ANIMAL NATURE OF DIATOME^, 479 



by the flexuose state of the surface. A line traverses 

 the middle of these triangular spaces. The mar- 

 ginal perforations correspond to the lines which border 

 the triangular spaces in A. sedenarius, to which again 

 they run in the middle of those of A. duodenarius 

 and of A. ocfodenarius ; and the internal septa {setti), 

 according to Ehrenberg, always correspond to those radii 

 which do not terminate in an aperture. We learn, too, 

 from the figures, that the areolation of the surface is con- 

 tinued without interruption even over the radii, and that, 

 besides the indicated apertures, there are visible also 

 those of the opposite surface, M'hich alternate with the 

 former. And Ehrenberg observes, that on this account 

 there occurs an optical illusion not easy to explain. 

 Hence arises a suspicion that an equal number of septa 

 {setti), but alternating with each other, project from the 

 inner superficies of both surfaces without reaching the 

 opposite surface, and so the marginal apertures belong 

 sometimes to one, sometimes to the other. 



All the species do not seem to have the same structure. 

 In some, at least, {iernarius, senarius, octodenariws,) the 

 radii are figured smooth, as in Actinocydus ; it is not 

 said that the triangular spaces are alternately clear and 

 obscure, nor that they are traversed by a longitudinal 

 line; and there remain doubts as to the marginal 

 apertures. 



As to the internal organisation, little can be collected 

 from the observations of Ehrenberg. 



Finally, it deserves attention, that the fourteen species 

 enumerated in this genus differ almost exclusively in 

 the number of radii. Two only, {senarius, Jiexapterus,) 

 having the same number, are much different from each 

 other. Three have unequal numbers. 



The family of Coscinodisceaj ( Coscinodiscus, Actinocy- 

 dus, ActinoptycJius) is, even in the opinion of Klitzing, 

 more nearly allied to that of the Melosirese than to any 

 other. I believe that both the Campylodisci and flexuose 

 SurirdlcB are to be regarded as allied to this family. 



