The Hilleli Codex: The Lost Master Copy That Quietly Rewrote the Torah Scroll

An article drawing on the study “Praised from the East” together with published scholarship on the Codex Hilleli.

A Reputation Built on Precision

In the entire history of the Hebrew Bible, few manuscripts acquired the aura of the Sefer Hilleli, the Hillel Codex. Medieval Spanish Jewry regarded it as the gold standard of the biblical text: a complete, vocalized Bible of such exactness that, in the words of the chronicler Abraham Zacuto in his Sefer Yuchasin (circa 1500), it was extremely correct and all other codices were revised after it. Spanish scribes did not merely admire it; they proofread their own Torahs against it, and grammarians cited its readings as decisive evidence. Its prestige in Sepharad rivaled that of the Aleppo Codex in the East.

And yet the book itself has vanished. What survives is a trail of colophons, marginal glosses, masoretic lists, and a handful of medieval copies, from which scholars have tried to reconstruct what the Hilleli was, where it came from, and what it did to the text of the Torah scroll we read today.

The Source: Legend Versus Evidence

Zacuto’s famous account places the codex in the kingdom of León. During a violent persecution in the year 4956/4957 (1196/1197), he reports, the twenty-four books were carried out of a besieged fortress. They had been written, he claims, some six hundred years earlier by R. Moshe ben Hillel, after whom the codex was named “Hilleli.” Zacuto adds that he personally saw two of its volumes, the Former and Latter Prophets, in large and beautiful script, brought out of Portugal after the expulsion and sold at Bougie (Bejaia) in North Africa. David Kimchi, in his grammatical work, testifies separately that the Hilleli Pentateuch was kept in Toledo, and reports that the grammarian Jacob ben Eleazar consulted it there in the twelfth century.

The traditional dating of roughly 600 CE, still repeated in reference works such as the Encyclopaedia Britannica and the Jewish Encyclopedia, is almost certainly legendary. As the e-codices catalog of the Braginsky Collection notes, a codex of that antiquity would predate every known Masoretic Bible manuscript by centuries, and scholars treat Zacuto’s “six hundred years” or “nine hundred years” as pious exaggeration. Some have even questioned whether a single original Hilleli ever existed, since evidence shows “Hilleli” manuscripts circulating in several communities at once.

The study “Praised from the East” pushes the inquiry a decisive step further, in two directions.

First, the physical candidate. Scholarship (noted already in the volume Mechkarim veIyunim) pointed to a fragment in the Russian National Library in St. Petersburg, Ms. EVR II B 115 from the second Firkovich collection, containing portions of Chronicles, Psalms, Job, Proverbs, Esther, Ecclesiastes, Ezra, and Nehemiah, written in the Eastern (Oriental) scribal style. Its colophon reads: “I, Moshe the scribe, son of Hillel, wrote and vocalized the entire Bible, the twenty-four books, for R. Yosef the Sephardi son of Isaac,” dated 4754, that is 994 CE. The scribe’s name matches the Hilleli tradition exactly; the owner’s epithet “the Sephardi” suggests the route to Spain (Jacob Mann conjectured that this Yosef is the famed liturgical poet Yosef ben Avitor). If the identification holds, the celebrated Hilleli was not a relic of late antiquity at all, but a tenth-century Bible produced in the East and imported whole into Spain.

Second, and more consequentially, the study argues that the Hilleli was an anomalous Eastern codex: a book that diverged in a small but consistent set of readings from the authoritative Tiberian tradition represented by the Aleppo Codex, the great early Eastern manuscripts, and the Yemenite scrolls that preserve that tradition. Because Spanish Jewry adopted the Hilleli as its master copy, those divergences were copied into the Sephardic masorah, from there into the rulings of R. Meir HaLevi Abulafia (the Ramah) of Toledo, and onward into every Sephardic and Ashkenazic Torah scroll written since. Only the Yemenites, who never depended on the Hilleli, escaped its influence.

