The action of the katharometer as an instrument for gas analysis depends essentially upon the thermal conductivity of the gas mixture examined. One method of calibration for a given pair of gases is to make a number of mixtures of known composition by volume, and to obtain from them a curve showing how the deflection θ of the galvanometer in the bridge circuit of the instrument depends upon the composition of the mixture which surrounds one of the fine platinum helices. By reference to this curve, any other mixture of the two gases can be analysed when its deflection is known. A typical calibration curve is shown in fig. 1, which is for mixtures of helium and argon. The direction of the galvanometer deflection depends on whether the gas is a better or worse conductor than air. A useful convention is to regard the deflection for gases which are better conductors than air as positive. Daynes has examined the nature of the heat losses in the katharometer cell, which are due to ( a ) radiation, ( b ) convection, ( c ) conduction by the gas, ( d ) cooling of the platinum helix by metallic conduction along the copper lead. Even at the highest temperature used in the katharometer, the effect of ( a ) is very small, and will not be influenced directly by the nature of the gas surrounding the wire. Experiments have shown that the effect of ( b ) is also small. The effect of ( c ) on the temperature of the helix is large, and will depend upon the nature of the gas. The effect of ( d ) is also considerable, and will depend upon the temperature of the helix. This effect will consequently vary with the nature of the gas under examination, but the magnitude of the effect in the steady state which is reached depends upon the effect of ( c ) (the small effects of convection and radiation will similarly depend upon the effect of ( c )). Thus in practice, the thermal conductivity of the gas controls the temperature of the helix, and the instrument will give the same reading for all gases or mixtures having the same thermal conductivity. Owing to the complicated design of the katharometer cell, it is not possible to make a simple calculation of the heat loss due solely to the conductivity of the gas, or to devise a method of converting katharometer readings directly into thermal conductivities. It should be noticed also that the calibration curves for different instruments are not all of quite the same form, owing to small differences in the winding of the helix, or of its position in the cell.