put it in a form which will render it at length available for critical purposes.

Of the extant мs L, I have made an entirely new collation. That this was not superfluous the facts will show. The мs was transcribed in the first instance for Ussher, and (so far as I can learn) has only been collated twice for subsequent editions. Of the numberless inaccuracies of the transcript from which Ussher derived all his knowledge of this Ms I have already spoken. Moreover he has not (except in a very few instances) distinguished the respective readings of the two MSS which he employed. And lastly, his printed text contains several readings which are not found in either, and which (in some instances at least) have slipped in through mere inadvertence.

Of the subsequent collations the earlier was made by T. Smith for the text which accompanied his edition of Pearson's notes (A.D. 1709). After describing the мs in his preface, he adds, 'quem ego quoque ea qua potui accuratione contuli, correctis illius [collationis?], cujus opera usus est D. Usserius, aberrationibus.' The result is a much better text of this Latin version than Ussher's; but for critical purposes his collation is quite inadequate. He has not recorded a quarter of the various readings of L,. Though he has corrected some of Ussher's worst mistakes, he has sometimes given readings for which there is no authority either in the MS or in Ussher's printed text; e.g. Smyrn. 3 carne ipsius et spiritu' for 'carni ipsius et spiritui,' and Smyrn. 6 'qualitate' for 'qualiter'; in neither instance giving any various reading, and in the latter distinctly stating that this is the rendering of the Latin translation.

The second collation to which I referred was made for Dr Jacobson's edition, and is thus described by him (Patr. Apost. 1. p. xxxvii); 'Hujus codicis lectiones variantes humanissime ad usus meos exscripsit vir reverendus Johannes Jacobus Smith A. M., Coll. Caiensis Socius.' This collation is in many respects more correct than Ussher's transcript, and more complete than T. Smith's collation. But how far it is from being trustworthy, the following list of errors, gathered from the first six chapters of the Epistle to the Smyrnæans alone, will show.

As this collation coincides with the Caius transcript, where it goes most wrong, as for instance in panem qui (§ 5) and qualesque facti sumus dei (§ 6), I suppose the collator must have allowed himself from time to time to consult the transcript instead of endeavouring to decipher the MS itself.

MS.

These two collations moreover, inadequate as they are in themselves, were confined to the seven epistles mentioned by Eusebius. The text of the other epistles has remained in the same state in which it was left by Ussher, without any fresh examination of the Thus for instance, in Ign. Mar. 2 Ussher accidentally omitted the word 'impellor,' and the omission has been repeated by all subsequent editors, though the sense of the passage is destroyed thereby, and a reference to the мs would at once have supplied the missing word. In some respects the text has even deteriorated since Ussher's time, for later editors have introduced errors of their own. Thus in Hero 3 a whole sentence, 'Saluta deo decens presbyterium,' is omitted in Cureton's text (Corp. Ign. p. 146).

Of the disappearance of the other Ms L, I have spoken in an earlier part of this work, where also I have described Ussher's collation, which is preserved among his books and papers in the Library of Trinity College, Dublin, but has been strangely overlooked by pre

vious editors. By the kindness of the Provost and Fellows I have been allowed the use of this collation, which is indispensable for the criticism of the Latin text; and thus the readings of L, are given in the present edition for the first time. Hitherto they have been left to conjecture, except in the very few passages where Ussher has distinctly mentioned it by name.

This collation however only commences in the middle of the Epistle to Polycarp, § 1 [in] orationibus vaca indesinentibus,' the earlier leaves of the transcript having been lost. For the previous portion, the whole of the Epistle to the Smyrnæans and the commencement of the Epistle to Polycarp, I have supplied the defect by a collation of Ussher's printed text of this version, which I have designated L. As Ussher had only the readings of these two MSS before him, it may be presumed that his printed text, wherever it differs from L,, gives the reading of L,. This rule however can only be accepted as roughly and approximately true. Large allowance must be made for inadvertences and inaccuracies. For instance, in Smyrn. 9 Ussher omits 'Bene habet et Deum et episcopum cognoscere,' and possibly these words may have been wanting in L,; but, when we find him leaving out whole clauses elsewhere, where we are able by means of his own collation to convict him of inaccuracy, e. g. Polyc. § 2 'ut gubernatores ventos,' and Ephes. I 'ut potiri possim discipulus esse' (not to multiply examples), the inference will appear highly precarious'.

Where a reading of this мs is distinctly given by Ussher in this collation, it is marked L, simply; where it is only inferred from his silence, i.e. where he has not noted any divergence from the reading which he had before him in the transcript of L,, it is given as L,s.

In the following recension I have endeavoured to restore the text of the version to the condition in which it left the translator's hands. Thus I have not scrupled to make an alteration here and there, where the Latin text itself had obviously been corrupted in the course of transmission. Thus, for instance, I have cast out two apparent glosses, Ephes. 1, Magn. 2. Thus again in four passages, Smyrn. 8, Ephes. 3, Magn. 13, Philad. 3, I have substituted 'episcopi,' 'episcopo,' for 'ipsi,' 'ipso,' the corruption having arisen from an easy confusion of the Latin contractions, ipi, epi, ipo, epo, and the Greek

1 The first of these two omissions is supplied by Ussher in his table of emendunda, p. 241, but inaccurately, ut

IGN.

gubernares ventos': the second is not mentioned at all.

