The Draw-Boy's Machine
In 1755 a three-year-old began work in a Lyon silk mill. He put in ten hours a day as a draw-boy: the child who sits at the side of a figured loom and lifts the warp cords by hand, one combination at a time, on the master weaver's call, so that a pattern can be woven into silk. He had no schooling. He learned to read at thirteen, from his brother-in-law. His name was Joseph-Marie Jacquard, and forty-six years later he built the machine that abolished his own job.
That machine gave computing the punched card. It is the ancestor of Babbage's Analytical Engine, of Hollerith's census tabulator, of the IBM card decks that ran payroll into the 1970s. The instruction set of the modern world was designed to weave silk. This paper traces the line, marks precisely where the popular version of the story overstates itself, and then follows it forward to July 2026, when a Dutch couturier put a dress through a particle accelerator and named the collection after a phenomenon on the crust of a neutron star.
§1 · The job the machine replaced
Figured weaving — brocade, damask, a woven portrait — requires that each pass of the shuttle lift a different subset of warp threads. A plain weave needs no such thing, and neither does a check: plaid is the pattern that never needed a child pulling cords, because a tartan is a simple repeat in warp and weft colour, executable on a loom with four shafts and a foot treadle. The draw-boy existed for the patterns a repeat cannot produce. His labour was the program — a human register holding one row of the design, re-entered by hand for every row, for every copy of every bolt.
Understanding this makes the invention legible. Jacquard did not automate weaving. Weaving was already mechanical. He automated the boy — the memory, not the motion.
§2 · The Lyon lineage
Jacquard was neither first nor alone. Every step happened in the same city, inside the same industry, across three generations of the silk trade.
| Year | Who | What changed |
|---|---|---|
| 1725 | Basile Bouchon, Lyon | A needle-based selector reading a roll of perforated paper. The first time a pattern left the weaver's head and became a physical medium. |
| 1728 | Jean-Baptiste Falcon | Replaces the fragile tape with linked punched cards laced into an endless chain. Random-length programs, and a damaged row can be replaced without discarding the whole. |
| 1745 | Jacques de Vaucanson | Automates the selection mechanism itself with a metal cylinder, building on Falcon. Removes the operator from the reading step. |
| 1801 | Joseph-Marie Jacquard, Lyon | Combines the chain of cards with a practical head that a single weaver can run. One card equals one row of the design; the deck is the pattern, and it is detachable, storable, copyable, mailable. |
| 1806 | The French state | Declares the loom public property. Jacquard receives a pension and a royalty per machine. |
§3 · What Lyon did not invent
The popular telling — that binary code was born in the silk industry — claims too much, and the claim is easy to dismantle, so this paper does not make it.
| Component | Origin |
|---|---|
| Binary arithmetic — number as a base-2 quantity | Leibniz, Explication de l'Arithmétique Binaire, 1703. Almost a century before Jacquard. |
| Binary logic — truth values under algebraic operations | Boole, 1854. Half a century after Jacquard. |
| Binary instruction encoding — hole or no hole, as the unit a machine reads | Lyon. Bouchon 1725 through Jacquard 1801. |
| The stored, detachable program | Lyon. Falcon's chained cards, 1728. |
The correct statement is narrower and loses nothing: the silk trade did not give us the mathematics of two states, it gave us the practice of encoding an instruction as a physical two-state mark and handing a stack of them to a machine. Leibniz had the number. Boole had the logic. Lyon had the card. Verified — documented history
§4 · The transmission
Charles Babbage travelled to France in 1840 and saw a Jacquard loom working. He owned a woven silk portrait of Jacquard — an image produced entirely by punched cards, a photograph made of holes. The Analytical Engine took the card from the loom directly: Babbage's design distinguishes a store from a mill and feeds it operation cards and variable cards. Ada Lovelace's note on that engine — the subject of Vol VII, Ch A — is the first description of a program as an object of study rather than a procedure to be run. The chain continues through Hollerith's tabulator for the 1890 United States census, into IBM, and out the other side into the punched decks of mid-century computing.
§5 · The return — Sonic Starquakes, Paris, 6 July 2026
Iris van Herpen's autumn couture collection was shown in Paris on 6 July 2026 under the title Sonic Starquakes. Its stated sources are sonic vibrating stars, the branching structure of exploding supernovae, the spiral geometry of galaxies, and the turbulence of plasma.
The garments carry plasma sealed in waning-crescent glass tubes that respond to the wearer's own magnetic field, sweeping from the shoulder of a tulle dress surfaced in hand-blown glass bubbles; some thirty thousand glass spheres across the collection. The centrepiece look is called Fractal Universe. In the days before the show it was charged inside a particle accelerator and cryogenically preserved, so that it walked as a metastable reservoir of trapped electrons holding an electric field within its own structure. Verified — press record
§6 · The starquake is not a metaphor
A magnetar is a neutron star with a magnetic field of order 1014–1015 gauss. The field stresses a crust of crystalline nuclear matter until it fractures — a starquake — and the star rings. In the decaying tail of the 2004 giant flare from SGR 1806−20, X-ray telescopes recorded quasi-periodic oscillations at approximately:
The sub-100 Hz modes are identified as torsional crustal shear modes — the crust twisting against itself — and magnetar asteroseismology now uses them to constrain mass, radius, and the magnetic coupling between crust and fluid core. So the title is literal in a way a fashion title has no obligation to be: a neutron star fractured and sounded a note in the range of a bass guitar. Verified — literature Mode identification contested
§7 · The operator reading, and one thing this paper refuses to do
Both ends of this history are the same chain. K is a gate that holds: the draw-boy's hand on the cords, admitting one combination per pass; the magnetar crust holding magnetic stress. F is the fold at which the held thing gives way: the card replacing the boy, the crust fracturing. U is what the system settles into afterwards and what it emits — a deck that can be copied and posted, a spectrum of torsional modes. The pattern survives the medium in both cases, which is the whole point of a stored program and the whole basis of asteroseismology. Model — a reading, not a derivation
Historical claims in §§1–4 are drawn from the standard record (Computer History Museum, Britannica, Science Museum Group) and were checked individually; the card count usually quoted for the woven Jacquard portrait could not be verified and is therefore omitted rather than repeated. Jacquard was forty-nine at the 1801 demonstration, not a boy — he had been a draw-boy, which is the substance of §1 and is stronger than the version in which a child invents the loom. §5 follows the July 2026 press record for a collection this author has not seen in person. §6 reports published observational values; the assignment of specific frequencies to specific crustal modes remains an open question in the astrophysics literature. §7 is a reading of the material in this series' vocabulary and derives nothing. Nothing in §§1–6 depends on it.