292 Profs, Percy Frankland and Marshall Ward. 



It may seem a remarkable coincidence that this partial filament 

 was the same length as that measured previously, but the explanation 

 is very simple, though I did not discover it till later : it is that the 

 long filaments tend to begin breaking up into isolated segments of 

 this length (each consisting of several cells) very early under certain 

 circumstances. 



I have given the measurements taken at 2, 3, and 4 P.M., but these 

 are far less useful than those preceding them, for two reasons : (1) the 

 intervals are too long, and (2) the filament underwent considerable 

 curvatures from about 2 o'clock onwards, so that I cannot insist on 

 the accuracy of the measurements so strongly as on those made pre- 

 viously. 



If, now, we examine these results, it is clear that the filament grows 

 more slowly at 15 C. than it does at 20 C., as was to be expected. 

 It is also evident that considerable variations in the rate of growth 

 occur during the whole period, and since I could not refer these to 

 the action of any external causes, it seemed necessary to assume that 

 we have here a case of periodicity due to internal causes of growth. 

 Thus, there was an acceleration between 10.49 and 11.2, and another 

 between 12.6 and 12.25. I have as yet failed to correlate these with 

 any observed phenomenon, but it is clearly a question worth asking, 

 whether the slower periods intervening were not perhaps the periods 

 during which new septa were put into the filaments i.-e., period of 

 cell-division. I was strongly inclined to think that is the case, 

 though I had hitherto been unable to actually satisfy myself by 

 directly observing the phenomenon. It is certain, however, that 

 the filament at the. beginning of the observations had fewer septa 

 than at the end : for at 10.18 there were only 16 septa, whereas 



