B1OT ON THE EMPLOYMENT OF POLARIZED LIGHT. 293 
studied under more simple circumstances, had previously pre- 
sented to me. The existence of such laws was not improbable, 
considering the almost immaterial sensibility of the agent em- 
ployed; and the hope of discovering them alone led me to perse- 
vere in the laborious task which their investigation required. 
2. I may be allowed to indicate here one of the points of view 
under which this kind of experiments seems to offer the greatest 
utility to science, and under which I propose to regard it spe- 
cially in what follows. Experimental chemistry is at present very 
rich in complex facts ; but these facts presenting, for the most 
part, only reactions already complete, throw no light upon the 
previous physical state of the material systems in which they are 
produced, any more than upon the mechanism of the molecular 
forces which determine them. This twofold knowledge would 
however be indispensable to connect hereafter chemical phzeno- 
mena with strict mechanical theories, just as the movements of the 
planets and capillary phenomena have been reduced to similar 
theories, by deducing the first from attraction varying inversely 
as the square of the distances, and the latter from attractive forces 
becoming insensible at sensible distances, in which respect they 
are analogous to affinities. We might, at the first view, hope 
to find bases on which to rest such calculations, in the succes- 
sion of definite products, proceeding by multiples of simple 
proportions, which the same material substances form on com- 
bining one with another, under various circumstances. But 
although this important remark is extremely useful to prac- 
tical chemistry, to decide the formation of these multiple com- 
binations, and even to predict, in many cases, their possibility, 
it offers no hold for mechanical calculus; first, because the 
diversity of circumstances may have essentially modified the in- 
dividual constitution of the combined substances, or that of the 
product resulting from their reaction, by changing the absolute 
quantity of imponderable heat which forms a part of it, that 
which would alter the supposed identity ; and secondly, because, 
even admitting this scarcely probable identity, calculation, which 
can be established generally only upon continuous data, could 
not deduce the forces from such intermittences, except by con- 
necting them by ideal conceptions supported only by isolated 
verifications ; and the chance of ascertaining in this manner the 
general mechanism of the reactions would be single, whilst the 
chance of error would be infinite. Thus, notwithstanding the 
importance of such a problem, no geometrician has hitherto 
