What does this quotation, merely by its form and place at the top of this page, suggest about the education and scholarly training of me, the author of this book? It may suggest that I have enjoyed a certain degree of education, since I am apparently familiar with the work of Dorothy L. Sayers and her locked-room mystery, Have His Carcase (1932), and can cite it in English. The quotation may further suggest that I have the leisure to read. When compared to other academic books, placing a quotation at the beginning of an introduction seems an acceptable convention. That I followed this convention suggests that I, the author, was either trained to do so or have absorbed the habit by imitation.

Some of these assumptions are true; others are not. I chose to begin my introduction with a quotation because I have seen this practice elsewhere and have found it to be a pleasant, low-threshold way to start a conversation. Yet I have not read this or any other of Sayers’s books. Rather, I came across another quote by Sayers in the header of an introductory chapter in an academic book. I then looked the name “Dorothy Sayers” up using a search engine and found a website with her quotations. I skimmed the quotations, chose a fitting short one that said something about quotations, and copied and pasted the quote at the top of this page using the appropriate function of my MacBook Air. I have no idea what the rest of the book is about; I just used the excerpt. It may even be possible that the attribution is wrong and that it is a quote from some other book or author.

This type of background information is usually withheld from the reader, and for good reason: It is tedious and breaks the spell of reading. It may even harm my reputation as a serious scholar. For someone interested in the history of the book, however, such information is key to understanding the intellectual, physical, and material processes that have generated a certain book. The Babylonian Talmud and Late Antique Book Culture explores such background information about text production and how missing information may be reconstructed. The book under investigation here is the Babylonian Talmud (henceforth “the Talmud”), a text that offers no or lacunose information as to how it was composed, by whom, or why.

How can answers to these questions be derived from a text that is obviously unwilling to share these secrets? By analyzing content, structure, or form. Traditionally, studies that have inquired into the Talmud’s formation have prioritized content and structure over form. This book takes the reverse approach, prioritizing form over content – so much so that I will quote talmudic passages simply to discuss their form, even their size and physicality, while discussing the content of those passages merely to explain compositional strategies. In the same vein, I have not yet discussed the content of the above quote by Sayers but rather the implications of the quote’s position and its function in marking the beginning of a chapter, and in asserting that I, the author, am well read, thereby revealing at least partly my intellectual background.

Although somewhat randomly chosen, the content of Sayers’s quotation is, of course, not entirely unrelated to the concerns of the present book. The Babylonian Talmud and Late Antique Book Culture is about quotations and how the use of citations as excerpts from someone else’s work may reflect hard work and original thinking rather than help avoid it, as Sayers implies. Indeed, Sayers’s assertion reflects the early-twentieth-century notion that late antique habits of working with excerpts were dull, repetitive, and synonymous with the decline of the Roman Empire. The last century, however – and remember that the quote dates to 1932 – has almost completely inverted this understanding. Scholars are now of the opinion that excerpt literature had its own aesthetics, and that authors often made ingenious use of excerpts, sometimes collating pieces as small as half-sentences.Footnote ¹ Because imperial period and late antique authors tended to work with excerpts – that is, already written text – content was often subordinate to form and method or equivalent with them.Footnote ² This book will explore the historical implications of considering the Talmud a piece of such excerpt literature.

I suggest that we can learn much about how the Talmud was made by focusing less on its content and more on its form. In other words, I suggest that the form of the Talmud, as a whole and in its parts, tells the story of the education of the authors of its texts, and the material and organizational challenges faced by its composers. Education provided the intellectual tools people needed to create or contribute to such a work. The form and structure, that is, the work’s makeup, tell us about the materiality, methods, and technology in play to produce a monumental work such as the Talmud. Form and structure make us think further about the material resources at the disposal of composers and authors and raise questions about libraries, archives, and data management and possible links to everyday bookkeeping, letter writing, book acquisition, and storage.

