"Guessing" in the Septuagint

by JAMES BARR, Oxford

What is guessing, and how does guessing work as a phenomenon of intel-

lectual activity?
It is not to be denied that guessing is important. Those who teach univer-
sity students, and who grade their examinations, are very much aware of its
existence and indeed its ubiquity-not least in a subject such as Hebrew. I
remember, as a boy, being instructed by teachers: always write something
down, there is after all a chance that it may be right; the examiner cannot
give you any marks for a blank sheet of paper, but your guesses, however
unlikely and remote, may reveal that you have at least some sort of knowl-
edge about the subject being examined. Guessing, then, it seems, may be an
important pedagogical principle. ,
And for the learned also it may be very necessary. Guessing seems to be a
central hermeneutical approach. I recently sat through a full-length lecture,
which was on my own subject, but I understood very little of it. I knew, as
well as anyone could, the general area within which the lecturer was speak-
ing, I had read the books and articles to which he referred. I could (I
thought) follow a reasoned argument. But all the time I found I was making
guesses. "What on earth is he trying to say?" I kept asking myself. "How is
this point connected with the last one?" "Since he is not a native English-
speaker but is lecturing in English, can I surmise what are the words of his
own language which he has in mind and which, if I knew them, might make it
clearer what he is trying to communicate in English?" All these connections
and interpretations seemed to depend on guessing. Doubtless the lecturer, a
distinguished person, himself knew what he was saying, but for me as a
hearer guessing was the best I could do. In an absolutely clear lecture, per-
haps, guessing falls to a minimum; but when linguistic communication is not
well organized, and not presented along with all the necessary information,
then the amount of guessing rises to a high proportion. Without guessing, it
seems, we cannot cope with certain problems. There should, perhaps, be a
science of guessing: or maybe there already is, but if so I cannot guess what
its name would be.
Guessing thu_ has many interesting aspects about it. For one thing, guess-
ing usually goes with haste. One is asked a question, which has to be
answered at once. One has no time to look up the dictionaries, no time to go
back over all one's past knowledge and bring it into order, no time to reason
out all the possibilities: so one guesses. If someone has worked over a ques-
tion for two or three days, considering all the possible aspects, his answer

TYNDALE HOUSE LIBRARY

36 SELWYN GARDENS
,t.V JJ\.Mt.:> .OJ\KK vuessmg m me ;>epruagmc .Ll

may indeed be wrong, but we would not say that he "just guessed": on the medium ("translating" is Haas's term for this); 2 (2) understanding its mean-
contrary, he thought about it, he used the available resources, he did not ing, a different matter from converting it into the medium of sound; (3)
guess. Guessing goes with the need for a quick decision. Television competi- translating it into another language, in our case Greek.
tions, which allow only a few seconds for an answer, display a great deal of Method A. Under Method A the reader worked from the then written text.
guessing: if the people had longer to think about it, they would not guess so The forms as written were often ambiguous or of multiple possible identity,
badly. in relation both to the sounds of the Hebrew words and to the meanings.
Another characteristic of guessing is randomness. It has no method, no Knowing the variety of possibilities, the reader worked by a process of
organized or logical manner of exploring the different possibilities. Guessing semantic/syntactic scanning, looking at each word within a series and consid-
can go anywhere, in any direction. Unless a person actually knows the ering what meanings and functions they, as spelt, might have within that con-
answer, and has it ready for immediate presentation, he can snatch at any- text. When he decided on a meaning through this process, he was also able to
thing at all that may seem to have some vague connection with the problem. provide the full phonic realization of the words. Whether he actually pro-
Guessing, also, seems to be detectable only when it is wrong. One may not nounced the words aloud makes no difference. The essential point is that the
actually know the answer, but if one guesses rightly one is credited with semantic/syntactic scanning precedes the full pronunciation of the words and
knowing. If one guesses wrongly, it is likely to be very obvious that it was a provides the basis for it. The reader may say to himself, "this is lana, i.e.
guess and that one did not know at all in the first place. 'year' - no, sorry, it's obviously 'sleep', which will be lena"; he will say this to
Guessing, then, it seems, can be a very interesting subject. And it has a himself, and will certainly not say "on semantic/syntactical grounds we can
possible involvement in the study of the Septuagint and other ancient ver- be sure that only a noun will fit in here etc. etc". Nevertheless reasonings of
sions of the Bible. The LXX rendered the Hebrew words into Greek. Did the latter type are normal in the reading of unpointed Hebrew text, can be
they know, or did they guess? And, in either case, can we tell whether it was made very fast and easily (provided that the grammar and vocabulary of the
one or the other? language are well known), and in any case were the only way with most texts
in ancient times, there being no other.
