While the apparent volume of the air remains constant throughout the whole experiment, by the conditions of the experiment itself the absolute amount may change considerably, owing primarily to the fluctuations in barometric pressure and secondarily to slight fluctuations in the temperature of the air inside of the chamber. Although the attempt is made on the part of the observers to arbitrarily control the temperature of this air to within a few hundredths of a degree, at times the subject may inadvertently move his body about in the chair just a few moments before the end of the period and thus temporarily cause an increased expansion of the air. The apparatus is, in a word, a large air-thermometer, inside the bulb of which the subject is sitting. If the whole system were inclosed in rigid walls there would be from time to time noticeable changes in pressure on the system due to variations in the absolute volume, but by means of the tension-equalizer these fluctuations in pressure are avoided. The same difficulties pertain here which were experienced with the earlier type of apparatus in determining the average temperature of the volume of air inside of the chamber. We have on the one hand the warm surface of the man's body, averaging not far from 32° C. On the other hand we have the cold water in the heat-absorbers at a temperature not far from 12° C. Obviously, the air in the immediate neighborhood of these two localities is considerably warmer or colder than the average temperature of the air. The disposition of the electric-resistance thermometers about the chamber has, after a great deal of experimenting, been made such as to permit the measurement as nearly as possible of the average temperature in the chamber. But this is at best a rough approximation, and we must rely upon the assumption that while the temperatures which are actually measured may not be the average temperature, the fluctuations of the average temperature are parallel to the fluctuations in the temperatures measured. Since every effort is made to keep these fluctuations at a minimum, it is seen that the error of this assumption is not as great as might appear at first sight. However, the calculation of the residual amount of oxygen in the chamber is dependent upon this assumption and hence any errors in the assumption will affect noticeably the calculation of the residual oxygen.