What gematria is, and what it is not

Seven reckonings, a Talmudic pedigree, and an honest account of where the ground stops being firm.

Hebrew has no separate numerals. Alef is one and also the letter alef; tav is four hundred and also the letter tav. Any Hebrew word is therefore already a number, without anyone having to encode it, and gematria is the practice of noticing what that number is and of setting words with the same number beside one another.

That is the whole of the mechanism. Everything difficult about the subject is a question of what the mechanism is allowed to be used for.

Seven reckonings

There is no single gematria. There is a family of reckonings, and a value means nothing until you say which one produced it. This site computes seven, over the consonantal written text with the vowels and cantillation stripped, and here is each of them worked on the same word, שלום.

• Standard, mispar hechrachi. Alef is 1 up to yod at 10, then by tens to kuf at 100, then by hundreds to tav at 400. Final forms count as their base letter. שלום=376
• Gadol, mispar gadol. The same, except that the five final forms continue the series: final kaf 500, mem 600, nun 700, peh 800, tzadi 900. שלום=936 Gadol
• Katan, mispar katan. Each letter reduced to a single digit by dropping its zeros, so shin is 3 rather than 300. שלום=16 Katan
• Siduri, mispar siduri. Position in the alphabet: alef 1 through tav 22. שלום=52 Siduri (ordinal)
• Atbash. Each letter swapped for its counterpart from the far end of the alphabet, alef for tav and bet for shin, and the standard value of the result taken. The swap is old enough to appear in Yirmiyahu, where Bavel is written Sheshach. שלום=112 Atbash
• Perati, mispar perati. Each letter’s standard value squared, then summed. שלום=92,536 Perati
• Milui, mispar shemi. Each letter is worth the standard value of its own name written out in full, so alef is not 1 but אלף, which is 111. It is the one reckoning that depends on a spelling: several letter names are written more than one way, and the tradition keeps all of them — the four fillings of the Divine Name, ע״ב 72, ס״ג 63, מ״ה 45 and ב״ן 52, are the same four letters under four spellings of heh and vav. This site’s default is the milui alfin, the filling of שם מ״ה, so the Name itself comes to 45; the calculator lets you take any of the others. שלום=527 Milui

One word, seven answers, all of them correct. This is the first honest thing to say about the subject: a claim that two words “are the same number” is incomplete until it names the reckoning, and a writer free to choose among seven after seeing the words has not discovered anything.

Where it comes from

The word gematria is Greek in origin, and its etymology is disputed. It enters rabbinic literature as the name of an interpretive technique rather than as a mystical discipline: the baraita of thirty-two rules attributed to Rabbi Eliezer ben Yose HaGelili lists it among the ways aggadah is derived, in company with other techniques of the same practical sort.

The Talmud uses it, and reading the places where it does is more instructive than any general description. Four of them, quoted from the text this site holds.

Berachot 8a: nine hundred and three kinds of death were created in the world, learned from a verse in Tehillim whose last word, תוצאות, comes to that figure.

תוצאות=903

Shabbat 10b, in a discussion of why a man should settle in a young town: Rabbi Avin reads the word נא in Lot’s plea as fifty-one, against Sodom’s fifty-two, the town’s one fewer year of settlement standing for its lighter account.

נא=51

Shabbat 145b: Rabbi Yehudah teaches that for fifty-two years no man passed through Judea, and finds the number in the word בהמה in Yirmiyahu’s lament.

בהמה=52

Shabbat 149b: an aggadah about the king of Bavel takes three hundred from ערל in Chabakkuk.

ערל=300

Look at what those four have in common. In every case the Gemara already holds the teaching on other grounds, and the gematria is offered as a hook to hang it on. Not one of them establishes a law. Not one of them is the only support for the statement it accompanies. That restraint is the tradition’s own, and it is the part most often left out.

What the numbers cannot do

This site is in an unusually good position to say why, because it has counted. The index behind the search box holds 2,581,248 spans, every word, every run of adjacent words within a verse, and every whole verse of the 23,213 verses of Tanakh, each valued seven ways. Against a haystack that size, a match is not a coincidence to be marvelled at. It is the expected result.

The value 26 is carried by 6,601 spans. The value 913, which is בראשית, is carried by 942. Pick any number under about a thousand and you will find hundreds of things that add to it, some of them meaningful-looking, because Hebrew is a compact language and there are a great many spans. A gematria that would be striking in a paragraph is unremarkable in a corpus, and anyone building an argument out of one equality should first ask how many others were available.

בראשית=913

And what it is not at all

Gematria is not prediction. Nothing in the classical use of it looks forward; the readings above all explain something already known.

Gematria is not a code. It is not the practice sometimes called Torah codes, in which letters are read at fixed intervals through the text to yield words that are not written in it. That is a different operation altogether: it works on a letter stream rather than on words, it depends on a particular letter-for-letter edition being exactly right, and the best-known statistical claim made for it was published in a scientific journal and then contested at length in the same journal by other statisticians. This site does not do it and has no plans to. What is indexed here is contiguous spans of the written text, in the order the text stands.

Gematria is not a way of making a verse say something the verse does not say. A remez that contradicts the plain sense has gone wrong, in the same ordinary way that any reading can go wrong.

The line

What separates a gematria that belongs to the mesorah from one that does not is never the arithmetic. The arithmetic is the cheapest part; a calculator does it, and this site does two and a half million of them before breakfast. What makes a reading traditional is that a tradition carries it: someone taught it, someone else recorded it, and it has been argued over for a long time by people who could also count.

So the useful question about any gematria is not “does it work out? ” but “who says it, and what did they think it showed?” The first question is arithmetic. The second is learning.

This site answers the first exactly, for the twenty-four books and nothing else, on the consonantal written text, in canonical order from Bereshit 1:1 forward, and it ranks nothing by relevance because it has no opinion about which occurrence matters. The second question is yours.

On the sources, editions and licences behind every text quoted here, see About.