A copper cylinder was constructed and to it four layers of oaktanned leather were applied. The thermal gradient through the leather was measured by means of thermocouples on the cylinder and between the layers of leather. By suitable extrapolation of this gradient the temperature of the outside leather surface was determined accurately and constituted the standard surface temperature to which the readings of the surface temperature measuring devices were compared. Several types of surface temperature measuring devices were compared and evaluated on the basis of: (1) accuracy as determined by agreement with the standard surface temperature, (2) adaptability to use in a variety of experimentally produced environmental conditions similar to naturally occurring environmental conditions. It was concluded that: (1) The radiometric type of skin thermometer was the most dependably accurate instrument of those tested, although requiring corrections when used in ``sunlight.'' (2) Under normal conditions, the #40 gauge bare wire thermocouple was of the same order of accuracy as the radiometer and was only slightly affected by environmental changes. The accuracy of the coarser bare wire thermocouple was considerably more impaired by such changes. The thermocouple glued to the surface, or embedded in solder and taped to the surface, or mounted on copper mesh and tied to the surface was less accurate than the bare wire thermocouple. In general, increasing the effective mass of the thermocouple decreased the accuracy of the instrument under normal conditions as well as under the changed environmental conditions. (3) The Dermalor resistance thermometer was next in order. Errors were not greater than two percent under certain conditions but increased to four or five percent during radiation. (4) The errors in the disk thermistor readings ranged from −1.2°C to −1.9°C under the conditions studied. The accuracy of this instrument could be increased by calibration in situ against a radiometer. However, this would not obviate the error produced by infra-red radiation, for instance. This error amounted to −0.7°C. Other types of thermistors having much smaller contact surfaces are available and it is reasonable to expect that these instruments would yield readings as accurate as those of a thermocouple. However, they too would be subject to the errors exhibited by thermocouples or any other small instruments in actual contact with the surface. The feasibility of employing a thermistor as a micro-radiometer is under consideration. It presents special advantages to use in this manner in that thermistors can be constructed with a rapid response time, and as they have a high resistance they are particularly well adapted for electronic recording. (5) The Roll and Mold pyrometers were entirely unsuited to this measurement in the temperature range studied, showing errors of from 3°C to 7°C depending upon the experimental conditions. (6) Mercury-in-glass thermometers were unsuited to this type of temperature measurement because of the long contact time required which affects the temperature itself, and because of the marked impairment of their accuracy by environmental changes.