Sorry, you need to enable JavaScript to visit this website.

The case of a missing second

One tiny leap

It may seem minuscule, but a row over a missing second could cause chaos in financial markets and deprive Britain of its position at the centre of the world’s time. Tom Whipple investigates

It may seem minuscule, but a row over a missing second could cause chaos in financial markets and deprive Britain of its position at the centre of the world’s time. Tom Whipple investigates

Tom Whipple | June/July 2016

On June 30th last year, midnight happened twice. The clocks reached 00:00:00, a second elapsed, and they still read 00:00:00. It was not the first such anomaly. Strictly speaking, on December 31st 2008 the ten-second countdown to the new year took 11 seconds. The same happened in 2012, 2005 and 1998 – even if most of the world remained unaware.

In fact, every other year or so since 1972 has contained a 61-second minute, as the world’s timekeepers inserted a leap second into the planet’s clocks.

Unlike a leap year, we don’t know in advance when these will happen. The next one may be this December, or it may be in 2017. But whenever it comes one thing is sure: it will be largely because of the dogged insistence of Britain – and against the express wishes of most other countries, some of which think Britain wants to keep the leap second simply because losing this oddity would deprive it of its place as the centre of the world’s time.

Or as Judah Levine, America’s chief timekeeper, put it at a global time conference, “I have bad news for you, Britain: Victoria isn’t queen anymore.” Then, leaning in closer, “We are stealing the Greenwich Meridian.”

Britain isn’t the only reason the leap second persists. The reason we have it at all is because humans have become too good at telling the time. Clocks are, and always have been, a mechanical device for modelling the rotation of the Earth. If this is not immediately clear to 21st-century humans, it would have been obvious to our medieval ancestors. Look at the Wells Cathedral clock, by some accounts the oldest clock face in the world, and you see not an hour hand but a sun. Slowly throughout the day, that sun processes around the Earth, at its centre. A clock’s purpose is to tell the time, and for the medieval congregation gazing on it, time was the position of the sun.

Through the centuries that followed, time-keeping devi­ces became more abstract but their goal was the same: to mirror the time the sun takes to travel across the sky and arrive back in the same place the next morning. As we got better at doing this, so the increments of time got smaller. First the hour was divided into 60 minutes, then the minute into 60 seconds.

Then, around the middle of the last century, scientists decided a second was not 1/86,400th of the time from noon to noon at all. They decided instead it was something that was ostensibly its equivalent: 9.2 billion oscillations of a microwave beam, when tuned to the frequency of a caesium atom. The advantage of atomic time was that you no longer needed to look at the sun to set your clocks. The disadvantage?

In every Victorian natural-history museum there was a clue, if we had known where to look, that introducing “atomic time” in this way would bring problems. That clue is fossilised coral. Coral grows so fast that they have not just annual rings – as trees do – but also daily rings. This means that in coral for every 365 daily rings, there is one annual cycle.

If any Victorian fossil collectors had bothered to count the rings on their fossilised coral they would have seen something puzzling: there were too many days in a year. Indeed, go back a few hundred million years, and there were 400. The reason? The day, back then, was a few hours shorter.

The leap second is a fudge. It happens because the world’s rotation is slowing. Every tide that hits the world’s beaches takes energy out of its spin, through the action of the moon and sun. And so, slowly – irregularly – we spin a little less. That was what we found when we switched to atomic time. Atoms aren’t affected by tides. The Earth’s rotation, for so long the governor of our time, is several million times less stable than the caesium atom. So, to correct for the discrepancy between the two, we have added, every now and then since 1972, a second.

At the end of the chosen day, in a few hundred of the world’s most accurate atomic clocks, a small intervention is made just as the clocks tick round to 00:00:00. In the global positioning satellites silently travelling across the night sky, and the servers gently humming beneath New York, London and Tokyo, the world’s time standard – the standard that keeps our markets trading, our planes flying and our computers talking – is incremented by a second. And it is still 00.00.00.

A leap second has just been added, and in the past not everyone has been prepared. At the end of June 30th 2012, before the next 9.2 billion oscillations were complete Reddit, Foursquare and Gawker, among many websites, crashed. On June 30th 2015, some financial markets chose to close for a minute, while engineers at Amazon, Apple, Netflix and Twitter all had to scramble to restore and repair disrupted services.

Now, most of the world is increasingly of the view the leap second causes more problems than it solves. Too much has changed since 1972. Then, the time standard was a technical concept – linked to a few government computers and national observatories. Today, it runs the world. So representatives from the top scientific nations, meeting in Geneva, have been trying, and failing, to kill the leap second.

