Old Testament Textual Criticism

Contents: Introduction * The Materials of Old Testament Criticism * The Methods of Old Testament Criticism
Appendix: Textual Criticism of LXX * Appendix: The New Testament in the Old * Appendix: Important Manuscripts (Hebrew and Other) * Appendix: Greek manuscripts cited by BHS and Rahlfs

Introduction

Trying to divide textual criticism into completely separate subdisciplines is not really a useful business (since all forms of TC have large areas in common), but if categories must be devised, the obvious categories would be New Testament criticism, Classical Textual criticism, and Old Testament criticism. And the division has some justification, because the differences between the fields are significant. For reasons of space (plus the author's ignorance, plus the fact that criticism of the Hebrew Bible is, in the author's opinion, an incredible mess with no signs of breakthrough), we can only touch briefly on OT criticism here.

In terms of materials, Old Testament criticism resembles New Testament criticism in about the eighteenth century: There are many manuscripts, but all of the same Majority recension, and there are a few versions, some of which differ significantly from the Hebrew, plus a handful of fragments of older materials. Since the manuscripts of the Majority recension appear not to preserve the original Hebrew and Aramaic with complete accuracy, there is an obvious need for textual criticism. This forces us to use rather different methods than we currently use in the New Testament.

To begin with, let us review the materials.

The Materials of Old Testament Criticism

The first and most important source is, of course, the Hebrew manuscripts. With a very few exceptions (which we shall treat separately), these were copied in the Middle Ages by scribes known as the Masoretes or Massoretes (hence the name Massoretic Text, frequently abbreviated MT or even 𝔐). The Massoretes were trained with exquisite care to preserve the text in all its details (down to such seeming minutae as the size of certain letters in the text and their position above or below the line). They also followed very exacting techniques of checking their manuscripts. The result is a text which shows almost no deviation, and manuscripts which reproduce it with incredible precision. Had such techniques been in use from the very beginning, textual criticism of the Hebrew Bible would be a trivial task.

The Massoretic Text contains a handful of carefully preserved variant readings, the K^etib and Q^ere. The K^etib ("written") are the readings of the text; the Q^ere are marginal readings which the reader is instructed to substitute for the text. Such noted variants are, however, relatively rare, and many of the Q^ere readings correct places where the text is so bad that it could hardly stand in any case. Thus the K^etib/Q^ere variants add very little to our knowledge of the ancient text, and the accidental variants of Massoretic copyists add even less. The latter should generally be treated not as authoritative variants but as conjectural emendations or simple errors; they have no genetic significance.

Our earliest substantial MT manuscripts date from about the tenth century. Prior to this, we have only a handful of Hebrew manuscripts. The best-known of these are the Qumran manuscripts (the "Dead Sea Scrolls"), though there are others such as the relics from the Cairo Genizah. With only a handful of exceptions, such as the Qumran Isaiah scroll, these manuscripts are damaged and difficult to read, and the portions of the OT they contain are limited. In addition, many have texts very similar to the MT -- but a handful do not. Perhaps the most important of all are the Qumran scrolls of Samuel, 4QSam^a and 4QSam^b, as they represent a tradition clearly independent from the MT, and apparently better (as the manuscripts lack many of the defects which afflict MT Samuel).

Also in Hebrew, but with differences in dialect, is the Samaritan Pentateuch. The production of a sect considered schismatic by the Jews, the text (which survives mostly in recent manuscripts, and in rather smaller numbers than Hebrew Bibles, as the Samaritan sect is nearly extinct) shows definite signs of editing -- but also seems to be based on a Hebrew text which predates the Massoretic recension. This makes it potentially valuable for criticism of the Pentateuch (the Samaritans did not revere the other portions of the Hebrew Bible) -- as long as we remain aware that it has been edited to conform to Samaritan biases. (We should also allow the possibility that the MT has been edited to conform to Jewish biases!)

There are many ancient versions of the Old Testament. These fall largely into two categories: Those translated directly from the Hebrew, and those translated from the Greek version. (There are, of course, versions which do not come directly either the Hebrew nor the Greek, but from some intermediate version; examples include the various Western European versions translated from the Vulgate. These are, however, of almost no interest in textual criticism of the Hebrew Bible. If they have any significance at all, it is for Vulgate criticism.)

Setting aside the Greek version and its descendents for the moment, the most important versions descended from the Hebrew are the Latin and the Syriac/Aramaic. As in the New Testament, the Latin actually went through two stages: An Old Latin phase (these versions being translated from the Greek) and the Vulgate Revision. The Vulgate was translated by Jerome in the fourth century (just as is true of the New Testament Vulgate) -- generally from the Hebrew, and with less attention to previous versions than Jerome showed in the Gospels. The result is a text generally quite close to the Hebrew. It appears, however, that the MT was well evolved by this time; Jerome's translation rarely departs from the MT, and the differences we do see may be the result of attempts to clarify obscurities or simply alternate interpretations.

The Aramaic Targums also are translations from the Hebrew, and are generally believed to be older than the Vulgate. They are also the work of Jewish scholars, meaning that they typically embody Jewish understanding of the Hebrew text. This does not, however, make them more valuable than the Vulgate. The Vulgate was translated by one man, Jerome; the Targums are multiple (e.g. the "Targum of Jonathan" and the "Targum of Onkelos"), making it harder to control for the translator's idiosyncracies. The most noteworthy characteristic of the Targums, however, is their freedom. Often they do not even qualify as translations. They paraphrase, they expand, they even include commentary. Thus it is better to treat the Targums as commentaries by Jewish Fathers than as actual translations.

The Syriac Peshitta is the final major version to derive from the Hebrew. Its history and origin is disputed, but it is clear that several hands were involved, and there are also indications of revisions from the Greek. This mixed text makes the use of the Peshitta somewhat problematic.

Which brings us to the earliest and greatest of the versions, the Greek. It should be noted that there is very little scholarly consensus on what follows; if there is any fact universally accepted about the Greek version (other than the bare fact of the existence of Greek translations), I don't know what it is. What follows is the most cautious of outlines, with conclusions postponed as best I can.

The Greek translation of the Hebrew Bible is often called the Septuagint, or LXX. This name derives from the so-called "Letter of Aristeas," which gives an official pedigree to the LXX. According to Aristeas, the LXX was prepared at the instigation of Ptolemy II Philadelphus of Egypt (reigned 285-246 B.C.E.), who wanted a version of the Jewish scriptures for the Alexandrian library. Seventy (in some versions, seventy-two) scholars were commissioned to translate the Pentateuch, hence the name LXX.

The story of Aristeas is, obviously, legend (though not the most extreme legend; Philo had it that the translators all translated separately, then compared their work and found the separate translations identical!); while Ptolemy II probably would have liked a copy of the Jewish scriptures in the Alexandrian library, there is little chance he would have supplied the funds needed for the translation project described by Aristeas. If there is any truth in Aristeas, it is only this: That the Pentateuch was translated in Egypt, probably during early Ptolemaic times.

It is noteworthy that the LXX of the Pentateuch is a careful, skilled translation. It also conforms relatively closely to the Hebrew as we have it (there are exceptions, e.g. in the ages of the Patriarchs and in the order of a few chapters, but these are quite slight compared to what we see in the rest of the Old Testament). Thus it is possible that it was an organized project of some kind. Still, it cannot be considered an official Jewish product, as the primary language of the translators appears to have been Greek.

And as we move away from the Pentateuch, the situation becomes much more complex. The LXX version of the Pentateuch seems to have been generally acceptable. The same cannot be said for the remaining books.

The term "LXX" is rather misleading, as it strongly implies that there was only one translation. This is simply not the case. The Greek Old Testament clearly circulated in multiple editions -- the earliest of which may in fact have preceded the LXX Pentateuch described by Aristeas. It is not clear whether these divergent renderings were actually independent translations (as a handful of scholars hold) or whether the text simply underwent a series of revisions. But that the "final" LXX text differed recensionally from the earliest is absolutely certain. This is perhaps most obvious in the Book of Judges, where Rahlfs (even though he is really citing only two manuscripts, the Alexandrinus/A and the Vaticanus/B) was forced to print two different texts. Few other books show such extreme variation (except in Daniel, where the version of Theodotian has replaced the original text of LXX), but all show signs of editorial work.

What's more, the direction of the recension is clear: The translation was made to conform more and more closely with the late Hebrew text, even to the extent of taking on some features of Hebrew grammar at the cost of good Greek style. Secondarily, it was made to be smoother, more Greek, and possibly more Christian and theologically exact. (This process very likely was similar to that which produced the monolith of the Byzantine text of the New Testament.) These two processes, however, were probably independent; the former resulted in the so-called kaige text, the latter perhaps in the "Lucianic" text.

We cannot detain ourselves here with the various later recensions of the LXX, which are of little value for textual criticism. A statement by Jerome has led many scholars to believe that there were recensions by Hesychius (associated with Egypt) and Lucian (associated with Constantinople), as well as an edition by Origen.

All three are somewhat problematic. The recension of Hesychius is the worst; it has never been identified. (Alberto Vacarri, " The Hesychian Recension of the Septuagint," reprinted in Sidney Jellicoe, Studies in the Septuagint: Origins, Recensions, and Interpretations, argues that there are strong indications that it has survived -- some have seen it in Vaticanus or Alexandrinus. But we can hardly be sure, since so little from Hesychius's pen has survived. Vacarri has only a few readings to work from, and he seems to think an Arabic translation is the best example. Not much help even if true.)

The text of Lucian has been pretty clearly identified, but we now have enough early manuscripts to make it clear that many of its readings predate the historical Lucian -- some have been identified in the Dead Sea Scrolls. Thus the difficulty with regard to Lucian is to identify the various strands within its text. Wevers, for instance, suggested that the basic text of Lucian, the "proto-Lucianic" text, is quite old and based directly on the original Old Greek, but prettied up, while the actual work of Lucian was to include the pluses found in Origen's "Hexaplairic&qot; tradition (below). However, I note a number of readings where Lucian moves closer to the Hebrew in places where there does not seem to be a Hexaplairic variant.

Origen's edition was the "Hexaplar" recension, which placed in six columns the Hebrew text, a Greek transliteration, and the translations of Aquila (a woodenly literal Jewish translation said by Epiphanius to have dated from the second century though there are hints that portions of it are older), Symmachus (a late translator who provided a clear rendering), LXX, and Theodotian (also thought to be older than its historical second century date; it seems a revision of LXX which is freer in style but closer to the MT in text). Origin is known to have revised his LXX text to more nearly match the Hebrew (while incorporating critical symbols to show what he had done), but later copyists simply took the text without copying the symbols. Thus Origen's text, although based on the Septuagint, quite literally eliminates most of what is of interest to us.

It should also be remembered (as S. P. Brock pointed out in "Origen's aims as a Textual Critic of the Old Testament," reprinted in Sidney Jellicoe, Studies in the Septuagint: Origins, Recensions, and Interpretations) that Origen wasn't really trying to look at the original Hebrew. He was concerned, at least in part, with controversies with the Jews. So, as Brock says (p. 344 in the Jellicoe reprint), "[Origen] was concerned with finding out what was the text of the Old Testament as used by the Jews of his own day." Origen's text testifies only to the third century. Its content shows that the MT was already largely established in his time, but we would have suspected this from other evidence.

It might help if we had accurate copies of Origen's various symbols showing his changes -- but we don't. Most copies of Origen's text either omit his symbols or, at best, bollix them badly. Without those symbols, we can never be sure if a reading of the hexaplairic type comes from the Old Greek, or derives from the Hebrew (usually through the medium of Theodotian), or is something else. Origen's was one of the greatest critical efforts of antiquity, and if we had it intact, it would be very interesting -- but his edited text in isolation is of very little use. And Origen's edition was produced before the great LXX codices were written, so there is always the danger that it has influenced a later copy. One of the major forms of drudge work in LXX studies, in fact, consists of trying to identify hexaplaric readings and manuscripts.

Modern scholars tend to use different names for the recensions -- typically "Old Greek," "Lucian," and kaige. The Old Greek is believed to be the earliest rendering. "Lucian," although in fact it predates Lucian, is an often-fuller type which remains close to the Old Greek. The kaige is so called because of one of its translational quirks -- using και γε as a hypertranslation of a particular Hebrew construction. The creators of kaige probably had some version of the Old Greek at hand, but they produced a rendering at once more literal and closer to the Masoretic Text of the Hebrew.

Each of these versions has quirks. All bear the signs of "translation Greek" -- a wooden sort of rendering that is unnatural, at times even incomprehensible, to Greek speakers. Kaige is worst in this regard, with Lucian (probably a modification of the Old Greek by someone who did not know Hebrew) being the most natural.

It has been suggested that kaige may be Origen's "Quinta." This has some problems, since kaige seems to have extended beyond the known content of Quinta.

But kaige is truly a mystery. The mystery being at its most extreme in 1 and 2 Samuel. In these books, we find an Old Greek text in large parts of Vaticanus. But, at other points, Vaticanus is kaige. It has been suggested that the original Old Greek of Samuel did not contain the missing material and that these sections (2 Samuel 10:1 to end, plus 1 Kings 1:1-2:11 and 1 Kings 22-end and all of 2 Kings) were filled in with kaige material. The problem with this is that Lucian preserves a text independent of kaige. How could this be if there had been no Old Greek?

The diagram at right shows one possible interpretation of the relationships between Greek (and Hebrew and Latin) versions. Other relationships have been proposed. In the diagram, a black line indicates a primary source; a lighter line indicates a source of mixture or correction.

The versions of Aquila, Theodotian, and Symmachus (often collectively known as "the Three") are now largely lost (Symmachus in particular has vanished almost completely). From what we can tell, their disappearance -- although sad for the historian -- is small loss to the textual critic. They are too close to the MT. As J. Reider confesses in "Prolegomena to a Greek-Hebrew and Hebrew-Greek Index to Aquila," reprinted in Sidney Jellicoe, Studies in the Septuagint: Origins, Recensions, and Interpretations, "the textual identity of Aquila's Hebrew and our own, as far as consonants are concerned, is proved in a preponderating number of cases" (pp. 318-319), although he goes on to note a handful of divergences -- and to observe on p. 320 that Aquila usually (although not always) renders the Q^ere rather than the K^etib.

Theodotian, with its links to kaige, is also close to MT. Of Symmachus we can say little except that there is no reason to think it diverged from MT in any notable way.

For the most part, the Codex Vaticanus is considered the best representative of the Old Greek, although there are places where there are other good witnesses, and some places where Vaticanus contains kaige or other sorts of texts, such as the places in Samuel noted above.

The question then arises, why did the LXX undergo such extreme revision? Why did later scholars see the need to revise, and even offer different translations? Why was this version different from all the other versions?

The answer: While there may have been many reasons, such as an uneven Greek style, or perhaps multiple translations of certain books which had to be reconciled, there seems to be only one basic one: Unlike the other versions, the early LXX does not agree entirely with the MT.

The nature of the difference between LXX and MT varies from book to book. In Isaiah, it may simply be the incompetence of the original translator. In Job and Jeremiah, however, the LXX is shorter than the MT by more than 10%. And while it is possible that LXX Job was reduced because of the damage to the Hebrew text, this cannot account for Jeremiah -- nor for the smaller reductions found in LXX Ezekiel and many of the minor prophets. (In Ezekiel 40-48, the translator may not have understood his text very well, but the MT is badly damaged and the LXX text of Ezekiel very important in trying to repair it.) In Samuel, on the other hand, the earliest LXX text is slighly longer (except that it omits a large portion of the story of David and Goliath; for a discussion of the folklore aspects, of this point, see the article on Oral Transmission), and in Kings we find many rearrangements of material. Lesser differences occur everywhere.

