CHAPTER V.

TEXTUAL CONDITION OF THE LXX., AND PROBLEMS ARISING OUT OF IT.

1. WHEN the work of the Seventy-two had been accomplished, the Jews of Alexandria (so the legend goes) were bidden to invoke curses, after their manner, upon any who should dare to add to the version or take from it, or alter it in other ways (Aristeas ad fin.: ἐκέλευσαν διαράσασθαι, καθὼς ἔθος αὐτοῖς ἐστιν, εἴ τις διασκευάσει προστιθεὶς ἢ μεταφέρων τι τὸ σύνολον τῶν γεγραμμένων ἢ ποιούμενος ἀφαίρεσιν). The imprecation, it has been acutely observed, may point to an early deterioration of the text of the Greek Pentateuch, which the Pseudo-Aristeas desired to check. This inference is insecure, for the story is sufficiently explained by a reference to such passages as Deut. iv. 2, xii. 32⁹⁵⁹⁹⁵⁹Cf. Apoc. xxii. 18 f.; but it is certain that textual corruption began before the Christian era. There are traces of it in the writings of Philo, which cannot be due to blunders in Philo's own text.

E.g. in quis rer. div. her. 56 Philo quotes Gen. xv. 15 in the form now universal in MSS. of the LXX. (μετ᾿ εἰρήνης τραφεὶς ἐν γήρει καλῷ), adding the comment: οὐκοῦν . . . τὸ τέλειον γένος . . . εἰρήνῃ καὶ ἐλευθερίᾳ βεβαιοτάτῃ ἐντρεφόμενον κτλ. This is perhaps the most convincing example, but we may add Gen. xvi. 14 Βαράδ = ἐν κακοῖς (de fug. 38), i.e. Βαράκ (Luc.); xxi. 6 οὐ χαριεῖταί μοι (de mut. nom. 24, where however, as in legg. all. ii. 21, iii. 78, quod det. pot. insid. sol. 33, Cohn and Wendland read συγχ. μοι with cod. A ^phil); Exod. xvii. 6 ἕστηκα πρὸ τοῦ σε ἐπὶ τῆς πέτρας ἐγχωρεῖν (de somn. ii. 32, cf. B πρὸ τοῦ σε⁹⁶⁰⁹⁶⁰Thackeray, however, points out that this may not be textual corruption; cf. Lev. xviii. 30 πρὸ τοῦ ὑμᾶς, Numb. xiii. 23 πρὸ τοῦ Τάνιν Αἰγύπτου. . . . ἐν 479Χωρήβ, AF πρὸ τοῦ ἐλθεῖν . . . ἐν Χ.). Similar corruptions probably exist in some of the N.T. citations, e.g. σῷμα⁹⁶¹⁹⁶¹As in all our MSS. of Ps. xxxix. in Heb. x. 5 (Ps. xxxix. = xl. 7), and ἐνοχλῇ⁹⁶²⁹⁶²See codd. BAF in Deut. l.c.. for ἐν χολῇ in Heb. xii. 15 (Deut. xxix. 18 (17)). It may be added that double renderings already appear in Philo. E.g. in citing Deut. xix. 14 his MSS. give οἱ πατέρες σου (B) in de post. Caini 25, but οἱ πρότεροί σου (A) in de justitia 3.

Justin, as we have seen⁹⁶³⁹⁶³Above, p. 424., charges his Jewish contemporaries with the deliberate excision of numerous passages in the LXX. which were favourable to their Christian antagonists (dial. 71 πολλὰς γραφὰς τέλεον περιεῖλον ἀπὸ τῶν ἐξηγήσεων τῶν γεγενημένων ὑπὸ τῶν παρὰ Πτολεμαίῳ γεγενημένων πρεσβυτέρων)⁹⁶⁴⁹⁶⁴Cf. dial. 120; Iren. iii. 21. 1, 5; Eus. dem. ev. vi. p. 257 c, d.. But of the four passages produced in proof of his assertion three are mere glosses, probably of Christian origin; while the fourth, a genuine part of the book of Jeremiah (xi. 19), is now found in all MSS. of the LXX. The charge, though made in good faith, seems to have rested on no better foundation than a natural distrust of the Jews, who in Justin's time were active and bitter opponents of the Church. It is equally improbable that the Greek O.T. was wilfully interpolated by Christians, or that, if they attempted this, the existing text has been affected by it to any appreciable extent. A few traces may be found of the accidental influence of N. T. citations, e.g. the interpolation in Ps. xiii. 3, and perhaps also the reading σῶμα in Ps. xxxix.; but apart from these, the Septuagint, during the first two centuries after Christ, suffered little from Christian hands beyond errors of transcription. What Dr Hort has written in reference to the N.T. is doubtless true also of the LXX.: "accusations of wilful tampering with the text are . . . 480not unfrequent in Christian antiquity . . . but with a single exception, wherever they can be verified they prove to be groundless, being in fact hasty and unjust inferences from mere diversities of inherited text⁹⁶⁵⁹⁶⁵Intr. to N. T. in Greek, p. 283. The one exception which Dr Hort mentions in connexion with the N.T., the excision practised by Marcion, finds no parallel in the Christian history of the Greek O.T.."

