text of the Georgian version, showed me a collation of Mk. i. in the Adysh MS. (dated AD 897) and in the recently discovered Xcent. Chanmeti fragments, which appear to represent an older form of that version than that reproduced in the printed editions. [Of the Adysh Gospels there is a photographic facsimile by E. S. Tagaiahoibi (Moscow, 1916). The two Chanmeti fragments are dated respectively AD 914 and 995; edited by V. N. Benesovie (Petropoli, 1908-9).] The MSS. frequently differ from one another; but the remarkable fact stands out that in the majority of cases in which one or more of these Old Georgian MSS. differs from the T.R., its reading is supported by fam. Θ. In Mk. i. in the Georgian there are altogether 83 variants from the T.R. Of these 28 are found nowhere else; and most of them look as if they were due to a translator's freedom. Of the remaining 55, no less than 38 occur in one or more of the seven main authorities from fam. Θ; and 5 others occur in MSS. classed by von Soden as minor supporters of the I text. [Old Georgian has the notable readings (p. 87) in Mt.xxvii.16-17; xxviii.18; Mk.x.14.   Lacking photographs of the MSS. those in John were not checked.] If, on further investigation, it should appear that this close relation between fam. Θ and the Georgian holds throughout all four Gospels, the Old Georgian version will become an authority of the first importance for the text of the Gospels; for it will enable us to check and supplement the evidence of Θ and its allies much as the Old Latin does for that of D.


Seeing that fam. Θ includes the main authorities for what von Soden calls the "I text," with the three all-important exceptions of D, the Old Latin and the Old Syriac, it seemed worthwhile to ask whether his theory that this text represents a recension by Pamphilus, the friend of Eusebius, would hold good, provided the authorities for it were restricted to fam Θ. I, therefore, turned to his discussion (vol. i. p. 1494) of the quotations of Eusebius of Caesarea, whom he regards as the leading patristic authority for the I text. It appeared, however, that, though a majority of the readings quoted from Eusebius are to be found in fam. Θ, a notable feature of the text of Eusebius is the number of passages in which he gives a reading found only in D. From this follows the important negative conclusion—fam. Θ does not represent the text of Eusebius. The facts, however, would be quite consistent with the hypothesis that the text of Eusebius had much the same relation to that of fam. Θ as א C L 33 bear to B, i.e. that the text of Eusebius represents a somewhat degenerate form of the text found in fam. Θ a degeneration largely due to mixture with a text of the D type.

At this point there flashed across my mind the distinction between the two texts used by Origen which was worked out as long ago as 1811 by Griesbach in his Commentarius Criticus [Part II. pp. x-xxxvi.] a book to which my attention had been called by Prof. C. H. Turner some months before. Griesbach's thesis was that Origen in his Commentary on John used an "Alexandrian" text of Mark for Mk. i.-xi., and a "mixed text" for the remainder of the Gospel, but that he used a "Western" text of Mark in his Commentary on Matthew and in his Exhortation to Martyrdom, both of which belong entirely to the period when he lived in Caesarea. It occurred to me to review the evidence submitted by Griesbach in the light of MSS. of the Gospels which have only been discovered or properly edited since his time.
The results were astonishing.

Two points became clear,

  1. The difference noticed by Griesbach between the use of an "Alexandrian" and of a "mixed" text of Mark corresponds to the change, not from the earlier to the later chapters of Mark, but from the earlier to the later books of the Commentary on John,
  2. Both this "mixed" text of Mark and the so-called "Western" text used in the Commentary on Matthew and in the Exhortation to Martyrdom are practically identical with the text of fam. Θ.

At once we notice the salient fact that the change in the text used corresponds, roughly speaking, to a change of residence. Origen himself tells us that the first five books of the Commentary on John were written before he left Alexandria for Caesarea, in 231. The Exhortation to Martyrdom was written shortly after the outbreak of the persecution of 235; the Commentary on Matthew (about 240) is probably one of the works taken down by shorthand from lectures delivered on weekdays in the church at Caesarea.

I proceed to submit statistics in support of the above conclusions.