One of the most important results of the finding of the Dead Sea Scrolls was to verify that many of these Greek readings which differ from MT go back to Hebrew originals. While 4QSam^a does not entirely agree with LXX, e.g., it does reveal that many if not most of the LXX readings go back to a Hebrew form. Similarly, 4QJer^b, although very short and fragmentary, shows that the LXX form of Jeremiah has a Hebrew relative. Most Hebrew texts from Qumran agree with the emerging MT, but the exceptions are very informative.

The table below shows the number of Qumran manuscripts of each book, as listed in Patrick W. Skehan, "The Biblical Scrolls from Qumran and the Text of the Old Testament" (1975). Note that some of these manuscripts, such as the Isaiah scroll, are extensive, while some are mere scraps. The intent is simply to show which books seem to have been popular in the Qumran period.

Book	Number of copies
Genesis	16
Exodus	16
Leviticus	8
Numbers	6
Deuteronomy	25
Joshua	2
Judges	3
Samuel	4
Kings	3
Isaiah	18
Jeremiah	4
Ezekiel	4
The Twelve	8
Psalms	27
Job	4
Proverbs	2
Ruth	4
Song	4
Ecclesiastes	2
Lamentations	4
Daniel	8
Ezra/Nehemiah	1
Chronicles	1
Esther	0

The average number of copies of a particular book is 7.25 -- but the median is only 4.0. So we have almost seven times as many copies of Psalms as we have of the median book, and more than six times as many copies of Deuteronomy. The mode is 4 -- there are seven books of which we have four copies.

We cannot detain ourselves with the arguments over the detailed relationships between the texts, since the experts are not agreed. The basic question for most people is, How do we deal with the divergences between the MT and the Old Greek?

The Methods of Old Testament Criticism

At this point we need to step back a little and examine the situation at a higher level of abstraction. What are the basic materials for criticism of the Hebrew Bible? Throwing out all revisions and minor translations, we come down to at least two and at most four separate sources:

A "Majority Text" -- the Hebrew tradition of the MT, found primarily in late manuscripts but universal in those late manuscripts.
The Old Greek -- a version, but made at a relatively early date, from materials clearly distinct from the MT, and surviving in manuscripts earlier than the oldest copies of the MT
A handful of Hebrew fragments (e.g. the Dead Sea Scrolls and the Genizah fragments), some of which agree with MT, some with the Old Greek, and some with neither.
For the Pentateuch, we also have the Samaritan version

Since in most places we are confronted with only two independent witnesses (MT and Old Greek), scholars have to decide what to do with them. Generally speaking, they choose one of two courses -- both of which, unfortunately, are logically flawed.

One course is to treat the MT as the basic text, preferring it at all points where it can be construed. (This is the stated policy, e.g. of Ernst Würthwein in The Text of the Old Testament, although his examples prove that he is slightly more nuanced than his statements imply.) The LXX is used only where the MT is corrupt. This, we can say unequivocally, is fallacious. If the LXX has value at all, it has value everywhere. If it is too faulty to consult for the ordinary run of the text, there is no reason to consult it where the MT is corrupt -- probably it was corrupt when it came to the LXX translators, too, and they were guessing. If the LXX is that worthless, we should simply resort to conjectural emendation. Housman, in his "Preface to Manilius" (I, p. 36) had this to say about this sort of reliance upon a single source (in this case, a single manuscript, but the principle applies well to OT criticism): "To believe that wherever a best MS gives possible readings it gives true readings, and only where it gives impossible readings does it give false readings, is to believe that an incompetent editor is the darling of Providence, which has given its angels charge over him."

The other extreme is to treat the MT and LXX exactly equally, as different witnesses to the original text, or even to treat LXX as superior. There was a tendency toward the latter in the nineteenth century. This, unfortunately, has the defect that it treats a version as a text in the original language. This can hardly be allowed; one must know the method and style of the translation.

As Karen H. Jobes and Moisés Silva write on page 90 of Invitation to the Septuagint (Baker Books, 2000), "Certain scholars tend to argue that a given Greek reading is certainly due to the translator's technique. Other scholars argue with equal enthusiasm that the same reading is due to a variant Hebrew text. The tension between these two opposing solutions constitutes what is perhaps the weightiest problem in Septuagint scholarship."

The correct answer doubtless lies somewhere in between the extremes. The LXX must be consulted. From the standpoint of readings, it is often as good and valuable as the MT (in some cases, such as Samuel, it is more valuable). But the form of the translation must be examined (e.g. a reading which would be accepted based on the Greek of the Pentateuch, which is carefully translated, might not be accepted for Isaiah, which is badly translated). Great care must be taken to be sure we know the nature and style of the Hebrew behind the LXX, and only then to compare it to the MT. Much of what we know about NT criticism still applies, but care must be taken to understand the peculiar circumstances of each section, each book, and even each part of a book (as some books seem to have been translated by more than one person). For details and examples, one must refer to specialized studies.

But let me give an analogy. I once had an argument with an Israeli who was convinced of the exact and literal truth of the MT Hebrew, who could not believe that the Greek could ever correct the Hebrew. I would agree with this if we had the original Hebrew -- but we clearly don't; the damage to Samuel and Job shows that there has not been some sort of providential preservation.

The analogy I would make is to an outdoor landscape. Suppose you had two people make images of it. One paints a painting, in colour; the other takes a black and white photograph. (Perhaps this was in the nineteenth century.) The painting preserves colours, but will probably oversimplify and make errors in detail. The photograph will be entirely accurate to the limits of its resolution, but won't show colour.

So suppose you want a fully detailed, colour image of the scene. How can you possibly get one? Answer: You must use both the painting and the photograph. The photograph preserves the exact details, but the painting supplies the colours.

Similarly, to get as close as possible to the original Hebrew, one must combine the extant Hebrew and the Greek. The Hebrew, in terms of the details of language, is of course closest to the original Hebrew -- it preserves nearly all the colour. But sometimes it is defective. At these times, we must take at least some details from the Greek. It lacks the colour, but it has the details. (Sometimes. At other times, of course, the Hebrew will preserve the colour and details.) Keeping this balance in mind almost certainly gives us the best chance to get back to the original Hebrew.

To put it another way: The Hebrew must be our copy text. Where the Greek is ambiguous and admits of two Hebrew originals, one found in MT, we should always accept as original the reading found in MT. But where the Greek clearly implies a different Hebrew text, we must give fair consideration to both readings, trying to determine which one best explains the other.

Having decided how to use our materials, we also need to decide on our canons of internal criticism. In dealing with individual Hebrew and Greek manuscripts, of course, all the familiar rules of NT criticism apply: Scribes make generally the same sorts of errors. And the basic rule of internal criticism remains "That reading is best which best explains the others." But we should note that certain errors seem to be more common in the OT than the New. Hebrew, for instance, is prone to interchanges of consonants -- a side effect of the lack of vowels. There are perhaps more instances of repeated letters, and there is a certain tendency for two of a letter to be reduced to one. There are also, perhaps, more errors resulting from misdividing words, or taking a consonant-used-for-a-vowel as a consonant (that error, of course, cannot occur in Greek). These are small errors, but because they are small, they can be very widespread -- and even, perhaps, affect all our witnesses. It is worth remembering that, because we have so few sources, conjecture is a more reasonable resort in the Old Testament than in the New.

Another thing that is perhaps worth noting is a rather different method of abbreviation which may have been used in Hebrew manuscripts. MS. Oxford Heb. e 30 contains a leaf of the Minor Prophets which is dramatically shortened. The first word of each sentence or verse is written out in full, but the remaining words are represented by a single consonant plus pointing -- as if the phrase "Sing a song of sixpence, pocket full of rye" were reduced to "Sing a s o s, p f o r." The reader was clearly expected to remember the rest of the verse, with the individual letters as reminders. It will surely be evident that this might cause verses to be misremembered. We don't know how many books of this sort there were -- only a few leaves have survived -- so they may not have had much influence. But the possibility should be considered.

Appendix: Textual Criticism of LXX

Several times in the section above, I make disparaging reference to the textual criticism of LXX. This is a clear and necessary task, and it's being conducted very slowly. In recent decades, we're seeing the publication of the Göttingen LXX, which is a huge advance -- but the individual volumes are edited by individual scholars, with idiosyncratic ideas; there is no truly unified edition of LXX, and no unified textual theory. And, sometimes, people adhere much too closely to out-of-date and disproved ideas.

Much attention has been given to a comment of Jerome's that there were recensions associated with Hesychius, Lucian, and Origen. This may be true, it may not. But there is no great value in naming text-types; what matters is finding them. Some editors have sought to do this. No one has really integrated the results.

Even the underlying assumptions are not entirely agreed. For example, most scholars follow Lagarde in believing that there was an original text of the Greek translation, now typically called the "Old Greek" text of LXX. Not only is this the true LXX translation, it is also the one most divergent from the Hebrew -- and hence the one most useful for textual critics. But not all accept this! Paul Kahle argued that there were several independent Greek translations of the Hebrew Bible.

Ironically, although most scholars disagree with Kahle, they still spend a lot of time talking about his positions. There is no need for this. Whether Kahle is right or wrong, those alternate translations are mostly close to the Hebrew of the MT. From the standpoint of textual criticism, they don't matter. What matters is that one translation (which for purposes of convenience we can call LXX) which isn't translated from a text effectively identical to the MT.

What's more, if the kaige text and the Hexaplar text are not independent translations of the Hebrew (and the latter certainly isn't and the former may not be), they are undeniably recensions -- that is, there has been editorial activity. Thus they can never be entirely trusted as witnesses to the Old Greek. From the standpoint of either criticism of the Old Greek or criticism of the Hebrew, we should be treating them as if Kahle is right and they are different translations. They can of course serve as a source of conjecture, but not as direct witnesses to the original text of the LXX (and probably not to the Hebrew either).

The Lucianic "recension" is a much more difficult issue. For starters, many Lucianic readings actually predate Lucian. Also, from what we can tell, the Lucianic text looks a lot like the Old Greek -- but much prettified. The grammar is smoother and less like translation Greek, and many of the monotonous repetitions of the Old Greek are eliminated. This could, of course, be the result of someone going over the LXX text to make it sound better, probably without reference to the Hebrew (something which has certainly happened more than once with, for instance, the English Bible -- Living "Bible," anyone?). But, taken individually, the vast majority of the differences between the Lucianic and Old Greek texts could be explained by assuming that a scribe merely accidentally smoothed the reading. Thus it is not in fact evidence that Lucian's text is actually recensional. The answer to this question will distinctly affect how we use the type.

One thing that I personally wish had been studied more is the use of the divine name in LXX manuscripts. Most, of course, render it by κυριος. But Jerome mentions manuscripts that simply write the tetragrammaton in Hebrew letters (or accidentally turn that into Greek ΠΙΠΙ, those being the Greek letters which most nearly resemble the Hebrew). One of the earliest Greek manuscripts from Egypt, Papyrus Fouad 266 (Ralphs 848) from probably around 100 B.C.E., follows this usage and includes the Hebrew letters (oddly enough, the original scribe left blanks in the manuscript so a different writer could add in the Hebrew letters). Emanuel Tov studied early LXX manuscripts and found that P.Fouad was the only one to include the name in square script, but six included it in paleo-Hebrew. One (4QpapLXXLev^b) offered ΙΑΩ. P.Oxy 656 spelled out κυριος. All the other manuscripts (almost forty) used the nomina sacra. The fact that we have two consistently-used alternate forms of such a common word implies some sort of editorial activity. What? One possibility is that the usage derives from from the Other Versions; according to the Milan hexaplar fragments, Aquila, Symmachus, and Quinta used the Hebrew letters (we find the ΠΙΠΙ form in the margin of Codex Q of LXX, e.g.). So the LXX might have adopted it from them -- but how then could it be found in Fouad 266, which predates the other versions? More likely is that the use of κυριος (abbreviated or unabbreviated) is Christian usage, the use of the Hebrew spelling Jewish (with ΠΙΠΙ being just a flat error), but this doesn't tell us which form was the original.

It should also be kept in mind that all the major manuscripts of LXX are Christian (some of Tov's texts were pre-Christian, but they were all fragments). The original LXX was a Jewish work, and some of the early fragments are of Jewish origin (including a few scraps from Qumran, although most of the surviving pre-Christian LXX material is from Egypt), but as time passed, the Jews became more and more dismissive of LXX, first replacing it with Aquila and then abandoning Greek renderings altogether. This of course does not mean that the Christians altered the LXX tendentiously, but it is worth remembering that most of the copying errors that were made were made by Christians, not Jews, who would suffer an obvious if unconscious temptation to Christianize the readings and conform them to New Testament passages. (Note that this is not the same as the editorial process which added large sections to the books of Daniel and Esther; those changes probably go back to the original translations, and are the work of the Jewish translators.)

There is also the complication that LXX, unlike most classical literature, is a translation. This poses an interesting dilemma for "users": should a Greek reader want a text of the Old Greek, or of the accepted text of the Orthodox church, or a text that is a good translation of the Hebrew? This admits of no answer -- but it isn't our worry. We are not users, we are textual critics. What matters to us is getting at a source of Hebrew variants. That's the Old Greek, plus possibly the "Lucianic" recension of boc₂e₂ etc.

We continue to see volumes of the Göttingen LXX. These give the raw material for a good textual history. But stopping with their texts, or Rahlfs, is not sufficient. Just as OT criticism is in some ways analogous to NT criticism a few centuries ago, LXX studies are in a state about equivalent to NT studies at the time of Tregelles: A lot of material, and no real organization of the texts or theory on how to use them.

English-speaking readers should perhaps note that Göttingen is a purely German edition. There is no English parallel to the introduction or discussion of the text, even for the books (such as the Pentateuch) where the Göttingen editor was a native English speaker. Obviously a good textual critic should know German as well as Hebrew, Greek, and Latin -- but the German used for textual criticism isn't exactly what you will find in German for Dummies! (Suggestion for someone: How about a manual of text-critical German terms and grammar.)

Another difficulty for the casual critic is the lack of affordable materials. There is only one portable, affordable critical edition of LXX -- that of Rahlfs. And it has only a very minimal apparatus and does not represent the latest critical thinking. But the alternative, the still-incomplete Göttingen edition, is large and forbiddingly expensive. We need a Nestle-type edition, and there isn't one. This is a rather desperate need -- but don't hold your breath.

I would note (for English readers) one recent tool that can be of some help: Albert Pietersma and Benjamin G. Wright, editors, A New English Translation of the Septuagint. This is a (mostly woodenly literal) translation of the Göttingen text where available (the rest mostly translates Rahlfs), often with separate translations of the Old Greek and kaige versions. It also has introductions to the individual books, discussion the translation technique and the text. These cannot be considered the final word, of course, but they are a nice portable set of notes. While NETS and Rahlfs together hardly add up to a true critical edition and apparatus, they are far more useful together than either individually.

Having said all that, let us assume that our goal is to reconstruct the Old Greek, and perhaps also the Lucianic recension. For the most part, our tools are the same as for New Testament criticism: We use the standard rules for evaluating internal evidence; we also classify manuscripts.

