About this project

The Gematria is the definitive gematria tool for the Hebrew Bible — accurate, fast, and strictly faithful to the authentic Hebrew text of the 24 books of Tanakh.

The app lives at TheGematria.com.

Created by Aleksandr Bodnar.

The corpus

The corpus is the 24 books of the Tanakh and nothing else — no prayers, no rabbinic texts, no apocrypha. The Hebrew is the Westminster Leningrad Codex (WLC) as published by the Open Scriptures Hebrew Bible project. Books are ordered in the traditional Jewish order: Torah, Nevi’im, Ketuvim, from Bereshit 1:1 through Divrei HaYamim.

The index covers 23,213 verses, 305,507 word tokens and 2,581,248 precomputed spans — every single word, every run of adjacent words inside a verse, and every complete verse, valued under all six methods.

How the values are calculated

Gematria is computed on the consonantal written text (ketiv). All vowel points and cantillation marks are stripped before calculating, as are maqaf, sof pasuq and paseq. Words joined by maqaf are separate tokens (they can still combine as a phrase). Where ketiv and qere differ, the written form is what is indexed; the qere is kept as metadata.

• Standard (מספר הכרחי) — א=1 … ת=400; final letters count as their base letter.
• Gadol (מספר גדול) — as standard, but ך=500 ם=600 ן=700 ף=800 ץ=900.
• Katan (מספר קטן) — standard values reduced to 1–9.
• Siduri (מספר סידורי) — ordinal position, א=1 … ת=22.
• Atbash (אתבש) — swap א↔ת, ב↔ש …, then take the standard value.
• Perati(מספר פרטי) — each letter’s standard value squared, summed.

Every result list is in canonical order — book, chapter, verse, position — from first appearance to last. Filters narrow a list; they never reorder it. Only the aggregate tally view sorts by frequency, and expanding any tally row returns you to canonical order.

Reverent display

Divine names are always calculated on their true letters. The optional reverent display setting changes only what is printed on screen (יהוה shown as יקוק, the Elohim family with a dash); no value and no stored text is affected.

Sources & attribution

Hebrew text: Westminster Leningrad Codex via the Open Scriptures Hebrew Bible (OSHB / morphhb). Public domain.

English translation:“The Holy Scriptures: A New Translation (JPS 1917)”, Jewish Publication Society — public domain, obtained from Sefaria. This project uses authentic Jewish translations only.

Lexicon:Strong’s Hebrew Dictionary JSON edition from Open Scriptures, licensed CC BY-SA. Compiled by Open Scriptures, David Troidl and David Instone-Brewer. Strong’s 1894 text itself is public domain. These glosses are used as English→Hebrew search keys and lexicon labels only.

The 613 mitzvot (Sefer HaChinuch):Hebrew titles from Sefer HaChinuch (13th c.), obtained from Sefaria; English from “Sefer HaChinukh”, translated by Rabbi Francis Nataf (Sefaria, 2018), released CC0.

The 613 mitzvot (Rambam): Minyan HaMitzvot titles from the Wikisource Mishneh Torah edition, obtained via Sefaria, licensed CC BY-SA. English renderings of these titles were prepared for this app and are not a historical translation.

Rashi on the Chumash:“Pentateuch with Rashi’s commentary”, translated and annotated by M. Rosenbaum and A. M. Silbermann, London 1929–1934 — public domain in both its Hebrew and its English, obtained via Sefaria. Rashi is an eleventh-century commentary, not one of the 24 books, so it is not indexed, valued, searched or counted here: the Daily Torah Learning page displays it under the verses it explains, for reading only, and only when the reader turns it on.

Daf Yomi text: Talmud Bavli, Wikisource edition, obtained via Sefaria, licensed CC BY-SA 4.0. The Talmud is outside the 24 books of Tanakh, so it is not indexed, valued or searched here — the Daily Torah Learning page displays it for reading only, with no gematria figures. English is not included in this edition.

Siddur — Ashkenaz: the Metsudah linear siddur (Avrohom Davis, 1981), Hebrew and English, obtained via Sefaria and published there under CC BY. It is the whole of the Ashkenaz nusach here, Hebrew and English both: the two are segmented alike in that edition, so every paragraph’s two languages are the edition’s own pairing and never a guess of ours. Its English also appears in the Chabad nusach, which has none of its own — but only where the Chabad paragraph’s Hebrew is the SAME HEBREW, letter for letter once the vowels, the cantillation, the punctuation and the printed form of the Divine Name are set aside. Where the two texts differ by so much as a word, nothing is carried across automatically: the word that differs is very often the reason the two nusachim are two nusachim, and an English that is right about everything except the difference would be worse than none.