Method B. In Method B the reader, having before him the then written or
"consonantal" text, asked for, or had available, the full pronounced form for
In two or three articles 1 I distinguished between two different ways in each word (in traditional terms, the "vocalization"; or, in more modern
which an ancient Hebrew text might be read, calling them Method A and terms, the reading tradition). He combined the written text with the reading
Method B. I applied the distinction to the procedures of the ancient transla- tradition. In other words, Method B is in principle the same as what happens
tors, although it is in essence derived from, and applicable to, the reading of with a pointed text, except that the same kind of information that is now
texts in Hebrew itself, quite apart from the special task of translation. derived from the points was then derived from an orally transmitted tradi-
In ancient times, as we all know, the biblical text did not have the "point- tion.
ing" that is now taken as normal. What a reader had before him was a text, The above has been expressed in terms of the "reader": but the same
commonly but not quite accurately called the "consonantal" text, which regis- applies to one who is translating the text into Greek. Under Method A the
tered the main consonant phonemes (except that i, and vi were not marked) translator, who had to carry out a semantic/syntactic scanning in any case, in
and also gave partial and partially optional indication of some of the vowels, order to read the text as Hebrew, used this same process to obtain the neces-
under conventions that are familiar. Many of the vowels were not marked at sary meanings and syntax for expression in Greek. He may or may not have
all in writing or only vaguely indicated. How did the reader read this pronounced to himself the Hebrew, but that makes no difference: the vocali-
text?-and that in any of the three main relevant senses of "reading", i.e. (1) zation, so called, was not available to him as a datum from which to work.
reading it aloud as Hebrew, i.e. converting it from a written into a spoken Under Method B the translator knew, or obtained from someone else who
knew, the "correct" vocalization (more exactly, the pronunciation of the
entire word), and used this, along with the written text, as combined datum
for the rendering of the text into Greek.
1 Principally in "Vocalization and the Analysis of Hebrew among the Ancient Transla-
In the articles mentioned above, I tried to show that certain LXX render-
tors", in: Hebriiische Wort/orschung, FS W.Baumgartner, VT.S 16, Leiden 1967, 1-11; also
ings are highly compatible with the use of Method A. Thus at Gen 47n
"Reading a Script without Vowels", in: W. Haas (ed.), Writing without Letters, Manchester
1976, 76-100; also, though less directly, "The Nature of Linguistic Evidence in the Text of
the Bible", in: H.H. Paper (ed.), Language and Texts: the Nature ofLinguistic Evidence, Ann 2 See Haas's own article, "Writing without Letters", p.189; cf. also his own earlier vol-
Arbor 1975, 35-57. ume in the same series, deliberately entitled Phono-graphic Translation, Manchester 1970.
 22 JAMES BARR "Guessing" in the Septuagint 23

"Israel bowed himself upon the head of the bed" (MT miffa) but LXX r}a/3- lated me to explore the problems from some different angles. One point
80~ implies that this was read as ma!fe, i.e. with the same consonants. It is where some clarification might be made concerns the "knowledge" of the
not probable that the translator knew, or was given, an authoritative pronun- "vocalization" at the time when the LXX translation was done. There was, as
ciation as ma!fe. Much more likely, he looked at the text, "saw" there the Tov savs, an "awareness of vocalization", as is indicated, among other things,
familiar word which he recognized as "staff", and rendered it so into Greek by the ·fact that texts ("unvocalized") were read aloud in public. Quite so: I
without further thought. 3 I gave a number of other similar examples, and never supposed otherwise, and indeed this observation is one of the founda-
suggested that Method A could be detected at a variety of places in the LX,X, tions of my own opinion. For there are two or three different things involved
fitted well with what is known of its circumstances of origin, and provided a here.
useful link with the "etymological" style of translation which seems to have Firstly, the fact that texts were publicly read aloud does not of itself settle
developed increasingly in the later tradition of translation into Greek. the question, by what method they were read. My Method A is fully consist-
This account of the matter, however, did not receive universal acceptance. ent with public reading. Especially in the texts that were most commonly read
People had doubts about it. Surely it was strange to suppose that the transla- in public, the historical and legal materials of the Torah, this was not necessa-
tors worked· "without taking the vocalization into account". Is not the idea of rily very difficult to do, and could easily be done fast enough by a practised
"Method A" a most implausible piece of theory? Barr has gone too far (as reader. The reader had the "consonantal" text in his hand, and as he read it
usual!). he did the necessary semantic and syntactic parsing, which dictated the pro-
Much the most important reaction is that expressed by Dr. Emanuel Tov nunciation which he then orally produced. Perhaps here and there he read a
in his fine book, The Text-critical Use of the Septuagint in Biblical Research word differently from the way in which he, or someone else, had previously
(Jerusalem: Simor, 1981 ), pp. 160-65, and since this book is likely to be very read it, but as long as it made reasonable sense the public would quite prob-
widely used by students I would like to return to the subject once more. It is ably accept it. The fact of public reading forms no obstacle at all to Method
from Tov's assessment, in particular, that I derive the connection of the ques- A unless we suppose that at that time already all public readings of text were
tion with guessing, and hence the title of this article. After giving a number of identical in vocalization and had to be so. There is no reason to believe this
reasons against my opinion, which will be mentioned in due course, he writes for a time as far back as the third century B. C. I do not suggest that this
(p. 164): argument demonstrates that Method A was factually in use; but it does show
It may therefore be assumed that, as a rule, translators were aware of the that it was perfectly compatible with public reading.
