PLANKTON 251 



three chief groups of organisms, and also in the case of atypical 

 species of each, has been calculated, by means of a formula 

 for obtaining the probable error, with the following results. 



The total number of Diatoms on April 3 varied in the 

 six hauls from 3,880 to 10,020, the mean being 8,055. 

 Two of the hauls are below the mean and four above. The 

 smallest haul is 52 per cent, below the mean, and the largest 

 haul is 24 per cent, above. The question is : Do these 

 variations in the catch come within the limits of the probable 

 error of the experiment ? If we assume that the estimation 

 of the number of Diatoms in each haul is correct, then the 

 possible errors are those inseparable from all such collecting 

 at sea — sUght movements of the boat, unknown currents 

 in the water, irregularities in the verticality of the line, 

 etc. In this case of the Diatoms on April 3, the " probable 

 error " is found to be = 1,458, and the " range " is the mean 

 d= the probable error, that is from 6,600 to 9,500. Compar- 

 ing this range with the estimated results of the hauls, we 

 find that three of the series are within the range and three 

 are outside it, and two of the latter (3,880 and 10,020) are 

 very considerably beyond the limits of the probable error 

 of the experiment. 



The Diatoms of the other hauls give much the same result 

 when treated in the same manner — that is, roughly 50 per cent, 

 or rather more of the observed variation in the catches is not 

 covered by the calculated range of error of the experiment. 



A series of detailed tables are given in the full report 

 from which the above is summarized, in which each of the 

 principal groups of the plankton, and also three prominent 

 organisms, the Diatom Coscinodiscus radiatus, the Dino- 

 flagellate Ceratium trifos and the Copepod Pseudocalanus 

 elongatus, are shown for all seven series of hauls treated as 

 in the case of the Diatoms of April 3 discussed above, and 

 giving in each case the figures necessary to make a com- 

 parison between the range of variation in the catches and the 

 calculated range of error. These tables show that in each 



