A.C.  These two MSS present the normal uncial text just as 046 and in some degree 025 present the normal cursive text. But whereas C is most carefully written, this is not true of A, which is seriously affected by copyists' blunders. C exhibits fewer singular readings than any other uncial (about 67), and these singular readings, moreover, with a single exception, possess no special interest. Here it is that it differs in kind from A and calls for different classification. A contains over 150 singular readings, and of these 56 (if not 63) preserve the original. Thus {clxxii} whereas C's singular readings take no particular direction, A's are pre-eminent as being certainly right in over 60 passages.

a.  This MS "is of all the five MSS far the least worthy of regard as representing a defensible form of the text; it is aberrant rather than divergent from the rest, to the point of eccentricity." So Gwynn (op. cit. p. xliv) rightly judges. When it stands alone, it is only right in four passages. The bulk of its variants are, unquestionably scribal blunders and corruptions of an early date, and call for no further consideration. A considerable part of the remainder represents an ancient element foreign to the normal uncial text and finds large support in the versions and to a less extent in certain cursives. Other variants connect a with the normal cusrive form of text, but these are not numerous.

025. 046.  These MSS are so widely sundered that they differ from each other in kind. While 025 represents on the whole the uncial type of text, 046 represents the cursive type. While slightly over half the variants of 025 from the other uncials find support among the cursives, more than four-fifths of the variants of 046 find such support.

But though 046 is largely cursive in character, its record compares favourably with a, considering its late date. We have already seen (see Table I. p. clxiv) that whereas a alone preserves 6 right readings (reckoning together columns one and two) against the rest of the uncials, 046 preserves 3. Again Aain combination are right 33 times, A 046 are right 31 times. Once more, from the results arrived at in § 4 we learn that, whereas a enters into groups of three or more MSS attesting the right text 45 times, 045 does so 40.

025 nad 046 are to be further distinguished from each other in this respect, that whereas 046 represents the close amongst the uncials of a long process of correction which began in the 2nd century, 025 represents to a considerable extent a deliberate recension of the texts of the 8th cent. or earlier. That 025 is the result of a deliberate recension is easy to prove. Nearly forty times it differs from the other uncials in correcting or improving the Greek text from the standpoint of Greek syntax. Thus in 1⁴ we have a[ + evstin evnw,pion) 1⁵tw/| avgaph,santi) 1⁶ basilei/j kai. i`erei/j) 1<sup>9</sup> sugkoinwno.j evn th/| qli,yei kai. $+ evn th/|% basilei,a|) 2<sup>9</sup> th./.n blasfhmi,an tw/n lego,ntwn) 2<sup>13</sup> evn t) h`me,raij +  evn ai-j) 2¹⁷ dw,sw auvtw/| + fagei/n) 2²⁰ th.n gunai/ka ) ) ) th.n le,gousan) 4¹ h` fwnh. ) ) ) le,gousa) 5<sup>2</sup> khru,ssonta fwnh/| mega,lh|) 5<sup>6</sup> avrni,on ) ) ) e;con) 7<sup>9</sup> o;cloj ) ) ) e`stw/tej( ) ) ) peribeblhme,noi) 8¹³ avgge,lou petome,nou) This change is due to the scribe's idea of syntax, but of the sense of the passage. 9¹⁴ fwnh.n ) ) ) le,gousan) 10¹ kai. i=rij, corrected {clxxiii} according to sense of context. The scribe knew no better. 11⁴evlai/ai -- e`stw/sai) The above examples are sufficient to prove the fact of a deliberate recension. On the influence of this recension on 35. 205 and other cursives, see under 35. 205, p. clxxv sq.

