My Codes
Search the Torah for a name or word spelled out at equal letter skips (an ELS). Every result is shown next to what shuffled text gives by chance, because long texts contain many such patterns by chance.
Loading the Torah text…
How often does appear?
–
–
…
Gematria of
Gematria values are arithmetic on letters, shown for study. A matching value is not a sign of anything.
Where it appears
Sorted by the shortest skip first. Positive skip reads forwards, negative backwards. Tap a row to open its letter grid.
English under each verse: JPS 1917 (public domain), via Sefaria. The real published translation, never machine-translated; if a verse's English is unavailable, none is shown.
Optional: your Hebrew birth date
Converted with the Hebcal API (CC BY 4.0). The date is sent to Hebcal only to convert it; nothing is stored.
Week of Commentary & reflection · not reporting · not prediction
Parsha Lens
This week’s news, read beside the weekly Torah portion
Parshat Week of
Parsha & the World
Also This Week
More stories this week
Where two themes meet
← Back to This Week · Archive
Earlier test, 25 Oct 3:02 PM:
This is study and exploration, not prophecy. The terms were fixed before any search was run, and every term is listed below, including the ones that came out at chance (misses).
Pre-registered terms
What the columns mean
- Found: ELS (skip ±2 to ±1000) with at least one letter inside the parsha.
- Expected: the formula estimate if letters were random with the Torah's letter frequencies.
- Shuffled parsha: the parsha's own letters shuffled times (the rest of the Torah unchanged). This is the main baseline. p = share of shuffles that matched or beat the real count (smallest possible = ).
- Shuffled words: the same with the parsha's word order shuffled (real words kept, order destroyed).
- Window rank: how this parsha ranks among all same-length stretches of the Torah (1 = most ELS).
- Corrected p: Holm correction for testing 5 terms at once. Below 0.05 = "above chance", which here still means only "a curiosity to re-test", never evidence.
Public log
Method & Sources
What an ELS is
Remove spaces from the Torah to get one string of 304,805 letters. An equidistant letter sequence spells a word by taking every d-th letter, forwards or backwards. Final letter forms (ך ם ן ף ץ) are treated as their regular forms for searching and kept for gematria.
Why a baseline is always shown
A long text contains many short words at some skip purely by chance: a 3-letter word typically appears thousands of times. So every count here sits beside (1) a formula estimate from letter frequencies and (2) the same search on shuffled copies of the text. A count that matches the shuffles is what chance gives. A count above every shuffle is a curiosity worth re-testing with fresh, pre-registered terms, not a message.
This Week: the news front page
The home page is a weekly news page for the Torah portion. We choose 5 to 8 stories from the week’s frozen theme list that have a real textual or thematic link to the portion’s verses or to classic commentary, and we list the stories we left out and why. Each article is commentary and reflection, not reporting. News facts come only from the linked reports and are cited with outlet and date. Verses are quoted in Hebrew and in the JPS 1917 English, with links to Sefaria. Rashi is quoted word for word from the public-domain Rosenbaum–Silbermann translation. We never say the Torah predicts events. Inside each article, a collapsible panel, “The codes behind this story”, shows that theme’s precomputed results from the Codes Desk, misses included.
The Codes Desk: how the code results are made
Each week, real headlines are grouped into about ten news themes, each with one to three Hebrew spellings. The list is frozen with a timestamp and a SHA-256 hash before any search. Then, for every spelling, we count the ELS (skips ±2 to ±1000) that touch the weekly portion and compare that count with three baselines: (1) the portion's letters shuffled 2,000 times, (2) the portion's word order shuffled 2,000 times, and (3) every other stretch of the Torah of the same length. The main figure is the share of letter shuffles that did as well or better, corrected for testing all spellings at once (Holm). Verdicts: Unusual — worth re-testing (corrected p < 0.05), Within chance, or Not found.
Where two themes meet is a simplified version of the "close together" idea in Witztum, Rips & Rosenberg (1994) and McKay et al. (1999), not their exact statistic. For two themes, we lay the portion out in rows as wide as one of the two skips, wrapping like a cylinder, and measure the smallest box (rows × columns) holding both words. The same is done on 2,000 shuffled copies of the portion and on the other same-length stretches, with Holm correction across all pairs. All results are computed in advance on a server and saved as a static file, so the page loads instantly.
What this site does not claim
We never say the Torah predicts events. Researchers disagree sharply about Torah codes. The original paper (Witztum, Rips & Rosenberg 1994) reported a significant effect for rabbis' names and dates. McKay, Bar-Natan, Bar-Hillel & Kalai (1999) argued that the result came from choices in the data (spellings, appellations, dates) and reproduced a similar effect in a Hebrew translation of War and Peace. Witztum and Rips published replies defending their work. This site shows both sides and lets the baselines speak.
Text
The standard Sefer Torah text (Ashkenazi/Sephardi scrolls), 304,805 letters: Genesis 78,064 · Exodus 63,529 · Leviticus 44,790 · Numbers 63,530 · Deuteronomy 54,892. Written text (ketiv) only; vowels, cantillation and paragraph marks removed. Deuteronomy 23:2 reads דכה.
Sources and credits
- Torah text: Miqra according to the Masorah, Hebrew Wikisource, licensed CC BY-SA 4.0, accessed via the Sefaria API, with the edition's own notes for the Ashkenazi/Sephardi scroll readings applied. Our letter-only derivative file is shared under the same licence.
- Hebrew dates and the weekly Torah portion: Hebcal (converter and leyning APIs), CC BY 4.0.
- English verses: The Holy Scriptures: A New Translation (JPS 1917), public domain, via the Sefaria API (Sefaria lists its licence as Public Domain).
- Rashi in English: Pentateuch with Rashi’s commentary, M. Rosenbaum and A.M. Silbermann, 1929–1934 (public domain), via Sefaria.
- News articles (This Week): Times of Israel, Ynet News and BBC News reports, each linked and dated inside the article that uses it.
- News themes (Codes Desk): headlines from the Times of Israel, Ynet News and BBC News World RSS feeds, linked to the original articles. We group them into themes; we never write or change a headline.
- Witztum, D., Rips, E. & Rosenberg, Y. (1994). Equidistant Letter Sequences in the Book of Genesis. Statistical Science 9(3), 429–438. doi:10.1214/ss/1177010393
- McKay, B., Bar-Natan, D., Bar-Hillel, M. & Kalai, G. (1999). Solving the Bible Code Puzzle. Statistical Science 14(2), 150–173. doi:10.1214/ss/1009212243
- Witztum & Rips: replies to McKay et al. (published by the authors; see the discussion around the 1999 paper).