1904.] Cotton ly Water and ly Water Vapour. 243 



times at which these 6 values are obtained on the ascending and 

 descending slopes of the curve, and the fifth and sixth columns show 

 the corresponding values of t/r. All the and t values may be taken 

 as experimental numbers, having been found by careful interpolation 

 from the recorded readings. <f> is the result of actual observation ; but 

 T is from its nature not sharply defined, for the temperature always 

 remains sensibly constant at its highest point for a considerable fraction 

 of a minute. The values of r used in the tables, and shown at the 

 head of the second and third columns, were therefore obtained by 

 plotting the arithmetic means of ^ and 2 2 for values in the immediate 

 neighbourhood of <f> and then slightly extrapolating the curve so formed. 

 No appreciable error is introduced by treating this short portion of the 

 curve as a straight line, though in fact it is part of a rectangular 

 hyperbola. Practically the same T value is obtained by assuming that 

 it is the geometric mean of the t\ and t- 2 of any closely contiguous 

 value. The r so found is of course well within the limits prescribed 

 by actual observation. 



For the comparison of saturated air immersion with water immersion, 

 the figures in the fifth and sixth columns of Table VI should be read 

 together with the corresponding figures of Table VII. The differences 

 are such as would be accounted for by comparatively small experi- 

 mental error, and are indeed not greater than might occur in a similar 

 comparison of two saturated air tests. The results point therefore 

 to the identity of the effects, and so of the causes at work, in the 

 two cases. 



For a comparison of either case with the results indicated by the 

 Equation (III), the figures in the last four columns of Tables VI 

 and VII may be read with the calculated values shown in Table VIII. 

 It is evident that there are greater differences between the calculated 

 and found values than between the found values of saturated air and 

 of water immersion, and that the tendency in both cases is for the 

 smaller t/r to be in practice a little too small, and the larger tfr to be a 

 little too large for the theory, while their product still remains of 

 practically unit value, as indicated by the equation. Such differences 

 are somewhat more strongly marked when the experiments at the 

 lower bath temperature (18) are studied in the same way; but here, 

 again, the close similarity between the water and saturated air tests is 

 conspicuous. Whether the agreement between the equation values 

 and those of the experiments would be improved by elimination of 

 such errors as may arise from the short initial exposure of the cotton 

 to outside air, from the lag of the thermometer, or from other causes 

 (some of which may work in opposite directions), it is impossible to 

 say without further data. 



