This chapter will discuss how imperial period and late antique authors of erudite compilations, that is, authors composing works out of excerpts of variegated content and size, tackled their projects. At the same time, there can be no discussion of the issue of data management unless we also address concepts of orality, oral transmission and formation that pervade the talmudic text as well as its scholarship. A discussion of the most pertinent arguments for an oral transmission of the Talmud will therefore open the chapter and propose ways in which the talmudic evidence for such a transmission could be read considering the work’s cultural context. To some extent, then, this chapter will open a world of fragments and fragmentary knowledge that are still challenging – but that enable contemporary scholarship to arrange them in many different ways and to different ends.

The Perennial Appeal of Orality

The blind, unfit, or unlearned teacher represents the ideal of unmediated knowledge that persisted throughout antiquity and late antiquity. The blind Homer, innocent children as arbiters of oracles, Moses with the “heavy tongue,” a simple carpenter and a fisherman from Galilee, the anchorite monks of the Egyptian desert, female martyrs expounding Christian doctrine, and the illiterate Muhammad are but a few examples. The Mishnah makes a somewhat different case, yet with similar ideological consequences: the work claims to be Oral Torah, a recording of the laws that God gave orally to Moses on Mt. Sinai and that were transmitted from Moses all the way down to rabbinic sages.Footnote ¹ Contrary to the above examples, knowledge does not appear as necessarily unmediated in the Mishnah and later rabbinic texts. Yet, along similar lines, oral transmission makes this particular kind of knowledge available only to an exclusive group of people. The basic question, then, is whether this tenet of an oral tradition is mainly aetiological or whether it implies an actual prohibition of writing down decisions, even narratives pertaining to rabbinic law.

The reasons for such a tenet have been variously discussed. Its cause could have been the necessity to distinguish rabbinic (Oral) Torah from Mosaic (Written) Torah, especially vis-à-vis the nonrabbinic environment, or for the sake of the reinforcement of the teacher–student relationship.Footnote ² The problem is that neither the Mishnah nor the later Talmuds are in any way specific about the implications of Oral Torah. The concept is clearly a rabbinic invention, however, as Second Temple sources are silent on the subject.Footnote ³ Scholars are thus left with their own judgement regarding how to evaluate the evidence.

A dominant pedagogical device in the Mishnah and, since they are based on it, also in both the Palestinian and the Babylonian Talmud, is repetition. The root of the Hebrew verb šnh (שנה) is also the basis of the word mishnah (lit., “repetition,” “repeated tradition”). Another prominent verb in the Mishnah as well as the Talmuds is tny (תני), which similarly refers to “repeat, learn a Tannaitic tradition, to recite, to report a tradition.”Footnote ⁴ Individuals are repeatedly called reciters of tradition, that is, tannay (in the Palestinian Talmud) or tanna (in the Babylonian Talmud). As Moulie Vidas recently pointed out, there is a notable distinction between the Palestinian Talmud’s tannay and the Babylonian Talmud’s tanna.Footnote ⁵ Being a tannay is part of being and certainly of becoming a sage, whereas the tanna seems to be a person with a distinct occupation.Footnote ⁶ Some Babylonian households appear to have had a tanna in residence.Footnote ⁷ Maybe the task of the Babylonian tanna was similar to the precious (and expensive) educated slaves who were capable of reciting the entire work of Homer, Hesiod, or other lyrical poets.Footnote ⁸ These slaves served as aide-mémoires to their masters or performed at banquets.Footnote ⁹ Although sometimes brought into the conversation about the formation of the Talmud, the tanna appears to have served a specific function in his own time rather than working towards a future project (the Talmud) by serving as “human tape recorder.”Footnote ¹⁰

The idea of an oral learning culture and a prohibition against committing to writing anything associated with it is strongest in the Babylonian Talmud.Footnote ¹¹ Evidence for an actual ban, however, is very sparse. Only in two instances does the Talmud refer to such a ban, and these instances are reworkings from parallel passages in the Palestinian Talmud (b. Tem. 14b; b. Git. 60b). Tendentious reworkings are generally ascribed to authors’ changing attitudes and life circumstances. In this case, the foregrounding of orality seems to relate to the pedagogical standards imposed by the Sasanian cultural hegemony.Footnote ¹² The Palestinian Talmud allows at least occasional or private documentation of law.Footnote ¹³

Yet both passages in the Babylonian Talmud that are critical of writing Oral Torah or halakhot allow for interpretations that do not constitute a general ban. The passage in b. Gittin 60b is concerned with Oral Torah and may refer not to the interpretation of law (halakhah), but, rather, simply to the public translation of the Hebrew Bible into Aramaic during the Shabbat liturgy. The reader of the assigned portion from the Torah was not allowed to recite from memory, while the translator was not allowed to translate from writing. Rather, translators prepared themselves in the Beit Midrash with the help of written translations, glossaries, and commentaries.Footnote ¹⁴ The maxim in b. Temurah 14b, again, prohibits the writing down of halakhot, in contrast to the already written Torah.Footnote ¹⁵ Halakhot were, as is implied by the word’s root “to go” (הלך), inquiries sent to experts of Torah. Terminologically, halakhot were distinguished from the mitzvah, the ruling with a biblical basis.Footnote ¹⁶ The situation seems mirrored in Roman law, where laws registered in a codex were occasionally enhanced and adapted for certain cases and individuals. Yet these private rescripts and “letters from the emperor responding to legal questions from private citizens” were collected separately from the codex.Footnote ¹⁷ The prohibition of writing down halakhot might similarly have referred to the insertion of such situational rulings into a document of collectively sanctioned rulings.Footnote ¹⁸

Undeniably, there are many implicit references to oral transmission.Footnote ¹⁹ The Talmud clearly imagines the rabbinic world to be an oral one.Footnote ²⁰ Students are sitting at the feet of their masters; they are listening and repeating. The setting is rather pastoral, no libraries are described, and references to writing material are, although present, marginal. In fact, we do not even know if there was an economic benefit to this learning, because the picture that is raised is one of men “immersed and soaked in learning” with nothing in their lives other “than Torah – day and night.”Footnote ²¹ The talmudic schooling system that emerges from these depictions caused David Goodblatt to conclude that it looked rather underdeveloped compared to other contemporary institutions.Footnote ²²

Yet this picture, although uncontested given the lack of archaeological and textual evidence for the educational culture in Sasanid Mesopotamia, does not seem to do justice to the result. The Talmud is too elaborate and too deeply in conversation with the literary standards of its time to be the result of an educational system that focuses on the memorization of laws that either were or were not, or maybe only temporarily, relevant. Moreover, the model for oral transmission developed based on this evidence cannot account for the formation of the Talmud unless we assume that the sages were working towards this collaborative outcome from the very beginning. In that case, as David Weiss Halivni suggests, professional memorizers would have traveled from the school of one rabbinic sage to the next while memorizing the focal points of the ongoing discussions.Footnote ²³ Memorizing a live discussion is, however, something different than learning an epic song (Homer) or any other fixed sequence of information. The raw memories of these memorizers would then have been smoothed out by a severe redaction, when, under circumstances that remain unclear, the sages decided to write down the oral recollections.Footnote ²⁴ The model is shaky in its reliance on human capacity: Not only would the memorizer’s recollections have been imprecise at times, but the sudden death of a memorizer would have further jeopardized the transmission of knowledge and the project as a whole.

Up to this day, spontaneous retrieval of knowledge is what makes or breaks the impression of an accomplished scholar.Footnote ²⁵ Imperial period and late antique pedagogy fostered and relied on memorization to a much greater extent than does its modern counterpart: today, digital tools, dictionaries and handbooks, help to bypass embarrassment. In late antiquity, on the other hand, if people wanted to appear erudite, they were usually entirely dependent on their memory. Mnemotechnics were therefore essential for personal and social advancement, and elaborate systems of remembering knowledge based on places, symbols, and letters were developed since antiquity.Footnote ²⁶ What the ancients were able to master from memory – judging from their written record – was certainly impressive. But there is a difference and a certain imbalance between, on the one hand, recognizing elaborate techniques for memorizing and retrieving knowledge and, on the other, the claim that Quintilian’s “equation of treasury directly with memory and only indirectly with writing depends on the fact that it is memory and not a superior filing technique that allows the classical writer to retrieve the appropriate excerpt” (emphasis added).Footnote ²⁷ Filing and notation techniques underwent many improvements from the early imperial period onward, as will be shown in the second half of the chapter. They did not and could not replace memorization for the obvious reason that filing and notation techniques could not be deployed as spontaneously. But they made elaborate written productions – such as the Talmud – possible.

Mental capacities and oral cultures have fascinated ancient and recent thinkers alike, and for much the same reason: Prominent examples of texts that teem with sayings are monastic and rabbinic ones. Both text corpora suggest that the knowledge they portray is the result of oral transmission. The anchorite monks are said to have been illiterate, whereas rabbinic literature is said to be the result of oral transmission. For both corpora, sayings have been interpreted to be a sort of an oral recording and hence the earliest layer, while more elaborate stories and homilies are thought to form the latest stratum.Footnote ²⁸ With regard to monastic literature, however, Lillian Larsen has convincingly shown that the use of sayings does not attest to oral culture, but, rather, to students trained according to Greek and Latin writing practices, who made extensive use of chreiai.Footnote ²⁹ Scholars of Talmud may never be as fortunate as Larsen, who was able to prove her claim with exercises, ostraca, wooden tablets, and papyri found in monasteries. Yet they can still follow her proposition and see what happens if they read the Talmud “in light of ‘the literary genre to which [it] belong[s].’”Footnote ³⁰ In this case, the analysis of the talmudic genre in the previous chapter suggests a comparison with symposiac compilations and erudite commentaries. Interestingly, oral transmission has not been suggested for any of the books classified under these labels. This contrasts with texts of religious standing, such as the Mishnah and the Talmuds, monastic literature, the New Testament, and the Qur’an.Footnote ³¹

The next section will consider the arguments that have been raised so far for an oral transmission and formation of the Talmud from a comparative perspective. Yaakov Elman, who argued for an oral transmission history of the Talmud, corroborated his argument by comparing the work to the Zoroastrian compendium Dēnkard. I will contest his conclusions with observations derived from a comparison of the Talmud with erudite compositions from the Roman Empire. This is again due to the lack of comparable sources in Sasanid Mesopotamia, although we should assume that they existed. The Babylonian Talmud was most likely not the region’s sole monumental compilation.

Arguing with Yaakov Elman

In a lengthy article titled “Orality and the Redaction of the Babylonian Talmud” (1999), Yaakov Elman advanced several arguments that seem to speak for an oral tradition and even formation of the Talmud. Elman’s arguments summarize the main ideas about the oral nature of the Babylonian Talmud and thereby provide a good platform to reflect on the implications of these objections for a written model.

Elman’s first argument concerns the position of the Talmud as a written document between the oral culture of the Amoraim, the assumed originators of the sayings, and the Geonim, post-talmudic rabbinic scholars living under the Abbasid Caliphate. Since both of these generations of sages emphasize orality, Elman does not see how a work the size of the Talmud could have originated in written form in between the two eras.Footnote ³² Elman’s second argument is based on the fact that rabbinic literature is replete with variant readings, which, according to him and many other scholars, are a sign of an oral transmission that led to the transformation of an original version through the loss or addition of pieces of information.Footnote ³³ In his third argument for an oral genesis of the Talmud, Elman points to the absence of “a terminology for copying, arranging, editing, and redaction” and argues further that the size of the Talmud does not comport with late antique writing technology.Footnote ³⁴ Finally, Elman noted that the talmudic lines of argument (sugyot) are often formulaic and stereotypical. Ring structures, chiastic structures, and the segmentation according to numbers are encountered: features that facilitate memorization.Footnote ³⁵ Elman’s arguments are very suggestive – not only in support of oral transmission and formation but also as a basis to discuss alternative interpretations.

The first argument relies on the assumed oral culture of the Amoraim and Geonim. The Amoraim bear that name because they are the originators of sayings in Aramaic. The name derives from the standard use of amar (אמר), meaning “he said.” Since “XY says” is the earmark of sayings, however, the formulation may have been generated by style and convention rather than actual speaking by word of mouth. Indeed, the sayings are concise and very much to the point, so much so that heavy reworking or editing would have to be assumed as an intermediary step between the actual uttering of the content and the version that ended up as saying or maxim in the Talmud. The intellectual work necessary to mentally turn an utterance into a saying would have been enormous. It is much easier to work with templates and to arrange thought in written form. Martin Jaffee and Steven Fraade have therefore suggested that orality in this case should be imagined as a process of alternating between writing and oral discussion.Footnote ³⁶ In particular, Fraade’s model, characterized by Elizabeth Shanks Alexander as an “orality that lies both behind and in front of the extant rabbinic texts,” seems appealing here.Footnote ³⁷ The orality before the saying would be a discussion, the solution to a puzzle assigned by a teacher, or even a game. Then again, purely written settings, such as personal musings over other texts, reformulated quotes, or summaries, are also feasible.

The post-talmudic generation of sages, the Geonim, promoted oral tradition over the written. This tendency is best seen in the context of the theological discussions of their time and place: oral versus written transmission preoccupied Islamic and rabbinic scholars alike.Footnote ³⁸ Moreover, when the Geonim speak about their own oral culture, they usually refer to the memorization of the Mishnah and corresponding drills, not an oral composition.Footnote ³⁹ As Uziel Fuchs has recently shown, they were most likely in possession of the Talmud in written as well as oral form.Footnote ⁴⁰

The second argument is more focused on the talmudic text and is based on the multiple variants within the Talmud, and between the Babylonian Talmud and rabbinic literature from Palestine. These variants gave rise to the above-mentioned models by Jaffee and Fraade whereby written texts were transformed through oral transmission before they were written down again. Acknowledging the limits of human memory but also the marks of Roman writing habits within the text, these models consider writing to be an intermediary stage that is then transformed again through oral transmission.Footnote ⁴¹

One would expect variants resulting from oral transmission to be arbitrary. Yet, as it turns out, in most cases (if not all of them), the variants are not haphazard deviations but, rather, versions exposing a new take on the subject. With regard to talmudic stories with obvious parallels in the Palestinian Talmud, for example, Jeffrey Rubenstein was able to establish a list of recurring features and devices used to give these stories a new twist.Footnote ⁴² Shamma Friedman recognizes a “typical intervention” in stories (as well as in the legal parts) by the commentators or redactors.Footnote ⁴³ As will be discussed in more detail in Chapter 4, the creation of variants was a lesson in its own right in the Roman curriculum and was artfully professionalized. To bring a story to a different conclusion, for example, was a way of learning how to turn a given argument in one’s own favor, just as the original purpose of rhetorical training was juridical argumentation.Footnote ⁴⁴ The techniques were fairly standard and resonate with the way in which talmudic stories were recast. Rather than the product of a commentator or redactor, these story variants look like the result of rhetorical exercises, or the implementation of this very learning, both trained and executed in writing. This does not, of course, rule out the possibility that an author may, at times, have relied on additional oral information regarding the case described in a story.

