1888.] Prafulla Chandra Ray Copier-Magnesium Group. 273 
Discussion of Results. 
The following table contains all the necessary data. Of course, 
my analyses do not prove that in every case the com]3osition of the 
crystals is really represented by the formulae which I have assigned 
to them. This no mere analysis can do, hut there is an unmistak- 
able and remarkable agreement between theory and experiment, the 
differences between which in no case exceed the range of unavoid- 
able experimental error. The conditions under which I worked 
sometimes necessitated my using rather small quantities for analysis, 
but as a rule in the case of the metal present in larger quantity, the 
weight of the substance estimated was about 1 decigram. 
In the majority of cases, I have not contented myself with the 
determination of one metal only, but I have estimated both. It 
stands to reason that the estimation of that metal which is the pre- 
dominant one should have the principal say in determining the con- 
stitution of the salt. For instance, in No. X. the ratio of metals 
Cu : Cd is 12 : 1, and therefore the estimation of the copper is 
more to be relied upon than the estimation of the cadmium. From 
Column II. it will be seen that in the case of cadmium the in- 
fluence of a possible error of weighing is about eight times greater 
than in that of the copper. As pointed out by Professor Pickering, 
the estimations of sulphuric acid, of the alkalies, and of the water of 
crystallisation, are of little moment in the determination of the con- 
stitution of these salts, and I have made these estimations in the 
case of one salt only, viz. 11a, zinc-manganese-ammonium sulphate. 
I. 
II. 
III. 
Mean 
Theory. 
Zn : Mn Zn : Mn 
9-2 2-1 
Zn =13-33 
13-20 
13-69 
13-41 
13-36 
10-93 
Mn = 2-56 
2-50 
2-53 
2-50 
4-61 
(NH^)^ 8-88 
8-92 
8-90 
9-01 
9-04 
(SOJ =48-10 
47-72 
47-91 
48-08 
48-27 
H,0 = 
(by difference) 
27-25 
27-05 
27-15 
100 00 
100-00 
100-00 
From this it will be seen how little the estimations of the con- 
stituents other than the two metals Zn and Mn are to be relied 
upon in determining the constitution of the salt. The differences 
