The Turing Zeta Rotor Model: Building Turing's 1939 zeta-function machine from his notes, and listening to it with a dial machine

In 1939 Alan Turing set out to build a machine of gears, weights and cords that would find the zeros of the Riemann zeta function, on the idea of Lord Kelvin's tide-predicting machine. The war came and it was never finished. This paper builds it as a toy in software from the published accounts of his plan, turns it, and then improves it as a tinkerer might have, keeping the restoration and the overhaul apart. What it found:1. Turing's own gearbox works on his range. Read from the blueprint's gear table, as transcribed by the Museum of Mathematics and Wolfram, and turned over the heights Turing proposed, his gears find 5,596 of the 5,598 zeros below height 6,000, losing only one pair whose zeros sit a ninth of the average spacing apart; re-set at each doorway, they find every one and move the zeros between three and ten times less than our best overhaul, the string machine. Four of his gears, wheel for wheel, are ones our machines had already chosen by rule before we read his table.2. The machine works. Its turning wheels and one small correction find every one of the first 1,600 zeros, up to height 2,090, each within 0.0001 of its true place, and the right number of them, 5,598, up to height 6,000.3. Without its correction the machine loses zeros only in pairs, and a rule worked out on the first stretch named every pair lost on the second before the machine ran there. Each of the thirteen lost pairs did not vanish but stepped off the critical line as a mirror pair; putting the correction back brings it home.4. The zeros the wheels lose are the ones through which the dial bank of The Two-Needle Compass hears the primes most loudly. Crowding does not explain it; the low hill between two close zeros does, because that is where the small primes line up.5. A tinkerer's overhaul rescues a coarser gearbox. Our first guess at Turing's gears, one pair of wheels to each wheel with at most 120 teeth, loses 1,078 zeros; re-set at each doorway, where the machine already stops, it loses 28. A string machine, one carriage per prime that may stop only at the gears continued fractions must find (read through Ford circles), loses none, and still loses none when cut as trains of small brass wheels with at most 60 teeth. Every prediction was written down before its run: of 60 scored, 49 landed and 11 missed, each miss followed by a redesign check or explained by a result already scored, and all 12 redesign statements landed; two more fell between the odds and stay undecided. The record holds the paper; a toy page that opens in any web browser, where you can turn the machine yourself, take away its correction, or swap in Turing's own gears; and a zip of every script, prediction file (as written before its run), output and data file, with a full rerun in a fresh copy that reproduced all 51 output files exactly. A preprint, not peer reviewed, with a limited literature search; built with Claude (Anthropic) as research partner for algorithm and program development. It claims nothing about the Riemann hypothesis. Matthew "Bear" Schulz, Belmont, CA · ORCID 0009-0000-0956-5743 · [email protected] · CC BY 4.0.

Authors

Publication Details

Journal
Zenodo (CERN European Organization for Nuclear Research)
Published
2026-11-14
DOI
https://doi.org/10.5281/zenodo.24233198
Primary Topic
Analytic Number Theory Research
Type
preprint
Controls
|||
ALL TIME
JAN
FEB
MAR
APR
MAY
JUN
JUL
AUG
SEP
OCT
NOV
preprint

The Turing Zeta Rotor Model: Building Turing's 1939 zeta-function machine from his notes, and listening to it with a dial machine

Matthew Schulz
Zenodo (CERN European Organization for Nuclear Research)
Analytic Number Theory Research
preprint

The Turing Zeta Rotor Model: Building Turing's 1939 zeta-function machine from his notes, and listening to it with a dial machine

Matthew Schulz
preprint en

Abstract

In 1939 Alan Turing set out to build a machine of gears, weights and cords that would find the zeros of the Riemann zeta function, on the idea of Lord Kelvin's tide-predicting machine. The war came and it was never finished. This paper builds it as a toy in software from the published accounts of his plan, turns it, and then improves it as a tinkerer might have, keeping the restoration and the overhaul apart. What it found:1. Turing's own gearbox works on his range. Read from the blueprint's gear table, as transcribed by the Museum of Mathematics and Wolfram, and turned over the heights Turing proposed, his gears find 5,596 of the 5,598 zeros below height 6,000, losing only one pair whose zeros sit a ninth of the average spacing apart; re-set at each doorway, they find every one and move the zeros between three and ten times less than our best overhaul, the string machine. Four of his gears, wheel for wheel, are ones our machines had already chosen by rule before we read his table.2. The machine works. Its turning wheels and one small correction find every one of the first 1,600 zeros, up to height 2,090, each within 0.0001 of its true place, and the right number of them, 5,598, up to height 6,000.3. Without its correction the machine loses zeros only in pairs, and a rule worked out on the first stretch named every pair lost on the second before the machine ran there. Each of the thirteen lost pairs did not vanish but stepped off the critical line as a mirror pair; putting the correction back brings it home.4. The zeros the wheels lose are the ones through which the dial bank of The Two-Needle Compass hears the primes most loudly. Crowding does not explain it; the low hill between two close zeros does, because that is where the small primes line up.5. A tinkerer's overhaul rescues a coarser gearbox. Our first guess at Turing's gears, one pair of wheels to each wheel with at most 120 teeth, loses 1,078 zeros; re-set at each doorway, where the machine already stops, it loses 28. A string machine, one carriage per prime that may stop only at the gears continued fractions must find (read through Ford circles), loses none, and still loses none when cut as trains of small brass wheels with at most 60 teeth. Every prediction was written down before its run: of 60 scored, 49 landed and 11 missed, each miss followed by a redesign check or explained by a result already scored, and all 12 redesign statements landed; two more fell between the odds and stay undecided. The record holds the paper; a toy page that opens in any web browser, where you can turn the machine yourself, take away its correction, or swap in Turing's own gears; and a zip of every script, prediction file (as written before its run), output and data file, with a full rerun in a fresh copy that reproduced all 51 output files exactly. A preprint, not peer reviewed, with a limited literature search; built with Claude (Anthropic) as research partner for algorithm and program development. It claims nothing about the Riemann hypothesis. Matthew "Bear" Schulz, Belmont, CA · ORCID 0009-0000-0956-5743 · [email protected] · CC BY 4.0.

Zenodo (CERN European Organization for Nuclear Research)
Analytic Number Theory Research
AI Navigator

Ask Laika to Summarize, Analyze, and Connect papers live on the map.

Summarize Papers & Methodologies

Extract key findings, datasets, and comparative methods across publications.

Benchmark Rankings & Visual Analytics

Rank top research institutions, authors, funders, topics, and journals by Field-Weighted Citation Impact (FWCI) and paper volume with instant charts.

Connect Distant Disciplines

Bridge topological clusters on the map to find hidden collaborative intersections.

Laika access coming soon