IN STUDYING QUESTIONS OF CHEMICAL MECHANICS, 363 
actual case, in order to ascertain to which of those forms it should 
be referred, and to deduce from it the system of combination 
produced. 
50. I have not been able to preserve, without precipitation, 
the tartaric solutions in the combined conditions of a sufficiently 
low temperature, with a sufficiently small proportion of water, to 
observe directly the rotatory power [«] becoming negative by the 
predominance of the coefficient (A) over the term (B) e. But, 
guided by the preceding indications, I produced an analogous 
effect, by introducing into these solutions sulphuric acid, which, 
from its affinity for water, rendered it less free, without opposing 
an absolute obstacle to the state of liquidity. For, in an experi- 
ment made thus at the temperature of + 14° C., I succeeded in 
rendering the deviation of the red rays towards the right almost 
null, and in making that of the violet rays pass to the left; so 
that I cannot doubt that the deviations of all the rays, in such a 
system, would have become in like manner negative if they had 
been observed at lower temperatures. I obtained this complete 
inversion by adding to the tartaric solutions large proportions of 
solutions of potass, and also by dissolving in very small quan- 
tities of water tartrates of alumina properly prepared. But al- 
though the tendency of the tartaric acid to invert its power, when 
deprived of water, or when this is withdrawn from its affinity, is 
evident in these last pheenomena, it is complicated by the union 
which it contracts with the other bodies placed in its presence, 
so that the even partial inversion effected by the presence of sul- 
phurie acid affords a better demonstration. 
51. From the mean and constant value + 14°3154 which the 
coefficient (B) is found to possess in the purely aqueous solu- 
tions of tartaric acid, the right line represented by [«] forms 
with the axis of the abscissz e an angle of 86° 0! 15", so that it 
is almost perpendicular to it. For this reason, in fig. 1, Plate 
V., where we have represented it for a temperature at which 
the coefficient (A) would be negative, it was necessary to trace 
it under an imaginary inclination far less than the real one, in 
order to distinguish it sensibly from the axis of the[«]. This alte- 
ration is without inconvenience in a purely conventional graphic 
representation, since it amounts to the constructing the e and 
the [@] on a different scale of lengths; but the circumstance 
which renders it necessary, points out a physical peculiarity 
which it is very essential to examine. At the mean temperatures 
2D2 