Recent scholarship on ancient literacy has repeatedly referred to empirical research among illiterate and semiliterate people to strengthen the argument for a literate mindset. Paul Evans, for example, drew attention to the work of Aleksandr Romanovich Luria, who “found that non-literate persons strongly resisted requests for word definitions.”Footnote ⁴⁵ Not only did they find the task of defining a word nonsensical but they were also unable to describe a word without using it. Literate people, on the other hand, solved the same problem with considerable ease. The same was true for syllogistic exercises.

The question remains, however, to what extent the execution of these tasks relates to literacy qua alphabetization or, rather, to the syllogistic sense of how to handle the range of exercises that accompany alphabetization in Westernized societies. For example, Luria asked non- and semiliterates the following syllogistic question: “Bears in the north, where there is snow, are white. The city XY is in the north, where there is always snow. What color are the bears there?”Footnote ⁴⁶ The answer was unanimously “brown.” The questioned people relied on their own knowledge about bears instead of focusing on the syllogistic and tricky nature of the question. The results are interesting but perhaps in a slightly different way than Luria and, for that matter, Evans, used them. The definition of words and the solving of simple syllogisms may be part of the primary curriculum in Westernized societies but they are not related, per se, to the basic ability of reading or writing. The distinction, then, should not be between literates and illiterates but between those who received literacy training beyond mere recognition and use of letters and those who did not. In fact, Luria’s fieldwork shows to what extent continuous exercises based on the same syllogism affect the mind. Indeed, when he performed the syllogistic exercise with people whose reading and writing abilities had dwindled during years of neglect, they were still able to follow this distinct pattern of thought and to understand the task.Footnote ⁴⁷ In addition to proving what he anticipated, namely, that knowledge of writing had a decisive effect on reasoning processes, Luria’s study also highlightes the lasting effect of repetitive exercises.Footnote ⁴⁸ Late antique teachers envisioned exactly this effect on their students’ minds when they anticipated that “Dexterity of mind and an almost mathematical ability in dealing with the elements of learning” would result from their letting the students toil through endless repetitive exercises.Footnote ⁴⁹

David Olson’s empirical research has shown that literates are much more sensitive toward language. Illiterate people, for example, are not capable to the same extent as literates of associating letters with words (e.g., b with baby, ball, or rabbit) or of isolating a letter from the rest of a word (e.g., f-ish). Olson’s most telling example is an exercise he performed with his preliterate grandchild: “I showed her a card on which I had written ‘Three little pigs.’ I read it to her and had her say back to me what it said. I then covered up the last word and asked her to tell me what it now said, to which she replied, ‘Two little pigs.’ She assumed that the written marks represented objects, pigs, not words, a kind of picture writing.”Footnote ⁵⁰ Olson’s examples substantiate the claim of the literacy hypothesis that “a writing system and a tradition of writing is not a neutral practice.”Footnote ⁵¹ Rather, language and mind are connected in ways that are still to be further explored.

Oral transmission and formation are, of course, not necessarily tied to illiteracy. Indeed, regarding rabbinic sages, many scholars appear to assume a voluntary refusal to read and write halakhot or Oral Torah more broadly. The mindset of rabbinic sages would in that case be literate and explain the saturation of rabbinic literature with comments and jokes that rely on wordplay and paronomasia, which are inextricably related to the ability to understand the anatomical makeup of a word or phrase. Indeed, puns, explanations based on homonyms, mute letters, and the like are exactly the features that dominate talmudic stories and arguments.Footnote ⁵² If these features were later redactional additions, it needs to be asked what the original message of these stories would have been. Equally obtrusive as these wordplays are certain types of syllogistic reasoning that the progymnasmata introduced at an early stage of education.Footnote ⁵³ Thinking along the oral tradition model, these features may have diffused into oral culture through a literary education gained elsewhere. Still, it needs to be asked how these linguistic adornments came into being if not through writing and how they lasted transmission by tradents, who did not necessarily have the same education and could not imagine a silent letter in a word.Footnote ⁵⁴ Would it not have been more efficient to focus on rhythm and meter if the goal were to remember and retain the old traditions, and even continuously add new elements, instead of preserving linguistic puns that primarily excite the one who sees them in written form?

It is indeed quite difficult to imagine that a completely alphabetized person would refuse to write down their insights for posterity and, instead, prefer to rely on the fragility of another mortal’s mind or on occasional notes. Now that we have the Talmud in front of us as a complete work, it is easy to muse about a possible oral transmission and formation, possibly with a redaction of some sort. But the rabbinic sages could not anticipate that this project would succeed; maybe they would not even have dared. They were confronted with many hazards, not least a higher mortality rate. It seems more likely that the sages were not, from their perspective, working toward the or even a Talmud, but that they studied, composed, and taught for their own benefit, not knowing what would become of their efforts.

“Learned orality” in late antiquity can generally be described in terms of the declamation of a previously memorized text or as the reading of a text before an audience. Such performative reading is rendered as reciting (recitatio) in Latin.Footnote ⁵⁵ Audiences immersed themselves in content to observe the reciter’s skill in making an argument rather than focusing on the memorization of the exact content of the performance. Indeed:

Still, Theon’s – and only Theon’s – progymnasmata suggest an exercise in attentive listening, or akroasis (Progym. 106–107 P). Students were trained to listen so carefully that they were able to recall the structure of a speech and the most important arguments. The purpose, however, was not memorization but imitation of style, which in this context can also refer to the succession of arguments, intonation, and gesture: “Some younger orators acquired so good an ability by listening to famous orators that their works were attributed to their masters.”Footnote ⁵⁷ Although students recalled speeches in written form, and not orally, quickness of the mind and eidetic memory was obviously the ideal – then and now.Footnote ⁵⁸

This ideal brings us to Elman’s third argument for an oral transmission of the Talmud, the absence of “a terminology for copying, arranging, editing, and redaction.”Footnote ⁵⁹ Several factors may account for this absence without necessarily implying actual oral transmission. First, the texts collected in the Talmud reflect a belief in an originally oral conception of the early texts. Following up with this textual “truth,” the talmudic texts imagine pastoral settings in which genuinely wise and quick-witted teachers instruct their students. Second, antique and late antique texts do not seem to make “the slightest distinction in kind between writing on the memory and writing on some other surface,” as Mary Carruthers observed.Footnote ⁶⁰ This “exact correspondence between the material and the mental library” is then also reflected in the vocabulary used for book production, which converges with the processes of memory and memorization.Footnote ⁶¹ A separation between the two is not always possible. Third, terminology of book production is generally absent from imperial period and late antique literature, a fact that will be the subject of the second half of this chapter. Apparently, processes of book production were so evident that they did not need to be discussed (just as I do not see a reason to inform the reader about how I produced this manuscript).