Accidental corruptions⁹⁶⁶⁹⁶⁶A good example of corruption in the Greek is to be found in Num. iii. 24, where all Greek MSS. and the O.L. (Lyons Pentateuch) read Δαήλ Dael for Λαήλ (לאל). The name of Joshua's father in the LXX. is Ναυή (0. L. Nave), probably in the first instance an error for Ναύν (ΝΑΥΗ for ΝΑΥΝ) = נון). Another well-known instance is the A text of Jud. v. 8 σκέπη νεανίδων σιρομαστῶν ἀνήφθη καὶ σιρομάστης, which, as Ewald pointed out, conceals the doublet (1) σκέπην ἐὰν ἴδω καὶ σιρομάστην, (2) σκέπην ἐὰν ὀφθῇ καὶ σιρομάστης. In 1 Esdr. v. 34 Σαφάγ B is an orthographical error (cf. A)., however, and variations of reading and rendering grew apace, and in the third century Origen complains of the uncertainty of the Biblical text in both its parts⁹⁶⁷⁹⁶⁷Though he is referring especially to MSS. of the N.T. his next words shew that the remark is meant to include the LXX.: τὴν μὲν οὖν ἐν τοῖς ἀντιγράφοις τῆς παλαιᾶς διαθήκης διαφωνίαν κτλ.. (see, for the rest, above, p. 60). (comm. in Matt. t. xv. 14 δηλονότι πολλὴ γέγονεν ἡ τῶν ἀντιγράφων διαφορά, εἴτε ἀπὸ ῥαθυμίας τινῶν γραφέων εἴτε ἀπὸ τόλμης τινῶν μοχθηρᾶς τῆς διορθώσεως τῶν γραφομένων εἴτε καὶ ἀπὸ τῶν τὰ ἑαυτοῖς δοκοῦντα ἐν τῇ διορθώσει προστιθέντων ἢ ἀφαιρούντων⁹⁶⁸⁹⁶⁸The gravest instance of ἀφαίρεσις was found in the book of Job; see above, p. 255.). Besides intentional changes he notices elsewhere (1) double renderings: hom. in 1 Regn., i. 4 "non me latet . . . quod in aliquibus exemplaribus habetur erat vir quidam ἄνθρωπός τις ἦν, codd. M, 44, &c.), sed in his exemplaribus quae emendatiora probavimus ita habetur, erat vir unus (A, ἐγένετο ἄνθρωπος εἷς)"; (2) transpositions: on Jer. x1vii. 4 he has the note ἡ τῶν οʹ ἔν τισι τόποις μετατεθεῖσα ὥστε τὰ πρῶτα ὕστερα καὶ τὰ ὕστερα πρῶτα γενέσθαι; (3) errors of transcription: in Jer. xv. 10, where most of his copies read, as ours do now, ὠφέλησα, ὠφέλησεν, he maintains that this reading is a γραφικὸη 481ἁμάρτημα for ὠφείλησεν. Such faults were specially common in the case of proper names: in Joann. t. vi. 41 τὸ δ᾿ ὅμοιοων⁹⁶⁹⁹⁶⁹In the context Origen refers to the apparent confusion of Γάδαρα and Γέργεσα in the Gospels. περὶ τὰ ὀνόματα σφάλμα πολλαχοῦ τοῦ νόμου καὶ τῶν προφητῶν ἔστιν ἰδεῖν, ὡς ἡκριβώσαμεν ἀπὸ Ἐβραίων μαθόντες, καὶ τοῖς ἀντιγράφοις αὐτῶν τὰ ἡμέτερα συγκρίναντες.

In these criticisms Origen makes no attempt to distinguish between supposed errors which are properly textual, and those which belong to the translation itself. His sole criterion of error was divergence from the official Hebrew, and he assumed that all divergences were textual only, the translation having been originally exact. Nevertheless there can be little doubt that in the course of four centuries many actual corruptions such as he describes must have accumulated in the MSS. of the LXX. The κοινὴ ἔκδοσις⁹⁷⁰⁹⁷⁰Ἡ κοινὴ ἔκδοσις was also used of the LXX. as compared with the Hebrew text and the other Greek versions: see Nestle in Hastings, D.B. iv. 438., as the uncorrected MSS. were called, needed revision, and the literary activity of the third century endeavoured to supply it. At Caesarea in Palestine, at Antioch, in Egypt, independent attempts were made to restore the Septuagint to its primitive purity. But the remedies which were adopted unhappily increased the disease. "The Hexapla, from its very nature, encouraged the formation of mixed texts⁹⁷¹⁹⁷¹Driver, Samuel, p. x1vii."; the Hexaplaric recension, divorced from the rest of the work, accentuated this tendency, and the other recensions had a similar effect, although they aimed at the simpler task of correcting the errors of the κοινή,

2. Of the Hexaplaric, Lucianic, and Hesychian recensions some account has been given already⁹⁷²⁹⁷²See above, Part I. c. iii.. In this place we have only to consider how far it is possible to employ them in the criticism of the text. Their importance to the critic of the LXX. lies in the fact that they were based upon copies of the κοινή, as it was read in Palestine, Syria, and Egypt during the 482third century. But in order to recover from them this unrevised text, two preliminary tasks have to be undertaken. The recensions themselves must first, as far as possible, be restored from existing materials, and we must then proceed to eliminate from them such elements as are recensional, or are due to the reviser's hand.

