256 
DR, H. T. BARNES ON THE CAPACITY FOR HEAT OF WATER 
different experiments, and all give identically the same temperature coefficient of 
variation. For the ice experiment on March 24, of Series I., the value of h is very 
nearly in agreement with the extrapolated value from Series I. and II. Hence the 
above method of treatment for this experiment would give an almost identical value 
of cl to the one obtained by eliminating the heat-loss from the two flows. 
It is important to notice that the value of cl in this experiment, obtained from the 
two flows, is more in agreement with the mean value of cl for the other ice experi¬ 
ments, obtained by extrapolating from the values of h in Series VI. than for the 
values obtained by eliminating the heat-loss in the usual way. This points to the 
fact that the values given above are more nearly correct than the values given in the 
tables for the same experiments. If this be so, the indication is, that the value of the 
specific heat of water rapidly approaches an exceedingly high value at 0°, and in a 
remarkable way substantiates the suggestions made by Howland in his memoir in 
regard to this. Further investigation is needed, however, in the neighbourhood of 
the freezing-point of water, before we can say that the specific heat of water 
approaches an infinite value as that point is reached. Such questions, as the con¬ 
tinuity of the curve for under-cooled water, render the idea quite unthinkable at 
present. In view of this uncertainty, I have adhered to the lowest of the values of 
the specific heat given by these measurements, and have consequently included them 
both in the tables and plot. Even in this case, the change of specific heat with 
temperature is very rapid, and no effect is shown by the observations taken below 4 C 
which would indicate a change at the point of maximum density. This, however, is 
not surprising when it is considered that the point of minimum specific heat in no 
way corresponds to the density curve for water. 
Unfortunately, only one complete set of observations could be obtained with 
Calorimeter D, with the device for getting rid of stream-line motion, owing to a crack 
which, shortly after, started in the fine flow-tube inside the vacuum-jacket, and 
admitted water into the jacket. This calorimeter is of sjiecial interest, as the vacuum- 
jacket was supplied with a quantity of phosphorus pentoxide. The value of the 
heat-loss is larger than for any of the other calorimeters, including calorimeter A, 
which we exhausted ourselves to a vacuum of at least '002 millim. of mercury. This 
*J 
indicated that the Po0 5 , instead of improving the vacuum as we at first thought, was 
really a disadvantage. The values of the heat-loss for the four calorimeters included 
in these measurements are, at 30° C. :— 
Calorimeter A . 1'8 millim. flow-tube . . . '0509 watt. 
„ C .... 2 „ „ ... '0500 
D .... 2'8 „ „ ... -0726 
„ E . . . . 3'1 „ „ ... '0600 
33 
