The subject of this column is time.

Don't worry, I do not propose to become philosophical and tackle the central mystery: What is time? Thousands of years and the best human minds have been devoted to that question without providing a satisfactory answer. "Time is what prevents everything happening at once" is about as good as the weighty, 500-page analysis.

What I want to talk about is the way that we measure time, and our changing view of the need for accurate measurements. My father saw no point in having a watch with a second hand. His own pocket watch was an old-fashioned clockwork wind-up model that, depending on how recently it had been wound, was usually either a few minutes slow or a few minutes fast. Why fool with seconds, he said, when chances are you were minutes off anyway? Even the term, "second hand," is short for "second minute hand" and is a relatively recent addition to both our timepieces and our vocabularies.

His father's great-grandfather (about whom I know nothing whatsoever) surely had the more relaxed attitude of his era. I doubt that he owned a watch of any kind and daily schedules of work and play ran by the sun or by the chimes of churches and town halls. True, for certain activities, notably navigation, the accurate measurement of time had enormous importance, and led John Harrison in the 1760s to construct the world's first accurate chronometer. But here we are talking long periods, weeks and months, and the idea that it was important to measure times as short as tenths of a second would have seemed preposterous. What could possibly be accomplished in a tenth of a second?

If we mean accomplished by a person, the answer is still, not much. But you probably have in your home a computer that performs at least a hundred thousand operations in a tenth of a second, and you can bet that the number will be a factor of ten higher in another year or two. This raises the question, how do we measure short periods of time, not just tenths of a second, but thousands, and then millionths? We didn't need these a century ago, but we certainly need them now. The fastest computer circuits run at a trillion cycles a second, and must be synchronized at least that accurately. Will we ever need to be able to slice time into pieces much finer than that?

Before we can hope to answer that question, we must first deal with another problem. If we are going to measure time, we need an absolutely uniform and unvarying standard, something that will give the same value of a time interval tomorrow as it does today. Again, a century ago there seemed to be no problem. The absolute standard was provided by the Earth itself, turning on its axis at a uniform rate. It was a big shock to scientists like Lord Kelvin and Simon Newcomb to discover that the Earth was an unreliable timekeeper, varying unpredictably in its rotation rate by a few parts in a billion.

Could anything be more steady and reliable than the Earth, spinning sedately on its axis? It was realized by the 1940s that atomic clocks provide a more stable system for measurement than anything offered by astronomical measurements. In 1967, the second was defined as the time taken to make a specified number of transitions between energy levels of a particular element, caesium-133. That gave a standard for time measurement accurate to one part in a hundred billion.

Good enough? Well, certainly good enough for all practical purposes, but nowhere near adequate if we are working in the transient world of subatomic particles. For example, a free neutron decays to form a proton and an electron in the relatively long time of twelve minutes. But a muon decays to an electron in two millionths of a second; a neutral pion decays in less than one one-thousand-trillionth of a second; and all of these are listed, in the catalog of subnuclear particles, as "long-lived" or even "stable." Many transitional particles created in high- energy accelerator experiments have lifetimes much less than one trillion trillionth of a second.

We do not "measure" such particle lifetimes in the conventional sense, with micro-stopwatch in hand. The lifetimes have to be inferred using other methods. However, it is reasonable to ask if we will ever need to deal with such short times directly. Computers are the obvious candidates. They have gone from millisecond to picosecond switching speeds (from a thousand operations per second to a trillion operations a second) in half a century. Will the computers of the future reach attosecond speed (a million trillion operations a second) just as quickly, which means we would have them in another generation? And will we be seeing a trillion trillion operations a second, by 2150?

I'm reluctant to say yes. At a "mere" thousand trillion operations a second, the energy needed for circuit switching is enough to tear electrons out of atoms. However, I'm also reluctant to say no. Computer speeds and power have marched ahead uniformly for more than 30 years, doubling according to "Moore's Law" in a little less than two years in spite of all physical or manufacturing obstacles.

This raises one final question. Could progress in subdividing time into smaller and smaller intervals go on, in principle, forever? That will be possible only if time is itself a continuous variable. However, might there be some minimum, a "quantum of time" which cannot possibly be further subdivided?

The smallest time which ever occurs in today's physical arguments is known as the "Planck time" or the "Planck-Wheeler time," and is the place where the quantum nature of time might be revealed. Measured in seconds, it is a decimal point followed by 42 zeroes and then a one. This is an interval enormously shorter than anything we've encountered so far. Whether such a minute interval exists as a significant practical part of the scientific future is anyone's guess.

Meanwhile, the favorite clock in my house is like my father's watch. It lacks a second hand, has to be wound with a key, and depending on the day of the week it is a little slow or a little fast. Someday I'm going to get rid of that minute hand.

"Borderlands of Science" is syndicated by: