422 DISCOVERY REPORTS 



to take place on a major scale. The gap is spread over the period April to July, there being no effective 

 coverage in April,^ only six observations in May and none in either June or July. Unfortunately this 

 period is critical, for clearly the absence, apart from the rare instances noted above, of young not long 

 hatched swarms from this region from November to March, when the vertical net was used no fewer 

 than 1177^ times (Fig. 156, unbracketed numbers) in South Georgian waters as against only 465 times 

 (Fig. 156, numbers in brackets) in Weddell West, and the widespread occurrence there (Figs. 150 



60 

 56 



52 



48 



44 



42 



40 



38 



36 



34 



32 



30 



28 



26 



24 



22 



20 



18 



16 



14 



12 



lO 



8 



7 



6 



5 



4 



3 



2 



Fig. 156, Growth of the swarms at South Georgia and in the Weddell drift, showing how the island population is 

 derived from incursions of surface-borne larvae from the Weddell Sea and is not of local origin. 



and 156) of late Furcilia swarms in August, when observations resume, postulates that it must be 

 some time during the period covered by this 'observational gap' that the first major incursions take 

 place. In four of the samples they analysed from five stations made off the east coast in May 1927, 

 Hardy and Gunther (1935, Appendix 11, Table iii) record moderate to substantial numbers ofE.superba 

 ' Cyrtopias ' ^ ( = almost certainly Furcilias) which suggests that it is possibly in that month (Fig. 150), 

 or even in April, that the young swarms carried in such profusion in the Weddell stream first begin to 

 arrive on a substantial scale. The observations of Hansen (p. 334) also suggest that it is in May that 

 the first large-scale incursions take place — of larvae perhaps in the early Furcihastate. Our own 

 observations round the island (Fig. 150) also point rather strongly to its being in autumn that a major 

 invasion is going on by young swarms sprung from risings in the Weddell Sea. For, although few and 

 rather widely scattered, especially on the east side, they do nevertheless show, at every station falling 

 fairly inside the Weddell stream in April and May, in three instances substantially, the presence of a 

 larval population, and at the same time that the south-western backwater, in striking contrast, is 

 barren. 



1 Not effective because it was confined (Fig. 148) principally to the south-western backwater. 



^ This figure includes vertical net hauls from which samples were analysed by Hardy and Gunther. 



^ Since no more precise developmental data are available for these May samples I could not plot them in Fig. 156. 



