Alan R. Millard
[Alan R. Millard is professor of Hebrew, Akkadian and Near Eastern Archaeology at the University of Liverpool. He has worked on numerous excavation projects in the Near East and currently is epigraphist with the British Archaeological Expedition at Tell Nebi Mend (Qadesh on the Orontes) in Syria.]
Every activity concerned with Old Testament study, owes its existence entirely to generations of Jewish scribes, who copied and recopied the books of the Old Testament for more than 1, 500 years. Until recently only the products of the last third of that time were available. The most extensive example is the Aleppo Codex. This manuscript represents at its fullest the meticulous concern of the scribes for the accurate transmission of the sacred text. Their activity in copying the text followed long-established patterns, eventually codified in tractates appended to the Babylonian Talmud (Soferim, Masseketh Torah).
The question of how old these practices, or the attitudes they embody, might be has received only limited attention, partly because of the lack of early material. Respect for small details of the text characterized the teaching of Rabbi Akiba (died ca. A.D. 133) and Aquila’s even earlier Greek rendering of the Old Testament. Care for the precise wording of the biblical text is attested,
BSP 11:2-3-4 (Spring-Summer-Autumn 1982) p. 34
therefore, at the start of the Christian era. The application of this care to the copying of texts is thought to have been Jewish imitation of Greek custom, in the course of this paper a different origin will be indicated. With the discovery of the Dead Sea Scrolls, the last three decades have given scholars the privilege of studying Hebrew manuscripts of the Old Testament much older than any previously known. Investigations of scribal techniques in the Scrolls have been published, but an overall and balanced evaluation has to wait until all the texts are made available. In the famous Isaiah Scroll from Cave I the obvious corrections display the faults of the original scribe and the attention of another. Other fragmentary manuscripts, varying from the traditional “Massoretic” text, have given rise to various hypotheses about earlier stages of their history and the fluid situation at Qumran. Without older copies, any opinions remain hypothetical. Although earlier copies of any part of the Bible are denied us, neighboring cultures can show how ancient scribes worked, and such knowledge can aid evaluation of the Hebrew text and its history.
Babylonian Scribal Practices
The most prolific source of ancient documents is Mesopotamia. There the practice of writing can be observed from before 3000 B.C. Almost from the start customs arose which endured until the demise of the cuneiform script at the beginning of our era. Scribes categorized and listed words in regular order, probably to be learned by rote. From the middle of the third millennium B.C. a significant number of literary compositions survive, written in Sumerian, but in some cases copied by scribes with Semitic names. Their names are known because they are given in colophons, concluding the copies. Here, at an early date, is a sign of responsibility; a signed copy could be traced to its writer for credit or reproof, or to check a source. A few works recently assigned to this era, the Early Dynastic III period, prove to be the ancestors of several copies previously known from Old Babylonian times, some seven or eight centuries later. Now the textual history of one or two compositions can be investigated. In editing a hymn in praise of the city of Kesh, R.D. Biggs commented that “there is a surprisingly small amount of deviation” between copies of the two periods, and “The Old Babylonian version is a faithful reflection of a text that had already been fixed in the
BSP 11:2-3-4 (Spring-Summer-Autumn 1982) p. 35
Sumerian literary tradition for centuries.” The archives of Ebla are now revealing that the basic scribal conventions and textbooks were common to that Syrian city as well as to the cities of Sumer about 2300 B.C.
It is the old Babylonian period, the age of Hammurabi, that has bequeathed to us the largest collections of early literature. The principal finds have been made at Nippur, Ur, and Kish, but it is clear that the material was known over a wider area. So far as can be determined, these tablets are the exercises of students in schools. That is why many duplicate texts are found, enabling the reconstruction of whole compositions from numerous incomplete copies. It is worth emphasizing the number of manuscripts available for individual compositions, in some instances 20 or 30, occasionally 50 or 60, all of approximately the same date. When they are set side by side in a critical edition the scribal errors are made plain, and they fall into the recognized classes. Large numbers of differences appear which are not errors. The majority are variants in orthography1 ; the minority, a relatively small number, are true variants which occasionally allow manuscripts to be grouped by type of text. Colophons occur in some of these copies, though not frequently. Most common is a note of the total number of lines. In a long text, every tenth line might be marked, and subtotals entered at the foot of each column. Evidently a check was made with an exemplar after the copy had been completed. Sometimes a correction was made in the text, and if a line was found to have been omitted, it was written on the edge of the tablet with a horizontal line marking its correct position in the text. (This appears to have been done on the Snake Charm text from Ras Shamra.) If a composition occupied more than one tablet, the last line of the tablet would stand as the first of the next. The Old Babylonian manuscripts of the Atrahasis Epic display these points, each ending with a comprehensive colophon: 1st tablet, “When the gods like man” (the title), number of lines 416, scribe’s name, month, day, year.
Just as third-millennium works were copied in Old Babylonian times, so compositions of the early second millennium were copied in the first. Again opportunities arise for comparison of copies made many centuries apart. There are compositions which were copied for a millennium or more with minimal change. The “Laws of
BSP 11:2-3-4 (Spring-Summer-Autumn 1982) p. 36
Hammurabi” exemplify this. The latest edition lists over three dozen manuscripts, many only small fragments, ranging from Hammurabi’s days until Nebuchadnezzar’s. Variations are basically in spelling: there are examples of “modernization” in grammatical forms and a few small differences of wording. Another example of faithful transmission is the poem edited as “The Return of Ninurta to Nippur.” The editor listed 64 variants from the 207 lines of Sumerian text attested by 54 manuscripts from Old Babylonian, Middle Babylonian, Middle Assyrian, Neo-Assyrian and Neo-Babylonian periods. Of those 64 variants he stated that “only twelve can be said to involve a real alteration of the sense of the line in question, and in no case is the sense of the text as a whole affected.” On the other hand, some works show major differences between the earlier and the later copies. In none is this more obvious than the Epic of Gilgamesh. However, the differences in this case are not simply the result of scribal error; they are due in large part to deliberate editorial activity. Reasons for some of the changes can be proposed in the light of known developments in religious thought; for the majority no reason can be offered, and indeed, it is hard to find any significance in them. Perhaps it is pertinent to observe that when a manuscript of only one period survives, it is impossible to predict whether an earlier or a later copy might or might not differ, and if it were to differ, how it would do so. But this is a matter that rises beyond our primary concern, the activity of the scribes as copyists. The tradition of the colophon persisted throughout the first millennium B.C., sometimes with the name of a scribe’s colleague or senior as the inspector or collator of the copy following the scribe’s name. In the later period, also, there are added details of the exemplar or exemplars; for example “copied from a tablet from Babylon,” providing a pedigree, as it were, for the text.
Certain other points illustrate the scrupulosity of the scribes in handling texts, their traditionalism, and their care as glossators2 attempting to elucidate texts. First, scribes copying from clay tablets might find their exemplars damaged. In some cases they may have been able to restore the damaged text and hide the fact from us. Sometimes the scribe simply recorded the damage by writing “break” or “recent break” in smaller script on his copy, even when the restoration seems obvious to us. Second, scribes were careful not to split a word between the end of one line and the start of the
BSP 11:2-3-4 (Spring-Summer-Autumn 1982) p. 37
next; in fact, they normally avoided breaking phrases. Where there were insufficient words to fill a line fully, the scribe would space his signs and ensure that there was one at the end of the line. Occasionally two lines of an exemplar might fit onto one line of the copy. Third, when the two lines were complete in themselves, a “colon” in the copy would mark the division. This “colon” varies its form between one vertical or diagonal wedge, and two diagonal wedges. If a scribe was forced by exigency of space to break a word or a phrase, he could write it below the far end of the line, sometimes preceded by the “colon.” The “colon” also served to mark glosses. From an early date, scribes adopted various orthographic techniques to ease the reader’s task, spelling syllabically words written with word-signs, for example marking them off with this sign. The Amarna Letters and the Ras Shamra Akkadian texts provide many examples of Akkadian writings with words glossed in a local language, the gloss usually being marked by the “colon.” Finally, certain copies of literary texts made in the first millennium B.C. have doublets: a word is followed by a synonym or variant, separated from the main text by the “colon.” The explanation offered is that these are the readings of different exemplars. This becomes a regular feature for distinguishing the text from the comment in the learned commentaries of the Babylonian academies.