But -- to repeat -- we must keep in mind the fact that the LXX is a translation. One cannot simply reconstruct what seems like the most likely text. One must also refer back to the Hebrew to determine what appears to be the basis of the translation. This will obviously influence the reading one adopts.

It will be evident that there is a dependency problem here: If we need the Hebrew to reconstruct the Greek, and the Greek to reconstruct the Hebrew, then we have no place at which to start our cycle. Fortunately this problem is not quite as bad as it sounds -- most of the time. Since the Hebrew is usually intact, and the Greek can usually be reliably reconstructed even without the Hebrew, and the corruptions often occur at different points, we can usually safely use one to reconstruct the other. Not always, sadly, but usually.

It should, incidentally, be noted that LXX criticism has at least five different aspects, depending on the book:

Reconstruction of books which are in the Hebrew Bible
Reconstruction of books which were originally in Hebrew but of which are not in the Hebrew Bible and for which no Hebrew text survives (e.g. Tobit, Judith, 1 Maccabees)
Reconstruction of books which were originally in Hebrew but which are not part of the Hebrew Bible but for which some Hebrew material survives (e.g. Sirach)
Reconstruction of books which were composed in Greek rather than Hebrew (e.g. Wisdom of Solomon)
Reconstructions of additions written in Greek to books which were written in Hebrew (e.g. the additions to Esther and Daniel)

For some, of course, it is only the first of these which matters, since the goal is to reconstruct the Hebrew Bible, and the additions make no difference. On the other hand,b the existence of the other LXX materials might influence the translations of the canonical books -- e.g. the translation known as 1 Esdras might have influenced the translations of Ezra and Nehemiah (since parallels might be assimilated), and the insertions into Esther and Daniel might influence the text around them. Thus one who wishes to use the LXX in criticism of the Hebrew Bible still needs to be aware of the portions of LXX which are not translations of canonical Hebrew books. This is most obviously true in the case of addition "D" in Esther. The other additions are just plopped into the middle of the text. But "D" is carefully woven into the Hebrew; trying to decide what parts of the addition are actual additions and which are derived from the archetype is, at best, tricky.

And the distinction between these types of criticism is real. Wisdom of Solomon is in very good Greek. Sirach is in Aquila-like translation Greek. The canons of internal evidence are very different in these two cases!

A final note, repeating part of the above: The Göttingen LXX is a noble critical effort. But it should be kept in mind that it is only a single edition -- and it isn't even the unified work of a single editor or committee of editors. It can hardly be considered the final word on the text of LXX; LXX studies are probably closer to the situation for the New Testament in Lachmann's day than to the equivalent of the UBS edition. Of course, one could argue that the Hebrew is in even worse state -- after all, we need good data on LXX before we can produce a truly critical edition of the Hebrew Bible!

Appendix: The New Testament in the Old

We are used to thinking of the New Testament quoting the LXX. And, indeed, this is the usual sort of cross-contamination: Either the LXX reading influences the text found in the NT, or the NT reading, if it mis-quotes the LXX, causes the LXX to be corrupted.

But there are a few instances of LXX manuscripts explicitly quoting the New Testament, in the book of Odes. The Göttingen edition of Psalms and Odes has a fourteen chapter set of Odes. The majority are Old Testament texts, but it also includes as Chapter 9 the poems of the Magnificat (Luke 1:46-59) and Zechariah (Luke 1:68-79) and as Chapter 13 the Nunc Dimittis (Luke 2:29-32). This raises the question, which I have not seen considered elsewhere, of how the New Testament was taken into the Old. Were the poems in Odes copied from the New Testament once, or were they repeatedly cross-compared?

This Appendix can hardly offer a complete examination of something that is surely worth a Ph.D. thesis, but let's take a brief look. For starters, let's compare the Rahlfs/Göttingen and UBS texts of the Nunc Dimittis. For this particular chapter, unfortunately, Rahlfs has only three witnesses: A (the same as A of the New Testament, of course), T (VII), and 55 (X; this manuscript often stands quite close to B, although there are books where they are quite different). He ends up with a text of the Ode that is identical to the UBS text. But let's attach two apparatus. The first apparatus is simply variants in A, the only (cited) manuscript to have both versions. The second apparatus shows all the variants of major manuscripts in these verses.

	Odes 13 LXX	Luke 2:29-32
29	Νυν απολυεις τον δουλον σου, δεσποτα κατα το ρημα σου εν ειρηνη	Νυν απολυεις τον δουλον σου, δεσποτα κατα το ρημα σου εν ειρηνη
30	οτι ειδον οι οφθαλμοι μου το σωτηριον σου	οτι ειδον οι οφθαλμοι μου το σωτηριον σου
31	ο ητοιμασας κατα προσωπον παντων των λαων	ο ητοιμασας κατα προσωπον παντων των λαων
32	φως εις αποκαλυψιν εθνων και δοξαν λαου σου Ισραηλ	φως εις αποκαλυψιν εθνων και δοξαν λαου σου Ισραηλ
Variants in A
(none; A's texts are identical)
All variants
31	παντων των λαων A 55 ] παντος του λαου T	omit των 1
32		omit εθνων D

This obviously is not very revealing; A's text is identical in both versions, and while the data from Odes indicates a possibility that the original LXX had a significantly different version of verse 32, we can't prove it.

If we turn to the Magnificat, where Rahlfs is able to cite the bilingual uncial R as well as A T 55, here is what we find:

	Odes 9 LXX	Luke 1:46-55
46	Μεγαλυνει η ψυχη μου τον κυριον	Μεγαλυνει η ψυχη μου τον κυριον
47	και ηγαλλιασεν το πνευμα μου επι τω θεω τω σωτηρι μου	και ηγαλλιασεν το πνευμα μου επι τω θεω τω σωτηρι μου
48	οτι επεβλεψεν επι την ταπεινωσιν της δουλης αυτου ιδου γαρ απο του νυν μακαριουσιν με πασαι αι γενεαι	οτι επεβλεψεν επι την ταπεινωσιν της δουλης αυτου ιδου γαρ απο του νυν μακαριουσιν με πασαι αι γενεαι
49	οτι εποιησεν μοι μεγαλεια ο δυνατος και αγιον το ονομα αυτου	οτι εποιησεν μοι μεγαλα ο δυνατος και αγιον το ονομα αυτου
50	και το ελεος αυτου εις γενεαν και γενεαν τοις φοβουμενοις αυτον	και το ελεος αυτου εις γενεας και γενεας τοις φοβουμενοις αυτον
51	εποιησεν κρατος εν βραχιονι αυτου διεσκορπισεν υπερηφανους διανοια καρδιας αυτων	εποιησεν κρατος εν βραχιονι αυτου διεσκορπισεν υπερηφανους διανοια καρδιας αυτων
52	καθειλεν δυναστας απο θρονων και υψωσεν ταπεινους	καθειλεν δυναστας απο θρονων και υψωσεν ταπεινους
53	πεινωντας ενεπλησεν αγαθων και πλουτουντας εξαπεστειλαν κενους	πεινωντας ενεπλησεν αγαθων και πλουτουντας εξαπεστειλαν κενους
54	αντελαβετο Ισραηλ παιδος αυτου μνησθηναι ελεους	αντελαβετο Ισραηλ παιδος αυτου μνησθηναι ελεους
55	καθως ελαλησεν προς τους πατερας ημων τω Αβρααμ και τω σπερματι αυτου εως αιωνος	καθως ελαλησεν προς τους πατερας ημων τω Αβρααμ και τω σπερματι αυτου εως τον αιωνα
Variants in A
49	μεγαλα A*	μεγαλεια A
50	απο γενεας εις γενεαν A	εις γενεας γενεαν A
54	∿ μνησθηναι ελεους after (55) πατερας ημων A
All variants
47		επι ] εν D a aur b c d f ff² g¹ l q r¹ vg
48		επεβλεψεν + κυριος D d
49	μεγαλεια A^c R T 55 ] μεγαλα A*	49 μεγαλα ℵ^,2 B (D d μεγαλα ο θεος) L W aur b c f ff² g¹ l q r¹ vg ] μεγαλεια ℵ¹ A C (D^c) K Γ Δ Θ Ξ Ψ ƒ¹ ƒ¹³ 33 565 579 700 892 1241 1424 𝔐 e ◉ ονομα ] ελεος ℵ*
50	εις γενεαν και γενεαν R T 55 ] απο γενεας εις γενεαν A	και το ελεος αυτου ] omit 579 1010 ◉ εις γενεας και γενεας B C* L W Ξ (d) vg^{amiat,cav,durmach,harl,mediol,sangall} pesh bo ] απο γενεας εις γενεαν 565 1241 vg^split sin geo; εις γενεας και γενεων C³ Δ* Λ 1365; εις γενεας γενεαν A C² D^c K Γ Δ (Θ) 33 𝔐 (a) (b) (c) hark; εις γενεαν και γενεαν ℵ F M S Ψ ƒ¹ ƒ¹³ 157 579 700 1071 1424 aur f ff² g¹ l q r¹; εις γενεας γενεων D*
51	διανοια A T ] διανοιας R 55	διανοια ] διανοιας ℵ^c E F H S 047 0211 118 205 209 579 828 983 1071
52	και A T 55 ] omit R	και rell ] omit sa^mss bo^mss geo²
54	μνησθηναι ελεους here T 55 ] ⥤ after (55) πατερας ημων A ◉ ελεους + αυτου αμην R	ελεους + αυτου 372 aur c e f g¹ r¹ vg^{mediol,sanger,split,val} sin pesh geo¹
55		εως τον αιωνα ] εως αιωνος C F M S Ψ ƒ¹ ƒ¹³ 700 1071 1241 1342 1424 pm aur; εις αιωνα 579 1319

We don't have many differences between the LXX and NT texts here, either, and they perhaps aren't all that informative anyway; A has a singular reading in verse 54 LXX, and a subsingular in verse 50 LXX. But it is fascinating to note that, in verse 49, A* LXX seems to preserve the original reading (at least as understood by the UBS of Luke) whereas A's New Testament text has the Byzantine reading. This seems to imply that, at minimum, A's LXX text is not copied from its NT text; it also hints that the LXX text is older than the Byzantine text!

Finally, let's look at the Prayer of Zachariah, which in Odes follows the Magnificat. R does not contain this poem, so the witnesses to the LXX text are A T 55.

	Odes 9 LXX	Luke 1:68-79
68	Ευλογητος κυριος ο θεος του Ισραηλ οτι επεσκεψατο και εποιησεν λυτρωςιν τω λαω αυτου	Ευλογητος κυριος ο θεος του Ισραηλ οτι επεσκεψατο και εποιησεν λυτρωςιν τω λαω αυτου
69	και ηγειρεν κερας σωτηριας ημιν εν τω οικω Δαυιδ του παιδος αυτου	και ηγειρεν κερας σωτηριας ημιν εν οικω Δαυιδ παιδος αυτου
70	καθος ελαλησεν δια στοματος των αγιων των απ αιωνος προφητων αυτου	καθος ελαλησεν δια στοματος των αγιων απ αιωνος προφητων αυτου
71	σωτηριαν εξ εχθρων ημων και εκ χειρος παντων των μισουντων ημας	σωτηριαν εξ εχθρων ημων και εκ χειρος παντων των μισουντων ημας
72	ποιησαι ελεος μετα των πατερων ημων και μνησθηναι διαθηκης αγιας αυτου	ποιησαι ελεος μετα των πατερων ημων και μνησθηναι διαθηκης αγιας αυτου
73	ορκον ον ωμοσεν προς Αβρααμ τον πατερα ημων [74] του δουναι ημιν	ορκον ον ωμοσεν προς Αβρααμ τον πατερα ημων του δουναι ημιν
74	αφοβως εκ χειρος των εχθρων ημων ρυσθεντας
75	λατρευειν αυτω εν οσιοτητι και διακοσυνη ενωπιον αυτου πασας τας ημερας ημων	λατρευειν αυτω [75] εν οσιοτητι και διακοσυνη ενωπιον αυτου πασαις ταις ημεραις ημων
76	και συ δε παιδιον προφητης υψιστου κληθηση προπορευση γαρ προ προσοπου κυριου ετοιμασαι οδους αυτου	και συ δε παιδιον προφητης υψιστου κληθηση προπορευση γαρ ενωπιον κυριου ετοιμασαι οδους αυτου
77	του δουναι γνωσιν σωτηριας τω λαω αυτου εν αφεσει αμαρτιων ημων	του δουναι γνωσιν σωτηριας τω λαω αυτου εν αφεσει αμαρτιων αυτων
78	δια σπλαγχνα ελεους θεου ημων εν οις επεσκεψατο ημας ανατολη εξ υψους	δια σπλαγχνα ελεους θεου ημων εν οις επεσκεψαται ημας ανατολη εξ υψους
79	επιφαναι τοις εν σκοτει και σκια θανατου καθημενοις του κατευθυναι τους ποδας ημων εις οδον ειρηνης	επιφαναι τοις εν σκοτει και σκια θανατου καθημενοις του κατευθυναι τους ποδας ημων εις οδον ειρηνης
Variants in A
68		επεσκεψατο ] επισκεψατο A
69		τω οικω A ◉ του παιδος A
70		των απ αιωνος A
74	∿ ρυσθεντας after (75) ημερας ημων A	εχθρων ] των εχθρων ημων A
75	ημων add (from 74) ρυσθεντας A
All variants
68		omit κυριος 𝔓⁴ W a b c ff² g¹ l r¹ vg^{am,cav,durmach,harl,sangall} sin ◉ επεσκεψατο ] επισκεψατο A ◉ τω λαω ] του λαου W aur b c f ff² q r¹ vg^val geo
69		οικω 𝔓^4-vid ℵ B C D L M W ƒ¹ ƒ¹³ 22 28 33 565 579 700 892 1071 1241 ] τω οικω A F G H K S Γ Δ Θ Ψ 0130 157 1010 1424 𝔐 ◉ παιδος 𝔓⁴ ℵ B C D L W 565 579 892 ] του παιδος A C F G H K M S Γ Δ Θ Ψ 0130 ƒ¹ ƒ¹³ 33 157 700 1071 1241 1424 𝔐
70		απ αιωνος 𝔓⁴ B L R Δ 0130 ƒ¹³ 33 579 (ℵ W των αγιων απ αιωνος αυτου προφητων) (D αγιων προφητων αυτου των απ αιωνος) ] των απ αιωνος A C F G H K M S Γ Δ Θ Ψ ƒ¹ 157 565 700 892 1071 1241 1424 𝔐
71		εξ εχθρων ημων και εκ χειρος ] εκ χειρος εχθρων ημων και D d sin
72		και ] omit D aur sa geo
74	ρυσθεντας here T 55 ] ⥤ after (75) ημερας ημων A	εχθρων ℵ B L W ƒ¹ ƒ¹³ 565 892 (e) ] εχθρων ημων D 33; 0130 22 των εχθρων 0130 22; των εχθρων ημων A C (K 579 παντων των εχθρων ημων) Γ Δ Θ Ψ 0177 700 1010 1241 1424 𝔐
75	ημερας ημων T ] ημερας ημων ρυσθεντας A; ημερας της ζωης ημων 55	πασαις ταις ημεραις 𝔓⁴ B L W 565 579 e aur b c f ff² g¹ l q r¹ vg ] πασας τας ημερας ℵ A C D F G H K M S Γ Δ Θ Ψ 0130 0177 ƒ¹ ƒ¹³ 33 157 700 892 1071 1241 1424 𝔐 ◉ πασα(ι)ς τα(ι)ς ημερα(ι)ς 𝔓⁴ ℵ A B C D F K L R W Δ Π Ψ 0130 0177 22 33 565 579 892 1010 pm it vg pesh hark sa bo goth ] add της ζωης G H M S Γ Θ ƒ¹ ƒ¹³ 157 700 1071 1424 pm sin
76	δε A ] omit T 55	δε 𝔓⁴ ℵ B C D L R W 0177 33 579 d ] omit A F G H K M S Γ Δ Θ Ψ ƒ¹ ƒ¹³ 28 157 565 700 892 1010 1071 1241 1424 𝔐 ◉ ενωπιον (𝔓⁴ ενωπιον του) ℵ B W 0177 ] προ προσοπου A C D K L Γ Δ Θ Ψ 0130 ƒ¹ ƒ¹³ 28 33 157 565 579 700 892 1071 1241 1424 𝔐
77		αυτων ℵ B D F G H K L R* (W 0177 565 αυτου) Γ Δ Ξ 0130 ƒ¹³ 33 157 700 892 1241 1424 𝔐 ] ημων A C M R^c Θ Ψ ƒ¹ 22 28 579 1071 1342 geo²; omit 2542; omit εν αφεσει αμαρτιων αυτων 047
78		επεσκεψαται ℵ* B (L επεσκεψαιται) W Θ 0177 sin pesh hark^marg sa bo ] επεσκεψατο ℵ^c A C D F G H K M S Γ Δ Ξ Ψ 0130 ƒ¹ ƒ¹³ 33 157 565 579 700 892 1071 1241 1424 𝔐 it vg hark^txt
79		επιφαναι (L 543 επεφαναι) ] επιφαναι φως D d (r¹)

This time, we have a rather mixed bag. In verse 75, LXX disagrees with UBS and has a reading found in part but not all of the Byzantine text. In the first variant in verse 76 (add/omit δε), the LXX text of A agrees with UBS, against its own NT text and the Byzantine text, but LXX is divided; in the second text, LXX agrees with A and the Byzantine text against UBS. In verse 77, LXX agrees with A against both UBS and the Byzantine text. In verse 78, LXX agrees with the Byzantine text. It's pretty clear that A did not copy its LXX text from the NT, or vice versa, but it looks as if the LXX text was taken from something that was starting to move toward the Byzantine text without being all the way there.

