582 MR EDMUND J. MILLS’S 
were soon afterwards undertaken independently, but practically at the above 
temperatures. These gave 
Cubic Centims. fi Be ales ee ue ii 
115°33 5 16 
116-04 5 10 
115-48 4 04 
115-06 4 07 
After keeping for some time at the ordinary temperature, the zero was 114:02 
(probable error of 6 determinations -08), with a tendency to further contraction. 
The probable error of measuring the air was inconsiderable. 
In this instrument, 1° C. corresponds to about ‘3 c.c. It would therefore 
require to have its volume multiplied about ten times in order to be sufficiently 
delicate, especially if we have regard to probable error. It would then be too 
unwieldy to employ; and even were this not the case, its susceptibility to irregular 
zero variation is a fatal defect. 
The method finally adopted was the one ordinarily used, in which the air is 
not allowed to escape, but caused to remain at constant volume, and raise a 
column of mercury above a fixed point by its tension. The air-vessel is usually 
a glass flask; but there is some ground for objecting to a flask, on account of 
its allowing convection in its contents, and so giving rise to unsteady indications. 
I have therefore employed a glass helix. The bath in which this stands is of 
peculiar construction. It consists, as may be seen in fig. 4, of a thin iron pot 
a, placed within a larger one 6, which is equidistant from all parts of it. The 
inner pot is wholly enclosed by pure tin tubing, through which an air supply 
passes from outside the entire apparatus inwards to the bottom, where it bubbles 
up below the helix, and so acts as an excellent stirrer. The temperatures of 
the outer bath are integrated by the air which traverses the tin tubing. If a 
liquid be placed in both pots, a body in the inner pot can only change its tem- 
perature very slowly unless the operator chooses to introduce air, or to introduce 
it at arapid rate. If, while a moderate current of air is passing, the outer pot 
is heated for some time, the inner pot will be heated not quite so much; and 
if the source of heat be now removed, the cold air entering will somewhat cool 
the outer liquid. A time will thus soon arrive when the outer and inner vessels 
are exactly at the same temperature, and therefore in the best possible condi- 
tion for giving a steady heating effect. Using a very strong solution of sodic 
nitrate in both pots, I have succeeded in keeping a delicate mercurial ther- 
mometer almost stationary, for a quarter of an hour, at about 120°. The 
comparisons of the mercurial with the air thermometer were effected in the 
following way. The glass helix already referred to carried at its hinder ex- 