The following cursives -- the list is provisional -- agree with 046 in giving the latest form of text: 
 

046 contains many readings of so late a date that they are not supported by any version. These are of the inferior cursive type. A few examples will suffice. Thus in 112 046 with cursives reads kai. + evkei/: 1¹⁶ceiri. auvtou/ th//| dexia/|: 2²⁵ avnoi,xw (for a'n h[xw): 3²avpoba,llein for avpoqanei/n: 3⁴ovli,ga e;ceij ovno,mata (order): 3⁷ eiv mh. o` avnoi,gwn)
 

§ 9. Cursives collated for this edition. -- The list of the 22 cursives colalted for this edition is given in vol. ii. p. 234, where attention is drawn to such as defective. Of thse the most interesting and valuable are 2020. 2040. 2050.

2020 is a good cursive and would stand close to 025 a in the third class. It agrees with A 2019 in 2¹⁸ and 1¹⁰ save that for o;pisqen it reads ovpi,sw, and with A and certain cursives in 1⁶. Over against seven agreements with A, it supports a in 18 passages and 025 in 13.

920. 2040 2040 (xi-xii cent.) 920 (x cent.). Though 2040 is written by the same hand throughout, it exhibits two distinct types of text. From 1-11⁷ it is of the late cursive type and seems to have been copied from 920 (x cent.). These two MSS contain unique readings in the following passages: 3⁵ tw/n zw,ntwn: 3⁸ ta. e;rga (for to.n lo,gon%: 3¹² tw/| ovno,mati (for tw/| naw/|): 4⁹ + kai. proskunh,swsin (-sousin, 920) tw/| zw/nti and another addition in 8². In 4¹⁰ they omit evnw,pion t) qro,nou and have other omissions in 4⁴ 5¹² 7⁴ 9⁹. They invert the order in 3⁸ and attest the same impossible readings in 5¹ 6¹⁴ 7¹ 9⁵.

From 11⁹ to 20¹¹ where it ends, the text is largely free from corruptions of the later cursives. It often supports A against most other authorities (cf. 11¹¹eivsh/lqen evn auvtoi/j, 12¹² oi` ouvranoi) and a and less often 025. But its excellence is still more clearly shown by the fact that in 11⁹-20¹¹ it agrees with the majority of uncials against the majority of cursives. The latter half, therefore, of 2040 is of so high a character as to entitle it to be ranked with 046, and after a.

2050. This MS, which consists only of 1-5, 20-22, and was clearly copied from a defective MS, stands in point of excellence alongside the uncials. In about 80 passages it agrees with the {clxxiv} majority of the uncials against the majority of the cursives. Thus in 1⁴ it reads avpo. o` w;n with AaC 025 al<sup>20</sup> fl gig vg s<sup>1.2</sup> bo against 040 and most cursives. In 1<sup>9</sup>evn VIhsou/ with aC 025. 2020 gig vg s<sup>1</sup> bo against the rest; VIhsou/ (without Cristou/) with Aa* 025 al<sup>5</sup> fl gig vg<sup>-d</sup> arm<sup>a</sup> against the rest. In 1<sup>12</sup> kai. (without evkei/ Aa 025. 045 al Tyc Pr fl vg s<sup>1.2</sup> bo against the rest. In 1<sup>13</sup>lucniw/n (without preceding e`pta,) ACP al¹⁰ Tyc Pr fl s^1.2 arm^1.2.4.a bo against the rest. In 2¹³ > ta. e;rga sou kai, (added by 046 al^pl s² arm^3.a) with AaC 025. 2020 and versions (-- s² arm^3.a): o` pisto,j mou AC 61. 69 Or<sup>s</sup> s<sup>2</sup> against rest. These suffice to show the character of this cursive. This cursive shows some slight affinities with A, as in 1<sup>13</sup> 4<sup>4</sup> 5<sup>4</sup> 22<sup>11</sup> etc., and still more with a. Thus with the latter it agrees in 1<sup>8</sup> (+ h` avrch. ktl)), 1¹⁵ pepurwme,nw| (a correction), 1¹⁷ evpe,qhken( 2²⁰ 4²⁰ etc. It agrees with 025 in 1¹⁵ca,lkw| liba,nw|( al⁶: 2²⁰ th.n le,gousan (also a^c al⁵), etc.

