144 ROYAL SOCIETY OF CANADA 
For this purpose he prepared pieces of river ice, formed during 
very sudden and severe cold weather. This ice he found could be 
procured in sizes suitable for his experiments, without flaw or crack 
of any description. He devoted some time and attention to the pro- 
duction of artificial ice free from air, but was not successful in obtain- 
ing any sufficiently free from the flaws and cracks which usually 
characterize this kind of ice. 
The weighing was first conducted in air, corrections being made 
for air displacement. Then the specimen, which was suspended from 
the arm of the balance by a single human hair, was immersed in refined 
petroleum oil. The specific gravity of this oil was determined with 
the utmost accuracy, the method employed being that of weighing 
in air and in the petroleum a piece of glass, of which the coefficient 
of expansion had been determined very carefully. The loss of weight 
undergone by the specimen of ice when immersed in this liquid afforded 
the data for the computation of its density, which was found to be 
0-9180 + -000039. 
It is worthy of note here that this series of experiments, while 
involving the determination of the ice density, was not undertaken 
for this especial purpose, but rather for the refutation of a statement 
that the density decreased with a decrease of temperature. This, 
needless to say, was entirely disproved, the linear coefficient of expan- 
sion obtained being 0-0000375. (Experiments carried on from —1° to 
—20° C.) 
After Brunner, the next experimenters of note to undertake 
determinations of this quantity were Pliicker and Geissler. The 
principle employed by them was essentially that of Robert Boyle, to 
whose experiments reference has been made, the method, of course, 
being subjected to refinements not attempted in the earlier experi- 
ments. 
The instrument employed was a dilatometer of exceedingly deli- 
cate construction, shown in fig. 2. First, the instrument was completely 
filled with mercury, after which water was introduced through the small 
opening at C; when the inner bulb B had been almost completely filled 
with water, the inlet C was sealed off in a flame. 
The introduction of the water into the instrument forced the 
mercury up into the capillary tube A, which had previously been very 
carefully calibrated. The instrument was now ready for use, and 
was placed in a bath at 0° C. and the height to the mercury thread 
noted; then the apparatus was transferred to a freezing mixture, com- 
posed of alcohol, cooled by ice and calcium chloride. As one would 
expect, the walls of the instrument gave way under the strain when 
congelation took place ; however, the mercury was forced up in the 
