170 MR JAMES DEWAR ON THE 
in Plate XII. It held conveniently 100 grammes of water,and was inserted in the 
middle of a stout brass envelope, thoroughly exhausted of air, the whole being 
placed in the middle of a large cylindrical tin vessel (E), having an outer annular 
compartment. The tin vessel was filled with water, and a constant current 
from the town supply was kept circulating by means of a syphon through the 
outer chamber. To the thermometer a ring of thin sheet india-rubber was 
attached as a stirrer, and when it was removed before the immersion of the 
palladium was placed in the little tin tube represented at (B). Immediately 
before an observation the whole apparatus was moved below the stéam bath 
(C), and the palladium dropped in. This arrangement of the calorimeter is very 
convenient in a small chemical laboratory, where uniformity of temperature 
cannot be easily commanded. The constancy of the radiation in the calori- 
meter makes the correction for cooling very exact. 
The same arrangement of the calorimeter is employed for registering very 
small amounts of heat by placing in (A) bisulphide of carbon or chloroform, 
packing the middle and outer compartments of the tin vessel full of pounded 
ice, and covering the exposed surface with sawdust. The thermometer is now 
provided with a thin sheet copper stirrer, instead of the india-rubber one 
formerly used. The rate of cooling is in this case determined once for all for a 
range of 5° above zero, and plotted in a curve. From this curve the correction 
for the radiation is determined for all subsequent experiments. 
Two series of experiments were made. In the first series bar, and in the 
second plate, palladium was used, three different hydrides* of each. The 
experimental results are given in Tables II. and III. 
In the case of the bar the specific heat of the occluded hydrogen increased 
as the charge diminished, the extreme values being 3°79 and 5:05. Similarly 
with the plate the values are greater, and have a wider range, viz., from 3°93 to 
5°88. These results are calculated in each case for the excess of heat given to 
the calorimeter by the hydride in excess of that of the original palladium. But 
if a comparison is made between the different hydrides in both series of experi- 
ments, then the specific heat is found to exhibit no such regular increase as in the 
former series. In the bar the extreme values are 3°21 and 3°77, and the mean 
of the three results is 3-47. The plate, on the other hand, gives a range of from 
2°7 to 3:94, the mean value being 3°31. The amount of variation in the plate is 
very great as compared with the bar, and is clearly due to some secondary 
action taking place. 
The increase of specific heat for small charges of hydrogen, when comparison 
is made with pure palladium, is clearly due to some regular increase of the spe- 
* In this paper the term hydride is not used in its strict chemical sense, but as a convenient 
abbreviation. 