This cursive has a conflate reading in 2²⁷kai. suntri,yei auvtou.j w`j ta. skeu,h ta. keramika. suntri,betai) Such a conflation is not found in any other MS or in any version. But gig arm⁴ bo eth read suntri,yei auvtou,j) Is 2050 influenced by gig or some ancestor of these versions? In 1¹⁶ 2050 with 920. 2040 Tyc fl gig vg read dexia/| auvtou/ against all other Greek authorities. Is there a trace of Latin influence here?

149. 386. 201.  Of these 201 was not collated for this edition. The first of these cursives, 149 (xv cent.), is a slavish copy of 386 (xiv cent.). It reproduces it where it is absolutely wrong: cf. 2¹⁴ evdi,dasken t) Balaa,m( 3¹⁴ h` avrch. th/j pistewj( 14<sup>19</sup> 18<sup>4</sup> la,qhte) In 13<sup>6</sup> it reads katoikou/ntaj with 201 against 386. 2019 oivkou/ntaj) Where 386 is quoted in the Appar. Crit. it carries 149 with it, unless 149 is quoted to the contrary. 201 (xiii cent.) is a member of this group. It agrees with 149. 386 in unique (or almost unique) readings in 3<sup>2</sup> (> peplhrwme,na): 3<sup>14</sup> h` avrch. th/j pi,stewj: 10² evpi. th.n gh/n (also 1): 11⁴ oi` evnw,pion: 14<sup>18</sup> bota,naj: 15<sup>6</sup> oi` e`pta. a;gg) evk tou/ naou/ oi` e;contej t) e`pta. plhga,j (also s¹ bo): 1617 tou/ qro,nou + tou/ qeou/) This is a conflation of tou/ qro,nou, A 046 al^pl, all versions (-- gig) and atou/ qeou/( 18⁷eivmi. kaqw,j( 20⁴evdo,qh kri,ma( and others. This group gives a late cursive text.

175. 617. 1934.  These cursives form a group, but one much less closely connected than the one immediately preceding. In 2¹⁹ they stand alone in reading cei,rona tw/n prw,twn( and in 17¹⁵ a] ei=dej + kai. h` gunh,: with 141. 242 in 6¹⁷ in reading swqh/nai) In the following passages these cursives attest the same text in conjunction now with one set of authorities now with another -- not consistently with any -- 10⁸ 17³ 18^8.22 19^7.11.13 20^{12 216.27} 22^{8.12.13.16.20.21}. 175 and 617 several times agree where 1934 diverges: 18¹⁶ 19²⁰ 20⁵ 21³ 22⁵ etc. and generally in conjunction {clxxv} with 025 text. This group gives a very late form of the cursive text, except in chapters 16-22 where they agree generally with 35. 205.

325. 456. 468.  The first two members of this group are closely connected. They stand alone in adding in kata, sou in 2⁵ and the marginal note evn a;llw| B in 14²⁰, in omitting kai. evnw,pion ) ) ) auvtou/ in 3⁵ and e;cwn ) ) ) te,tarton zw/|on in 4⁷, and reading (325**) dw/| in 4⁹ and cro,non for e;ti cr) mikro,n in 6¹¹, in omitting gemou,saj in 15⁷. In very many passages these two cursives attest the same text in conjunction with a variety of others: cf. 6¹⁷ 7⁵ 8² 9^2.9 14⁸ etc. 468 agrees frequently (but apparently always in conjunction with others except in 15⁶oi` a;gg) oi` e`pta,) with 325. 456.  See 1<sup>6</sup>kai. poih,santi h`mi/n basi,leion i`era,teuma and > eivj t) aivw/naj( 2²² ba,lw( 3²th,rhson( 7²tou/ qeou/ zw/ntoj) See also 9^6.11 14¹⁴.