How Was the Talmud “Made”? Models of Formation

This book argues that existing models of the formation of the Talmud might benefit from engagement with intellectual and material aspects of late antique book production more broadly. Previously, models of formation have been based almost exclusively on the talmudic text, with occasional comparisons with the text of the Palestinian Talmud and other rabbinic texts. This somewhat “intra-familial” perspective has contributed to the – often subconscious – notion “that the Babylonian Talmud is indeed sui generis.”Footnote ³ Other books with long reception histories, not least the Hebrew Bible and the New Testament, are perceived in similar ways.Footnote ⁴ The unique reception history of these works seems to suggest that not only their reception history but the works themselves are singular and that they came into being in ways that differed considerably from the production of ordinary books. As a result, these texts have, for a long time, not been analyzed as material artifacts. Recent awareness of this neglect has caused scholars to develop historically more embedded models for the genesis of the Hebrew Bible or the New Testament, for example, thereby advancing these works’ integration into the material and scribal culture of their time.Footnote ⁵

The present scholarly consensus as to how the talmudic text came into being and how it must be analyzed leans toward the “two-source theory.”Footnote ⁶ This theory basically divides the text into three layers: an early layer, which attributes sayings and tenets formulated in mishnaic Hebrew to scholars classified as Tannaim; a later Aramaic stratum of sayings, which are attributed to the scholarly generation of the Amoraim; and a final closing layer, which negotiates anonymously between the different dicta.Footnote ⁷ Although these seem to be three sources, the decisive divide is the one between attributed statements (i.e., tannaitic and amoraic) and an anonymous voice that comments upon them, often bringing these quotations into conversation with each other, thereby contributing to the Talmud’s characteristic dialectic form. This mediating, explanatory layer must obviously be the latest layer. Additionally, one can differentiate between concise tenets and sayings attributed to rabbinic sages, short stories, and lengthy narratives. The latter are usually also attributed to the latest layer. The dating of this final layer is a matter of debate. Since it connects to the final formation of the Talmud, the stratum is usually seen as a lengthy process that scholars place somewhere between 450 and 750 CE.Footnote ⁸

Because the earlier two layers are traditionally attributed to generations of scholars (i.e., Tannaim and Amoraim), David Weiss Halivni proposed to attribute the final, unattributed stratum similarly to such an intellectual generation. He called this generation the Stammaim, after the Aramaic setam, or “anonymous,” the name also given to the mediating voice. Unlike the generations of the Tannaim and Amoraim, then, the Stammaim were not identified by medieval historiographers. According to Halivni’s thesis, the Stammaim reconstructed the dialectical argumentation that had been lost in the process of oral transmission. In this process, he argues, reciters had mostly memorized concise dicta by Tannaim and Amoraim.Footnote ⁹ Based on the knowledge of these reciters, then, the Stammaim completed the arguments and wrote down the Talmud.

Other scholars, most notably Shamma Friedman and Jeffrey Rubenstein, have combined the thesis of the stammaitic redaction with tools of higher criticism developed in biblical studies. These tools have proven helpful for isolating certain patterns and, especially, for systematizing a set of questions with which to confront the text and to distinguish between earlier and later stammaitic narratives.Footnote ¹⁰ Friedman disagrees with Halivni over the origins of the dialogue structure, which he does not understand as an artificial stammaitic reconstruction of a lost discussion. Friedman, rather, attributes the characteristic dialectic structure to the creativity of the stammaitic “commentators” who redacted the Talmud.Footnote ¹¹

Richard Kalmin has proposed yet another way to disentangle the Talmud’s obviously quite disparate – in terms of language, style, and content – pieces. Kalmin’s model mediates between the medieval tradition and higher criticism. He uses attributions to certain rabbis to identify the chronological and local background of the material. In his words, he looks for “general patterns characterizing Palestinian and Babylonian and early and late rabbis, all the while remaining alert to the possibility that the transmitters and editors of these traditions altered them in subtle or not so subtle ways.”Footnote ¹²

These models are in continuation of earlier ones that stressed the chronological succession and local nature of certain compositions and editorial processes. Isaac Halevy and Zacharias Frankel, for example, emphasized the contributions by scholarly generation: each generation would have their own Talmud, since they continued working on the one transmitted and organized by the previous generation.Footnote ¹³ Jacob Epstein and Eliezer S. Rosenthal broke this model down to local teachers, each of which taught his own version of the Talmud. Although eventually merged into a single work, every tractate was a book on its own and with its own editorial story.Footnote ¹⁴

