36 



DISCOVERY REPORTS 



Rhincalanus gigas 



This large Copepod usually occurred in the N 70 V net samples in small numbers 

 when they were picked out and counted separately so that an error did not arise. But 

 when it occurred in large numbers the sample would be sub-sampled by the stempel 

 pipette. A sample known to contain 222 specimens was sub-sampled forty-eight times 

 by eight different methods; i.e. six times by each method. The numbers estimated as 

 being in the sample were as follows, according to the sub-samples taken: 



50 per cent of the sub-samples show a range of variation from the actual number in 

 the sample of from — 46 to + 49 per cent. 



75 per cent of the sub-samples show a range of variation from the actual number in 

 the sample of from — 77 to + 71 per cent. 



90 per cent of the sub-samples show a range of variation from the actual number in 

 the sample of from — 77 to + 98 per cent. 



The total of the sub-samples show a range of variation from the actual number in 

 the sample of from — 100 to + 143 per cent. 



The error with this large species is seen to be great. The pipette generally takes more 

 than a fair sample ; the mean of the estimated numbers is 287, whereas the actual number 

 was 222, an error of 29 per cent. The mean variation was 122, or 55 per cent of the 

 actual number present. In actual practice, when it was necessary to use the pipette for 

 this species, particular care was taken to see as far as possible that a fair proportion was 

 taken up, so that the error would not be so great as represented in this test, nevertheless 

 we must assume that the error was often as high as 50 or even 70 per cent. 



Chaetognatha 



These organisms, like Rhincalanus, are not adequately sampled by the pipette method, 

 and were nearly always picked out and counted separately. But occasionally when there 

 were a large number of small specimens they would be sampled by the pipette. A sample 

 containing 184 specimens was sub-sampled forty-eight times by eight different methods, 