As to the first of these processes, the materials from which it is proposed to recover the recensions are fairly abundant and varied, but there is much uncertainty as to the attribution of some of them; whilst others present a particular recension only in certain books or portions of books, or with more or less of mixture. The principal authorities for each recension have already been mentioned, but it may be well to collect them here in a compact form.

Hexaplaric⁹⁷³⁹⁷³For fuller information see pp. 78, 112 ff., 118 ff., 137 f., 140, 148 ff.. Codd. G, M, Q; 15, 22, 38, 58, 72, 86, 88, 135, 137, 138, 139, 161, 248, 249, 250, 252, 255, 256, 258, 259, 264, 268, 273; Paris Nat. Reg. gr. 129, 131, 132, Ars. 8415, Escurial Σ. t. 16, Leipzig gr. 361, Zurich c. ii, Athos Vatop. 516, Pantocr. 24, Protaton. 53, Laur. γ. 112. Versions: Sahidic (in part), Armenian (in part), Syro-hexaplar.

Lucianic⁹⁷⁴⁹⁷⁴See pp. 82 ff., 93, 116 ff., 148 ff.. Codd. 19, 12, 36, 48, 51, 62, 82, 90, 93, 95, 108, 118, 144, 147, 153, 185, 231, 233, 245, 308; Parts Coisl. gr. 184, Athens bibl. nat. 44. Versions: Old Latin, Philoxenian Syriac, Gothic, Armenian (m part), Slavonic. Fathers: Chrysostom, and other writers of the School of Antioch⁹⁷⁵⁹⁷⁵Lagarde would add (Ankündigung, p. 27) the writings of the Emperor Julian..

Hesychian⁹⁷⁶⁹⁷⁶See pp. 80, 107 ff., 145, 148 ff., and on the recensions generally cf. Ceriani in Rendiconti d. R. Ist. Lomb. (18 Feb. 1886).. Codd. Q, 26, 44, 49, 68, 74, 76, 84, 87, 90, 91, 106, 107, 134, 198, 228⁹⁷⁷⁹⁷⁷228, and 238 to some extent, fluctuate between Luc. and Hes.; see Oesterley, Amos, p. 19 f., 238, 306. Paris suppl. gr. 609. Versions: Bohairic, Armenian (in part). Cyril of Alexandria; other Egyptian writers.

The fragments of the Hexapla have been collected by the labours of a succession of scholars such as P. Morinus, Drusius, Montfaucon, and especially Field, in whose Origenis Hexaplorum quae supersunt may be found all the remains of 483Origen's works which were available in 1875. These editions do not aim at restoring the text of the Hexaplaric LXX. in a connected form. Such a restoration, however, has been attempted in the case of Lucian's recension by Lagarde⁹⁷⁸⁹⁷⁸See above, p. 83 f., who desired to see a similar work accomplished for the recension of Hesychius, and an edition in which the two texts should appear facing one another on opposite pages. When this had been done, he proposed (1) to eliminate from these any Hexaplaric matter, by comparing them with the fragments of Aquila, Symmachus, and Theodotion; and (2) to collect the readings which departed most widely from the M. T. By this process he hoped that a point of departure would be reached from which the reconstruction of the LXX. might begin⁹⁷⁹⁹⁷⁹An earlier scheme is set forth in Genesis Graece, p. 21: "primum molior librum a codicum uncialium qui hexaplares non sunt . . . consensu haud raro certa coniectura emendando edendum . . . deinceps propositum est . . . editionem hexaplarem curare . . . tertio loco . . . adparatum criticum integrum adiungere cogito.".

This scheme is worthy of the great scholar who initiated it, and it was the first serious effort to grasp the problem of scientific reconstruction. But its progress has been checked and perhaps finally stopped by its author's premature death, and its successful accomplishment under any circumstances was at least problematical. So long as no MS. or version presents an unmixed text of either Lucian or Hesychius, and much uncertainty remains as to the exact sources from which they are to be recovered, restorations of this kind cannot be regarded as more than tentative or provisional. Meanwhile, such attempts are not free from danger. Since the publication of Lagarde's edition, there has been a tendency on the part of Biblical students to cite it as 'Lucian,' without reserve. Lagarde himself is careful not to claim finality for his work; he describes it as "editionem . . . in gravioribus omnibus satis fidam," and looks forward to a more exact 484representation of Lucian's text: "conlatis codicibus versionibusque eam praebentibus et patrum ea utentibus excussis efficiendum erit ut etiam in minutioribus adcurate edita dici merito possit⁹⁸⁰⁹⁸⁰Praef. xv.." But this hope has not been fulfilled, and an edition of Lucian which falls short of exactness in smaller details cannot be directly used for the critical editing of the LXX. It has rendered valuable services in other departments of Biblical study, exhibiting sufficiently the characteristics of this recension, and repeatedly offering, especially in the four books of Kingdoms, renderings of a Hebrew text distinct from ⁹⁸¹⁹⁸¹See Driver, Samuel, pp. lii. f., lviii.: I. Hooykas, Iets over de grieksche vertaling, van het 0. T., p. 12 ff.. But in the delicate task of reconstructing the Greek text, recourse must be had to the actual evidence which lies behind Lagarde's work. For this purpose it would seem to be more important to provide texts based upon groups of MSS., somewhat after the manner of the Collection of four important MSS. (the Ferrar-group) published by Dr T. K. Abbott. Doubtless such groups would mainly follow the lines of the ancient recensions, but the identification would not be complete, and the student would have before him not only the general result, but the whole of the evidence upon which it was based.

