98 TRANSACTIONS LIVERPOOL BIOLOGICAL SOCIETY. 
or some other substance present in very small quantity, that 
is oxidised. 
It was also suggested that the deficiency was due to the 
imperfect drying of the wet substance, but reflection upon this 
possible source of error does not support the suggestion. 
Finally there are possible errors in the estimation of nitro- 
gen by the Kjeldahl process. Now it is perhaps unnecessary 
to state that blank and control experiments were made. A 
piece of N-free Swedish filter paper was substituted for the 
proteid-containing residue, and the operations were then carried 
out in all respects as if an actual estimation were being made. 
Then distillations and titrations were made, using pure 
ammonium chloride instead of the diluted acid from the Kjeldahl 
flask. In both cases small departures from zero or the theo- 
retical percentage were obtained, but these sources of error were 
so small that it was not regarded as necessary to correct the 
results stated in the Tables. 
However, we see from Table IV, that the iin herrings 
show more satisfactory totals. The mean of 5 analyses is :— 
61-8 % water, 21-9 % oil, 14:9 % proteid, 1-1 % ash = 99-7 %, 
and it is therefore evident that we have to deal with a real 
deficiency, that is, there is, in the flesh of the Manx Summer 
herrings some substance not found by the usual poe of 
estimating fat or proteids. 
We discuss later what this substance may be. 
Differences between Summer and Winter herrings will also 
be noticed with respect to the percentage of ash recovered 
from the residues. The Manx Summer herrings of 1916 and 1917 
gave 2:33 and 2-75 per cent. respectively, while the Winter- 
caught shop and Welsh herrings gave 1-08. But this difference - 
is easily to be accounted for by the fact that the Manx fish were 
slightly salted. Thus some sodium chloride was added to the 
tissues, and some water withdrawn. The difference in ash in 
the two kinds of fish is about 15%. 
