This sum, which is the mechanical energy of the system, is therefore
an invariable quantity in all the states to which it may be brought by
the interaction of its various parts, and the word energy well
expresses a capital property of this quantity. For if two systems are
connected in such a way that any change produced in the one
necessarily brings about a change in the other, there can be no
variation in the characteristic quantity of the second except so far
as the characteristic quantity of the first itself varies--on
condition, of course, that the connexions are made in such a manner as
to introduce no new force. It will thus be seen that this quantity
well expresses the capacity possessed by a system for modifying the
state of a neighbouring system to which we may suppose it connected.

Now this theorem of pure mechanics was found wanting every time
friction took place--that is to say, in all really observable cases.
The more perceptible the friction, the more considerable the
difference; but, in addition, a new phenomenon always appeared and
heat was produced. By experiments which are now classic, it became
established that the quantity of heat thus created independently of
the nature of the bodies is always (provided no other phenomena
intervene) proportional to the energy which has disappeared.
Reciprocally, also, heat may disappear, and we always find a constant
relation between the quantities of heat and work which mutually
replace each other.

It is quite clear that such experiments do not prove that heat is
work. We might just as well say that work is heat. It is making a
gratuitous hypothesis to admit this reduction of heat to mechanism;
but this hypothesis was so seductive, and so much in conformity with