The Babylonian Talmud names no author and carries no date. It runs to something near two million words, and about half of it is spoken by nobody. Named sages make short statements. Around those statements sits an anonymous voice that asks the questions, raises the objections, proposes the answers, and decides what the named men must have meant. That voice has no name because it never gives one.
Who spoke in that voice, and when, is the central unsettled question in the academic study of the Talmud. The answer determines what world the book belongs to, which comparative literature is admissible in reading it, and which languages a scholar has to learn before he can be taken seriously. Nothing in Talmud study rides on more.
The traditional answer is short. Bava Metzia 86a says that Rav Ashi and Ravina were the end of hora’ah, authoritative instruction. Sherira Gaon (d. c. 1006), writing from Pumbedita in 987 to Jacob ben Nissim of Kairouan, fills that out with archival material. Rav Ashi died in 427 by the traditional reckoning. Ravina died in Kislev of the Seleucid year 811, which Sherira gives as December of 499. After Ravina there was no hora’ah. There were the Savoraim, who explained what had been left unsettled and offered rulings close to hora’ah without adding anything of their own. Sherira names some of them. Seder Tannaim ve-Amoraim ends the Savoraic period with Rav Gida and Rav Simona. Maimonides took Rav Ashi to have composed the book. On this account the Talmud is a fifth-century work with a short editorial tail.
Academic Talmud study has been dismantling that account for a century, and the dismantling has a technique behind it. Abraham Weiss (1895-1970) worked on the sugya as a composed literary unit. Shamma Friedman (b. 1937) gave the field its working criteria in a long methodological essay on the sixth chapter of Bava Metzia, published in 1977, which set out how to separate the anonymous discursive material from the attributed dicta and showed that the anonymous layer regularly comments on the attributed layer while misconstruing it. David Weiss Halivni (1927-2022) reached the same terrain from a different direction across the volumes of Meqorot u-Mesorot, beginning in 1968 and running for four decades, and gave the anonymous men a name. He called them the Stammaim, after stam, the anonymous.
Halivni’s claim is that the Stammaim are the authors of the Talmud in any sense that matters. On his model, transmission preserved terse rulings and lost the argumentation that produced them. The Stammaim reconstructed the lost reasoning, and where they got it wrong the forced explanation shows in the text. They then arranged what they had built. The apparatus that makes the Talmud feel like a conversation is theirs.
Then the dating. Halivni moved the Stammaim later across his career, and in the summation that Jeffrey Rubenstein translated as The Formation of the Babylonian Talmud in 2013 he puts the Stammaitic era at 550 to 750. He treats the Savoraim of the Geonic sources as the last of the Stammaim, around 700 to 750. He holds that the Talmud was never closed by any decision or agency, and that it closed on its own around 770 when sages began writing independent books. That is three centuries of anonymous editorial work after the death of the last named Amora.
An arithmetic objection to the late dating circulates, and it is worth stating because it sounds decisive and is not. Count the generations of named Babylonian Amoraim. There are six or seven. Give each generation thirty years. Start the clock around 220, or be generous and start at 250. You reach 430 or 460. So the attributed material ends in the middle of the fifth century, exactly where Sherira put it, and a very late redaction requires a century and a half or more of undocumented work by men who left no names.
The trouble is that the objection describes the late-stam position rather than refuting it. Halivni’s whole claim is that named teaching stopped and anonymous work continued for a long time afterward. The anonymity is the point. These men have no names because they fall outside the chain of transmission that generates names, so counting generations of named figures tells you nothing about how long unnamed men worked. The tradition itself has a label for the interval and Sherira supplies names for some of them. To land, the objection would have to say that a long stretch of undocumented anonymous editorial labor is implausible on its face, which is a claim about plausibility and can be met with other claims about plausibility.
