52 



to cause a pulse of development of Fragilana, for nitrates are 

 abundant when the diatom declines and is at its minimum. It 

 does not seem possible to find in the unstable environment of this 

 organism any external factor which shows a causal connection with 

 its periods of growth. 



Apstein ('96) found that this diatom reached its major pulse 

 in March and April in Dobersdorfer See, and a minor one in 

 November. 



The cells of this diatom form long twisted bands, visible to the 

 unaided eye. They reach a much greater length in this species 

 than in the preceding one, and are longest during the height of the 

 growing period, decreasing rapidly in length as it declines. The 

 average number of cells in a ribbon at the time of the maximum lies 

 between 150 and 200, and at other times is usually below 100 and 

 often below 2 5 . 



The vernal pulse of this species coincides with that of F. croto- 

 nensis, and appears either with or just after the volumetric pulse. 

 The December pulses may in part serve as primal food sources for 

 the fairly constant minor volumetric pulse of December. 



Gomphonema constrictuvn Ehrbg.* Average number, 501,923. 

 This species appears irregularly, with a predominance of occur- 

 rences in May and November, and is apparently adventitious. 



Melosira granulata (Ehrbg.) Ralfs var. spinosa Schroder. 

 Average number of cells, 1,181,125 (filter-paper, 34,762,365). 

 In 1897 it was more than five times as abundant. In the 

 filter-paper collections as a whole it is about fifty times as abun- 

 dant as in those of the. silk net. A much greater proportion 

 of single cells and short filaments occurs in the latter collections, 

 since the longer filaments are the more readily retained by the silk. 

 In the discussion which follows, the data from the silk collections 

 will be used, since they cover the whole period. The data from 

 the filter-paper collections indicate very nearly the same seasonal 

 routine, and the differences between the results by the two methods 

 lie in the proportions of the numbers rather than in the direction of 

 movement in the fluctuations. The pictures of the seasonal 

 changes in occurrence of the diatom given by the two methods are 

 essentially alike aside from greater irregularity during minimum 

 periods, resulting from the larger margin of error in the filter-paper 

 method as I used it. 



