The existence of such manuscripts, and several more not yet discussed in the scholarly literature, proves that Sefer Hasidim began not as a single original composition, but as many single paragraphs that the author and others combined differently into several editions of varying lengths. SHB and SHP were but two published versions of several different editions. It was an accident that they and not the others were published first. The more Sefer Hasidim manuscripts were published, the more obvious it became that neither SHB nor SHP could be the “real” Sefer Hasidim in any objective sense because they were outnumbered by several other editions from some twenty manuscripts (see Catalog and below).
The most ambitious set of publications of Sefer Hasidim traditions appeared online in 2007 as the Princeton University Sefer Hasidim Database (PUSHD), an invaluable research tool. For the first time, scholars could digitally search and compare sixteen versions of Sefer Hasidim. In 2015 the editor added three more: Moscow 103, Frankfurt am Main 94, and the Zurich Fragment. A few others remain to be added.9 These manuscripts confirmed further that Sefer Hasidim traditions were preserved in many short and some long editions made up of parallel and unique short passages arranged in different sequences and not as a single original composition that corresponded to either SHB or SHP, for example, or to some lost manuscript that the others resembled in structure.
But PUSHD also inadvertently reinforced the binary Bologna-versus-Parma character of Sefer Hasidim traditions in the editor’s analysis of the manuscripts, “The Recensions of Sefer Hasidim.” Despite wanting to transcribe all the manuscripts as equal, parallel editions of Sefer Hasidim, PUSHD divided the manuscripts into two groups. It then classified and described all of the manuscripts as they were supposedly related to either or both of the two printed editions (SHB or SHP). The editors of PUSHD apparently did this because most of the manuscripts have at least some parallel passages that are shared with the familiar Bologna edition or Parma manuscript or with both of them. The editorial decision to input SHB and SHP first because of their size or familiarity was one thing. To compare all the other manuscripts to one or both of them separately results in privileging those two editions as somehow being more Sefer Hasidim than all the other manuscripts, including former JTS Boesky 45, for example. This was an unintended but unfortunate consequence of how the database was structured. In the description section, the editors also refer to SHP and SHB as the “principal” versions of Sefer Hasidim, a misleading claim if truly parallel editions were intended.
Size alone does not make Bologna and Parma the “principal” manuscripts of Sefer Hasidim any more than a specific content justifies referring to SHP as “the comprehensive statement of German Pietism recorded in the Parma manuscript.”10 That formulation would exclude from German Hasidism not only the majority of Sefer Hasidim manuscripts but also the other pietistic works of R. Samuel he-hasid, of R. Judah he-hasid, and those of R. Eleazar of Worms as well.
Most of the Sefer Hasidim manuscripts contain some paragraph passages that are not found in either SHB or SHP, and each edition or small set of editions is independent of SHB and SHP even when one or more blocks of passages appear in more than one of them. The editions differ with respect to the order of the paragraphs as well as the contents. Since neither SHB nor SHP was the author’s original composition or derived from a lost original composition, all of the manuscripts need to be treated independently. This also means that each paragraph of any manuscript needs to be compared to all of its parallels.
With the exception of the Parma manuscript that was published in facsimile, we do not have much-needed published facsimiles of the other manuscripts.11 Without them, scholars must rely on transcriptions that are of varying degrees of accuracy. PUSHD notes that it does not correct letters it transcribes even when it is obvious that they are scribal errors; the other short published texts have several errors of transcription.
Single Paragraphs as the Initial Unit of Composition
A clue to Sefer Hasidim’s origins in short paragraph units is suggested by the fifteenth-century moralistic work from medieval Germany called either Sefer Hasidim Qatan or Sefer ha-Maskil by R. Moses b. Eleazar ha-Kohen. The author admonishes his reader not to use a codex as a place to store “scraps of your written ideas” (pitqei ketavekha ve-‘inyanekha).12 That may be a good way of thinking about the beginnings of Sefer Hasidim and other Ashkenazic Hebrew books. Authors wrote down short paragraphs on scraps of parchment and then copied the paragraph units on pages or gatherings of parchment pages.
The process of rabbinic authors and students keeping notebooks (yalqutim) of passages that they wanted to recall or recopy, commonplace books, suggests how Judah might have composed, rather than collected, short passages of Sefer Hasidim, and how students wrote down what they heard from him directly, as we learn from his student, R. Isaac b. Moses of Vienna, and from Judah’s son, R. Moses Zaltman, or from what they read. The fact that most of the short manuscripts or parts of longer ones are referred to as liqqutim or yalqutim may but need not mean “miscellany” in the sense of an anthology taken from one or more authors’ writings. It may also refer to scholars and students writing down passages or notes for future reference, regardless of their source.13 Many of the short manuscripts are introduced by the title “liqqutim.”14 The so-called liqqutim of Sefer Hasidim are not taken from the longer manuscripts of Sefer Hasidim but are independent gatherings taken from many hundreds of single paragraphs, some of which also appear, with variants, as parallels in other editions. Any paragraph may contain a better reading than its parallels in other editions, regardless of the date of the manuscript in which it is preserved.
Another sign that editions of Sefer Hasidim were composed in small units of text is found in a report by Judah he-hasid’s son, R. Moses Zaltman, who preserved his father’s comments on the Humash (Pentateuch) that they studied together. In one passage, R. Moses Zaltman reports that just before his father died in 1217 he “wrote two pages of Sefer Hasidim” (katav bet dappim mi-sefer hasidim).15 Groupings of up to a dozen paragraphs, enough text to occupy two pages, are common in parallels between SHP and former JTS Boesky 45; Cambridge Add. 379 and Oxford Add. Fol. 34 (Neubauer 641); and SHB, the editions with the most topically arranged passages.
For the idea of a bi-folio as the meaning of “bet dappim,” compare in Sefer Hasidim: “If a person comes across a bi-folio (shenei dappim), on one page of which there is writing and one that is blank, and he needs the blank page, he should not cut it off.”16
The process of how some of the longer editions of Sefer Hasidim were written from small text units is suggested in a passage in Sefer Hasidim that describes how the Talmud was put together:
If a man has to sell books, he should sell books of Oral Torah rather than books of Written Torah. For books of Oral Torah are like wool and flax that people work and weave. That is why (a tractate of Talmud) is called a masekhta, a term taken from weaving, as in “into the web” (‘im ha-masekhet) (Judges 16:13–14), as is written about Samson. Laws (halakhot) (are combined) into chapters (li-feraqim), and one gathers together everything pertaining to a subject and that is called a tractate (masekhta).17
This passage appears only in SHP 667 and SHB 932 but not in former JTS Boesky 45, 290, even though parallels to SHP 666 and to SHP 668, the passages immediately before and after SHP 667, are found in it. This pattern is another sign that individual paragraphs migrated from one edition to another. The passage refers to the building up of the Talmud text from smaller to larger units. By analogy, Sefer Hasidim is made up of the author’s single paragraphs or groupings of them, many of which he then further combined into topical booklets (mahbarot). This is the case in Parma and former JTS Boesky 45; Cambridge Add. 379 and Oxford Add. Fol. 34 (Neubauer 641); and eventually in SHB.
Paragraphs