Throughout the history of cuneiform writing there was a tradition of care in copying. Babylonian scribes were aware of their weaknesses and established various conventions to overcome them. No one could claim they always succeeded, but it is important to be aware of the fact that they tried.
Early West Semitic Scribal Practices
After the end of the Late Bronze Age (ca. 1200-1100 B.C.), Babylonian influence in the Levant grew weak. The political situation was one cause of this, and another, in the sphere of writing, was the rise of the alphabet. With the simple script of 22–30 letters, writing ceased to be a scribal monopoly. Nonetheless, scribes still held a major place in the production of documents, and doubtless they were responsible for introducing and maintaining various conventions that are apparent in surviving texts. Unlike the Babylonian scribes, early Hebrew clerks and their colleagues did not
BSP 11:2-3-4 (Spring-Summer-Autumn 1982) p. 38
hesitate to break a word between one line and the next if space ran out. The likelihood of misunderstanding was minimized, however, by the habit of dividing each word from its neighbor. Continuous writing, without spaces between words, familiar from Greek manuscripts as a fruitful source of error, was avoided. This practice of word division was noted by some modern Old Testament scholars but ignored by others who sought to emend the Hebrew text by dividing the words differently. Ten years ago it was demonstrated that scribes who wrote Ugaritic, Early Phoenician, Hebrew, and Moabite were accustomed to word division by a point. Where Aramaic dominated, the word-divider was not usual, but from the Persian Empire onward spaces were left regularly between words.
To date, no preexilic Israelite literary manuscript is available. The longest early Hebrew text in its contemporary form is the Siloam Tunnel Inscription. Longer compositions from adjacent regions do exemplify the work of scribes using the alphabet. There are many early Hebrew ostraca (Andre Lemaire collected 250 or so in his valuable Inscriptions Hebraiques I: Les Ostraca [Paris, 1977], many of them illegible) and several dozen graffiti. Yet strangely, longer texts are few. In contrast, early Aramaic texts of some length have been found, but few ostraca or graffiti. Only time may tell whether this situation is the accidental result of chance discovery or has other causes.
In these longer Aramaic texts some indications of techniques that would have been equally at home in the process of writing or copying a book may be seen. One reservation is necessary: texts written on stone are likely to have been traced by a scribe in ink, then engraved by a sculptor or mason, a technique apparently visible on some Assyrian stonework. Therefore, some irregularities and errors may not be truly scribal.
The three stelae from Sefire near Aleppo, bearing the treaties Bar-Gayah king of KTK (a place of uncertain identity) made with Matiel of Arpad about 750 B.C., are the most extensive inscriptions, about 175 lines preserved to some extent. In his recent edition of the stelae, John Gibson has noted “several mistakes, certain or probable, by the stone-cutters.” In all he lists fourteen, but the number that can be counted as “certain” is very much smaller, possibly no more than three or four. The presence of an ancient correction is as interesting an error as modern scholars can detect. Face B of Stele II reads: “the treaty and favour which the gods have
BSP 11:2-3-4 (Spring-Summer-Autumn 1982) p. 39
Facsimile of an Aramaic treaty text, Sefire stele II, face B, showing inserted line, ca. 750 B.C. Copied by J. Starcky, in A. Dupont-Sommer, Les inscriptions araméennes de Sefire, 1958.
