Friday, February 29, 2008

The Science Of Leap Time

It took astronomers 5,000 years or so to figure out how leap years work - and with every technological leap, we're becoming increasingly dependent on ultra-precise timekeeping.
As a result, even tiny leaps in time are becoming just as important - and just as controversial - as leap days and leap weeks must have seemed back in the days of Julius Caesar and Pope Gregory XIII.

We have Julius and Gregory to thank for Friday's leap day, the extra day that's periodically tacked onto the month of February.

Even before Julius Caesar's reign, the ancients had figured out that a 365-day year came closest to matching the annual round of equinoxes and solstices, which were so important for planting schedules and holy rites linked to astronomical observations. But over the years, the seasons gradually fell out of sync, and extra days had to be stuck in haphazardly to put the year back on track.

With an assist from the astronomer Sosigenes, Julius Caesar instituted the much more reliable Julian calendar, which stuck a 366th day in the month of February every four years. (OK, when the Romans started the leap-year habit in 45 B.C., they added a leap day every three years. Which was bad. But eventually they got it right.)

There was still a slight discrepancy between the calendar cycle and Earth's actual solar year, however, and by the time Gregory entered the picture in the year 1582, the difference added up to 11 days. Armed with updated astronomical advice, the pope revised the calendar again, and added an exception to the four-year leap rule. Leap days would be added to century years divisible by 400 (such as 2000), but not to the other century years (such as 1900 or 2100).

And that's where we're at right now. Even Gregory didn't completely solve the problem. There are still little discrepancies that crop up between the calendar and our planet's yearly rounds. If you could check Earth's precise position in its orbit year over year, you'd find an average yearly discrepancy of 26 seconds.

John Lowe, leader of the atomic standards group at the National Institute of Standards and Technology's Time and Frequency Division, said that gap isn't close to requiring any correction ... yet.

"It's going to take 3,300 years for one day of slippage," Lowe told me today. "I doubt anybody is worrying about that right now."

So don't expect any additional tweaking to the Gregorian calendar until, say, the year 5000 or so. There is another leap-time issue that will have to be resolved much sooner than that, however. One of the biggest dilemmas facing timekeepers today has to do with the leap seconds that have been added to the year periodically since 1972.

Leap seconds aren't directly related to the length of the year, but rather to an infinitesimal difference between the length of a 24-hour day (that is, 86,400 seconds) and the international atomic standard for the length of a second. The difference (known in geekspeak as DUT1) accumulates because our planet's rotation is gradually slowing down.

Every time the gap between atomic and astronomical time reaches 0.9 seconds, a leap second is added to bring the two standards back into sync. Right now, the gap is almost 0.4 seconds - but Geoff Chester, an astronomer at the U.S. Naval Observatory, says the discrepancy could start picking up speed.

"Fifty years or so from now, it's entirely possible that we may be adding two or as many as four leap seconds per year," Chester told me.

That's very inconvenient for the folks who rely on the precise, steady timekeeping of the atomic system - such as the people in charge of the world's computer networks, power grids and even the Global Positioning System.

"Sooner rather than later, this is going to become a real nuisance, so there is debate right now over how we're going to reduce this nuisance factor," Chester said. "The simplest solution is just to do away with the leap second."

However, that doesn't sit too well with the folks who deal with timekeeping in the astronomical world. Without the leap seconds, the atomic time standard would run further and further ahead of natural time cycles.

"If we did away with leap seconds, we would find ourselves at some time in the same position that they did in the 1500s - one minute off, two minutes off, five minutes," NIST's Lowe said.

"All of the charts and tables stating sunrise and sunset, those would all have to be adjusted periodically to accommodate the slipping time scale."

Earth's time lords, at the International Earth Rotation and Reference Systems Service as well as at the International Telecommunications Union, have been talking about proposals to do away with leap seconds, or at least wait until the discrepancy adds up to a leap hour. Lowe and Chester say their organizations have no position for or against the idea, but will do whatever the international standard-setters tell them to do.

Lowe doesn't expect a resolution of the dilemma anytime soon - and in the meantime, there's likely to be more leap seconds added to the clock. "This has been going on for the last dozen years," he said. "Large international bodies like these never proceed too quickly."

But one thing is for sure: Every year, society is becoming more and more dependent on precision timekeeping. For Lowe and his colleagues, the most important job isn't whether an extra day is added to a year, or whether an extra second is added to a minute. Rather, it's to make sure that every second of the year is accurate and accounted for.

"We build these highly precise clocks not to define the time of day, but to define the length of a second," Lowe said. "That defines frequency, so many cycles per second - and frequency is what drives our technological world, from television and radio to our global telecommunication systems. The big satellites that carry massive amounts of data operate at a very high frequency, so they need very stable calibration.

"That's why we build these incredibly accurate instruments," he said. And that's why every second counts.

Update: I revised my reference to the 26-second annual discrepancy between the Gregorian calendar year and the tropical year (which adds up to one day in 3,300 years or so), because the actual discrepancy can't be judged on a year-over-year basis. For more about the so-called 4,000-year rule, check out this Web page.

Also, to celebrate leap day, you can either propose to your beau (if you're a woman) or try out the Project Leap Year Web site, which gives you the opportunity to share your leap-day dreams.

Alan Boyle covers the physical sciences, anthropology, technological innovation and space science and exploration for

No comments: