A test has been made of the accuracy with which an ice‐water system establishes the equilibrium temperature. A device consisting of a bulb containing ice and water with provision for volumetric measurement is used as a thermometer. This was placed in the re‐entrant well of a chamber similar to those used to establish the triple point. The temperature established by the well chamber was varied by varying the pressure of the water in it. It was found that when the temperature was made to differ from the equilibrium temperature of the thermometer by as little as 3×10−6°C, growth or melting occurred in the thermometer.