35. 205.  205 may be directly derived from 35, though other links may have come between. They stand alone in 3² kuri,ou tou/ qeou/( 9¹⁸ tw/n triw/n tou,twn plhgw/n) In conjunction with a variety of uncials, these two cursives agree in over 110 passages. This number would be still greater but that 18¹⁴-20⁹ (= one page of 205) was not photographed through an error of the photographer. Hence for the number 110 we should read 120 or thereabouts. But dealing with the passages actually given in the Appar. Crit. 35. 205 agree 20 times with each of Aa 025 and AaC 025; 3 times with each of Aa and AaC; 2 times with AC 025; 5 with A; 1 with A 046. All these are first class groups, and nearly all the readings so attested are right. Thus so far 35. 205 exhibit a good uncial type of text. But 35. 205 show affinities with another type of readings, a considerable number of which have originated with the recension of 025, which they have followed 28 times, and almost always wrongly. 

The influence of this recension of 025⁽¹⁾ is seen clearly in 1. 35. 67^mg(?). 104(?). 205. 468**. 620(?). 632**. 1957. 2015. 2019(?). 2023. 2036. 2037. 2038. 2041. 2067, etc.  I add here three examples of the influences of 025 on later MSS. 2⁵evkpe,ptwkaj (instead of pe,ptwkaj) 025. 1. 35. 104. 205. 620. 1957. 2015. 2023. 2036. 2037. 2038. 2041. 2067.  2¹⁷ + avpo, before tou/ ma,nna 025 (where the slip xu,lou in 025 is rightly corrected in later MSS). 1. 35. 61^mg. 104. 205. 468**. 620. 632. 2015. 2023. 2036. 2037. 2038. 2041. 2067.  2⁹blasfhmi,an evk (> 025) tw/n lego,ntwn) Here this obvious correction is followed by 1. 35. 205. 1957. 2015. 2019. 2023. 2036. 2037. 2038. 2041. 2067 Or^s.

Of groups of the second or third class 35. 205 follow aC {clxxvi} 025, a 025. 046, a 046 once each: a (or a^c) C 025 3 times: a 025.  11: a 6.

205 presents two conflate readings in 13¹⁴ 14⁶.

Thus group (35. 205) has quite the value of an uncial -- superior in the main to 046, but falling short of 025.

§ 10. Origen's so-called text -- in this edition Or^s. -- Whether the text which accompanies undoubted scholia of Origen is really the text of Origen, Harnack in his edition (Der Scholienkommentar des Origenes zur Apokalypse Johannis, 1911), p. 81, leaves undecided. He claims that it is a text of the highest character of the 10th century, whicih "though it may not prove to be even a rival of C, perhaps even not of A, is at all events on an equality with a and 025, while it is certainly superior to the text of 046 and Andreas."

But this text is not deserving of such praise. (a) It has nothing to do with the text that Origen used. I will compare the texts in a few passages. In 37 Or^s reads: ta,de le,gei o` a;ggelo,j avlhqino,j ) ) ) o` avnoi,gwn kai. ouvdei.j klei,sei auvth.n kai. klei,wn kai. ouvdei.j avnoi,gei( eiv mh. o` a`noi,gwn kai. ouvdei.j avnoi,xei) Here, as the Appar. Crit. in loc. shows, the text which Origen used differed in two respects (see heavy type) in this verse, and agreed in these with the text of this edition. Or^s alone is conflate. It combines kai. klei,wn ) ) ) avnoi,gei(the text of A 025) and eiv mh. o` avnoi,gwn ) ) ) avnoi,xei (the text of 046 and most cursives). Again Origen > avkou,sh| t) fwnh/j mou kai, always when quoting 3<sup>20</sup>, but not so Or<sup>s</sup>. This may be an accident. In 5<sup>1</sup> Origen reads e;swqen k) o;pisqen and also e;mposqen k) o;pisqen, but Ors e;swqen k) e;xwqen) In 5<sup>5</sup> Origen rightly reads avnoi/xai( but Or<sup>s</sup> o` avnoi,gwn with 046 and cursives. In 7³ Origen reads mh,te t) qa,lassan( but Ors kai. t) qa,lassan( and a;cri against Ors a;crij ou-) In 1⁶ Origen (c. Celsum, viii. 5) has basilei,an where Or^s gives merely a cursive reading. A multitude of such divergenes will be found in Harnack's work (p. 76 sqq.). In the face of such divergences it is impossible to identify Or^s with the text of Origen.⁽²⁾