essential phenomena of the vocalization, except for one situation where Secondly, Dr. Tov distinguishes, and rightly, between "all the details of
Barr's theory must be accepted. It is probably correct to assume that vocalization", which the translators certainly did not possess, and "its main
translators often merely guessed at the meaning of the consonants and features", which they did recognize (his p.162). Quite so. We are not con-
did not think of any particular vocalization. In these cases the transla- cerned, therefore, with such matters as the difference between mobile shewa
tors indeed proceeded from the semantic identification directly to the and the hatephs. What matters is the set of distinctions which govern major
rendering, often disregarding such details as prefixes or suffixes. semantic and svntactic distinctions between forms that would otherwise be
written alike: thus between natan and noten, between malko and mal'ku.
My view, then, is a correct one where guessing is involved. Examples given by "Main features" of this kind of course they "recognized", and this again is
Dr. Tov include cases where the rendering depended on Aramaic meanings fundamental to my own original conception. In effect it just means that,
rather than Hebrew, and two of these are cited. It is added that "The number apart from occasional rarities, they knew Hebrew grammar, and without this
of translational guesses is much larger than is generally thought" (p. 164, thev could not, of course, have done their work.
n. 2). In etymological exegesis also (p. 165) the translators often proceeded But this sort of "recognition of vocalization" is not the point at issue. The
from the semantic identification directly to the translation. So there is some question is not that of knowing how the vocalization works within the gram-
room for the view that I have expressed, but it belongs to special cases such mar and syntax of the language, or of knowing what the various vocalized
as guessing and etymological exegesis. It is likelv that Dr. Tov's judgement, word forms are. These are (apart from details) matters of regularity, running
expressed by a scholar of his extreme distinction in LXX studies and within a through the entire language system, recurrent, familiar, and easy to remem-
book of such high value, will be very influential. ber for a native speaker. For Method B to work, something quite other than
Unfortunately, I still think that I was entirely right. But Dr. Tov has an awareness of the essential phenomena of the vocalization is necessary.
expressed some issues in a very suggestive way, and his comments have stimu- Method B requires not a grammar of the language, an understanding of its
working, but a text. What is necessary for it is the possession in memory,
' For further detail, see my VTS article, p. 3. known by heart, of the phonic text of the entire Hebrew Bible, or at least of
24 JAMES BARR "Guessing" in the Septuagint 25

the book being translated, at every point, or the accessibility of a person who ble: both make sense, and both in fact occur. 5 We leave aside cases such as
had that same knowledge and could be conveniently and immediately con- 'anoki n-t-n, since by normal syntactic scanning this must be the participle
sulted. Method B depends not on an awareness of vocalization but on the and not the perfect, and thus must be noten and not natan. We concentrate
possession of a complete phonic text: for in text, unlike grammar, we are deal- on the cases which are, on the basis of the consonantal text alone, ambiguous,
ing not with regularities but with unpredictabilities, and even if one has and these are many, perhaps around fifty in this one book: the cases I have
known the vocalization of (say) n-t-n, and through it the meaning, correctly judged relevant are in fact 49. Of these in the MT 36 are the participle noten
the last ten times, this does not tell you how it will be the next time, as will and thirteen are the perfect natan; of the cases of noten, none is written with
shortly be illustrated. waw. All are very similar in general type of content, so that from the general
Knowledge of "vocalization" in this sense is an enormous task: to know by sense either possibility would seem reasonable.
heart, for example, the phonic text of the entire book of Isaiah or Eze- Of these 49, LXX conforms closely to MT in 46, while in three it deviates:
kiel? - and this at a time when no graphic marking of these vowels existed, or at 1111 it has fom:xr::v where MT has J-!:ll, and at 121 and 20u it has ofommv
had ever existed. That there could have been persons who had this ability I where MT has 1m- Moreover, a study of Wevers' careful edition in the
do not question. But that the LXX translators, always or even normally, were Gottingen series shows that textual variation in the Greek manuscript tradi-
these same persons seems to me doubtful and improbable. tion, in respect of the difference between past and presenu'future in these
Dr. Tov writes as if I thought that my Method A was the one solely or many places, is truly minimal. .
exclusively used by the LXX or by some of the translators. This was never A case like this confirms Dr. Tov's observation of the substantial agree-
my opinion. I only maintained that it was intelligible, was highly likely, and ment between the MT and the Greek. But it also illustrates the problem; it
explained a considerable number of cases. The degree to which it was actu- leaves some questions; and it furnishes some suggestions.
ally used I left vague, because I did not know how to determine the answer. It illustrates the problem, for we can see how difficult it might be for any
Thus I wrote: 4 reader, by either method: the same written form occurred about fifty times, in
There is, I believe, evidence that among the ancient translators of the contexts where either reading was entirelv possible and the two are entirely
Bible some worked by one of these methods and some by the other, or mixed up, 1.ri; and HJl sometimes both occurring in close proximity (e.g.
that they varied back and forward between one and the other. l611-1s). Knowledge of how it had been the last time, or the last ten times,
did not at all predict how it would be the next time. As I have said above, to
It is not, in fact, difficult to imagine ways in which the passage back and for- work by Method B required a totally perfect identification as between parti-
ward between the two methods could take place. And this brings us to
ciple and perfect at every single point.
another difference of emphasis which may help us forward.