Elman sustains this argument further with the observation that the verb “to write” is used approximately 3,000 times, while the verb “to say” appears over 70,000 times by his count.Footnote ⁶² Again, I would hold that the structure “XY says” is, first and foremost, the defining structure of the literary unit “saying,” rather than a reflection of actual speech. The saying as a rhetorical device was very popular in the imperial period and late antiquity and was highly theorized.Footnote ⁶³ The declarative saying, for example, the apophantikon, was conceptualized as consisting of a speaker (prosopon) and a meaningful sentence (logos). Speaker and content are unrelated in the sense that the same sentence could be attributed to various people to suit different contexts.Footnote ⁶⁴ Indeed, a saying could make “different points on different occasions. But they are only used to make one point on any one occasion.”Footnote ⁶⁵ The saying had the pedagogical function of associating figures of the past with prevailing standards of correct behavior and speech in the present.Footnote ⁶⁶ Because of these advantages, sayings were used from the very beginning of education. Wooden tablets used by students testify to their struggle to conjugate the verb “to say” in all its variants in order to produce appropriate chreiai.Footnote ⁶⁷

Chreia is the generic term for a literary form that is best explained in the words of a first-century author, since it is a form that is no longer distinguished in this way. The progymnasmata of Hermogenes explain it as follows:

Aelius Theon further explains in his progymnasmata that the chreia, contrary to the maxim (gnōmē) or reminiscence (apomnēmoneuma), is always attributed to a person. Theon praises the usefulness of the chreia not only for pedagogical purposes but for life in general: “A chreia is given that name par excellence, because more than the other [exercises] it is useful [khreiôdês] for many situations in life, just as we have grown accustomed to call Homer ‘the poet’ because of his excellence, although there are many poets” (Progym. 97).Footnote ⁶⁹

Seen from this perspective, if certain rabbinic sages are portrayed to have said something, this relates foremost to the author’s choice to cast a certain logos, often a maxim, as a chreia. The logos benefits from the attribution in at least two ways: First, it is enhanced with an esteemed authority that is thought to support its content. Second, the logos is more easily memorized if it can be associated with the mental picture of a person. Yet, as pointed out above, there is no natural connection between the speaker and the logos. The speaker may, therefore, easily be substituted if they are no longer suitable. Similarly, identical maxims are often attributed to different rabbinic sages within the Babylonian Talmud but also between the Talmud and other rabbinic texts. These changes never affect or distort the content of the sayings, since the characters are, in their function as speakers, without character.Footnote ⁷⁰

In his fourth argument, Elman discusses the size of the Talmud. According to Elman, the Talmud’s size, slightly over 2,000 folia in a codex, does not comport with the writing and book production technology of late antiquity.Footnote ⁷¹ Based on word count, Elman provides an estimate of the length of the Babylonian Talmud in Torah scrolls. In his 1999 article, the estimate was eighteen Torah scrolls; another estimate, in 2007, based on a large, possibly eighth-century fragment of the Talmud, yielded ten and a half scrolls.Footnote ⁷² If we follow another suggestion by Elman, namely, that “each tractate would have been copied separately,” we have thirty-two scrolls or “books,” for that matter, of different sizes.Footnote ⁷³ Elman’s comparison with the Zoroastrian compendium Dēnkard, composed in the ninth and tenth centuries, supports the notion that the Talmud is indeed of an impressive size, since the Dēnkard consists of 169,000 words, whereas the Talmud contains 1,836,000 words.Footnote ⁷⁴ If, on the other hand, the Talmud is compared to Greek and Latin oeuvres, the former’s size is put into considerable perspective.

The draft commentaries of Pliny the Elder (first century), for example, on which his Natural History is based, were written in tiny script on both sides of 160 papyrus scrolls, each 6–10 meters in length.Footnote ⁷⁵ Except for this example, however, calculations as to the original sizes of works are rare.Footnote ⁷⁶ Nevertheless, Manfred Landfester’s Autoren- und Werklexikon gives a good impression of the productivity of Greek and Latin authors or, more likely, authors and their teams of slaves and hired personnel.Footnote ⁷⁷ To give just a few examples, Columella (first century) wrote twelve books on agriculture and one on trees.Footnote ⁷⁸ Josephus Flavius (first century) wrote The Antiquities of the Jews in twenty books, The Jewish War in seven books, and the treatise Contra Apionem.Footnote ⁷⁹ In addition to the thirty-seven books (including the book-length introduction) of Natural History, Pliny the Elder, who died at the age of 55, authored one book on javelin throwing from horseback, two books on the life of Pompenius Secundus, twenty books on the Germanic wars, three books (covering six scrolls due to their bulky nature) called Studiosus, eight books on grammar, and thirty-one books continuing the historical work started by Aufidius Bassus.Footnote ⁸⁰ That is a total of 102 books! Aelian (second to third centuries) composed seventeen books on the Nature of Animals and fourteen books on Variegated History.Footnote ⁸¹ The physician Galen (second to third centuries) is said to be the author of over 250 works.Footnote ⁸² Libanius (fourth century) wrote 1,544 letters (though the originality of two may be doubted) and 144 school exercises (progymnasmata). Augustine of Hippo (fourth to fifth centuries) wrote his Confessions in thirteen books, Contra academicos in three books, De civitate Dei in twenty-two books, On Christian Doctrine in four books, De Trinitate in fifteen books, and five single books. That is a total of at least sixty-two substantial works.Footnote ⁸³ John Chrysostom (fourth to fifth centuries) is the author of 700 orations, twenty sermons, six books of theological discourse, and probably 241 epistles.Footnote ⁸⁴

Another striking feature of these lifetime achievements, apart from their impressive size, is the wide range of topics they cover. A certain polymathy was clearly the intellectual ideal. Given these numbers, Elman’s comparandum, the Dēnkard, may have been an unfortunate choice, since it appears that the time’s book production technology was perfectly capable of composing a work the size of the Talmud, so much so that it even becomes feasible to argue that a single person, with the appropriate amount of help, would have been capable of composing a work the size of the Talmud (and more!) during their lifetime. This would challenge the traditional assumption that generations of sages were involved in the work’s final process of redaction and formation.

Elman’s last argument concerns structures that seem to have a mnemotechnical background, such as stereotypical structures and lists, ring and chiastic structures, and segmentation according to numbers.Footnote ⁸⁵ It is during this discussion of complex structures, however, that Elman concedes that “some written components may well have played a role in the ultimate form” of the Babylonian Talmud.Footnote ⁸⁶ Indeed, while clear structures may facilitate memorization, their artificial makeup does not reflect speech and is more easily conceptualized in writing. Orators composed their speeches in written form precisely because this allowed a conceptualization according to structures that facilitated easy memorization and apparent sophistication.Footnote ⁸⁷ Mnemotechnical structures, therefore, refer foremost to a carefully designed written composition intended to be memorized. However, since these patterns were the ones according to which people learned how to write, and which also defined style, authors customarily used them even when they wrote commentaries or letters.Footnote ⁸⁸ Cause and effect of this entanglement was that, “[e]ven in their most literary moments, Romans preferred imagining texts (at least potentially) as speech acts.”Footnote ⁸⁹

The formulaic, stereotypical formulation of lines of argument in the Talmud, called sugyot (sing. sugya), is another feature Elman raises to sustain his argument. The sugya is a post-talmudic expression used for units that resemble the classical juridical argument composed of “a statement with a support (usually a scriptural or Tannaitic prooftext) followed by a challenge (qushya, קושיה), a resolution (teiruts, תירוץ) of the challenge, another challenge, another resolution, and so forth.”Footnote ⁹⁰ David Brodsky has shown that this structure parallels the teaching in the progymnasmatic exercises “On Thesis” and “On Introduction of Law.”Footnote ⁹¹ These were written exercises for students in a developed stage of the Greco-Roman curriculum. The stereotypical structure of the sugya is, if anything, primarily indicative of rhetorical training and not of an oral learning culture.

The discussion of Elman’s arguments for an oral Talmud has shown that the Talmud can be considered congruent with much of imperial-period and late antique literature precisely because of its preference of the spoken word over the written one. Like other works, the Talmud is silent about technologies of data management and composition. The Talmud further anatomizes language in ways that can only be achieved by literate and particularly schooled minds that “see” words. It takes similarly schooled minds to appreciate respective puns. One would need to assume a voluntary waiver of literacy, which seems tricky in the face of the mortality rate in late antiquity. Indeed, the size of the Talmud seems to suggest a composition from written excerpts rather than oral tradition. And even in that case, data management was a highly sophisticated matter that left clear marks on the text.