  1. In books i.-x. of the Commentary on John, Origen quotes the greater part of Mk. i. 1-27 and the whole of Mk. xi. 1-12, besides a few odd verses. The number of variants in Mk. i. 1-27 cited by Griesbach is 36. For 2 of these there is no support in MSS. of the Gospels; but in one, and perhaps both, of these cases Origen seems to be paraphrasing rather than quoting his text. In the remaining 34 readings Origen is supported 31 times by one or both of the MSS. B א and once each by the other Alexandrian MSS. C, L, Δ; but in only17 of the 34 is he supported by fam. Θ.
    [Of these readings 16 occur in B or א and the remaining 1 in D; so there are none distinctive of fam. Θ. N.B. In two cases, where Griesbach, using the Benedictine text, cites variants of Origen which differ from B א, in Brooke's edition Origen's reading agrees with those MSS.]
    From the shorter passages quoted in books i.-x. (i.e.; x.18; xi.15-17; .26-27; xiv.60) Griesbach cites 16 variants. Origen is supported in 10 of these by B, and in 1 each by C, Δ; for 3 there is no MS. support, and 1 occurs in the T.R. The continuous passage Mk. xi. 1-12 is specially important, for it is so long that by no possibility can it be a quotation from memory; it must therefore represent the third-century MS. of the Gospel used by Origen. Apart from an accidental omission (I think in some ancestor of our copy of Origen on John)  the variants noted by Griesbach number 31; [In Mk. xi. 7-8 Origen omits καὶ κάθισεν ... ὅδον, but to make sense adds the last four words after ἀγρῶν. Burkitt (J.T.S. xvii. p. 151) thinks this was a defect in Origen's MS. of the Gospel. As, however, there is a similar omission (through homoioteleuton) in the quotation of Mt. xxi. 8 on the same page in Brooke's edition (i. p. 208), it seems to me more likely to be a defect in the MS. of the Commentary on John, an ancestor of which was prone to omission.] in 29 of these the reading of Origen is supported by one or both of the MSS. B א; in 1 by fam. Θ), and in 1 by the T.R., where these texts differ from B א. It may be of interest to note that in the passages examined above, where B and א differ, Origen has 6 agreements with B as against 7 with א.
  2. The number of variants in Mark cited by Griesbach from the later books of the Commentary on John is 43. For 5 of these the text of Origen has no MS. support; in 6 cases it agrees with the T.R. We have seen (p. 45 ff.) that when a quotation by an ante-Nicene Father agrees with the T.R. against earlier texts, there is always a possibility that this may be the result of later scribal alteration in the MS. of that Father; again, whenever a reading in a patristic quotation is not supported by a single MS. of the Gospels, there is a presumption, either that the author is quoting from memory or paraphrasing, or that it is an error in the MSS. of his work. In view of these considerations it is highly significant that of the remaining 32 variants no less than 30 are found in fam. Θ (10 occurring only in the MSS. of this family), while Origen is supported only once each by א and D, and never by B, against the family.
  3. From Origen's Commentary on Matthew, Griesbach cites 99 variants in Mark; of these 8 are peculiar to Origen, and 13 occur in the Byzantine text. [The suspicion that some, if not all, the 13 Byzantine variants do not represent what Origen actually wrote, is partially justified by the fact that one of them (Mk. xiv. 62) is quoted in Origen's Commentary on John (bk. xviii.) according to the reading of fam. Θ. Of these 13 variants 2, though found in the great majority of Byzantine MSS., do not occur in the printed T.R., which has, besides some very late readings, a few derived from Codex 1. Since this MS. was used by Erasmus, the T.R. occasionally supports fam. Θ against the Byzantine text. For the purpose of the above calculations, the only MS., except those mentioned in the table on p. 83, which I have reckoned as evidence for fam. Θ is 644.] Of the remaining 78 as many as 74 are found in one or more members of fam. θ; while Origen has one single agreement with each of the four MSS. B, C, Δ, D, where these differ from fam. Θ.
  4. The figures in regard to the passages of Mark quoted in the Exhortation to Martyrdom are, if anything, more striking. Of the 15 variants instanced by Griesbach, 11 occur in fam. Θ; 1 goes with the T.R. against fam. Θ; 3 are unsupported by any MS., but of these 1 is practically the reading of fam. Θ, and the other 2 spoil the sense and are obviously errors in the MS. (or printer) of Origen. Figures like these amount to demonstration. The text of Mark which Origen used at the time he wrote these works was that of fam. Θ.

