ACCOMPANYING BASIC SUBSTITUTIONS. 
29 
Salts of Silver. 
The nitrate is the only salt of silver I have examined, and the experiment was per- 
formed in order to ascertain the position of the oxide of silver as a thermal base. 
The full equivalent 42*48 grains decomposed by potash, gave 3°*88, or referred to 
the resulting liquid 4°*17, or to water 3°*95. Two similar experiments with half an 
equivalent gave nearly similar results, viz. 3°*90 and 3°*94, all corrections being made 
and the final results doubled to bring them to the ordinary standard. 
Sesguisalts of Iron. 
The bases hitherto examined being all of the form MO, it was of importance to 
determine how far the same principle would apply to bases of the form M 2 0 5 . The 
sesquisalts of iron appeared to be the best adapted for experiment, but it is difficult 
to procure them in a neutral state. The most certain method of effecting this object, 
is to pass a current of chlorine gas to saturation through solutions of the protosalts, 
expelling afterwards the excess of chlorine by heat. In this way a solution of sesqui- 
chloride may be easily procured from the protochloride, and a mixture of sesqui- 
sulphate and sesquichloride from the protosulphate. But as the resulting sesqui- 
compounds require one half more potash for decomposition than the protocompounds 
from which they are derived, it was necessary, in order to preserve the usual quantity 
of potash, to take only two-thirds of an equivalent of the latter. Accordingly 16*40 
crystallized protochloride of iron, converted into sesquichloride and decomposed by 
potash, gave, in different experiments, 3°*83, 3°*75 and 3°*74, which, all correc- 
tions made, correspond to 3°*97, 3°*89 and 3°*88. Of the crystallized protosulphate, 
23*00 grains, treated in the same way, gave 4°*09, 4°*11 and 4°*12, corresponding to 
4°*25, 4°*27 and 4°*28. These results, although not identical, present a sufficiently 
close approximation, particularly when the uncertainty of the original composition of 
the crystallized salts, and the difficulty of expelling without decomposition all the 
excess of chlorine, are considered. Another circumstance, whose influence it is diffi- 
cult to appreciate, but which will tend to modify these results, is this, that the preci- 
pitate obtained always contains potash united by so strong an affinity to the hydrated 
sesquioxide that the action even of hot water produces only a partial separation. It 
is very probable that this may occur to a different extent with different salts, and 
hence may be one source of variation in the thermal effects. 
On reviewing the foregoing results it will be observed, that while the effect of the 
substitution of potash for different bases produces thermal changes, varying from 
— 0°*34 to + 4°*28, the greatest uniformity prevails in those obtained with the salts 
of each separate base. It is true that in some instances slight differences do manifest 
themselves, but these differences, I apprehend, are in general not greater than occur 
in the chemical reactions. It is of course essential to a perfect uniformity of result, 
that exact equivalents of the salt and base should be employed, and that a complete 
