94 On definite Proportions. 
siderably colder, 1 repeated the experiment with a part of the 
same oxide, which had remained on the sand. It now lost 9-66 
per cent.: another portion, left for a longer time, lost 10°8 per 
cent. by ignition, and the next morning, when every thing had. 
been completely cold for some hours, the loss amounted to 12°5 per 
cent. ; the weather being in the mean time very damp and rainy. 
It is impossible to determine which of these experiments ought 
to be preferred: and although this oxide, which approaches so 
nearly to the acids in its nature, when it retains the greatest 
portion of water, eontains itself nearly twice as much oxygen as 
this water, still we can form no very decided conclusion from this 
circumstance. 
It may be hoped that similar experiments on the water contain- 
ed in the oxides of titanium and tantalium, on the tungstic acid, 
and other similar substances, will lead us to a more general and 
decided knowledge of the relation of water to these bodies, which 
occupy a middle rank between acids and bases. 
But what are we to think of the attraction of a body for water, 
which retains this fluid so slightly as to be overpowered by its ex- 
pansive force at a moderate temperature? It must manifestly be 
the same power that causes filtering paper to attract moisture 
from the atmosphere, and sometimes to become heavier even 
during the operation of weighing it, and that exhibits the well 
known effects of hygroscopical substances. Whether this force 
differs only in degree from that which is characterized by regu- 
larly defined relations between the bodies that are united, or 
whether it merely consists in a modification of superficial attrac- 
tion, I do not venture to decide. With respect to the modifica- 
tions of the force of elective attractions, there still remains much 
for us to investigate . for example, in what the force which causes 
a salt to be dissolved in water, in quantities dependent on the 
temperatures, differs from the force by which the same salt is ca- 
pable of rendering solid a greater or less quantity of water of 
crystallization. Between these quantities there is no determinate 
and unalterable relation; for salts, which contain no water of 
crystallization at all, may be very soluble in water, for example, 
nitre, and muriate of potass ; while others, which contain it very 
abundantly, are completely insoluble, as, for example, subsalts 
of iron and copper, carbonate of magnesia, and some others, 
C. Combinations of Water with Salts. 
Water of Crystallization. 
Salts of potass. Inno salt of potass that I have examined, 
the sulphate, nitrate, muriate, nor tartrate, have I found a trace 
of water of crystallization. If these salts are finely powdered, 
and dried in the sunshine, or in any other moderate warmth, 
they 
