134 ROYAL SOCIETY OF CANADA 
being assigned in the ratio of 2 to 1 in favour of our own seven mea- 
surements, as compared with his one. 










Exp ter. Method. Kind of Ice. SA ook Calcul. Obs. 
Brunner | Double weigh.| Fresh surface 0°9180 + °000039 | High | + ‘0015 
ice 
ee Dilatometer | Interior Form | 0°91580 | + ‘000008 | Low — ‘0007 
an 
Geissler 
Kopp re se 0°9078 + ‘0007 Low — ‘0087 
Bunsen * Column 0°91685 | + ‘000035 | High | + ‘00035 
Nichols | Double weigh. Mantles 0°9163 — ‘00009 High | — ‘0002 
EMMA . Icicles 091807 | + -00004 | High | + -0015 
oe " Surface (fresh) | 0°91804 | .......... High | + ‘0015 

The results may in the main be said to agree very well with the 
theory advanced, but, although some of the observed phenomena seem 
strikingly indicative of the truth of this explanation, it can hardly be 
said at present to be more than a suggestion. It is the writer’s inten- 
tion to pursue the matter experimentally as soon as possible; some 
preliminary measurements would seem to indicate that artificial ice 
of the interior formation tends to gain in density when allowed to 
stand in an ice bath, but the experiments were too rough to be 
conclusive. 
Ice may undergo compression arising from other causes than 
unequal expansion and contraction, as when specimens are kept by 
being buried deep in ice houses, or are formed under a head of mercury 
in dilatometers. Thus, ground ice, which can undergo no strains 
due to unequal temperature, is always formed under a considerable 
hydrostatic pressure. However, the writer has purposely only treated 
the one aspect of the matter in this paper. 
With regard to the Bunsen Ice Calorimeter, the conditions which 
affect the density of the mantle would seem to be its thickness, the 
diameter of the tube around which it is formed, the thermal conduc- 
tivity of the same, and the temperature of the refrigerant; also the 
outer layers are probably less dense than the inner ones. This instru- 
ment would thus appear to have many disadvantages when used to 
obtain absolute calorimetric measurements. However, if the ice man- 
tles are formed by the use of a comparatively mild refrigerant, and 
allowed to stand before being used, much greater accuracy should be 
attainable. 
In conclusion, I wish to thank Professor Barnes for his kind 
supervision and suggestions in the preparation of this paper. 