The next step was to test the character of the text of Matthew that was used by Origen. I recollected that books on textual criticism commonly speak of the reading "Jesus Barabbas" in Mt.xxvii.17 as found "in MSS. known to Origen," as if this characteristic reading of fam. Θ was one which, though known to Origen, did not occur in the text he ordinarily used. But on turning to the passage in the Commentary on Matthew  I found to my surprise that this reading occurs in the text recited and commented on by Origen. It is the omission of the name Jesus before Barabbas that should properly be described as a reading "found in MSS. known to Origen." Origen dislikes the reading of the text he is using, and suggests that the name Jesus may be an heretical interpolation; but it is in his text. He informs his readers that it is absent from MSS. known to him, but, presumably, not equally well known to them.
[The mystical interpretation, verum mysterium, of the contrast of the two prisoners Christ and Barabbas which he proceeds to develop has much more point if regarded as his way of making the best of a text which gave to both the name Jesus; though the meaning is slightly obscured by the fact that it is introduced by the word enim, which I cannot help thinking stands for γοῦν, misread or misrendered as if it were γάρ.]

An investigation of several sections in this Commentary (chosen for the exceptional length of the quotations they included) revealed the facts set out in Appendix III. Briefly, the majority of readings in Origen are found in one or more members of fam. Θ; but a minority are not. Further examination, however, showed that, where the text of Origen deserts that of fam. Θ, it is almost always in order to agree with the Byzantine text. In Mt. x. and xxv., which were selected for minute study, Origen's quotations differ from the T.R. in 45 variants. In 37 of these, his reading is supported by one or more members of fam. Θ.  Clearly we must make a choice. Either Origen used a text which in the main was that of fam. Θ, but occasionally went over to the side of the Byzantine text, or the Gospel quotations in the MSS. from which is derived the printed text of Origen have been to a slight extent assimilated to the Byzantine standard. This is obviously the more probable alternative, and affords one more example of that assimilation of biblical quotations to the standard text, which is one of the principal causes of corruption in the text of the Fathers (cf. p. 45 ff.). That this assimilation has affected the MS. tradition of Origen's quotations from Matthew more than those from Mark is only what we should expect; for precisely the same distinction is found in the textual traditions of the Gospels themselves.

The evidence given above as to the assimilation to the Byzantine text of the quotations of Origen in the Commentary on Matthew compels us to discount the appearance in other works of Origen of occasional readings of the Byzantine type. In particular we can disregard the Byzantine readings that occur here and there in the Gospel quotations in the Commentary on John—more especially as that work depends upon a single MS. of the tenth century.    Bearing this in mind I proceeded to test the quotations in the Commentary on John of Gospels other than Mark, selecting for the purpose a number of the longer, and therefore presumably more representative, passages. The tests, though by no means exhaustive, all pointed in one direction.    Origen, so long as he was at work on the Commentary on John, continued to use his Alexandrian MS. for John (and in the main, I think, for Matthew); and where א B differ, Origen's MS. of John more often agreed with B than with א.    But at some point or other he seems to have changed his MS. of Luke, as well as that of Mark, for one of the type of fam. Θ.   Incidentally, we may infer that for some time after he reached Caesarea Origen read the Gospels, not in a Four-Gospel Codex, but on separate rolls. After reaching these results, it occurred to me to check them by the discussion on "the Bible-text of Origen" by E. Preuschen in the Berlin edition of the Commentary on John, 1903.   Preuschen shows conclusively that Origen frequently quotes from memory, conflating, for example, the Matthean and Lukan versions of the Parable of the Supper.    From this it follows that we cannot indiscriminately take all his quotations as evidence of the text he used;  we must be careful to use only passages where it is evident from the context that he is commenting on a MS. open before him. But Preuschen goes on to argue that, even where it is clear that Origen is using a written copy, the text from which he quotes does not correspond at all closely with that found in any extant family of MSS. To prove this point Preuschen (p. xciv) selects three passages (all from torn, xix.), and gives the variants with the MS. evidence for each. The central column in the tables given below reproduces his statement of the facts. The right-hand column is my own addition, and gives the MS. evidence (much of which, of course, was not available when he wrote) for the readings of fam. Θ. It will be seen at once that this fuller statement of the evidence points to a conclusion very different from that which Preuschen draws.

