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.



Home | The textual homogeneity of the Θ family | the purple MSS., fam. 1424 & א | the K Π group | U Λ etc. | Table 1 (Lk.ii.1-25) | Table 2 (Jn..1-6) | top

the textual homogeneity of the Θ family

No early MS. has a text entirely homogeneous.
Even B, as Hort insists, has not escaped "sporadic" corruption, while א has a considerable infusion of "Western" readings. In view of the statistics as to the differences of these two Alexandrian MSS., which nevertheless are the most closely related of all MSS. earlier than AD 500, given on p. 329, the hypothesis that the various members of fam. Θ represent a single local text would not be seriously imperilled unless the number of variants within the family exceeded the number of the differences between א and B. So far, however, as I have been able to test it, they are very few; from which we may infer that the extent to which the leading MSS. of fam. Θ have been crossed by any text, other than the Byzantine, is very small.

Lake's Table1 of readings in Mk.i. exhibits 102 variants [Harvard Journal of Theology, July 1923, p. 270 ff.]; but in only 5 of these do members of the family give a reading found in any text other than that of the family or, of course, in the Byzantine text. In regard to the same set of 102 variants א differs from B 16 times. In the Table for Luke ii.1-25 at the end of this Appendix there appear 44 departures of fam. Θ from the T.R.; out of these 44 variants there are 5 in which members of this family differ from one another, and 6 in which א differs from B.&nbsp The similar Table for John .1-6 shows 3 differences between א and B, but not a single instance of one member of fam. Θ opposing another in any non-Byzantine reading. The Table of variants common to the family and the text used by Origen in his Commentary on Matthew tells the same tale. It may, however, be worthwhile to supplement this evidence by the result of a few preliminary tests, which I essayed before drawing up these tables.

(a) The lists of classified readings in the Introduction to Lake's Codex 1 and its Allies afforded another means of testing the relation of Θ to fam. 1. List G in the Introduction (p. Ixxi) contains readings peculiar to fam. 1. In Matthew there are 8 of these. I find that Θ agrees with fam. 1 in 3 of these readings, but in the other 5 has been conformed to the Byzantine text. According to von Soden, 2 of the 5 not supported by Θ are found in 1424, and 2 more in one of the Purple MSS. List F gives readings "which are supported by a few other MSS., but cannot be identified with any authority generally recognised as primary." I checked this list against Θ for the part of Matthew which is extant in that MS. (much of Matthew i.-v. is lost). Twenty-four readings of fam. 1 are concerned. Ten of them appear in Θ; in all the other instances Θ gives the Byzantine reading. List E gives readings in which fam. 1 agrees with B א against D Old Lat., Old Syr., and Byz. In Matthew there are 23 of these; 12 of them are found in Θ, which otherwise follows the Byzantine text. List B gives the readings found in fam. 1, for which the Old Latin is the chief ancient authority. In Matthew there are 11 of these, 4 of which are found in Θ. It appears, then, in whatever direction we look for the readings which are in some special way characteristic of fam. 1, we find that a large proportion of them appear in Θ; and, where that does not happen, the occurrence of the Byzantine reading in Θ shows that in these passages it does not represent its own characteristic text.

(b) Proceeding to test the relation of Θ and fam. 13, I at once noted that Θ exhibits the famous Ferrar reading (found in 346—828—828) ᾧ μνηστευθεῖσα κτλ., Mt.i.16. Then, by way of a fair test, I opened my copy of Beerman and Gregory's edition of Θ at a venture in that part of the volume that gives a collation of the readings of Θ with the other MSS. The book opened at p. 657, and I worked through the next six pages, which happened to include Mt.xviii.25 to xi.2. In this short section there are a very large number of readings of Θ in support of which one or more members of the Ferrar group are quoted; but since the majority of these are also found in at least two of the great uncials B א L D, they afford no evidence of a special connection between Θ and fam. 13. There are, however, 9 readings in which Θ is the only uncial (apart from fragments) supporting the reading of fam. 13; 4 readings found in both Θ and fam. 13 but supported by B only of uncials; 4 ditto supported by D only; 2 ditto supported by D Δ, and 2 ditto supported by inferior uncials. This seems strong evidence of a very close relation between the text of Θ and that of fam. 13.

