92 TRANSACTIONS LIVERPOOL BIOLOGICAL SOCIETY. 
proteid. The amount of residue in the single samples was 
rather small, so those for each month were mixed together and 
the composite samples so formed were estimated (as shown in 
Tables II-III). In all cases 0-500 gram of the dried residue was 
taken and digested in 20 c.c. of pure sulphuric acid, using potas- 
sium sulphate and a small globule of metallic mercury. A first 
complete series of estimations of the 1916 and 1917 samples 
was made using copper sulphate instead of mercury, but since 
several of these estimations appeared to be erroneous the whole 
series of analyses was repeated, varying the process in detail, 
using mercury instead of copper sulphate, and employing 
different reagents and standard solutions—with two or three 
exceptions the results were nearly the same, but the latter series 
of estimations is that quoted. 
The process was perfectly straightforward. There was 
very little frothing of the mixture of residue and acid, and the 
reduction to a colourless solution took about two hours, as a 
rule. Some difficulty was, however, experienced in the cases 
of the residues from the flask of salted herrings and the re- 
duction took the greater part of a working day. In two cases 
there was no precipitate of mercury sulphide on adding sodium 
sulphide solution to the diluted acid, and the cause was apparent, 
the residues contained from 20 to 40% of salt and a chloride of 
mercury was produced which volatilised during the reduction. 
Afterwards the residue was digested with sulphuric acid and 
potassium sulphate for about half an hour and then the mer- 
cury was added. In these cases the whole process then went 
normally. 
It was found advantageous to boil the diluted acid solution, . 
after adding the sodium sulphide, so as to get rid of H,S. This 
seemed to improve the end-point of the titration. 
Litmus was used for an indicator, and decinormal solutions 
of sulphuric acid and sodium hydrate were employed in the 
titration. 
