The Bible Codes — the whole story

Proponents

Ivan Panin, Bible Numerics (1890s–1942); E. W. Bullinger, Number in Scripture (Eyre & Spottiswoode, 1894). Both public domain.

The critique, at full strength

The counts are true. Genesis 1:1 really does have seven Hebrew words and twenty-eight (4 × 7) letters, and its letter-values really do total 2701 = 37 × 73. The arithmetic is not in dispute; the inference is. Oswald T. Allis — a conservative, inerrantist Old Testament scholar and a co-founder of Westminster Theological Seminary, which makes this an inside-the-house refutation rather than a skeptic's — showed in Bible Numerics (1944) that the “features” are selected after the fact from an unbounded menu: count letters, or words, or vocabulary items, or grammatical forms, or nouns, or place-values, or odds and evens, or beginnings and endings. Failures are silently discarded. Allis also documented outright errors: the celebrated feature that “exactly 7 of the 21 Old Testament writers are named in the New Testament” quietly omits Solomon and Jonah, who are on the list of twenty-one and are named in the New Testament — correcting the count invalidates three claimed features. He documented spelling-rigging (Ἰάκωβος valued at 833, using the Old Testament Jacob spelling without the -ος ending, while Peter and Paul were counted with their -ος endings, because that combination is what supports the theory), and text-rigging: Panin adjudicated New Testament textual variants by whether they produced sevens, then cited the resulting text as proof of that text. The combinatorial point stands on its own. Any verse offers dozens of independently countable quantities, each with roughly a one-in-seven chance of being divisible by seven, so twenty or thirty hits amid an unrecorded pile of misses is expected rather than miraculous. The same holds for the 2701 = 37 × 73 geometry that later writers built on this verse: the space of “remarkable” properties — triangular, hexagonal, star, centred-anything, prime mirrors, digit sums, standard versus ordinal letter-values — is effectively unlimited, the same figurate marvels can be produced from secular sentences, and no proponent has ever stated in advance what would count as a failure.

Proponents

Rabbi Michael Dov Weissmandl (d. 1957; published posthumously by his students); popularized by Chuck Missler, Cosmic Codes (Koinonia House, 1999) — cite-only.

The critique, at full strength

This is a real historical observation, and it was never proof. Finding a short word as an equidistant letter sequence is mathematically trivial: תורה is four letters, and four-letter strings appear at some skip in any sufficiently long text. That is why the scientific claim was never “equidistant letter sequences exist” but the far narrower claim that related sequences sit closer together than chance allows. The freedom in the method is visible in the demonstration's own table: the five columns use four different skips (50, 50, 7, 50, 49), two of the four TORAH readings run backward, and each column picks its own anchor letter. Which anchor, which skip, which direction — every one of those is a degree of freedom, and degrees of freedom are precisely what a control text exists to measure. Then there is the edition. The classical presentations use the Koren (1962) Hebrew text. The Westminster Leningrad Codex — the public-domain base of nearly every open digital Tanakh, and the only Hebrew text MannaFest can ship — differs from Koren in roughly a hundred places, about forty-five extra letters in the Torah, almost all of it plene-versus-defective spelling. One inserted or deleted letter shifts every subsequent position and destroys every sequence that spans it, so a long-skip code requires letter-perfect transmission from autograph to printed edition, and the manuscript record does not show that. The Talmud concedes the point: “we are no longer expert in defective and plene spellings” (b. Kiddushin 30a). Defenders answer that Genesis's Koren-versus-Leningrad differences are few and that the results survived them; MBBK's deeper answer is that the real problem was never the edition but the tuning. Both replies belong in view.

Proponents

Chuck Missler, Cosmic Codes (Koinonia House, 1999) — the founder's on-ramp, cite-only; the glosses draw on standard Hebrew lexica.

The critique, at full strength

This is the one item on the page that takes a Hays note, because it is an interpretive reading rather than a statistical claim, and the Hays rubric applies to echoes rather than to permutation tests. Availability: yes, the names are in the text. Volume: low — the sentence depends on choosing one gloss per disputed name. History of interpretation: modern, nineteenth and twentieth century, not patristic and not rabbinic. Satisfaction: undercut by the glosses themselves. The etymologies run from solid (Adam, Enosh, Noah) to genuinely contested: “Methuselah = his death shall bring” is disputed, and the Mahalalel and Jared glosses are selective rather than settled. Change one contested gloss and the sentence stops being a sentence — which is the tell that it is being assembled rather than read out. None of that makes it worthless. It is a devotional observation worth telling, with the etymological caveats kept visible, and it is not a demonstrable authorial encoding.

Proponents

Yacov Rambsel, Yeshua: The Hebrew Factor (Frontier Research, 1996); Grant R. Jeffrey, The Signature of God (1996) — cite-only.

The critique, at full strength

ישוע is a four-letter string, and four-letter strings are dense at some skip throughout any long Hebrew text — contrary and even blasphemous strings surface in the same passages by exactly the same method, which is what a method with no control produces. Independent investigators who ran it reported no signal. Randall Ingermanson, a physicist and a believing Christian writing for a Christian publisher, ran the method himself and concluded in Who Wrote the Bible Code? (WaterBrook, 1999) that it does not hold. David E. Thomas set out the same arithmetic in Skeptical Inquirer 21(6) (1997). This entry matters most on a confessional site, and it is the reason the honesty mandate exists: the temptation to keep an “our-side” code because it flatters the faith is precisely the temptation the mandate is for. The Suffering Servant passage does not need it. Isaiah 52:13–53:12 is preserved in the Great Isaiah Scroll (1QIsaᵃ, c. 125–100 BC) essentially as the Masoretic Text carries it, roughly two centuries before the crucifixion — a plain-text, manuscript-attested fact, and a far stronger one than any skip.

Proponents

E. W. Bullinger, The Companion Bible, Appendix 60 (1909–1922) — public domain; the manuscript marking tradition is older.

The critique, at full strength

The letters are where Bullinger says they are. That part is checkable, and it is true. What is not demonstrable is the selection. Esther offers thousands of windows of four consecutive words, and the four that yield יהוה were picked out of that pool after the fact — with the searcher free to accept either initial letters or final letters, and either reading direction. How often four consecutive Hebrew words happen to begin or end with yod, he, vav, he is exactly the quantity a control test would measure, and no one has measured it. The tidiness of the result, two forward and two backward at four narrative hinges, is what a post-hoc search produces when it has that much freedom, not evidence against the freedom. Note too that this is acrostic-adjacent rather than gematria proper: no letter-values are involved at all. The honest classification is a speculative observation — devotionally suggestive, genuinely charming, and unprovable. It sits on this page immediately before Psalm 119 in the Tier-A band below so that the difference between a pattern the author built and a pattern a reader found is something the eye can feel rather than something the page has to assert.