3. Perhaps a more lasting service was rendered to the textual criticism of the Septuagint by the axioms and principles which Lagarde's long study of the problem enabled him to lay down for the guidance of the student and the future editors. His early book Anmerkungen zur griechischen Übersetzung der Proverbien (1863) starts with the following axioms: (1) Since the MSS. of the LXX. are all directly or indirectly the result of an eclectic process, any attempt to restore the original text must also proceed on eclectic principles; and the critic must chiefly depend upon (a) his acquaintance with the style of the 485several translators and (b) his faculty of referring readings to a Semitic original or, when they are not of Semitic origin, recognising them as corruptions of the Greek archetype. (2) Where the critic has to make choice between two readings, he will do well to prefer (a) a free translation to one which is slavishly exact, and (b) a translation based upon another Hebrew text to one which represents the M. T. In the preface to his Lucianic Septuagint, published twenty years later, three principles are asserted: (1) A critical text of the Greek O. T. cannot be based on the authority of any one MS. or without regard to the grouping of MSS.; (2) the restoration of the text common to any one family must not be regarded as more than a step forward in the right direction; (3) even a critical text, when reached by these or other means, will not be free from the element of uncertainty.

Lagarde's own words are as follows: Anmerkungen, p. 3: "nur drei axiome schicke ich voraus: I. die manuscripte der griechischen übersetzung des alten testaments sind alle entweder unmittelbar oder mittelbar das resultat eines eklektischen verfahrens: darum muss, wer den echten text wiederfinden will, ebenfalls eklektiker sein. Sein maasstab kann nur die kenntniss des styles der einzelnen übersetzer, sein haupthilfsmittel muss die fähigkeit sein, die ihm vorkommenden lesarten auf ihr semitisches original zurückzuführen oder aber als originalgriechische verderbnisse zu erkennen. II. wenn ein vers oder verstheil in einer freien und in einer sklavisch treuen übertragung vorliegt, gilt die erstere als die echte. III. wenn sich zwei lesarten nebeneinander finden, von denen die eine den masoretischen text ausdrückt, die andre nur aus einer von ihm abweichenden urschrift erklärt werden kann, so ist die letztere für ursprünglich zu halten." Libr. V. T. can. i. p. xvi.: "tenenda tria esse aio: [1] editionem veteris testamenti graeci curari non posse ad unius alicuius codicis auctoritatem, sed conlatis integris codicum familiis esse curandam: nam familiis non accedere auctoritatem a codicibus, sed codicibus a familiis: [2] unius alicuius familiae editionem nihil esse nisi procedendi ulterius adminiculum: [3] errare qui si quando ipsa manus veterum interpretum inventa sit, in ea legenda adquiesci debere perhibeant, quum conlatis vetera emendandi periculis omnibus indagandum sit quae explicationis veteris testamenti per quatuor saecula fata 486fuerint, ut tandem aliquando pateat quam incerta in hoc literarum genere omnia sint, et quam multa nulla alia re nisi coniectura nitantur sciolorum, superstitiosorum, desperantium."

4. These principles have been stated at length, because they are fruitful in themselves, and they mark an important step in the progress of LXX. textual criticism. But it is obvious that they do not form a complete and coherent code of critical canons. Indeed, Lagarde's later axioms to some extent limit and correct the earlier, for the recognition of the principle of grouping the MSS. and taking their evidence according to families evidently serves as a check upon the extreme eclecticism recommended in the first axiom of 1863. Nevertheless the series forms an excellent starting-point for a brief discussion of the problems which lie before the future critical editor of the LXX. and the principles by which he must be guided.

By a singular accident the first two printed editions of the Greek Old Testament exhibit on the whole the Lucianic and Hesychian texts respectively⁹⁸²⁹⁸²Cornill, Ezechiel, p. 79: "ein wunderbar glücklicher Zufall hätte uns somit in der Aldine im Grossen and Ganzen den Hesych gegeben, wie die Complutensis im Grossen and Ganzen den Lucian darstellt.", whilst the Roman edition of 1587 and the Oxford edition of 1707—20 are roughly representative of the two great uncial codices, B and A. Thus the earlier editors anticipated, though imperfectly and (in the case of the Complutensian and Aldine Septuagints) unwittingly, the two methods of editing the Greek O. T. which are still in use. Of the advantages and disadvantages of the recensional method, enough has been said. The other, which consists of printing the text of a single MS., with or without an apparatus criticus, is clearly desirable only in the case of a MS. which sufficiently represents an important type of text, and may thus be profitably used as a standard of comparison. Such are the two great uncials already mentioned.

