To leap or not to leap: scientists debate a timely issue.

Did last New Year s Eve seem a trifle tedious. I Did your celebration go on a little too long? Maybe that s because just before midnight Greenwich Mean Time--6:59:59 Eastern Standard Time to be exact--the international authority on timekeeping ordered everyone to wait a second. For the 23rd time since 1972, the International Earth Rotation and Reference System Service added an extra second to the time standard, a worldwide network of some 200 atomic clocks.

The clocks, most of them governed by the ultrasteady vibrations of electrons in cesium atoms, are accurate to a tenth of a billionth of a second a day. However, humankind's oldest clock--Earth's rotation--isn't nearly so precise. Primarily in response to the moon's tidal pull on the oceans, our planet isn't turning quite as fast as it used to. To keep Earth time and atomic time in sync, experts have agreed to insert a leap second every few years into the official atomic-based standard, which is called Coordinated Universal Time.

Because the rate at which Earth slows isn't perfectly predictable from year to year, leap seconds are announced only 6 months in advance. That's a concern for software designers, operators of satellite-based systems, and anyone else who relies on split-second communications. Six months isn't much warning for engineers who operate computer programs or types of equipment that require precise time information and are intended to last for at least a decade. Some operations, such as the Global Positioning System, use custom time scales that eschew leap seconds entirely.

A glitch in inserting a leap second, these researchers say, could throw everything off, whether it's the timing of an international business deal, the location that a missile hits, or the star that the Hubble Space Telescope observes. "A 1-second hiccup in the phasing of North American power grids would likely cause a hemispheric blackout" notes Daniel Kleppner, director of the Massachusetts Institute of Technology-Harvard Center for Ultracold Atoms in Cambridge, Mass., in the March Physics Today.

Inserting a leap second "is a little bit like walking along the San Andreas fault," comments Tom Van Baak, a self-described precision-time hobbyist from Bellevue, Wash. It's typically an innocuous experience, but there's always the potential for catastrophe lurking beneath the surface.

With Earth continuing to grow more sluggish, scientists note, leap seconds will have to be introduced more and more frequently. "Eventually, you get to the point that the paradigm involved in this won't work," says Dennis McCarthy, a time specialist now retired from the U.S. Naval Observatory. "You've got to do something different. The addition of leap seconds is going to be an increasing nuisance for people who are counting on a time scale where a minute actually contains 60 seconds."

That's why a group of U.S. time-communication specialists, part of the International Telecommunications Union, proposed in 2004 to do away with leap seconds altogether. Let atomic time be out of whack with Earth rotation-based time, these scientists say. Their proposal is now under review by a working group of the union.

But to many astronomers, doing away with the leap second is anathema. It would be a major headache for hundreds of observatories to keep track of the heavens using a time measure that no longer had anything to do with the Earth's rotation, says astronomer Steve Allen of the University of California, Santa Cruz.

Then, there's the philosophical objection. For thousands of years, he notes, a clock was set by where the sun was in the sky. It was morning when the sun rose, noon when it was directly overhead. If your clock didn't agree with that phenomenon, you reset it.

If leap seconds were abandoned, noon atomic time might eventually correspond to sunset on Earth. It all boils down to "what should a clock tell you and what it [traditionally] has told you," Allen notes.

So, whose time is it anyway?

SLUGGISH EARTH A variety of competing effects, including the moon's tug on the oceans and the melting of glaciers, combines to slow Earth's spin. A day now is about 0.002 second shorter than it was a century ago. Some 150 million years ago, dinosaurs had to jam a full day of foraging and killing into what is now only 22 hours.

Observations of solar eclipses and comets recorded over the past 4,000 years provide graphic evidence of the slowdown. Tracing these events back in time, modern astronomers can account for the locations of eyewitnesses only if Earth had been rotating faster in the past than it does today, McCarthy notes.

Although scientists more than a century ago discerned Earth's sluggishness, the rotation of the planet remained the unbeatable standard against which all mechanical clocks, from the pendulum to the marine chronometer, were set. The second was defined in terms of Earth's rotation in 1900, as 1/86,400 the length of the average day, as indicated by when the sun set.

In 1949, physicists developed the first type of atomic clock. Still one of the standards, this clock is based upon a transition between two closely spaced energy levels in cesium atoms. The transition occurs at a frequency of 9,192,631,770 cycles per second. That frequency is accurate to 2 nanoseconds per day, so it provides a fundamental measure of time.

Another type of atomic clock, based on the coherent excitation of large numbers of hydrogen atoms, is even stabler than the cesium clock on time scales of about a week, though less stable on longer times.

