Medieval Mechanical Clocks: Weight-Driven Timekeeping and Early Automation

A witness who says the job is not finished

In 1271, Robertus Anglicus, lecturing on Sacrobosco’s Sphere, says the clockmakers are trying to make a wheel that turns once with the equinoctial day. He also says they cannot quite perfect it. Lynn Thorndike printed that passage in 1941 under a title that sounds like an invention date. Read the passage, not the title. The man is describing an attempt that has not yet kept the day. He does not describe an escapement. He does not name a town that already has one working.

Later chronicles get busier. The Dunstable annals for 1283 say the priory made a horologium and set it above the pulpitum. Exeter and St Paul’s are mentioned soon after, in the same kind of language. Dohrn’s warning still sits on those lines. Between about 1270 and 1330 you often cannot tell a water clock from a wheel clock from a bell that belonged to either. I will not pick Dunstable as the first mechanical clock. The sentence does not show you the works.

What I will say is smaller. By the fourteenth century, Europe has weight-driven public clocks that can be described as machines, and the descriptions are no longer only a wish. The step from “they are trying” to “here is a tower that strikes” is a generation or two of shops, not a spark.

The weight wants to fall. The escapement will not let it

A raised weight is a store of pull. Left alone on a drum and a gear train, it would empty itself in a rush. The escapement is the part that refuses the rush. It stops the train, then releases it one beat, then stops it again. That alternation is the whole trick. Time, in this machine, is a counted series of mechanical permissions.

When the early European works can actually be described, the common regulator is a verge and a foliot. The verge is a shaft with two pallets. They take turns catching a crown wheel. The foliot is a horizontal bar on that shaft, with a weight you can slide toward the middle or out toward the end. Slide the weights out and the beat slows. Slide them in and it quickens. That is how the clockkeeper aimed the rate. It is not a pendulum, and it is not a steady oscillator. Friction, a dirty pivot, a rope that does not pull the same all the way down, and a cold morning will all move it. Someone compared it with the sun, or with a sundial, and set it back. I will not hand you a minutes-per-day figure. The honest statement is that it wandered enough to need a person.

The gears are a ratio, not a mind

A gear passes a turn and sets a ratio. If one wheel has more teeth than the next, the next turns faster, and with less force. Stack those choices and you can make a slow hour out of a quicker beat, or drive a second train that lifts hammers for the bell. The count of teeth is the design. The wheels do not decide anything. They do what the count makes them do, until a tooth breaks or the weight sits down.

That is the family resemblance to later calculating machines, and it is a thin one. A ratio cut into brass can represent a cycle somebody already understood. It does not take in a new number and change its own rule. The Antikythera mechanism, which is a Greek instrument and not a medieval tower clock, gets the same warning on the engineering page. Do not promote a gear train into a computer because the diagram looks busy.

A lot of the early public clocks were built to be heard

Where the records are clear enough to trust, a good many of the first town clocks existed to strike. The bell is the output. People who never looked at a tower still heard the hour, if they were inside the reach of the sound and if they had agreed what the sound meant. A painted dial facing the square is a real later development. It is not a part you can assume was there because the word “clock” is in a document.

A striking train is a second mechanism, locked to the first so that the hammers fall at a chosen point in the cycle. That is a sequence. It is still open-loop. The clock does not listen for whether the town heard it. If the bell cracks, or the rope slips, a person finds out the old way.

Shared time was a slow argument, not a switch

A tower clock is expensive. It takes a patron, a mason, a smith, and a keeper. Churches, monasteries, and town governments are the customers we can see, which is why the survivors and the contracts cluster there. The bell could mark a market, a service, or the start of a work period. It did not, by itself, put every household on one schedule.

The older day was not cut into equal hours year-round. Daylight was divided, and the winter hour was not the summer hour. A machine that beats a steady rate is pushing toward equal hours, day and night. Dohrn-van Rossum’s history is the careful account of how that public time actually spread: unevenly, beside the old ways, and not as a single reform that covered Europe because a gear had been cut. David Landes tells a larger story, in Revolution in Time, about clocks and the making of modern life. Read him. Do not let one book turn a bell into the birth of the factory whistle. The two clocks — the church’s unequal hours and the tower’s steadier beat — overlapped for a long time.

Two astronomical machines we have on paper

Richard of Wallingford, abbot of St Albans from 1327, wrote a Tractatus Horologii Astronomici for a clock that would show more than the hour: the sky as he understood it, tied to the same works. He died in 1336, before that clock was finished. It was built afterward from his directions, and it does not survive. John North recovered the treatise, which had been sitting out of order in a manuscript, and published it. God’s Clockmaker is his book for a reader who is not going to edit the Latin. The title’s “invention of time” is his frame. The finding is a designed machine, known from a text, not a case we can open.

Giovanni Dondi, a physician of Padua, spent years on an astrarium and wrote the Tractatus astrarii so the machine could be understood from the page. It indicated the hour and the motions of the sun, moon, and the planets then counted, using the astronomical tables of his day. The original does not survive. What stands in Milan’s science museum is a reconstruction by the clockmaker Luigi Pippa, finished in 1963, from Dondi’s text. It is a serious modern build. It is not the fourteenth-century object, and a plate that calls the replica a medieval masterpiece is talking about the design, not about the brass in the case.

Neither machine is an early computer. Each one embodies cycles the maker already accepted, cut into ratios. Feed it a different astronomy and you have to cut different wheels. That is the limit. Sophistication is not the same thing as a program you can change.

The day they started it was the day the work began

A public clock is a machine that lives in the weather and in other people’s hands. Pivots wear. Ropes stretch. The foliot gets nudged. Oil gums. The weight has to be raised, by hand or by whatever hoist the keeper has, on a schedule shorter than a modern eight-day spring. A clock that is not wound is a sculpture. A clock that is not set is a confident liar.

The pendulum clock, which Christiaan Huygens set out in the 1650s, is a different regulator. It belongs to that century, not to the first tower clocks. So do the later escapements that made a marine chronometer possible. Do not smuggle their accuracy back into a foliot. The medieval machine earns its place without them: a repeating mechanical step, bought and maintained by an institution, heard by a town.

Where this sits in the history of automation

A modern line uses time constantly. A controller waits, then fires an output, or it compares a measurement with a setting and changes a drive. The medieval clock did the first kind of thing in hardware. It counted its own beats and, on a striking clock, released a signal at a chosen count. It did not do the second. Nothing in the verge looked at the result and corrected the next beat. The correction was a person with a sundial.

I am describing equipment the trade uses now, not a rule that every shop owns a PLC. The kinship is stored energy, a ratio, and a repeated step. The break is the closed loop, the program, and a time base that does not ask to be set every morning. Do not call a foliot a controller because the bell kept ringing.

The long use, without the straight line

Clockwork kept getting better at one problem: make the next beat closer to the last one. Springs, pendulums, temperature compensation, and then electrical time took that problem to places a tower keeper would not recognize. None of those later answers was hiding in the 1271 lecture. Each one showed up when somebody could build it and somebody would pay for it.

The medieval clock earns its place because it was the first widely attested European way to store a pull, meter it with an escapement, and let a bell repeat the result. It does not earn the claim that it created the modern workday, or the computer, or the factory. The weight does not care what we call it. It only falls until it reaches the floor.