72 
Proceedings of the Royal Society 
then, this number would give a rough explanation in admitting 
condensation in many of the metals. 
I have thought that it would be interesting to compare this 
volume with the volume of oxygen when it is combined with solid 
substances other than metallic, and to take a series of analogous 
combinations. For this purpose the chlorine family is well fitted 
in their respective combinations with potassium, and these with 
oxygen. The following table contains the best known density 
determinations and volumes of chloride, bromide, and iodide of 
potassium, compared with the densities of chlorate, bromate, and 
iodate. 
The total volume of the oxygen in chlorate of potash, on the sup- 
position the chloride of potassium retains its original volume in 
combination, is 15 ; whereas it is only 7 in bromate of potash, if 
we allow that the bromide of potassium retains its original volume ; 
and it appears to occupy no volume in iodate of potash, assuming 
that iodide of potassium maintains its original volume. The 
apparent disappearance of the volume of the oxygen, in changing 
iodide of potassium into iodate, is analogous to the apparent loss 
of volume of many salts in their water of hydration, the salt occu- 
pying the volume of the crystal water taken as ice, as pointed out 
many years ago by Playfair and Joule. It is clear that, in assum- 
ing the halogen compounds of potassium as retaining their primi- 
tive volume in their oxidised derivatives, we place these compound 
substances in the same position as the metals in the simple oxides. 
Now, we saw that in many oxides the volume of the oxygen 
varied, and that, in all probability, from metallic condensation 
taking place during the act of combination. The metals having 
the lowest density and the greatest atomic volume condense the 
most in combining. Generally speaking, if we examine the 
