420 DISCOVERY REPORTS 



insufficient to justify any of the more desirable statistical elaborations. One fact, how- 

 ever, seems to show that these arithmetical means have some real significance : the very 

 marked maximum in February was partly due to exceptionally rich hauls obtained in 

 1 93 1, but if this series is excluded it still remains obvious that the maximum is reached 

 in February although its numerical value is greatly reduced. It should be noted that all 

 tables and diagrams relating to seasonal effects have been arranged as if the year began 

 on July i . This seems the easiest way in which to indicate the reversal of the seasons in 

 the southern hemisphere. 



The average positions of the Antarctic and subtropical convergences shown on the 

 distribution charts are taken from Deacon (1937). I have also benefited greatly from 

 the opportunity of discussing some of the hydrological points involved in this work 

 with him and with Mr A. J. Clowes, from whose notes the probable positions of the 

 Antarctic convergence on our several series of observations in longitude 8o° W have 

 been taken. 



I was led to study the size relations of R. curvata by Mr R. S. Wimpenny, who very 

 kindly allowed me to read the proofs of his most interesting paper on R. styliformis in 

 the North Sea (Wimpenny, 1936). R. curvata is circular in section, and apart from its 

 regular curvature it closely resembles a robust individual of R. styliformis in general 

 form. It was therefore considered that of all possible measurements the diameter was 

 most likely to bear a reasonably constant relation to the volume, since the organism is 

 essentially cylindrical. 



In attempting to trace the seasonal variation in size, and the presence or absence of a 

 correlation between size and temperature, I have followed Wimpenny's methods so far 

 as the limitations of my material permitted. Whereas Wimpenny (1936) was dealing 

 with abundant material from localized populations, I was confronted with scanty 

 material distributed all round the world. The methods adopted in the endeavour to 

 make the correlations fairly comparable were as follows : 



In studying seasonal variation in size, data from the Scotia Sea only have been used, 

 except for the autumn and winter months of May, June and August. At these times 

 I was forced to include material collected in the South Atlantic and Indian Oceans in 

 order to get a sufficient number of observations. Between five and ten stations spread 

 over each month were selected, within the organism's normal temperature range, and 

 additional measurements over and above those made in the ordinary routine examination 

 of the samples were taken, so that the number from each station was nearly equal and 

 the total number over 200. The months in which it was impossible to obtain strictly 

 comparable data are indicated by an asterisk in Table VIII. The measurements for each 

 month were divided into 10/x diameter classes: < 29^, 30-39/x, 40-49/n and so on. The 

 percentage frequency of each class was calculated and the results for each month 

 entered in the table. 



The measurement data obtained in the initial routine examination of the samples in 

 which R. curvata occurred were used to compute the mean diameter of the species. 

 These data amounted to 3024 measurements distributed irregularly over 113 stations. 



