A high-temperature specimen chamber is described in which continuous gas circulation is employed in heating a specimen 1 in. diameter × 0 020 in. thick to a uniform and accurately known temperature. Using techniques sensitive to 0.1° C no temperature differences could be detected between the centre and periphery of the specimen in the range 20-200° C and the difference had a maximum value of less than 1° C at 600° C. In practice the true temperature of the specimen can be measured throughout the range 20-650° C. Temperature fluctuations due to the control circuit never exceeded ±0.5° C as measured on the specimen. The general reliability of the chamber has been critically examined. Slight modification to the instrument should raise the maximum attainable temperature to approximately 900° C.