Appendix: Important Manuscripts (Hebrew and Other)

An irony of the nature of the Hebrew Bible tradition is that it renders individual manuscripts relatively unimportant. Probably the only truly irreplaceable Old Testament manuscript is Vaticanus -- and it's LXX, not Hebrew! But the list below catalogs a few widely-used texts, both Hebrew and versions.

It is worth noting that the catalogs of manuscripts of the Hebrew Bible are not nearly as tidy as those of the New Testament. There is no really universal nomenclature for Hebrew manuscripts, and LXX manuscripts have a rather incoherent system. Qumran manuscripts are usually noted by cave number, book, and a letter (e.g. 1QIs^a means "First Qumran cave, Isaiah, manuscript A"; 4QSam^b is "Fourth Qumran cave, Samuel, manuscript B"). This notation, however, isn't used in the Biblia Hebraica, which uses the notation Q^a, etc. BHS uses letters for a few major Hebrew manuscripts (e.g. L=Leningradensis), or letters with superscripts (e.g. V^p is the Petersburg codex of the Prophets). Many manuscripts aren't even directly cited; they are simply cited as variants in some edition or other.

LXX manuscripts have sometimes been known by letters and sometimes by numbers; at present, the Rahlfs system is most widely adopted, but it is (to say the least) confusing. Major uncials are known by letters, just as in the New Testament, so A, B, C, and ℵ are the same manuscripts as in the New Testament (although it is common to use the symbol S for ℵ, to avoid confusion with the Hebrew letter and, I suspect, make typesetting easier). Fragmentary uncials and papyri are known by numbers, from 800 to 1000 -- e.g. 911 is the third century papyrus frequently cited for Genesis by Rahlfs. Minuscules occupy the remaining Arabic numbers -- the numbers 1001 to 2000 are for late witnesses to the Psalms, numbers from 2001 on up are early witnesses to the Psalms, and numbers from 1 to 799 are minuscule witnesses not to the Psalms (which are far fewer than Psalms manuscripts -- Psalms witnesses outnumber other LXX texts by perhaps 10:1. LXX Psalms manuscripts are more numerous than witnesses of most parts of the New Testament, although of course Gospels manuscripts are more common still).

1QIs^a * Hebrew: The great Isaiah scroll from Qumran. The largest and most complete of all the Dead Sea Scrolls. It is not dramatically different in text from the Masoretic Text, but there are some divergences, and the orthography is quite different. The scroll thus serves both to affirm the general quality of MT and to demonstrate that transmission was not entirely perfect.
The scroll contains all of Isaiah apart from some very slight damage. The scroll uses 17 sheets of leather to make a scroll 0.26 meters tall and 7.34 meters long. There are 54 columns of about 30 lines. Peculiarly, the form of the text changes right around the end of chapter 33; the orthography is much fuller in the second half. It has been speculated that two different exemplars were used. Since Isaiah is a long book, it is not unlikely that it was sometimes copied onto two scrolls, and that 1QIs^a was copied from two such.
4QSam^a * Hebrew: One of three Qumran manuscripts of (1-2) Samuel. It is relatively fragmentary (contains, according to Eugene Charles Ulrich, Jr., The Qumran Text of Samuel and Josephus, p. 271, 1 Samuel 1:11-13, 22-28, 2:1-6, 8-11, 13-36, 3:1-4, 18-20, 4:9-12, 5:8-12, 6-1-7, 12-13, 16-18, 20-21, 7:1, 8:9-20, 9:6-8, 11-12, 16-24, 10:3-18, 25-27, 11: 1, 7-12, 12:7-8, 14-19, 14:24-25, 28-34, 47-51, 15:24-32, 17:3-6, 24:4-5, 8-9, 14-23, 15:3-12, 20-21 25-26, 39-40, 26:10-12, 21-23, 27:8-12, 28:1-2, 222-25, 30:28-31, 31:2-4; 2 Samuel 2:5-16, 25-27, 29-32, 3:1-8, 23-39, 4:1-4, 9-12, 5:1-16, 6:2-9, 12-18, 7:23-29, 8:2-8, 10:4-7, 18-19, 11:2-12, 16-20, 12:4-5, 8-9, 13-20, 30-31, 13:1-6, 13-34, 36-39, 14:1-3, 18-19, 15:1-6, 27-31, 16:1-2, 11-13, 17-18, 21-23, 18:2-7, 19:7-12, 20:2-3, 9-14, 23-26, 21:1-2, 4-6, 15-17, 22:30-51, 23:1-6, 24:16-20).
Despite its fragmentary state, 4QSam^a is probably textually the most important of the Dead Sea Scrolls. The importance lies in the fact that it contains a Hebrew text clearly closer to the LXX than to the MT (although not identical to either). According to McCarter, following Cross, it is especially close to the type called "Lucianic," with some similarities also to the Old Greek of Vaticanus. The differences from the Old Greek are however substantial -- for example, after 1 Sam. 10:27, it includes a long addition about Jabesh-Gilead and Nahash the Ammonite (adopted by NRSV even though both MT and LXX omit it). On the other hand, it omits 2 Samuel 5:4-5, found in both MT and LXX. It is extremely unfortunate that it lacks 1 Samuel 18 and 19 in their entirety, so we cannot compare its version of the story of David and Goliath with that of Vaticanus. Still, the existence of this third text makes it clear that both LXX and MT are valuable witnesses to Samuel, even where 4QSam^a is not extant, and it gives us a third independent witness where it is extant.
To have some idea of the textual influence of 4QSam^a, it is instructive to look at the footnotes of the NRSV. In all of Isaiah, only 20 readings are adopted based on either of the Qumran Isaiah scrolls, and in one other case, the editors of NRSV were influenced by a Qumran manuscript (that is, while they did not follow it against all other witnesses, its reading was part of the basis for their decision). By contrast, my count was that, in 1 Samuel alone, 27 readings were adopted in NRSV based on 4QSam^a, and seven others influenced by it. Given that Isaiah is about twice as long as 1 Samuel, and that 4QSam^a covers at most a quarter of 1 Samuel, 4QSam^a was roughly eleven times as influential as the Isaiah scrolls. Hence the claim that 4QSam^a is the most textually important scroll. Patrick W. Skehan, "Qumran Manuscripts and Textual Criticism" (reprinted in Frank Moore Cross and Shemaryahu Talmon, eds, Qumran and the History of the Biblical Text) says "From the twofold viewpoint of the amount of text preserved, and the direct value of that text as a source of fruitful criticism of the Masoretic tradition, the Books of Samuel are in fact unique at Qumran" (p. 214). There are several Qumran scrolls of Samuel, all of which contribute somewhat to that assessment, but 4QSam^a is the most important of them all.
4QJer^b * Hebrew: Now listed by Tov as 4QJer^b, 4QJer^d, and 4QJer^e; also 4Q71 and 4Q72a. Like 𝔓⁵² in the New Testament, these manuscripts are more important for their existence than for their actual content: Just as 𝔓⁵² proves that the gospel of John was in existence early in the second century, the Jeremiah scraps prove -- or at least provide strong evidence -- that the LXX version of Jeremiah is based on a Hebrew original; it is not just the result of some strange translation practice. (Recall that LXX Jeremiah is substantially shorter than MT Jeremiah, and also has many changes in order -- notably, it relocates the oracles against foreign nations, but there are other rearrangements of lesser scope.) The three Jeremiah fragments are all very small; the largest has only seven lines of about thirty characters each, and the top three lines are all badly damaged. But even these tiny scraps are sufficient to show an arrangement of text which agrees with LXX against MT, and also to show some readings characteristic of LXX. The effect of this is to demonstrate that LXX Jeremiah must be taken seriously as a source, since that recension did exist in Hebrew. The three fragments contain small portions (in no case as much as a complete verse) of Jeremiah 9:22-10:21, 43:3-9, and 50:4-6.
Murabba'at 88 * Hebrew: A scroll of the Minor Prophets found in 1955 at the Wadi Murabba'at. The caves there were in use as late as the Bar Kochba revolt. The scroll is much damaged, but it agrees strongly with MT in the text (although not in details such as paragraphing). Textually perhaps not of great importance, but if several of the Qumran scrolls show that the MT was early, Murabba'at 88 confirms that it was early and used in several places.
The Cairo Codex * Hebrew: A Hebrew manuscript of the Former and Latter Prophets. The colophon says it was written by Moses ben Asher in 895 C.E. This makes it the earliest of the surviving manuscripts of the Ben Asher family -- generally considered to be the embodiment of the Masoretic Text (although questions have been raised about whether the Cairo manuscript truly belongs to the Ben Asher recension). It is kept in Cairo, and as a result it has not always been available for the use of Jewish or Christian scholars.
The Petersburg Codex * Hebrew: The title of this codex is obviously somewhat confusing; it is not the same as the Leningrad Codex described below. (In a critical apparatus, this manuscript may be referred to as V^p; the Leningrad Codex may be L.) A codex of the Later Prophets (Isaiah, Jeremiah, Ezekiel, and the Twelve). Its colophon is dated 916 C.E., making it the earliest of the major Masoretic Hebrew codices. Interestingly, instead of the common Tiberian pointing, it uses the Babylonian pointing. This does not affect the text, but it gives us our best insight into this alternative pointing scheme.
The Aleppo Codex * Hebrew: Originally a complete Old Testament, which acording to its colophon was written by Shelomo ben Buya'a (who lived in the early tenth century; another of his manuscripts is dated to 930 C.E.) and had the pointing and masora supplied by Aaron ben Moses ben Asher, the last of the great scribal line, himself. It seems to have been kept as a model codex -- used only on special days and for reference in matters of particular textual uncertainty. Maimonides seems to have referred to this volume as being particularly accurate. Thus the Aleppo Codex could almost be said to be the Masoretic Text.
Sadly, it has not been as available as might have been hoped. Originally located in Jerusalem, it apparently travelled to Cairo, and came to Aleppo by about the fifteenth century. There it was kept secret, so that no photographs could be taken. In the mid-Twentieth century, it was feared lost due to the strife in Syria, Jordan, and the new state of Israel. It has now been recovered, but about a third of its contents appear to have been lost. (A lesson, surely, for all who try to keep manuscripts from the scholarly world!) It is believed that it originally had 487 pages; 193 were missing when it was rediscovered, including the beginning and end and some pages in the middle. Most of the book is written in three columns per page, although some sections (mostly poetic) are written in two columns.
The Leningrad Codex * Hebrew: Other than the Aleppo Codex, considered to be the best text of the Masoretic Text in existence (and more widely cited because it has been available to scholars). The colophon says it was written in around 1009 C.E. (there are five dates in the colophon, based on the creation, the exile of Jehoiakim, the Seleucid Era, the destruction of the second Temple, and the beginning of the Islamic Era; four of the five dates indicate a time between 1008 and 1010; the date based on the exile of Jehoiakim seems to claim a date of 846 C.E., but since the Jews lost track of time in the second Temple era, this can safely be discounted). The copy says it was written by Samuel ben Jacob based on manuscripts attributed to Aaron ben Moses ben Asher. There has been dispute about its text, because it shows many corrections; it is possible that it was originally copied from some other manuscript and corrected toward the ben Asher type. But there seems no doubt that the final text is of the ben Asher recension.
The Cambridge Codex * Samaritan Hebrew: Probably the oldest copy of the Samaritan Pentateuch now in existence, since it contains a note saying it was sold in 1140-50 C.E. and was probably written somewhat earlier, making it about as old as the earliest manuscripts of the MT Hebrew. It is in codex form.
Note that this is not the famous Abisha' scroll of the Samaritan Pentateuch, which contains an inscription claiming it was written by Aaron's great-grandson in the thirteenth year of the settlement of Canaan. The Abisha' manuscript was not made accessible to scholars until quite late, and the evidence seems to indicate that it is composite. The beginning part of Abisha' (through Numbers 34) is thought to come from around the fourteenth century; the last section of Numbers and all of Deuteronomy is older -- but probably no older than the Cambridge codex.
The Nahal Hever Minor Prophets * Greek: This manuscript is known by many symbols; it is Rahlfs 943; its official designation is 8HevXIIgr (i.e. Nahal Hever, Cave 8, 12 prophets, Greek) or sometimes 8Hev1. Barthélemy called it R.
This manuscript, although rather short (as it stands, it has only 24 columns, containing portions only of Jonah, Micah, Nahum, Habakkuk, Zephaniah, and Zechariah, written in the hands of two different scribes) has revolutionized LXX studies almost as much as Qumran revolutionized MT studies. Studied by D. Barthélemy in particular, it is clearly based on a Hebrew text which is approaching MT. But it is the translation technique which is most important. This was the first clear example of the kaige recension, which has now been recognized in many so-called LXX manuscripts and which perhaps gave rise to the "Theodotian" text. Thus, although the manuscript is not particularly important textually (it's too close to MT, and too short), it is a key to the history of the Greek versions.
It is fascinating to note that this Greek manuscript was found in the "Cave of Horrors" at Nahal Hever, where many Jews died, often horribly, during the revolt against Rome. It was a significant surprise to find a Greek Bible there -- but it is noteworthy that it was not the (Old Greek) LXX; it was much more precise and Aquila-like.
Nahal Hever, on the west shore of the Dead Sea some distance south of En-gedi, is a wadi with a series of caves along the side. These caves, including the "Cave of Horrors" where the manuscript was found (so named because about forty people starved to death within it), were occupied during the second Jewish revolt (the Bar Kochba rebellion), meaning that the manuscript must come from before 135 C.E.. But it appears to have been well-used before the revolt, and this plus paleographic examinations have caused most scholars to suspect a date in the first century C.E. or even earlier. The most typical estimate would probably be the first half of the first century. It is clear that it is a revision of the Old Greek, intended to be at once a more literal rendering of the Hebrew and closer to the MT; it is also evident that, unlike the original LXX, this translation was still accepted by Jews in the second century.
Codex Vaticanus (B) * Greek: The earliest nearly-complete codex of the LXX Greek, it dates from the fourth century (being, of course, the same as B/Vaticanus of the New Testament). It now lacks Genesis 1:1-46:28; (2 Samuel=) 2 Kingdoms 2:5-7, 10-13; Psalms 105:27-137:6 (=106:27=138:6), and it seems never to have contained 1-4 Maccabees, or the Prayer of Manasses. Although for much of the Old Testament it is in terms of age the oldest surviving witness, its great value lies in the fact that, with some exceptions, it represents the Old Greek text, not the Hexaplairic recension. There are however a few places where its text is considered inferior -- notably Deuteronomy, Isaiah, Chronicles, and 1-2 Esdras, but see below. The exceptional quality of its text in the other books is accepted by every source I have checked. Its great importance springs from the fact that its text is independent of the Hebrew and relatively free of later "improvements."
P. Kyle McCarter, Jr. considers Vaticanus so important that it is the only Old Testament manuscript he examines book-by-book. In his book Textual Criticism, he has an appendix titled "Textual Characteristics of the Books of the Hebrew Bible." This describes the general nature of the MT and LXX of those books, and lists significant other witnesses such as Qumran scrolls. In general he calls Vaticanus the best witness to the Old Greek. There are exceptions: In Deuteronomy B has hexaplairic readings. In Joshua, it is shorter than MT. In Judges and Ruth, it is kaige. In 2 Samuel 10:1-1 Kings 2:11 and 1 Kings 22:1-2 Kings, it is again kaige. Chronicles is kaige and perhaps "troubled"; Ezra and Nehemiah are also kaige. In Isaiah he calls B expansionistic and with hexaplairic readings. Jeremiah is mixed, probably not Old Greek in chapters 29-52. Lamentations is kaige. Ezekiel 28-39 may not be Old Greek. In Job there are kaige insertions into the short Greek text. The Song of Songs is kaige. Ecclesiastes may be Aquila (there does not seem to be a true LXX rendering of this book). Daniel (as in most LXX MSS) is kaige or perhaps Theodotian.
Codex Alexandrinus (A) * Greek: The most substantial early codex of the LXX Greek, dating from probably the first half of the fifth century. Its primary lacuna is at the beginning; it has lost the first few chapters of Genesis. Its text in general is not as valuable as that of Vaticanus, frequently revealing hexaplairic influence, but it some books, such as Isaiah, it is superior.
Codex Sarravianus (G) * Greek: Also sometimes called Codex Colberto-Sarravianus, and dated from the fifth, or possibly late fourth, century. Symbol is G. The varying names arise because portions are in Leyden (130 leaves), Paris (22 leaves), and a single leaf in St. Petersburg. Its contains Genesis through Judges with many lacunae (we have portions of Genesis 31-36, Exodus 36-end, Leviticus 1-18, 24-end, Numbers 1-7, 11-20, 25-26, 29-end, Deuteronomy 4, 7-19, 18-21, Joshua 9-19, Judges 9-10, 15-17, 19-21). Its text is not in itself particularly important, being primarily hexaplaric (note that Rahlfs cites it only very rarely, and BHS not at all), but it is important for the fact that it includes Origen's hexaplairic signs, although seemingly not with perfect accuracy. This means that it, along with the Syro-hexaplar, can be used to try to detect hexaplairic contamination in other Greek manuscripts. Tischendorf identified no fewer than seven correctors, the first three of which (who did the largest share of the work) were all early: A was contemporary with the scribe, B worked on Deuteronomy and Judges probably in the fifth century, and C went over Numbers in the sixth century.
Codex Marchalianus (Q) * Greek: Vatican Library, MS. Greek 2125. Along with G, one of the most important sources for the Hexapla. Written in a beautiful hand usually dated to the sixth century. The letter forms are described as being of the Coptic style. It contains the Prophets, with the minor prophets preceding the major. The original text is thought to be Alexandrian/Hesychian, with some Hexaplar influence. More important is a work of a very early corrector, who went over the manuscript to add Hexaplaric symbols and marginalia (including many readings from the other translations). It is these marginal materials which give the manuscript most of its value.
The Milan Palimpsest * Greek: Milan, Ambrosian Library, O 39 Sup. A palimpsest. The upper text is of little value, bueing merely a late (thirteenth century or later) copy of the Orthodox service book known as the Οκτωηχος. But two of the lower leaves are copies of a text based, in some form, on Origen's hexapla. The Hexapla of course contained six columns: Hebrew (in Hebrew letters), Hebrew (transcribed in Greek letters), Aquila, Symmachus, LXX, and Theodotian (plus occasional other versions). The Milan fragments include five of these columns: Hebrew in Greek letters (except that the tetragrammaton is written in Hebrew), Aquila, Symmachus, LXX, and -- it is believed -- Quinta. (This ia part of why there are so many vexed questions about Theodotian, Quinta, and kaige.)
The total text is minimal -- about 150 verses of Psalms. And the copy is much later than Origen's original -- it's thought to be ninth or tenth century. But it gives us one of our few looks at the actual format of the Hexapla.
b o c₂ e₂ * Greek: This set of letters may be the most fetishistic in all Old Testament criticism. These four minuscules (actually five manuscripts) of the former prophets (which I group because they form a family and are almost always referred to together) properly should be referred to by their numbers, 19 (Chigi R.vi.38, X? century), 82 (Coisl. Gr. 3, XII century), 93 (British Library Reg.i.D.2, XIII century), 108 (Rome, Vatican Library Greek 330, XIII century) and 127 (Syn. 31a Vlad. 1, X century) -- but almost all articles use the letters, the symbols used by Brooke and McLean. The importance of the four lies in the fact that they seem to preserve Lucian's text. This text is of great importance in 1-4 Kingdoms, especially in the places where Vaticanus is kaige and we have little evidence for the Old Greek except the Lucianic text. The true core of the group is b(=19+108) and e₂(=93), because both o(=82) and c₂(=127) sometimes depart the group -- e.g. 127 is the closest ally of B in Chronicles.
55 (h) * Greek: Rome, Vatican Regin. Greek 1. The "h" of Brooke and McLean. Dated century X by Kenyon, XI by Brooke and McLean. Poetic and Historical books: Kings, Ezra, Nehemiah, 1 Esdras, Judith, Esther, Tobit, Maccabees, Job, Psalms. Although a minuscule, the text seems to be very old; in Ezra, it stands very close to B, and in the Psalms, Kenyon says it is part of a group consisting of A and the Freer Psalter. It probably deserves to be examined in greater detail. My feeling, in examining its text in the historical books, is that it was copied from a manuscript that was corrected from a B-like text toward the standard late text of the LXX. This correction was much more thorough in the earlier historical books than the later; the corrector I think became much less diligent as he worked his way through Chronicles. At the beginning of 1 Chronicles, e.g., 55 is almost worthless; by the end of 2 Chronicles, it is second only to 127 as an ally of B, and in Esdras, as noted, it is B's closest supporter.
c e * Greek: Rahlfs 64 (Paris, National Library 2) and 52 (Florence, Laurentian 44). I've examined the text of these two only in Chronicles, but in those books they form a very interesting pair, often going with the late majority but also having many early and unusual readings which do not go with any of the LXX uncials or with the Lucian group or the Complutensian group.
d (j) p q t (z) * Greek: In Rahlfs numbers, 107 243 106 120 134 554. A group of manuscripts which seem to be associated with the Complutensian Polyglot. The best of the group are probably p and q; d is prone to very erratic errors, t seems slightly mixed, and j and z belong to it only part of the time.
Codex Amiatinus (A) * Latin: Yes, the same manuscript as A of the Vulgate in the New Testament. Just as in the New Testament, Amiatinus is perhaps the most important witness to the version (although the Vulgate is of generally less value in the Old Testament than the New). The text is slightly mixed, and rather inferior in the Psalms (where it is said to be Irish), but mostly very good.

