Molecular physics, as I have already remarked, early claimed his
attention. Various papers on electrolysis of liquids, and on the
constitution of gases, have been the result. A very interesting paper
on "Heat and the Constitution of Elastic Fluids" was read before
the Manchester Literary and Philosophical Society in 1848. In it he
developed Daniel Bernoulli's explanation of air pressure by the impact
of the molecules of the gas on the sides of the vessel which contains
it, and from very simple considerations he calculated the average
velocity of the particles requisite to produce ordinary atmospheric
pressure at different temperatures. The average velocity of the
particles of hydrogen at 32 deg. F. he found to be 6,055 feet per second,
the velocities at various temperatures being proportional to the square
roots of the numbers which express those temperatures on the absolute
thermodynamic scale.

His contribution to the theory of the velocity of sound in air was
likewise of great importance, and is distinguished alike for the
acuteness of his explanations of the existing causes of error in the
work of previous experimenters, and for the accuracy, so far as
was required for the purpose in hand, of his own experiments. His
determination of the specific heat of air, pressure constant, and the
specific heat of air, volume constant, furnished the data necessary for
making Laplace's theoretical velocity agree with the velocity of sound
experimentally determined. On the other hand, he was able to account
for most puzzling discrepancies, which appeared in attempted direct
determinations of the differences between the two specific heats by
careful experimenters. He pointed out that in experiments in which air
was allowed to rush violently or _explode_ into a vacuum, there was a
source of loss of energy that no one had taken account of, namely,