In a recent paper by Glazebrook, Bousfield, and Smith some doubt has been east on the accuracy of my absolute measurements of the specific heat of water by the continuous-flow calorimeter. It is stated that an error of as much as 6 parts in 10,000 might have occurred in the superheating of the oil-stirred platinurn-silver resistances from which the values of the electric heating current were obtained. On the other band, it is also stated that this error might have been as small as 2.5 parts in 10,000. It was with some surprise that I read this statement, inasmuch as the authors could not have been aware of the rapidity of oil circulation which I used. It has been shown by Osborne Reynolds and others that the heat loss from a surface immersed in a liquid moving in turbulent motion is directly proportional to the velocity of flow. The degree of superheating of a wire immersed in oil will depend then directly on the rate of circulation, bully realising this fact, the resistances which I used were designed by Prof. Callendar and myself to be immersed in oil which could be circulated with great rapidity, These resistances are described in our papers, where the illustration shows approximately to scale the relative sizes of the stirrer and resistances. The paddle was rotated at a high speed by a powerful water motor, and the oil was thrown down with such force that a considerable depression was made in the surface. The oil, thrown sideways, passed up around the bars wires, which were wound loosely on the mica frames. I do not know by what standard Glazebrook, Bousfield, and Smith decide what is "normal stirring" or "very vigorous stirring," but it was evident to me at ones that I must have bad much more rapid circulation than anything used by these authors. In testing the accuracy of my experiments special attention was taken of possible superheating, and tests of stirring were made at the time. The good agreement of the various determinations of the specific heat, made with such different values for the heating current and flow of water, make evident that no large superheating error could have existed without being detected.