Then follows the obvious question: why, given that the LXX was so very
Dr. Tov is impressed by the degree of agreement between the LXX and the close to MT, did it three times diverge from it? It is not difficult to offer an
MT in relevant regards. He writes (p. 164 ):
explanation. Let us agree, with Dr. Tov, that there was a common tradition
There is considerable agreement between the LXX and MT in the un- of vocalization here. The translator knows this, and knows it quite well. But
derstanding of the consonants, which amounts to agreement in vocal- even so there comes a time when he has forgotten, or is not sure, or the infor-
ization. This agreement can hardly be coincidental, and it probably re- mant on whom he relies is absent, and he simply renders according to what
flects a common tradition of vocalization. the consonantal text will permit, which is what he had to do in any case in all
Yes, quite so: yes, and no. The same observation is one of the foundations of translation work from written texts other than the Bible. In other words, he
my own thinking. But there is another side. The more one is struck by this drops into Method A.
degree of agreement between Hebrew and Greek, the more one has to ask There is, indeed, another path to go: one can say there were two or more
for an explanation for the existence of places where deep and flagrant traditions of reading. The LXX followed, by Method B, one such tradition
breaches of this agreement take place. Let us take two illustrations. which happened to differ from that of MT. This is of course possible and I
A. The reader of Deuteronomy is familiar with expressions such as "the am sure that it sometimes happened. But observe the consequence of going
land which the Lord your God n-t-n to you". This is very common in the along this track. The necessary consequence is the multiplication of variants.
book. But both "gives", which would be noten, and "gave" (natan) are possi-
5 For another study of the relation of verb tense to translation technique see my article

"Translators' Handling of Verb Tense in semantically ambiguous Contexts", in: Claude E.

Cox (ed.), LXX: VI Congress of the International Organization for Septuagint and Cognate
' Writing without Letters, p. 86. Studies, Jerusalem 1986, SCSt 23, Atlanta, Georgia: Scholars Press, 1986, 381~403.
 26 ]AMES BARR
"Guessing" in the Septuagint 27
At 12t there was in ancient times an actual, spoken text natan, as in our MT, ning that I have suggested. Thus at the end of v. 2 the translator took the
and also an actual, spoken text noten, reflected by LXX o[owaiv. If all such phrase as me- 'olam 'ad 'olam 'atta "from eternity to eternity thou art", a per-
differences reflect actual differences in the ancient reading tradition, then fectly possible construction. Taking it in this way, however, he has next the
their number will be large indeed. Like Dr. Tov, I am reluctant to go along sequence 'l ts"b 'nws" 'd-dk'. Since the '/ has not been used in his construction
that path, except where there is substantial evidence to support it and no of the previous clause, it has to fit into the syntax of this next one. The con-
alternative explanation at hand. What I have suggested is an alternative tent of the next clause suggests something undesirable, against which the
explanation. It helps us to understand how the LXX text can be very close in Psalmist might well pray. The '/ therefore is the negative particle and is trans-
such matters to the MT and yet have great variations. lated in Greek as µ7]. It is frankly impossible to believe that the Greek trans-
B. Yes, great variations: for this is what is insufficiently expressed by lator could have gone this way if he had known of 'el as the authoritative
observations about the considerable agreement between MT and the Greek. pronounciation of the first word. Much the most natural interpretation is
Take a text of quite different character from the familiar prose of Deuteron- that this Psalm, or these portions of it, were translated by Method A. Note,
omy, such as Ps 90 (LXX 89). Here, in a quite short poem of about seventeen incidentally, that, where we can follow the renderings of this poem, they
verses, we have a series of drastic differences between MT and LXX, of often show signs of careful precision and literalness. Even if the variations
which the most relevant are listed below: 6 listed above are serious mistakes, some of them are very sensible suggestions
MT LXX LXX implies and by no means wild guesses, for example both "years" in v.5 and "our age"
in v. 8, in view of the general content of the poem.
?l! µ1] ?l! The idea of Method A, therefore, has positive evidence in its favour in that
:i~w frr; :ir~. crit it has explanatory power in numerous passages, explanatory power that no
u1;7~ 6 alciv 77µwv 1Ji;iii
other suggestion can deploy. Having said this, however, it is worth while to
lJ'?:;> tt&:tfnoµ&v lJ'?:;> go back over some of the background of this discussion.
