Ross: Classification of Diatoms 407 



of characters to some extent culling across the present classification. Ultra- 

 structure, however, provides few characters and a system based solely upon it 

 would be as open to criticism as one based solely upon the structure of the 

 valve as seen under the light microscope. All of the information about the 

 frustule, whether obtainable with the light microscope or the electron micro- 

 scope, must be taken into consideration with any that can be obtained about 

 other characters. 



A few authors have attempted to use characters from the cell contents, in 

 particular the form of the chromatophores, for delimiting genera within the 

 Naviculaceae (Pfitzer, 1871, Mereschkowsky, 1901a,6, 1902, 1903(7,6) or sub- 

 genera within Navicula (Karsten, 1899). However, except where these groups 

 could also be readily distinguished by characters of the valve, e.g., Anomoeoneis 

 Pfitz. and Neidium, they have not been adopted by subsecjuent authors. The 

 principal reason that there has been no further work along these lines is a matter 

 of technique. The greatest possible amount of detail in the structure of the 

 valves of diatoms can be seen most easily under the light microscope if all of 

 the organic matter is removed and the frustules mounted in a medium of high 

 refractive index. Diatomists have rarely used any other method of making 

 preparations and all collections of diatoms consist almost entirely of specimens 

 treated in this way. They provide information perfectly adecjuate for identifi- 

 cation, and hence workers on floristics and ecology have had no incentive to 

 change their technique. These have been the chief fields of work of virtually 

 all diatomists throughout this century and even when they have turned their 

 attention to true taxonomy they have not altered their methods. It may be 

 that it would not have been possible before the phase-contrast microscope was 

 available to devise a technicjue which made both the fine detail of the valve 

 structure and the cell contents visible in the same specimen. It would seem, 

 however, that it was not attempted. The justification for ignoring the cell 

 contents in taxonomic work has been the contention, also used in connection 

 with ultrastructure, that a classification by chromatophore number, shape, and 

 disposition within the cell runs counter to the currently accepted one (Peragallo, 

 1907). This criticism is valid insofar as it is directed against a classification in 

 which characters of the chromatophore are accorded overriding importance, but 

 it is not a reason for ignoring the cell contents completely. 



It has been pointed out that the classification of the Naviculaceae is on a very 

 unsatisfactory basis, at least above the specific level, and there is no reason for 

 supposing that it is much better in other families of diatoms. Cell contents 

 and ultrastructure provide characters of which the distribution does not, in 

 places, accord with the current classification. There is no justification for 

 arguing from this that variations in these features occur at random and have no 

 taxonomic significance. To do so is to attach overriding importance to the 

 particular characters of the frustule on which emphasis is placed in the current 

 classification; not even a priori grounds have been advanced for this. Instead 

 of arguing in this way from the lack of correspondence between the current 

 classification and the distribution of types of cell contents and ultrastructure, 

 this discrepancy should be regarded as an indication that there is a need for a 

 new classification based upon the extent of overall resemblance with these 

 features taken into account. 