The Most Famous Changes: Nine Letters That Divide the World’s Torah Scrolls

Today’s Ashkenazic and Sephardic Torah scrolls differ from the Aleppo Codex and the Yemenite scrolls in exactly nine spelling variants (plene versus defective spellings, that is, with or without a vowel letter, plus one final-letter change). The central discovery of “Praised from the East” is that for each of the nine, a source explicitly connects the divergent reading to the Hilleli, whether Menachem de Lonzano’s Or Torah, Yedidya Norzi’s Minchat Shai, the masoretic list copied by Christian D. Ginsburg headed “From the book of Hilleli the scribe,” or the 1241 Toledo copy itself. The nine are:

1. Minsoa (מנשוא), Genesis 4:13: “my sin is too great to bear,” written plene with vav in our scrolls; defective (מנשא) in the Aleppo Codex and Yemenite scrolls. Or Torah reports the plene spelling from the Hilleli.
2. Ma’yenot (מעינות), Genesis 7:11: “the fountains of the great deep,” plene in our scrolls per Minchat Shai, Ginsburg’s Hilleli list, and a marginal note in the Damascus Crown (“in the Hilleli: ma’yenot”); defective in the Eastern tradition.
3. Vayehi kol yemei Noach, Genesis 9:29: our scrolls read vayehi (singular form); the Eastern tradition reads vayihyu. The Hilleli reading is attested in Or Torah and in the Damascus Crown’s margin.
4. Tei’aseh (תיעשה), Exodus 25:31: the menorah “shall be made,” written with yod in our scrolls, a spelling that follows an aggadic midrash; the masorah and the Eastern manuscripts have it defective. Minchat Shai traces the plene form to a copy of the Hilleli.
5. HaEphod (האפוד), Exodus 28:8: plene in our scrolls, attested in Ginsburg’s Hilleli list and in the body of the 1241 Toledo copy; the Crown reads it defective.
6. Besamot / Basmat (בשמות): plene in our scrolls against the defective Eastern reading; reported by Or Torah.
7. Chodshechem (חדשכם), Numbers 28:14: here the influence ran the other way; the defective spelling adopted in our scrolls rests on the Talmud (Sukkah) against the masorah, and the 1241 copy attests the defective reading “in the muggah,” which its scribe identifies as the Hilleli.
8. Be’or (בעור), Numbers 22:5: the father of Balaam. The Eastern masorah explicitly counts three plene spellings of the name in the Torah (Vayishlach, Mattot, Ki Tetze), making the Balak occurrence defective, and so it stands in the Aleppo Codex tradition and the Yemenite scrolls. Our scrolls write it plene, following a rival masoretic note of “two defective” found in no early Eastern source. The study traces this orphan masorah to the Hilleli, where two contradictory counting traditions met “at one inn.”
9. Daka / Dakah (דכה), Deuteronomy 23:2: our scrolls end the word with heh; the Aleppo Codex and the Yemenite tradition end it with aleph. Minchat Shai and Ginsburg’s Hilleli list attest the heh.

The study also documents a tenth battleground, the word alehen/aleihen (“to them,” feminine) in Exodus 1:19. The Aleppo Codex apparently wrote it plene with yod, yet in a survey of twenty-nine early Eastern manuscripts nearly all read it defective, and remarkably, in at least seven of them (including the Leningrad Codex and the Washington Pentateuch) the word was originally written plene and then physically corrected, the yod scraped away. This is a rare case where one can watch, on the parchment itself, one masoretic tradition overpowering another.

The Influence: How One Book Reshaped a Continent’s Scrolls

The Hilleli’s authority radiated through every channel of medieval Spanish text criticism:

• Toledo’s scribal dynasty. The colophon of the celebrated Toledo Pentateuch of 1241 states outright that its scribe corrected his work from “a muggah called Hilleli.” That scribe, Israel ben Isaac ben Israel, was no ordinary copyist: the Meiri, in his halakhic work Kiryat Sefer, reproduces the certification of this very scribe as the model of an exemplary Torah text.
• The Ramah and Burgos. A letter of R. Meir HaLevi Abulafia (circa 1180 to 1244), preserved by the Meiri, shows the sages of Burgos checking section divisions against a book they called the “Haleleia,” that is, the Hilleli, alongside books of the Geonim, when Maimonides’ rulings reached them. The Ramah’s authoritative treatise on the Torah text, Masoret Seyag laTorah, became the legal anchor for Sephardic (and ultimately Ashkenazic) scrolls, and the study shows that its deviations from the Eastern tradition track the Hilleli readings.
• The classic text critics. Menachem de Lonzano’s Or Torah (early 1600s) and Yedidya Norzi’s Minchat Shai, the two works that effectively fixed the printed and scribal Torah text used to this day, both cite the Hilleli repeatedly as a prime authority, though, as the study argues, de Lonzano likely knew it only through secondhand copies and lists.
• Marginal afterlives. A Sephardic masoretic Bible now in Munich carries marginal glosses drawn from the Hilleli; the First Ibn Merwas Bible (Toledo, 1300) contains a gloss showing a Hilleli copy still in use there; and the responsa Be’er Mayim Chaim of R. Shmuel Vital records readings copied from an “old book” signed Moshe ben Hillel, which the study identifies as a Hilleli-corrected copy completed in 1229.