origen. support quoted by preuschen. support from fam. Θ.

Mk. .


v.41. καὶ ἐστὼς

1, 69, Syr. Sin.,

Θ, 1, 13, 28, 69, 565.

(for  καθίσας)

Hclmg., Arm.


κατέναντι (T.R.)

א A D L al

(T.R.) Θ, 1, 69, 124, 565.

(ἀπέναντι B 33 579)


κατενώτιον 13 &c.


ἀπέναντι 1424 &c., U, 544.

Καὶ πᾶς... (for  πῶς)

Solus *


ἔβαλλεν... (for  βάλλει)


13, 691 (ἔβαλε), 124.

v.42.  ἐλθοῦσα δὲ

D, Latt., Boh. Sah.

Θ, 565, 700.

(καὶ ...)


V.43.  εἶπεν… (for  λέγει)

B א A D L, Δ, 33

K U a. k. verss.

Θ, 565, 700.

ἡ πτωχὴ αὔτη ...


ἔβαλεν ... (for  βέβληκεν)

D, a, b, ff, g2, i

B A D L, Δ, 33

Θ, 565, 700.

Θ, 565.

* πῶς mis-spelt πᾶς: καὶ added to restore grammar.

Luke xxi.1-4

origen's text. support as quotes by preuschen. support from fam. Θ.

Lk. xxi.


v.1. εἰς ... εἰς ... (for  εἶδε ... εἰς) T.R.

Syr. Cur.

13 and 124 read εἶδε.

Was εἶδε a marginal   note, cor­recting the first

εἰς  to εἶδε , which has been applied to both occurrences of εἰς ?


v.1.   εἰς  τὸ γαζοφυλάκιον τὰ δῶρα αὐτῶν πλουσίους (order)

B א D L X 1, 33, 69. e, Syr. Pesch.

1, 69.

v. 2. om. δὲ ...

om. (καὶ before τινα.

(T.R.  καὶ τινα)

S, a, Boh., Arm.

B א L Q X, 33.

K M,  Γ, c, ff, I, Syr. Hcl.text, Aeth.


(Θ, 1, 700 τινὰ καὶ)

Om.  ἐκεῖ

D, Syr. Sin. Cur. Pesch. Aeth.

2193 (of fam. 1).

λεπτὰ δύο ... (order)

B א Q L X, 33.

Vulg. Syr. Sin. Cur. Pesch. Boh.


v.3.  ἡ πτωχὴ αὕτη (T.R.)

(for αὕτη ἡ πτωχὴ  א B D, 13 &c.)

A X, 1, E G H al. a, Syr. Hcl.

(T.R.) Θ, 1, 565, 700.

v.4.  πάντες ... (for ἅπαντες)

τὰ δῶρα τοῦ θεοῦ (T.R.)

(om. Τοῦ θεοῦ  B  1 &c.)

πάντα ... (for ἅπαντα)

B א D Δ

A D Q? E G H
al. Latt. Verss.

B א D Q L X 33, 69

(T.R.) Θ, 13&c., 700.

13 &c., 1071

As displayed by Preuschen in the central column, the MS evidence appears amply to justify his conclusion that the text of Origen does not correspond to any of the recognised families But the right-hand column tells a different tale.    It shows that, so far as these particular passages are concerned, the text used by Origen has the closest resemblance to that of fam Θ though I suspect that the two readings in which Origen and Θ both agree with the T.R. against a few members of fam. Θ are not original, but the result of the text of both having been conformed to the Byzantine standard.    Now, if the above passages had been selected by myself to substantiate the conclusion for which I have argued above, the remarkable coincidence they exhibit between the text of Origen and fam. Θ would have been impressive.     But they are passages specially selected by Preuschen in order to prove a thesis precisely opposite to mine, viz. that Origen's quotations correspond to no known form of text.   The fact, therefore, that they so exactly bear out my own conclusion is, I venture to think, a strong confirmation of the correctness of this conclusion.

The third passage, which Preuschen selects, is Jn.vii.40-46. In this he quotes five variants.    In three of these the reading of Origen is supported by B T L; in the fourth, by T.   There remains the substitution of the perfect γεγένηται for the aorist ἐγένετο, a reading of Origen found in no extant MS. of the Gospels.    Seeing that the text of the Commentary on John depends on a single copy of the tenth century, our confidence that this last variant really stood in the passage as originally quoted by Origen must be very small.    Since of the four other variants three are found in B T L and the fourth in the Graeco-Sahidic MS. T-
the text of which, so far as it survives, is even nearer to B than א the passage merely serves to corroborate my own observation that the text of the Fourth Gospel used by Origen throughout the Commentary was on the whole nearer to B than to א.

I have not found leisure to test the scattered quotations from Luke or John, which occur here and there in the Commentary on Matthew a peculiarly delicate task, since most of them are short passages likely to be quoted from memory. But I have noted one passage where the context makes it clear that he is quoting from a written text and is contrasting the readings of John with the Synoptics. This occurs in his comment on Mt.xvi.24 (tom..24). Origen here (Greek and Latin support one another) quotes John with the addition of the words "and they laid upon him the cross." This addition is one of the most remarkable of the Ferrar readings and is only found else­where in Syr. Hier. This in itself is almost enough to prove that, whether he always quoted from it or not, Origen at this time certainly had access to a copy of John with the fam. Θ text.

A further question must now be raised.
Does fam. Θ represent a text which Origen found already in possession in AD 231 when he moved to Caesarea? Or is it a recension which he himself made at a subsequent date? There can, I think, be no reasonable doubt that fam. Θ represents the old text of Caesarea and not a recension by Origen.
The following are relevant considerations.

  1. In his Commentary on Matthew (tom. xv. 14), Origen, after deploring the number of variants between texts of the Gospels, gives a brief account of the efforts that he had himself made, θεοῦ διδόντος, towards the restoring the true text of the LXX; but adds that he had not dared to do the same thing for the text of the New Testament.
    In exemplaribus autem Novi Testamenti hoc ipsum me posse facere sine periculo non
    This passage would be decisive evidence that, at the time of writing the Commentary on Matthew, Origen had produced no recension of the Gospels, but for the fact that the words I have quoted from the old Latin version are not found in the Greek. But the Greek MSS. of the Commentary on Matthew ultimately all go back to a single much mutilated, and possibly intentionally abbreviated, archetype; also the clause in question, read in its full context, seems essential to the point which Origen is wishing to make. Hence it may be taken as reasonably certain that an equivalent clause originally stood in the Greek. In view, however, of the margin of uncertainty, further considerations may be adduced in support of that conclusion.
  2. Eusebius devotes a large part of the sixth book of his Ecclesiastical History to a description of the work of Origen; but, though he expatiates at length on Origen's critical labours on the text of the Greek Old Testament, he says nothing at all about any such work on the New.
  3. Origen's discussion of the reading Jesus Barabbas, Mt. xxvii. 16, 17, makes it quite clear that he objected strongly on theological grounds to the idea that the sacred name of Jesus should be borne by a robber. He affirms that it was absent from many MSS., and suggests that it was an heretical interpolation in the text on which he is commenting. This surely implies that it was in the text most familiar at Caesarea. Indeed, as he was obviously lecturing with a copy of the Gospel open before him, one would naturally suppose that it was the copy ordinarily used for public worship in the church in which his homiletical lectures were delivered. In a recension made by himself the offending passage would have been omitted, since he had excellent MS. authority for so doing.
  4. Jerome twice alludes to exemplaria Adamantii, that is, "the copies of Origen" (Adamantius was another name of Origen). The allusion is obscure. Jerome, however, shows no knowledge of any reading characteristic of fam. Θ; [This statement is based on a fairly thorough examination of the lists of non-Byzantine readings, either expressly cited by Jerome or introduced by him into the Vulgate, given in Wordsworth and White's Vulgate, pp. 653-671.]
    but he frequently appeals to MSS. practically identical with א.
    Reasons will be given (cf. App. IV.) for the view that if Jerome associated any text with the name of Origen it was that of א.

We conclude that fam.Θ represents the text which Origen found already established in the Church of Caesarea in 231.  This affords another fixed point for the history of the text of the New Testament.

The text of the New Testament is a subject about which so many theories have been spun that it may be well to recapitulate the evidence that this particular conclusion is not a matter of theory but rests on definitely ascertained fact,

  1. Θ and the group of allied MSS. contain between them an enormous number of readings which deviate from the standard text; in the great majority of these deviations different members of this group support one another in the readings substituted for those of the standard text,
  2. The readings in which one or more of this group of MSS. disagree with the Byzantine text are a definite set, to be ascertained by purely objective observation; it is found that these residual readings correspond to the text of the Gospels of Matthew and Mark used by Origen in his Commentary on Matthew, and to the text of Mark and Luke used in the later books of his Commentary on John. There is no room here for subjective judgements, the facts either are, or are not, as I have stated.

the MSS. sent to constantine

Caesarea and its Library had a considerable reputation in the Nicene and early post-Nicene period. Nevertheless, the number of MSS. which show a larger or smaller admixture of the Θ text is larger than we should have antecedently expected if it represented merely the local text of Caesarea. Again, the very different way in which the Caesarean and the Byzantine texts are mixed in the different members or sub-families of the Θ group suggests that these MSS. represent different mixtures current in several different localities. This implies that the Θ text was at one time very widely circulated. Here, I believe, von Soden is on the right track.

When Constantine rebuilt the old city of Byzantium hoping by magnificent buildings and imported works of art to make it worthy to replace Rome as the capital of the Empire, from policy and conviction he showed himself specially lavish towards the Church. About 331 he wrote to Eusebius the correspondence is still extant desiring him to prepare at the Imperial expense fifty copies of the Scriptures on vellum for the use of that number of churches in the new city, von Soden suggested that what he calls the "I text" is descended from a recension made by Eusebius and disseminated through these copies, von Soden's "I text," however, never existed, nor is there any evidence that Eusebius undertook a recension of the Gospels. But the natural thing for Eusebius to do would be to have the copies asked for by Constantine made from the oldest copies in the Library of Caesarea. The text of these would differ very little from that of the MS. used by Origen a century earlier in the same Church, and this MS., we have seen, had a text like that preserved in fam. Θ.

Some fifty years later, c. 380, Jerome was at Constantinople. He found that the authorities there advocated the text of the martyr Lucian a text that, as we shall see later, was practically identical with what I have called the Byzantine text. We can readily understand their preference of the Lucianic recension; it includes the longer conclusion of Mark and so many other interesting passages omitted by the Caesarean text. (Cf. the list, p. 88.) Assuming, then, that the authorities at Constantinople had decided to adopt it, what would become of the fifty copies given by Constantine? They were not written on perishable papyrus, but on vellum; and the vellum on which the two contemporary MSS. B א were written is still in excellent preservation after the lapse of nearly 1600 years. They would not be destroyed, they would be corrected some copies more thoroughly than others, some in one place, some in another. In the course of time the wealthier churches of the city would desire clean new copies, undisfigured by constant correction. They would get these from the best reputed copying establishments, whether secular or monastic, in Constantinople. Such establishments would have been careful to provide themselves with copies of the standard text; so the new copies would represent the Lucianic text. What would become of the old ones?
Most probably they would be given away or sold cheaply to smaller churches or monasteries in the provinces, who could not afford to buy new and clean copies of the standard text. Thus many of the fifty copies originally made for Constantinople, more or less corrected to the standard text, would get into the provinces. Some of them in all probability are the ancestors of some of the mixed MSS. we now possess.
[The peculiar mixture of texts in MSS. like
33 and 157 would be easily explained on the hypothesis that they are descended from Alexandrian ancestors sporadically corrected by MSS. of this mixed Lucianic-Caesarean type.]

