168 



second division resulting in the four primary neck cells, which 

 in cross section present the usual quadrant arrangement (Fig. 97). 

 The middle cell next divides by a transverse wall into the 

 primary neck canal cell and the central cell (Fig. 103) from which, 

 subsequently a small ventral canal cell is cut off. The canal cell 

 pushes up between the four primary neck cells, which presently 

 divide by nearly horizontal walls (Fig. 104), so that there are 

 two tiers of neck cells. One or both of these divide again, and 

 each row of neck cells consists of three or four (Figs. 105 — 108). 

 Rarely there may be five cells in one or more of the rows. 



The neck canal cell is very conspicuous, its base being broad 

 and the upper part narrower and extended to the uppermost 

 neck cells. The large and conspicuous nucleus soon divides 

 into two, but as a rule there is no division wall formed. In 

 one case, however, (Fig. 107) there were two distinct neck 

 canal cells. Sometimes both of the nuclei remain in the broad 

 basal part of the cell, and sometimes one is at the base and 

 one nearer the apex. This arrangement seems to depend upon 

 the direction in which the nuclear division takes place. 



The basal cell divides by a vertical wall at about the same 

 time that the primary canal cell is cut off from the central 

 cell. The basal cell subsequently undergoes further divisions, 

 but its limits are readily distinguishable up to the time that 

 the archegonium is mature (Figs. 106, 109). 



In its earlier stages the archegonium of 0. pendulum bears 

 a striking resemblance to that of the Marattiaceae, which the 

 mature archegonium. more closely resembles than it does that 

 of Botrychium. Compared with O.vuk/atum, the neck is decidedly 

 shorter, and this difference is still more marked when com- 

 pared with Botrychium, especially Botrychium virginianum. Even 

 when mature, the neck projects but little above the surface 

 of the prothallium, although there is an elongation of the cells 

 at the time of dehiscence (Figs. 108, 109). 



The ventral canal cell is extremely difficult to demonstrate, 

 and one might be inclined to doulit whether it is formed at 

 all, in some cases, were it not that it is always present in 