BSP 11:2-3-4 (Spring-Summer-Autumn 1982) p. 40
made in Arpad and among its people; and if Matiel will not obey, and if his sons will not obey, and if his nobles will not obey, and if his people will not obey….” The repetition of “will not obey” lends itself easily to the error of haplography3 , and, in fact, the words “if his sons will not obey” in the second phrase were omitted originally. After the third line had been incised in the stone, the missing words were squeezed in between lines 2 and 3.
A similar error was made by the person who wrote the Aramaic dialect text about Balaam on the plaster of a temple wall at Tell Deir Alia in the Jordan valley about 700 B.C. (see Bible and Spade, Autumn 1977, pp. 121-124). The first line of the text, as restored by A. Caquot and A. Lemaire on the basis of Hoftijzer’s edition, reads, “The record (spr) of Balaam, son of Beor, the man who saw the gods. Now the gods came to him by night…” Writing the text on the vertical plastered face of the wall, the scribe omitted “to him” before “the gods” and had to insert it above the line. (Similar omissions were rectified in two other places.) This restoration involves an adjustment to Hoftijzer’s edition and is attractive, yet leaves a space at the beginning of the first line. Indentation was not normal at the beginning of a text, so another word should be supplied at the start, and the most likely word is the demonstrative pronoun, “this” (znh). The narrative might then commence: “This is the record of Balaam, son of Beor, a man who saw the gods was he. Now the gods came to him by night….” This inscription from Deir Alla probably represents a column of a scroll. It has the upper and left-hand margins ruled (the right was provided by the corner of the plastered face) and headings written in red ink in Egyptian style. It is the nearest we can come to the appearance of a book in Palestine about the time of the prophet Isaiah.
The oldest actual example of West Semitic literature in book or scroll form so far recovered is the “Proverbs of Ahiqar” from among the papyri from the island of Elephantine at Aswan. Epigraphic study has dated the manuscript late in the 5th century B.C.; thus, it reflects book production at the time of Ezra, the time when traditionally, the Aramaic or square script (called “Assyrian”) was adopted by Jewish scribes. Here it is interesting to see how the introductory narrative is written in long lines, each one filled, the words separated from one another by small spaces, and not broken
BSP 11:2-3-4 (Spring-Summer-Autumn 1982) p. 41
Facsimile of the opening section of the Balaam text from Tell Deir Alla, showing inserted word in line 1, ca. 700 B.C. The writing appears to be laid out as a column of a scroll. (J. Hoftijzer, Aramaic Texts from Tell Deir Alla, Brill, pl. 29).
between one line and the next. (The scribe was not concerned to justify his left-hand margin!) In the section containing the proverbs the scribe often ended one proverb and left the rest of the line blank, starting the next one on a new line. Sometimes he marked the end of a proverb with an alep-like sign, whether or not the next proverb followed on the same line. Other proverbs are distinguished from each other by a horizontal stroke between the lines. The commencement of each proverb on a new line is not regular, however, nor is the insertion of the terminal mark or the bar.
Scribal Accuracy
These diverse examples of extrabiblical documents reveal how ancient copyists wrote their texts, and how they tried to write them
BSP 11:2-3-4 (Spring-Summer-Autumn 1982) p. 42
so they would be readily legible to anyone trained in the same conventions. In this atmosphere, too, the early copyists of the Old Testament books were bred. That they maintained similar high standards of careful and accurate copying is proved, at least in certain respects, by the following collection of examples.
