436 DISCOVERY REPORTS 



rafts are of very different sizes. Usually the smaller individuals lie along the convex 

 sides of the larger ones. It may be mentioned that rather similar rafts of R. semispina 

 have also been seen rarely in the Southern Ocean. 



REPRODUCTION 



As already mentioned, the commonest method of reproduction of R. curvata is by 

 binary fission, as in most other true solenoid diatoms. Continued binary fission, how- 

 ever, results in a gradual diminution of size of the individuals ; and auxospore formation, 

 by means of which their size is again increased, must sooner or later intervene or the 

 species would die out. Auxospore formation is well known in other members of the 

 genus such as R. styliformis, R. alata and R. bidens, but was rarely seen in our relatively 

 scanty material of/?, curvata. The measurement investigations described later show that 

 it tends to take place most frequently at the height of summer, in December and 

 January, and to a lesser extent in spring and early autumn. We have comparatively few 

 observations from the optimum zone of the species at these times, which probably ex- 

 plains the scarcity of auxospores in our samples. A third method of reproduction, by 

 the formation of microspores, is common in some solenoid diatoms (e.g. Corethron 

 criophilum and Rhizosolenia semispina), but has not been observed in R. curvata. It is of 

 interest to record, however, that in the course of working through this material I came 

 across some very beautiful examples of microspore formation in R.polydactyla, a species 

 with a similar but less restricted distribution. So far as I have been able to determine, 

 microspore formation had not previously been seen in this species. 



In R. curvata then, only two methods of reproduction, binary fission and auxospore 

 formation, are known. When an individual frustule is about to divide, the endochrome 

 accumulates in two ovoid masses towards the middle of the cell. At this point the 

 minute granulations of the outer wall of the old frustule become indistinct, rendering 

 it more transparent, and the intercalary bands also become much less readily visible. 

 This stage is represented in Plate XIV, fig. 5. Later the inner apexes of the two new cells, 

 each with its mucron, are clearly laid down before the outer wall of the end of the new 

 frustule, which forms immediately within the old one, is nearly complete (Plate XIV, 

 fig. 6). The old frustule breaks along one of the original intercalary bands, and often 

 persists as an outer degenerating sheath long after fission has been completed, so that at 

 one end of a solitary frustule a sort of collar can be seen. This is shown in Karsten's 

 original figure of the species (1905, Taf. xi, fig. 2). 



Auxospore formation cannot be followed so closely owing to the scarcity of examples 

 of it in our material. T 1 appears, however, that after central aggregation of the endo- 

 chrome, similar to that observed in the first stages of binary fission, the old frustule 

 suddenly breaks at one of the intercalary bands. From the broken end a much larger 

 bag-like extension with plastic walls develops, which presumably takes on the cha- 

 racteristic shape and slowly becomes silicified, as in other solenoid diatoms where auxo- 

 spore formation is well known. An auxospore of R. curvata is shown in Plate XIV, fig. 7. 



The methods of reproduction adopted by R. curvata provide an important line of 