The cumulative result, in the study’s striking formulation: subtract the Hilleli, and the Torah scrolls of all communities would agree letter for letter. One eccentric Eastern codex, carried to Spain around the turn of the millennium, is the reason that an Ashkenazic or Sephardic scroll today differs in nine letters from a Yemenite one.

Where Is It Today?

The original is lost. Zacuto last saw its Prophets volumes in Bougie, Algeria, around 1500; the Pentateuch volume is last heard of in Toledo. Scholars suggest it may have been consulted for the final time for a Pentateuch edition produced in Guadalajara, Spain, shortly before the expulsion of 1492. After that, the trail goes cold. If the St. Petersburg fragment EVR II B 115 of 994 CE is indeed a surviving piece of the original (a remnant of its Writings section), then a sliver of the actual Hilleli sits today in the Russian National Library, awaiting a full collation of its readings against the known Hilleli traditions.

The copies, however, survive, and they are scattered across the world:

• New York. The Jewish Theological Seminary holds Ms. 401 (Lutzki 44a), the complete Pentateuch copied in Toledo in 1241 by Israel ben Isaac and corrected against the Hilleli itself. It is the single most important surviving witness; a facsimile was published in Jerusalem in 1974 with an introduction by Nahum Sarna, and a scan circulates freely on the Internet Archive.
• Oxford. The Bodleian Library’s Kennicott 7, a copy of the Latter Prophets completed in 1222, is the other surviving manuscript by the same Toledo scribe.
• Zurich. The Braginsky Collection includes a Spanish Pentateuch of the second half of the fourteenth century copied, per its own testimony, from what was held to be the original Hillel Codex; its catalog ranks its text-critical value alongside the JTS manuscript. The Jeselsohn collection in Zurich holds a Bible completed in 1477 by Moshe ibn Zabara, who states that the exemplar he relied on was the work of “R. Israel the scribe,” extending the chain one generation further.
• Parma. The Biblioteca Palatina’s Cod. Parm. 1996, completed in 1468 by Benjamin ben David Gabbai, declares in its colophon that its twenty-four books are “from the text of the Hilleli.” The study checked all nine variants in it and found them matching today’s scrolls in eight of the nine, with the ninth showing a telltale interlinear correction.
• Munich and beyond. A Sephardic masoretic Bible in Munich preserves Hilleli readings in its margins, and Ginsburg printed a manuscript list of the Hilleli’s plene and defective spellings in his monumental work The Massorah.
• On the auction market. In 2018 Sotheby’s offered a rare medieval Spanish Torah scroll whose hand, in the judgment of the paleographers Malachi Beit-Arié, Benjamin Richler, and Shlomo Zucker, closely resembles that of Israel ben Isaac himself, the man who touched the Hilleli.

Conclusion

The Hilleli Codex is a paradox: a book that no living person has seen in five centuries, whose very existence as a single volume has been doubted, and which nonetheless governs nine letters in every Ashkenazic and Sephardic Torah scroll being written this morning. The research synthesized in “Praised from the East” gives the legend a plausible body: a Bible written in 994 by Moshe the scribe, son of Hillel, in the East, for a Spaniard named Yosef; carried to Iberia; canonized in Toledo and Burgos; copied in 1241, 1468, and beyond; and finally dissolved into the margins, lists, and scrolls of the communities that revered it. The original may be gone, but in a precise, countable sense, the Hilleli is still being read aloud in synagogues every week.

---

Principal Sources

• The study “Praised from the East” (Mimizrach Mehulal), on the nine spelling variants and the Hilleli hypothesis (the uploaded document).
• Abraham Zacuto, Sefer Yuchasin, fifth essay (as cited in the Jewish Encyclopedia, “Bible Manuscripts” and “Hilleli”).
• Encyclopaedia Britannica, entries “Codex Hilleli” and “Codex Mugah.”
• Sotheby’s, Important Judaica, 19 December 2018, lot 213 (on the scribe Israel ben Isaac of Toledo and JTS Lutzki 44a / Bodleian Kennicott 7).
• Braginsky Collection / e-codices (Virtual Manuscript Library of Switzerland), “Pentateuch Based on the Hillel Codex.”
• Center for Jewish Art, Hebrew University, study of the Munich Sephardi Massoretic Bible (on Hilleli copies in Toledo and Burgos).
• McClintock and Strong, Biblical Cyclopedia, “Hillel Manuscript” (on Jacob ben Eleazar, Norzi, and Lonzano).
• Wikipedia, “List of Hebrew Bible manuscripts” (current locations of the 1241 copy and related codices).