Within the Old Testament are numerous foreign names, many of them alien to the western Semite. (Foreign names pose problems in all languages and scripts; the various spellings of East European or Oriental names in our newspapers illustrate that.) Where ancient writings of these names are available, detailed study shows the Hebrew writings represent the contemporary forms very closely. Thus the names of the Assyrian kings Tiglath-pileser and Sargon, as handed down through the Old Testament, turn out to be accurate reflections of the Assyrian dialect forms of these names. Tiglath-pileser is found in an almost identical spelling on the Aramaic Bar-Rakkab stele from Zinjirli, carved during his reign, or very shortly after. Sargon, occurring in Isaiah 20:1, has become familiar in Akkadian dress as Sharru-ken, but in Assyria during the king’s rule, the sh was pronounced s and the k as g as in Tiglath-pileser. These are normal sound-shifts between the Babylonian and the Assyrian speaking regions in the early first millennium B.C. They are demonstrated by the way Sargon’s name is spelled in Aramaic letters on two documents. In the Aramaic letter written on a potsherd sent to Ashur, the old Assyrian capital city, from southern Babylonia, Sargon appears as sh, r, k, n, shar-ken, while on the Aramaic seal of one of his officers, known from an impression found at Khorsabad, Sargon’s new city in Assyria, it is s, r, g, n, sar-gon. It is exactly that spelling that has been preserved in the traditional Hebrew text of the Old Testament. A comparable precision can be argued for other foreign names throughout the Old Testament, as continuing study and discoveries indicate. In a recently published papyrus fragment from Elephantine, dated 484 B.C., the name of king Xerxes (Ahasuerus) is seen for the first time written in Aramaic with prosthetic alep as in the Old Testament and in Akkadian. From the same age there also survives a seal now in the British Museum. According to its Aramaic inscription, this cylinder seal belonged to a Persian, Parshandatha son of Artadatha. Where an identical name is read in Esther 9:7, the likelihood that the Jewish scribes correctly preserved a good Persian name seems high.
Now these minutiae may not seem to be of great consequence,
BSP 11:2-3-4 (Spring-Summer-Autumn 1982) p. 43
Facsimile of a stele of Bar-Rakkab from Zinjirli, giving Tiglath-pileser’s name in Aramaic letters in line 3, ca. 730 B.C. (F. van Luschan, Ausgrabungen in Sendschirli 4, 1911, p. 379).
BSP 11:2-3-4 (Spring-Summer-Autumn 1982) p. 44
and may simply show the scribes could transmit names with precision. There is a corollary, however, which deserves emphasis: in each case mentioned, the Septuagint differs considerably from the Hebrew. Sargon, in Isaiah 20:1, became Arna; Parshandatha was distorted through Pharsannestain to become two names, Pharsan and Nestain, in Codex Vaticanus. These cases, not confined to one book, should at least warn against reliance on the Septuagint for emendation of proper names in the Old Testament, unless the evidence against the Hebrew text is very strong indeed.
Indeed, the purpose of this paper is to point to the care which was an integral part of a scribe’s skill in the ancient Near East. The practices of scriptoria in imperial Rome offer a strong contrast, as the complaints of several ancient authors reveal, but the mass production techniques applied there were probably never at home in the world of the Old Testament. Rather, from the examples presented, and from many others, a copying process can be discerned that included checking and correction, a process that had built-in devices to forestall error. Some of these, the counting of lines or words in particular, reemerge in the traditions of the Massoretes in the early Middle Ages. That device is so obvious that a connection with Babylonian practice is unlikely. It is part of an attitude which was common: the copyist’s task was to reproduce his exemplar as faithfully as possible.
Be Wary of Emendations!
In this light the way the Old Testament text is viewed by scholarship seems to need some modification. The Dead Sea Scrolls make explicit what had previously been supposed by many, that the Massoretic text preserves an earlier text-type current in the century or so prior to the Fall of Jerusalem. Between the completion of some books of the Old Testament and the Scrolls there is a relatively short period of time. (How short will depend upon opinions about the age of each book.) Only in that period can the great majority of the errors textual critics and commentators claim to find in the Hebrew text have arisen. Is it conceivable that those who copied Jeremiah’s prophecy for over four centuries made so many mistakes as to require an average four to six lines of textual apparatus to every page in the current critical edition of the text, the Biblia Hebraica Stuttgartensia? Jeremiah may be peculiar in respect
BSP 11:2-3-4 (Spring-Summer-Autumn 1982) p. 45
to its Septuagint version, but the problems involved are such that to emend the Hebrew on the basis of the Greek would seem a very risky business indeed.