12 :i;i:;iry :,,;? rfj xa{)Ot(l tv ao<p{(l :i;i:;iry:;i :,, 1;,
Firstly, in expressing the view I did, I had no intention of propounding a
14 lJz':'/,W tv&:r:trya{/rJµ&v 1Jp~ novel or revolutionary theory ("Barr's theory"). On the contrary, I thought
UIJ?pW 6V(f){)aV{}17µ&v 1J~7;~ that I was only expressing in a more up-to-date and linguistically correct
16
':!TJ~ OOTJYTJOOV -rrr::r fashion what was already accepted as obvious by scholars. Take for example
the classic opinion of Swete, published in 19C0: 7
Now all the examples listed above have the same quality: they are render- Lastly, almost every page of the LXX yields evidence that the Hebrew
ings that fit quite well with conceivable Hebrew terms as indicated by the text was as yet unpointed. Vocalisation was in fact only traditional until
consonants of MT, but only so if they are provided with a phonic text, a set the days of the Massora, and the tradition which is enshrined in the
of vowels and other audible features, that is drastically different from that of Massoretic points differs, often very widelv, from that which was inher-
MT. The "considerable agreement" between MT and LXX has here disap- ited or originated by the Alexandrian translators.
peared. Now of course it is logically conceivable that LXX was translated, by
Method B, from a reading tradition that was in vowels widely divergent from - and Swete then goes on to cite some of the same examples that I have used.
MT, for instance that the translator heard an authoritative oral reading of ?X Now no one today doubts that the Hebrew text had no points at the time
as 'al, of:,,:,,, as leb b•, of 7,,:, as hadrek, and so on. Some few of these possi- when the LXX originated. But, that being so, how did the LXX give evi-
bilities may be valid, and in that case they are real variants, correctly trans- dence, "on almost every page", that the text was then unpainted? Solely
lated as such by the LXX. But I doubt if any scholar takes more than a few of because it had many words where the Greek agreed with a meaning of the
them in this way. It is far more convincing to understand them, in most cases, Hebrew consonantal text but only if the implied vowels of that word were
as renderings that worked from the signs of the written text, producing other than those of MT. Swete expressed this in an old-fashioned way, using
meanings that were consistent with that text and ignoring the question of an approach predicated on the completed and pointed text. My own
what the vocalized text was like. And note in particular how well the LXX approach by contrast accepted that a reading tradition could be and had been
rendering at certain points fits in with the scheme of semantidsyntactic scan- preserved over centuries before the vowel points were first used, and that the
points were for the most part merely registering that older tradition. On the
• I have not included those cases, fairly plentiful in this Psalm, where it is extremely dif- other hand I was deeply interested in the dynamics of reading an unpainted
ficult to see what the implied Hebrew (or Aramaic) might have been, as in various parts of
verses 8-12. 7 H.B. Swete, Introduction to the Old Testament in Greek, Cambridge 1900, p. 322.
28 JAMES BARR "Guessing" in the Septuagint 29

text. Given these interests, and given the actuality of the evidence that Swete ary break with tradition. There remain a series of further aspects which may
had cited, I simply reformulated his thoughts in the form of a reading tech- also support the validity of my account of Method A.
nique which worked from the then written text. I did not think of this as (a) Method A and Method B leave behind them evidences of very differ-
either revolutionary or controversial: it seemed to me to be both traditional ent kinds; this, I think, has not been noticed before. Many average pieces of
and obvious. LXX prose could have been translated by either Method, and from the result
Equally, I was much influenced by Chaim Rabin's fine article The Transla- one cannot tell. This is one reason for the lack of positive evidence or "abso-
tion Process and the Character of the Septuagint (Textus 6, 1968, 1-26) and his lute proof" (Tov, p.163). But, when something like substantial evidence is to
picture of the earlier translators as being something like the "dragoman" or be found, it is found in quite opposite ways. Method A leaves evidence of its
practical, commercial and diplomatic, translator of the Orient. This seemed existence mainly, perhaps only, when things go very far wrong, e.g. when the
to fit well with the facts and has been rightly and widely acclaimed. But con- translation at Gen 1511 tells us that Abraham "sat down with" the wild birds
sider the implications of Rabin's view. The dragoman was an ad hoc practical or when a w~rd meaning "God" is rendered as "not" (Ps 902, cf. above, p. 26).
translator. He might work with oral communication, converting it into oral When Method A works rightly, there is no evidence of its existence.
speech in another language. He might also deal with letters, commercial doc- Method B works in the opposite sense. With it, too, in a normal piece of
uments, diplomatic messages and the like. These he would receive in written prose no actual evidence of its existence is to be found. The best evidence of
form and translate into the receptor language, whether in oral or in written it is seen when it works extremely well, that is, when it produces results so
form. But in either case, if it was a written text, he had no choice but to work precise and good that they are difficult to explain except on the supposition
by Method A. There was no other way to go. If a Greek-speaking client in that a phonic ("vocalized") text was used by the translator: in other words, in
Jerusalem brought to him a letter in Hebrew from someone in Hebron, the sort of situation I have described for n-t-n in semantically ambiguous
would he go to Hebron or send a messenger to find the writer and ask him contexts in Deuteronomy (above, pp. 24 ff.). When things go badly (and
for the "vocalization" of it? Of course he would not. He worked out the "badly" here means, roughly, a large semantic/syntactic divergence between
meanings of the Hebrew from the semantic and syntactical possibilities of the MT and LXX), Method B commonly implies the recognition of more vari-
written text, and translated the whole thing into Greek, whether oral or writ- ants than before. At Ps 90s, for example, it is conceivable that the translator,
ten made no difference. What lay outside the dragoman-translator's normal working by Method B, heard the phonic realization of the word as 'olamenu,
experience was the idea that, in order to understand a written text, you had to and correctly rendered it as "our age". If this is so, then we have to recognize
go to some authoritative person and ask that person to provide the correct that 'olamenu, existing in the valid reading tradition as heard by the transla-
vocalization. tor, is a valid variant of the Hebrew text.
