a very early Harmony of the narrative of the Passion are very premature, and in his judgment should either not have been put forward at all in a _Preliminary Study_ or suggested with more deliberation. G. Krüger also puts “this Combined Gospel written in Syriac (Greek?)” in his _History of Early Christian Literature_, § 37. On the other hand, Hogg in § 12 of his Introduction,[229] _Non-Syriac Texts of the Diatessaron_, says nothing of a Greek text, and in § 19, where he raises the question, “In what language was it written?”, he speaks only of the “view favoured by an increasing majority of scholars, that it was written in Syriac,” and then asks, on this view, “was it a translation or simply a compilation?” and lastly, which is the main question, “what precisely is its relation to ... the Western text generally?” In his first work, written prior to the publication of the Arabic text, Zahn very frequently pointed to the fact that the so-called =Western= witnesses—_i.e._, Codex D and the Old Latin manuscripts, agree so often with Tatian.[230] His explanation of this phenomenon is very simple—viz., that Tatian returned from Rome to his old home in Syria about the year 172, and took with him from the West his text, which was just the Western text. [Sidenote: Tatian and the Western Text.] This view would present no difficulty if it were only the case that the Diatessaron shared the peculiarities of the Western text, but is the fact not rather the converse of this—viz., that D, the leading representative of that text, shares the peculiarities of a Harmony of the Gospels, might we say, in short, of the Diatessaron? Not only are certain readings the same in both texts, but the Western text seems actually to exhibit features which can scarcely be regarded otherwise than as the outcome of a Harmony. I have given expression to this opinion ere now; it struck me forcibly when I was collating the Codex Bezae for my _Supplementum Novi Testamenti Graeci_. In order to afford a more convenient survey of the vast number of variants, I followed the paragraphing of Westcott and Hort’s edition. Now look at the variants there. Whereas the majority consist of quite separate and disconnected readings, I was obliged at the beginning of the pericopæ regularly to copy half a line or even a whole line from D, its text differed so much from that of our present editions at the beginnings of the pericopæ, and there only to the same extent. See, _e.g._, Luke v. 17, 27; vii. 1, 18; ix. 37; x. 1, 25; xi. 14; xii. 1 to the end; xxiv. 13. It is true this phenomenon is most frequently observed in Luke, where I had previously explained its appearance in another way by supposing like Blass that it was due to the author having issued two editions of that Gospel. But neither is it altogether absent from the other Gospels. It occurs most seldom, as might be expected, in Matthew, but examples may be seen in xvii. 22, 24; xx. 29. In Mark see iii. 19; iv. 1; vi. 7. There are other