Looking Over and Beyond Pliny’s Shoulder: Data Management in the Imperial Period and Late Antiquity

How are we to imagine the process of compiling in the sense of writing with excerpts? There was no formal training for compiling; at least, no school curriculum attests to such. Additionally, authors of compilations often used metaphors when describing their procedures. Macrobius, for example, describes his plan of action for the Saturnalia as follows: “We ought to imitate bees if I can put it that way: wandering about, sampling the flowers, they arrange whatever they have gathered, distributing it among the honeycomb’s cells, and by blending in the peculiar quality of their own spirit they transform the diverse kinds of nectar into a single taste” (Sat. praef. 5 [Kaster, LCL]). If we are to make something out of this metaphor, then Macrobius collected data, stored it, and subsequently redistributed and reorganized it in his book, together with his own thoughts.

The image of the honeycomb is also found among the titles of miscellanies known to Gellius:

Titles like “Meads,” “The Field,” and “Fruit-basket” relate graphically to the etymology of Latin lego, to read (lit., “to collect” or “to cull, pluck”), as well as to the variegated and colorful nature of the miscellany.Footnote ⁹² Other titles reflect the reason for the production of the book, what it means to the author, or what the book should come to signify to the reader. The honeycomb, like the others, is a repository and display of personal collectanea, “the fruits of reading.” One work simply refers to its material form, the wooden tablets from which it is made, hence “Woods.”

As already pointed out in the previous paragraph, data management of literary excerpts was apparently not an ingenious invention but, rather, something quotidian that was not worth an explicit outline. Scholars interested in how ancient authors progressed in fashioning what we, in the last chapter, termed an erudite compilation must cull information from indirect references by authors, the makeup of the text, that is, remains of its original physicality, the text’s shape, its regularities and irregularities, and also from archaeological cues.

Pliny the Elder’s preface to his Natural History and a letter by Pliny the Younger are the most explicit literary sources at our disposal about data gathering in the imperial period. Still, they do not paint a clear-cut picture of how Pliny the Elder managed the production of such a complex work. In the (book-long) preface to Natural History, Pliny states the following with regard to the goal he pursued with his ambitious work: “From a reading of approximately 2,000 volumes … written by one hundred select authorities, I have comprised in thirty-six volumes 20,000 things worthy of consideration – since, as Domitius Piso says, we have need of storehouses, not of books.”Footnote ⁹³ So, Pliny uses a metaphor to describe the type of book he wants to create, a storehouse, and provides a specific number of items he wants to store in this house, 20,000. Obviously, he tried to keep track of the information he had gathered. Judging from a modern recount, however, it appears that Pliny lost track of his data at some point. The numbers, which appear highly exaggerated, are, in fact, not exaggerated at all. The things (rerum) Pliny discusses amount to approximately 34,000.Footnote ⁹⁴ Considering this extremely high number of topics, it is not surprising that Pliny’s method of keeping track collapsed at some point. Nevertheless, he managed to write a pretty consistent work from the thousands of excerpts that he collected over the years. How did he do it?

A letter by Pliny’s nephew and adopted son, Pliny the Younger, is the only testimony for the elder’s method. Alas, many questions remain open since the letter focuses on work ethic and not on compilatory techniques. At least, the information indicates how the elder was able to gather such a vast amount of material within a reasonable time frame: Pliny the Younger describes his uncle (eulogizing) as a ceaseless and driven student who would not waste a single minute. He had the notarius (secretary) by his side at all times with books to read from and wax tablets (pugillares) on which to copy relevant excerpts. A slave read while Pliny indicated the passages that should be copied by the secretary, and although it is possible that Pliny also read by himself, Pliny the Younger generally refers to the books as being read.Footnote ⁹⁵ The relief from reading enabled him to concentrate solely on the content, which he scanned for references to natural substances according to his book project. What can be gathered from this account of Pliny’s method is lacunary, but it nevertheless offers the scaffolding from which to paint a more coherent picture. Based on the information in the letter, a close analysis of NH’s text, and archeological finds, Albrecht Locher and Rolf Rottländer have made specific suggestions as to how Pliny organized his data and finally arranged it into a coherent text.Footnote ⁹⁶

Locher and Rottländer begin their analysis by noting that it is inconceivable that Pliny owned all his 2,000 sources. Otherwise, Pliny the Younger would certainly have mentioned the vast private library he had inherited to a friend in one of his many letters. Instead, he mentions “only” the 160 commentarii with notes that came down to him.Footnote ⁹⁷ Since books needed to be returned to their owners, to a public library, or to a friend, it was necessary to copy relevant passages. Indeed, recurring mistakes such as wrong associations and the misclassification point to the fact that the context of the excerpts was no longer available to Pliny when he finally composed Natural History.Footnote ⁹⁸ Excerpting information was thereby an everyday practice, even a necessity that Pliny turned into the basis and goal of his work, which should function as a storehouse of organized, preexisting knowledge. Excerpting and compiling were both a stylistic choice and a necessity.

Since many of Pliny’s excerpts were apparently taken en route, the notarius must first have made a copy of a certain passage on a wax tablet that could conveniently be carried along. Generally speaking, wooden tablets, waxed or unwaxed, were predominantly used for notes because they could either be covered with another layer of wax or simply be scrubbed off for reuse. Papyrus, by contrast, did not allow for as many reutilizations, and parchment was too expensive.Footnote ⁹⁹ Yet only a restricted number of wooden tablets could be carried along together with book manuscripts; consequently, a single tablet may have served for the copying of several different excerpts. In this manner, completely unrelated excerpts were collected on a single tablet. Descriptores (keywords) had to be added immediately to an excerpt in order for Pliny and his servants to associate them later with the correct main topic and entry. These keywords were important to remember the reason why a particular text had been excerpted, and they helped to distinguish between the mix of excerpts that ended up on the same tablet during one reading session.

Locher and Rottländer suggest that the process of excerpting and organizing the collected data did not end with the wax tablets. It could not have ended there, since the information on the tablets was mixed and the tablets themselves bulky: It would have been difficult to keep track of the keywords, and storage would have taken up a considerable amount of space. In particular, wax tablets need to be protected from anything pressing down on them. This was usually achieved with a rim between the two waxed surfaces that faced each other in a diptych, or with a small wooden cube in the middle of the tablet.Footnote ¹⁰⁰ This characteristic did not pose a problem in everyday use but it is rather impractical for storage. It appears more likely that the excerpts were copied, one at a time, onto another writing surface, before sorting and storing them according to the keywords.Footnote ¹⁰¹

For this intermediate step, scholars have proposed different kinds of writing material, such as “papyrus off-cuts, slates, ostraca, or individual palimpsest sheets of parchment” – materials widely used for taking notes.Footnote ¹⁰² Depending on the size of the literary enterprise and the financial situation and preferences of the author, different and even mixed writing surfaces are indeed conceivable for this step. In the case of Pliny and the enormous number of excerpts he used, Locher and Rottländer assume that a uniform writing surface that facilitates storage and review would have been most suitable. Based on discoveries of wooden slats in the Roman military camps of Vindolanda (England) and Vindonissa (Switzerland), the two scholars propose that Pliny copied individual excerpts onto such thin (0.25 mm/0.01 inch) and very small (20 cm/7.9 inch by 10 cm/3.9 inch) “wooden leaves.”Footnote ¹⁰³