(c) To test the text of these MSS. in Luke, I turned to the Introduction to Hoskier's collation of 700, in which he gives the MS. support for all variants of this MS. that are in any sense uncommon. In the first chapter of Luke there are 26 such variants. I found that, if the readings of Θ were added to those of the MSS. cited by Hoskier, every single one of these readings of 700 was supported by at least one other member of the family.

(d) So far as Mark is concerned, 565 would appear to be slightly superior to Θ; but in the other Gospels it has suffered more from Byzantine revision than any other of the group with which we are concerned. It was for that reason that I selected Luke ii.1-25, John vi.55-vii.3, and John .1-6 for the Tables in the Text and Appendix. A glance through Belsheim's collation had shown me that the proportion of non-Byzantine readings in 565 was above the average in these passages, and therefore their character could be most easily tested here. It so happens, however, that all these are passages in which Θ and 700 have suffered rather heavily from Byzantine correction so that the Tables somewhat understate the value of these two MSS. as compared with other members of the family,

Theoretically, of course, no proposition as to the homogeneity of the texts of these MSS. can be held to be proved until every reading in all four Gospels in each MS. has been compared. Practically, the chances are very small that the various tests enumerated would have come out as they did unless the fundamental text were unusually homogeneous.

the purple MSS., fam. 1424, and אc

There are four MSS. of the sixth century, Ν—Σ—Ο and Φ, written on purple parchment in letters of silver—except Ο, which is in letters of gold. Ο contains only a fragment of Matthew; Φ and Σ contain Matthew and Mark almost complete ;
contains portions of all four Gospels. N—Σ—Ο are so much alike that Mr. H. R. S. Cronin, who has made a special study of the group, believes that they were copied from the same exemplar [Texts and Studies, v. 4 (1899), and J.T.S., July 1901, p. 590 ff.]. In general they present the Byzantine text; but there is a small proportion of earlier readings. The text of Φ is closely allied to that of N—Σ—Ο, but with an additional infiltration of D readings.

When I first began testing the homogeneity of fam. Θ there happened to be on my table Mr. Cronin's edition of Codex N (Texts and Studies, v. 4). Recollecting that the books speak of a connection between this MS. and the Ferrar group, it occurred to me to use some of his lists to test the text of Θ, choosing Luke for the investigation. One list (p. lix) gives the readings in which N agrees with fam. 13, against the T.R. and all the leading uncials. There are 7 such readings in Luke. Of these 3 occur in Θ; of the remaining 4, 1 occurs in 131 (fam. 1), 1 in 565, and 1 in M (fam. 1424). Another list (p. Ix) gives agreements of N with the texts of the "Better Uncials"—meaning either B or א L combined—against the majority of MSS. In Luke 22 instances are given; Θ supports N in 10 of them. Noticing that for most of the 22 Mr. Cronin quotes the support of fam. 1 or fam. 13, I proceeded to test the cases where he does not quote such support by means of von Soden's Apparatus, which of course gives the readings of some new members of the Ferrar group and of 700. It then appeared that in only 3 of the 22 readings did N lack the support of either Ο or one of the Θ family. That is to say, N hardly ever agrees with א B L except where these support fam. Θ.

The early date of the Purple MSS. made it seem specially worth while to explore still further the relation between their pre-Byzantine element and the text of fam. Θ. And, as Mark is the Gospel where the characteristic text of both groups of MSS. is best preserved, I proceeded to test the text of N in that Gospel.

Mr. Cronin gives (p. li f.) a list of 48 readings in Mark where N and Σ agree together, and are supported by a very few MSS. against all the leading uncials and also against the Byzantine text; in each he cites the MSS. which support them. From these it appeared that 31 of the 48 readings in N, 2 occur in one or more then known members of fam. Θ. But, checking the list by the new evidence of Θ 700 and von Soden's revised collations, I found the number rose to 37. This struck me as remarkable. I then noticed that, of the remaining 11 readings, 9 were supported by the group of MSS. which von Soden classes together under the symbol Iφ, but which on the analogy of the accepted usage in similar cases I have called fam. 1424. This group he defines as one intimately related to fam. 1 and fam. 13, but preserving a few readings that have been eliminated from these MSS.

