A famous debate
The Bible Code and the puzzle it posed to statisticians
In the 1990s a claim about hidden patterns in the letters of the Torah reached one of the world's leading statistics journals, and then the bestseller lists. This page tells the story, from both sides, and lets you search the text yourself.
I. The idea
Letters at equal intervals
Write the Torah as one unbroken line of 304,850 letters, without spaces. Choose a starting letter and a fixed skip, and read every 50th letter, or every 8th. The result is an equidistant letter sequence (ELS). The best-known examples sit at the very start of each book of the Torah:
| Genesis 1:1 | ראשיתבראא |
| Genesis 1:2 | ניתהומורו |
| Genesis 1:4 | ורויראאלה |
| Genesis 1:5 | ראאלהימלא |
Each row shows one letter of the sequence with its neighbors; the rows skip the letters in between.
| Exodus 1:1 | השמותבניי |
| Exodus 1:2 | נלויויהוד |
| Exodus 1:5 | צאיירכיעק |
| Exodus 1:6 | הדורההואו |
Each row shows one letter of the sequence with its neighbors; the rows skip the letters in between.
| Leviticus 1:1 | ויקרא |
| Leviticus 1:1 | אלמשהויד |
| Leviticus 1:1 | בריהוהאל |
| Leviticus 1:1 | יומאהלמו |
| Leviticus 1:1 | עדלאמרדב |
| Leviticus 1:2 | ראלבנייש |
| Numbers 1:1 | אלמשהבמדב |
| Numbers 1:1 | רצמצרימלא |
| Numbers 1:2 | פרשמותכלז |
| Numbers 1:3 | קדואתמלצב |
Each row shows one letter of the sequence with its neighbors; the rows skip the letters in between.
| Deuteronomy 1:5 | אראתהתורה |
| Deuteronomy 1:6 | בתבהרהזהפ |
| Deuteronomy 1:7 | שפלהובנגב |
| Deuteronomy 1:8 | ראהנתתילפ |
Each row shows one letter of the sequence with its neighbors; the rows skip the letters in between.
II. The story
From hand counts to a journal puzzle
Before WWII
Patterns counted by hand
Rabbi Michael Ber Weissmandl (1903–1957) spent years tracing letters at equal intervals through the Torah by hand, including the patterns shown above. His notes were published after his death in Torat Chemed. Decades later, computers made it possible to search the whole text and to test such patterns statistically.
1994
The Great Rabbis experiment
Doron Witztum, Eliyahu Rips and Yoav Rosenberg published “Equidistant Letter Sequences in the Book of Genesis” in Statistical Science, a leading peer-reviewed journal. They took famous rabbinic figures from Margaliot's Encyclopedia of Great Men in Israel and tested whether each name and its Hebrew dates of birth or death appeared as letter sequences closer together in Genesis than they would by chance. On a second list of 32 personalities, chosen to test an earlier result, they reported odds of about 1 in 62,500 that the effect was chance. The journal's editor wrote that the referees could not find an explanation and offered the paper to readers as “a challenging puzzle.”
1997
A bestseller, and a step back
Journalist Michael Drosnin's The Bible Code brought the idea to millions, claiming the codes could foretell events; he said he had warned Prime Minister Yitzchak Rabin in 1994, a year before his assassination. Rips publicly stated that he did not support Drosnin's work or his conclusions. Drosnin also issued a challenge: find a message about the assassination of a prime minister in Moby Dick. Mathematician Brendan McKay took it up and found similar letter sequences about several assassinations in the novel. Researchers on both sides noted that this exchange concerned Drosnin's predictions rather than the 1994 experiment.
1999
“Solving the Bible Code Puzzle”
In the same journal, Brendan McKay, Dror Bar-Natan, Maya Bar-Hillel and Gil Kalai argued that the 1994 result rested on many small decisions: which names and titles to use for each rabbi, how to spell them, and which forms of each date to include. Made with the outcome in view, they argued, such choices can produce a striking result in any long text, and they demonstrated a comparable one in a Hebrew translation of War and Peace.
Since
Replies and continuing research
Witztum and other supporters published detailed responses, arguing that the original spellings and names had been supplied independently and that the critique's own choices were the biased ones. Harold Gans, a former cryptologist at the U.S. National Security Agency, reported a related experiment using the rabbis' cities, which he said confirmed the effect; critics disputed that work as well. The two sides continue to disagree: critics consider the 1999 paper to account for the 1994 result, while supporters maintain that the effect is real, point to further experiments, and continue to publish research.
III. The positions
The questions at the heart of the debate
How should each name be written?
Critics argue
Rabbis are known by first names, surnames, acronyms and honorifics, each with spelling variants, and dates can be written in several forms. Critics argue that this flexibility leaves room for choices that favor the result.
Supporters respond
Supporters respond that the names and spellings were supplied by an independent expert according to set rules before the test, and that the critics' alternative choices introduce their own bias.
Which text should be used?
Critics argue
Torah scrolls and printed editions differ in a small number of letters. Critics ask how patterns that depend on exact letter positions should be read when editions disagree.
Supporters respond
Supporters note that the experiments use a standard edition fixed in advance, and that the question of textual variants applies to any study of the text.
What counts as a meaningful pattern?
Critics argue
Any long text contains an enormous number of letter sequences (the search below shows how many). Critics stress that only carefully pre-specified tests can separate a real effect from the sheer number of possibilities.
Supporters respond
Supporters agree, and point out that the 1994 experiment was not about finding single words: it measured how closely pre-chosen pairs of related words appear, compared against many controls.
Was the experiment fixed in advance?
Critics argue
Critics argue that some decisions in the 1994 study, from the lists to the measures used, were not fully fixed before the data were examined.
Supporters respond
Supporters maintain that the second list was a genuine fresh test run with the earlier procedure unchanged, which is exactly what a sound replication requires.
IV. Try it yourself
Search the Torah for a word
Enter a Hebrew word to find every place it appears as an equidistant letter sequence, and choose any result to see its letter grid. For context, the counts are shown next to the same letters shuffled at random. Single words appear about as often in both, which is why the research described above measured something different: how closely pairs of related words appear.
Searching every starting letter and skip…
V. Read more
Sources, from both sides
- Witztum, Rips & Rosenberg, “Equidistant Letter Sequences in the Book of Genesis,” Statistical Science 9(3), 1994
- McKay, Bar-Natan, Bar-Hillel & Kalai, “Solving the Bible Code Puzzle,” Statistical Science 14(2), 1999
- Brendan McKay, “In Search of Mathematical Miracles” (critiques and Moby Dick examples)
- Torah Codes Primer and Research Survey (research by proponents of the codes)
Letters are counted from the written text (ketiv) of the Leningrad Codex, with final letters folded into their regular forms. Explore other ways letters and numbers meet in the history of gematriya and the statistical atlas.