most Jewish books written in northern France or Germany in the late twelfth and early thirteenth centuries, Sefer Hasidim is not a commentary on another book, nor is it primarily devoted to Jewish law or the liturgy. It contains thousands of interpretations of biblical verses, but it is not a new biblical commentary nor is it an anthology of earlier ones. A book of religious instruction, it contains hundreds of stories that mirror situations in everyday medieval social living unparalleled in Hebrew literature or in most contemporary Christian sources, but it is not an anthology of Hebrew stories.
Sefer Hasidim as an Inverted Pyramid
To date, scholars have addressed different religious and social questions and selectively quoted from Sefer Hasidim or anthologized it. They have written about its ascetic ideals and penitential practices; its egalitarian views about society; the production of the medieval codex as a physical object; women in business and family life; and Jewish-Christian relations, among many other topics. In each case, scholars selected passages from some versions of Sefer Hasidim for evidence about the subject that interested them as historians or folklorists or students of halakhah (Jewish law). The working assumption behind these studies is that Sefer Hasidim is an important source about the everyday life of the Jewish world of thirteenth-century Germany. And it is.
But it is not the contents alone that make Sefer Hasidim important. Although Sefer Hasidim and other medieval Hebrew books have long been studied critically over the last two centuries for their contents—the message—and in the last few decades for their properties as physical objects—the medium—only recently have they also been examined for their literary form—the medium of the message.9 By this I refer not to the shapes of the letters (paleography) or to the properties of the material on which a text is written (codicology) or to the design of the page (mise-en-page), but to how units of text were composed and put together.10
Few studies have been devoted to the form of the text of Sefer Hasidim that survived in several manuscripts since the thirteenth century. Early on, scholars observed that Sefer Hasidim exists in two editions or versions. One was published in Bologna in 1538 (SHB), based on a manuscript that no longer exists. It consists of 1,176 paragraphs that are defectively numbered so that it concludes with paragraph 1,178 (see Catalog). A version of this text was reissued some sixty more times in the ensuing years, down to 1924 and 1957, when Reuven Margoliot published his two annotated editions, based on Bologna, still in print (SHM 1 and SHM 2).
A much longer version appeared in Berlin in 1891 based on an undated Ashkenazic Hebrew manuscript from around 1300, now located in the Biblioteca Palatina in Parma, Italy (SHP). It, too, is based on a manuscript that is now lost. It was annotated by Jehuda Wistinetzki and reissued with a new scholarly introduction by Jacob Freimann in 1924 in Frankfurt am Main. It consists of 1,999 paragraphs but because of errors in the numbering it ends with number 1,983.11 These are sometimes referred to in the scholarly literature as Bologna or Parma, or as the short or the long text, since Parma is about twice as long as Bologna.
Charts of the paragraph numbers from the Parma manuscript to parallel passages in the familiar Bologna edition appeared in Wistinetzki’s edition of Parma in 1891. Freimann produced a reverse table of paragraph numbers in his “mavo” (introduction) in 1924 so that readers of Bologna could find parallels, when they existed, in the longer new edition of MS Parma.12 Scholars now could find parallel passages that existed in both directions.
But no one asked why Sefer Hasidim was written in such a way that tables of parallel passages were needed in the first place. They were needed because the parallel paragraphs are not in the same sequence in the two versions.13
Generally, we assume that an author composed a book and that it is preserved in one or more manuscripts. Scholars examine the manuscripts, divide them into families, based on copying errors, create a stemma or map of resemblances among them, and ultimately try to reconstruct an author’s original version or lost urtext. This model of a book may be compared to a pyramid with the author’s unique original version at the top and various manuscript witnesses of it at the broad base. This is true of most Western classical books, of the Latin and Arabic traditions in the Middle Ages, and of the Hebrew or Judeo-Arabic books Jews produced in the Muslim world.14
It is not true, however, of Sefer Hasidim or, it turns out, of many Hebrew books that Jews wrote in early Ashkenazic Europe. Authors there composed books in small-paragraph units and then combined them into different editions in the form of an inverted pyramid. At the broad top are the multiple acts of an author’s composition preserved as single paragraphs that make up the particular work. The author (or sometimes a student or relative) combined these short units of text into different parallel editions of Sefer Hasidim and of some other Ashkenazic Hebrew books. We can think of these parallel editions as the narrower bottom of the inverted pyramid. Sefer Hasidim never was written as the author’s single, original composed work that others copied.
A good way to think of Sefer Hasidim and many Hebrew books produced in medieval Ashkenaz is as an “open text” or “open book.” Umberto Eco introduced the concept of “open text” in 1962 and meant by it the multiple readings that a reader brings to a work or the way some modern artists require the reader or performer to complete a work left semi-finished.15 Israel Ta-Shma used the term “open book” in 1993, without reference to Eco, to mean an author like Maimonides who writes a version of his Commentary on the Mishneh and then revises it himself, as demonstrated by surviving drafts from the Cairo Geniza, for example.16 Others did it as well, even in Ashkenaz.17 Had the passages in both editions of Sefer Hasidim been arranged in the same sequence, even with additions or deletions, this might point to an original book that an author revised, the kind of composition that Israel Ta-Shma referred to as an “open book.”
The meaning I am giving to “open book” in the case of Sefer Hasidim and of many other Ashkenaic books refers to an author:
• composing a work in short text units that he sometimes rewrites;
• combining them disjunctively (without linear coherence); and
• producing more than one parallel edition, as opposed to composing one edition that the author or someone else revises one or more times. The term “open book” here refers to writing parallel editions of a book so that there never was only one original edition from which the others are derived.18
A good example from classical literature of an open book created in parallel editions is discussed in David Konstan’s study of the Latin Romance of Alexander. He refers to it as “a special kind of text, which I shall call an open text.”19 That term refers to “the way in which a certain kind of literary work is produced.”20 For example, the Greek Life of Aesop “is composed in segmented fashion” and “the textual history of the Alexander Romance confirms the impression that the text presents of its segmentary composition, which may accrue or lose elements without damage to its structure.”21
The implications about this episodic form of composition and variability in the transmission of these texts means that there never was an urtext: “In the case of what I am calling open texts, such as the Alexander Romance, the History of Apollonius King of Tyre, or the Life of Aesop, however, the effort to retrieve an original form is not only futile but detrimental. For such a procedure would generate a text less authentic than any of the diverse recensions transmitted by the manuscript tradition—a work that in fact no one had ever read or written.”22 He concludes that “an editor’s responsibility is to present one or more of the existing versions as independent texts. … The mistake, however, is to suppose that the various existing versions are false or inferior forms. … Open texts, then, are by nature multiple”;23 “they