Appendix: Greek manuscripts cited by BHS, BHK, and Rahlfs

To completely catalog the materials cited in BHS, BHK, and Rahlfs is beyond the scope of this article (and to do Göttingen or Brooke and McLean even harder), but I will attempt to supply a useful quick reference. To begin with, let us examine the use of the Greek manuscripts in Rahlfs. Following the tables is a description of these manuscripts, plus others cited in the Hebrew critical editions.

Frequency of Citation in Rahlfs

This is not a perfect table. Ideally, one would cite places where each manuscript agrees with the others, or goes with or against the Rahlfs text, or something of the kind. That is not what this table does. Instead, it simply counts how many times Rahlfs explicitly cites each manuscript. In the books where Rahlfs cites only two uncials (Exodus, say), this can to a first approximation be treated as a count of the number of places each manuscript disagrees with the Rahlfs text. But even this is imperfect, for two reasons. First, I counted variants in both the main text and correctors. Second, there is the issue of which variants count. Many LXX variants repeat (e.g. two different manuscripts may have two different spellings of a particular name, and use their own spelling repeatedly). In general I have tried to count such variants only once -- but often this depends on how Rahlfs sets up the apparatus. Then, too, Rahlfs sometimes includes one of the uncials in one of his manuscript groups (e.g. in 1 Kingdoms, he includes A with O, the Origen group). When Rahlfs cites O without A (i.e. O^-A), I have not counted this as a variant involving A. I have, however, counted the rare occasions when Rahlfs counts one of the other members of the O group (e.g. O³⁷⁶), because they are rarely cited and it is perhaps significant when they are.

To emphasize, this is not a full analysis of the Rahlfs apparatus -- and, in any case, Rahlfs is not the last word on the text of LXX. But this gives some idea of what he considers worth citing, and it is a comprehensive list of which manuscripts are cited in each book.

Pentateuch

Note that B is defective for most of Genesis, and ℵ for almost the whole of the Pentateuch. N+V (which Rahlfs always cites as V), which is overall his most-cited manuscript other than A, B, and ℵ, does not begin until Leviticus 13:59, and has other lacunae in the Pentateuch. In Genesis, therefore, A, M, and 911 are the primary sources -- and even A fails for a few chapters.

	BOOKS
MSS	Genesis	Exodus	Leviticus	Numbers	Deuteronomy	Totals
A	619	925	451	887	908	3790
B	62	404	289	398	804	1957
D	13	--	--	--	--	13
F	--	1	3	2	4	10
G	--	--	--	1	2	3
L	5	--	--	--	--	5
M	127	14	27	6	6	180
S (ℵ)	15	--	--	3	--	18
(N+)V	--	--	--	5	7	12
W	--	--	--	--	21	21
911	126	--	--	--	--	126

Former Prophets and associated books

Note that ℵ is defective for these books. Also, Rahlfs cites several versions in this section -- most notably the Syriac in connection with the Hexapla, but even versions as obscure as the Ethiopic. I have not counted these witnesses. Also note that Rahlfs prints two texts of Judges, the B-text and the everything-else text, and does the same with parts of Joshua; this reduces the number of citations in both books, especially of B, since there can be few variants in a text which is mostly a diplomatic edition of the B text!

	BOOKS
MSS	Joshua	Judges	Ruth	1 Kingdoms	2 Kingdoms	3 Kingdoms	4 Kingdoms	Totals
A	1008	411	76	486	495	505	584	3565
B	374	67	25	316	473	638	634	2527
G	4	1	--	--	--	--	--	5
M	11	19	3	20	40	7	--	100
V	26	4	--	18	19	28	66	161
W	69	--	--	--	--	--	--	69
93	--	--	--	--	1	--	1	2
127	--	--	--	1	1	3	1	6
247	--	--	--	9	24	42	--	75
376	--	--	--	20	8	--	--	28

Other Historical Writings

Note that ℵ is defective for large portions of 1 Chronicles -- and seems to have a rather wild text even where extant; at least, Rahlfs cites far more variants for it than for A and B. In 1 Chronicles in particular, Rahlfs places extra weight on V as a result. This is not a possibility in the books of Esdras; V is defective for the end of 1 Esdras and the first several chapters of 2 Esdras.The rate of variants in 1 Chronicles 1-9 is in any case very high (note how many more variants there are in 1 Chronicles than in 2 Chronicles), due to all the Hebrew names so easily distorted in Greek. Similarly, V gains in importance for Maccabees because B lacks all four books and ℵ has only two. The low number of variants for ℵ in Tobit is due to the fact that Rahlfs prints two texts of that book, one from ℵ and one from all the other witnesses.

	BOOKS
MSS	1 Chronicles	2 Chronicles	1 Esdras	2 Esdras	Esther	Judith	Tobit	1 Maccabees	2 Maccabees	3 Maccabees	4 Maccabees	Totals
A	1030	623	290	507	406	246	143	733	488	224	447	5137
B	1282	666	569	901	48	165	134	--	--	--	--	3765
S (ℵ)	419	--	--	704	127	416	45	948	--	--	400	3059
V	213	46	17	100	12	10	8	180	297	109	21	1013
19	--	--	--	1	--	--	--	--	--	--	--	1
58	--	--	--	--	151	62	--	--	--	--	--	213
93	--	--	--	2	158	--	--	--	--	--	--	160
108	--	--	--	2	--	--	--	--	--	--	--	2
583	--	--	--	--	6	--	--	--	--	--	--	6

Wisdom Books

It is well known that the Old Greek of Job was much shorter than the Hebrew. It is not so often mentioned that this short text is preserved primarily in the Old Latin and Sahidic Coptic. The surviving Greek manuscripts have generally been based on Origen's hexaplairic text, meaning that they are contaminated by Theodotianic readings. But little attention has been paid to the way in which this was done. Working over the critical apparatus of Job, one cannot help but observe the extent to which we find B and ℵ on the one hand going against A on the other. V, insofar as it is extant (it has only the last dozen or so chapters) goes with A. The differences are so substantial that I have to suspect recensional activity.

	BOOKS
MSS	Psalms	Odes	Proverbs	Ecclesiastes	Song of Solomon	Job	Wisdom of Solomon	Sirach	Psalms of Solomon	Totals
A	674	58	297	145	49	1427	139	576	--	3365
B	503	--	197	209	56	318	138	502	--	1923
C	--	--	6	2	--	34	6	16	--	64
R	31	33	--	--	--	--	--	--	--	64
S (ℵ)	908	--	386	251	90	648	310	1040	--	3633
T	--	13	--	--	--	--	--	--	--	13
U	7	--	--	--	--	--	--	--	--	7
V	--	--	39	52	11	394	28	96	--	620
55	--	7	--	--	--	--	--	--	--	7
1098	1	--	--	--	--	--	--	--	--	1
2013	4	--	--	--	--	--	--	--	--	4
2018	1	--	--	--	--	--	--	--	--	1

Prophets

It is interesting and somewhat surprising to look at how Rahlfs edited the prophetic books. In Isaiah, he has an apparatus which quite regularly cites four witnesses, A B Q S. It almost resembles the early Nestle editions. But he largely ignores Q in the other major prophets. (This is apparently because Q is widely regarded as Alexandrian/Hesychian in Isaiah and the minor prophets, Hexaplairic elsewhere.) We see similar variations in the importance he gives to the different witnesses of the various minor prophets. To account for this, I have made up separate tables for, for instance, The Twelve and the books associated with Jeremiah, and then created a master table with totals for each of these categories.

