
TAKING THE DENSITY OF VAPOUR OF VOLATILE LIQUIDS. 443 
The cause of the increased coefficient of expansion at temperatures approach- 
ing those of condensation appears to be, that the atoms of gases, under these 
conditions, approach each other too nearly, so as to come under the influence of 
cohesion.* If this cause be the true explanation, the tendency of the atoms of 
vapours to cohere at low temperatures may be combatted in two ways—either 
by diminishing the pressures to which they are subjected, or by separating the 
atoms by mixture with another gas of a more permanent character. 
Various experiments on vapours have been made at diminished pressures. 
ReEGNAULT} has examined the vapour of water under these conditions, and BrnEAu 
has determined the vapour densities of acetic and formic acids at pressures much 
below that of the barometric column. Indeed, Gay-Lussac’s method of taking 
vapour densities is, to a certain extent, one of estimating them at diminished 
pressures, as there is generally a considerable column of mercury within the 
apparatus to be deducted from the column of mercury in the barometer. The 
practical objection to the correction of the error by diminishing the pressure is, 
that slight diminutions of it seem to exercise little effect in counteracting the 
effect due to the cohesion of the particles of gas on approaching to the condition 
of liquefaction, whilst the working at extremely low pressures is surrounded with 
great experimental difficulties. 
The second remedy to which we have alluded, viz. the separation of the par- 
ticles of the vapour by a considerable excess of a permanent gas, having in itself 
a normal coefficient of expansion, such as hydrogen or air, appeared to be one 
which promised good results, and has chiefly engaged our attention—especially 
so, because it offered a means by which both the difficulties to which we have 
alluded might be overcome—viz., the danger of incipient condensation on the 
surface of the containing vessels, and the danger of contraction by the action of 
cohesion when the particles of gas approach each other too closely. The first 
danger can be wholly avoided by always having an excess of gas, so that it does 
hot become saturated with the vapour under examination. The second danger 
is also greatly reduced by the admixture of the gas; for we shall show, by expe- 
rimental evidence, that, although the coefficient of expansion of a vapour taken 
under ordinary circumstances is higher than that of air, it undergoes an alteration 
when the vapour is mixed with hydrogen gas—this alteration making it, if not 
identical, at least approach closely to that of air and other non-condensible gases. 
Before we proceed to describe the peculiar apparatus which we use in deter- 
mining the vapour densities of bodies which are either decomposed at their 
boiling points, or which it may be desirable to examine at lower temperatures, 
* The researches of Recnavutt (Annales de Chim. et Phys., v.), showing that at diminished pressure 
gases have very nearly the same expansion-coefficient, whilst. at increased pressure considerable 
divergency is observable, lends experimental support to the hypothesis we have adopted. 
+ Reenavtt, Annales de Chim. et Phys., xv. p. 146. 3™* séries (1845). 
