200 DISCOVERY REPORTS 



division of the vegetable kingdom under the name of Bacillariophyta ; but the funda- 

 mental idea of Schiitt's two suborders was allowed to remain unchanged. 



For some considerable time I have felt dissatisfied with Schiitt's method of classi- 

 fication and with every modification of it that insists upon the two suborders based 

 upon either radial and concentric structure on the principal axis or isobilateral structure 

 upon the polar axis of the valve, for a large number of genera that have been included 

 in Centricae possess neither radial nor concentric structure, and their construction can 

 in no way be referred to a central point. Mention has only to be made of Biddulphia, 

 Chaetoceros, Anaulus and Rhizosolenia to bring to mind a large group of forms which 

 strictly speaking cannot be brought into line with the general idea of radial symmetry 

 so clearly portrayed in such genera as Coscinodiscus, Actinoptychus, etc. 



The shortcomings of the scheme are not apparent so long as one's attention is con- 

 fined to the study of fossil or freshwater material, which is usually strongly siliceous 

 and in the main composed of forms which lend themselves readily to such a classifica- 

 tion. But when the weakly siliceous plankton diatoms are studied, one is immediately 

 struck with the inadequacies of Schiitt's scheme, which does not pay sufficient attention 

 to the structure of several groups of marine diatoms that constitute a considerable pro- 

 portion of the phytoplankton. With the increasing importance of the study of marine 

 diatoms, and the advance in our knowledge of their structure, it is necessary to revise 

 the classification and the following scheme is therefore proposed. 



Diatoms are considered as a class of Algae, Bacillariophyceae, comprising one order, 

 Bacillariales, which is divided into the following ten suborders. 



Discineae. Valves disciform, usually flat or convex, but occasionally concave. 

 Frustules usually much greater in diameter than in thickness. Valves sometimes 

 hemidisciform, frustules cuneate. Valves usually covered with puncta or areolations 

 which radiate from the centre. Connecting zone usually simple, spinulae present or 

 absent. Structure equal on all radii. 



Aulacodiscineae. Valves disciform, flat or convex, punctate or areolate, markings 

 usually radial. Valve furnished with a number of pediform processes. Connecting zone 

 simple. 



Auliscineae. Valves circular, oval or triangular, usually flat, furnished with large 

 ocelli. Valve covered with punctation or sculptured lines, usually arranged with 

 reference to the ocelli. 



Biddulphiineae. Valves angular, seldom round, angles furnished with strong pro- 

 cesses ; spines present or absent ; markings granular or hexagonally areolate, sometimes 

 radial, usually arranged with reference to the angles of the valve. Valve usually com- 

 pound, laminate. Thickness of frustule usually more than the diameter of the valve. 

 Connecting zone usually simple. 



Soleniineae. Valves circular or subcircular, domed or conical, seldom flat, frequently 

 apiculate ; apicule eccentric or marginal ; markings fine, granular, often absent. Valves 

 frequently armed with a circlet of spines. Connecting zone usually complex, composed 

 of numerous scale-like intercalary bands, annular, imbricate or squamose, usually con- 