Jacob Neusner contested the positivistic interpretation of medieval historiographers and criticized the practice of focusing on details within the text before proceeding to the work as a whole.Footnote ¹⁵ He held that the whole of the composition should be considered before moving on to a detailed analysis. Following this path, he pointed to the distinct overall style of the Talmud and argued that the Talmud had been written and composed according to an identifiable set of rules.Footnote ¹⁶ These rules produced two different sets of documents: documents that concentrated on the Mishnah and documents that focused on other things, which Neusner called “miscellanies.”Footnote ¹⁷ According to his thesis, then, the Talmud’s authors composed the work from documents of various sizes according to a detailed and specific program responsible for the characteristic pattern in the arrangement of the different documents.Footnote ¹⁸

The models obviously agree that the Talmud is a composite document, a patchwork made of many different sources. These sources have been interpreted in different ways, as being reflective of different places of origin, times, or teachers. Unarguably, the production of this material, whether written or oral, took time and was carried out in different locations. The question that remains to be answered is how the Talmud was actually produced. Only Neusner’s admittedly vague “documentary hypothesis” differs in that it reckons with an active, strategic production of the Talmud out of written texts. The other models have a rather sedimentary view of how the material came together, maybe intermitted by an occasional organization and systematization, and a final overhaul by the Stammaim. The reason for this complicated, undecided, and somewhat singular model is the fact that many scholars reckon with an oral tradition of the Talmud.Footnote ¹⁹ If, of course, the bits and pieces that constitute the Talmud were not written texts but oral traditions, the production of the final written composition of the work is a highly complex project.

Indeed, oral transmission may explain the overall oral notion of the Talmud’s dialectic form, the sayings, the reciting, and, most of all, the concept of “oral Torah” that has lingered over rabbinic literature since mishnaic times.Footnote ²⁰ Then again, oral transmission is suggested by a text that is unwilling to tell us anything about its genesis; a text that is more often than not not corroborated by archaeological evidence such as inscriptions or graffiti;Footnote ²¹ a text whose historical reliability has been questioned in many ways.Footnote ²²

What if the Talmud Had Been Composed like a Late Antique Compilation?

This study considers the possibility of bringing the Talmud’s characteristic features, its overall structure and outlook, into conversation with imperial period and late antique literary production. For this purpose, I will have to reconfigure some of the models discussed above. I am aware that this may be a stretch in some areas and for certain readers. Yet such a turnaround might offer possibilities for expanding the tools currently available for analyzing the Talmud. These tools, as Moulie Vidas has insightfully observed, have been shaped in ways that direct the user, apparently inevitably, to see layers, and, especially, the seemingly earliest ones among them.Footnote ²³

There is, in fact, good reason to approach the Talmud simply as a late antique compilation, that is, a book assembled according to an elaborate plan that followed upon a period of sorting excerpts according to keywords. First of all, compilations were a popular genre from the imperial period through late antiquity. They ranged from a simple mix of excerpts from other works and personal notes to structured compositions in which an explanatory voice guided the reader or listener from one excerpt to the next where necessary. Excerpts from the same source ended up in different places: divided and yet connected through style and content, same-source excerpts covered compilations with a net of recurring motives and linguistic tropes that sometimes ran counter to the structure and topic of their newly assigned place in a compilation. A similar connectivity throughout the work can be observed in the Talmud, where words, phrases, bits, and pieces of the very same source span an interlocking web over the text and, in fact, define it as a “book.”Footnote ²⁴ Indeed, the overall organization, the use of the very same or slightly adapted narrative in different places just because it makes a point in both cases, is stunning.Footnote ²⁵ Then again, logical gaps as well as stylistic and linguistic differences point to the fact that the material had not been written for the particular place where it ended up.Footnote ²⁶ All of these observations give reason to compare the Talmud to late antique compilations and the material and intellectual preconditions for book production.