Now the Bible was a different and special case, for here a reading tradition Thus the kinds of evidence left behind them by Method A and Method B
with full pronunciation no doubt existed. It is perfectly likely that the drago- are not commensurable and cannot be simply weighed against one another.
man-translator took this into account where and when he knew about it. But In respect of critical strategies they lead in different directions.
I cannot believe that, even if the dragoman-translator knew of this tradition (b) There is, in addition, one piece of evidence which is definite proof of
and took it into account, he allowed this (to him) novel procedure, applicable the use of Method A, and by a wide variety of LXX translators: that is, the
to no other text than the Bible, to eradicate completely the habits of his own existence of d/r variations. The widespread existence of d/r variations is
professional lifetime, which were, necessarily, to work out the meaning from recognized, but the implication they have for our question is less commonly
the consonantal text alone. This was the way in which contemporary docu- noticed. 8 For a moment's thought makes it obvious that a d/r difference
ments like Ben Sira and inscriptions were read, just as it is the way in which a between the Hebrew and the Greek could have taken place only where the
Modern Hebrew newspaper is read. Even if a translator took a reading tradi- text is visually read but without an oral tradition of its pronunciation. If a trans-
tion into account, at all sorts of points he could and did still conclude to the lator looks at a text that has ,,,, and translates it with the meaning of 1Y1',
meaning from the consonantal text. Paradoxically, though the Bible was a far then it is clear that he has not heard the reading of the text, or taken into
older text than a commercial letter written in the third century B. C., from the account what he then heard: he has translated on the basis of the written
point of view of translation methods and the translator's experience it was signs alone, and has misunderstood them precisely because he did not hear
the Bible that was the novelty. Hence it is not surprising if Method A is best them read aloud. If the translator did not know the phonic realization of the
represented in the earlier translations, while Method B becomes more domi-
nant later.
All these thoughts, therefore, seemed to me to follow naturally from • Dr. Tov pays proper attention to the d/r variations (e.g. pp. 229ff. of h,s boo~), but
widely accepted scholarly positions, and not at all to constitute a revolution- seems not to take into account the repercussions they have upon our present quest10n.
 30 JAMES BARR
"Guessing" in the Septuagint 31
consonants, then a fortiori he was not much governed by the vowel patterns. phonic reading tradition, and that comes very close to my own position, that
This argument seems to me to be irresistible. . in such cases the rendering was worked out from the written text itself, with
Now indeed the d/r confusion exists already within the Hebrew text and 1s the use, no doubt, of some extra imagination.
plentifully exemplified there. So it is not a matter of the translators only. (d) Various other arguments were already taken into account in my orig-
Quite so: but that only strengthens my argument. It shows that, within the inal articles and need not be repeated here. For instance, the fact that names
transmission of the Hebrew text itself, d/r differences took place, and could of persons and places were transliterated in the LXX does not at all prove
have taken place only on the basis of the graphic appearance of the text: no that a definitive phonic text of these names was known to the translators. 9
one, hearing way-yede; would have transcribed it as way-yera' (Ps 1811, 2 Sam And, especially in the case of place names, the realization and transmission of
2211) or the other way round. This kind of difference, which existed within these names in the Greek is notoriously wild and inaccurate, so that it is not
the Hebrew written text itself, existed in even greater amount as between the surprising that Swete, in the passage already quoted above, went on to cite
Hebrew and the LXX. In a few cases we may reasonably suppose that the precisely this phenomenon as a good argument for the lack of a standard
difference existed already in the written Vorlage of the LXX, but surely no vocalization as basis for the LXX. And place names form a link with the
one will argue that this explanation is true of all of them, or even of the argument already stated above, for within them the d/r difference is partic-
majority. ularly prominent: e.g., Num. 33 has in v. 12 Raphaka but MT Dophka, and
Let us repeat: confusions of this kind were numerous. Consider, m the v. 21 f. LXX Dessa but MT Rissa. On the whole, the presentation of personal
major Prophets alone, these cases, all with y1vwax(l) in Greek: and place names in the LXX fits better with the idea that a firm phonic text
was not recognized except for the more important persons and places, whose
Is 89 111hJC'T;l~ 1:;', - yvwv: lih•'l xai 1-rraa{Je
names would be known in any case, independently of their representation in
154 ill :if;,; iv,i~J ~ 1] lf/VXJ7 avTT!; yvwm:ra1 the text.