The Twelve

	BOOKS
MSS	Hosea	Amos	Micah	Joel	Obadiah	Jonah	Nahum	Habbakuk	Zephaniah	Haggai	Zechariah	Malachi	Totals
A	106	107	91	51	25	44	36	43	48	22	197	44	814
B	104	112	79	35	18	26	23	32	30	18	148	34	659
Q	38	59	56	38	15	28	15	28	29	10	124	29	469
S (ℵ)	--	--	--	61	26	55	50	69	51	38	205	61	616
V	7	8	4	10	2	3	3	3	6	1	16	3	66
W	--	44	54	30	13	31	15	29	29	19	123	23	410
393	--	--	--	--	--	--	--	--	--	--	3	--	3

Writings Associated with Jeremiah

	BOOKS
MSS	Jeremiah	Baruch	Lamentations	Letter of Jeremiah	Totals
A	1320	108	101	85	1614
B	631	30	60	22	743
Q	64	2	1	3	70
S (ℵ)	1248	--	60	--	1308
V	17	1	--	--	18
86	1	--	--	--	1
393	1	2	--	--	3

Writings Associated with Daniel

Note that Rahlfs has two texts of Daniel and the related books, the LXX text of 88 and the Syriac and the Theodotianic text of everything else. He in fact uses 88 more than any other witness (since his LXX text is almost identical to that of 88), but cites it less often because there are no other witnesses to cite against it.

	BOOKS
MSS	Daniel	Susanna	Bel and the Dragon	Totals
A	395	25	29	449
B	258	18	10	286
Q	45	1	1	47
V	10	1	1	12
88	97	3	6	106
393	2	--	--	2

The Prophets Summarized

	BOOKS
MSS	The Twelve	Isaiah	Jeremiah	Ezekiel	Daniel	Totals
A	814	789	1614	1348	449	5014
B	659	1095	743	828	286	3611
Q	469	482	70	115	47	1183
S (ℵ)	616	1101	1308	--	--	3025
V	66	50	18	22	12	168
W	410	--	--	--	--	410
86	--	--	1	--	--	1
88	--	--	--	--	106	106
393	3	21	3	2	2	31
918	--	4	--	--	--	4

Grand Totals

The table below shows all the witnesses cited in Rahlfs, with the number of times cited.

A	20871
B	13783
C	64
D	13
F	10
G	8
L	5
M	280
Q	1183
R	64
S (ℵ)	9735
T	13
U	7
V	1974
W^Octateuch	90
W^Prophets	410
19	1
55	7
58	213
86	1
88	106
93	162
108	2
127	6
247	75
376	28
393	31
583	6
911	126
918	4
1098	1
2013	4
2018	1

If we wish to know which witnesses Rahlfs is most dependent upon, the list below sorts the above in descending order of citations.

A	20871
B	13783
S (ℵ)	9735
V	1974
Q	1183
W^Prophets	410
M	280
58	213
93	162
911	126
88	106
W^Octateuch	90
247	75
C	64
R	64
393	31
376	28
D	13
T	13
F	10
G	8
U	7
55	7
127	6
583	6
L	5
918	4
2013	4
108	2
19	1
86	1
1098	1
2018	1

The total of the above citations is 49284. Thus, counting the citations of L, O, and the versions, Rahlfs has over 50,000 citations.

Descriptions of the Cited Manuscripts

BHS and BHK use a number of different symbols for the Greek versions -- e.g. α' for Aquila, σ' for Symmachus, θ' for Theodotian, ε' for Quinta, and sometimes οι γ' for the non-LXX translations collectively. LXX as a whole is referred to by a (Gothic) G (ℭ). G* is sometimes used for what this article tends to call the Old Greek. In addition, particular LXX manuscripts are sometimes referred to individually by G with a superscript -- so G^A would be Alexandrinus, while G^B would be Vaticanus. Rahlfs cites Greek manscripts simply by letter or number.

In this list, a witness marked * is cited, at least occasionally, by Rahlfs. Two stars, **, indicates a witness which he cites quite frequently in at least one book.

