64 
Proceedings of the Royal Society 
of an observation is almost exactly proportional to x. After giving 
to each result a weight proportional to x, the mean thus obtained 
was 
k = 0-134. 
This is given in terms of the units mentioned above, to which also 
Neumann’s and De La Rive’s results have been reduced. Other 
experiments confirmed this result when reduced in the same 
manner. 
In the course of these experiments some facts were noted, which, 
though not belonging exactly to the subject of this communication, 
are yet worthy of being recorded. 
1 st, A number of measurements were made in the temperature of 
salt and fresh snow, mixed in different proportions, with the 
following results : — 
4 parts (by weight) of salt + 1 
part of snow gives - 20°*3 C. 
2 „ 
>> 
„ + 1 
„ - 21-1 
1 „ 
)> 
» +1 
>j >> 
„ - 21-4 
1 
» 
>> +2 
>> >> 
„ - 21-66 
1 
>» 
„ -f 3 
„ - 21-72 
1 „ 
» T 
» + 4 
>> 
„ - 21-4 
2d, The blocks of ice formed were frozen in a cylinder, with air 
above and water below. These blocks were cut out by means of a 
chisel and hammer. In spite of the great force used, there was not 
the slightest tendency in the ice to split when thus compressed by 
the walls of the containing vessel, although the finest point split 
the ice with great ease when the block had been cut out. 
Having completed the investigation with respect to ice, it occurred 
to me to extend the same process by means of a slight modification 
to the conductivity of other substances. The method employed was 
as follows : — A tin canister of about 3 inches diameter was filled 
with a freezing mixture whose temperature was frequently read. 
This was placed above the substance to be examined, which itself 
was laid in a flat tin dish resting on supports in water cooled to 0° 
cent. If the substance examined were a powder or soft material 
like cotton wool, it was made to rise to the level of two pieces of 
glass laid on the flat dish. So soon as ice began to be formed, 
it was considered that the statical state of temperature was 
