Expert Advice: A Leap Second — One More Time!

From left: Dennis McCarthy, Wayne Hanson, Ronald Beard and William Klepczynski

By Dennis McCarthy, Wayne Hanson, Ronald Beard and William Klepczynski

Once again we are going to adjust the world’s clocks by one second. This time it will happen on June 30, when we insert another leap second in Coordinated Universal Time (UTC), the standard international time scale. In theory, all UTC clocks should insert a second labeled 23h 59m 60s (the leap second) following one labeled 23h 59m 59s UTC. This is equivalent to having all of the clocks in the world stop for one second at that time.

Are you ready for it?

The last leap second occurred two years ago on June 30, 2012, and the continuation of the process of making these one-second adjustments has stirred a growing controversy over the last few years.

How did the leap second come about — and why do we continue making these sporadic adjustments?

From Sun to Caesium

Historically, it has been easy to make use of the apparently uniform repetition of various astronomical phenomena to measure the passage of time. We’re familiar with the Sun rising and setting, and this regularity provides us a convenient measure of time: the solar day. In recent times until 1960, the average solar day was used as the basis for timekeeping, and if we divide the day into 24 hours, each containing 60 minutes made up of 60 seconds, we can define the second as 1/86,400 of the mean solar day. This meant that the length of the second depended on the Earth’s rate of rotation because it is the rotating Earth that causes the Sun to appear to move across the sky.

In the mid-1930s, astronomers concluded that the Earth did not rotate uniformly as measured by the most precise clocks then available. This causes the duration of a second to vary as the Earth’s rotation rate varies. We now know that a variety of physical phenomena affect the Earth’s rotational speed, and consequently this definition of a second became impractical for applications that require a truly uniform time scale. So, in 1960, the second was redefined in terms of the Earth’s yearly orbital motion around the Sun. The time scale provided by this astronomical phenomenon was called Ephemeris Time (ET), to call attention to the fact that its realization depended on the conventionally adopted positions and motions (that is, the ephemeris) of the Sun (or Moon) that was used in the analyses of the required astronomical observations. The second defined in this manner was called the Ephemeris second.

Although Ephemeris Time does provide a more uniform measure of the duration of a second, it is inconvenient to make the necessary astronomical observations that would be required to maintain a practical time scale for applications that demand high precision. So, in 1967, the second was redefined again, this time in terms of the frequency of an energy level transition in the Caesium atom, which had already been calibrated with respect to Ephemeris Time by using astronomical observations of the Moon’s motion. Caesium frequency standards, by the early ’60s, had become known as reliable, uniform, accurate and precise clocks. The second defined in this way provided, and continues to provide, a uniform standard of time that can easily be measured in a laboratory with greater precision and accuracy than any astronomical phenomena.

Lab Clocks Rule

Although the second defined using the frequency of an atomic energy level transition does provide a unit of time duration that is precise and uniform, it does mean that the passage of time measured in this way is no longer connected to astronomical phenomena. Indeed, with the advent of more accurate observational techniques, astronomers could measure variations in the Earth’s rotation rate by measuring its changing orientation in space and comparing the rate of change with laboratory clocks. They established that among the various variations in the Earth’s rotation rate is the gradual slowing down with respect to a uniform atomic time scale. This deceleration is consistent with theoretical tidal effects and observed terrestrial deglaciation.It is also apparently consistent with ancient observations of solar eclipses, indicating that that this slowing has been going on for thousands of years

As a result, if we were to observe a recurring astronomical event, we would see it happening earlier from day to day. To bring our clock back into agreement with the astronomical event, we would have to add some time to the face of our atomic clock. While astronomers can cope with this situation by applying the appropriate corrections derived from astronomical observations that measure the Earth’s rotation rate, navigators that relied on astronomical observations to determine their positions considered this situation problematic.

When the definition of the second based on the Caesium atom was introduced, it was known that there would be a time varying discrepancy between a clock running at a uniform rate and a theoretical one using a second defined by the Earth’s rotation rate. Starting from 1961, the observed discrepancy was modeled by making small adjustments on the order of a few milliseconds (thousandths of a second) to our clocks at first, and later by making small adjustments to the frequency of the atomic clocks from time to time, usually on an annual basis. This meant that the duration of a second could vary depending on when it was measured.

No More Changes

In 1970 the International Radio Consultative Committee (CCIR and now known as the International Telecommunications Union Radiocommunications Sector, or ITU-R) in collaboration with other international agencies adopted a definition of UTC that did away with any periodic changes to the duration of the second. Instead it was decided that the discrepancy between UTC and the observed rotation angle of the Earth would be accounted for by making one-second adjustments when needed, so that the absolute difference between UTC and the Earth’s rotation angle measured in time units would always be less than 0.9 seconds. A finer correction would also be provided frequently so that the Earth’s rotation angle in time units designed as Universal Time 1 (UT1) could be derived to 0.1 second precision.

It was specified that the one-second adjustments, either positive or negative, were to be made preferably at 23h 59m 59s on the last day of the months of December or June, but could also be made, if necessary, at 23h 59m 59s on the last day of the months of March and September, and further if required at 23h 59m 59s on the last day of any month. The implementation of this definition actually began in 1972, a year in which two leap seconds were introduced.

These one-second adjustments came to be known as “leap” seconds by analogy with the “leap” day inserted in calendars. This definition then fixed the second in UTC to be uniformly established as the international standard atomic second defined by the resonance frequency of Caesium and known as the SI (Système International) second.

