32 DISCOVERY REPORTS 



Six out of these twelve sub-samples give a range of variation from this mean of 

 from — 3 1 to + 5 per cent. 



Nine out of these twelve sub-samples give a range of variation from this mean of 

 from — 42 to + 53 per cent. 



The twelve sub-samples give a range of variation from this mean of from 81 to 

 + 1 1 1 per cent. 



Thus in the majority of instances the error is of the order of 50 per cent but may be 

 as high as 100 per cent or over. The mean variation is 41 per cent of the mean. 



Oithona 

 The numbers estimated from sub-sampling the St. 121 sample were as follows: 



Sub-samples 1/40 of whole sample 18,080 14,960 16,640 



1/50 „ ,, 15,100 18,700 12,200 



„ 1/60 „ ,, 17,640 16,800 12,800 



,, 1/100 ,, „ 16,200 14,600 15,200 



giving a mean of 15,660. 



Six out of these twelve sub-samples give a range of variation from this mean of 

 from — 7 to + 6 per cent. 



Nine out of these twelve sub-samples give a range of variation from this mean of 

 from — 7 to + 15 per cent. 



The twelve sub-samples give a range of variation from this mean of from — 22 to 

 + 18 per cent. 



The sub-sampling of the St. 41 D sample gave similar results: 



Sub-samples 1/40 of whole sample 4040 



3040 

 1/60 „ „ 3742 



4342 

 giving a mean of 3816. 



Six out of these twelve sub-samples give a range of variation from this mean of 

 from — 7 to + 9 per cent. 



Nine out of these twelve sub-samples give a range of variation from this mean of 

 from — 17 to + 11 per cent. 



The twelve sub-samples give a range of variation from this mean of from — 21 to 

 (13 per cent. 



The fractions employed in the sub-sampling are those most frequently used in 

 practice. No doubt over the whole survey somewhat larger variations will occur, but 

 we may take it that when large numbers are concerned the majority of instances will 

 have an error of about 15 per cent and a minority with as much as 25 per cent. When 

 very small numbers are concerned no reliance can be placed upon the figure ; it is merely 

 an indication of the presence of the organism in the sample ; this applies to all organisms. 



