In the earlier apparatus 10-liter samples were used, and the volume of the respiration chamber was so large that it was necessary to multiply the values found in the residual sample by a very large factor, 500. Hence, the utmost caution was taken to procure an accurate measurement of the sample, the exact amounts of carbon dioxide absorbed, and water-vapor absorbed. To this end a large number of corrections were made, which are not necessary with the present type of apparatus with a volume of residual air of but about 1,300 liters, and accordingly the manipulation and calculations have been very greatly simplified. While formerly pains were taken to obtain the exact temperature of the air leaving the gas-meter, with this apparatus it is unnecessary. When the earlier type of apparatus was in use there were marked changes in the temperature of the calorimeter laboratory and in the water in the meter which were naturally prejudicial to the accurate measurement of the volume of samples, but with the present control of temperature in this laboratory it has been found by repeated tests that the temperature of the water in the meter does not vary a sufficient amount to justify this painstaking measurement and calculation. Obviously, this observation also pertains to the corrections for the tension of aqueous vapor. It has been found possible to assume an average laboratory temperature and reduce the volume as read on the meter by means of a constant factor. The quantity of air passing through the meter is so adjusted that exactly 10 liters as measured on the dial pass through it for one analysis. The air as measured in the meter is, however, under markedly different conditions from the air inside the respiration chamber. While there is the same temperature, there is a material difference in the water-vapor present, and hence the moisture content as expressed in terms of tension of aqueous vapor must be considered. This obviously tends to diminish the true volume of air in the meter. Formerly we made accurate correction for the tension of aqueous vapor based upon the barometer and the temperature of the meter at the end of the period, but it has now been found that the reduction of the meter readings to conditions inside of the chamber can be made with a sufficient degree of accuracy by multiplying the volume of air passing through the meter by a fraction, _(h-t)/h_, in which _h_ represents the barometer and _t_ the tension of aqueous vapor at the temperature of the laboratory, 20° C. Since the tension of aqueous vapor at the laboratory temperature is not far from 15 mm., a simple calculation will show that there may be considerable variations in the value of _h_ without affecting the fraction materially, and we have accordingly assumed a value of _h_ as normally 760 mm., and the correction thus obtained is (760 - 15)/760 = 0.98, and all readings on the meter should be multiplied by this fraction.