But a more important task awaits us. We have to define the relations of Or^s and determine its position with reference to the main texts of J^ap. We shall find that this position is not high amongst the uncials, as Harnack would have it, but low amongst the cursives. It will not be necessary to bring forward the entire evidence, but the following will suffice.

(a) Or^s is full of corrections like 046, or rather in dependence on it. -- In 1²⁰ it reads avste,rwn w-n with 046. But our author never uses the attracted relative. After 046 it corrects 2²⁰ th.n{clxxvii}gunai/ka ) ) ) h` le,gousa into th.n gun) ) ) ) h] le,gei( and 3<sup>12</sup> th/j kainh/j VIer) h` katabai,nousa into t) kainh/j VIer) h] katabai,nei) With cursives only it corrects 10⁸ la,lousan ) ) ) le,gousan into la,lousa ) ) ) le,gousa) Now this last correction is most probably the correction of an original slip of the author, but the other two constructions are Hebraisms in the text and should not have been altered. 5¹⁰ basilei,an kai. i`erei/j into basilei/j k) i`erei/j)

(b) It makes additions to the text with 046: 2¹³ + ta. e;rga sou kai,: and with a 046: 2⁹ + ta. e;rga kai,)

(c) In 8¹² we have a conflation of A and 046: kai. to. tri,ton auvth/j mh. fa,nh| h`me,ra kai. h` h`me,ra mh. fa,nh| to. tri,ton auvth/j( where 046 comes first and A second. Another conflation appears in 4⁸ (see (g) below).

(d) A few of the passages where it follows 046 and some cursives. -- 1¹⁰ fwnh.n ovpi,sw mou mega,lhn: 1¹² kai. + evkei/: 2¹⁰paqei/n: ivdou. + dh,) dh, does not belong to our author's vocabulary. 2¹⁴ + kai, before fagei/n: 4⁴tou.j qro,nouj + tou,j: 4⁷ > w`j before avnqrw,pou: 4<sup>11</sup>h`mw/n + o` a[gioj: 5<sup>5</sup>p` avnoi,gwn (where the text is avnoixai): 9²kami,nou kaiome,nhj)

(e) Directly or indirectly it follows 025 in the following corrections: -- 2⁹ th.n blasfhmi,an tw/n lego,ntwn: 2¹⁷ dw,sw auvtw/| + fagei/n: 7⁹ o;cloj ) ) ) peribeblhme,noi)

(f) Or^s is not unfrequently without any support but that of cursives. -- 1¹⁶ dexia/| auvtou/ ceiri,: 2¹⁴ o]j evdi,daxen to.n bal): 3⁷ tou/ before Dauei,d: 3¹⁸i[na evgcri,sh|: 5¹³o[sa evsti,n: 6⁹evsfragisme,nwn (for evsfagme,nwn!): 10⁴gra,fh|j with only 205: 11⁷ > kai. o[tan tele,swsin with 617. 920. 2040 arm^2.3: 13⁷po,lemon poih/sai.

(g) Thus every step we have taken proves in an increasing degree the secondary, eclectic and cursive character of the text. It now remains to define the group of cursives with which it is most intimately connected. These are 61 (xvi cent.) and 69 (xv. cent.). With these cursives it agrees against all other authorities in the following: 4⁵ kai. (for a[ evstin%: 4⁸kuklo,qen e;swqen kai. e;xwqen( where 61. 69 have kukl) e;xwqen k) e;swqen -- conflations of kukl) k) e;swqen Aa etc., and kukl) k) e;xwqen 1957. 2050: 11⁵ evkporeu,setai: 13⁵ polemh/sai (instead of poih/sai): 13¹⁵ avpoktanqh/nai (instead of i[na ) ) ) avpoktanqw/sin). In 3¹⁸ with 69 alone Or^s reads fanh/| for fanerw,qh|.
 

