The Torah is the first five books of the Hebrew Bible — Genesis, Exodus, Leviticus, Numbers, Deuteronomy — traditionally attributed to Moses. In Jewish tradition it is not five separate books but one continuous scroll: the תורה (Torah, “instruction”). It contains 304,805 Hebrew consonants written in an unbroken stream — no vowels, no punctuation, no chapter breaks.
What most readers don’t know is that the Torah has a hidden structure. The word תורה (Torah) is spelled forward in the letter spacing of Genesis and Exodus. In Numbers and Deuteronomy, it is spelled backward: הרות. Both halves point inward toward the center book — Leviticus — where the divine name יהוה (YHWH) appears in the ELS structure. The Torah literally points to God.
This pattern is invisible in any translation. It exists only in the Hebrew letters. And it raises a question: if the large-scale structure encodes information in the letter spacing, what else is hidden inside individual verses?
Place your finger on a letter in the Torah. Count forward 7 letters. Then another 7. Then another. If those letters spell a real Hebrew word, you have found an Equidistant Letter Sequence — an ELS code. A word hidden in the spacing of the text, invisible to any reader, visible only to someone counting.
This is not mysticism. It is arithmetic. Any long text contains ELS patterns. English novels have them. Legal documents have them. The question is whether the ones in the Torah are meaningful, or just noise.
That question has been debated since 1994. The problem has always been selection bias: every tool ever built — CodeFinder, Keys to the Bible, Advanced Bible Decoder — asks you to type in the word you want to find. You search for what you already suspect. You find it. The critic shrugs.
The critic is right. That is a real problem.
Darash, built by Publifye AS in Norway, does something no previous tool does. Its blind scan takes a Torah verse — nothing else. No words to search for. No hypothesis. It checks all 19,321 Hebrew words in the lexicon at every skip from 2 to 49, and reports what is actually there, ranked by rarity.
Here is what happened when we pointed it at Deuteronomy 21:23:
“His body shall not remain all night on the tree, but you shall bury him the same day, for a hanged man is cursed by God.”
This is a law about criminal execution. The author was writing legal code for ancient Israel. We gave Darash zero vocabulary input — just the verse reference. The tool scanned the letter spacing and returned these words, among others, at zero baseline rarity:
Hang. Curse. Iniquity. Jeer. Deceit. A wreath. Collapse. Rise. Distressed.
No human told the algorithm to look for crucifixion vocabulary. It scanned every Hebrew word that exists and returned these — from a verse about hanging a criminal on a tree. Paul quotes this exact verse in Galatians 3:13: “Christ redeemed us from the curse of the law by becoming a curse for us.”
The tool did not know that. It simply measured what is there.
You can run this yourself. The results are reproducible.
Darash offers two ELS tools. The blind scan (els_discover) asks: “what is hidden here?” The verification scan (els_verse_signal) asks: “are these specific words statistically present?”
The blind scan is what you just saw — zero input, the algorithm searches the entire dictionary. This is the tool used in the 40 ELS evidence findings.
The verification scan lets you test a hypothesis. You provide a verse and a set of Hebrew words. The tool measures how tightly those words cluster around the verse on a cylindrical letter grid, then runs the same search on 1,000 random Torah verses as controls. Words that cluster tighter than 99% of random locations are flagged as statistically significant.
The two tools complement each other. Discover first, then verify. Below, we used the verification scan to test whether specific thematic vocabulary — chosen by us, not by the algorithm — clusters around four Torah verses that the New Testament claims are prophetically significant.
ELS findings are not mathematical proof. A rigorous critic can ask whether any sufficiently long Hebrew text would produce similar patterns. That is a fair question, and any honest presentation of this data must acknowledge it. The Torah is 304,805 consonants long. At skip intervals of 2 to 49, the number of possible letter sequences is vast. Some coincidences are inevitable.
What Darash demonstrates is a reproducible pattern: theologically significant verses produce coherent, thematically relevant vocabulary from the blind scan, while insignificant verses produce noise. Forty verses were scanned. Ten controls were scanned. Every result is reproducible by anyone with the tool. The signal-to-noise distinction is consistent, not occasional.
Whether that pattern is signal or coincidence is a question Darash surfaces. It does not answer it for you.
For each verse below, we chose the Hebrew vocabulary ourselves — words drawn from the New Testament event the verse is said to foreshadow. The tool measured whether they cluster more tightly than chance allows. No parameters were adjusted between runs.
“He shall bruise your head, and you shall bruise his heel.”
God speaking to the serpent after the fall. Christianity reads this as the first promise of a redeemer — one who defeats evil but is wounded in the process. We tested eight crucifixion and resurrection words. Six clustered around the verse:
The word for serpent — נחש — has a gematria value of 358 (gematria sums the numerical value of each Hebrew letter). The word for Messiah — משיח — also equals 358. The serpent and the anointed one share the same number.
The verb “he shall bruise” — שוף (shuf) — equals 386. The same value as ישוע — Yeshua.
The gematria values above are a traditional Jewish observation about Hebrew letter arithmetic — not a result produced by Darash’s ELS engine. Included here as additional context.
“Aaron shall lay both his hands on the head of the live goat, and confess over it all the iniquities of the people of Israel, putting them on the head of the goat.”
The Day of Atonement ritual. One goat is slaughtered. The other carries all sins into the wilderness. We tested eight words. Seven out of eight were found:
Carry. Iniquity. Yeshua. Blood. Forgive. Exchange. Atone. The word carry is woven into the grid of a verse about a goat that carries sin away.
“If the ox gores a slave, the owner shall give to their master thirty shekels of silver.”
A law about livestock damage. Matthew 27:9 connects it to Judas — who betrayed Jesus for thirty silver pieces, threw the money into the temple, and the priests bought a potter’s field. We tested eight words from the Judas narrative. Seven were found — the strongest statistical clustering of all four examples (p = 0.079):
Kiss. Innocent. Bribe. Prophet. Potter. Blood. Betray. Every element of the Judas story — in the letter spacing of a law about what happens when an ox hurts a slave. Written over a thousand years before the event.
“The scepter shall not depart from Judah, nor the ruler’s staff from between his feet, until he comes to whom it belongs.”
Jacob’s deathbed blessing over Judah. Both Jewish and Christian tradition read it as messianic. We tested eight words. All eight were found:
Eight for eight. King. Messiah. Priest. David. Throne. Redeem. Forever. Yeshua. The name David — who would descend from Judah a thousand years later — appears on the verse grid, five letters from center.
The blind scan found crucifixion vocabulary in a criminal execution law — with zero input. The verification scans found thematic clusters in four different verses, from four different books, each tighter than 1,000 random controls.
The tool does not know which verses are theologically significant. It was not told. It measured. The verses that carry theological weight produced signal. The ones that don’t, produce noise.
Darash does not tell you what to believe. It tells you what is in the text.
See all 40 blind discovery findings with AI analysis →
View ELS Evidence