282 Proceedings of Eoyal Society of Edinhurgh. [apb.il 2, 
In preparations XX. I have similarly been able to prove the 
same thing. The first three fractions have all the composition, 
2Fe . 3Mg . 5[S04 . (XH4)2S04 . GHgO]. The next salt which forms 
is 3Fe. 4Mg. 7[S04. (XH4)2S04. GHgO], and this came down as a 
whole in the fourth fraction (8), but with greater care it also 
might have been fractionated. The next salt which formed was 
4Fe . 5Mg . 9[S04 . (XH4)2S04 . 6H2O], and this I was able to collect 
in two fractions (e, ^). The next fraction (rj) appears to consist of a 
mixture of the two salts — 4Fe. 5Mg . 9[S04. (XH4)2S04 . 6H2O] and 
Fe . Mg. 2[S04. (XH4)2S04 . 6H2O]. The percentage of iron is greater 
than what is required for the former, hut it is less than what is 
required for the latter. This is a case of overlapping, and we have 
further instances of the same thing in XVIy and XYIII^. In 
preparation XVI. we have no fraction which agrees with the 
formula, 2Fe . 5Zn . 7[S04. (XH4)2S04 . 6H2O] ; hut as in preparation 
XVIII. we have undoubted proof of its existence, it must have 
been present in XVIy, mixed with the salt Fe . 3Zn. 4[S04 . 
(XH4)2S04.6H20]. 
In preparation III. as compared with preparations IV. and V., 
and in preparations XVI. and XVII. as compared with preparation 
XVIII., there are apparent anomalies. Why has the salt No. III., 
the ratio Fe : Cu = 4 : 5, and the salts for Nos. IV. and V. the ratio 
Fe : Cu = 2 : 3 1 And again, why have the first salts deposited from 
XVI. and XVII. the ratio Fe : Zn = 1 : 4, and the salt from XVIII. 
the ratio Fe : Zn = 1 : 3 ^ The answer to these queries probably lies 
in the fact that in the solutions, as originally made, the ratio of the 
metals must have varied slightly in the respective cases, as the 
salts were weighed out only very roughly. Further, the temperature 
may have been different in the two cases. I am far from denying 
that slight changes in the various conditions under which the 
crystals are formed may have a determining influence upon the com- 
position of the compounds separating out from solution, but main- 
tain that by proceeding cautiously, and especially, as above described, 
by taking advantage of spontaneous evaporation, I have succeeded 
in proving the existence of definite compounds of the class dis- 
believed in by Aston and Pickering. The reason why these 
investigators failed to obtain definite compounds seems to me to 
lie in their modus operandi, wPich necessarily brought about the 