{clxxviii}§ 11. Some account of the Versions.

(i.) Latin Versions: (a) Tyconius; (b) Primasius; (c) Codex Floriacensis (= fl); (d) Codex Gigas (= gig); (e) Vulgate.

(a) Tyconius. -- There is no critical edition of this text. Dr. Prinz has such a text in preparation. The readings in the Appar. Crit. of the present work are taken from Professor Souter's "Tyconius' Text of the Apocalypse, a partial restoration," J.T.S., April 1913.

(b)  Primasius (= Pr). -- Haussleiter has published a critical edition of Primasius' text in his work, Die lateinische Apocalypse, 1891, pp. 80-175.

(c) Codex Floriacensis (= fl). -- Only fragments of this Latin version made in Africa survive. These amount to 61 verses: 1-2¹, 8⁷-9¹², 11¹⁶-14¹⁴, 14¹⁵-16⁵. They are preserved in a palimpsest in the National Library of Paris -- No. 6400 G (formerly in the library of Fleury). This palimpsest has been deciphered and published by Vansittart, Journal of Philology, vi. (1872) pp. 219-222; Omont, Bibliotheque de l'ecole des chartes, xliv. (1883) pp. 445-451, Belsheim in 1887; Berger, Le palimpseste du Fleury, 1889; Haussleiter in his edition of Primasius, 1891, and a recent collation in 1906, J.T.S. p. 96 sqq.

Pr and fl render mutual service to each other. They make the detection of intrustions of vg in one or other of these two versions an easy task. The canon of criticism here is that where Pr and fl differ, such variants as agree with vg are to be rejected and the remainder to be retained as the older text.

(d) Codex gigas (= gig). -- This codex of the xiii cent., formerly in Prague, is now in Stockholm. It contains the whole Bible, but only Acts and the Apocalypse are Old Latin. This codex was edited by Belsheim in 1879, but inaccurately. For the collation used in the present work I am indebted to Professor White, who has put at my service the fresh collation made by Dr. Karlsson in 1891 for John Wordsworth, bishop of Salisbury. It appears to have an Italian character (Gregory).

(e) Vulgate (= vg). -- I have used Professor White's Editio Minor of the Vulgate -- Novum Testamentum Latine, Clarendon Press, 1911. In this edition the following seven MSS (vg^{a.c.d.f.g.h.v}) are used:
 

a -- Amiatinus (vii-viii) cent.	g -- Sangermanensis (ix).
c -- Cavensis (ix).	h -- Hubertianus (ix-x).
d -- Armachanus (812 A.D.).	v -- Vallicellanus (ix).
f -- Fuldensis (vi).

ii. Syriac Versions: (a) Philoxenian, (b) Harkleian or Syriac Vulgate.

{clxxix} (a) Philoxenian (= s¹). This version was discovered and edited by Professor Gwynn in 1897. He ascribes it on good grounds to the 6th century. It is perhaps the most valuable of all the versions, its only rival being arm⁴ (see p. clxvi sqq.). It is remarkable that with the Armenian versions it has many readings in common with the Latin versions (see Gwynn, p. cxliii), where these differ from all Greek MSS (though the list is not quite correct). Thus in 5⁴ s¹ arm¹ Pr read lu/sai ta.j sfragi/daj auvtou/ for ble,pein auvto,: in 13¹⁰ s¹ gig sa eth read evn macai,ra| avpoktanqh,setai: in 9¹⁷ s¹ Tyc Pr gig vg arm^1.2.3.a read tou/ sto,matoj; but this is found in one Greek cursive -- 35. The presence of a common Latin (?) element in s¹ arm sa eth calls for investigation. Most of this element, no doubt, goes back to lost Greek MSS, but there appears to be a residuum of Latin readings which made their way into s¹ arm and other versions.