442c :i1z "19~ :,~i ~ yvwre on ano8o; ,j xa~d{a aurwv It is reasonable to claim, therefore, that the reading of the text by Method
Jr 216 ~1);~' ~ &yV(l)<J{XV <JE: A was indeed used by the translators to a substantial, if still only vaguely
1512 1;n~ ~'i;;J - d yvwafN/aE:Tal a{8'7(!0;; determined, degree, and that in some cases at least, where d/r differences are
involved, it must be considered to be proved beyond reasonable question.
This gives us five cases, within two books, of words in which the written
sequence r- '-h or r- '-' corresponds to LXX yivwaxw. Notice, incidentally,
how in all the above cases the r of MT is realized as a d: y-d-' is a commoner
and more familiar verb than the likely combinations with r. Moreover, an This being so, it remains to consider how far the use of Method A is really
important point of principle: the part played by the d/r differences makes it close to guessing. The question is a good one, and suggestive; one is grateful
clear why it is mistaken to formulate the question in terms of the "v?:aliza- to Dr. Tov for having formulated it in this way. But the answer seems ambig-
tion": the question is not rightly to be understood as one of the prov1s1on of uous: it goes a long way in two opposite directions. Dr. Tov seems in one
vowels, it is a matter of the existence of a full phonic text. If the translators respect to ascribe too little place to guessing, in another respect too much.
did not have a phonic text, which would have differentiated clearly between d On the one hand, Dr. T ov's statements suggest that he thinks of guessing
and r, then it is vain to argue that the vocalization must have been known. as something quite different from normal reading. Sometimes, perhaps
To sum up, the existence of the d/r differences, except in those limited "often", the translators merely guessed at the meaning of the consonants, and
cases where the confusion can be supposed to have pre-existed in the Vor- then they did not think of any particular vocalization. Now I do not think
lage, seems to be definite proof that Method A was used, at least at these that this is the case. All reading of unpointed text has a certain exploratory
points. and divinatory quality. There is seldom any positive, absolutely definite, evi-
(c) Roughly the same point is made if we consider the category of what dence which can mean absolutely only one thing: or, if there is, it is so only
Dr. Tov calls "pseudo-variants". These are cases where the LXX gives a ren- after one has scanned a large area of text, large enough to make it possible to
dering which can easily be retroverted into Hebrew, and if so done would eliminate all possibilities other than one.
differ from MT, and yet is not a true variant, in that it never existed in the Reading works by trial and error. The reader will usually try the more
Vorlage, but existed only in the translator's mind (Tov, p.228). I find it difficult obvious possibilities first. Seeing :,Jv;,, a reader will usually think immediately
to understand how Dr. Tov, who has particularly elaborated the category o.1
pseudo-variants, combines this with doubt about my Method A. For, if a vari
' So already my VT.S article, p. 5; cf. more recently Tov, p. 164, "After all, personal
ant existed only in the translator's mind, that means that it did not exist in th1 names were transliterated in the LXX".
 JAMES BARR "Guessing" in the Septuagint 33
32
of Iana "year"; only when he sees that thi_s will not fit ~oes ~e thi~k of less The first word was :,u•, and for it the translator hypothesized the most com-
frequent possibilities, of which_ the _most hkdy :111 be se~,, sleep • Show a mon form and meaning, i.e. :,~¥ "answered". He did not think, or did not
Hebrew speaker c•,:i, and he will thmk of hanm mounta1~s , _only second_ar- think quickly enough, of the pie! :,H'· The result was that, having taken l"IJY as
ily will he go on to think of "to raise up". Seeing :l'/7', he will ~1rst of all thmk qal "answer", he did not have a place for a direct object, and so •n:i (in:i) was
of "he sat, dwelt" (yasab), then of the future of the same (yeseb), only the;e- connected in a construct relationship with the preceding 7,,,the result of
after of more out-of-the-way realizations such as "he will brin~ back" (yasib) which is that the noun has no article and the phrase is read as b'-derek. Simi-
or "he will make to dwell" (yosib). The method looks complicated but can larly, further on, 'ly is taken as "to me" through contiguity with the preceding
actually be carried out with great speed and accuracy by anyone who knows '-m-r (cf. the second-person prayer that follows), and•~• ii;, is "the shortness
the grammar and the vocabulary of He?re:" well. ~here the rea~er goes of my days" to go along with "the middle of my days" in the next line. We
wrong is usually where he, having made his fmt surmise or ~ypothesis, trans- can offer explanations of this kind again and again. These are not cases of
lates that first divinatory possibility into Greek and leaves it at that, goes on pure or blind guessing, but of a quite sensible and rational procedure. It is a
to another section of the text, and fails to see that the context after further procedure dictated by necessity where no phonic text ("vocalization") is
inspection requires something different. " . ,, . available, and it works quite well most of the time. But it is fallible, and the
Method A, then, in a certain sense does belong to . guessmg : it do~s main cause of fallibility is that readers ( or translators) conclude too quickly
indeed work by hypothesis of a kind, it does belong to tnal and error. But it to the most obvious identification of the words, which usually means the
is not guessing in the sense of an exceptional ?r abnormal procedure: ra~~er, most common realization of these letters. 10 A mistaken identification, once
it is something that was very normal and, given the nature of the wntmg made, and once fixed either by pronunciation in Hebrew, i.e. in the Hebrew
system of the time, necessary and inevitable. . that it implies, or by translation into Greek, at once sends a series of shock
In another respect, however, Method A did not belong to guess11:g at a~l. waves along the line of contiguous words.
