: 
692 Scientific Proceedings, Royal Dublin Society. 
great number of crystals have been determined, it so happens that 
few of these are available for the purpose of testing this question, 
as most of these substances are not such as readily yield crystals 
from fusion." 
The conductivity of ice has been determined by G. Forbes to 
be greater in the direction of the hexagonal axis than at right 
angles to it, in the proportion of 22 to 21. Under favourable 
conditions even this slight difference would appear to be sufficient 
to determine the arrangement of ice-crystals, so as to have the best 
conducting axis parallel to the flow of heat. Mr. M‘Connel’ and 
others have drawn attention to the curious hexagonal arrangement 
of vertical crystals to be observed in the ice formed on a sheet of 
water under particularly calm conditions. 
On the other hand, bismuth, which crystallizes in the same 
system, was found by Jannetaz to have a conductivity less in the 
hexagonal axis than at right angles, in the ratio of about 2 to 3. 
It is not difficult, as is well known, to get crystals of bismuth 
formed at a surface through which heat is passing out, by allow- 
ing a vessel of the molten material to cool down a certain amount, 
till crystals have formed over the sides of the vessel, and then to 
pour the rest of the still molten bismuth out. The crystals so 
obtained are rhombohedrons, and are seen, in the great majority 
of cases, to point with one corner inwards. The rhombohedron is 
so nearly a cube that it is only by exact measurement it can be 
ascertained whether it is the corner at the hexagonal axis (which 
would not agree with the above supposition), or one of the others — 
which would permit the maximum rate of dissipation of energy. 
Several of these crystals were measured with the goniometer, 
and in each case the corner pointing inwards was that agreeing 
with our theory. 
Tt is obvious that further observations are most desirable — 
before accepting any conclusion of this character, but this can 
only be effected on the accumulation of further thermal con- 
ductivity data for crystals. 
1 The author hopes, when opportunity offers, to make determinations of the relative 
conductivities of certain crystals, such as, say, sulphur, &c. 
2 Nature, vol. xxxix. 
a 