s¹ exhibits conflations in 5¹⁰ 6² 11¹¹ 18¹⁷ o` evpi. tw/n ploi,wn evpi. to,pon ple,wn) 

Gwynn puts forward two hypotheses to account for the form of the text of s¹. The translator formed the text for himself, taking as basis our main exemplar, but modifying it to the extent of about one-third by the introduction of readings from a secondary subsidiary examplar. Otherwise he followed a single exemplar in which the primary and secondary factors stood to each other in the ratio of two to one.

(b) The Harkleian (= s²). -- This version was made about 616. As yet no critical edition of the text has appeared. It preserves very ancient readings lost in most of the Latin versions, but it is decidedly inferior to s¹. See above, p. clxviii, and Gwynn (op. cit.), pp. lxxi-lxxv, lxxxi-lxxxiv.

iii. Armenian Versions. -- The Armenian version was admitted into the Armenian canon in the 12th century through the agency of Nerses. But the Armenian version was known in the earliest years of the 5th century. There are in reality two distinct Armenian versions. The first is exhibited in arm¹, arm², arm³, arm^a, which on the whole form, notwithstanding many differences, a homogeneous whole over against arm4. Arm^1.2.3 represent the sources of the older and unrevised text, and arm^a the Nersesian 12th century recension, which was based on arm^1.2.3 etc. Arm⁴ and arm^1.2.3 represent, according to Conybeare, "two independent renderings of a common Greek text." But this statement needs drastic revision. The Greek source of arm⁴ differed very much from that of arm^1.2.3. Conybeare ascribes arm^1.2.3 to a 5th century text and arm⁴ to a redaction of the early 8th.

As in the case of s¹, so here the Latin element is evident. In 19¹ arm² this influence is undeniable. Thus, where the {clxxx} Greek has o;clou pollou/( vg^a.c.v have tubarum multarum, and so arm². This corruption could only have arisen in Latin, i.e. tubarum corrupt for turbarum. The same corruption reappears in 19⁶, where o;clou pollou/ is rendered by Pr vg^a.c.d.f.v by tubarum (-ae, -vg) magnarum (-nae vg).

Conybeare thinks that the early Armenian version "was made from an old Latin copy, or perhaps from a bilingual Greco-Latin codex." The latter appears the more probable, but the question requires thorough investigation, not only in regard to arm, but also in regard to s¹ bo sa and eth.

It is much to be regretted that Conybeare did not print in its entirety arm4 alongside arm^1.2.3.a, seeing that it represents a more ancient type of Greek text than arm^1.2.3.a. Arm⁴ is alone complete, and yet neither is its text nor even a single variant from it given in Armenian. Only English renderings of the variants and of 16¹⁷-19¹⁸ are supplied. It is rather strange for a scholar , who is editing both a text and a translation, to translate two chapters (16¹⁷-19¹⁸) from a text which he does not give, and print a text (arm²) of these chapters, which he does not translate save in the case of its variants. For the text of arm⁴ he refers his readers to Dr. F. Murat's edition of it "in the great university libraries of our country," or "to the Armenian Convent of St. James in Jerusalem."

Students of the J^ap cannot be other than most grateful to Dr. Conybeare for his edition of the Armenian version, but it does not bear the character of a final one.

(d) Bohairic Version (= bo). -- The Bohairc (or Memphitic) version has been edited with great care by the Rev. G. Horner. This editor prints J^ap from the Curzon MS 128 with variants from other MSS. He has provided an English version of this MS, but unfortunately the variants are not translated. The result is that the reader who does not know Bohairic cannot get to know anything beyond MS Curzon 128.

(e) Sahidic Version (= sa). -- The same scholar is engaged on an edition of the Sahidic. He has most generously supplied the present editor with some hundreds of readings from this fragmentary version. This version appears to agree more with A and its allies than do bo eth.