The next step was to test the combination N Σ by the Table of readings characteristic of fam. Θ given in the article by Lake and Blake in the Harvard Theological Review.
is not extant for the beginning of Mark, but Σ is. The results of a scrutiny of the 102 variants in Mark i. there tabled, tested against the collation of Σ by Gebhardt and Harnack, may be succinctly presented as follows:

Agreements of Σ with T.R. 84
Agreements of Σ with fam. Θ 14
Conflations of text of fam. Θ with T.R. 2
Agreement of Σ with fam. 1424 in a reading not preserved in other members of f fam. Θ 1
Agreements with other MSS. (A 33) 1
. 18

These figures materially strengthen the conclusion that the ground text of N Σ was identical with that of fam. Θ, only that it has suffered a much larger amount of revision to the Byzantine standard. Obviously, however, they would amount to demonstration if it could be shown that von Soden was right in view of the relation of fam. 1424 to fam. 1 and fam. 13.

Accordingly I proceeded to test the character of fam. 1424 by reference to Lake and Blake's Table for Mark i. Assuming that where von Soden fails to cite the evidence of any of these MSS. it agrees with the Byzantine text, it appears that in 52 of the 102 variants cited fam. 1424 represents the T.R., but in 40 it goes with one or other member of the Θ family; 9 times it has readings differing from the T.R. but not found in any of the six representatives of fam. Θ cited in the Table. One of them, however (the omission of εὐθύς; i.43), is found also in 828, a Ferrar MS. quoted by von Soden but not included in Lake's citations; and one is practically the fam. Θ reading (i.e. the addition καὶ τεσσάκοντα νύκτας; i. 13, only with a transposition of the last two words); and two are found only in one MS. of fam. 1424 and look like errors of the individual scribe. In one fam. 1424 agrees with D, but all members of fam. Θ except Θ itself here go with T.R., and the reading of Θ ἧλθον for ἀπῆλθον, found in no other MS., is probably a slip, so that here fam. 1424 may well preserve, the original fam. Θ reading. In all other variants, so far as one can infer from von Soden's general system of citing (and I know of no collation by which to check him), fam. 1424 gives the Byzantine reading.

The Tables of readings in Luke and John, given p. 83 and p. 582 ff., show similar results, and justify us in treating fam. 1424 as a genuine and important constituent of fam. Θ. I may add that in the course of writing this book I have had to study the MS. evidence given by von Soden in innumerable cases up and down the Gospels, and have found nothing to conflict with the results obtained above. Accordingly, though it may be that a few of the less important of the twenty-eight MSS. which he groups as Iφ ought not to be included, he has discovered a real group; and fam. 1424 must be treated as an important con­stituent of the Θ family. I have also found reason to accept his view that 544 (ε 337) is a true member of the same family. [A partial collation of 544 is given in Scrivener's Adversaria Critica, p. 1-liv (Cambridge, 1893), under the number 567.]

It occurred to me to test the readings of the corrector of א whom Tischendorf cites as אc or אca, and who probably belongs to V or VIcent.
[Cf. the facsimile edition of א by K. Lake (Clarendon Press, 1911), p. xvii f.] The chapters of Matthew which I tested showed a predominantly Byzantine text with a sprinkling of readings definitely of the fam. Θ type. This is interesting, as in the O.T. and in the Epistles אc seems to have used a MS. in the hand of Pamphilus. [Bousset, op. cit. p. 45 ff.] Unless, as is possible, but not probable, Pamphilus in prison copied a MS. of Lucian's recension, this shows the Byzantine text dominant in Caesarea by V or VIcent.

the K Π group

Von Soden classes M, which has a very small non-Byzantine mixture, as an inferior member of fam. 1424; Bousset (Text-kritische Studien, chap, iv.) regards M as a poor relation of K Π. The lists of the readings in his chapter on "The K Π Group" favour von Soden's view that these two uncials (which are supported by a number of cursives) have the same relation as have N Σ to the fam. Θ text, except that they are more predominantly Byzantine and have a smaller admixture of the older text, von Soden classes A: the famous Codex Alexandrinus, with K Π; and, so far as I have tested the suggestion, I think he is probably right in supposing that the non-Byzantine element in A represents mainly, if not wholly, the fam. Θ text. W. C. Braithwaite (Expository Times, i. pp. 114 ff.) says that the recently discovered uncial Y has affinities with K Π; and a hurried look at the collation of the MS. given in Gregory's Texikritik (pp. 1928 ff.) seemed to show that its non-Byzantine element (perhaps 10 %) is at any rate closely connected with fam. Θ. Since, however, all the sub-families of fam. Θ overlap one another, it is not of much importance, especially where the non-Byzantine element is small, whether in border-line cases, like A or M or Y, a MS. is included in one sub-family or another, or regarded as forming a class by itself. The same consideration applies to the question (discussed p. 80, note) whether 22 should be included in fam. 1 or not. What does matter is to know whether the non-Byzantine element in a mixed MS. belongs mainly or entirely to the Θ family.