Cod. B, as was pointed out by Dr Hort⁹⁸³⁹⁸³See O.T. in Greek, p. xi. f., "on the whole 487presents the version of the Septuagint in its relatively oldest form." Taken as a whole, it is neutral in its relation to the recensions of the third and fourth centuries; its text is neither predominantly Lucianic nor Hesychian⁹⁸⁴⁹⁸⁴This however has been doubted; see Nestle, Introd. to the Textual Criticism of the N. T., pp. 61 f., 183 f. nor Hexaplaric. Cornill, indeed, was at one time led by certain appearances in the B text of Ezekiel to believe that in that prophet at least the scribe of B had extracted his text from the fifth column of the Hexapla, or rather, from the edition of Eusebius and Pamphilus⁹⁸⁵⁹⁸⁵See his Ezechiel, pp. 84, 95. The theory was suggested by an early hypothesis of Lagarde (Anmerkungen, p. 3) that the text of B was extracted from a glossed codex.. Lagarde, however, at once pointed out the difficulties which beset Cornill's theory⁹⁸⁶⁹⁸⁶In Gött. gelehrte Anzeigen, 1886 (reprinted in Mittheilungen, ii. p. 49 ff.)., and Hort, in a letter to the Academy (Dec. 24, 1887), dismissed it with the remark, "What Cornill does seem to me to have proved is that in Ezekiel B and the LXX. text of the Hexapla have an element in common at variance with most other texts"; adding, "The facts suggest that B in the Septuagint was copied from a MS. or MSS. partially akin in text to the MS. or MSS. from which Origen took the fundamental text for the LXX. column of his Hexapla⁹⁸⁷⁹⁸⁷On the provenance of B and א see Hort, Intr.², p 264 ff., Harris, Stichometry, p. 71 ff., Robinson, Euthaliana, p. 42 ff., and the summary in Kenyon, Our Bible and the Ancient MSS., p. 128.." Eventually Cornill withdrew his suggestion, observing that the forms of the proper names in B shew no sign of having been influenced by Origen's corrections⁹⁸⁸⁹⁸⁸Gött. gelehrte Nachrichten, xxx. (1888, 8, p. 194 ff)..

If we accept Dr Hort's view, which at present holds the field, the Vatican MS. in the O. T. as a whole carries us back to the third century text known to Origen, and possibly to one much earlier. In other words, not only is the Vatican MS. our oldest MS. of the Greek Bible, but it contains, speaking quite generally, the oldest text. But it would be an error to suppose that this is true in regard to every context or even every book, 488and a still graver error to treat the text of B as necessarily representing everywhere the original Septuagint. As Mr Burkitt has pointed out⁹⁸⁹⁹⁸⁹Tyconius, p. cxvii., "the O. L. and the Hexaplar text convict B here and there of interpolation, especially in Isaiah." "Certainly (he writes in another place⁹⁹⁰⁹⁹⁰Aquila, p. 19.) in the books of Kings it is free from some of the gross interpolations which have befallen most other MSS. But it cannot claim to transmit to us an unrevised text of the κοινὴ ἔκδοσις. Many of its readings shew marks of irregular revision and the hand of an editor. As a result of this critical process, B sometimes tends to agree with the Massoretic text where other LXX. authorities represent a widely different underlying Hebrew. B also contains a certain number of widely spread corruptions that are of purely Greek origin, which are absent from earlier forms of the LXX. such as the Old Latin⁹⁹¹⁹⁹¹An interesting and plausible specimen of this class of errors occurs in 4 Regn. iii. 2 B καὶ εἶπον Ὤ (A, with , καὶ ἐπάνω. The process of corruption is evident (, , ). In Sirach instances are especially abundant, e.g. xliii. 17 ὠνείδισεν (A, ὠδίνησεν); 23 ἐφύτευσεν αὐτὴν Ἰησοῦς (H. P. 248 ἐφ. ἐν αὐτῇ νήσους); 26 εὐωδία τέλος (248 εὐοδοῖ ὁ ἄγγελος). ." In certain books the general character of B breaks down altogether, i.e. the archetype of B in those books was of another kind. Thus in Judges B was formerly suspected of representing the Hesychian recension⁹⁹²⁹⁹²Grabe, ep. ad Millium (1705)., whilst a living scholar has hinted that it may give the text of a translation not earlier than the fourth century A.D.⁹⁹³⁹⁹³Moore, Judges, p. xlvi. The Cambridge editors of the A text of Judges wisely content themselves with "the surmise that [as regards B and A in this book the true text of the Septuagint is probably contained neither in the one nor in the other exclusively, but must be sought for by comparing in detail, verse by verse, and word by word, the two recensions, in the light of all other available evidence, 489and especially of the extant remains of the Hexapla⁹⁹⁴⁹⁹⁴A. E. Brooke and N. M cLean, The Book of Judges in Greek acc. to the text of Cod. Alexandrinus (Cambridge, 1897), p. v.'—a remark which is capable of a much wider application⁹⁹⁵⁹⁹⁵On the B text of Sirach and Tobit see above, pp. 271, 274..