(f) Ethiopian Version (= eth). -- Only two uncritical editions of this version exist -- that of Platt and that contained in Walton's Polygott. I have used the edition of Platt published in 1899, and only consulted the other version that is printed in Walton's Polyglott.

Bo sa and eth form one group as we have already seen, but their exact relations cannot be determined till critical editions of the three are accessible, and a scholar who has a mastery of the three languages takes the task in hand.

{clxxxii} For the meaning of the above symbols and abbreviations of MSS and versions, see vol. ii. pp. 227 sqq., 234 sqq. For F^1.2.3.4 (i.e. Papyri Fragments), see vol. ii. pp. 447-451.

Though the above table must in many of its features be regarded as purely hypothetical, the editor is convinced of its general accuracy down to Aa F^1.2.3.4; also that, though C belongs to the family of A, it has been influenced by that of a, besides showing signs of frequent correction.

So far the evidence is on the whole clear. Henceforth the relations of the MSS and versions can only be partially and, until several important questions are investigated, provisionally represented. 025 and 046 are certainly descendants of A and a, or of the families of which these are representatives; for 025. 046 preserve primitive readings lost in Aa. Thus in 4⁴ evpi. t) qro,nouj (+ touj 046) ei;kosi te,ssaraj presbute,rouj is undoubtedly right where Aa are wrong and C is defective; for s^1.2 arm^2.3.4.a Pr gig vg bo eth here support 025. 046. In 6⁸ o` qa,natoj of 025. 046 is right, where A is corrupt and Ca wrong. In 9<sup>10</sup> ouvra.j o`moi,aj skorpi,oij of 025. 046 is again right against the greater uncials, and also in 19¹⁸ tw/n kaqhme,nwn evpV auvtw/n) This fact cannot be represented in the above table.

Further, a study of 025. 046 shows that these two MSS are connected; for they have 36 (more or less) readings in common against AaC. This connection is accordingly represented in the above table. But 025 and 046 are related differently to A and a. Moreover, 025 shows signs of a deliberate recension, whereas 046 exhibits rather signs of a progressive correction. But these MSS have other connections. Thus in 14¹⁸ 025 unites with C in reading kraugh/| (a wrong reading) against fwnh/| of Aa 046: in 14¹³ in reading evn kuri,w| of all other MSS. This connection is represented in the above table.

Certain cursives, i.e. 35. 205. 2040 (11⁸-20¹¹ only). 2050 preserve some original readings lost wholly in a 025. 046 (see clxxiii sqq.). These cursives are in many respects as valuable as the later uncials, while in a few they are superior.

Of the remaining cursives a considerable number follow for the most part 025, while the main body appears to follow 046. But the exact differentiation of these cursives has not yet been investigated.

Turning from the Greek MSS to the versions, we enter on a more difficult task. Of the versions, Tyc sa eth and s² have not yet been critically edited. All the materials for such a critical edition of bo are given in Horner's edition of the Bohairic N.T., but they are accessible only to Coptic scholars. The internal relations of the Latin versions Tyc Pr fl gig which are still undetermined, {clxxxiii} and likewise the influence of the Latin versions (or of the Greek MSS from which a large part of this peculiar (?) Latin element may be derived) on arm s¹ bo eth from attractive problems for future researchers.

Since we know that the Latin versions (or their Greek progenitors) exercised some influence on arm and s¹, I have placed these versions in close connection on the above table. But the Latin influence on bo eth is not represented, nor is s² even mentioned.
 

NOTES
1. 35, but  not 205, adopts the correction of 046 in 3¹², i.e. h] katabai,eni) Some 20 other cursives do likewise.
2. Naturally some points of agreement are found. Cf. the addition with a alp in 1⁸ avrch. kai. te,loj and others, for any MS of J^ap has of necessity many points of contact with every other.
 
 
 

Scanned and edited by Brad Johnson