A brief history of timekeeping: from railway time to scientifically measured, digitally coded time

Author: Richard Hoptroff, CTO and Founder, Hoptroff

Measuring when events take place is fundamental to all businesses, and a tighter grasp on that sword has repeatedly proven to earn commercial advantage over the centuries. 

But time is not as simple as it seems: it’s just perplexing and exciting, but also surprisingly political. Within this paradigm, knowing the right time can improve financial services performance, and can keep your business practices ahead of the game.

Why time synchronisation can help in different industries

The need to quantify time was agricultural: knowing when to plant and reap. This gave rise to calendars, where counting days was all that was needed to work out the seasons by counting the days that have elapsed.

Since then, more and more benefits of time synchronisation have materialised; transportation and communications have improved dramatically; life started moving at a relentlessly increasing pace; and with these changes, the need to agree and quantify the passage of accurate time became more and more acute. But each time we try to peel back a layer of time’s onion, we discover that time is not an onion, in fact, a can of worms.

8 September 1752 never existed

Julius Caesar’s 365-day calendar wasn’t quite a full year, so Pope Gregory XIII introduced leap years in 1582, happening roughly every 4 years.  Adjusting to this across Europe was a monumental feat politically and religiously. By the time England switched over in 1752, it had to make up for lost time, literally. So, Wednesday 2 September had to be immediately followed by Thursday 14 September.

Bristol had no clock synchronisation and ran 8 minutes slow compared to London 

In 1840 they introduced Railway Time, a single time for all of the United Kingdom. As railways propagated across the globe, time synchronisation became more important. But the idea of midday being at sunrise wasn’t going to wash with the Americans, and France insisted on doing her own thing, so the idea of time zones was developed, in sync but spaced apart by exact hours. Interestingly, Russia’s Trans-Siberian Railway timetable remains all in Moscow time, despite passing through 8 time zones.

Politicians shift time

Daylight saving time is the sudden switching of time zones by an hour according to frequently changing legislation. An hour is skipped or repeated. The only way to avoid confusion is to reference a global meridian time, which also avoids time zone confusion.

France insisted on her own meridian 

For a while, as she couldn’t palate the Greenwich Observatory defining time. When you’re next in Paris, search online for the “Arago Route” to discover a line of diamond-shaped studs in the streets defining the longitude of the now defunct méridien de l’Observatoire de Paris.

France used decimal time 

For a short while, as Napoleon Bonaparte needed to assert himself on the international stage. There were 10 decimal hours per day, each with 100 minutes composed of 100 seconds. This only lasted a few years, though his similar insistence that carriages pass on the right remains to this day.

Days are getting longer

Today will be about 65 billionths of a second longer than yesterday, because the earth is being dragged by the tidal action of the moon. That may sound small, but it adds up: Tyrannosaurus rex’s day was a mere 19 hours.

Some minutes have 61 seconds

These leap seconds are largely because the earth is slowing down, but the rate of slow-down depends on the weather. The earth’s rotation slows down when snow is stuck on mountains and speeds up when it melts. So, leap seconds need to be added every few years to realign clocks. Since the processes involved are unpredictable, we timekeepers only get 5 months’ notice of these events.

This plays havoc with computer systems, which are designed for 60-second minutes. Suddenly they have to jump back in time, creating negative time intervals and software can crash in a tearful heap of stack dumps unless managed correctly. The alternative, which has been adopted by some scientific and computing fields, is TAI, which is defined as what UTC would have been if the leap seconds hadn’t been added.

The French won the time wars

Sort of. BIPM is the Bureau International des Poids et Mesures in Paris. TAI stands for temps atomique international. The global meridian is no longer Greenwich Mean Time, but temps universel coordonné, or Coordinated Universal Time, which is somewhere in a rather pleasant field a bit to the east of Greenwich. In the worst compromise in political correctness history, this meridian is known by the acronym UTC, correct neither English nor French.

But the first working atomic clock was developed by Essen and Parry at the National Physics Laboratory in the UK; I produced the first atomic watch in 2014, accurate to one second every thousand years. So the UK remains well ahead where it counts.

Time is uniquely cumulative 

Unlike other weights and measures. A pound of butter is a pound of butter whenever you weigh it. But to know the time, you have to count the total number of seconds that have elapsed. Even tiny errors quickly become noticeable and clocks disagree, especially when what you’re interested in is the intervals between events measured by different clocks.

