212 DISCOVERY REPORTS 



The classification, as I see it, cannot be based upon such variable features as the num- 

 ber of sides possessed by the frustules, the outline, form of process, arrangement of 

 markings, or modus vivendi, but rather upon some feature or features which show con- 

 stancy from group to group, making due allowance for variation of degree but not of 

 kind. The only feature that satisfies these demands is the structure of the valve. 



If we examine the type species, T.favus, we find that the valves are of a very charac- 

 teristic structure, one which allows the species to be recognized instantly. This structure, 

 which has been so ably elucidated by the work of Miiller (1871) and Flogel (1884), is 

 constant in the favus group, which contains about thirty species. The valve is strongly 

 siliceous and covered entirely with a rigid hexagonal loculation, in transverse lines, open 

 on the upper surface, but closed on the lower surface. The lower surface or floor may 

 or may not bear markings which have been variously interpreted, the walls of the loculi 

 may or may not be furnished with secondary markings or small spines, particularly at 

 the points of confluence. The processes of the valve are cornutate or subcornutate. 



The loculation of the entire surface of the valve is a constant character of the favus 

 group, a feature which sharply defines it from the rest of the biddulphioid diatoms, and 

 endows all the members with an appearance that is unmistakable. It is essential that 

 the integrity of such a group should be preserved and that its absorption into another 

 and older genus should not be permitted. 



Another section of the genus which requires attention is that described as the 

 arcticwn group. This consists of about thirty species or varieties that are structurally 

 allied to T. arcticwn Brightwell. Here again, as in the favus group, to which the group 

 is closely allied, the structure is found to be very characteristic. Members of the group 

 exhibit polymorphism in the same way, and the surface structure of the valves exhibits 

 a hexagonal aerolation. One or two important differences, however, occur. The hexa- 

 gonal loculation is not always entire over the whole surface of the valve. In some species 

 the loculation breaks down in the middle of the valve to a number of coarse granules ; 

 in others it is entire over the central area of the valve face, and is reduced to granulation 

 at the valve mantle. The cellulation is less regular than that in T. favus, and as a rule 

 radially arranged instead of being in parallel lines. In the arcticum group the loculi in 

 the valve are closed both upon the outer and inner surfaces, with the exception of 

 poroids upon the lower surface. The angles of the frustules of the arcticum group are 

 furnished with a pad of minute pores, which is only very slightly raised above the level 

 of the rest of the valve surface in some species, while in others the angles are actually 

 lower than the central area of the valve. The pads of fine pores secrete mucus by means 

 of which the frustules attach themselves to the substratum. Members of the group are 

 solitary, littoral, and seldom occur in great numbers. 



Research into the evolution of the biddulphioid diatoms has revealed that there exists 

 a relation between the geographical distribution of the favus and arcticum groups and 

 climatic conditions. Presumably these two groups emerged at about the same time, for 

 they both make their first appearance in the same geological formation, namely, the 

 Oligocene at Oamaru, New Zealand. Members of the genus Triceratium Ehrenberg 



