F. W. Clarke on the Atomic Volume of Compounds. 179 
Phosphates and arsenates have been compared by others, 
Where their values differ, that of the arsenate is the highest. 
Silicon and titanium. The acids of this group will be cited far- 
ther along. Fayalite, 2FeO SiO, 4° oe 2FeO TiO, 4:370 
a definite exception is found in ae 2KF Sif, ; 
sp. 2°6652, Stolba; and 2KF TiF,, 2080, Bodeker. At 
To 82°5 and 116°. 
These examples will suffice for the present, although I shall 
have occasion to cite others to illustrate another poin on 
comparing all the material which I have collected, I find the 
exceptions to be quite rare, although, perhaps, I have given 
them undue prominence here 
Second,. similar compounds of similar metals often have 
equal atomic volumes, even when those of the metals them- 
selves are unequal. The corresponding compounds of the iron, 
Of the rule ander sorters however, the most striking 
case hitherto adduced is that of the vitriols. As I have already 
stated, the sulphates with 7 aq. of Fe, Ni, Co, Zn and Mg, have 
equal atomic volumes, while the double salts of the same class, 
containing in addition to the above metals Cu and Cd, follow 
the same rule. similar regularity connects the anhydrous 
ete of most of these metals. But an equally remarkable 
series is formed by the chlorids of the same metals, which, of 
the formula MCl,, have, as far as they have been studied, 
atomic volumes Siok are very nearly equal. If we present 
these chlorids in tabular form, taking the average of their ob- 
served atomic volumes as the real value, we shall find that the 
variations from their average are wholly within the limits of 
experimental error. 
ae Sp. gr. 2°528, Filhol. At. vol., found, 50°2 ( Sp. gr. cale., 2-668 
mic 2°560, Schiff. 507 2°726 
CoC, “ 2937,P.&5. * we co YS 136 
CuCl, “ “ “ 44:0 oe te 2-895 
ZnO, (4 os ‘Bodghee. # “49-4 } Mean, 47°6.4- “ 9 9-857 
Pls; * 68 = ART ee 
Mg(l “ 2117,» “ “43-6 r+ 1996 
CdCl,“ 3.6204, Backer, " See hs oe ee Rae 
HgCl, “ 2, Karsten, “ be 5071 lL 4“ t 5 693 
Now, if we ri into account the difficulty of obtaining per- 
fectly 1 accurate determinations of sp.gr. for some of these sub- 
stances; and also bear in mind that for most of them only sin- 
