jaculiferum var. subexcavatuw , forms of >S7. O'Mearii, etc., and 

 also some of the larger forms of Arthrodesmus IIU-HK, are very 

 confusing in tlieir relationships, and are only placed with 

 considerable difficulty. In some cases it is necessary to 

 examine a laro-e series of forms from various districts in 



n 



order to arrive at a decision concerning the genus. 



Just as there are fusiform or biradiate forms of Staurastrum, 

 so there are triangular forms of Arthrodesmus, such as A. 

 triangularis var. sub triangularis forma triyut.'trd (cf. PL CXV, 



Notwithstanding the difficulties concerned with the placing 

 of these transitional forms, and the fact that its definition 

 cannot be made very exact, the genus Arthrodesmus is a 

 great convenience for the reception of those Desinids which 

 do not rio-htlv belono- either to Staurastrum or Xanthidium. 



*j O 



The genus has been subdivided into the sections ' Tetra- 

 canthinm 3 and c Octacanthium 3 by Hansgirg, into Euarthro- 

 desmus 3 and f Centrenterium 3 by Eaciborski (who included 

 Boldt's subgenus of Xautltidium}, and into ' Aplodexmus 3 and 

 ( Schizodesmus ' by Turner. It is not possible, after a careful 

 consideration of the species of this genus, to accept the 

 suggestions either of Raciborski or of Turner, but Hansgirg's 

 two sections are quite satisfactory for most species of the 

 genus. To these we have added a third section for the 

 reception of the species which do not appear to have been 

 sufficiently well studied by the above-mentioned authors. 



Both Borge and Liitkemiiller have suggested that Ichthyo- 

 crrcny should probably be placed as a section of Arthrodesmus, 

 but that genus is nearest to Tetmemorus and has no close 

 affinity with Arthrodesmus. 



SECTION A. Semicells with a single spine on each side, attached 

 to the lateral angle. [TetracanlJtitnn (Nag.) Hansg.] 



* Semicells obsemicircular or in the form of an inverted 



triangle, sinus open. 

 t Angles of semicells furnished with long or very long 



spines. 



+ Apex of semicell straight or convex (or very faintly 

 concave), spines very rarely horizontal, usually 

 divergent, rarely convergent. 



1. A. Incus. 



+ + Apex of semicell elevated, and retuse in the middle 

 (except in one variety), spines horizontal. 



2. A. triangularis. 

 Apex of semicell retuse in the middle, spines of 



great length and widelv divergent. 



T* O O 



3. A. qiiirtferus. 



