7 
. “2 
84 H. Ss. H. WARDLAW. 
- This final titration was always performed on the acid solu- 
tion while still nearly boiling, and although cold alkali was. 
added, no error worth considering could be introduced in 
this way as the amount of alkali used was only about I cc. 
and the amount of carbonate introduced by it into the hot 
solution, if any, would be very small. (b) What has already 
been said about the composition of the substances in the 
solution applies here too. 
All the errors set forth above about which we have 
definite information can thus only be of small amount, 
and of them three would tend to give high, two low results. 
The analysis of standard phosphate solutions showed us, 
however, that positive errors of considerable magnitude 
were invariably met with, and as all sources of error except 
those depending on the composition of the precipitate and 
the substances derived from it have been accounted for, we 
are forced to conclude either that the precipitate obtained 
has not the composition assumed, or that the reaction with 
the alkali does not take place according to the equation 
given above, or that the products formed are subsequently 
modified in such a way as to alter the amount of alkali 
combined with. These factors may also act simultaneously. 
It has been shown by Baxter, Kilgore, Baxter and Griffin, 
Chesneau, Hissink and van der Waerden, Lagers, Artman, 
and by other workers, that the precipitate of ammonium 
phosphomolybdate as prepared by them invariably contained 
an amount of molybdenum in excess of that required by the 
formula given by Hundeshagen, (NH,),.PO,.12 MoO, 
.2HNO,. It must be remembered that in preparing this 
substance Hundeshagen was careful to avoid excess of 
molybdic acid, whilst the precipitate obtained in the course 
of analysis is formed in the presence of considerable excess 
of molybdate. MHissink and van der Waerden, Richardson, 
and Artman have also shown that the presence of sulphuric. 