Recent scholarship has pointed to the talmudic texts’ multiple entanglements with its literary co-world, and that the rabbinic movement itself may be framed as an association in the form of an exclusive study group.Footnote ²⁷ Whether such groups had a wider social impact or not, their members tended to entertain and challenge one another not just with arguments and expositions but also with riddles or astute stories, which were prepared in advance and then read or recited from memory.Footnote ²⁸ It is also conceivable that people took notes from such meetings and transferred the most compelling contributions into more concise forms, that is, sayings or maxims, which ended up in collections at a later date. Most likely, the members of this rabbinic association were also members of other consortia, and their personal notebooks may have offered an interesting mix of topics. The synagogue, for instance, does not seem to have been identical with rabbinic forms of organization. Still, some rabbinic sages appear to have given public lectures in synagogues, given legal (halakic) advice, or consulted with teachers of children.Footnote ²⁹ The preparations for such lectures may have yielded some form of text, which eventually provided teachers with a model or exercise text, thereby multiplying its influence. The cases brought before the rabbinic sage may have been cause for halakic discussion with colleagues, which also resulted in the jotting down of some thoughts.

I do not claim here that rabbinic sages composed elaborate texts the length of a scroll or even a whole tractate, as cautioned by Sussman.Footnote ³⁰ Rather, I think of tablets, ostraca (pottery shards used for writing), and rotuli (a long, narrow strip of [waste] parchment or papyrus that opened vertically), or papyrus scraps, material suited for short compositions, and private notes. These were not fair copies destined for formal use and dissemination (“publishing”); rather, they were forms of texts that even today would not be considered “real writing.”Footnote ³¹ Nevertheless, as will be discussed in Chapter 2, these notes reflected one’s personal achievement and were held dearly. The compilation of the Talmud would have required that these compositions were eventually gathered in an archive or a sort of library that served students and scholars even prior to this endeavor.

For the purpose of producing the Talmud, the material was sorted, significant passages were excerpted and these were arranged according to keywords. Since the work was to follow the text of the Mishnah – which was maybe only available from memory, maybe in the form of the notes just described – lemmas were identified. Keywords were assigned to the lemmas, and commentaries were crafted with the material yielded through the keywords. Although connected through keywords, the material assembled in this way was, of course, inconsistent, and the composers had to add editorial notes in order to connect the pieces. Questions, objections, and clarifications seem to have been quick strategies for solving these problems. Lengthy excerpts such as stories were taken apart when needed and rearranged. Names could easily be exchanged or added as another means to create connectivity through association.

This model for the formation of the Talmud would account for several of the work’s main features observed in earlier models: The used texts were chronologically and geographically diverse and there were older texts and more recent ones, although style should not be used as the only decisive factor for dating, as Robert Brody and Vidas have pointed out.Footnote ³² An active composition process, which left the excerpts mostly intact, took place.Footnote ³³ The composers had an overall plan, a point on which Neusner insisted, and they applied the same strategies over and over again, as observed by Friedman. The excerpts are apodictic in form, and the explanations added by the composers are often forced, since they are no longer aware of the context, both points that are important to Halivni’s thesis. This model agrees further with Boyarin and Vidas that the authors of stories are not, at least not necessarily, identical with the composers of the Talmud.

Method and Chapter Outline

This book explores avenues into background information about the production of the talmudic text. As noted, the text itself is reluctant to provide such information, and where it seems to do so, we may be facing ideology, literary convention, or imagination rather than a historical account. The method suggested here is, therefore, a focus on form and convention rather than on content, the content’s attribution to a certain sage, his generation, or his geographical location. This approach allows us to identify the intellectual and material preconditions that are responsible for the text’s composition and structure. The way I look at the texts, then, is informed by form and source criticism and thus focuses on literary patterns, style and vocabulary, and genre. This focus on form cannot, of course, happen in a vacuum if there is to be any historical validity to it. In this regard, I clearly must compare the Talmud to other late antique works as the scaffolding for a thesis about the Talmud’s production.

