Observations on Certain Problems Connected with the So-called Septuagint
This article was originally published in TSF Bulletin 56 (1970), pp. 8–13.
It is not the purpose of this article either to give a general account of the Septuagint or to relate the results of recent research in this field. Several such accounts with extensive bibliographies are already available in the up-to-date Bible Dictionaries, in Professor Jellicoe’s recent work The Septuagint and Modern Study (Oxford University Press, 1968), and, for those who read German, in Professor J. Wevers’s ‘Septuaginta Forschungen seit 1954’ (Theologische Rundschau, NF, 33, Mai 1968, 18–76). It would be pointless, and indeed impracticable, to summarize these accounts here. Rather the purpose of the present exercise is to ventilate two problems raised in the minds of some by views on the nature and quality of the Septuagint which, if not new, find nowadays greater publicity as the result of more recent research. Nor is it the intention of this article to offer broad, general statements as a solution to numerous particular difficulties which can only be solved individually and that by detailed investigation. Broad general statements about the Septuagint are themselves the source of some of the difficulties. Rather is it proposed to offer some observations which may help to provide a realistic background against which individual problems should be seen, and which may, at the same time, suggest some lines of approach to the investigation, if not the solution, of those problems.
Problems of text
Problem number one stems from the often-made observation that the Septuagint is in places both inaccurate and tendentious. In itself, of course, this observation, however true, would present no real difficulty. Inaccuracies are found in greater or lesser quantities in all Bible translations. And since a certain amount of interpretation is unavoidable in the process of translating, scarcely any translation could everywhere escape the charge of being intentionally, or unintentionally, tendentious. But with the Septuagint there enters a special consideration: the New Testament writers frequently quote it. A problem therefore arises in some people’s minds: how can the New Testament writers quote and weave into their arguments an inaccurate and tendentious translation, and still lay claim to our belief that they were inspired?
Problem number two is raised by the claim, increasingly heard nowadays, that the Septuagint has in many places preserved the text of the Old Testament better than the Masoretic Text has. Again this in itself would constitute no difficulty. In principle the same thing could happen here and there with the New Testament and it would trouble nobody. Once it is accepted that the original autographs of the New Testament have disappeared and that the original text must be recovered from copies of the original, then obviously it is possible that a given reading, accurately copied in a Greek manuscript, should later have been accurately translated into some other language. In this case the translation will have preserved the original reading equally well as the Greek copy. And once it is accepted that no one Greek copy, nor any one copy of a translation, is everywhere and in all places an accurate copy, and that the original text of the New Testament must therefore be reconstituted by comparing all the Greek copies, and all the copies of all the translations, and following for each detail that copy, or copies, which has best preserved the original reading, then no new or strange principle is involved if on the odd occasion it is a copy of a translation that alone appears to have preserved the true reading. And it stirs nothing but gratitude if there exist copies of translations that render such a service.
So, then, with the Septuagint. No strange or disturbing principle is invoked, nor should anything but gratitude be aroused, by the claim that in places copies of the Greek translation of the Hebrew Old Testament have preserved the original better than the Hebrew copies have. But there are nevertheless other factors which seem to make the case of the Septuagint somewhat special. First and foremost is the great frequency of the Septuagint’s divergences from the Masoretic Text, and second is the size of many of these disagreements, which exceeds by far anything one finds in the New Testament textual traditions. If in a number of these larger disagreements, to say nothing of the hundreds of smaller ones, the Septuagint is to be preferred to the Masoretic Text, then the suspicion is created that the latter must be a very unreliable text indeed. Unfortunately this, as anybody can see who reads the Septuagint, does not mean that the Septuagint in its turn is everywhere a consistent and uniformly good witness to the original. Quite the reverse. To recover with certainty the original Septuagint itself from the mass of variants in the Septuagint manuscripts is in places exceedingly difficult; and in other places, where the original is easily recoverable, the sense is poor, at times poor enough to be called nonsense. One can see, therefore, how some people might form the hasty impression that the Septuagint and the Masoretic Text are both very poor and unreliable representatives of the original Hebrew Scriptures, and might conclude that our knowledge of that original must be a very doubtful and uncertain thing.
At this point another, relatively minor, factor may increase their unease. For centuries until the discovery of the Dead Sea Scrolls the Masoretic Text was the only Hebrew text tradition of the Old Testament. It therefore enjoyed until recent times an authority which no New Testament text tradition has enjoyed since the Textus Receptus was challenged by the discovery of earlier and better manuscripts. For this authority of the Masoretic Text now to be called in question at this late stage by the discovery of Hebrew biblical manuscripts belonging to a different text tradition more in line with the Septuagint might at first sight appear as something hostile to the integrity of the Old Testament text. It is not, of course. As in the New Testament, the possession of differing text traditions increases rather than decreases the possibility of accurately reconstructing the original text. But then that is not all. So long as the Masoretic Text was the only Hebrew text tradition, and the Septuagint had no extant Hebrew manuscript to support it in its large divergences from the Masoretic Text, it was easier in any dispute between them to give the Masoretic Text the benefit of the doubt. True, some scholars showed an astonishing readiness to prefer the Septuagint, and make it the base for their numerous suggested emendations of the Hebrew text, and constantly cited its evidence in support of their source-critical analyses. Kittel’s Biblia Hebraica and the older commentaries are storehouses, not to say museums, of the results of this preference and procedure. On the other hand more extensive study of the Septuagint often tended to show that these deductions from the Septuagint were either not valid or did not constitute evidence worthy to be preferred to that of the Masoretic Text. (Contemporary study very often points in the same direction. See, for instance, D. W. Gooding, The Account of the Tabernacle, Cambridge University Press, 1959, pp. 29–39, 66ff.; VT, 1965, 405ff.; 1967, 145ff., 173ff.). And so the Masoretic Text’s reputation tended to rise (D. W. Thomas, The Recovery of the Ancient Hebrew Language, Cambridge University Press, 1939, p. 37). But then the Dead Sea Scrolls appeared, and among them, as we now know, are Hebrew Bible-texts which agree with the Septuagint, not merely in minutiae, but in some of its characteristic and significant differences from the Masoretic Text (see F. M. Cross, The Ancient Library of Qumran, Duckworth Press, 1958, pp. 130–5). Now these Hebrew texts are comparatively small in extent and fragmentary, too; but they show us that the Septuagint’s larger disagreements with the Masoretic Text, even where they are unsupported by any extant Hebrew text, cannot be automatically attributed to the whim and inventiveness of the translators; they may be founded on Hebrew manuscripts that have just not happened to survive. And if these Hebrew manuscripts were straightforward Bible texts, their evidence must be heard, albeit through the medium of the Septuagint, on equal terms with that of the Masoretic Text. Intrinsic merit alone can settle the dispute between them. But if this be so, consideration of the size and number of the Septuagint’s peculiarities might well lead some to imagine that the question of the original text of the Old Testament is now in a grave state of uncertainty.
So much then for the unease that some may feel. Perhaps this account of it is to some extent imaginary or exaggerated. Nonetheless the frequency with which the RSV has chosen to depart from the Masoretic Text and follow the Septuagint, to say nothing of the multitudinous suggestions in the learned commentaries, has been sufficient to make a wide public aware at least that a problem exists.
The use of the LXX in the New Testament
To come now to some general observations. Let us take first the implications of the fact that New Testament writers quote the Septuagint. In the modem world, if we found a writer frequently quoting the AV, we should not be justified in concluding that this writer necessarily approved of the AV’s translation of every verse in the Bible. Still less should we be justified in such a conclusion, if the writer on times quoted translations other than the AV. Unless he were very unintelligent, one would infer from his use of more than one translation, that he was aware that more than one translation existed, that these several translations did not always agree, and that the translations were not necessarily all equally accurate. If further our writer knew Hebrew, Aramaic and Greek and sometimes offered his own translation of a biblical verse, it would be altogether unwarrantable to deduce from his frequent quoting of the AV that he approved of the AV everywhere, mistakes and all. What then we should never dream of deducing from our modem writer’s use of the AV, we should not deduce from the New Testament writers’ use of the Septuagint, especially when they do not always quote the Septuagint but, as the evidence suggests, make on times their own ad hoc renderings (see R. H. Gundry, The Use of the Old Testament in St Matthew’s Gospel, Brill, Leiden, 1967, p. 171).
But to develop the analogy further. The AV is a homogeneous translation, of uniform style throughout, made at one time, and from the first issued in codex form within two covers. In spite of this, quotation of passages here and there cannot be taken as implying approval of the translation of every verse. How much less can the New Testament quotations of the Septuagint be taken to indicate approval of its renderings everywhere. The Septuagint is not homogeneous. It was not translated all at one time, nor all by the same person or persons, but by different people in the course of many decades (D. Barthélemy reckons to have proved that the translation of Ecclesiastes is the work of Aquila, whose floruit is ad 117–138; see Les Devanciers d’Aquila, Brill, Leiden, 1963, pp. 32–33). The translation-styles are also widely different, ranging from good koine Greek in some books to renderings in other books so literalistic as to be unintelligible to anyone who did not know Hebrew (see H. St J. Thackeray, Grammar of the OT in Greek, Cambridge, 1909, p. 13). Add to this even in pre-Christian times continuing, though spasmodic and partial, revisions of some books, and remember also that the New Testament writers would have known these Greek translations of the Old Testament books as separate scrolls. (The putting of them all together in codex form along with apocryphal and pseudepigraphical works was, of course, a post-New Testament development.) Remember, too, that the Greek Esther eventually circulated in shorter and longer editions, and that our so-called Septuagint now contains two different translations of Daniel. It will then be apparent that the fact that New Testament writers on numerous occasions quote verses or longer passages from some of these Greek translations cannot fairly be taken to mean that they approved of every rendering in every scroll which has happened to find a place in what we now call the Septuagint. This is, of course, a very elementary observation; but the habit of extending to this motley collection of translations the name Septuagint, which originally applied only to the Greek Pentateuch, does sometimes betray us into thinking and speaking as if the ‘ Septuagint’ were a homogeneous whole, which the New Testament writers would have carried around with them in codex form as one distinct, unvarying and unalterable translation.
But to say, however truly, that the New Testament writers’ quotation of parts of these Greek translations does not imply approval for all of them, leaves untouched a far bigger question: when the New Testament writers quote from these Greek translations, is the translation in every individual passage quoted a fair and true rendering of the original Hebrew? Are there ever places where the writer quotes an inaccurate Greek translation (and here it does not matter whether it comes from the so-called Septuagint, or from some other source, or whether he made it up himself), and knowingly or unknowingly exploits its inaccuracy to secure a point in his argument unfairly? This is the real problem; the fact that the ‘Septuagint’ is inaccurate, misleading, and unfairly tendentious in other passages which the New Testament writers do not quote is irrelevant. The real problem is very important, but for two reasons I do not propose to answer it here: (a) because it cannot be answered by making broad generalizations. Every quotation must be examined individually; ten quotations, examined and proved to be correct translations of the original, are no proof that the eleventh quotation is a fair and true translation. (b) The examination must go beyond the question of translation. Even if the New Testament were written in Hebrew and its quotations of the Old were accurate citations of the original wording, one would still have to examine the contexts in the Old Testament from which the quotations were drawn to see if the use made of individual verses by the New Testament writers was always fair to the original contexts.
How much more so if one is dealing with quotations of a translation of the original. Obviously a task of this magnitude is beyond the possibilities of this brief article, but, lest I should seem to have raised the question merely to gloss over it, I repeat that this is the real problem. It has, of course, long been discussed in the learned commentaries, and, in the light of quotations and exegetical practice as found in the Qumran literature, is still being discussed, both in the newer commentaries and in monographs like Dr Gundry’s work mentioned above. At this point, however, it may be worth while to make a few general observations on the topic of the fair and unfair use of exact and inexact translations, so long as it is understood that these observations are offered not as a blanket solution of the above-mentioned problem but as some considerations worth remembering by any who attempt to investigate the problem. First the mechanical difficulty of hunting up a phrase in a scroll, as distinct from the (comparative) ease of doing this in a book. This difficulty of checking quotations made the ancients rely more on memory than we do (their memories were remarkably good) and be content with accuracy of sense rather than demand verbal accuracy.
Next the mode of quotation expected in the ancient world by the ancients themselves. On this topic F. Johnson’s The Quotations of the NT from the Old considered in the Light of General Literature, published as long ago as 1896, is still worth reading. Dr Gundry (p. 171) puts the matter thus:
However, it is common knowledge that the ancients did not scruple against quoting interpretatively. Neither the historian in the Graeco-Latin classical tradition nor the Jewish targumist had the modern concept of the sacrosanctity of direct quotation. Rather, a certain freedom of interpretation and adaptation was expected in order to show one’s grasp of the material, to bring out its inner meaning and significance, and to apply it to the subject at hand.
Thirdly, consider a modern analogy. A missionary goes in 1969 to an African country. The local Christians have a New Testament in their own language, but the translation was made eighty years ago by a pioneer missionary, not experienced in the art of translation, and in many places, therefore, it is not exact. The missionary wants to impress a point of doctrine upon the local church, and to prove that point by quoting a New Testament passage. Knowing Greek himself he can see that the translation of this passage in the African New Testament is by no means accurate. While it does not pervert the sense, it contents itself with vague paraphrase which conveys no more than seventy per cent of the original meaning. He could himself supply a better translation which would support his point more powerfully. But if he supplies his own translation, the Africans, not knowing Greek, might suspect that he was manipulating the translation for his own advantage. On the other hand if he uses their translation, he gets no unfair advantage, for the proof is not so strong as it might be; nevertheless it is strong enough to prove his point and it does not wrest the general meaning of the New Testament context. Is he acting unfairly, if he uses the African translation?
The Septuagint and the Masoretic Text
To come now to the second main problem: the comparative value of the Greek translations and the Masoretic Text as witnesses to the original text of the Old Testament. Here again to get a true picture one must beware of generalizing, as if what is true of some books, or parts of books, were true of the whole Old Testament. This applies not only to the motley collection of Greek translations but also to the Masoretic Text. Take the latter first. Professor F. M. Cross, as strong a champion as any of the value of the Septuagint as against the Masoretic Text, has this to say about the text of Jeremiah:
The text of Jeremiah is of particular interest. In the recension underlying the Septuagint text it is one-eighth shorter than in the Hebrew Bible . . . . From Qumran comes a fragmentary Hebrew manuscript, which, where preserved, follows the short text of Jeremiah found hitherto only in Greek. In Chapter 10, for example, the Septuagint omits no fewer than four verses, and shifts the order of a fifth. The Qumran Jeremiah (4 Q Jerb) omits the four verses and shifts the order in identical fashion’ (The Ancient Library of Qumran, p. 139).
Elsewhere he comments:
Those who have defended the originality of the traditional text by arguing that the Greek translator abbreviated the Hebrew text before him are proved wrong. The Septuagint faithfully reflects a conservative Hebrew textual family. On the contrary, the Proto-Masoretic and Masoretic family is marked by editorial reworking and conflation, the secondary filling out of names and epithets, expansion from parallel passages, and even glosses from biblical passages outside Jeremiah. (‘The Contribution of the Qumran Discoveries to the Study of the Biblical Text’, Israel Exploration Journal, 16, 2, 1966, p. 82)
On the other hand, having had this to say about the Proto-Masoretic Text of Jeremiah, Professor Cross says of the Proto-Masoretic Text of the Pentateuch: ‘To be sure, there are secondary expansions in the Pentateuch, but by and large it is a superb, disciplined text’ (‘The History of the Biblical Text in the Light of the Discoveries in the Judean Desert’, HTR, 57, 1964, p. 289).
Or to take an example of similar unevenness in the Greek tradition. Thousands of students have become aware of the value of the so-called Lucianic recension of the Books of Samuel from S. R. Driver’s words (Notes on the Hebrew Text . . . of the Books of Samuel, p. xlix):
But what imparts to Lucian’s work its great importance in the criticism of the OT, is the fact that it embodies renderings, not found in other MSS of the LXX, which presuppose a Hebrew original self-evidently superior, in the passages concerned, to the existing Masoretic Text . . . Lucian’s recension contains elements resting ultimately upon Hebrew sources, which enable us to correct, with absolute certainty, corrupt passages of the Masoretic Text.
But this does not mean that Lucian’s recension is like this throughout the whole Old Testament. In some books, like Isaiah, it is a poor text; in Genesis it is virtually indistinguishable. (For the better understanding that Qumran has given us of this phenomenon see both D. Barthélemy, Les Devanciers d’Aquila, pp. 89–139 and F. M. Cross, The History of the Biblical Text, pp. 292 ff.).
Another point to remember is that the statement that in, say, Jeremiah and some of the historical books, the Greek translation is founded on a better Hebrew text than the Masoretic, does not mean that the Greek is automatically superior in every reading. Far from it. Which tradition is better in any given verse is a matter that has to be decided detail by detail. The Masoretic Text could still be superior in many, or even in a majority of the details, if for no other reason than that the Greek is often a poor translation.
Moreover, it is good not only to avoid the temptation to generalize, but also to make some conscious effort to look at text-critical statements in proper proportion. In the nature of things, a textual critic speaking of two textual traditions will talk most of the, say, twenty per cent difference between them, rather than the eighty per cent agreement. It is a healthy thing, therefore, to compare the Greek translations with the Masoretic Text in order to see how much they have in common. It varies considerably but in many books the common element is, of course, very large.
Again, as the old dictum reminds us, variant readings should be weighed, not counted. An expansionist addition may involve a score or more words which statistically speaking, will bulk large; and yet it may not impair any of the original sense of its context. On the other hand a corruption involving only two words can destroy the sense of a whole sentence. Or again, apparently similar phenomena can have different significance. In the books of Kings there is a dispute between the Greek and the Masoretic Text over the position of the paragraph describing Solomon’s palace. The Greek is clearly secondary, but the difference in ultimate meaning is comparatively small. On the other hand, in those same books there is a dispute over the positioning of several other paragraphs, which is based on a considerable disagreement over chronology. This is important (see J. D. Shenkel, Chronology and Recensional Development in the Greek Text of Kings, Harvard University Press, 1968, and my review in the forthcoming issue of JTS, October 1969).
