When a 13th-century author writes “this equals that”, does it? We machine-recomputed 8,629 gematria claims from 834 works. Three in four hold exactly.
6,494 of 8,629 attestations hold with no adjustment. The remaining 2,135 — 24.7% — need im hakolel: plus or minus one. This is not an error of the tradition; it is its tolerance, and nobody had measured it before.
| Value | Attestations | Works |
|---|---|---|
| 86 | 155 | 57 |
| 13 | 154 | 56 |
| 65 | 129 | 45 |
| 345 | 127 | 45 |
| 72 | 110 | 51 |
| 110 | 106 | 20 |
| 12 | 104 | 69 |
| 52 | 97 | 53 |
| 45 | 92 | 42 |
| 70 | 90 | 44 |
| Value | Attestations | With adjustment | Share |
|---|---|---|---|
| 110 | 106 | 90 | 85 % |
| 67 | 23 | 14 | 61 % |
| 161 | 22 | 11 | 50 % |
| 74 | 22 | 11 | 50 % |
| 31 | 37 | 17 | 46 % |
| 6 | 22 | 10 | 45 % |
| 111 | 39 | 15 | 38 % |
| 346 | 33 | 12 | 36 % |
Especially 110: of 106 attestations, 90 rest on the adjustment alone. The equality the tradition repeats most often is almost always one with a tolerance.
| Century | Values first recorded |
|---|---|
| XII | 5 |
| XIII | 39 |
| XIV | 251 |
| XV | 182 |
| XVI | 212 |
| XVII | 58 |
| XVIII | 119 |
| XIX | 35 |
| XX | 8 |
| XXI | 1 |
For each of the 1,023 values, the earliest attestation found. Two peaks: the 14th century and the 16th. Dates of works are approximate, so this is a picture, not a chronometer.
Mispar hechrachi: alef 1 … tav 400, final letters equal to the ordinary ones, vowel points removed. “Exact” — both sides equal. “With adjustment” — a difference of exactly 1. Every row recomputed by machine; the CSV is open.
Download the database (CSV): value, attestations, exact, with adjustment, works, earliest
The other study: Coincidence is cheap · /method · /research/spec
The SAPIR 861 app
An icon on your home screen. Opens without the address bar and works offline.