SILICIOUS AND HOBNY SPONGES WILSON. 339 



T. diploderma, but there is nothing to show that this statement is 

 more than a conjecture. 



If the species enumerated above are to be merged, as I think 

 should be done with perhaps the exception of D. multifida, so pe- 

 culiar in its shape of bod}', it would be more justifiable to use for 

 the group the name of the oldest recognizable form — namely, that 

 of D. ingalli (Bowerbank, 1872), as Hentschel, 1912 (see above) has 

 done. Hentschel in his text (p. 317), although not in his definition, 

 would also include D. japonica Sollas, a species which Topsent 

 (1918) holds to be distinct. In regard to the latter three of this set 

 of forms, D. ingalli, niaza, seychellensis, and japonica, Sollas long 

 ago (1888, p. 428) stated that he was "inclined to regard them as 

 varietal modifications of a single species." 



I may be allowed a word or two as to what I understand by this 

 expression "varietal modification " or variety (compare what is 

 said on the separation of genera and subgenera, farther on under 

 Tetilla.) 



Certainly the phenomena of variation in sponges are coming closer 

 to our eyes as many, myself among the number (1904, 1919, 1919 b), 

 have pointed out in recent years. The various diagnostic features 

 which have been picked out as characterizing natural races all vary 

 and vary, apparently, independently of one another, so that many 

 kinds of combinations come into existence. Moreover they do not 

 vary as definite, relatively unalterable, features (unit characters of 

 the older Mendelian terminology) which are present or absent, as in 

 the t} T pical Mendelian crossings, but any one "character" varies 

 quantitatively so as to produce a series only the extremes of which 

 contrast sharply. Hence as the number of combinations known to 

 us increases — in other words, as the number of carefully studied in- 

 dividuals from various regions, intergrading more or less and in 

 very complex fashion, increases — it becomes more and more difficult 

 to apply our customary method of classification, which is essentially 

 a splitting method. Any attempt to cleave the whole of one of these 

 assemblages into subgroups (subspecies, varieties, etc.) is arbitrary. 

 It can only lead to erroneous conceptions. Nevertheless within a 

 group of related natural objects, such as an assemblage of Donatio, 

 individuals, we see certain distinct types to which numbers of in- 

 dividuals conform, but we also see numerous individuals which con- 

 form to none of these and which in respect to one character are in- 

 termediate between certain two types while in respect to another 

 character they may be intermediate between two other types. If 

 now we are to express objective fact in our classifications, we have to 

 recognize by name certain somewhat heterogeneous groups {Donatio 

 species) and the types within these (varieties), understanding by the 