The International Telecommunication Union’s conference hall in Geneva seems designed for deadlock. Underground, overheated, its windowless wood-panelled walls enclose concentric semi-circles of bored delegates. There are dozens of basements like this in Geneva – where people must feign interest in “the conclusions of working group 7a” or turn when instructed to “the summary of paragraph 3.2, discussion document”. Here, over the past half-dozen years, a series of gatherings of the world’s chief time scientists – journalists like to call them the Time Lords – has sought to reach unanimity, so that a new global time system can be developed. These meetings have been the Bretton Woods of time and, for America’s Time Lord Judah Levine, they have been about recognising that, as with Bretton Woods, the point is the economy, stupid.

Every second, from his laboratory in the mountains of Colorado, Levine receives 160,000 requests for a time stamp from his atomic clocks. These include those given to official commercial clients, who need time verified to sub-second accuracy in order for their businesses t0 exist. If time is money then his time signal is its watermark.

“These are people working at sub-second accuracy,” he said. What is absolutely crucial, for financial clients of Levine, is latency – in other words the time that elapses between ordering a trade, and the trade actually happening. It may be only a tiny fraction of a second, but when the best trades go to the fastest traders, a fraction of a second costs. “One of the customers for our time took out big web adverts saying they had decreased the latency between the New York and Brazilian stock exchanges by two milliseconds. That was regarded as a big improvement – these rapid trading companies make many trades a second, spotting small differences in price.”

This is not a system that has time zones, or daylight saving time. In summer and winter, in Vladivostok or Vancouver, it works on Universal Coordinated Time – effectively Greenwich Mean Time. So it certainly does not welcome the ad hoc introduction of seconds, as and when the Earth’s messy rotation requires it. Indeed, some fear that if we get a leap second wrong – if even for a second Tokyo, say, is on a different time from New York – the global financial consequences could be catastrophic.

The worry is not just that the leap second may cause computers, markets – or for that matter aeroplanes – to crash, but that even if all goes to plan for a two-second period it will be impossible to distinguish between 00:00:00 the first time round, and 00:00:00 after the leap second. The order of trades will be impossible to decipher – causality itself, at least as far as financial regulators are concerned, will be reversed. “It’s only a matter of time”, said Levine, who knows about matters of time, “before there’s a real problem.”

To which Britain – and in particular its Time Lord, Peter Whibberley – says: nonsense. The British position can best be summed up by his peroration in 2013, when he spoke to the floor at yet another failed Geneva conference. Removing the leap second would not be, “a minor technical change, but rather a fundamental change. Ever since mechanical clocks were created, we have always adjusted to model the rotation of the Earth.”

When he was Britain’s science minister, David Willetts got to decide how Britain would vote on the leap second. He sees the battle as one between time geeks and the rest of humanity, and is unrepentant about British intransigence. “If we just fix time by the atomic clock it will, over the centuries, diverge from time as measured by the rotation of the Earth. The sun would not be at its highest in the sky at noon: the Meridian would no longer be at Greenwich but slowly heading towards Paris. Leap seconds enable us to...keep the time of day linked to the rotation of the Earth even as it slows. The purists disapprove of this but we should not sever the link between time and the pattern of our lives on Earth.”

Some of the world’s time scientists believe that Britain is motivated less by the belief that the abstract notion of time should stay moored to the physical reality of the planet than by the idea that Britain might lose control – and worse, sacré bleu, that midday might, in a few centuries, end up in France. After all, the last occasion the time community had a disagreement this intractable was during the 30-year period when both Paris and London laid claim to be the centre of time.

Whatever the reason, there was no agreement in Geneva in 2013, nor at the subsequent conference in 2015. A decision has now been put off until 2023 – seven years and (depending on the tides) four leap seconds away.

2 Readers' comments

Sign in or Create your account to join the discussion.

zleo99 - December 30th 2016

If the length of the day has been changing throughout Earth's existence/(religious) time, how many days were there in a year/ how long was a day, at the time of the Creation??

Steve Allen - May 27th 2016

The UK is not who kept the leap second in place. At the 2015 March CPM15-2 meeting http://www.cept.org/ecc/groups/ecc/cpg/client/introduction/report-of-the-second-conference-preparatory-meeting-for-wrc15-(cpm15-2)-(23th-march-%E2%80%93-2nd-april-2015) "The CPM also considered the proposal from six Arab group countries and after the discussions the new method D was incorporated in the CPM text." And method D is at http://www.itu.int/md/meetingdoc.asp?lang=en&parent=R12-CPM15.02-C&PageLB=125 "Proposed addition of a new Method (No Change) - Agenda item 1.14"