on Atomic Volume and Specific Gravity. 531 



whatever (anhydrous or hydrated) is a multiple of 11, or of a 

 number near to II, or a multiple of 9*8 (the atomic volume of 

 ice); or again, the sum of a multiple of 11 or of 9'8. 



This law appears to us to resemble very much the prece- 

 ding, except that the indecision as to the choice between the 

 multiples of two different numbers renders it still less pro- 

 bable. 



We do not in this case meet in the same degree with the 

 objections above set forth for the case of dissolved salts; the 

 temperature cannot cause any great variation in their density, 

 and the experiments were made at temperatures varying too 

 little to have any separate influence, — if it be admitted, which 

 however is far from being proved, that with respect to the 

 solid bodies their densities should be compared at the same 

 temperatures. 



We will however make one remark relative to the process 

 by which the densities were determined ; it appears to us little 

 suited to give accurate results. It is not stated what was the 

 volume of the liquid to which the salt whose density was to be 

 determined was added ; but as it was the same apparatus which 

 had served for the preceding experiments, we may suppose 

 that it contained at least 1000 grains of water. The quantity 

 of salt employed in each experiment was from 40 to 60 grains, 

 and there thence resulted an increase in the volume of the 

 liquid corresponding to about 20 to 40 grains of water; in a 

 great number of cases indeed we find an increase of only 10 

 to 20 grains, that is to say, of from 1 to 2 per cent, of the total 

 volume. It is evident that by this process it is extremely dif- 

 ficult to avoid serious errors produced by the slightest varia- 

 tions of temperature, which tend to alter the volume of so 

 large a liquid mass, and of errors probably still more import- 

 ant, which might result either from the solution of a portion 

 of the salt in the liquid, if this was not accurately saturated, 

 or from the precipitation of a portion of the salt contained in 

 the liquid, if it were more than saturated. The experiments 

 of Gay-Lussac prove indeed that both these circumstances 

 may readily occur. 



These causes of error might perhaps be neglected if the vo- 

 lume of the liquid were very inconsiderable; but when, on the 

 contrary, its proportion is so large relatively to the solid salt, 

 they become too serious for any confidence to be placed on the 

 densities obtained by this process. 



We should add, that on reviewing the formulae which 

 Messrs. Playfair and Joule have established, based on the pre- 

 ceding law, they do not appear to us to indicate any very great 

 probability for this law. Along with certain analogies which 



