RHIZOSOLENIA CURVATA 417 



to a large proportion of our enormous collection of samples from the more southerly 

 areas. It is unthinkable that Zacharias, whose description of the species is mainly 

 accurate, should have confused it with the warm-water species R. robusta Norman, 

 though the superficial resemblance between the two is sometimes strong, as is shown by 

 one of Karsten's figures of the latter (1906, Taf. xxix, fig. 10). The dimensions Zacharias 

 quotes for his single supposedly tropical specimen of R. curvata are also against such a 

 view. The most probable explanation is that an error was made in collection, and that 

 the specimen did not originate at the locality from which it was recorded. 1 



It is possible that the previous record of a single specimen obtained by the ' Valdivia ' 

 at the island of St Paul, which is just north of the subtropical convergence in the Indian 

 Ocean, may have been due to an error of this kind. This is unlikely, for it is evident that 

 the 'Valdivia's' collections were most carefully made; but I do not believe that these 

 two northerly records, one of them almost certainly made in error, invalidate the 

 observation that R. curvata is essentially a sub-Antarctic species. 



Apart from the St Paul Island specimen, Schimper's records, quoted by Karsten 

 (1905, 1907), show that R. curvata occurred at four other stations worked by the 

 ' Valdivia '. Two of these were just to the south of the Antarctic convergence north-east 

 of Bouvet Island, one just south of the convergence off Kerguelen, and the last in the 

 sub-Antarctic Zone in the Indian Ocean some 300 miles north-east of that island. Thus 

 three of the four records lie south of the average position of the Antarctic convergence 

 as we know it to-day. However, apart from minor changes in the position of the con- 

 vergence, which are known to occur, it seems highly significant that all three records 

 are from localities in which sharp changes in the relief of the sea-floor may be expected 

 to complicate the movements of the water layers. The two Atlantic stations were very 

 near the Meteor bank, while that to the south of Kerguelen lies almost on the Kerguelen- 

 Heard Island ridge. 



Mangin (1922) records the species from two of the stations worked by the ' Scotia' 

 in the middle of the South Atlantic, one to the north of the convergence, where it was 

 "fairly common"; one to the south, where it was "rare". The positions of all these 

 early records, except that of Zacharias 's first sample, which is not accurately known, 

 are shown by distinctive symbols on the general distribution chart in the concluding 

 section of this paper (Fig. 7). 



1 The samples upon which Zacharias's two papers are based were obtained for him by Herr Wahlmann, 

 a seaman on a sailing ship, who collected them as opportunity offered — in port, or when the ship lay be- 

 calmed. We are justified in assuming that his collecting was carried out under difficulties. From the sequence 

 of dates and localities of the samples, it seems fairly certain that Herr Wahlmann's ship was a nitrate clipper 

 on the regular run between Chilean ports and Europe, round Cape Horn. It is probable that this tropical 

 sample was the first one that he had been able to get after his unusual opportunity of collecting south of 

 Cape Horn, and his net, which may have been stowed away wet, may have retained some specimens of the 

 southern species. The fortuitous occurrence of diatoms from previous hauls is a potent source of error in 

 studying their distribution, unless the nets are carefully washed down and dried after use. Zacharias does not 

 give the date at which the first sample containing R. curvata was obtained, but it was certainly prior to the 

 tropical haul in question, and none of the other catches he describes intervened, for he gives the dates of all 

 of them. 



