476 



Transactions. 



Discussion. 



The deductions of Part I are true for the further twelve samples 

 analysed this year, though it may be noted that in the two samples of 

 red cod the fat-percentage is practically the same (0-47), and the sample 

 with 82 per cent, water has 16 per cent, protein, while the other with 

 80 per cent, water has 18 per cent, protein. 



The following points are worthy of notice : — 



1. The analyses of two samples of whitebait are almost identical. 



In these analyses a large number of the fish was used, giving as a result 

 the average composition of that particular species. This eliminates to a 

 large extent the variations noticeable in different samples of the same 

 species of fish even when bought at short intervals. It indicates that in 

 order to obtain the average analysis of any species a sample should be 

 taken from the well-mixed muscle of as many fish as possible. That 

 method would eliminate (1) the difference due to the different metabolism 

 of the specimens taken, and (2), to some extent, the difference due to 

 varying richness of feeding-grounds. 



2. The analyses of the roes of groper and flounder show differences from 

 the general analyses of fish-roe given in various books : e.g. — 



The figures 23-95 and 24-88 in (3) and (4) include all nitrogenous 

 matter. In the case of the roe, the method of extracting the fat with 

 ether alone is not entirely satisfactory, and in future analysis of this class 

 of material it is intended to use Kosenfeldt's method (alternate extraction 

 with boiling absolute alcohol and chloroform). 



Some Experiments on the Seasonal Variation of Composition of 



Groper and Kingfish. 



In Part I of this paper reference was made to the variation in fat 

 content in different samples of the same variety of fish. It was suggested 

 that it was a matter, say. of metabolism or of seasonal variation. Further 

 work was done in this connection with the object of ascertaining what 