Time is a consensus, not an absolute 

Since we can’t trust the stars or the earth’s rotation to tell us the time anymore, all we have is our own clocks. Some super-duper optical lattice clocks, sure, but who’s is right? Right now, time is defined by comparing the clocks from around 70 national standards institutes worldwide and reaching a consensus, moderated by the BIPM standards institute.

If we are to understand what is happening in a fast-moving world, we need to be plugged into that consensus, just as we plug into the power grid, and trust it, just as we trust the peer review of audit processes. So as suppliers of time, we are in an odd space somewhere between being a telecoms utility and a forensic auditor.

Time’s sell-by date is now 

To know the time requires traceability: a known and assured continuous chain of comparisons back to BIPM Stratum Zero sources. If that is not maintained, trust in your time rapidly evaporates.

Knowing the time matters

As complicated as time is, politicians have for the time being agreed upon some kind of global consensus of what the time is.  Why is that consensus important? Why is so much money and political capital invested in it? I call the reasons the eight CRESCENT factors:

C – Causality 

Time has an arrow. A cause can only have an effect if it happened first. (Actually, if you’re a physicist, it’s the other way round – causality, the measurable sequence of events, defines time.)

For example, on 6 May 2010, around 2.30pm New York time, automated trading systems somehow locked into an ecstatic tryst and the markets went nuts. Accenture’s shares traded for a cent  and for one glorious moment, Sotheby’s, the auction house, had a market capitalisation of six trillion dollars. A major embarrassment for the traders, but nothing compared to the repercussions of the failure of forensic oversight. It was impossible to work out when who did what. Trades executed in less than a thousandth of a second were timestamped with clocks out by as much as a second. We still don’t know the sequence of deals – what triggered what. This is why the MiFID II and Consolidated Audit Trail timestamping regulations were developed: to establish an auditable chain of causality.

R – Recognition 

Information sources are distributed and unless we can stitch their information together in a time synchronised way, the information is useless. Hence oil exploration seismologists being the biggest customers of atomic clocks. Recognition is also great for spotting outliers, detecting where the world is not behaving as it should. Is this a potential fraud? Does this jet engine need a service? etc.

E – Efficiency 

Collecting intelligence on how long processes take is the first step towards improving the productive use of our resources, reducing bottlenecks and speeding up work-in-progress. But in digital services, “how long processes take” involves looking at the difference between timestamps created by different clocks. If the clocks don’t agree closely enough, the intel evaporates.

S – Synchronisation 

Our virtual world only works because gazillions of photons and electrons constantly negotiate and weave around each other. Unless they know when to jump on the wire, internet packets collide. Radio frequencies stamp on each other’s feet. Our virtual world crunches to a halt. Our data has no value if it arrives late or not at all.

C – Control

We’re capturing and processing all this data because real-time automation allows us to take actions and achieve things that otherwise would be impossible. Land airplanes in fog. Ride hoverboards. Trade on information ahead of the next guy.

E – Expectations  

Time-accurate data is needed before you can start to predict what might happen next. Whether it’s millisecond market swings or how many sandwiches to stock on the shelves tomorrow, if you can’t be sure when your data points come from, you can’t start to use forecasting to improve business performance.

N – Navigation 

Ever since Harrison’s Sea Watch No.1, which allowed longitude to be determined, knowing where you are depended on knowing the time. This is true of GPS satellites today, and it will be in future as new, terrestrial systems provide the time resilience that satellites, highly accurate as they are, cannot provide.

T – Traceability

Just as important as the use of double-entry and hash ledgers are in bookkeeping, events need to be timestamped, and those timestamps need to be trustable. This can be done by using distributed ledger technology (DLT).

“He who controls the past controls the future” – George Orwell, Nineteen Eighty-Four

Time on tap

It’s clear that there are many business advantages to accurate timing – saving money by improved use of bandwidth and computing resources; being able to audit events in time, place and identity; using real time intelligence to optimise business execution; detecting behaviours that are not business-as-usual; and so on.

It’s also clear that accurate timing is not an easy thing to obtain. That’s why we created Traceable Time as a Service (TTaaS®), a global network of high-accuracy time delivery over internet protocol, originally to enable the MiFID II and Consolidated Audit Trail regulatory timestamping requirements to be fulfilled at low cost. Since then, the potential is becoming more widely recognised.