The arithmetic has two other soft spots. Thirty years per generation is a choice. Rabbinic generations are teacher to student, and Rav Ashi is said to have led Sura for something near sixty years. Loosen to forty and the named chain reaches past 500 on its own, which shortens the gap from one end while weakening the calculation from the other. And the traditional sources do not agree with each other. Sherira begins the Savoraim after the death of Rav Yose, Seder Tannaim ve-Amoraim begins them after the death of Ravina, and the two do not fall in the same year. Dorot ha-Rishonim puts Ravina’s death in 475 against Sherira’s 499. The endpoint of the named period is itself a reconstruction from tenth-century testimony about a fifth-century event.
The resistance to the Stammaitic model comes from elsewhere, and its most effective form belongs to Robert Brody of the Hebrew University. In an essay published in the first volume of Iggud in 2008, Brody presses four points. Some anonymous material is early, because Amoraim address it and argue with it. The Palestinian Talmud, whose redaction is placed toward the end of the fourth century, contains a substantial anonymous component, and portions of that material are shared with the Bavli, which means anonymity by itself carries no chronological weight. Attributions drop off in transmission, so a statement can be anonymous because a name fell away rather than because a late man wrote it. And the gap between the plain sense of an Amoraic statement and the anonymous interpretation of it does not establish distance in time, since contemporaries misread each other constantly. Brody also observed that the Stammaitic hypothesis has become the foundation for an entire secondary literature, which means that a great deal now rests on a claim that has never been proved.
Moulie Vidas of Princeton pressed a different objection in Tradition and the Formation of the Talmud in 2014. Vidas noticed that the anonymous voice sometimes strips attributions on purpose, to hold the received tradition at a distance and to mark its own voice against it. If anonymity is a designed effect, then the layering that source critics read as chronology may be an artifact of the composition rather than a record of it. Daniel Boyarin (b. 1946) called this an upsetting of the applecart, and he was right about the scale of it. The distinction between building blocks and construction turns out to be one of the things constructed.
Richard Kalmin framed the question in its blunt form as the title of a book in 1989, asking whether the redaction of the Babylonian Talmud was Amoraic or Saboraic. Nearly forty years on, the field has a consensus about method and no consensus about dates. The stammaitic toolbox is the source-critical method in general use. The chronology it is supposed to underwrite remains contested, and the contest runs along a rough geographic line, with more resistance in Israel and more acceptance in America.
Now the part that has changed the stakes.
Over the past twenty-five years a subfield has grown up around reading the Bavli against Sasanian Persian material. Yaakov Elman (1943-2018) at Yeshiva University drove it. Shai Secunda, who holds the Jacob Neusner chair at Bard, gave it a programmatic book in The Iranian Talmud in 2014. Geoffrey Herman worked the exilarchate and Persian narrative, Kalmin wrote Jewish Babylonia between Persia and Roman Palestine in 2006, Jason Mokhtarian worked the polemical material, and a volume edited by Carol Bakhos and Rahim Shayegan in 2010 gave the enterprise a name in print. The claim is that Babylonian rabbis lived inside a Zoroastrian empire, argued with Persian legal categories, and can be read against Middle Persian sources the way Palestinian rabbis have long been read against Greek and Roman ones.
The chronology decides how much of that reading is available. Put the anonymous layer at 600 or later and the Bavli is a document of the late Sasanian world, and reading it well requires Middle Persian, the Pahlavi legal compilations, and Zoroastrian jurisprudence. Put it at 500 and the Bavli is a late antique Jewish text with Persian contact rather than Persian formation, and the comparanda shift back toward Palestinian rabbinics, Syriac Christianity, and Roman law. The chronology decides which competence a young scholar must acquire to be hired. Fields fight about that with feeling.
It would be easy, and lazy, to explain each side by its interest. A specialist in late Sasanian material has a professional stake in a late Bavli. A traditionalist has a stake in Sherira. A man whose commentary rests on the Stammaim has a stake in the Stammaim. All of that is probably true and none of it is an argument. Attributing an incentive disposes of a position without meeting it, and both sides here have the evidence available to meet each other.