Both types of atomic clocks are used to determine the international standard. It relies on comparisons among clocks at 55 locations worldwide, including about 100 clocks located at the U.S. Naval Observatory in Washington, D.C., and at its facility at the Schriever Air Force Base in Colorado Springs, Colo.

The biggest challenge in keeping Earth time and atomic time in harmony is that Earth doesn't decelerate steadily. While the friction generated by the sloshing back and forth of tides dominates the braking action, other, smaller effects work in the opposite direction.

One confounding factor, notes McCarthy, is the melting of glaciers since Earth's last ice age. Under gravity's influence, the water from melting ice flows away from high-altitude regions and packs additional mass onto lower-lying regions. The flow of material from high to low elevations causes a tiny increase in Earth's rotational speed. Uncertainties in predicting ice melts add to the difficulty in predicting the planet's spin.

THE CORRECTION To determine exactly how Earth's rotation varies from week to week and month to month, NASA and the Naval Observatory use a network of radio telescopes to precisely locate some 600 quasars, the brilliant beacons at the centers of distant galaxies. Because Earth is slowing down, the positions of the quasars appear to shift ever so slightly over time. The telescopes can discern the shift over a period of weeks.

Using these data, the International Earth Rotation and Reference System Service determines when the slow-down is large enough to warrant the introduction of another leap second. The service won't decide until June whether timekeepers will need to insert a leap second at the end of 2006.

It's the unpredictability of leap seconds that creates the potential for problems, says McCarthy. People "want to be assured that there's a uniform time that they can make use of," he says.

One of the issues facing timekeepers is that there's no standard way to insert a leap second. Although the extra second is usually inserted as the 61st second of a minute, some software and some digital clocks don't implement the leap second in that way, notes McCarthy. Some clocks either go blank for a second, read the 60th second twice, or stay at zero for 2 seconds. Differences in adding the leap second increase the likelihood of errors or confusion.

"If I were a communications company and wanted to make sure I never got bothered [with a leap second], I'd create my own sort of internal time scale," McCarthy says. "Then there's a concern that if everyone started doing this, there'd be a [complete] lack of standardization."

Banks, armies, or any group of institutions depending on close coordination could start acting "like a dysfunctional family," says Allen.

FUTURE TIME McCarthy and other scientists propose a compromise in the leap-second debate: Continue to make corrections but at longer intervals. The challenge would be to balance the inconvenience of the two types of times drifting apart with the chance that the adjustment would introduce glitches in time-sensitive communications.

One suggestion is to insert a bundle of leap seconds in official worldwide time only once every decade and to give everyone at least several years' advance warning of the total. If it turns out that the guesstimate of Earth's rotation-time change were wrong, an adjustment could be made in the next go-round.

Another possibility would be to avoid any change in standard atomic time until the disparity between the clocks and Earth's rotation becomes, say, an entire hour. That wouldn't happen for another 400 years.

In the meantime, a new problem looms. The inextricable link between gravity and time becomes increasingly apparent as atomic clocks become more and more precise. Every decade since the mid-1950s, the accuracy of atomic clocks has improved tenfold, notes Kleppner. The clocks are approaching an accuracy of 1 part in [10.sup.16], and newer systems, based on the vibrations of laser-cooled atoms and ions, are expected to eventually attain 1 part in [10.sup.18].

Einstein's general theory of relativity predicts, and atomic clocks have confirmed, that clocks at higher elevations run slightly faster than do those closer to the ground. Given the current accuracy of clocks, this gravitational effect requires that researchers know the altitude of timekeeping laboratories to within a few meters. Ultimately, altitudes would have to be measured to within a centimeter.

That becomes tricky because gravitational theory dictates that the altitude isn't measured relative to average sea level, but to the geoid, a hypothetical surface that approximates the shape and size of Earth. The geoid's size fluctuates in response to, for example, ocean tides and the redistribution of water due to climate changes.

These "shakes and shimmies" would make comparisons of future, ultraprecise atomic clocks kept at different locations "no more meaningful than comparing the rates of pendulum clocks on small ships scattered in the oceans, each bobbing in its own way and keeping its own time," says Kleppner.

An alternative would be for nations to agree to define the second on the basis of clocks at just one terrestrial location. But the politics involved in such a decision could make this unrealistic, Kleppner adds.

"We may be making clocks that are more precise than time itself can be defined on Earth,' says Kleppner. The timekeeping of such clocks, he concludes, would be too good to be true.