1994: the claim at its strongest
Doron Witztum, Eliyahu Rips and Yoav Rosenberg published “Equidistant Letter Sequences in the Book of Genesis” in Statistical Science9(3):429–438. They took a list of thirty-two famous rabbis, the second of two lists a bibliographer drew up from a standard encyclopedia, paired each man's names with his Hebrew dates of birth and death, and measured how closely each name's letter-sequence sat to its matching date in the Koren text of Genesis, against 999,999 random re-pairings. The true pairing came fourth out of the million; the paper reports its significance as 1.6 × 10⁻⁵, about one in 62,500, that rank multiplied by four for the four statistics computed. Rips is a respected Hebrew University group theorist. This is the serious version of the case and the one worth engaging. The journal's editor, Robert Kass, published it explicitly as a challenging puzzle offered to readers rather than as an endorsement.
1997: the popularization, and its repudiation
The journalist Michael Drosnin turned the subject into a bestseller with The Bible Code, claiming the Torah foretold the Rabin assassination. Witztum, Rips and Rosenberg publicly denounced the book; Rips stated he did not support its conclusions and that its predictions had no scientific basis. Drosnin's method, any cluster of related words, at any skip, in either direction, was never defended statistically even by the serious wing of the codes camp.
1997: the control
Drosnin issued his own challenge in Newsweek: when his critics found a message about the assassination of a prime minister encrypted in Moby Dick, he would believe them. Brendan McKay promptly published letter-sequence arrays in Moby Dick, using Drosnin's own method, “predicting” the assassinations of Indira Gandhi, Trotsky, Martin Luther King Jr., Rabin, with the assassin's name adjacent, Lincoln, Kennedy, Diana, and Drosnin's own death. That is a control experiment: if this kind of cluster appears in any long text, its presence in Genesis carries no weight by itself. McKay's grids are linked, never re-hosted.
1999: the reply, in the same journal
McKay, Bar-Natan, Bar-Hillel and Kalai published “Solving the Bible Code Puzzle” in the same journal. Three findings. Which of a rabbi's many names, titles and spellings to use was fixed by no rule, the space of legitimate-looking choices was enormous, and the choices actually made were consistently favorable to a positive result. With comparably mild adjustment they produced a “successful” rabbis experiment in the Hebrew translation of Tolstoy's War and Peace. And every variation that removed the freedom made the effect vanish: “all of our own experiments failed to detect the phenomenon.”
2004: the closest thing to an adjudication
A Hebrew University committee under Robert J. Aumann, who won the Nobel in economics the following year, and who began sympathetic, having sponsored the codes work for presentation, was formed in 1996 to test the Gans cities experiment. Its members were drawn from both camps: Aumann, Hillel Furstenberg, Dror Bar-Natan, Isaac Lapides and Eliyahu Rips himself. It ran two fresh experiments in the spirit of the Gans tests, with the decisions fixed in advance, and reported in June 2004 (Discussion Paper 364, Center for the Study of Rationality) that they failed to confirm the codes: the two runs came out at 0.463 and 0.617 against a threshold of 0.001. Barry Simon's independent collection of the same city data had likewise returned nothing. That is the weightiest single fact on this page, and it lands on the Gans exhibit above. Aumann's own summary is the honest landing and this page will not improve on it: the research failed to confirm the codes, it did not establish their non-existence, and he personally returned to his prior that the phenomenon is improbable.
The objection to Panin: from inside the house
Oswald T. Allis was a conservative, inerrantist Old Testament scholar and a co-founder of Westminster Theological Seminary, so this is not a skeptic's complaint. In Bible Numerics (1944) he argued that Panin's features are chosen after the fact from an unbounded menu: count letters, or words, or vocabulary items, or grammatical forms, or nouns, or place-values, with the failures silently discarded. He documented specific errors: the celebrated feature that exactly 7 of the 21 Old Testament writers are named in the New Testament omits Solomon and Jonah, who are on the list and are named. He documented inconsistent spelling choices, and a deeper circularity: Panin decided between competing Greek manuscript readings by which one produced sevens, then cited the resulting text as proof.
The general form of the objection stands on its own and is worth stating plainly, because it applies to this whole family of claims: any verse offers dozens of separately countable quantities, and each has roughly a one-in-seven chance of being divisible by seven. Twenty hits amid an unrecorded pile of misses is what chance produces, not what design proves. The Genesis 1:1 exhibit at the head of this page is deliberately built to be immune to that objection, it counts the whole verse, every word, with no menu to choose from, and the reader is entitled to apply the same test to everything else here.
Cited to: McKay, Bar-Natan, Bar-Hillel & Kalai, “Solving the Bible Code Puzzle,” Statistical Science14(2):150–173 (1999); Witztum, Rips & Rosenberg, Statistical Science9(3):429–438 (1994); Brendan McKay, “Assassinations Foretold in Moby Dick!” (Australian National University, 1997); the Aumann committee report, Discussion Paper 364 (Hebrew University, 2004); Oswald T. Allis, Bible Numerics (1944; the scan linked here is the 1952 printing); Randall Ingermanson, Who Wrote the Bible Code? (WaterBrook, 1999); David E. Thomas, Skeptical Inquirer 21(6) (1997).