The slats were found in large quantities in the camps; many of them are inscribed with lists and notes pertaining to the organization of the camp, thereby testifying to their usefulness for data management. One such slat also contained a line of Virgil’s Aeneid (9.473), which shows that they were also used for mnemotechnical purposes.Footnote ¹⁰⁴ The slats were light and easily inscribable with ink. Lengthier texts would run over several of these tablets. The sequence in which they had to be read was then marked by diagonal cuts in the corners.Footnote ¹⁰⁵ In other cases, the tablets were bound together in a concertina-like form, a method also known to the writers of the texts collected in the Babylonian Talmud.Footnote ¹⁰⁶

Locher and Rottländer assume that the slats were a phenomenon of the north with its suitable wood and that they served the special needs of the military camps in which they were found. They proposed that Pliny learned about their usefulness when he served as an officer in a camp that sent troops to Vindolanda. Since the tablets in Vindonissa can be dated to the middle of the first century CE, and those from Vindolanda to the latter part of the same century, Locher and Rottländer think they were a recent invention.Footnote ¹⁰⁷

There is at least one other first-century source that mentions small and thin slats and seems to corroborate Locher and Rottländer’s dating. The context of this mention, however, is far removed from the organization of military camps, although not necessarily from the private preoccupations of their inhabitants. In three of his epigrams, the Rome-based poet Martial mentions such wooden slats to which he refers as Vitellian tablets, a name that may have been derived from their manufacturer.Footnote ¹⁰⁸ In book 2, epigram 6, Martial describes how his friend Severus had been so fond of his, Martial’s, epigrams that he copied them on Vitellian tablets and carried them around in the puff of his garment. In a section of epigrams devoted to different forms of tablets, two refer to the usefulness of said tablets:

The delicate tablets were apparently the preferred means for conveying secret messages of love and desire – but not exclusively so, as Martial specifies in the next epigram:

Thin wooden slats thus seem to have served very different purposes throughout the Roman Empire. They met the administrative needs of military camps, served admirers of fine poetry as aide-mémoires, and helped lovers and embarrassed supplicants achieve their secret goals.

Locher and Rottländer are mistaken when they confine the slats to the northwestern Empire. The Palestinian revolutionary Bar Kokhba, for example, wrote a letter on such a thin wooden leaf.Footnote ¹⁰⁹ In fact, the talmudic use of the term pitqa (פיתקא), a loanword from Greek pittakion (πιττάκιον), may refer to a similar, if not identical, lightweight carrier of text. The Greek usage of the term refers to a “tablet, label, ticket, promissory note or receipt.”Footnote ¹¹⁰ Similarly, the Talmud mentions it as a carrier of a writ of summons (b. Rosh Hash. 31b; b. Qidd. 70a), an apotropaic pendant (b. Qidd. 73b), a note shot by an arrow (b. Sanh. 26a), a farewell note (b. Sanh. 96b), and a promissory note (b. Bek. 8b). In three stories, such notes fall from the sky, informing the protagonists about what to do (b. B. Metz. 86a; b. Sanh. 64a; b. Yoma 69b). In yet another story, such pitqa-tickets help two teachers who were banned from the study house communicate with those still inside (b. Hor. 13b). It seems, therefore, that the pitqa is a lightweight “slip.”Footnote ¹¹¹ Besides wood, other materials such as leaves, papyrus, or even parchment snippets, may also have been used to fashion suitable tickets.

Imperial period and late antique epitomizers possessed writing surfaces that supported the collection and storage of excerpts. Copied on lightweight and thin material, excerpts were much easier to handle than bulky wax tablets and could be stored according to main, secondary, and maybe even tertiary keywords. Locher and Rottländer base their thesis of keywords on the mistakes in Pliny’s text, which are likely to happen when using this method, such as wrongly interpreted descriptors, questionable connections, and false comments.Footnote ¹¹² Pliny must have gathered excerpts until his collection seemed exhaustive enough for the project he envisaged. By the time he finally started to arrange commentaries on specific lemmas for Natural History, he had long forgotten about the context of the excerpts, and the books had been returned. The only thing he could do at this stage was verify the keywords and reassign the excerpts, if needed. At this stage it was impossible to make the connection between the Greek name of a plant and its Latin counterpart, for example, if he had only noted either the Greek or the Latin name as a descriptor and stored the excerpts accordingly in separate places. This led to two separate entries on the same plant in two different locations.Footnote ¹¹³ The same happened at times with Greek and Latin city names or names of people. Keywords that could refer to either of two things also challenged this system. Electrum, for example, can refer to both an alloy of silver and gold and to amber.Footnote ¹¹⁴

Although not free from mistakes, the keyword system allowed Pliny to organize vast amounts of excerpts. When he finally started to compose Natural History, which begins with the cosmos and ends with precious stones, he was able to consult the excerpts referring to specific categories and subcategories according to keywords. The slats further allowed him to arrange and rearrange a selection of excerpts until the most meaningful and appealing composition for the entry on a given subject was achieved. It appears that Pliny generally attempted to follow the rhetorical structure (introduction, narration of the case, proofs, and peroration), adding excerpts that did not fit loosely at the end.Footnote ¹¹⁵ Frictions were glossed over by an original commentary. The literary style of the initial sources, however, still shines through in the final version.Footnote ¹¹⁶

The flexible and easily transferable nature of separately copied excerpts helps to account for the complex structure of many texts from the imperial period and late antiquity. Without wasting expensive writing materials, the slats allowed authors to experiment with different arrangements until they found the most suitable structure. To what extent and how – if at all – they included their own voice to link one excerpt with the next remained, of course, an individual choice.Footnote ¹¹⁷

Most of all, the model of excerpts written on light, easy-to-transport materials may account for the complex structures of excerpt-rich texts such as the symposiac works by Athenaeus and Macrobius, or the Babylonian Talmud. It explains how the composers of these works were able to create meaningful units out of a large selection of excerpts because they were able to juggle the excerpts and to experiment with different versions before settling for one and finalizing a subsection. Regarding the Talmud, this would explain subsections (sugyot) that are almost identical but differ in arrangement (e.g., reversed).Footnote ¹¹⁸

To make a case for the Talmud similar to the one made by Locher and Rottländer for Pliny’s Natural History, their thesis needs to be expanded and further substantiated. New evidence, together with a somewhat broader perspective on the subject, allows us to distance Locher and Rottländer’s thesis from an all-too-neat model of index cards and boxes. Rather than on leaf tablets alone, excerpts were probably stored on writing materials of very different shapes and sizes – mostly the material on which they were composed in the first place. This expanded thesis will be the subject of the next paragraph.

The Ubiquity of Excerpting and Reassembling: Writing Material in Late Antiquity

Two distinct practices governed the literary productivity of the imperial period and late antiquity: excerpting and reassembling. By focusing on the small and significant unit, by way of “fragmentation,” “miniaturization,” and “condensation,” authors produced massive works. The procedure appears to point to a Roman-era “connection between acquisition of territory and acquisition of knowledge.”Footnote ¹¹⁹ Yet neither this connection nor the elaborate methods used to excerpt and reassemble are mentioned by authors. This fact may be worthy of closer consideration since authors were otherwise rather keen to highlight the innovative traits of their works. Pliny, for example, praises the index of book titles he provides for the readers of Natural History with the aim of sparing their time. According to Pliny, only Valerius Soranus had provided such an index before he did.Footnote ¹²⁰ Only a few years later, Martial lists a number of reasons why he did not publish more than 100 epigrams in his second book: to save paper, to save on the expenses for a copyist, and to ensure that the book would be short enough not to anger the reader should it turn out that it was not worth the time spent reading it.Footnote ¹²¹ Eusebius, Bishop of Caesarea, consistently highlights the merits of his variously applied tabular methods.Footnote ¹²² But neither data management nor excerpting and reassembling was apparently considered innovative enough to be mentioned. Could it be that these practices were common knowledge, something quotidian, self-evident, and simple, as if they were the only way to produce respective books and texts?

