DIATOMACEJE 595 



until these again reach their minimum size. This theory has. 

 in the judgment of Count Castracane, deceived many botanists, 

 from the idea that it was founded on actual observation, and lias 

 at the same time been in harmony with the natural tendency to 

 generalisation, in attributing to the whole family of diatoms that 

 faculty of division which has been regarded as the universal property 

 of the vegetable cell. The ' auxospore ' theory rests on the supposed 

 inability of the siliceous walls of diatoms to expand ; and implies, 

 secondly, the idea that all diatoms are capable of binary sub- 

 division ; and thirdly, that there is no mode of reproduction except 

 by auxospores. That the siliceous walls of diatoms are capable of 

 distension seems to result from the examples already given of Melosim 

 sn.bfle.riUs and J/. varians, as also from some other species in which 

 there may often be observed a sudden variation in diameter in frus- 

 tules united together in a row. But the power of increase in size of 

 the siliceous diatom-cell is evidently proved by the spnrangial frus- 

 tules ofOrthosira Dickiei, 1 where, in the chain of cylindrical frus- 

 tules of the same diameter, the sporangia! frustule is dilated in its 

 equatorial axis, but much more so in its polar axis, pushing back the 

 base of the next cell and furring it to fold itself up so as to occupy 

 the whole cell-cavity, and sometimes even that of the next frustule. 

 The exactness and fidelity of the figure given in Smith's Synopsis,' 

 besides being guaranteed by the authority of the distinguished author 

 and by the signature of the celebrated artist Tufien West, Count 

 Castracane was able to confirm by a magnificent preparation 

 of these diatoms in which are a number of sporangia! frustules. 

 The auxospore theory supposes the fact that all diatoms are capable 

 of binary subdivision, since the auxospore is understood, according 

 to Pfitzer, to provide for the progressive decrease in size of the 

 frustules, with the production of larger sporangial frustules, destined 

 to commence a new descending series. But binary subdivision cannot 

 take place in genera with unequal valves, as it is universally 

 acknowledged that the two new valves which are formed in the 

 process of binary subdivision must stereotype themselves on the old 

 valves ; and for this reason the process cannot take place in thus, 

 genera in which the axes cross one another, like Campylodiscus, or 

 in those in which the two valves, although equal, yet constantly 

 unite in such a way that the similar parts alternate with one 

 another, as may be seen in Asteroktmpra. That it is impossible for 

 binary subdivision to take place in these three classes of forms. i> 

 confirmed by the fact that, notwithstanding that there are recorded 

 not less than seventy-five observations of the process of division in 

 them, not one affords an exception to the rule given above. 



Where multiplication by binary subdivision occurs among the 

 Diatomacece, it takes place on the same general plan as in the Des- 

 inidiacece, but with some modifications incident to peculiarities of 

 the structure of the former group. The first stage consists in the 

 elongation of the cell, and the formation of a 'hoop' adherent to 



1 See Castracane, ' The Theory of the Eeproduction of Diatoms,' Atti dell' Accad. 

 Pontif. del Ni/ori, Lhicci, May 31, 1874; and 'New Arguments to prove that 

 Diatoms are reproduced by means of Germs,' ibid. March 19, 1870. 



g Q 2 



