The People Who Keep Time

For much of her working life, Ruth Belville travelled around London selling people the time. She checked a pocket chronometer against the Royal Observatory at Greenwich, then visited customers who used it to set their own clocks. According to Royal Museums Greenwich, her service was accurate to about a tenth of a second.

This sounds like an occupation invented for a children’s book. It was, in fact, a practical answer to a serious problem. Owning a clock did not establish that it was right. For that, you needed a trustworthy connection to something better. In this case, the connection arrived carrying a handbag.

We now expect the correct time to appear on our phones without intervention. The work has become less visible, which is easily mistaken for its disappearance. There are still people keeping time for the rest of us. Their customers have simply become rather demanding.

When noon was a local matter

Before standard time, a town could organise its day around its own position relative to the Sun. This was reasonable until people began travelling quickly enough for the differences to become inconvenient. A railway timetable is less useful when every destination has an independent opinion about the present.

Royal Museums Greenwich records that most British railway companies had adopted Greenwich Mean Time by the 1840s. Greenwich time became the legal standard in Britain in 1880. Coordination had acquired enough practical value to overrule local astronomy.

The significant invention was not a better clock alone. It was an agreement about which clock everybody would follow. You could have an exquisitely maintained instrument and still miss the train through an excess of municipal loyalty.

That distinction survives. Accuracy concerns how well you measure something; coordination also requires agreement about the reference. A room full of accurate clocks is not necessarily a room full of people who will arrive together.

The planet has failed its appraisal

For a long time, the Earth supplied the obvious basis for measuring a day. Unfortunately, its rotation is not perfectly uniform. As our clocks improved, we discovered that the enormous celestial object beneath them was the less dependable part of the arrangement.

The modern second therefore takes its reference from atomic physics. The international definition fixes the frequency associated with a particular transition in caesium-133 at 9,192,631,770 hertz. Put more accessibly, the second corresponds to that many periods of the specified radiation.

The formidable number preserves continuity with the second we already used. Scientists did not decide that Tuesday needed to become a different length. They found a more reproducible way to measure the familiar unit.

There is something pleasing about this reversal. We began by asking the heavens what time it was. We eventually built equipment precise enough to conclude that the heavens were taking liberties.

The master clock does not exist

It would be satisfying if the world’s time came from a single magnificent clock in a guarded room. Perhaps someone would approach it each morning in special shoes. The actual arrangement is more distributed, more mathematical and considerably harder to photograph.

The International Bureau of Weights and Measures, usually known by its French initials, BIPM, computes Coordinated Universal Time from data contributed by approximately 450 atomic clocks at around eighty laboratories. Participating institutions maintain their own local versions in real time.

The international comparison is then published monthly in a document called Circular T. The definitive reckoning of how those clocks related to UTC arrives retrospectively. This does not mean nobody knows the time until next month. It means the most exact account requires measurements, comparison and calculation.

My interpretation is that the system’s strength lies partly in its refusal to appoint an infallible instrument. Clocks are checked against other clocks. Differences are examined. Confidence comes from a maintained process rather than the impressive appearance of one machine.

The people doing this are professionally interested in discrepancies most of us could not notice. It is an excellent quality in somebody responsible for making the rest of civilisation agree.

Einstein would like a word about your delivery

This precision is not merely a competitive hobby for laboratories. Satellite navigation depends on timing signals, and the clocks involved do not experience time at identical rates.

As NIST explains, GPS satellites’ motion would make their clocks lose about seven microseconds a day relative to clocks on Earth. Their position in a weaker gravitational field produces an effect in the opposite direction. Combined, the difference is approximately thirty-eight microseconds a day faster.

Those effects must be accounted for. Relativity is therefore involved in the apparently ordinary business of establishing where you are. A theory that unsettled our understanding of the universe now helps a delivery driver discover that your front door is behind the bins.

We tend to separate profound knowledge from useful knowledge. GPS makes the distinction awkward. Understanding that time behaves differently under different physical conditions turns out to be relevant to finding a roundabout in Milton Keynes.

A disagreement lasting one second

Atomic time creates a further difficulty. We still organise civil life around a rotating planet, even though we have stopped trusting it to define the second precisely. Two reasonable requirements have become slightly incompatible.

Leap seconds were the compromise: occasional adjustments keeping UTC close to time derived from Earth’s rotation. The inconvenience is that digital systems must accommodate an unusual event in something they would prefer to treat as a regular sequence.

In its 2022 resolution on UTC, the General Conference on Weights and Measures explicitly identified risks to digital infrastructure from leap-second discontinuities and inconsistent ways of handling them. It agreed to permit a greater divergence between civil and rotational time.

The disagreement is revealing. Astronomical fidelity and uninterrupted technical coordination are both defensible goals. Neither can simply be declared the only meaning of correctness.

We like standards to look inevitable once established. Here, the people responsible had to confront a less comfortable question: which imperfection should everybody agree to live with?

The infrastructure of agreement

There is a broader lesson here, although it does not require pretending that every political dispute could be resolved with better clocks. Physical measurement has advantages unavailable to a committee arguing about housing.

Still, shared time illustrates how much useful certainty must be actively produced. It depends on instruments, expertise, published methods and institutions willing to compare their results. The agreement is reliable because the differences remain visible to the people maintaining it.

We are accustomed to noticing infrastructure when it breaks. Timekeeping is particularly vulnerable to being taken for granted because its output looks so simple. A few digits change. Nothing appears to happen. That uneventfulness conceals a great deal of accomplished work.

Ruth Belville’s customers could see the service arriving. They understood that the correct time had to reach them from somewhere, and that somebody was responsible for the journey. The contemporary version is so convenient that we can forget there was a journey at all.

Tomorrow morning, your phone will display the time. Behind it will be atomic measurements, international comparisons and people investigating fractions of a second with considerable seriousness. You may look at the result, decide you have another five minutes, and go back to sleep. It is an extraordinary amount of precision to bring to being late.