Scholarship justifiably tends to focus on ancient texts that, at some point or another, were copied onto fine parchment or papyrus, either as scrolls or codices. Less prestigious materials, such as palm panicles, bones, soft and hard clay, loose stones (not graffiti on walls, rock, or statues), fabric, gems and semiprecious stones, (noble) metal, leather, skin, ivory, glass, and wood are usually not associated with the production of literary texts but, rather, with school exercises, notes, and amulets.Footnote ¹²³ Yet these materials, the use of which is also attested across texts in Hebrew square script, stand at the beginning of every lengthy text, guiding and shaping its development.Footnote ¹²⁴ Among these, the most common writing materials were wooden tablets, shards of broken pottery (ostraca), and papyrus scraps.

Ostraca were in use throughout the Mediterranean area, the earliest dating to the second millennium BCE and the latest to the eighth century CE, when paper replaced them.Footnote ¹²⁵ Although ostraca have been found with inscriptions in every language spoken in the Roman and Sasanid Empires (Greek, Hebrew, Aramaic, Parthian, Latin, Latin-Punic, Middle Persian, Demotic, and Coptic), the archaeological finds decrease significantly outside of Egypt but cover the modern territory of Iran.Footnote ¹²⁶ The geography of the finds, however, does not represent the actual use of ostraca, which were most likely equally ubiquitously used in the Mediterranean and adjacent areas. Rather, ostraca were subject to decay or the gradual degradation of the script due to unfavorably wet weather conditions. Most importantly, they suffered scholarly neglect until recently.Footnote ¹²⁷

For administrative purposes, such as bills, receipts (bookkeeping), lists, tokens, letters, exercises, testaments, and notes, ostraca were the preferred writing surface.Footnote ¹²⁸ Ostraca lent themselves to writing because of their smooth surface, because of their abundant availability, and because they were free of charge. These, however, were not the only qualities of ostraca. Compared to papyrus, parchment, or wood, they are very durable and therefore suitable for documents meant to last. This explains why they were predominantly used for civil affairs.Footnote ¹²⁹ Archives of ostraca show that ostraca were often further broken and shaped into rectangles to facilitate their use and subsequent storage.Footnote ¹³⁰ Since the ink could easily be washed off, ostraca often show signs of correction and reuse.Footnote ¹³¹ Changes and additions to content attest to the repeated consultation of some, while others were marked consecutively as belonging together.Footnote ¹³² This attests to the sophisticated ways in which pieces of information were collected and stored in a way that made retrieval possible. If necessary, the content of ostraca was copied and systematically gathered on papyrus, while the shards were discarded or reused.Footnote ¹³³

Wooden tablets also had their specific advantages and disadvantages compared to other writing surfaces. It was easier to write on them than on papyrus, they could be reused like ostraca, and they were not as heavy as the latter but more prone to decay. Like ostraca, wooden tablets served very different purposes. Testaments, birth announcements, bills, receipts, and contracts but also sermons, hymns, prayers, and excerpts of literature, exist on wooden tablets.Footnote ¹³⁴ Although wooden tablets continued to be in use at least until the fourteenth century CE, a sharp decrease is observable from the eighth and ninth centuries onward.Footnote ¹³⁵ It appears that everyday writing in late antiquity was governed by independent pieces of written information, which had to be managed and meaningfully stored.

Empirical support for Locher and Rottländer’s thesis may thus be found in the way people organized their daily receipts, that is, in their bookkeeping practices. The “Heroninos archive,” an exceptionally vast and intact collection of letters, documents, and accounts found in the Fayum area in Egypt, provides a rare glimpse into how an estate was managed.Footnote ¹³⁶ The documents, dating to the third century CE, are especially interesting for the present purpose, since erudite works similarly dealt with “big estates” in terms of the number of books and excerpts they handled.

Judging from that evidence, it appears that the managers of subunits of the estate had to account for their expenses and profits on a daily basis. They produced quite short accounts from receipts, which were most likely written on the small and cheap materials discussed above. These accounts were then collected by accountants and merged into a detailed monthly account of all the revenues and expenses effected by the estate. These monthly accounts were consolidated once more at the end of the year before a final fair copy of this annual report was produced for the landowner.Footnote ¹³⁷

This process parallels the one described above for the management of literary data in many ways. Accountants needed to identify a system by which to arrange and store different pieces of information, that is, agree on a shorthand for labeling receipts and entries.Footnote ¹³⁸ The system had to allow for a subsequent, sometimes much later, collation of the information. Similarly, composers sorted their excerpts according to keywords before arranging individual commentaries, drafting, and finally copying them onto a single writing surface. And like the receipts on shards, scraps, and slats that were eventually discarded, the excerpts of a composer, if not purposely discarded, fell prey to decomposition over time. The composers of erudite compilations may not have enjoyed as comprehensive trainings as the phrontistes themselves, but they were certainly aware of the standard methods for archiving documents.

The composition and success of lengthy works, whether yearly accounts or literary works, obviously relied on the production of multiple drafts. As Locher and Rottländer have pointed out, a mobile writing carrier that enabled the arrangement of excerpts before fixing them would have facilitated the production of such a coherent work made from excerpts considerably. Tablets, ostraca, and scraps of papyri would easily have allowed for this sort of mobility. Wooden tablets, in particular, are suggestive of such mobility and suited for convenient storage, since they allow for the drilling of sturdy holes and subsequent bundling by means of strings. Although tablets have been in use for a long time, the small format appears to have been established in the early imperial period.Footnote ¹³⁹

This technology had several advantages. Confidential texts were bound together with their messages facing inward and tied with strings on both sides and/or with a cord and sealed.Footnote ¹⁴⁰ The holes allowed the owner to attach a string and carry the notebook by it.Footnote ¹⁴¹ They could serve the purpose of suspending the tablet on a wall for storage in a school or at home to facilitate repeated reading and memorization.Footnote ¹⁴² Most significant for the present argument, however, is not the fact that tablets with individual content could be tied together but that they could be untied again. If one wanted to change the sequence of the content of such a wooden notebook, or add or remove content, the binding could simply be detached and then retied. In this way, personal collectanea could eventually be organized according to topic and disseminated as an encyclopedic work.

The compilation of notes, literary and otherwise, was thus an everyday practice, and it is not surprising that the codex used for literary texts was a development of the account book rather than a prestigious invention in itself.Footnote ¹⁴³ Indeed, references to such polypticha point out that literature often took form on this basis. Aulus Gellius’s reference to miscellany titled “Woods” (silva) was already mentioned. In his work On Grammarians, Suetonius similarly quotes from a letter that the philologus Lucius Ateius wrote to a certain Hermas: “Remember to recommend my Hyle [woods] to others; as you know, it consists of material of every kind, collected in eight hundred books [libros].”Footnote ¹⁴⁴ The seemingly exaggerated number of 800 books becomes more feasible if one conceives of the work as comprised of wooden notebooks of a more resilient making, holding together a mere handful of wooden tablets. Quintilian associates “woods” with draft versions of personal compositions: “An opposite fault is committed by people who elect to make a draft of the whole subject as rapidly as possible, and write impromptu, following the heat and impulse of the moment. They call this draft their ‘woods [silva]’” (Inst. 10.3.17).Footnote ¹⁴⁵ Still, some people obviously considered even their drafts worthy of a broader audience.