If, then, one were to make an evaluation, on this kind of qualitative basis, of the disputes between the Masoretic Text and the Greek translations throughout the whole Old Testament and then were to total the scores, there is no doubt what the result would be: the Masoretic Text would emerge overwhelmingly superior. In a very real sense, of course, it would be pointless to compile such a total, for our task is not to defend the Masoretic Text against all others (any more than we attempt to defend the ‘Neutral’ text or anyone other text tradition in the New Testament) but to reconstruct the original; and we value all the evidence available from whatever source. The more sensible assessment to make would be: With the help of the Masoretic Text, other Hebrew text traditions, the Greek translations, and their secondary translations, Targums etc., etc., what proportion of the original can be reconstructed? My impression is a high, a very high proportion.
Yet the fact remains that, for all the help they can give, the Greek translations suffer in differing degrees from grave defects that suggest the need for the utmost caution in using their evidence for the reconstruction of the original Hebrew. One of these disadvantages is that they are inadequate translations. Admittedly Qumran has shown that some features formerly thought to have been introduced by the translators, were in fact based on Hebrew texts. But there remain multitudinous places where the translations are inadequate, or positively mistaken, or so consistently paraphrastic that their evidence over a particular point is unreliable. Then there are the occasions when the Greek’s rewriting of the original goes far beyond the limits of legitimate paraphrase and can only be held to be a deliberate alteration of the sense of the original. The motive behind these alterations varies. For example, it may be piety, as in the oft-quoted Exodus 24:10, ‘And they saw the place where the God of Israel stood’, instead of ‘And they saw the God of Israel’. Or it may be nationalistic pride and prejudice, as in Isaiah 19:24–25, where the Masoretic Text has
In that day shall Israel be the third with Egypt and Assyria, a blessing in the midst of the earth: for that the Lord of hosts has blessed them, saying Blessed be Egypt my people, and Assyria the work of my hands, and Israel mine inheritance
but the Greek has
In that day shall Israel be third among the Assyrians and among the Egyptians blessed in the land which the Lord hosts blessed saying Blessed be my people that is in Egypt and that is among the Assyrians and my inheritance Israel
that is, the Jewish diaspora in Egypt and Assyria, and the Israelites in Palestine! (See I. L. Seeligmann, The Septuagint Version of Isaiah, Brill, Leiden, 1948, p. 117.) Or, again, it may be the cultural influence of the translator’s Hellenistic background that led the translator to rewrite the original. The Greek Proverbs is an example of this: it has been extensively rewritten throughout, and Hellenistic influence is both obvious and strong (see G. Gerleman, Studies in the Septuagint, III, Proverbs, Lund, 1956).
Moreover, while all translation involves some measure of interpretation, there comes a point at which, if interpretation is added needlessly, or expansions inserted gratuitously, the end result is no longer the original in translation, but a commentary on the original, or a historical novel. The Greek Esther, as distinct from the canonical Esther, and the Greek 1 Esdras, are examples of the historical novel; 1 Kings (3 Reigns) exhibits numerous features of midrashic commentary (see ZAW, 3, 1964, 269–80; VT, XV, 1965 153–166, 325–335; XVII 173–189; ‘Problems of Text and Midrash in the Third Book of Reigns’ Textus, 1969). Now this does not mean that a book like the Greek 3 Reigns is of no value in the reconstruction of the original 1 Kings, but it does counsel careful discrimination in the use of its evidence. To take an extreme analogy. A New Testament quotation embedded in the text of a commentary by an early church Father may be very valuable evidence for the original text of the New Testament; but it does not mean that the surrounding commentary, as well as the quotation, is to be regarded as a New Testament text in the sense that, say, Codex Vaticanus (B) is. And so when we find midrashic interpretations worked into, or simply inserted in, the Greek translations of the Old Testament, we should not proceed on the assumption that since they are found in a translation that elsewhere offers valuable evidence for the text of the original, they are themselves to be ranked as textual evidence of the same status as the Masoretic Text. And even should it be that the Greek translations were based on Hebrew texts that already had these midrashic elements incorporated in them, that does nothing at all to increase the value of these midrashic elements as witnesses to the original text of the Old Testament. A midrash on a biblical book, even if the midrash were in Hebrew, would not be the same thing as a biblical text.