not alone in noting the textual problems of Sefer Yeṣirah, which were, in fact, discussed by most of its early commentators.41 Indeed, the first three commentaries that were written on Sefer Yeṣirah—by Saadya, Dunash Ibn Tamim, and Shabbetai Donnolo—were written on the basis of different versions of Sefer Yeṣirah. The fact that there are three (or possibly more) main recensions of Sefer Yeṣirah42 raises fundamental issues: What is it exactly that we intend to date when discussing Sefer Yeṣirah? Is it viable to assume that there is one urtext written by a single author whose date needs to be determined? How can we establish the date of a treatise when we cannot reconstruct its earliest version and when there is not even scholarly agreement about the very existence of such an original?
Daniel Abrams, in his extensive and comprehensive study about kabbalistic manuscripts and textual theory, suggests an original path to investigate Sefer Yeṣirah. According to Abrams, since it is essentially impossible to reconstruct Sefer Yeṣirah’s urtext and since there are great differences between the manuscripts of this composition, Sefer Yeṣirah scholarship should focus on more valid evidence: the manuscripts themselves. He says, in other words, that there is no Sefer Yeṣirah (Book of Formation) but rather Sifrei Yeṣirah (Books of Formation) and therefore instead of trying in vain to establish the “original” Sefer Yeṣirah, one should trace the history of Sefer Yeṣirah’s acceptance and the ways that this fluid text had been modified over the years by its medieval commentators. Each recension reflects, according to Abrams, a certain moment in Sefer Yeṣirah’s history of acceptance, and that moment should be committed to scrutiny.43
Abrams did not offer textual evidence to support his argument, and although his theoretical suggestion appeals to me, I did not find much support for it in the manuscripts of Sefer Yeṣirah. In my opinion, the textual history of Sefer Yeṣirah should be divided into two stages: in the first stage, before the tenth century, there are indeed differences between the recensions of Sefer Yeṣirah. During that period, the book was edited and reedited by various redactors, and a few glosses were inserted. That was the reason for the discomfiture of its early commentators with regard to its correct version. Therefore, in analyzing the history of Sefer Yeṣirah before the tenth century, I used a similar method to the one that Abrams suggested.44 Nevertheless, in the second stage, after the tenth century, the three recensions of Sefer Yeṣirah remained the main ones, and it would be rare to find new glosses within Sefer Yeṣirah. Therefore, the assumption that the book continues to change during the High Middle Ages has no textual support. From a careful reading of tens of manuscripts of Sefer Yeṣirah, I have not found evidence of conspicuous interventions of late medieval commentators in the versions of Sefer Yeṣirah but rather, the contrary. New versions that combine the short and the long recensions of Sefer Yeṣirah constitute the main modification that can be encountered.
The differences between the versions of Sefer Yeṣirah, hence, occurred before the book was interpreted by its early commentators, and it seems that these commentaries framed its versions. Moreover, even if one scrutinizes the three main recensions of Sefer Yeṣirah, the differences between them are less crucial than might be assumed. At first glance, they are mainly differences in length and manner of editing that did not influence the structure of the book and its basic arguments. Ithamar Gruenwald, who published the first critical edition of Sefer Yeṣirah, has articulated it: “The three recensions differ from one another mainly in the length of the text and in inner organization of the material. The differences of reading between the three recensions are not as many as is generally assumed.”45
There are, as Gruenwald states, great differences between the image of Sefer Yeṣirah in scholarship and the reality of this book according to its manuscripts. We would not be wrong in saying that the textual problem of the version(s) is less complicated than assumed and that those problems were sometimes over-theorized in scholarship. From all the recensions of Sefer Yeṣirah known to me, the basic issues of the book remain stable: in all the recensions, twenty-two letters are divided into the same three groups: immot, “doubles,” and “simples.” Each of these groups contains the same letters without variations, and the discussions about the letters use identical terminology and symbolism. Similarly, in all the recensions, the first paragraphs of Sefer Yeṣirah deal with the ten sefirot, and only minor differences can be found between the recensions. For example, the differences between the long and the short recensions are related to the length of the discussion but are not reflected nor do they have any influence on the meaning or the symbolism of each letter. In the same vein, the great differences between Saadya’s recension and other recensions of Sefer Yeṣirah are related to the way in which the text is edited, but there are merely a few differences in terms of content and terminology.
A different methodology to analyze the textual labyrinth of Sefer Yeṣirah has been suggested by Gruenwald and Ronit Meroz. Forty years ago, Gruenwald suggested that there are thematic and terminological reasons for making a distinction between Sefer Yeṣirah’s first chapter and subsequent chapters of the book and that it seems that the first chapter reflects a different treatise, which was integrated into Sefer Yeṣirah.46 Such an approach can help explain, for example, the opening paragraph of the book by determining the odd number: thirty-two, as an editorial addition. This number thirty-two is not discussed throughout Sefer Yeṣirah; it was added by an editor of the book who combined together the main chapters of the book discussing the twenty-two alphabetical letters, with the new chapter about the ten sefirot. In an alternative suggestion put forth a few years ago, Meroz argues that Sefer Yeṣirah comprised three distinct compositions that are described in the opening paragraph as: a book, a book, and a book (ספר, ספר, וספר).47 If Gruenwald’s or Meroz’s hypothesis is correct, we must suppose, as Meroz noted,48 that the three main recensions of Sefer Yeṣirah all evolved from one branch—after the book was redacted, and that the sections of Sefer Yeṣirah known to us had already been edited at that juncture.49 It would be a mistake to assume that the early forms of Sefer Yeṣirah are merely a result of a redaction of various preexisting compositions. It would be more suitable to perceive it as a combination between original and eclectic materials. In comparison with other late antiquity and medieval compositions, such as the Hekhalot literature, Sefer haBahir, and the Zoharic literature, Sefer Yeṣirah, despite all the differences between its recensions, seems to have a coherent structure with unique and distinct terminology. Of course, Sefer Yeṣirah is a layered text, and preceding the tenth century, its readers edited it, reedited it, and added material. Nevertheless, we have to listen to the manuscripts themselves and observe the great similarities between the three recensions. We should conclude that there was an early composition from which the three main recensions of Sefer Yeṣirah developed, a composition that Peter Hayman tried to reconstruct as “the earliest recoverable text of Sefer Yeṣirah.”50
My main goal in this study is not to publish a new edition of the “early” Sefer Yeṣirah, so I will not discuss every word in the book with the purpose of determining whether it is part of that early version. My purpose is to date and locate the early version of Sefer Yeṣirah; in order to do so, I will determine the terminus a quo and the terminus ad quem of central themes and basic issues that relate to the core of Sefer Yeṣirah and that can be found in all its recensions.
Chapter 1
Discussions About Alphabetical Letters in Non-Jewish Sources of Late Antiquity
Sefer Yeṣirah’s assertions about the role of the twenty-two letters of the Hebrew alphabet were not conceived in a vacuum. Some scholars have argued that the engagement with letters in Sefer Yeṣirah and in other Jewish sources is a unique phenomenon