Cod. A, the great rival of cod. B, "exhibits a text which has been systematically corrected so as to agree more closely with the Hebrew⁹⁹⁶⁹⁹⁶Driver, Samuel, p. 1.." "In all four books of Kings and in some other parts A has been conformed to the Hexaplar text . . . In fact A is often little more than a transcript of the fourth column of the Hexapla, but without the critical signs by which Origen's additions were marked off from the rest⁹⁹⁷⁹⁹⁷Burkitt, Aquila, p. 19; cf. p. 53 f. Cf. Silberstein, Über den Ursprung der im cod. Alex. u. Vaticanus des dritten Königsbuches . . . überlieferten Textgestalt (Giessen, 1893).." In other words, adaptation to the Hebrew has been effected not by direct use of the official Hebrew text, but through the medium of Origen's work. Thus, if B represents in part the text which lay before Origen when he began his task, A, at least in the historical books, answers roughly to the result at which he arrived.

Yet A is very far from being, even in the earlier books, a mere reproduction of the Eusebian recension. It has been extensively hexaplarised, but it possesses a large element of ancient readings which are not Hexaplaric, and which it shares, to a great extent, with the Lucianic family. Moreover, as we have already seen, the citations of the LXX. in the N. T. and by Christian writers of the first three centuries, often support the readings of A with a remarkable unanimity⁹⁹⁸⁹⁹⁸Above, pp. 395 f., 403, 413, 422.. These phenomena point to the presence in A of an underlying text of great antiquity, possibly a pre-Christian recension made in Syria⁹⁹⁹⁹⁹⁹It is, however, possible that the readings in B, which have no such support and are indeed almost unique, belong to a still earlier text of the LXX., which had not received Palestinian revision. Cf. p. 429.. It must be observed, however, that the text of this MS. is not 490homogeneous throughout. The Psalms are evidently copied from a Psalter written for ecclesiastical use, and it is interesting to notice how constantly A here appears in company with the later liturgical Psalters R and T, and with the seventh century corrector of א known as א ^c.a. In the Prophets אAQ are in frequent coalition against B, and in agreement with the group which is believed to be representative of the Hesychian recension.

As to cod. א it is more difficult to form a judgement. We are still dependent for its text on Tischendorf's facsimiles. Moreover, with the exception of a few fragments of Genesis and Numbers, larger portions of 1 Chronicles and 2 Esdras, and the Books of Esther, Judith and Tobit, 1 and 4 Maccabees, this MS. is known to us only in the poetical and prophetical books. Notes at the end of 2 Esdras and Esther claim for the MS. that in those books it was corrected by the aid of a copy of the Hexaplaric text written under the supervision of Pamphilus^10001000See above, p. 75. The N. T. has now appeared in collotype, with introduction by Prof. K. Lake (Oxford, 1911).. But the first hand of א often agrees with A against B, and the combinations אART in the Psalms, אAC in the other poetical books, and אAQ in the Prophets, are not uncommon. In Tobit, as we have seen, א follows a recension which differs widely from B. On the whole, however, it comes nearer to B than any of the other uncials, often confirming its characteristic or otherwise unique readings. Cod. C is yet more fragmentary and its fragments are limited to the poetical books which follow the Psalter.

Thus if a single uncial MS. is to be adopted as a standard of comparison, it is obvious that either A or B must be chosen for the purpose, and B is to be preferred as being freer from Hexaplaric interpolations and offering generally a more neutral text. The latter MS. has therefore been employed by recent editors, and this course is probably the best that can be 491followed. But the method of editing the text of a single MS. leaves much to be desired, for, as Lagarde rightly insists, no single MS. and no single family of MSS. can be regarded as a trustworthy or sufficient representative of the original LXX.

5. There remains the alternative of constructing a critical text. This can only be done by the scientific use of all existing materials^10011001Cf. Nestle, Zur Rekonstruction der Septuaginta (in Philologus, 1899).. The task which lies before the critical editor of the LXX. is partly similar to that of the N. T. editor, and partly sui generis. The general principles which will guide him are those which have been expounded by Dr Hort in the second part of Introduction to the N. T. in Greek^10021002Ed. 2 (1896), pp. 19—72.. The documents moreover fall into the same three classes: (1) MSS., (2) versions, (3) literary citations; although in the case of the LXX., the versions are 'daughter-versions' and not based upon an original text, and the citations are not limited to post-apostolic Christian writers, but may be gathered also from Philo, Josephus, and the New Testament. But in the application of the principles of criticism to these documents the critic of the LXX. must strike out a path for himself. Here his course will partly be shaped by the fact that he is dealing with a version and not with an original text^10031003The original text may be regarded as the primary document for the text of the version., and by the history of the transmission of the version; which is only to a limited extent identical with that of the transmission of the Greek New Testament.

(a) The first business of the critic of the LXX. is to review the documentary evidence which is available for his use. This has been already described at some length (MSS., pp. 122—170; Versions, pp. 87—121; Citations, pp. 369—432). The preliminary work of preparing these materials for use is still in progress. We now have access to photographic reproductions 492of codd. ABGLQΘ, facsimiles or printed texts of אCDEFHKO RTUYZΓΠ, and collations of the remaining uncials, and of a large number of the cursives. But the facsimiles are more or less inadequate, and the older collations of unpublished MSS. need careful verification. To turn to the versions, the fragments of the Old Latin are now for the most part accessible in carefully edited but scattered texts, and the more important of the Egyptian and Syriac versions have received much attention; but the Armenian, Ethiopic, Arabic, Georgian and Slavonic are still but partially explored. Good progress is being made in the editing of Philo, Josephus, and the Christian fathers, both Greek and Latin. Thus, while much remains to be done in the way of perfecting the apparatus criticus of the Greek O. T., there is an abundance of materials ready for immediate use, and every prospect that in a few years the store will be largely increased.

(b) When an editor has been found who is competent to undertake reconstruction, he will probably desire to limit himself to that one task, after the example of the editors of the New Testament in Greek^10041004Cf. Hort, Intr.², p. 90., and his resources, if not as abundant as those of the N. T. editors, will be both sufficient and trustworthy. But with the materials thus ready to his hand, how is he to proceed? As in the case of the New Testament, he will begin by interrogating the history of his text. Here there are certain landmarks to guide him at starting. As we have seen, the three recensions which in the fourth century had a well-defined local distribution, have been connected with groups of extant documents—two of them quite definitely, the third with some probability. Other groups representing less clearly recognised families have emerged from recent enquiries, such as that which yields the text characteristic of the catenae (H. P. 14, 16, 28, 52, 57, 73, 49377, Paris Reg. Gr. 128, and many others), the pair H. P. 54, 75, with which Θ and 59 may also to some extent be classed, and the codices which correspond more or less closely with cod. A and cod. B respectively. It is probable that as the collation and examination of MSS., versions, and fathers proceed, other groups, or other members of the groups already mentioned, will come to light, leaving an ever diminishing number of documents which present a text either too mixed or too peculiar to be classified.

(c) In operating upon the groups thus obtained the critical editor will possess two chief aids towards the discrimination of ancient elements from those which are later or recensional. (1) While the East in Jerome's time was divided between the Lucianic, Hesychian, and Hexaplaric texts, the great Western dioceses, Carthage, Milan, and Rome, read the LXX. under the guise of a Latin version, beneath which originally lay a Greek text anterior to the Hexapla itself. Consequently, the Old Latin, in its purest types, carries us behind all our existing MSS., and is sometimes nearer to the Septuagint, as the Church received that version from the Synagogue, than the oldest of our uncial MSS. Readings which have disappeared from every known Greek MS. are here and there preserved by the daughter-version, and in such cases the 0. L. becomes a primary authority for the Greek text^10051005Burkitt, Tyconius, p. cxvii. f.. But besides these occasional contributions of a direct nature, this version is of the highest value as enabling the critical editor to detect pre-Origenic readings and to distinguish them from those which are later or recensional. In regard to the latter point the test is not an absolute one, because it is always possible that the reading on which an O.L. rendering is based was one of two or more that were both current in the κοινή before Origen's time. (2) But the 0. L. is not our only witness to the readings 494of the κοινή. Its evidence may often be checked and confirmed by that of the Syro-Hexaplar and the fragments of the Hexaplaric Greek, where the obeli and asterisks distinguish readings which existed in Origen's MSS. from those which were interpolated from other sources, or rewritten with their aid^10061006On this point see Burkitt, Aquila, p. 33 f..

(d) By such means the critic may often satisfy himself that he has reached the text of the Septuagint as it was found in Christian MSS. of the third, perhaps even of the second century. It is another question how far the κοινή ἔκδοσις of the Christian Church was identical with the pre-Christian text or texts of Alexandria and Jerusalem. Early citations from the LXX. suggest a diversity of readings and possibly the existence of two or more recensions in the first century, and lead us to believe that many of the variations of our MSS. have come down from sources older than the Christian era.

Here our documentary evidence fails us, and we have to fall back upon the 'internal evidence of readings.' The variants which remain after eliminating Hexaplaric matter, and recensional changes later than the Hexapla, resolve themselves into two classes; viz. (1) readings which affect merely the Greek text, such as (a) corruptions obvious or possible, or (b) doublets, whether brought together in a conflate text, or existing in different MSS.; and (2) readings which presuppose a difference in the original. In dealing with both classes much help may be obtained from Lagarde's earlier axioms^10071007Above, p. 484 f.. In detecting corruptions the student must chiefly depend on his faculty of recognising a Semitic original under Greek which does not directly suggest it; in deciding between double renderings, he will set aside that which bears marks of correction or of assimilation to the official Hebrew or to later Greek versions based 495upon it, choosing that which is freer, less exact, and perhaps less grammatical, as being probably nearer to the work of the original translator. Lastly, when the variants imply divergent Hebrew texts, he will prefer, ceteris paribus, that which departs from the Massoretic text. The application of these, rules, however, calls for knowledge and judgement of no ordinary kind^10081008On the scope for conjecture where evidence fails, see Hatch, Essays, p. 281, where some other remarks are to be found which deserve attention but need sifting and safeguarding..

