and on the refractive Powers of different Gases. 299 
sand times less than their primitive volume?-in the same 
manner as takes place, for instance, in the composition of 
water.” 
The well-known phenomena of the prismatic dispersion 
of light, a dispersion which varies in general with the re- 
fractive powers of the mediums, are also connected with the 
chemical composition of bodies, and seem to indicate that 
their molecules have an action a little unequal over that of 
light: this same inequality may arise from a method of 
analysis analogous to that now developed, but stil] more 
subtle; and by observing the dispersion which takes place 
in liquids, from it we might infer what would take place in 
the component gases if their smallness of density permitted 
its effect to be appreciable. Thus we should have, for in- 
siance, the dispersive power of the atmospheric air accord- 
ing to that of the oxygen and the azote observed in the h- 
quid composed of these two gases. 
_ The labours of which we have now given an account are 
as applicable to astronomy as they are to chemistry. The 
refractive force of the atmospheric air, which is given by 
these experiments, is one of the most delicate elements of 
the theory of refractions. It is capable of being determined 
in two ways; by a great number of celestial observations, 
and by the direct and terrestrial method employed_by our 
authors. M.Declambre, to whom astronomy is so much 
indebted, has just published on this subject new researches 
for his solar tables according to the formule of M. de La- 
place; and the coefficient he has deduced from the compa- 
rison of more than 500 observations is that which has been 
employed in the Mécanique Céleste. But it happens, by a 
very remarkable coincidence, and well calculated to merit 
confidence, that one single terrestrial observation of our au- 
thors is sufficient to give the coefficient of M. Delambre: 
not one of them differs from it in any sensible measure ; and 
the mean result deduced from their aggregate would not pro- 
duce a difference of one-¢ighth of a second upon the position 
of the stars observed at 45° of altitude *. 
The 
* In order to give an idea of the progress of these experiments, and the 
Presa 
