50 BRITISH DESMIDIACEJE. 



with furcate spines and the latter those with simple spines. 

 Wille, in 1890, elevated these to generic rank, but 011 most 

 inadequate grounds, as may be seen by mere reference to the 

 forms of the common species X. armatum. Moreover, in 

 some of the tropical and subtropical species partially divided 

 spines occur of such a nature that it is impossible to refer 

 these species with certainty to either of Wille ; s proposed 

 genera. It is also significant that in his revised account of 

 the Chlorophyceae in Engler & Prantl, ' Natiirl. Pflanzenfam.' 

 1909, Wille has himself submerged his proposed genera in 

 the genus Xanthidiiim. 



Boldt, in 1888, suggested that the genus should be sub- 

 divided into Euxanthidium, with parietal chloroplasts, and 

 Centrenterium, with axile chloroplasts. For the reasons that 

 the disposition of the chloroplasts of most species of Xanthi- 

 diiim is entirely unknown, and that the parietal disposition of 

 chloroplasts may be acquired at any time by any Desmid, and 

 is therefore no criterion of its affinities [vide Vol. II, p. 126], 

 these suggestions cannot be accepted. Moreover, the chloro- 

 plasts do not appear to be constant and may sometimes be 

 parietal and sometimes axile in the same species. 



Turner, in 1893, adopted Lundell's two subdivisions, to 

 which he added a third, Micracanthum j and this subdivision 

 of the genus into three sections we have adopted. 



The decisive combination of characters which we adopt for 

 this genus is the possession of spines and the presence of a 

 differentiated area in the centre of each side of a compressed 

 semicell. Even in the very rare triangular forms the differen- 

 tiated area is present in the middle of each of the three sides. 



This central area (whether protuberance or thickening) is 

 the only distinction which really separates Xanthidium from 

 Arthrodesmus. 



We arrange the British species as follows :- 



SECTION A. Schizacanthum Lund. Cells large or very large, 

 with tri- or quadrifurcate processes. 



1. X. armattnn. 







SECTION B. Holacantlium Lund. Cells of medium size or 

 small, with long simple spines. 



a. Semicells with one pair of spines on each side. 



Senricells elliptic, lateral spines in a vertical plane, cen- 

 tral area small and slightly thickened. 



2. X. subhastiferum. 

 ** Semicells elliptic-hexagonal (not regularly elliptic), 



lateral spines not in a vertical plane, central area 

 large, thickened, and scrobiculated. 



3. X. tetracentrotum. 