Personal notes, smaller or longer compositions, adapted or paraphrased excerpts, or actual copies appear to have been the basis for larger compositions. This further modifies Locher and Rottländer’s thesis about the practice of excerpting directly, and somewhat exclusively, from books. Indeed, erudite compilations often feature what appear to be distinct excerpts from a well-known author but in a version different from what is considered the original or standard version. This has been observed in Julius Africanus’s Cesti as well as in Ioannes Stobaeus’s Anthology. Africanus even acknowledges at some point that he uses his own version of the Nekyia (Odyssey).Footnote ¹⁴⁶

Through inheritance, endowment, or copying the collectanea of others, people came into the possession of a sort of prefabricated excerpt, which further facilitated the process of compilation and helps account for the speed and productivity of authors. Rosa Maria Piccione suggests that the use of such short anthologies was the reason for Stobaeus’s many dittographies.Footnote ¹⁴⁷ The same practice can account for parallel or almost parallel stories and sayings in rabbinic literature. If a composer (or someone of their staff or team) did not remember or know that the same or similar content they found on a tablet had already been stored in the archive and assigned a different keyword, the piece was inevitably going to end up in a completely different context than its cognate.

Whether draft or fair copy, tablets represented and were associated with people’s personal achievements. This notion manifests itself in the few instances in which the Talmud mentions tablets and notebooks, referring to them as pinqsa (פינקסא), a loan word from Greek pinax (πίναξ). First, the pinqsa are usually attributed to their owner. The notebook of Rabbi Yehoshua ben Levi (b. Shabb. 156a) is mentioned, or the one by Ilfa (b. Menah. 70a). Rabbi Hiyya is said to have had a notebook in which he wrote down his business transactions (b. B. Qam. 99b). The image of personal achievement being visible by the sheer possession of a notebook or through its content stretches into the heavenly realm, where everybody is said to have their own tablet on which their deeds are recorded. This confidential tablet is opened each time someone makes a vow (b. Ned. 22a).Footnote ¹⁴⁸

Beyond the association of tablets with memory as internalized knowledge (e.g., Prov. 3:3 and 7:3: “Write [my commandments] on the tablet of your heart”), the examples discussed here show that tablets were the material locus of personal knowledge and achievement.Footnote ¹⁴⁹ A similar notion can be observed in the Hadith collection Ṣaḥīḥal-Bukhārī (d. 870), which lists the different writing surfaces from which the Qur’an was compiled. The list moves from materials directly to the “hearts of men,” thereby again drawing a direct line between the physicality of writing and the physicality of the human being: “Then I searched out and collected the parts of the Quran, whether they were written on palm leaves or flat stones or in the hearts of men.”Footnote ¹⁵⁰ A later recension of Hadith by Ibn ‘Atiyya (d. 1147) expands the list of materials: “At the time of the messenger of God, the Qur’an was dispersed in the hearts of people. People wrote some of it on sheets, on palm-leaf stalks, on pumice stone, on baked clay, and on other items like that.”Footnote ¹⁵¹ Beyond their religious significance, these passages can be read as sources of information about writing culture, data gathering, and compilation during and after late antiquity.

It may be concluded that taking notes, or even composing lengthy literary pieces, on writing surfaces with limited space, such as slats, scraps, or shards, was a practice that accompanied almost any process of writing and studying in late antiquity. This “piecemeal writing practice” resulted from the material circumstances of the time, from the availability and cost of writing surfaces combined with the increased prestige of literacy and the political need for administrators. These givens simultaneously prompted, suggested, and enabled authors’ work with excerpts. Although the fair copies in the manuscripts or even prints before us still bear the imprints of these intermediate auxiliary steps, the auxiliaries themselves appear to have been left to decay or were reused once a project had been successfully finished. Regarding rabbinic texts that claim oral transmission, it should be asked to what extent these intermediary stages were considered proper writing at all.

Conclusion

This chapter has dealt with many intermediate and unobtrusive steps in the creation of complex works such as Pliny’s Natural History or the Babylonian Talmud. These steps relate mostly to data management which appears to have been a version of the methods applied in bookkeeping. Accountants needed to store receipts and other data retrievably so that they were able to draft a weekly, a monthly, and finally a fair copy of the yearly income and expenses for the landlord. The fact that data management is not discussed by any author of the imperial period or late antiquity confirms the perceived ordinariness of the procedure. A similar sense of ordinariness may also have pervaded the attitude toward the intermediary bits and pieces of information that preceded erudite literary compositions, or a fair copy of the revenue of an estate, for that matter. It should, therefore, be asked whether what rabbinic ontology considers “writing” includes these preliminary notes and compositions, or if the focus is only on fair copies or even only the process of copying the Torah.

This sense of ordinariness is also entangled with the observation that late antiquity’s dominant play with excerpts, its perfection of the “jeweled style,” is simultaneously the result and cause of the era’s material givens for writing.Footnote ¹⁵² Writing on tablets, ostraca, and scraps of papyrus requires precision and brevity, which was converted into a virtue, while the restricting shape of tablets and shards also stimulated creativity.Footnote ¹⁵³

Such small, intermediate writing surfaces are suggestive of the above-outlined thesis by Locher and Rottländer. According to their model, composers of erudite compilations (commentaries, miscellanies, encyclopedic works) collected excerpts on small surfaces and stored them according to keywords. Once composers set out to write an entry or commentary on a specific topic, they gathered the excerpts with the appropriate descriptor and assembled them in a meaningful way. Thanks to the flexible and loose nature of the excerpts, composers could play with different arrangements without spoiling parchment or paper, and without wasting much time until they settled for one. Finally, the composer compensated for any remaining friction, break, or contradiction between the excerpts by adding remarks, explanations, and objections, or by introducing the perspective that was to follow.

Because of the already-mentioned ordinariness of this method, it is only alluded to or described in metaphorical terms by imperial period and late antique authors. Macrobius, for example, describes the pedagogical program behind Saturnalia by invoking the image of the bee. A similar pastoral metaphor appears in the rabbinic treatise Avot de Rabbi Natan (A18), an extra-talmudic tractate concerned with the mishnaic tractate Avot (Sayings of the Fathers):

If the “basket” is taken as metaphor for a miscellaneous notebook, then Rabbi Akiva is here depicted as taking excerpts from different books, metaphorically referred to as wheat, barley, spelt, beans, and lentils. He pools them in a miscellany (the basket) before sorting them out again to provide a commentary on the Torah (the rings). This method matches the one presumably applied by Pliny the Elder in his Natural History.

So far, the metaphor concerning Rabbi Akiva has been read in reference to mnemotechnics and the ancient method of memorizing according to loci. As reported by Cicero, Simonides invented this method when he identified mutilated corpses by recalling where people had been reclining at the banquet before the roof buried them. Cicero concludes that:

The problem with ancient imagery of cognitive processes is that it is based on and shaped along writing processes as Cicero’s quote shows. If tablets are a symbol of personally acquired knowledge just like memory, the “tablets of the mind,” and if honeycombs refer as much to mentally created loci as to the material locus which is the tablet (or the like), it becomes almost impossible to distinguish where a text speaks of the cognitive ownership of knowledge and where knowledge is owned in a material way. In the end, the value of memorization and, even more so, exegesis remains questionable without the verifiable counterpart of a text.Footnote ¹⁵⁶

The next chapter will probe the relationship of the talmudic text to the method of data management described in the present chapter. Since the Talmud itself adheres to the ideology of Oral Torah, it is the text’s form and, quite literally, its “texture” that can tell us something about the processes that led to its construction.