6. It cannot be doubted that the future will produce a school of critics competent to deal with the whole question of Septuagint reconstruction, and that a critical edition of the Old Testament in Greek will hereafter take its place on the shelves of the scholar's library by the side of the present New Testament in Greek or its successor. Meanwhile some immediate wants may be mentioned here. (1) Several important uncial MSS. still need to be reproduced by photography, particularly codd. א, F, R, V, T; and the process might well be extended to some of the weightier cursives. (2) Texts of which photographs have been published, or of which verified transcripts or collations exist, deserve in some cases detailed examination, with the view of determining their precise character in the several books or groups of books, and their relation to one another and to a common standard; such as the text of B. (3) The stores of fresh Hexaplaric matter which have accumulated during the interval of years since the publication of Field's great book^10091009See the second fasciculus of Dr Redpath's Supplement to the Oxford Concordance., will soon be sufficient to form a supplementary volume, which might also contain the corrections supplied by photography and by the more exact collation of Hexaplaric MSS. (4) Is it too much to hope that the University which has the honour of having issued from its Press the Septuagint of Holmes and Parsons 496may see fit to reprint at least the apparatus of that monumental work with such emendations and abbreviations as it may be possible to adopt without seriously interfering with the scope and method of the edition? It is improbable that a collection of all the evidence on so vast a scale will ever be attempted again, and until this has been done, Holmes and Parsons cannot be superseded as a storehouse of facts. (5) A proposal was made by Dr Nestle at the London Oriental Congress of 1892 to compile a 'Variorum Septuagint,' giving the text of B with marginal variants sufficient to correct the errors of that MS. There can be little doubt that such an edition would be serviceable, especially if the scheme could be so far extended as to include a selection from all the variants, after the manner of the English 'Variorum Bible.' (6) Every student of the Old Testament will wish success to the undertaking which is now in progress at the Cambridge Press. Although the text of the Larger Septuagint will be simply that of the standard MS. employed in the manual edition, its apparatus will for the first time present to the critical scholar the essential documentary evidence, verified with scrupulous care, and arranged in a form at once compendious and helpful to research.

LITERATURE. W. Selwyn, art. Septuagint, in Smith's D. B. iii. (London, 1863). P. de Lagarde, Anmerkungen zur gr. Übersetzung der Proverbien (Leipzig, 1863); Genesis Graece (Leipzig, 1868); Ankündigung einer neuen Ausgabe der gr. Übersetzung des A. T. (Göttingen, 1882); Librorum V. T. canonicorum pars prior (Göttingen, 1883); review of Cornill's Ezechiel in Gött. gelehrte Anzeigen, June 1, 1886 (reprinted in Mittheilungen, ii. 49 ff., Göttingen, 1887). J. Wellhausen, Der Text der Bücher Samuelis (Göttingen, 1871); art. Septuagint in Encycl. Brit. ⁹ (London, 1886). C. H. Cornill, Das Buch des Propheten Ezechiel (Leipzig, 1886); in Gött. gelehrte Nachrichten xxx. (1888, 8, p. 194 ff.). A. Ceriani, Le recensioni dei LXX. e la versione latina detta Itala in Rendiconti del R. Istituto Lombardo II. xix., xxi. (1883—4); review of the O. T. in Greek in Rendiconti 497II. xxi., xii. (1888); De codice Marchaliano (Rome, 1890). W. Sanday and F. J. A. Hort, letters in Academy, Dec. 10 and 24, 1887. V. Ryssel, Untersuchungen über die Textgestalt . . . des Buches Micha, p. 175 ff. (Leipzig, 1887). I. Hooykas, Iets over de grieksche vertaling van het Oude Testament (Rotterdam, 1888). H. Oort, De Lagarde's plan van eene vitgaaf der Septuaginta (? 1882). E. Hatch, Essays on Biblical Greek, iv.—vii. (Oxford, 1889). S. Driver, Notes on the Hebrew Text of the Books of Samuel, Intr. p. xlvii. ff. (Oxford, 1890). A. Dillmann, Textkritisches zum Buche Ijob (in Sitzungsberichte d. k. P. Akademie d. Wiss. zu Berlin, 1890, liii.). E. Nestle, The Variorum Septuagint, in Proceedings of Oriental Congress held at London, 1892; Urtext p. 77 f. (1897); Zur Rekonstruktion der Septuaginta, in Philologus, N. F., xii. (1899) p. 121 ff. E. Klostermann, De libro Coheleth versione Alexandrina (Kiel, 1892); review of The O. T. in Greek in Gött. gelehrte Anzeigen (1895. 4). S. Silberstein, Über den Ursprung der im Cod. Alex. u. Vat. des dritten Königsbuches überlieferten Textgestalt (Giessen, 1893). Bleek-Wellhausen, Einleitung in das A. T., p. 549 ff. (Berlin, 1893). F. C. Burkitt, The Rules of Tyconius, p. cxlii. ff. (Cambridge, 1894); The Old Latin and Itala (Cambridge, 1896); Fragments of the Books of Kings according to the translation of Aquila (Cambridge, 1897). G. Moore, Commentary on the Book of Judges, p. xliv. ff. H. P. Smith, Commentary on the Books of Samuel, pp. xxx. ff., 402 ff. (Edinburgh, 1899). A. Rahlfs, Septuaginta-Studien, i.—iii., Göttingen, 1904, 1907, 1911; C. F. Burney, Notes on the Heb. Text of the Books of Kings, Oxford, 1903; W. O. E. Oesterley, Studies in . . . Amos, Cambridge, 1902; C. C. Torrey, Apparatus for the criticism of Chronicles, Ezra, Nehemiah, in O. T. and Semitic Studies (Chicago), xi.; H. St J. Thackeray in J. Th. St., xii. 46 ; and many of the works named on pp. 27, 28; 104; 191—4; 262—4; 285—8.