304 Phillips on a Comparative Study of the 



protoplast lays down a dense wall which protects it through- 

 out the resting period. The spore wall ruptures when the 

 conditions favorable for growth are restored at the close of 

 the resting period, and the protoplast pushes out, dividing 

 as usual (Fig. 45). 



Another method of multiplication by spore formation was 

 observed upon one occasion. At the time in question, while 

 preparing a life-slide for the continuous observation of 

 Oscillaria, a trichome of unusual appearance came into the 

 field of the microscope. The end, or cap, cell, from which the 

 finger-like hairs were protruding, was in place (Fig. 47), and 

 the third cell was perfectly normal, apparently having given 

 up none of its chromatin as in spore formation. The cell 

 between these two, however, — the second cell of the fila- 

 ment, — had every appearance of a spore with the exception 

 of the heavy wall. The trichome was actively moving for- 

 ward, which is not the case in plants that are forming, or 

 have formed, spores in the manner described above, the 

 cilia having been withdrawn and the end hairs absorbed. 

 While observing this trichome, the end cell and the adja- 

 cent enlarged spore-like cell separated from the rest of the 

 filament and moved away slowly by means of the slow 

 crawling movement of the finger-like processes. These pro- 

 cesses were finally absorbed (Fig. 46), and the spore-like 

 cell began to divide, the dense, diffused, nucleus-like centre 

 separating into halves (Fig. 46) and a dividing wall grow- 

 ing between them. Division continued until a six-celled 

 stage was reached, by which time the cilia were formed and 

 it could swim about freely in the water of the life-slide. On 

 account of these movements it could no longer be kept under 

 constant observation, but development had progressed far 

 enough to show that it was forming into a new trichome. 



In Cylindrospermum the process of spore-formation is 

 somewhat different, but follows the same general lines. 

 Usually the end cell of a filament which has no heterocyst 

 becomes strongly chromatic, probably through the chroma- 