Guessing there certainly was: and real guessing m~ant a sort ~f creative w~1t- Guessing undoubtedly did occur: its most obvious cause, no doubt, was
ing of the sort of sentiment that might ~e appropn_ate to the Bible, along with that rare words or locutions were simply unknown to the translators, or that
perhaps some sort of similarity to the likely meamn~ of two o_r th~ee cha~ac- the text was too badly defaced for them to know what signs it contained
ters in the text. But Method A is not really guessmg of _this kind. It is_ a (again, incidentally, something that could have been a problem only where no
proper and appropriate reading meth?d which was normal m a!most all_ wnt- phonic text was available to supplement the defective graphic text). Much
ten documents except (perhaps) the Bible itself. If proper!}'. carne~ out, 1t sur- guessing, we may add, was caused by exactly the same circumstance we have
veyed the various semantic possibilities, P?ssible sy~tact1c relations_ of ele- been discussing, namely the absence of a phonic text. It was not so much that
ments, and so on, in relation to a syntactl~ whole, m order t? arnv: at a translators worked from the written characters alone because they were
reading of that whole which would be mea?mgful and cohere_nt m rel~ti_on to guessing, but that they guessed because they worked from graphic characters
the string of written signs taken as the basis. It ma~ be_ guessmg, ~ut 1t 1s not which they could not readily interpret by the more obvious modes of reading.
mere guessing, it is a clear and powe~ul method wit~ 1~s own logic: This is true, also, I think, of many cases where the translators have under-
Thus even when it goes wrong, which, as I have s~id, 1s the only t1m~ when stood the Hebrew as if it was Aramaic (Tov, pp.164f.). It was not that,
we have clear evidence of its existence, Method A 1s not mere guessmg but because they were basing their renderings on Aramaic meanings, they worked
has a logical series of premisses and consequences. We have seen an example from the consonants directly: it was because they worked from the conso-
from Ps 90 already above (p.26). Or take one of the famous cases, where out nants directly that it was all the easier to be guided towards Aramaic mean-
_ of about seven words the LXX implies in five a reading contrary to the vow- ings. (This is confirmed, in cases like those of :i-;,:;i~/:,i::iY, [Tov, p. 165] by the
els of MT: existence of the d/r difference, cf. above, pp. 29 f.). Fully pronounced
Hebrew words can hardly have sounded as much like Aramaic words as writ-
Ps 10224 MT '7!! ,~i< : ·~: ,~p '!)~ 1J1; :,~:; ten Hebrew words could be similar to written Aramaic words.
Ps 10124LXX ancxg{i'h7 avr@ tv 6&p laxvo~ avroo·
To sum up, then, the translation of Hebrew texts through Method A had
TiJv 6}.,1y6rryra rwv r;µcpwv µou avayycuov µ01
one or two characteristics that come close to guessing. Like guessing, it began
LXX implies a reading as:
• 0 This is not always the case: deber "plague" is much less common than dabar "word",
'?~ ib~ •~; iJp : in~ 1JH :,~¥ and yet appears at places where MT dabar is intrinsically more likely. This is perhaps to be
explained, however, through the prominence of the concept of plague in the prophets,
where most cases occur.
34 ]AMES BARR

without full knowledge of the data, since the written text gave only a part of
the information that belonged to the full (phonic) biblical text. It was in a
certain sense exploratory, divinatory and hypothetical. And it may well have
shared with guessing the circumstance of haste. The translator who wrote
"staff" at Gen 4731 knew very soon afterwards that it was a bed, but he did
not go back to correct what he had earlier written. Many cases where mis-
takes were made through Method A can be understood as the result of tak-
ing the first, and easiest, identification that came into the mind of the transla-
tor, and leaping to the conclusion that it was the right one. On the other
hand, translation on the basis of reading by Method A did not at all share in
the wild randomness of straightforward guessing. Method A reading was a
rational and logical approach, working by trial and error rather than by mere
guesswork. Trial and error, however, have to be properly balanced. When the
trial is not carried out with sufficient thoroughness, the error takes over, and
hence come the numerous mistakes that can be explained in this way. How
many cases there were, however, in which this method worked correctly, we
are not yet in a position to say.
Doubtless all this discussion has a speculative and theoretical aspect; and it
will be appropriate in conclusion to apologize for presenting so vague and
general a discussion when writing in honour of Dr. Hanhart, whose main
life-work has been so dedicated to the magnificent Gottingen edition, upon
which we all so completely depend. Nevertheless the precise and meticulous
work of editing so complicated a text, and the more theoretical task of under-
standing what lies behind it, are interdependent, and it is hoped that the pres-
ent discussion may interact with the use and appreciation of the great text
edition upon which Hanhart has laboured so fruitfully.