At the Bureau International des Poids et Mesures (BIPM) in Sèvres, France, sits a cylinder of platinum-iridium alloy 39 millimetres high by 39mm wide. This small lump of metal, machined in 1878, is the very embodiment of the kilogram.

Though all the other units in the International System of Units (SI) have been redefined to be based on fundamental constants or atomic properties, the kilogram still takes its definition from this cylinder in Sèvres. The problem is, like any artefact, the kilogram must be dusted - and mass could be lost with every wipe.

"Over the years, several official copies have been produced and distributed to various national metrology offices," says Ali Eichenberger of the Swiss Metrology Office (METAS). "Although it is not yet possible to define the kilogram in an absolute way, modern technology makes it possible to compare different masses with microgram precision. There seems to be significant variation in the masses of the official copies."

An academic issue, one may think, but the mass of the kilogram is linked to other units, such as the ampère. An imprecise kilogram means imprecise ampère.

To fix the problem CERN is participating in a metrology project launched by METAS. The idea is to build an ultra-precise watt balance - an instrument that compares mechanical and electrical power (see box). Using the watt balance and its equations, it should be possible to relate the unit of mass to fundamental units and constants, such as the metre, the second and the Planck constant.

"One of the crucial elements of the watt balance is the magnetic circuit, which needs to be extremely stable during the measurement," says Davide Tommasini of the Magnets and Superconductors group in CERN's Technology Department. Tommasini is directly involved in the METAS project. "Using a correctly dimensioned 'magnet shunt' with a low Curie temperature, it is possible to drastically reduce the effects of temperature variation. The circuit must also provide a very homogenous magnetic field in the whole volume involved in the measurement." The magnet circuit will be assembled at CERN. "We expect the permanent magnet and the 'shunting' cylinder to arrive soon. We will then work on testing the performance of the circuit," says Tommasini.

"We are very happy that CERN agreed to take part in the project in the framework of its knowledge transfer activities," says Henri Baumann, a METAS physicist who launched the project together with Ali Eichenberger. "This measurement will lead to a significant improvement in the determination of the Planck constant. CERN theorists will be happy to know that!"