The K Π group is regarded by von Soden as a definite recension; he styles it the Ka text, and holds that it was used by Chrysostom in his Homilies on John and in the so-called "Antiochene Commentary" on Mark (by ? Victor of Antioch ±420), and on Luke by Titus of Bosra, 370. I am a little sceptical as to the clear-cut distinctions within the Byzantine text, which von Soden believes he can detect; but, if the Ka text was used by these fathers and is that of the Vcent. MS. A, may not this be the text of Lucian? The K1 text in other works of Chrysostom may be due to scribal revision to the VIcent. Byzantine text.

U Λ, etc.

Two other sub-families of MSS., regarded by von Soden as authorities for his "I text," are headed respectively by U and Λ. The non-Byzantine element in fam. U (in which he includes the interesting cursive 1071  (ε 1279)) [Described and collated by K. Lake in Studia Biblica, v. p. 132 ff., Oxford, 1903.] seems to be about as large as that in the Purple MSS.; and, so far as I have observed, it represents the fam. Θ text. Fam. Λ seems to have a smaller non-Byzantine element, and therefore is more difficult to test; but I do not happen to have noticed any readings which suggest that this element is other than the fam. Θ text; and I would say the same thing of 1604 (ε 1353).

An immense number of MSS. are assigned by von Soden to the I text. Unfortunately, however, his inclusion of D WMk, Old Lat., Old Syr. in the I text vitiates his principle of classification; for it would justify his assigning to that text a MS. containing a considerable mixture of specifically Syriac or Western readings. This consideration precludes one from the simple expedient of classing as authorities for the text of fam. Θ all MSS.—merely excepting D WMk, Old Lat., Old Syr. —cited in von Soden's Apparatus as authorities for the I text. They must be scrutinised again in every case. And this caution is the more necessary as von Soden is over-anxious to enlist MSS. in support of the I text. For example, 157 is reckoned as an I MS., and it undoubtedly has a number of readings characteristic of fam. Θ; but a much more striking feature of this curiously mixed MS. is its support of the Alexandrian text. The fragments P Q and R von Soden classes as authorities for the I text; Γ also is claimed as a weak supporter of the same text— perhaps rightly, but it also has some striking Alexandrian readings.
[My confidence in von Soden's classifications was seriously shaken by testing the VIcent. fragment 089 (ε 28 = Tischendorf's Θe) containing Mt.xxvi.2-4, 7-9, which he quotes (vol. i. p. 1350) as "a pure I text." Tischendorf gives 8 variants from this fragment, of which 7 occur in B א and the remaining 1 in B Δ.]

One naturally asks if all traces of the old text of Antioch have disappeared. If we are right in surmising that this was the Greek original of the Old Syriac, a predominantly Byzantine MS. in which the remnants of such a text survived as a small element of mixture might easily be mistaken for a weak member of fam. Θ, since fam. Θ and the Old Syriac have so much in common. Again, there must have been early local texts in Asia Minor and Macedonia the parts of the world in which the majority of our later MSS. were probably produced and it would be very strange if no readings at all from these texts had crept into the later MSS. The astonishing thing is that, of the sporadic non-Byzantine readings that survive in later MSS., there are so few which are not also found in one or other of the great texts which we can identify. I can only account for this by supposing that, at the time when the transition from papyrus to parchment was made, the smaller churches, instead of copying their local texts, obtained their new parchment copies from the larger centres. This change of material seems to have taken place early in the fourth century, that is to say, just after the revisions by Lucian and Hesychius were accepted at Antioch and Alexandria. And we know that the copies of the text of Caesarea with which Eusebius supplied Constantine (from which we have suggested many representatives of fam. Θ are descended) were written on parchment.

In order further to illustrate both the essential homogeneity of fam. Θ and the curiously sporadic and unsystematic character of the assimilation of earlier texts to the Byzantine standard, I append two Tables modelled on that drawn up by K. Lake to which reference has been so often made.