The Fire & the Veil · Foundation of Asha · dataset
The Persian Passband
Which Persian words entered the Hebrew Bible — and which did not
What this is. Every language that touches another leaves a fingerprint in what it lends. Borrow cooking words and the contact was kitchens; borrow court words and the contact was offices. This dataset tags 86 non-Semitic loanwords in Biblical Hebrew and Biblical Aramaic — Old Iranian, Akkadian, Egyptian and Greek — to the semantic fields of an external linguists' taxonomy, with every entry carrying its verse citations so you can check it against a Hebrew Bible yourself.
The Iranian result: 39.5% of 38 loanwords fall in law and political office, against an 8.3% expectation if they were spread evenly across the taxonomy — and the concentration rises to 50.0% when you keep only the etymologies nobody disputes.
⤓ Download CSV⤓ Download JSONThe pre-registration (DOI)
This was a prediction before it was a result. The claim that Iranian loanwords would cluster in legal-administrative fields was deposited to Zenodo on 10 August 2026 with a timestamp, four days before the test was built. The order is verifiable against the deposit record without taking our word for anything. The taxonomy is the World Loanword Database's 24 semantic fields, chosen because it is external to the hypothesis and its categories could not be shaped to fit.
The finding
Social & political relations
across 24 fields
one-sided
etymologies
The words themselves are the argument: dāt (royal law), pitgām (edict), 'aḥašdarpān (satrap), gizbār (treasurer), dətābar (judge, built on the same root as dāt), patšegen (the certified copy of a document). A civil service, borrowed whole.
- Social and political relations10
- Clothing and grooming5
- Law5
- Possession4
- The house3
- Cognition2
- Miscellaneous function words2
- Speech and language2
- Agriculture and vegetation1
- Emotions and values1
- Food and drink1
- The body1
- Time1
- Religion and belief0
Read the last row carefully. Zero Iranian loanwords fall in the taxonomy's Religion and belief field. That is reported here as a count, and as an existence proof — the same books borrowed cultic vocabulary from other donors, so the channel was open and simply was not used from this direction. It is not offered as a statistical result; see the limits below.
The control is twelve verses wide
The obvious objection is that Persian loans look administrative because they occur in books about the Persian court. Daniel 3 answers it, because two donor languages are exhaustively attested inside one scene. The officials summoned in verses 2–3 are satraps, prefects, governors, counselors, treasurers, judges and magistrates. The instruments listed in verses 5, 7, 10 and 15 are the horn, pipe, lyre, trigon, harp and bagpipe — and the Greek loans of the entire Hebrew Bible are essentially that list.
Same author, same chapter, same genre, twelve verses apart. Every Greek loan is a musical instrument; every Iranian loan is an office. Genre is held constant by construction and only the donor varies. Within Daniel the Iranian passband share is 50% against the Greek stratum's 0% — directionally exactly as predicted, and not statistically significant, because the entire Greek stratum of the Hebrew Bible is 4 words and no test on four words can be.
The four donor strata
| Donor | Lexemes | Coverage |
|---|---|---|
| Iranian | 38 | Aims at completeness for common nouns and adjectives of Old Iranian origin in the Hebrew Bible and Biblical Aramaic.… |
| Greek | 4 | COMPLETE for the Hebrew Bible on the standard view. The Greek stratum in Biblical Hebrew and Aramaic is famously confi… |
| Akkadian | 32 | A SAMPLE, not a catalogue. Restricted by the stated inclusion rule to lexemes attested in the six late books. The Akka… |
| Egyptian | 12 | A SAMPLE added in the v0.2 expansion, assembled by the same rule as the Akkadian stratum. Egypt is the third long-cont… |
Why the comparison strata exist. A passband is defined by what it filters, so the Iranian stratum means nothing without rivals to compare it against. Akkadian was the prior imperial language of the region and its stratum was expanded deliberately to include the strongest counterexamples — peḥāh (governor), səgan (prefect), sārîs (court official), tipṣar (scribe). Two borderline calls were resolved against the hypothesis and flagged in the data.
What this does not show
It does not show that Zoroastrian theology entered Judaism. A loanword passband identifies the channel a vocabulary travelled. The step from vocabulary to theology is an inference this dataset does not make and cannot license. Anyone citing this page as proof of religious borrowing is citing it for something it refuses to say.
One result here survives multiple-comparison correction, and it is the concentration finding above. Across the six instruments this dataset belongs to, 53 tests were reported and 2.7 false positives were expected by chance alone. The concentration result clears even the strictest correction at p ≈ 10−7. Several weaker results that were initially reported — including a statistical version of the Religion-field asymmetry — do not survive, and were retracted.
"Religion and belief" is a modern category. The idea of religion as a domain separable from law, kingship and craft is a product of early modern Europe. Achaemenid administration and Achaemenid cult were not separate spheres, and dāt means both royal decree and, at Ezra 7:12, the law of the God of heaven. The zero in that field may be measuring the absence of a category rather than an ancient filter. This objection is unresolved and is stated here rather than in a footnote.
The corpus is an open reconstruction, not a transcription. It was assembled from the standard published scholarship and is not claimed to be exhaustive; the comparison strata are declared samples. Every entry carries its attestations precisely so that inclusion, exclusion, gloss and field can be checked and corrected. Verse numbers follow the Masoretic text, where Daniel 6 runs one ahead of English editions.
One tagging call carries the whole Religion count. Rāz — the word Daniel uses for the mystery God discloses, and later a technical term at Qumran — is tagged Cognition, not Religion, by the rule that lexemes are assigned on core denotation rather than theological use. Moving it changes the count from zero to one. The rule was fixed before tagging; a critic who re-runs the file with rāz moved is running the analysis we invite, not cheating.
A dataset from the Foundation's instrument suite, testing a prediction pre-registered in The Library That Does Not Burn. See also Did Zoroastrianism influence Judaism?, the Eschatology Map and the Overwrite Timeline. Released under CC BY 4.0 — reuse it, with attribution. Corrections that damage the result are the ones we most want. · Foundation of Asha