The most accurate thermometer in the known universe looks nothing like a thermometer. It is a copper vessel the size of a large cantaloupe, filled with ultrapure argon gas and studded with microphones and microwave antennas. The purpose of the gadget, which sits on the campus of the National Physical Laboratory (NPL) in Teddington, England, is not simply to measure temperature, however. Rather the device and others like it may allow scientists to completely overhaul the concept of temperature and recast it in terms of fundamental physics.

The plan rests on linking temperature to energy via a physical constant. Today the international standard temperature unit, the kelvin, is based on the properties of water, but scientists would like to bring it in line with other measurement units that have been liberated from the vagaries of the macro world. The second is now defined by the oscillations of a cesium atom; the meter relates to the speed of light in a vacuum. “It's bonkers that the kelvin doesn't directly relate temperature to energy,” says Michael de Podesta, who leads the research team.

The NPL device measures the Boltzmann constant, which links changes in energy to changes in temperature. De Podesta's team and its competitors hope to nail down the constant well enough to relate one kelvin to a certain number of joules of energy.

The new thermometer—technically an “acoustic resonator”—rings like a bell when the physicists feed certain sound frequencies into its microphones. From that sonic resonance, the researchers can determine the speed of sound within the gas-filled cavity and thus the average speed of the argon molecules—that is, their kinetic energy. In July, de Podesta's team reported in the journal Metrologia the most accurate measurement yet of the Boltzmann constant.

The current temperature definition makes use of water's phase changes. One key threshold is the so-called triple point, 273.16 kelvins, where water ice, liquid and vapor can coexist. In 1954 an international agreement defined the kelvin as 1/273.16 the difference between absolute zero and water's triple point.

The 1954 definition works well in general but begins to break down for extreme temperatures, such as those found within stars. “It only happened this way because people started measuring temperature long before they knew what it actually was, before temperature was known to just be atoms and molecules buzzing around,” de Podesta remarks. “Now that we know better and have the opportunity to correct it, we should.”