1903.] RICHARDS—SOLVENT IN CRYSTALS. 31 
The results thus show a steady decrease in the amount of water 
as the powder becomes finer; hence each successive powdering 
must have opened new cells. 
When the figures are plotted a somewhat irregular curve is 
obtained, the study of which seems to show that further powdering 
would have but little effect upon the last-named amount of water 
held by the crystals. As a matter of fact, among many scores of 
determinations of the quantity I have never found in a pure speci- 
men of baric chloride less water than this. Is this limit then, 
observed in the very fine state of division, the true amount of water 
in the salt? This question is easily answered in the negative, for 
from the universally accepted atomic weights of barium, chlorine, 
oxygen and hydrogen it is easy to calculate that the theoretical 
amount of water is 14.744 per cent., an amount less by one part in 
a thousand than the lowest limit recorded above. 
Similar results were obtained from cupric sulphate, and it seems 
probable that any other salt would behave in the same way. Thus 
it appears that although powdering and drying will diminish the 
excess of solvent, it will not wholly remedy the error. It is prob- 
able that anything short of molecular division would fail to open 
the minutest enclosing cells, although the larger cells are broken 
up with comparative ease. 
The irregular shape of the curve drawn from the data given sug- 
gests that there is another cause of error superposed upon the effect 
just studied—an influence which grows in magnitude as the fineness 
of the powder is increased. A moment’s reflection serves to show 
what this new cause of error must be. 
It is well known that water adheres to or wets almost anything, 
except a few fats and oils ; and even these absorb or dissolve water 
to some extent. In consequence of this tendency to adhere, water 
is condensed or adsorbed in an invisible film, from the always 
slightly moist atmosphere, upon the exposed surface of nearly all 
substances. This adhering film increases the weight of these sub- 
stances. 
The extent of the adsorption varies with the nature of the sub- 
stance and in many cases is easily appreciable. With a given sub- 
stance it is of course dependent upon the pressure of the aqueous 
vapor and the temperature, as well as upon the extent of surface. 
Since the adsorption increases with the surface it is evident that 
fine pulverization, while tending to diminish the inclusion, will 
