133 
rated upon; if we abstract heat we cause the portion of 
matter to contract and slirink, and the bulk is increased 
and diminished in certain definite proportions ; and not only 
so, but the opposite holds good also, namely, that if we cause 
a portion of matter to shrink we must in doing so abstract 
heat from it, while if we increase its bulk in any way with- 
out adding more matter we must, willingly or unwillingly, 
add a corresponding portion of heat. I believe this is accepted 
as an axiom in physics by the best judges and accepted as uni- 
versally true. Formerly some substances, such as bismutln 
water between certain temperatures, &c., were quoted as 
exceptions to this rule, inasmuch as they expand in solidify- 
ing; but they are only apparent exceptions, for their aber- 
rant behaviour has been shown, as I think beyond doubt, 
to be a phenomenon of crystallisation. Speaking of the 
most notorious case, that of water, Tyndal says : “ The 
arrangement of the atoms of water when solid require more 
room than they need in the neighbouring liquid state. No 
doubt this is due to crystalline arrangement. The attract- 
ing poles of the molecules are so situated that when the 
crystallising comes into play, the molecules unite so as to 
leave larger interatomic spaces in the mass ; we may suppose 
them to attach themselves by their corners, and in turning 
corner to corner to cause a recession of the atomic centres. 
At all events their centres retire from one another when 
solidification sets in. By cooling then this power of retreat 
and of consequent enlargement of volume is conferred.” — 
Tyndalls Heat as a Mode of Motion, 107. These excep- 
tions then I hold to be no exceptions at all. There is one 
substance, namely indiarubber, whose behaviour is eccentric 
and not explainable in this way, nor at present, so far as I 
know, explainable at all ; but with this solitary exception, 
whose raison d’etre I have no doubt will be shown to be as 
consistent with the general law as those of substances such as 
water, bismuth, &c.> we may safely conclude from our experi- 