Siddur — Sefard: the same Metsudah linear siddur (Avrohom Davis, 1981), in its Siddur Sefard edition — a separate book with its own weekday order — obtained via Sefaria and published there under CC BY, with the licence and the version checked on every node in both languages before a word of it is kept. It brings its own English, segmented alongside its own Hebrew, so nothing in the Sefard nusach is borrowed, adapted or rendered by us: every paragraph’s two languages are that edition’s own pairing. Its tree is far coarser than the Ashkenaz book’s — a whole Amidah under one heading, and every psalm of Pesukei d’Zimra inside a single node — so the nodes that hold scripture are cut open at their own opening words, and each scriptural run is checked against this app’s index before it is replaced by it. Where the edition prints a note immediately above a passage that IS Tanakh, that note is not shown: a passage read from our own index has no paragraph to carry it, and the passage names itself.

Siddur — where a person judged, and what they judged: every prayer within two words of a Metsudah paragraph is written to a review list and attached to nothing until somebody reads it. Those answers are kept in the open, in data/siddur/borrow_rulings.json, one entry per prayer with its reasoning. Where the difference turned out to be spelling — מלא and חסר, a variant form of one word, or which reverent form of the Divine Name an edition typeset — the Metsudah’s English is shown, because those are the same words. Where the difference was an editorial heading the Metsudah’s edition printed and the Torah Ohr does not (“MISHNAH IV”, a “Chazzan and congregation” direction), the heading is removed and the rest shown. And where the difference was the words themselves — ובגלוי that Nusach Ari does not say, הנפלאות where Ashkenaz says המלחמות, ויתנשא added to V’al Kulam — the Metsudah’s English is NOT shown. Those prayers carry an English rendered for the Nusach Ari text: adapted from the Metsudah where its wording still serves, which CC BY permits and which is marked as an adaptation on the page, and written from the Hebrew where it does not.

Siddur — Chabad (Nusach Ari):“Siddur Torah Ohr”, the arrangement of Rabbi Schneur Zalman of Liadi (1803), in the Hebrew Wikisource transcription of the Schulzinger Bros. 1940 printing, licensed CC BY-SA 4.0 — the licence Hebrew Wikisource states in its own API, checked before any text is kept. No modern copyrighted edition is used, and none of its instructions or formatting. That transcription is Hebrew only; where a prayer’s Hebrew is word for word the Metsudah’s, the Metsudah’s English is shown beneath it under the CC BY credit above. Where no permissively-licensed English exists for a prayer, the page says so rather than leaving a blank; our own renderings are being prepared, and the ones written so far are described above.

The siddur and the corpus.A siddur is not one of the 24 books, so not one word of a blessing, a kaddish or an Amidah is indexed, valued into the index, searched, tallied or counted here — it is displayed. Its gematria is computed in your browser, from the text on the page, and stored nowhere. Where the liturgy IS Tanakh — the Shema, Ashrei, Hodu, the psalms of Pesukei d’Zimra, the Song at the Sea — those verses are not a siddur’s reprint of scripture: they are read from this app’s own WLC index, character for character, with the JPS 1917 translation beneath them and a door back into the book they came from.

Daf Yomi page scans: the Vilna Shas (Romm, Vilna, 1880s) — public domain; a faithful photographic reproduction of a public-domain flat work carries no new copyright of its own. Most volumes are ingested whole from a HebrewBooks.org scan of the printed Vilna set, ingested once per volume and served from this app afterwards; a tractate not yet ingested that way falls back to scans provided by E-Daf.com, obtained via dafyomi.org, fetched once per page and served from this app afterwards. No modern re-typeset edition is used, from either source.

Calendar and reading cycles: the weekly portion, the seven aliyot, the haftarot, the monthly Tehillim division and the Daf Yomi schedule are computed inside this app with hebcal — @hebcal/core (GPL-2.0), @hebcal/leyning and @hebcal/learning (BSD-2-Clause). Nothing is fetched at page load, so the calendar works offline.