Compromise Overcome by GNSS

The introduction of the concept of the leap second was historically a compromise with practitioners of celestial navigation who needed to base their observations on astronomical time to determine their longitude. If UTC doesn’t differ from the observed rotation angle of the Earth by more than a second, navigators could use UTC directly as a substitute without introducing a systematic error greater than a quarter of a mile. However, the routine practice of using celestial navigation has been overcome by the success of Global Navigation Satellite Systems (GNSS), inertial navigation systems, and radar navigation.

In fact, the U.S. Naval Academy stopped including celestial navigation in its curriculum in 1998. In the time span since the introduction of the idea of a leap second, computer networks, wireless telecommunication systems, satellite communications, telephone networks, air traffic control systems and even industrial processes have developed to the point where precise time is an essential component of their successful operation. Users and suppliers of these systems are concerned with the impact of sporadic, essentially unpredictable, one-second adjustments.

Most of these modern systems derive their time using GPS timing receivers. Although the navigational solutions make use of GPS System Time, these receivers provide UTC by means of a broadcast correction that provides the time-varying difference between GPS System Time and UTC. This correction normally provides the varying difference between the two times to less than a microsecond but must also keep track of when a leap second is introduced. As the leap second changes occur sporadically, there may be worries that problems could arise because hardware or software may never have been tested thoroughly for a leap second occurrence. As a result of these concerns, as well as the cost of stopping all of the clocks in the world for one second, the ITU-R has been discussing a possible revision of the definition of UTC by dropping the future use of leap seconds.

Leap or Not Leap?

The question of the future of UTC was raised in 2000 with the suggestion of modifying it to be a continuous timescale without leap seconds. Consideration of this question is still ongoing. The 2012 World Radiocommunication Conference (WRC-12) identified this issue as urgent, requiring further examination by the 2015 World Radiocommunication Conference (WRC-15) “to consider the feasibility of achieving a continuous reference time-scale, whether by the modification of Coordinated Universal Time (UTC) or some other method, and take appropriate action…”.

With the aim of providing adequate technical background for WRC-15 to make an informed decision on this issue, the International Bureau of Weights and Measures (BIPM) and the ITU agreed to organize jointly a workshop on the future of the international time scale. This workshop was held in Geneva, Switzerland, in September 2013. It provided a unique opportunity to present available information on current and possible future precise frequency and time standards, sources and their characteristics, time scales and dissemination systems and different views on the future of UTC.

Contributions to the workshop were specifically invited to ensure that the breadth of the issue would be covered. Included were the relevant international organizations (the International Astronomical Union, the International Earth Rotation and Reference Systems Service, the International Union of Geodesy and Geophysics, the International Organization for Standardization, the International Maritime Organization, the International Civil Aviation Organization, the Union Radio-scientifique Internationale), the providers of GNSS services (GPS, GLONASS, Galileo and BeiDou), the national metrology institutes that realize and maintain local representations of UTC, the ITU member administrations, and the ITU-T and authorities responsible for electronic time services. Information on the workshop, agenda and presentations is available.

Final Decision in November

A special issue of ITU News magazine dedicated to the workshop has also been published; an online version is available. It did not provide a decision on the issues, but rather a forum for issues to be discussed, since there is some controversy over modifying the global reference time scale. The final decision is to be made at the WRC-15 in November when the method for satisfying the feasibility of achieving a continuous time scale will be determined as well as how it would be implemented.

As preparations begin for the June leap second, hardware and software will undergo testing. This process is likely to be repeated for some time to come, even if the decision to eliminate the use of leap seconds in UTC is made. Legacy systems reliant on the use of leap seconds will require an adequate period of time to adapt to any change in the definition of UTC. If the suppression of leap seconds would be decided, it is recommended that a period of time no less than five years be allowed  before the Final Acts of the WRC-15 go into effect. So, leap seconds could be with us for some time yet.

Editor’s Note: For an earlier discussion on the leap second by McCarthy and Klepczynski, download the Innovation article “GPS and Leap Seconds: Time to Change?” from the November 1999 issue of GPS World.

Dennis McCarthy is retired, and serves as a contractor with the U. S. Naval Observatory, where he was science advisor, director of the Directorate of Time, and head of the Earth Orientation Department. Internationally, he has served as president of the Commissions on Time, Commission on Earth Orientation, and Division 1  (Fundamental Astronomy) of the International Astronomical Union (IAU). He was also secretary of Commission 5 of the International Association of Geodesy.

Wayne Hanson has been a consultant and president of Time Signal Engineering since his retirement in 2001 as chief of the Time and Frequency Services Group in the Time and Frequency Division of the National Institute of Standards and Technology. He is the U.S. chairman of the International Telecommunication Union – Radiocommunication Sector, Working Party 7A concerned with Time Signal and Frequency Standard Emissions.

Ron Beard is the head of the Advanced Space PNT Branch at the Naval Research Laboratory and International Chairman of ITU-R Working Party 7A, Precise Time and Frequency Broadcast Services. During the early development of GPS in the 1970s, he was the project scientist in the NRL GPS Program Office that developed Navigation Technology Satellites One and Two that operated the first atomic clocks in space.

William Klepczynski is now retired. During his career, he was a consultant to the Institute for Defense Analyses and the head of the Time Service Department of the U.S. Naval Observatory, where he managed the USNO Master Clock, timing operations for GPS and time distribution systems that utilize communications and navigation systems.