The preceding paragraphs suggest ancient copyists were not likely to be so careless. If this is true, then textual emendations should become rarities. Before charging an ancient and anonymous copyist with error, every possible explanation of the form that seems objectionable should be sought.
Emendations to remove hapax legomena4 should be a last resort. As more ancient texts are recovered, more of the unique words in the Old Testament gain satisfactory explanations without emendation. Even if there is no alternative evident, any emendation offered should be properly supported and compatible with ancient scribal practice. Further, there should always be a recognition that the text may be right after all. Of course the scribes made mistakes, and some of them were perpetuated. It is the scholar’s duty to try to discover them to correct them; it would be wrong to argue that we have received a perfect Hebrew text. The present argument is that we too freely underrate the ability and the accuracy of those copyists to whom we owe the Old Testament. There are no grounds for supposing they were less attentive to their task than those whose products have been recovered in modern times.
Scribal Alterations
For the work of scribes as copyists there is much informative material from the ancient world, from which a few pieces have been used here. With their copying, reliably or not, the scribes commonly face the accusation of altering or modifying the texts they copied. They are not reckoned as editors, a more wide-ranging activity and one beyond the scope of this study, but as glossators and interpreters, adding comments and explanations, applying the text to current circumstances. Obviously, without manuscript evidence it is almost impossible to prove that words have been added or altered. Again, the practices of ancient scribes, visible in extant specimens of their work, suggest caution should accompany every claim to detect glosses or interpretations in the biblical text. Scribes writing cuneiform normally signalled the presence of a gloss, as mentioned already, although no cases have come to light in early
BSP 11:2-3-4 (Spring-Summer-Autumn 1982) p. 46
West Semitic texts. The commentaries of the Akkadian scholars working in the first millennium B.C. interpreted standard works and applied them to the existing situation. But their interpretation and application were kept distinct; they were not incorporated into the text. Now it is possible that only after a text gained an authoritative status would the scribe provide a commentary, a process modern scholarship cannot observe.
Turning to the biblical books, it is noteworthy that the Septuagint and the Aramaic Targums also fit the text to their times. The simplest cases are the replacements of obscure place-names. In Isaiah 48:12 Sinim of the Hebrew (= Aswan) is represented by Persia in the Greek; the old name Aram in Isaiah 9:11 is replaced by Syria. In the Aramaic Genesis Apocryphon from Qumran, a more imaginative retelling of Genesis than the standard Targums, Kaptok (Cappadocia) replaces the Ellasar of Genesis 14:1. Such changes are not found in the Hebrew text. Therefore, application of the Hebrew text to current affairs, having an effect on the text itself, is unlikely to have occurred later than the making of the Greek translation or the Aramaic paraphrase. How much was done before those stages cannot be discovered at present; the customs of the Babylonian scribes, if the comparison is valid, suggests there was little done, if any at all.
Ancient Scribes Also Were Human!
Everyone who writes and copies is aware of the likelihood of mistakes in their own work. Ancient scribes were equally prone to failure. The conventional “introductions” to the Old Testament and handbooks of textual criticism instruct their readers in the categories of scribal error that appear in ancient manuscripts and may be detected in the Old Testament. There is no doubt that errors were committed by copyists and have passed into the printed text. The modern reader’s readiness to detect them should not be greater than his readiness to admit that ancient scribes and copyists could also be as precise and careful as he and may have known their business better than he. The ancient scribes deserve our thanks and praise!
(Reprinted by permission from the Biblical Archaeologist, Vol. 45, No. 3, Summer 1982, pages 143–153.)
BSP 11:2-3-4 (Spring-Summer-Autumn 1982) p. 47