I venture the suggestion that one of these discarded MSS. was used by St. Mesrop and St. Sahak to revise the Armenian version. These two, we are told, translated the Scriptures into Armenian about AD 400; but subsequently, receiving "correct" copies from Constantinople, proceeded to revise their earlier work. Dean Armitage Robinson [In "Euthaliana," Texts and Studies, iii. 3.] argues that the original translation was made from the Old Syriac. This has been lately disputed by the French scholar Prof. F. Macler a summary account and criticism of whose theories is given by R. P. Blake in the Harvard Journal of Theology, July 1922. Macler holds that the Armenian was derived directly from a Greek text of the type which von Soden calls the "I text," most nearly related to Θ, and lacking many of the characteristic readings of D a phrase which would serve as a description of fam. Θ. The question is one that hinges largely on linguistic considerations, a judgement on which demands a knowledge of both Armenian and Syriac, which I unfortunately lack. But it certainly fits in with statistics given on p. 90, which show that the Armenian is frequently a supporter of fam. Θ, not only where fam. Θ and the Old Syriac agree, but almost as often where they differ. The hypothesis that the Greek MS. used by St. Mesrop to revise his first translation had the fam. Θ text might, I think, explain the phenomena noted by Armitage Robinson, and also those brought forward by Macler. Its verification, however, must await the publication of a text of the Armenian version based on a critical study of the oldest MSS. with complete apparatus, which, to the best of my knowledge, does not yet exist. Meanwhile it would seem sufficiently plausible to justify us in provisionally regarding the Armenian as a supplementary witness for the text of fam. Θ.

Besides this, in Palestine itself, there would necessarily be in circulation many copies of the old text of Caesarea. These also would suffer correction from the standard text; and these half-corrected copies may be the ancestors of some surviving members of the Θ group. One such copy, very heavily revised, was, I believe, used by the corrector of א, known as אc, who worked in the library of Caesarea (cf. App. II.). Again, since Jerusalem, until the Council of Chalcedon, 451, was under the jurisdiction of the metropolitan of Caesarea, it is possible that Jerusalem used much the same text as Caesarea. It is at any rate an interesting fact that in 565, at the end of Mark, there is a colophon stating that it was copied from old MSS. from Jerusalem. If Constantinople, Caesarea, and Jerusalem were all centres of distribution, the evidence for a wide circulation of the Θ text is readily accounted for.

But though there is an element of speculation in any theory as to how this Caesarean text came to be propagated, there is none, I submit, in the conclusion that in fam. Θ this text is preserved. Superficially the MSS. of this family differ greatly from one another; but on examination it appears that this is solely due to the different degree to which they have been corrected to the Byzantine standard. Deduct the Byzantine readings, and the differences between these MSS. in regard to the residual text is very small. There are differences, but they differ far less from one another than do א B L. From this fact, and from the very close correspondence of this residual text with the quotations of Origen, we are entitled to infer that (however we may explain its preservation) the readings of this family give the text read at Caesarea about 230 in an extremely pure form.