Although comparison is probably the most ubiquitous scholarly practice, it is often not recognized as a method.Footnote ³⁴ Arindam Chakrabarti and Ralph Weber have recently rehabilitated the use of comparison as a worthy academic method that can even be used to compare the proverbial apples and oranges, if applied correctly.Footnote ³⁵ They emphasize the importance of defining the tertium comparationis, a third element regarding which a comparison is carried out. This “third in comparison” provides the analyst with “a neutral third place or at least a third philosophical point of view.”Footnote ³⁶ It defines and clarifies with regard to what two comparanda are compared to each other. The tertia used in this study to compare talmudic texts to others are hence primarily form and structure, genre, and practice.

By so doing, however, I also compare the textual productions of different cultures and subcultures to each other, Babylonian Aramaic texts to Palestinian Aramaic, late Hebrew, Syriac, Latin, Greek, and even Coptic texts. The tertium in that regard is similarity, that is, what is similar in form and practice, not necessarily (and, indeed, often not at all) in content. The liberty I take in comparing texts across geographical and linguistic boundaries, even across a certain time period (mostly the first through the sixth centuries CE), is motivated by two facts. The first and rather simple fact is that intellectual and technological inventions, the focus of the present book, travel notwithstanding their origin. A clear sense of origin often withers quickly; the source of innovation becomes intractable and is reattributed to the same degree its success grows.

Second, by comparing the Talmud as a book to other books, the study partakes in the recent scholarly endeavor to bring the Babylonian Talmud into conversation with texts written in other languages and under different ideological perspectives, that is, the work’s cultural context and social history.Footnote ³⁷ How cultural and ideological boundaries are defined and drawn and how the relationship between such entities is imagined governs the choice of comparanda, which are chosen based on aspects that are “presumed to be common to both” – the tertium comparationis.Footnote ³⁸ Thus, for example, it is well known that rabbinic literature expresses an ambivalent relationship toward Greek language.Footnote ³⁹ Together with the rabbinic emphasis of “oral Torah,” this can quite easily lead to the notion that rabbinic learning operated on completely different premises and in different settings, and that similarities came into being by way of an elusive and indescribable osmotic process. Recent comparisons between rabbinic literature and Roman schooling standards (i.e., rhetoric) have proven fertile and justify positioning the curricular standards described in the progymnasmata, treatises describing preliminary rhetorical exercises, as a tertium comparationis between Roman and Persian or rabbinic literature.Footnote ⁴⁰ Thereby discussion about ideology and language become secondary, since the focus shifts to style, and ultimately also to modes of production.

The problem with reconstructing the ways in which complex, excerpt-based books were produced in late antiquity is that there exists no account of how this was done. In this regard, the Talmud does not differ from other books. Not a single “author-composer” felt the need to inform posterity or even their contemporaries about how they came up with the idea for a certain book project, how they planned it, how much money and time it cost, whether they received help from slaves or hired personnel, how they managed their data, and how they finally went about composing their work. There is, therefore, no account that can be compared to the structure of the Talmud in order to see whether there are similarities. Comparison between the Talmud and other compilations has therefore led to observations regarding the production of compilations that are relevant to the study of book history more broadly.

The “third in comparison” that I use in Chapter 1 is genre, or, since genre is an elusive category, structure and outlook. In that chapter, I compare books that convey knowledge using mostly excerpts from other works. Some of these works adhere to a fixed structure, such as Pliny the Elder’s Natural History, which, in thirty-six books, proceeds through all kinds of natural substances, starting with the planets and ending with minerals. Other works, such as Aulus Gellius’s Attic Nights, are purposefully unstructured. Two works are particularly interesting comparanda, since, like the Talmud, they arrange excerpts into a long dialogue that covers all kinds of topics: Athenaeus’s The Learned Banqueters and Macrobius’s Saturnalia. The latter basically turned Gellius’s Attic Nights into a symposiac dialogue. The arranging of excerpts to form a conversation was a method that was widely known and praised for its pedagogical value. Analysis of these works shows that the compilation of pieces of knowledge into a meaningful, dialectical work had several possible motivations, including antiquarianism, a pressing need for preservation, personal ambition, and/or the wish to bequeath knowledge to the next generation in a simple and compact manner.

