Study 1 · the Esther pilot

What the Esther pilot found

The Greek Additions to Esther have no verbatim reception in ancient rabbinic Hebrew on this bench. What looks like it is a medieval return of the Additions from Latin. The one ancient witness at the level of wording is Aramaic.

The question

The Septuagint's Esther carries six long passages that the Masoretic text does not: Mordecai's dream, two royal letters, the prayers of Mordecai and Esther, Esther's audience with the king, and the dream's interpretation. Since Qumran, material found only in the Greek is presumed to preserve a lost Hebrew or Aramaic tradition. Talmai asks whether any of it survived into rabbinic memory: whether a rabbinic or medieval Jewish text cites it verbatim, reproduces its motifs, or shows no awareness of it at all. The Greek is the diff tool; the Hebrew tradition behind it is the object; the rabbinic corpus is the second witness.

The pilot divided the Additions into 80 units of a clause or a motif each, and searched them against 2,122 paragraphs of 27 rabbinic, targumic and medieval works, from the Bavli and the targums to Rashi and the late Esther anthologies. Every candidate was read against the unit and ruled.

Verbatim: two medieval families and one ancient witness

Esther Rabbah 8:5, 8:7 and 9:7 retell Mordecai's dream, the two prayers and Esther's audience in close Hebrew: the two great dragons drawn up for battle, the small spring that becomes a great river, "not from pride did I refuse to bow to Haman," the maid Esther leans on, the king springing from his throne. These are the parallels the literature names first, and the search found every one of them without being told where to look.

They are not ancient. They belong to the second part of Esther Rabbah, redacted around the eleventh century, and were taken from Sefer Yosippon (c. 953), a Hebrew chronicle that took the Additions from the Latin Vulgate. The Jewish Encyclopedia, the Encyclopaedia Judaica and recent studies say so, and the page itself agrees: in Breithaupt's printing of Yosippon (Gotha, 1707; reissued 1710), twelve phrases of Esther Rabbah 8:5 recur in order, including a frame the Greek does not have, and Breithaupt's own note sends the reader to the Vulgate "translated from the Greek" and says the passage is absent from the Hebrew Bible. Yalkut Shimoni and the Aramaic "Dream of Mordecai" belong to the same family. A second medieval line, Jacob ben Makhir's thirteenth-century Hebrew rendering of the Vulgate, reaches the bench through a modern anthology. Neither is evidence that the tradition survived; both are the Additions coming back.

Targum Sheni 8:13 is different. This Aramaic targum, a work of the seventh or eighth century, quotes the king's second letter clause for clause in the Greek's order: the malice that deceives rulers, the innocent blood, Haman "called our father" and seated at the king's right, his plot on the king's life, the Jews as sons of the Lord of all, Haman hanged. It even carries the first letter's opening, displaced into the second. None of those clauses is the Masoretic text's. Whether the Targum's letter descends from a Greek Esther, from Josephus's paraphrase, or from a Semitic form is open. The pilot does not settle it.

Motif: candidates, not a count

Below the level of wording, 284 pairs share a specific detail that the Masoretic text does not supply. The contact is real, and it gathers where the literature already knew to look: the Bavli's audience scene with its three angels (b. Megillah 15b–16a), Esther's refusal of the king's food (Targum Sheni 2:9, b. Megillah 13a), and the reasons Mordecai would not bow.

But the tier is not clean. An adversarial reviewer drew a seeded sample of forty motif pairs and ruled each against the study's own test; 58% failed, most because the shared detail was already in the Masoretic text, or the two texts answered the same question differently, or the shared word had a different referent. So the motif pairs are published as candidate parallels, labelled as such on every page, and no per-work or per-Addition motif count is claimed.

Where the parallels sit

Verbatim and motif pairs by Addition and by stratum of the witness. The last column is the medieval retranslations, counted apart because they are not reception.

What did not survive

14 of the 80 units have no rabbinic witness at all once the medieval retranslations are set aside. They include the opening of Mordecai's dream itself — the cosmic tumult, the nations mustering at the dragons' cry, the day of darkness — which survives only in the Yosippon family. The rest are Addition A's doublet of the eunuchs' plot, the first letter's decree, Esther's second faint, and a piece of Hellenistic rhetoric about "recent impious deeds." See negative results.

The answer, in one sentence

At the level of wording, no non-Masoretic Hebrew tradition from the Esther Additions is visible in rabbinic memory on this bench; at the level of motif, the audience-scene and prayer traditions are, through the Targum and the Bavli, by conduits this pilot cannot yet name.

That is a negative result, and it is published as one.

What the pilot shows about the method

• A cold machine search recovered every known Esther parallel on the bench, and found the one the model's own memory missed (Targum Sheni 8:13).
• Dates must be assigned per stratum, not per work. Esther Rabbah was first dated as a single late-antique work, and its eleventh-century half nearly carried a false headline.
• A scholarship check belongs before any ruling about how a tradition travelled. An encyclopedia entry overturned a conclusion that every other gate had passed.
• Reading for recall overproduces by an order of magnitude, and a reviewer who reads whole paragraphs does not measure pair-level precision.
• Negative controls, pairs the literature says are absent, are the only precision instrument at motif grade. Every later book needs more of them.

The full procedure, the controls and the verdict definitions are on the method page.