On the outskirts of Paris, in a locked vault to which only three people have the key, lies a treasure worth more than its weight in gold.

It's even worth more than its weight in platinum and iridium, which is what it's made of.

The squat metal cylinder weighs exactly 1 kilogram, as it should. It is the world's definition of mass, the standard kilogram against which all others are judged.

But now "le grand K," as the kilogram is known, is putting itself out of date. Since it was cast in the late 1800s, it has changed mass ever so slightly, drifting by a few millionths of a gram per year when compared with six copies made at the same time.

And that just won't do, physicists say.

"It's scientifically very unsatisfactory to have a mass standard that changes in mass," said Paul De Bievre, a standards expert at the European Commission's Institute for Reference Materials and Measurements in Geel, Belgium.

It's time for a new kilogram standard, researchers say -- one that won't depend on the vagaries of a single chunk of metal. So physicists are striving to replace le grand K with a fundamental physical measurement that will last forever.

Scientists have done so for other important units of measure. A second, for instance, is defined as 9,192,631,770 periods of a flickering between two levels of a cesium-133 atom. A meter is the distance light travels in a vacuum during 1/299,792,458ths of a second. (It used to be the distance between two scratches on a certain platinum-iridium bar, which is still kept next to le grand K at the International Bureau of Weights and Measures, or BIPM, near Paris.)

To fix the kilogram, one group of physicists is trying to define mass based on voltage, resistance and other electromagnetic measurements. A second group wants to make a perfect sphere of silicon; by counting the number of atoms in it, the scientists hope to arrive at a new mass standard.

One of these ideas -- or both, or neither -- may replace le grand K in the next decade or two.

"At least, we can do no worse than it's been for the last 100 years," said Richard Steiner, a physicist at the National Institute of Standards and Technology in Gaithersburg.

Scientists are driven by more than just curiosity. They need a mass standard for use in their precision experiments. And adopting a standard kilogram is important for international trade, Steiner said. Even tiny discrepancies in how much a shipment weighs can add up to headaches for people trying to sell goods, he noted.

"The thing that unites us in the world is the definition of the units," said De Bievre.

The metric system, used almost everywhere in the world except in the United States, was born during the French Revolution in an effort to unify the many weights and measures used at the time. In 1875, 17 nations signed the Metre Convention and adopted the metric standards.

Today, 51 countries have signed on to the International System of Units, or SI, after its French name. It defines seven basic units: meter for length, kilogram for mass, second for time, ampere for electric current, Kelvin for temperature, mole for the amount of a substance and candela for luminous intensity.

Over the years, all units but the kilogram have received physical definitions that don't depend on an object. The kilogram is the last because it's not an easy thing to define.

It was first conceived as the mass of a cubic decimeter of water at 4 degrees Celsius. (A kilogram equals roughly 2.2 pounds.)

Today, the kilogram is whatever the mass of le grand K is. It's a squat metal cylinder, about an inch-and-a-half high and wide, made of 90 percent platinum and 10 percent iridium. That mixture is particularly stable and dense, making it a good candidate for a lasting mass standard.

In 1889, the keepers of le grand K made six copies, which are kept in the same temperature- and humidity-controlled vault as the original. Most of the countries that signed the Metre Convention have their own copy of the kilogram, which they occasionally send to Paris for calibration.

Le grand K comes out of its vault only once every few decades -- "only when there's a scientific reason to think that your uncertainty is too high," said Richard Davis, head of the mass section at the international standards bureau. It hasn't been out since 1992, when it was cleaned with a chamois cloth coated with solvents, then steamed with doubly distilled water. It probably won't come out again until it is replaced by a new standard, he said.

Nobody knows why the mass of le grand K fluctuates compared to its copies, but some scientists think it might be absorbing atmospheric contaminants. Even weighing it is a delicate job, as bumping it against the scale flakes off tiny pieces, Davis said.

The new efforts aim to replace the kilogram standard with one that would vary by less than 1 part in 100 million each year.

For one group of scientists, that means defining the kilogram as a certain number of atoms of a particular element.

The scientists work backward, starting with a 1-kilogram mass and determining how many atoms are packed into its volume. The task is like trying to figure out, just by looking, how many gumballs are in a gigantic spherical gumball dispenser.

"You essentially add up this huge number of atoms in the crystal just by knowing what the spacing between the atoms is," said Davis.

Once the scientists know how many atoms are in 1 kilogram, that number could redefine the kilogram standard.

The team works with the element silicon, fashioning spherical crystals a bit bigger than a billiard ball.

"All of this has to be done very carefully" because of the precision required, said De Bievre, one of the project's leaders. "No national institute can do all of what is needed."

Laboratories in Australia, Belgium, Germany, Italy, Japan and the United States share the task of creating and polishing 1-kilogram spheres of silicon, then measuring their physical properties in great detail.

For years, the project was foiled by naturally occurring holes that dotted the silicon crystal like missing gumballs in a gum dispenser. Only recently have scientists realized that the presence of such holes could explain why silicon spheres made by different labs appear to have different densities.

The delay did have one spinoff, said physicist Frans Spaepen of Harvard University: Computer chip manufacturers now know exactly how many holes riddle silicon crystals.

The next step for the team is to make a sphere out of pure silicon-28. Until now, scientists have used a mixture of the three naturally occurring silicon isotopes -- silicon-28, -29 and -30 -- which are forms of the element with different masses.

Scientists at the Institute for Crystal Growth in Berlin will soon make test crystals of silicon-28 to see how the material behaves. The project could have its kilogram replacement as soon as six years from now, said De Bievre.

By then, the competing technique may also have an answer.

This second device, known as a "watt balance," was invented in the 1970s by Bryan Kibble of Britain's National Physical Laboratory.

The apparatus works by balancing the force of gravity pulling down a 1-kilogram mass against an upward-pulling magnetic force. The device can indirectly define the kilogram because all the other units measured -- such as time, length, voltage and resistance -- are already precisely defined.

The idea "has a beautiful exactness about it," said Kibble.

But it's not so easy to pull off. Scientists have built two big watt balances -- a two-story one at the U.S. standards lab in Gaithersburg, and a room-sized one in England. Both devices work pretty well, but so far neither is accurate enough to do better than le grand K.

The team behind the British watt balance reported in 1988 that it had some encouraging results, which were confirmed a decade later by the group behind the American one. But now the British team has redone its experiment and reached a different conclusion.

"We're still not there," said Kibble.

The American team has also torn down and rebuilt its watt balance. Preliminary results are expected by the end of this year, said Steiner.

"Back in '98, when we agreed with them [the British lab], it looked real neat," he said. "That's why everybody is looking to see, once we've rebuilt our system, will we get the same number that we did four years ago?"

In the meantime, a laboratory in Switzerland is working on a smaller watt balance of its own. Finnish scientists are trying to devise a related experiment that uses magnetic levitation. And German researchers are modifying the atom-counting idea, by spitting atoms into a container one at a time and measuring them as they go.

It's not clear which, if any, of these new approaches will finally go to the BIPM as the kilogram standard.

Whatever the new kilogram standard turns out to be, it must hold true for the next 100, or 1,000, or 10,000 years for measurers. It mustn't depend on a single thing sitting in a vault in Paris. And it must be reproducible anywhere in the world.

"It doesn't matter; you don't have to have it at the BIPM," said Davis. "Anyone could have one."

Zeina J. Jabbour of the National Institute of Standards and Technology in Gaithersburg holds the official U.S. kilogram. Copies of the kilogram are sent to Paris for calibration with "le grand K" -- the standard against which all others are judged.