Symbol		Manuscript
	ℵ	See under S (ℵ)
**	A	Codex Alexandrinus. Same as A of the New Testament. The most complete uncial copy of the Old Testament. See description above. In Esther, Hanhart pairs it with 311; in Psalms, Rahlfs calls it a mixed text and groups it with 55 2019; in Isaiah, Ziegler puts it in an "Alexandrian Group" with Q 26 86 106 710, with 106 being the closest to A; in Jeremiah, he links it with 106 410 and the Arabic version; in Ezekiel, Ziegler files it with 26 544, with 106 410 somewhat more distant; in Daniel, he puts it with 106 584; in the Twelve Prophets, he again makes it Alexandrian, with Q, W, the Coptic versions, and many others, 26 and 106 being closest of all.
**	B	Codex Vaticanus. Same as B of the New Testament. Mostly complete, although with large lacunae in Genesis and Psalms. See description above. It is usually regarded as "Old Greek"; in 1 Esdras, Hanhart pairs it with 55; in Psalms, Rahlfs calls it a a Lower Egyptian text and groups it with S bo; in Isaiah, Ziegler regards it as Origenic, and puts it in the core group O with V; in Jeremiah, Ziegler puts it at the head of a large group that also includes S, the Latins, and the Coptic versions; in Ezekiel, it is the closest to 967; in the Twelve Prophets, Ziegler links it with S V. In Daniel, of course, it has the θ' text, not LXX, and Ziegler links it most closely with 26 46 130 239, the Latins, and the Sahidic.
*	C	Codex Ephraemi Rescriptus. Same as C of the New Testament, although a much smaller fraction of the OT has survived. The only surviving portions are from Proverbs, Ecclesiastes, Song of Solomon, Job, Wisdom of Solomon, and Sirach; for details, see, e.g., Swete. Not considered to be of much value in the Old Testament, although this may be a side effect of its extremely fragmentary state.
*	D	(Not cited by symbol in BHS.) Cotton Genesis. Brought to England in the sixteenth century, and incomplete even then, it was burned in the 1731 Cotton Library fire. The surviving fragments are almost illegible, although one or two of the many illustrations survive well enough to show how beautiful they were before being burned. Most of what is known of the manuscript comes from early collations and descriptions, plus modern attempts to reconstruct the badly burned leaves.
	E	(Not cited by symbol in BHS.) Codex Bodleianus, or the Bodleian Genesis. Properly part of a curious manuscript of probably the tenth century (the British Library dates it to late X), with the beginning portion in uncials, the remainder in minuscules. Tischendorf (who thought it belonged to the ninth century) distributed the parts,leaving Genesis 1:1-14:7, 17:24-20:13, 24:54-42:18 in the Bodleian Library in Oxford, while Genesis 42:18-44:13 is on a single leaf in Cambridge, and contains the transition from uncials to minuscules. The rest of the manuscript, which is rarely cited, extends as far as 3 Kingdoms (1 Kings) 16:28, with lacunae. The larger and later portion is at St. Petersburg; it has the Rahlfs number 509. Scans of the Bodleian portion (Auct. T. infr. 2.1) are now available on the Bodleian site at https://digital.bodleian.ox.ac.uk/inquire/p/403c2c20-5072-4ad2-831d-a43edbdfcf47; it is often rather hard to read due to various sorts of damage.
*	F	Codex Ambrosianus, Milan, Ambrosian Library, S. P. 51. Fifth century. Genesis 31:15-Joshua 12:12, with large lacunae (some filled by a later hand; there are also many corrections). It is sometimes said also to include a tiny portion of Isaiah and Malachi, but Swete says this is a different manuscript. There seem to have been two scribes, one working on the Pentateuch, the other on Joshua.
*	G	(Not cited by symbol in BHS.) Codex Sarravianus. Note that, although this manuscript is described at (relative) length by Kenyon, Swete, and Würthwein, and Kenyon says that it is very important in its "own special department," its text is of no great value except as a witness to Origen's recension (in Leviticus, e.g., it is the only uncial that Wevers cites with a group rather than independently). As a result, Rahlfs and the Hebrew editions cite it very rarely. The Paris portion (fragments of Exodus 39; Leviticus 1, Number 23; MS. Greek 17) can now be viewed at https://gallica.bnf.fr/ark:/12148/btv1b10515747h.
	I	(Not cited by symbol in any major edition.) Codex Bodleianus. Oxfor, Bodleian Auct. D.4.1. Psalms, with catena. The Bodleian web site dates it to 951; other sources simply refer it to the ninth (sometimes tenth) century. The Bodleian web site has an image of a much-battered page showing an illustration of David at https://digital.bodleian.ox.ac.uk/inquire/p/4450d30a-0ee3-4551-823f-2f35bdb93913.
	K	(Not cited by symbol in BHS.) Codex Lisiensis, Leipzig University Library Greek 2. Seventh or century palimpsest with portions of Numbers, Deuteronomy, Joshua, Judges. Upper writing is Arabic. Discovered by Tischendorf. Rahfls lists this as a weak member of the Lucianic group L in Judges but does not cite it explicitly.
*	L	(Not cited by symbol in BHS or BHK.) Vienna Genesis. A purple manuscript, made with silver ink. Written in the fifth or sixth century, with many illustrations. Fragments of Genesis. Very beautiful but of relatively little textual value. We do note with interest that many of the illustrations are based not on actual information in the Bible but on legends which grew up about it; a folklorist might learn more from it than would a textual scholar.
**	M	Codex Coislinianus. Paris, National Library Coislin 1. Probably seventh century, although some have argued for the sixth. Contains Genesis-3 Kingdoms (1 Kings) 8:40, with relatively minor lacunae. Rahlfs cites it fairly often in Genesis, where other witnesses are few, but only very sporadically thereafter (53 times in the rest of the Pentateuch, which still makes it the most-cited witness after A and B, and about a hundred times in the historical books, which makes it the #4 witness behind A, B, and V). Some have said its text is similar to A; Lagarde called it hexaplairic, and there are some hexaplairic symbols in the manuscript. In Leviticus and Numbers, Wevers puts it in a group with 416. In Ruth, Quast places it in the large R group, the other leading members of which are N+V and 55. There are citations in the margin from the New Testament, sometimes cited in the older critical editions as F^a.
	N	Codex Basiliano-Vaticanus. Vatican, Greek 2106. Eighth or perhaps ninth century. Formerly at Basel; now in the Vatican. Half of the a two-volume set, the other half being V (which see, since Rahlfs always cites it under that symbol).
(*)	O	See 918
**	Q	Codex Marchalianus. Rahlfs cites this heavily for Isaiah and the Twelve, only occasionally for Jeremiah, Ezekiel, and Daniel. In Isaiah, Ziegler puts it in an "Alexandrian Group" with A 26 86 106 710; in Jeremiah, Ziegler links it with V and about eight minuscules; in Ezekiel, it plus 88 Hark are the core of the Origenic group O; in Daniel, Ziegler places it with 230 233 541 and the Ethiopic; in the Twelve Prophets, he again makes it Alexandrian, with A W etc.
*	R	(Not cited by symbol in BHS.) Codex Veronensis. Sixth century bilingual (Greek and Latin) copy of the Psalms, with the Greek on the left and the Latin on the right. The Greek, astonishingly, is written in Latin letters! (Presumably this is a copy used in the Latin church and written for someone who wants access to the Greek but can't read it; there are a few other Latin manuscripts of that sort.) The Greek occupies the left side of the page. A few minor lacunae have been supplied by later hands, one of which also added Psalm 151. There are eight miscellaneous poems which are taken from other portions of the Bible. Rahlfs in his hand edition cites it primarily for the Odes, which B and ℵ omit; in the major edition of Psalms, he groups it with several Old Latin witnesses.
**	S (BHS, Rahlfs) or ℵ (BHK)	Codex Sinaiticus. Same as ℵ of the New Testament. The Old Testament portion, however, is very fragmentary; apart from a few scraps of the Pentateuch, the surviving material is all from the later prophets and the Writings, and even these have several lacunae. One of the correctors is said to have worked from manuscripts corrected by Pamphilius based on Origenic manuscripts, but these corrections do not seem to have been based on a particularly good text. Where it is extant, it seems to diverge from Rahlfs's text more than does A, and substantially more than B. In Jeremiah, Ziegler nonetheless links it with B.
*	T	(Not cited by symbol in BHS or BHK.) Codex Turicensis, or the Zürich Psalter. A purple manuscript with writing in gold (headings and initial letters), silver (the main text), and vermillion (marginalia from the Gallican version of the Psalms). 223 of an original 288 leaves survive. Dated to probably the seventh century. Swete, followed by Kenyon, regards it as close to A and even closer to ℵ^c.a, with which it is almost contemporary; Rahlfs puts it with the Lucianic witnesses.
*	U	British Library Papyrus 37/London (cited as Pap. Lond. in BHK). A fragment of Psalms from the seventh century, noteworthy primarily as the first Biblical papyrus recovered (in 1836). It is said to be similar in text to the Sahidic Coptic. Swete says the surviving materials constitute 30 leaves; Kenyon and Würthwein say 32. The surviving material is all from Psalm 10 to Psalm 35. The writing is rather curious; there are accents and breathings, but the scribe writes continuously, not even starting each psalm on a new line! Rahlfs calls it Upper Egyptian, linked to 2013 Sahidic.
**	V	Codex Venetus. Eighth or perhaps ninth century. A two volume manuscript originally containing most if not all of the LXX. The volumes have been separated; the first half, now designated N, is in the Vatican, and is called Codex Basiliano-Vaticanus; the second half, at Venice, is Codex Venetus. Rahlfs and the Göttingen editions cite both halves as V; Brooke and McLean continued to call the Vatican portion N. Both halves are somewhat mutilated, but since N ends with Esther and V starts with Job 30:8, the division appears to have been at the end of Esther. V is effectively complete except for the initial lacuna; N begins with Leviticus 13:59 and has sundry lacunae after that, totalling perhaps the equivalent of a dozen chapters. Kenyon calls the text "Lucianic," but it is rarely linked with the group b o c₂ e₂. Whatever its text, it has significant importance as one of only five once-complete uncial copies of LXX. It has the unique distinction of being the only (formerly) complete uncial copy of LXX which does not contain the New Testament also. (Unless, as seems likely enough, there was a third New Testament volume now lost. The only extant manuscript which could possibly be this third volume is Ψ, which is about the right age but has a completely different format.) Rahlfs cites it extensively for the Maccabean books (all of which are omitted by B and two of them by ℵ), and rather often in Job as well. He cites it at least intermittently in almost every other book of LXX for which it is extant. The result is that, although there is no section where it is one of his primary sources, Rahlfs cites it more than any manuscripts other than A, B, and ℵ. In Ruth, Quast places it in the large R group, the other leading members of which are M and 55; in 1 Esdras, Hanhart pairs it with 245; in Isaiah, Ziegler regards it as Origenic, and puts it in the core group O with B; in Jeremiah, Ziegler links it with Q and about eight less well-known minuscules; in Ezekiel, Ziegler places it in the secondary Lucianic group 𝓁II with 46 449; in Daniel, he places V 62 147 in the core Origenic group O; in the Twelve Prophets, Ziegler puts it with B S.
*	W^Octateuch (Rahlfs) or Θ (BHK)	(Not cited by symbol in BHS.) Codex Freer, or Washingtonensis I. Deuteronomy (lacking 5:16-6:18) and Joshua (lacking 3:3-4:10), from probably the sixth century. Based on the quire numbers (we now have quires ΛΖ to Ν, or 37-60), probably originally contained the whole Octateuch (certainly it contained all five books of the Pentateuch as well as Deuteronomy and Joshua), or even more. Said by Kenyon to agree more with A, and especially with G and 963, plus 54 and 75, than B. This seems to be based on the excellent 1910 edition by Henry A. Sanders (The Old Testament Manuscripts in the Freer Collection: Part I: The Washington Manuscript of Deuteronomy and Joshua) -- which, however, describes a much more complicated situation than Kenyon describes. It is the opinion of Sanders that W (or Θ in Brooke & McLean's edition) has as much value as B or A, but does not align with either. (It would seem Rahlfs did not agree, since he didn't cite it as often as its value would require.) Around 600, a cursive hand added lectionary information. A sample page can be seen on the Freer/Sackler web site at https://www.freersackler.si.edu/object/F1906.272/.
	W (BHK)	(Not cited by symbol in BHS.) Codex Atheniensis. Various of the historical and pseudo-historical books. Not particularly important textually.
	W (BHS)	A fragment of 1 Samuel 18, from the fourth century.
**	W (Rahlfs) or X (Kenyon)	Freer Greek MS. V, at Washington. A very much damaged papyrus codex, originally containing probably 48 leaves; portions of 33 now survive (some relatively complete; others in small pieces). A manuscript of the Minor Prophets, of the third century. Lacks Hosea and Amos 1:1-10, but most of the other books survive nearly complete. A very long, narrow manuscript. Has occasional corrections. Ziegler makes it a loose member of the Alexandrian group also found in A Q etc., most closely linked to 407 410. Samples of three leaves (two nearly complete, one badly fragmented) can be seen on the Freer/Sackler web site at https://www.freersackler.si.edu/object/F1916.768/.
	Γ	Symbol used by Swete for 393 (which see).
	15	Paris, National Library Coisl. Gr. 2. Octateuch. Ninth or tenth century. Described as Hexaplairic, and Rahlfs lists it as a weak member of his group O in Judges, but never cites it individually. Wevers in Leviticus and Numbers places it in oI, a group of weak Origenic witnesses that also include 64 381 618; in Ruth, Quast again places it in oI with 64 381. A few pages have had their margins cut off. (I wonder if this might have been to cut off marginal notes, with which some pages are extensively supplied, in a different hand, although most are untouched except for chapter numbers in a late Latin hand presumably added by a modern owner in France.) There is very little ornamentation, even at the beginnings of books, which merely start at the top of pages with the title in a slightly different hand with dashed lines above and below. Full scans are now available on the Paris web site at https://gallica.bnf.fr/ark:/12148/btv1b105494378.
*	19	Rome, Chigi R.6.38. Swete suggested a tenth century date; more recent evaluations make it eleventh or twelfth century. Pentateuch and historical books. Sister of 108 or nearly; they are so close that Brooke and MacLean cited them under the joint symbol b (when cited individually, 19 was b') -- the famous b of the Lucianic group boc₂e₂. Although Rahlfs lists it as having a hexaplairic text in Judges, and includes it as part of his O group, he never cites it individually. The situation is altogether different in 1-4 Kingdoms, where 19 and 108 have a Lucianic text. Rahlfs lists both among the five Lucianic witnesses cited as L (the the five being 19 82 98 108 127) -- but explicitly cites both 19 and 108 very rarely (in fact, the only explicit citation of 19 appears to be in 2 Esdras 17:3, and even there it is cited with 108). In 1-2 Chronicles and 2 Esdras, Rahlfs lists as Lucianic 19 93 108 and sometimes 121. In 1 Esdras, Rahlfs lists only 19 and 108 as members of L. In Esther, Rahlfs lists 19 and the "textus prior" of 93 and 108 as constituting L -- that is, it is one of the manuscripts having the alternate translation of Esther which was formerly called Lucianic but is now often called "A."; Hanhart lists 19 93 108 319 as the manuscripts of this type (with 392 as a partial supporter), and lists 19-108 and 93-319 as subgroups. In Judith, 19 and 108 are the only witnesses to L. In 1 Maccabees, L is considered to include (19) 64 (93) 236. Lagarde cited it as h. Wevers in Leviticus and Numbers puts it in his b group along with 108 118 314 537; in Ruth, Quast places it in the Lucianic group L, linking it most closely with 108; in 1 Esdras, Hanhart puts it in L along with 108; in 1 Maccabees, Kappler lists it in the lesser Lucianic group l.
	22	British Library Royal 1 B II (Reg. i.B.2). Codex Pachomianus. Prophets, from the eleventh or twelfth century. In the Twelve Prophets, Rahlfs includes 22 (36) 48 51 (62) (147) among the witnesses to the Lucianic text L but does not cite them individually. In Isaiah, 22 (36) 48 51 (62 93 147) constitute L. In Isaiah, Jeremiah, Ezekiel, Daniel, and the Twelve, Ziegler makes it a member of the core Lucian group L. Full scans are now available on the British Library web site at http://www.bl.uk/manuscripts/FullDisplay.aspx?index=127&ref=Royal_MS_1_B_II.
	36	Vatican, Gr. 347. Prophets, from the thirteenth century. In the Twelve Prophets, Rahlfs includes 22 (36) 48 51 (62) (147) among the witnesses to the Lucianic text L but does not cite them individually. In Isaiah, 22 (36) 48 51 (62 93 147) constitute L. In Isaiah, Ziegler makes it a member of the secondary Lucianic group 𝓁III; in Jeremiah, Ezekiel, Daniel, and the Twelve Prophets, Ziegler makes it part of the core L group.
	44	Zittau, Stadtbibl. A.1. Complete Bible (same as 664 in the New Testament manuscript list). Fifteenth century. Listed by Rahlfs as a weak member of the Lucianic group L in Judges, but never cited individually. In Leviticus and Numbers, Wevers lists it as part of the d group along with 106 107 125 610; in Ruth, Quast places it in the Catena group C.
	48	Vatican, Greek 1794. Prophets, from the twelfth century. In the Twelve Prophets, Rahlfs includes 22 (36) 48 51 (62) (147) among the witnesses to the Lucianic text L but does not cite them individually. In Isaiah, 22 (36) 48 51 (62 93 147) constitute L. In Isaiah, Jeremiah, Ezekiel, Daniel, and the Twelve Prophets, Ziegler makes it a member of the primary Lucianic group L.
	51	Florence, Laurentian x.8. Prophets, from the eleventh century. In the Twelve Prophets, Rahlfs includes 22 (36) 48 51 (62) (147) among the witnesses to the Lucianic text L but does not cite them individually. In Isaiah, 22 (36) 48 51 (62 93 147) constitute L. In Isaiah, Jeremiah, Ezekiel, Daniel, and the Twelve Prophets, Ziegler makes it a member of the primary Lucianic group L.
	54	Paris, Nat. Reg. Gr. 5. Octateuch (with Theodoret's commentary, according to the Paris Library web site) plus Aristeas. Thirteenth or fourteenth century. Listed by Rahlfs as a member of the Lucianic group L in Judges, but never cited individually. In Leviticus and Numbers, Wevers lists it as part of the n group along with 75 127 458 767; in Ruth, Quast places it in the Lucianic group L.. Black and white scans (from a microfilm) are available at https://gallica.bnf.fr/ark:/12148/btv1b107217972.
*	55	Vatican, Regin. Gr. 1. Fourteenth or fifteenth century. LXX almost complete. Cited by Rahlfs only for the Odes (along with A R T), since B and (presumably) ℵ never contained this book. Nonetheless it has a very high quality text in many books. In Leviticus and Numbers, Wevers lists it among the mixed manuscripts; in Ruth, Quast makes it, along with M and N+V, one of the primary members of the group R; in 1 Esdras, its extremely high-quality texts causes Hanhart to link it with B; in Esther, he cites it individually (implying a high-quality text); in 1 Maccabees, Kappler lists it as mixed; in Psalms, Rahlfs calls it a mixed text and groups it with S 2019.
*	58	Vatican, Regin. Gr. 10. Thirteenth century. Pentateuch and historical books. Of all the minuscules, this is the one Rahlfs cites most. In Judges, curiously, Rahlfs calls it both a hexaplairic and a Lucianic manuscript -- but he does not cite it for that book, so it hardly matters. His citations are from Esther and Judith; in Esther, he actually cites it more often than ℵ. In Esther Rahlfs lists it exclusively as Hexaplairic, a description echoed, e.g., by Field and Swete. In Judith, he describes it as going with ℵ. Rahfls cites 58 and (the later text) of 93 as the O witnesses in Esther. Where 58 is defective (9:22-9:27), he uses 583 instead. In Leviticus and Numbers, Wevers lists it as part of the core Origen group O along with G 376 426; in Ruth, Quast places it in the large group R; in 1 Esdras, Hanhart pairs it with 340 but does not place it in a particular type; in Esther, Hanhart places it in O and pairs with 583; in 1 Maccabees, Kappler lists it as mixed.
	59	Glasgow, University, MS Gen 322 (fomerly BE 7b.10, and so listed, e.g., by Swete). Octateuch. Fifteenth century. Listed by Rahlfs as a member of the Lucianic group L in Judges, but never cited individually. In Leviticus and Numbers, Wevers lists it among the mixed manuscripts; in Ruth, Quast places it in the Lucianic group L.
	62	Oxford, New College 44. Prophets. Swete dated it to the thirteenth century, but Ziegler dates it XI. In the Twelve Prophets, Rahlfs includes 22 (36) 48 51 (62) (147) among the witnesses to the Lucianic text L but does not cite them individually. In Isaiah, 22 (36) 48 51 (62 93 147) constitute L. In 1 Maccabees, Kappler lists it in the lesser Lucianic group l. In Isaiah, Ziegler makes it a member of the secondary Lucianic group 𝓁I along with 147; so also in the Twelve Prophets; in Jeremiah, Ziegler puts it in the group l; in Ezekiel, Ziegler makes it part of the secondary Origenic group o with 147 407; in Daniel, Ziegler again places it in the Origenic group O with V 147.
	64	Paris, National Library Greek 2. Tenth or eleventh century (Wevers prefers the tenth). A casual glance at Swete gives the impression that it is a complete Bible, but in fact it contains the Octateuch and historical books (Samuel, Kings, Chronicles, Ezra 1+2, Esther, Tobias, Judith, 1-3 Maccabees) only. Wevers in Numbers places it in oI, a group of weak Origenic witnesses that also include 15 381 618; in Ruth, Quast again places it in oI with 15 381. Listed by Rahlfs as a member of the Lucianic group L in 1 Maccabees along with (19) (93) 236, but never cited individually. In 2 and 3 Maccabees, 64 and 236 are the only listed witnesses to L, and there is no Lucianic text of 4 Maccabees. In Leviticus and Numbers, Wevers lists it as part of the secondary Origenic group oI group along with 15 381 618 708; in Ruth, Quast again places it in oI, along with 15 381; in 1 Esdras and Esther, Hanhart puts it in the large b group; in 1 Maccabees, Kappler lists it in the Lucianic group L. Black and white scans (from a microfilm) are available at https://gallica.bnf.fr/ark:/12148/btv1b107218011.r=grec%202?rk=21459;2.
	75	Oxford, University College LII. Octateuch. Dated 1126. Listed by Rahlfs as a member of the Lucianic group L in Judges, but never cited individually. In Leviticus and Numbers, Wevers lists it as part of the n group along with 54 127 458 767; in Ruth, Quast places it in the Lucianic group L.
	82	Paris, National Library Coisl. Gr. 3. Twelfth century (with one replacement leaf at the beginning from XIV). Pentateuch and former prophets, with some disordering at the beginning of 3 Reigns. Also includes excerpts from Eusebius's Ecclesiastical History. One of the five manuscripts Rahlfs cites as part of the Lucianic group L in 1-4 Kingdoms, the five being 19 82 98 108 127 (although he lists 82 and 127 as weak group members in 3-4 Kingdoms). He does not however cite 82 individually. In Judges he lists it as a weak member of the Lucianic group L. In Leviticus and Numbers, Wevers lists it as part of the secondary Origenic group oII along with 29 72 707; in Ruth, Quast places it in the Lucianic group L.. Black and white scans (from a microfilm) are available at https://gallica.bnf.fr/ark:/12148/btv1b11004824x.r=coislin%203?rk=21459;2.
*	86	Rome, Barberini v. 45. Later Propets (Isaiah, Jeremiah, Ezekiel, the Twelve; not Daniel). Ninth century. Properly this should not be included among the manuscripts cited by Rahlfs; he does not cite it for the text of any book of LXX. Rather, he cites it at Jeremiah 38:15 with regard to the reading of Aquila. In Isaiah, Ziegler puts it in an "Alexandrian Group" with A Q 26 106 710; in Jeremiah, it goes with Q V etc.; in Ezekiel, Ziegler pairs it with 710; in the Twelve Prophets, Ziegler puts it in the secondary Lucianic group 𝓁II.
	87	Chigi MS. 2, Rome. Prophets, from perhaps the ninth century. Although Swete calls 87 and 91 Hesychian, Rahlfs in the Twelve Prophets lists 87 91 (97 490) among the witnesses to the major catena C (but does not cite them individually). In Isaiah, Jeremiah, Ezekiel, Daniel, and the Twelve Prophets, Ziegler makes it a member of the main catena group C.
**	88	Chigi MS. 3, Rome (Vatican Library R.VII.45). Dates from roughly the eleventh century. Contains Isaiah, Jeremiah, Ezekiel, and Daniel, as well as Hippolytus's commentary on Daniel. Strangely, the list of important codices in Rahlfs does not include this manuscript -- but he cites it very often, because 88 was, until the twentieth century and the discovery of 967, the only known copy of the original Old Greek text of Daniel; it is still the only copy to contain all of LXX Daniel. Rahlfs reconstructs LXX Daniel on the basis of 88 and the Syriac version; the Göttingen LXX of course adds the testimony of 967 (a papyrus of probably the early third century). Although 88 is usually mentioned solely because it contains "Δανιηλ κατα τουc ô," it also has some value in the other prophets, since it contains at least some of the hexaplairic signs. In Isaiah, Ziegler makes it a member of the secondary Origenic group oI with the Harkleian Syriac; in Jeremiah, Ziegler makes 88 hark the core Origenic group O; in Ezekiel, Ziegler puts 88 hark with Q in O. In Daniel, Ziegler files the non-LXX portion with 449 as 𝓁II.
	91	Vatican, Ottob. Gr. 452. Prophets, from the eleventh century. Although Swete calls 87 and 91 Hesychian, Rahlfs in the Twelve Prophets lists 87 91 (97 490) among the witnesses to the major catena C (but does not cite them individually). In Isaiah, Jeremiah, Ezekiel, and the Twelve Prophets, Ziegler makes it a member of the main catena group C.
*	93	London, British Library Royal. I.D.ii. Thirteenth or fourteenth century (the British Library site would place the main text in XIII with a supplement, containing roughly half the surviving portion of Isaiah, from the fifteenth century; Kappler says XIII for 1 Maccabees, and Hanhart dates it XIII for Esther and Ruth). Ruth, 1-4 Kings, Chronicles, 1-2 Esdras, Esther (two different texts), 1-3 Maccabees, and part of Isaiah. There are a many notes in the margin. A mostly Lucianic text. Other than 58, this is the minuscule Rahlfs cites most often. It is the e₂ of Brooke and MacLean's famous group boc₂e₂. It is one of the five manuscripts Rahlfs cites as part of the Lucianic group L in 1-4 Kingdoms, the five being 19 82 98 108 127 -- in other words, it has the alternate text of Esther (see under 19). In 1-2 Chronicles and 2 Esdras, Rahfls lists as Lucianic 19 93 108 and sometimes 121. In Esther, Rahlfs includes the "textus prior" in L, the later text in O; which essentially agrees with Hanhart (see above under 19 for the two texts). Hanhart also makes it Lucianic in Ruth; so does Quast. In 1 Maccabees, L is considered to include (19) 64 (93) 236; Kappler put it in the lesser Lucianic group l. In Isaiah, 22 (36) 48 51 (62 93 147) constitute L. In Isaiah, Ziegler makes it a member of the secondary Origenic group 𝓁III with 36 96. Lagarde labelled it m. High-resolution scans are now available on the British Library site at http://www.bl.uk/manuscripts/FullDisplay.aspx?ref=Royal_MS_1_D_II&index=0.
	97	Vatican, Gr. 1153. Isaiah and the Twelve Prophets, from the tenth century. Rahlfs in the Twelve Prophets lists 87 91 (97 490) among the witnesses to the major catena C (but does not cite them individually).
	106	Ferrera, Bibl. Comm. Comm. Cl. II, 187, III(Gr. 187). Dated 1334 by its colophon. Complete Bible; it is 582 of the New Testament. Swete calls it Hesychian, but Rahfls lists it as a weak member of the Lucianic group L in Judges. In Leviticus and Numbers, Wevers lists it as part of the d group along with 44 107 125 610; in Ruth, Quast places it in t group with 74 76 134 344 799; in 1 Esdras and Esther, Hanhart puts it in the very large a group; in 1 Maccabees, Kappler lists it as mixed; in Isaiah, Ziegler puts it in an "Alexandrian Group" with A Q 26 86 710, with 106 being the closest to A; in Jeremiah, Ziegler pairs it with 410 but links the two with both B and A; in Ezekiel, he simply links it with 410; in Daniel, he puts it with A 584; in the Twelve Prophets, Ziegler again make it Alexandrian and close to A and 26.
*	108	Vatican, Gr. 330. Thirteenth or fourteenth century. Pentateuch and historical books. Lucianic text. This is Brooke and MacLean's b -- the famous b of the Lucianic group boc₂e₂. In addition, it is believed that this manuscript was the single most important source used in the compiling of the Greek column of the Complutensian Polyglot. MS. 19 is a sister or nearly of 108, and was also cited by Brooke and MacLean as b (where they were different, 108 was cited as b and 19 as b'). But Rahlfs cites 19 only once, and 108 only twice as individuals -- both in 2 Esdras (at 17:3, which reading it shares with 19, and at 21:13). Otherwise, he cites in in 1-4 Kingdoms only as part of the group L, along with 19 82 98 127. In 1-2 Chronicles, and 2 Esdras Rahfls lists as Lucianic 19 93 108 and sometimes 121. In 1 Esdras, Rahlfs lists only 19 and 108 as members of L. In Esther, Rahlfs lists 19 and the "textus prior" of 93 and 108 as constituting L. In Judith, 19 and 108 are the only witnesses to L. Lagarde cited it as d. In Leviticus and Numbers, Wevers lists it as part of the b group along with 19 118 314 537; in Ruth, Quast places it in the Lucianic group L; in 1 Esdras, Hanhart lists only 19 and 108 as L witnesses. It is another of the manuscripts that have the alternate text of Esther, for which see under 19; indeed, Hanhart pairs it with 19 in Esther.
	121	Venice, St. Mark's Greek 3. Tenth century. Octateuch and historical books. In 4 Kingdoms, Rahlfs includes it among the witnesses to the Origenic text O, but does not cite it explicitly. In 1-2 Chronicles and 2 Esdras, Rahlfs says it goes mostly with the Lucianic text L, but still does not cite it, apparently because the witnesses 19 93 108 are more purely Lucianic. In Leviticus and Numbers, Wevers lists it as part of the y group along with 318 392; in Ruth, Quast places it in the large group R; in 1 Esdras, Hanhart puts it in the very large a group.
*	127	Moscow, Syn. Library Greek 11. Tenth century. Octateuch, Kingdoms, 1 Chronicles; 2 Chronicles breaks off at chapter 36. It is the c₂ of Brooke and MacLean's famous group boc₂e₂. Rahlfs cites it explicitly only in 1-4 Kingdoms, and very rarely even in those books. It is, however, one of the five manuscripts he cites as part of the Lucianic group L, the five being 19 82 98 108 127 (although he lists 82 and 127 as weak group members in 3-4 Kingdoms). In 1 Chronicles, Rahlfs says its character changes, going with B rather than L. I might be inclined to list it as going with ℵ instead, or the agreement of Bℵ, but certainly it is more like the Egyptian text than the Lucianic. In 3-4 Kingdoms, it has a unique chronological scheme, unlike that of either MT or the original LXX, which appears to be editorial. In Leviticus and Numbers, Wevers lists it as part of the n group along with 54 75 458 767; in Ruth, Quast places it in the Lucianic group L.
	134	Florence, Laurentian Library Plut. v.1. Octateuch and historical books; Swete dates the former to the eleventh, the latter to the tenth century. Swete lists the Octateuch as Hesychian, but Rahfls lists it as a weak member of the Lucianic group L in Judges (but does not cite it). In Leviticus and Numbers, Wevers puts it in the t group along with 74 76 84 370; in Ruth, Quast places it in the t group of Lucian and pairs it with 344; in 1 Esdras, Hanhart puts it in the very large a group; in 1 Maccabees, Kappler lists it in the large group q. Scans are available on the Laurentian web site at http://mss.bmlonline.it/s.aspx?Id=AWODjo-RI1A4r7GxL9Z3&c=Biblia#/book.
	147	Oxford, Bodleian, Laud 30. Prophets (Daniel is defective). In the Twelve Prophets, Rahlfs includes 22 (36) 48 51 (62) (147) among the witnesses to the Lucianic text L but does not cite them individually. In Isaiah, 22 (36) 48 51 (62 93 147) constitute L. In Isaiah and the Twelve Prophets, Ziegler makes it a member of the secondary Lucianic group 𝓁I with 62; in Jeremiah, Ziegler puts it in the group headed by Q V; in Ezekiel, Ziegler puts it in the secondary Origenic group o with 62 407; in Daniel, it is part of the core Origen group O with V 62.
	236	Vatican Library Gr. 336. Joshua, Judges, Ruth, and the rest of the historical books. Twelfth century. Listed by Rahlfs as one of the two primary witnesses to the Lucianic text L in 1 Maccabees, with 64 being the other and 19 93 being secondary witnesses, but Rahlfs never cites it explicitly. In 2 and 3 Maccabees, 64 and 236 are the only listed witnesses to L, and there is no Lucianic text of 4 Maccabees. In Ruth, Quast places it in the secondary Catena group cII; in 1 Esdras, Hanhart puts it in the large b group; in Esther, Hanhart has it in the even larger a group; in 1 Maccabees, Kappler lists it in the Lucianic group L.
*	247	Vatican Library Greek Urb. 1. Contains only 1-3 Kingdoms plus part of 4 Kingdoms. Although Rahlfs does not cite this regularly in any book, citations in 1-3 Kingdoms are not uncommon. Rahlfs links it with A, 376, and the Syriac among the hexaplairic witnesses, in 1-2 Kingdoms, and with A and the Syriac in 3 Kingdoms, citing it as O²⁴⁷.
	344	Athos, Παντοκρατορος 24. Rahfls lists this as a weak member of the Lucianic group L in Judges but does not cite it explicitly. In Leviticus and Numbers, Wevers lists it as part of the s group along with 30 85 130 321 343 346 730; in Ruth, Quast places it in the group t with 74 76 106 134 799.
*	376	Escorial, Real. Library Y-II-5. Listed by Rahlfs as having a hexaplairic text in Judges, but he cites it only in 1-2 Samuel (primarily in the former). Rahlfs links it with A, 247, and the Syriac among the hexaplairic witnesses, citing it as O³⁷⁶. In Leviticus and Numbers, Wevers lists it as part of the core Origenic group O group along with G 58 426; in Ruth, Quast again places it in O.
*	393	Grottaferrata, A XV. Eighth century (?). Codex Cryptoferratensis, usua. A few fragments of the Prophets, of the eighth or ninth century. Palimpsest, with the upper writing consisting of muscal texts of the thirteenth century. Sometimes listed under the symbol Γ. This is one of the four minuscules listed as most important on the reference card in Rahlfs, but he does not in fact cite it particularly often. In Isaiah, Jeremiah, Ezekiel, and Daniel, Ziegler regards it as mixed; in the Twelve Prophets, Ziegler puts it among the Alexandrian witnesses along with A Q (W) etc.
	426	British Library MS. Burney 34. A catena manuscript with some other material (e.g. Aristeas). Wevers lists it as fifteenth century; the British Library site says mid-XVI. Listed by Rahfls in Judges as one of the Origenic group O, but never cited individually. In Leviticus and Numbers, Wevers lists it as a mixed manuscript; in Ruth, Quast places it in the Origenic group O.. Scans are available on the British Library web site at http://www.bl.uk/manuscripts/FullDisplay.aspx?ref=Burney_MS_34&index=26.
	490	Rahlfs in the Twelve Prophets lists 87 91 (97 490) among the witnesses to the major catena C (but does not cite them individually). In Isaiah, Jeremiah, Ezekiel, Daniel, and the Twelve Prophets, Ziegler makes it a member of the main catena group C.
*	583	Apparently derived from 58; Rahlfs cites it (as part of the Hexaplairic group O) only for those verses of Esther where 58 is defective. In Esther, Hanhart puts it in the O group with 58 and 93.
	700	In 4 Kingdoms, Rahlfs includes it among the witnesses to the Lucianic text L, but does not cite it explicitly.
	911	Berlin Genesis papyrus, Staatsbibliothek Gr. fol. 66 I, II. A strange codex, probably originally of 32 leaves, thought to have been copied from a scroll of Genesis 1:1-35:8. The first nine leaves have two columns per page, the remainder only one; the hand is irregular and non-literary. Written on papyrus in the third or early fourth century, it is now badly mutilated, but its age alone makes it important; Rahlfs cites it extensively for Genesis.
*	918	Trinity College, Dublin, K.3.4. Confusingly, this is sometimes referred to as an uncial, O, but is treated as a minuscule, 918, by Rahlfs. The designation as an uncial is better, since it is of the sixth century. Both it and Z of the New Testament were used as under-writings for a manuscript of patristic writings. The surviving portins of O were originally four leaves, now folded as eight, containing Isaiah 30:2-31:7, 36:17-38:1. It has been suggested that the text is Hesychian. Rahlfs cites it only four times (Isaiah 30:14, 15, 36:9, 23), but all four variants are significant because in each case O/918 supports either A or B against all other witnesses known to Rahlfs. In Isaiah, Ziegler makes it a mixed manuscript.
	967	Chester Beatty Papyri IX and X; also Princeton University Library, Scheide Papyrus 1. An early third century papyrus of Ezekiel, Daniel, and Esther. Two different scribes wrote the book, with Ezekiel in one hand and Daniel and Esther in the other. The original probably had 118 leaves; a few have been lost and others mutilated. (With some of the damage thought to have been done by the finders, who tore it up to get a higher price.) Interestingly for a manuscript of this period, it uses dots to mark verse divisions. (Emaanuel Tov reports that some sort of indication of verses is found in most early LXX fragments, which are Jewish or Jewish-influenced, implying that Christians actually dropped verse indications!) In Ezekiel and Esther, it is said to agree primarily with B. Its great value is in Daniel, however, since it is a copy of the original LXX translation rather than of Theodotian. Thus it and 88 are the only extant copies of that version. Sadly, it was not discovered until too late to be used by Rahlfs, and parts were not available even to Ziegler when he produced the Göttingen Daniel. The known portions consisted of parts of Daniel 5-12 plus the Additions at the end (Susanna and Bel and the Dragon). The Chester Beatty portions can now be found at the Chester Beatty Library web site, https://viewer.cbl.ie/viewer/browse/-/1/SORT_TITLE/DC:biblicalpapyricollection/. (Be warned, however, that the site is not organized by manuscript but by content, one leaf at a time; if you want to see, say, a passage in Daniel in 967, you have to go through the list of pages of 𝔓⁴⁵, 𝔓⁴⁶, 𝔓⁴⁷ and all the other NT and LXX and apocryphal manuscripts until you get to the section on Daniel.)
*	1098	Milan, Ambrosian Library O 39 Sup. Palimpsest, written probably in the ninth or tenth century and overwritten around the thirteenth. Noteworthy because the text is not just the LXX but the (almost) complete Hexapla; the Hebrew column was omitted, but five others (Hebrew translated into Greek letters, plus Aquila, Symmachus, LXX, and a fifth column, which is generally said to be Quinta rather than Theodotian) are preserved. Only eleven psalms are preserved, but it is still probably the best evidence for the Hexapla we have available. This presumably explains why it is one of only four minuscules listed on the reference card in Rahlfs; it certainly isn't because he cites it frequently. The only citation is in Psalm 17:43, where Rahlfs adopts its reading χουν against the χνουν of all other witnesses. In the major edition of Psalms, Rahlfs files it with 2005 among Origenic witnesses.
*	2013	Leipzig Papyrus 39. A papyrus scroll -- in fact, an Opisthograph -- rather than a codex. Since the writing on the other side is dated 338 C.E., the document is presumably of the mid to late fourth century. Contains portions of Psalms 30 to 55, with the first several psalms badly mutilated. Given the length of the surviving portion of the scroll, and the usual maximum length of scrolls, it seems likely that it contained only about half the Psalms. This is another manuscript listed on Rahlf's quick reference card as an important minuscule, but he cites it only a handful of times. In the major edition of Psalms, he calls it an Upper Egyptian witness along with U and the Sahidic.
*	2018	London papyrus, of the seventh or eighth century. Greek/Sahidic psalter. Contains Psalms 10:2-11:5, 48:20-49:7, 118:24-38. Obviously such a small fragment is cited only very rarely. Rahlfs considered its text to be somewhat "upper Egyptian," but it is hard to know what that means in context; U 2013 and the Sahidic are the primary Upper Egyptian witnesses.

Old Testament Textual Criticism

Introduction

The Materials of Old Testament Criticism

The Methods of Old Testament Criticism

Appendix: Textual Criticism of LXX

Appendix: The New Testament in the Old

Odes 13 LXX

Luke 2:29-32

Odes 9 LXX

Luke 1:46-55

Odes 9 LXX

Luke 1:68-79

Appendix: Important Manuscripts (Hebrew and Other)

Appendix: Greek manuscripts cited by BHS, BHK, and Rahlfs

Frequency of Citation in Rahlfs

Descriptions of the Cited Manuscripts