A Roman clock at Vindolanda

It was not easy for the Romans to keep a close track on time. They were not, as we are, inveterate clock-watchers. They did not work by minutes, let alone seconds, their smallest unit being the hour. Even that was not standardised, but varied according to season and location. By day, a Roman hour was a 12th of the time between sunrise and sunset, and by night a 12th of the time between sunset and sunrise. The length of this hour therefore depended on the time of year and on how far north or south you were.

Other than estimating from the position of the sun in the sky or by the length of shadows, the Romans had two ways of telling the time. One was the sundial – which was not much use in cloudy weather and even less by night. The other was the water-clock or clepsydra – in its simplest form a bowl of water with a small hole at the bottom which emptied in a given time.

The simple water-clock was fine for measuring, say, the period allocated for a speech in the law-courts. But it did not empty at a constant rate: the fuller it was – the greater the head of water – the faster the outflow, and the emptier, the slower. Moreover, short of a hit-and-miss adjustment of the size of the hole with wax, it could not cater for seasonal hours. One Roman remarked that it was easier to find agreement between philosophers than between clocks!

But already in the 3rd century BC, the Greeks had begun to develop clocks which took account of the seasons and the latitude, and which displayed the time with a pointer on a scale. The final and most sophisticated version, the ‘anaphoric’ clock evolved somewhere around 100 BC. It was picked up by the Romans and described by the 1st century BC architect Vitruvius. Its usual Latin name was horologium, although that can also apply to a sundial or simple clepsydra.

A working model of the heavens

Water was fed from a header tank into a regulator tank. This was kept full to overflowing, which meant that under the constant head, the water emerged from the bottom at a constant rate. It fell into a receiver, in which was a float that, by means of a cord or fine chain with a sandbag as counterweight, turned an axle once every 24 hours. The end of the axle carried a bronze disc engraved with a map of the stars in stereographic projection (that is, the dome of the heavens represented on a flat surface) centered on the north celestial pole and extending to the Tropic of Capricorn. The ecliptic (the path of the sun through the sky) was represented on the disc by a circle of holes, ideally 365 in number. A peg to indicate the sun was plugged in the hole appropriate to the day (that is, in its proper place relative to the stars) and was moved forward by hand one hole per day. Thus, it simulated the sun’s annual passage through the heavens, just as the turning disc simulated the daily rotation of the heavens.

In front of the disc was a fixed grid of wires. One of them represented the horizon as seen at the place where the clock was located. This is where latitude mattered. Both the disc and the grid designed for a clock in Southern Italy, for instance, would be very different from those designed for one in Northern Britain, where the sun’s path through the sky is different, and the horizon is in a different place relative to the stars. An Italian clock in Britain would be wildly inaccurate.

The other important wires represented the hours, 12 above the horizon, 12 below. At the beginning of the day, the sun peg behind the grid climbed up above the horizon wire. One seasonal hour later, it passed the first hour wire, and so on until after 12 hours it dropped below the horizon. Thus it mimicked the actual position of the sun in the sky; and it did so even by night when the real sun was not visible. Provided the disc turned once in exactly 24 hours and the peg was advanced one hole per day, this elegant clock automatically adjusted itself to the changing seasons. It was a working model of the heavens.