8 V. T. OLEVK, SYNOI'SIS OF NAVK^ri.Oll) DIATOMS. 



puiicta. In many forms there is an interior stratum, wliicli may be isolated from the alveolar 

 network. It is freqiiently found in preparations of 'Trachyneis Dehyi and allied fortns, and has 

 been figured in A. Schmidts Atlas PI. XTiVITT f. 23. Also in Narirula distans' this interior stra- 

 tum ocu'urs frecjnently in an isolated form. 



In the most complicated cases, as in Tntrhyncis. we may distinguish tin-ee different 

 strata, thus 



1. The porons or ocelliferous interioi" stratum. 



2. The alveolar stratum of reticnlating costae. 



3. The exterior, minutely punctate stratum. 



The structure of the valves presents characteristics which are of great iniportantie for dis- 

 tinction l)oth (»f species and genera. The coarseness of the structure may vary in the same species, 

 though in most cases such variation is less than is usually believed. The direction of the striiv, 

 the arrangement of their ])uncta, are characteristics subject to only very slight variation in the 

 .same species and in groups of allied forms. 



CcU-ronfcnt.s. As is well known, the chromatophores of the diatoms present a great varia- 

 tion in different tribes, consisting in some of numerous granules, in others of one or two jjlates. 

 To the latter type belong almost all the naviculoid diatoms, which have either one or two cliromato- 

 phore-plates. The position and the form of these plates appear to be constant for groups of 

 allied species, as is proved by the excellent researches of Pfitzer. But the cell-contents are 

 known only in a very limited number of forms, and the characteristics dependent on the cell- 

 contents are, for the greatest number of forms, yet to be studied. The same is also the case 

 with the reproduction by means of au.rosporcs, and the characteristics, derived from the living 

 cell, cannot for the present be used in the systematic arrangement of species, as they are too 

 little known. 



The chai'acteristies derived from the structure of the valve are of the greatest im- 

 portance in the delinition of genera and families, and next to those, in my opinion, are the 

 presence or absence of longitudinal lines, and the nature of the non-striate parts of the valve, 

 or the areas. 



An ideal system should take in consideration the evolution of the different forms, but in 

 the present fragmentary state of oui- knowledge, such consideration cannot be more than an ap- 

 proximation wicli may become closer to the truth as our knowledge of the forms becomes mor'e 

 extensive and perfect. The ([uestion which of the groups of forms are the higliest and lowest may 

 be answered by considering as the lowest those in which the characteristics are the least developed, 

 and as the highest those in which the characteristics have attained their greatest evolution. But 

 a greater developnuMit in one direction usually involves the slighter develo])ment, or oblitera- 

 tion, of other characteristics, so that one group may be the highest in some respects, and 

 another in others. The changes of the oi'ganisms do iH)t always indicate ])rogress, but ai'e 

 frequently retrograde, especially where the mode of life has become ])arasitical. A natui'ai 

 system must consequently take account not only of the upward cxolution, but also of the 

 downward. 



As the known diatoms probably represent merely a fraction of those, which exist and 

 have existed, it will be necessary in constructing a natural system to till the gaps with coji- 

 jectures. 



There can be no doubt that the naviculoid diatoms are most nearly ccninei'ted with the Kitz- 

 chirce among the diatoms without median line. The genera Tropidoncis and Ampkiprora are very 

 closely allied to several forms of Nit.tchia. We meet in that genus with the carinated asymmet- 

 rical valves and the wing of Tropidoneis, also the complex zone and the carinal puncta of Amp/ii- 

 prora. In (jonstructing a natural system then we may arrange the genera and gi'oups in order of 

 their relationship to the two genera above mentioned, and the following schcinc shews how I sup- 

 pose the different groups of the naviculoid forms to be related. 