39

text in all these cases deciding the true reading. On the same principle in Mart. 4 I have substituted 'ipsi' (ipi) for 'Christi' (xpi). So too I have not scrupled to alter 'assensores' into 'assessores' Polyc. 6, 'sollicitudine' into 'insollicitudine' Polyc. 7, 'potiri' into 'per potiri' Ephes. 1, 'salvificemini' into 'salificemini,' 'exacuens' into 'exacescens,' 'perfari' into 'profari' Magn. 10, 'suadeo' into 'suadeor' Trall. 3, 'præcipue' into 'præcipio' Rom. 4, 'fidei' into 'Dei' Philad. 9, 'sciant' into 'sitiant' Ign. Mar. 1, and (having regard to the contractions) 'deo' into 'dicit' Tars. 7; in all which cases the corruption was easy in the Latin text and quite impossible in the Greek. On the other hand I have not attempted to correct those errors which must be traced to the faulty Greek text which the translator had before him. For example, in Mar. Ign. 1 'et Sobelum' is left; for though there can be little doubt that the correct reading is Kaooóßnλον οι Κασσοβήλων, it is equally clear that the Latin translator had καὶ Σόβηλον in his text.

In recording the variations of the MSS I have not (except in special cases and for particular reasons) included readings which are corrected prima manu. Nor again is any account generally taken of the punctuation of the MSS, which is arbitrary and valueless. The marginal glosses and notes moreover, of which a very few occur in L,, and which are frequent in L,, are not recorded, unless they have a bearing on the reading. Some of these, which have an interest of a different kind, are given in an earlier part of this work.

On the orthography of the MSS one or two points require explanation. In L, the diphthongs, a, a, are systematically disregarded and written e (e. g. eterne, penitet); and in this same мs ci is universally, or all but universally, written for ti (e. g. Ignacius, propiciacio). In both these cases the normal spelling is silently adopted. In other instances, where L, persistently departs from the normal orthography (e. g. misterium, carisma, ammonere), I have contented myself with noticing the fact at the first occurrence of the word.

Much error has arisen in previous collations from inattention to the contractions. Thus for instance, quando, quoniam, quum, qui, quæ, quem, quia, etc. have been confused; and again, ergo, igitur; and again, tamen, tantum. In this way various readings have been erroneously multiplied. In most cases there can be no doubt as to the force of the contraction. In some few instances, where a contraction in L, is ambiguous, I have given it the. interpretation which accords with the Greek text or with the reading of L..

It did not seem necessary to encumber the notes by pointing out

every instance where previous collators have misread L,. In one or two cases I have done so, because the error was sufficiently important to call for notice, e. g. Trall. 5 'scire celestia' for 'supercelestia,' Polyc. 7 in oratione' for 'in resurrectione'; but these are exceptional. As I have had the latest collation (that which is given in Jacobson's edition) constantly before me, the variations recorded in it and not noted by me have been deliberately rejected. Thus for instance the various readings, 'panem qui' for 'passionem quæ' Smyrn. 5 (see above, p. 592), 'optimum' for 'opportunum' Trall. 2, with many others, have disappeared.

On the other hand, some readings will appear in my text (on the authority of one or both of the MSS) for the first time; and in most instances these bring the Latin into stricter accordance with the Greek than it is in the text of the printed copies. Thus for instance, 'veneremini' (èvrρéreo0e) for veneremur' Magn. 6, 'ipsos' (avroús) for 'ipsas' Philad. 3, 'apponi' (#poσbeîvai) for 'opponi' Ign. Mar. 4, 'portus' (λuévas) for 'Portum' Mart. 5, ‘immunda inani gloria' (rŷs ákaláρrov piλoriuías) for 'mundi inani gloria' ib., are read by both L, and L. And again, in Rom. 7 'adjuvet; ipsi autem magis mei fiatis', and Mart. 5 'da ea quæ a nobis futura separatione; justo autem fieri ipsi secundum votum accidit,' the readings of L,, involving in both cases a transposition, produce exact conformity to the Greek. The text, thus restored, is 'adjuvet ipsi; magis autem mei fiatis' (ẞoŋleitw αὐτῷ· μᾶλλον ἐμοὶ γίνεσθε) in the one passage, and de ea quæ a nobis futura separatione justi fieri; ipsi autem secundum votum accidit' (T ἀφ ̓ ἡμῶν μέλλοντι χωρισμῷ τοῦ δικαίου γίνεσθαι· τῷ δὲ κατ ̓ εὐχὴν ἀπέβαινεν).

The correspondence with the Virgin and S. John, forming the second part of this collection, is comparatively unimportant. It is found in a considerable number of MSS besides L, L,; sometimes by itself, sometimes in connexion with the epistles of the Long Recension. In this latter case it sometimes precedes the twelve epistles of this Recension (e. g. Flor. Laur. xxiii. 20, Palat. 150, Oxon. Magd. lxxvi), and sometimes follows them (e. g. Bruxell. 20132). The various readings are very numerous, and the order of the four epistles is different in different copies.

For the sake of exhibiting the character of the variations, I have given a collation of three Oxford MSS besides the readings of L, L, taking the editio princeps (Paris, 1495) as the basis of my text.