Adjusting after a ‘long’ weekend at the Royal Observatory – Precision clocks and the leap second
It is unlikely that anyone came into work feeling more refreshed than usual following the insertion of an extra second into Saturday 30 June. While this event did not affect our body clocks, it did have an effect on the National Maritime Museum’s collection. At the Royal Observatory two clocks, in particular, needed adjustment: the Shepherd gate clock (ZAA0533) and the Hewlett Packard atomic clock (ZBA2599).
The Shepherd gate clock was installed at the Observatory in 1852 under the direction of the seventh Astronomer Royal, George Biddell Airy, and is named after its maker, Charles Shepherd. The clock’s installation marks the beginning of the Royal Observatory’s role as a provider of accurate time for the nation. Today it can be relied upon to show Greenwich Mean Time (GMT) accurate to half a second. The second clock was originally used at the National Physical Laboratory in the 1990s for its contribution to the international time-scale, known as Coordinated Universal Time (UTC).
Why do we need a leap second?
Believe it or not, the Earth is not as reliable a timekeeper as our atomic clock and the leap second is a fine-tuning that keeps UTC synchronized with the motion of the Earth. The speed of our planet’s rotation is variable for a number of reasons. Generally speaking, the Earth’s rotation is slowing down because the Moon is gradually moving away from us: there is no need for alarm as it is only moving away at about the same speed our fingernails grow, but as it does so its gravitational influence is lessened. This and other unpredictable influences, such as seismic activity, are monitored by the International Earth Rotation and Reference Systems Service (IERS), who decide whether or not a leap second is necessary.
The leap second has caused us to adjust the two clocks already mentioned and make a basic equivalent calculation when comparing the rates of precision pendulum clocks against a radio-controlled clock. But for those responsible for managing computer networks that handle large volumes of digital traffic the leap second poses a tricky problem. Networks depend on accurate timekeeping: every data transfer is time-stamped and any data logged during the leap second will be outside of the computer’s logic. This could potentially cause the system to fail. Indeed, there are reports indicating that some systems did stop working after last Saturday’s leap second and, as a result, some popular websites were inaccessible. Among various ways of overcoming this problem, there is one that bears an interesting parallel to operational principles used by Airy in the mid-1800s. Google, in preparation for the 2005 leap second, opted to add it, in advance and over a set period across their system, in tiny increments that did not upset communication with any other systems that they were connected to. They named their method the ‘leap smear’.
The same principle was applied to the system that drove the gate clock at the Royal Observatory. Most people do not realize that this is a slave clock, not a timekeeper in its own right: it was originally one of several slave dials on the site, all of which followed the instruction of Shepherd’s master clock (ZAA0531). The beauty of Shepherd’s master clock was not that it was an exceptional timekeeper but that it could be corrected electrically by means of electro-magnetic adjusters working on the pendulum. Small and precise adjustments to the hands could be made by causing the pendulum to move faster or slower over a fixed period of time; and because the master clock was continuously sending out pulses to all of the slave clocks, the slaves were automatically corrected. Today, the gate clock is driven by a radio-controlled Quartz clock and the hands are corrected manually.
Of course, the leap second was unheard of in Airy’s time; the first leap second was implemented in 1972 and, since then, a total of 35 have been inserted into UTC. The IERS gives six months’ notice of forthcoming leap seconds in their Bulletin C, so we will have at least another year before we can look forward to a ‘long’ weekend like that of 30 June – 1 July 2012.