1878.] 
On Residual Phenomena. 
3i 
per cent, and in the presence of which such errors as are at 
present regarded as tolerable in the atomic numbers of even 
the best determined elements shall be considered utterly in- 
admissible — I think can this question be settled ; and when 
such gigantic consequences — so entire a system of Nature — 
are to be based on a principle, nothing short of such evidence 
ought, I think, to be held conclusive, however sedudtive the 
theory may appear.” 
The so-called atomic heats of the elementary bodies bring 
us in contact with another instance of residual phenomena. 
It has been observed that within certain limits the atomic 
weights of the simple bodies vary inversely as their specific 
heats, and that consequently the products of these two 
numbers, or the so-called atomic heats, are approximately a 
constant quantity. In very many cases the numbers thus 
found range from between 5*9, as in aluminium and rhodium, 
to 6*9, as in iodine. This is certainly a rather- wide devia- 
tion from the requirements of the law. But there are 
several sources of uncertainty which may be supposed to 
affedt the results of the calculation. The atomic weights 
themselves, as we have just had occasion to remember, are 
not yet ascertained with absolute certainty. The purity of 
the various specimens of the elementary bodies operated 
upon is in some cases open to doubt. The determinations 
of the specific heats themselves may not have been entirely 
free from experimental error, and the circumstances under 
which these determinations were made cannot be pronounced 
stridtly comparable. The experiments were performed 
within temperatures nearly the same in an absolute point of 
view, but not relatively identical — i.e . 9 not equidistant from 
the fusion-points of the bodies concerned. The difficulty of 
obtaining the elements in comparable conditions is no little 
increased by the fadt that not the temperature of a substance 
alone, but its physical strudture and its state of aggregation, 
have a modifying adtion upon its specific heat. 
But there are three elements, mutually analogous in many 
respedts, which show a much wider discrepancy from the 
law than can be explained by such considerations. The 
atomic heat for boron in the graphitic state is 2^59, in the 
crystalline condition 2*7 5. Carbon, as wood charcoal, gives 
the number 2 ‘go, as graphite 2*41, and as diamond 1*76. 
These figures, moreover, widely as they differ from what the 
law might require, are obtained by taking the atomic weight 
of carbon — 12. Silicon when crystallised shows the atomic 
heat 4*97, and when fused 4*90. It is obvious that these 
three elements cannot be made to harmonise with the pre- 