The Persianists are also not uniform on the question, which the incentive story would not predict. Elman built much of his work on Rava, a fourth-century figure, read against Sasanian legal material attested far later. That approach needs continuity in Zoroastrian legal tradition rather than a late Talmud. And the methodological hazard in the subfield runs no matter where the stam is dated: the Pahlavi books were written down in the ninth and tenth centuries and get used as witnesses to what Zoroastrian jurists thought four hundred years earlier. Secunda has written about that problem in print, which is the right way to handle it.
What could move the question? Four things, none of them a matter of temperament.
The Geonic testimony can be worked harder. Sherira had archives. What he had, what he inferred, and where his two recensions differ are all recoverable questions.
Language can be tested. If anonymous editorial work continued into the middle of the eighth century, the Talmud sat in Iraq through the Arab conquest of 637 and a hundred years of Islamic rule. Scholars have long noted how little the Bavli shows of either. The absence of Arabic vocabulary and the near absence of Islam are the strongest available arguments against the very late end of Halivni’s range, and they are arguments from evidence rather than from plausibility.
Aramaic dialectology can be pushed. If the anonymous layer is separated from the attributed layer by two or three centuries, the two should differ in ways that a corpus study can measure across manuscripts.
And Vidas’s challenge can be tested rather than admired. If the anonymous voice removes attributions for literary effect, the cases can be counted and the criteria for identifying them stated. A few good examples establish a possibility. A large sample would establish a rate, and a rate would tell the field how much of its chronology is an artifact of its object.
The Talmud presents its sources as if they belonged to one room. Sages three centuries and a thousand miles apart answer each other in the present tense. That flattening is a stance the book takes, and the stammaitic hypothesis is at bottom the claim that somebody took it, deliberately, later than the men whose words were being flattened. Whoever those men were, they built a book that conceals the fact of its own building. It should surprise nobody that the date of the concealment is hard to fix. The concealment was the achievement.
Notes
Bavli Bava Metzia 86a on Rav Ashi and Ravina as the end of hora’ah. Sherira’s letter is edited by B. M. Lewin, Iggeret Rav Sherira Gaon (Haifa, 1921); an English translation by Nosson Dovid Rabinowich appeared in 1988. The Ravina dating and the discrepancy with Dorot ha-Rishonim are discussed at Encyclopaedia Judaica, s.v. Savora, Savoraim.
Shamma Friedman, “Pereq ha-Isha Rabba ba-Bavli,” in Meḥqarim u-Meqorot, ed. H. Z. Dimitrovsky (New York, 1977).
David Weiss Halivni, Meqorot u-Mesorot, multiple volumes from 1968; The Formation of the Babylonian Talmud, translated and edited by Jeffrey L. Rubenstein (Oxford University Press, 2013), especially chapter one on the dating and identity of the Stammaim.
Robert Brody, “Stam ha-Talmud ve-Divrei ha-Amoraim,” in Iggud: Selected Essays in Jewish Studies, vol. 1, ed. Baruch J. Schwartz, Avraham Melamed, and Aharon Shemesh (Jerusalem, 2008), 213-227.
Moulie Vidas, Tradition and the Formation of the Talmud (Princeton University Press, 2014).
Richard Kalmin, The Redaction of the Babylonian Talmud: Amoraic or Saboraic? (Hebrew Union College Press, 1989), and Jewish Babylonia between Persia and Roman Palestine (Oxford University Press, 2006).
Shai Secunda, The Iranian Talmud: Reading the Bavli in Its Sasanian Context (University of Pennsylvania Press, 2014), and “This, but also That: Historical, Methodological, and Theoretical Reflections on Talmudo-Iranica,” Jewish Quarterly Review 106:2 (2016): 233-241, part of a forum in the same issue.
Carol Bakhos and M. Rahim Shayegan, eds., The Talmud in Its Iranian Context (Mohr Siebeck, 2010).
Yaakov Elman, “Talmud ii. Rabbinic Literature and Middle Persian Texts,” Encyclopaedia Iranica.
Geoffrey Herman, A Prince Without a Kingdom: The Exilarch in the Sasanian Era (Mohr Siebeck, 2012).