It would be well worthwhile for some scholar to prepare a continuous text of fam.  Θ,after the model of Ferrar and Abbott's edition of 13 &c. It would then, I think, appear that a practically continuous text of the Four Gospels of this type has been preserved. And this text would rank alongside B and D as the third primary authority for the text of the Gospels.

concluding survey

If we look at the map we see at once that the Churches whose early texts we have attempted to identify stand in a circle round the Eastern Mediterranean Alexandria, Caesarea, Antioch, (Ephesus), Italy-Gaul, and Carthage. The remarkable thing is that the texts we have examined form, as it were, a graded series. Each member of the series has many readings peculiar to itself, but each is related to its next-door neighbour far more closely than to remoter members of the series. Thus B (Alexandria) has much in common with fam.Θ (Caesarea); fam. Θ shares many striking readings with Syr. S. (Antioch); Syr. S. in turn has contacts with D b a (Italy-Gaul); and, following round the circle to the point from which we started, k (Carthage) is in a sense a half-way house between D b a and B (Alexandria again).

Antecedently we should rather expect the text of any particular locality to be, up to a point, intermediate between those of the localities geographically contiguous with it on either side. But the exactness of correspondence between the geographical propinquity and the resemblance of text exceeds anything we should have anticipated. And this fact is, I feel, of some weight in confirming the general thesis propounded in these chapters.

There remains to draw a practical conclusion. In discussions of variants in commentaries and elsewhere it is usual, in quoting the MS. evidence for a particular reading, to cite first the uncials which support it in alphabetical order, then cursives in arithmetical order. This practice is fundamentally misleading, von Soden's method of quoting authorities in three great groups (K H I) would have been a great improvement had he divided his I group into three, corresponding to Θ, D, Syr. S., and their respective allies. What we want to know in any given case is the reading

  1. of B א and their allies,
  2. of D and its allies WMk. Old Lat.,
  3. of the leading members of fam. Θ,
  4. of the Old Syriac, Armenian, and Old Georgian, and
  5. of the T.R.

In subsequent chapters, therefore, I shall cite MSS. thus. Further, it is not as a rule necessary to cite all the evidence of each group. Thus, if a reading is supported by א B L, nothing is gained by adding C Δ Ψto the list; if it is supported by a, b, e, k,it is superfluous to add further Old Latin evidence. Only where the leading authorities of any of the great texts disagree with one another is it, for ordinary purposes, important to cite their sub­ordinate supporters. The method of citing all uncials, and that in alphabetical order, disturbs the judgement and inevitably gives an undue weight to mere numbers. The fallacy of numbers is insisted on by Hort (ii. p. 43 ff.), as it is only through a chapter of accidents, different in every case, that any MS. not representing the standard text has survived. The first principle of scientific criticism is that MSS. should be not counted but weighed. And the weight of a MS. depends on the extent to which it preserves, more or less, one of the ancient local texts.

P.S.—Since this chapter has been paged I have heard from Dr. Blake that further examination of the old Georgian confirms our first impression (cf. p. 91) of the close relation between the text of the version preserved in the Asyth MS. and that of fam. Θ.


(the student should memorise primary & secondary authorities.)



Syr. S.



kMk. Mt.


א L Sah. Boh.

Syr. C.

1 &c. 13 &c. 28

b a

WMk. e


C, 33, WLk. Jn.

ΔMk. ΨMk.

Syr. Pesh.

1424 &c. 544 N-Σ-O-

Φ Old Georgian

ff 2 h Mt. i r c Mt. Jn.

Frag: n (cf. a)

c Mk.


579Mk. Lk. Jn. 892 1241 157 X

Syr. Hcl.

Syr. Hier.

U Λ 1071 1604

ff, g, l, q

(?) f



Origen AD 230
Cyril Alex. 430


Origen AD 240 (Eusebius) 325

Tatian 170 Irenaeus 185

Cyprian 250

1 &c. = 1-22-118-131-209-872Mk.-1278-1582-2193.
13 &c. = 13-69-124-230-346-543-788-826-828-983-1689-1709.
1424 &c. = 28 MSS., including M, cited by Soden as IΦ.
Byzantine text: S V Ω; E F G H; (A, K, H, Y); (Γ); (WMt.).
Mixed Frags. P Q RLk.
1 &c. = fam. 1 = Sod. Iη;
13 &c. = fam. 13 = Sod. Iι;
Sod. Iα misleadingly includes D with Θ, 28, 544, 565, 700.