Chapter 2 is devoted to the data management needed to compile coherent works out of excerpts. So far, there is only one specific theory to this issue, established by Albrecht Locher and Rolf Rottländer based on Pliny the Elder’s Natural History.Footnote ⁴¹ They proposed that Pliny first set out to roughly list the categories he wanted to cover. Over many years, Pliny then collected passages from books he read or that were read to him, and that addressed a topic relevant to his project. Locher and Rottländer suggested that these excerpts were copied on thin, small wooden slats, such as those found in Roman military camps at the time. Because he stored the excerpts according to keywords, Pliny was able to retrieve the material once he started writing about a given subject. Every excerpt was written on an individual slat, which allowed moving them around until a rhetorically appealing structure was achieved. Without wasting much paper or even parchment, Pliny could now add complementary information or transitioning remarks to round off the paragraph.

The thesis is compelling but maybe a little bit too “neat” in the way it reckons with a wooden form of index cards and matching boxes. A closer look at the materiality of late antique writing culture shows that writing generally happened “on bits and pieces”: on wooden tablets, ostraca, or papyrus scraps. Much writing was thus already portioned and, as a result, could easily be stored according to keywords. Bookkeeping practices further substantiate Locher and Rottländer’s thesis. Receipts were collected and drafted into weekly and monthly accounts that were eventually assembled into an account of expenses and income for the whole year, which, in turn, was transmitted as a fair copy to the landlord.

Chapter 3 analyzes three talmudic passages, which I will call commentaries, that run from one mishnaic lemma to the next according to the model outlined in Chapter 2. Chapter 3 asks how keywords were assigned and whether the composers pursued a certain structure with the excerpts, such as the rhetorical four-part structure: proem, narration, proofs, and peroration. The chapter shows that the assigned keywords usually go far beyond the one suggested by the mishnaic lemma. This is most obvious in the cases in which a commentary to the same lemma exists in the Palestinian Talmud. In these cases, the keywords also relate to the issues raised by the commentary of the Palestinian Talmud to the respective lemma. This creates the oft-observed notion that the Babylonian commentary is, in very subtle ways, similar to the Palestinian one. Moreover, this move beyond the mishnaic lemma is responsible for the notion that the Talmud is a “commentary plus.”

Chapter 4 looks at the intellectual premises needed for a project such as the compilation of the Talmud and how they play out in other, smaller literary forms such as talmudic stories. The preliminary rhetorical exercises, the progymnasmata, provide insight into the late antique curriculum. The progymnasmata, of which copies from several different authors and in several translations survive, do not just offer exercises but discuss what these exercises are designed to achieve in students. They give us a glimpse into the intellectual principles according to which late antique authors operated. Taking the progymnasmatic principles as a tertium comparationis between late antique narratives and the narratives in the Talmud, I show how the latter were created according to the same principles, and how this reflects the training that the authors of these narratives received.

The chapter further shows how the methods applied to the sorting and arrangement of excerpts into a book were also used on a microlevel to fashion stories. Rather than writing a story from scratch, students learned to work with already existing plots and enhance them with quotations or to combine them with another plot, thereby working with excerpts from other texts. Like authors of whole compilations, authors of short compositions would start with an inquiry into other works, collecting small excerpts that would substantiate the case they were about to make with their story. The story about Ashmedai, Solomon, and Benaiah (b. Gittin 68a–b), for instance, turns out to have been crafted based on a Persian narrative, into which the biographical details of these three protagonists were meticulously integrated.

In Chapter 5, I reverse the process of excerpting and compiling according to the observed methods by following the structure of a particular medical recipe throughout the Talmud and by reassembling an Aramaic treatise of fifty-seven simple remedies. Such treatises were rather popular between the fourth and early seventh centuries CE, and the treatise presented here is the first Judeo-Aramaic exemplar of this kind. The reassembling of a source that was dissected for compilatory purposes reveals further strategies employed by the composers of the Talmud, who seem to have worked at quite a fast pace, often repetitively, but without neglecting the attempted impression of an overall conversation.

Seen from this perspective, the use of texts written by someone else, quotes and other excerpts, appears as highly sophisticated, with a lot of innovative potential, and not at all “unoriginal,” to paraphrase the introductory quote. Measured against the available technologies of the time, the Talmud appears to be one of the time’s finest compilations.