SPERMIOGENESIS OF THE PRIBILOF FUR SEAL 



483 



\fter the disappearance of the rodlet still further shrinkage 

 takes place, especially at the more posterior part of the con- 

 necting piece, with the result that the wall of the caudal tube 

 closely parallels the caudal filament but a short distance from it, 

 save at the anterior end where a marked swelling indicates the loca- 

 tion of the proximal group of the posterior centrosomes (figs. 34 

 and 35) This enlargement gradually becomes eliminated m later 

 stages and the connecting piece finally assumes its uniform cylin- 

 drical shape in the epididymis during the fourth period. The 

 cell membrane covering the caudal tube, has fused mdistingmsh- 

 ably with the latter, which terminates anteriorly at the beginmng 

 of the connecting piece. The membrane, however, is contmued 

 forward over the neck region and is lost to sight in its mtimate 

 fusion with the nuclear wall and with the perforatormm m front. 

 Within the connecting piece the cytoplasm increases m density 

 and appears more granular, while the tail filament also becomes 

 thicker and more strongly defined. These granules stain with 

 the Benda method and are apparently of a mitochondrial nature 

 but the fixation of the material did not permit the tracing of 



their complete history. . a .. a n-. 



The nucleus has progressively condensed and flattened, its 

 staining capacity with respect to iron hematoxylin has changed 

 from that of the previous period. The deep black colon 

 rapidly differentiated and the whole nucleus becomes a light 

 bluish gray, the darker transverse bands or zones of Valentin, 

 described in detail by Ballowltz ('91) for other forms, becoming 

 clearlv visible in most instances. 



When last considered in the previous period, the centrosomes 

 consisted of two parts; first, an anterior centrosome fused with 

 WucLr wall, from which the rodlet projected out obliquely 

 nto the cytopla m surrounded by the caudal tube, and, second, 

 he posterL centrosome, which had also divided into two parts 

 a proximal and a distal. The distal half soon starts on its 

 migration down the tail filament. The distinct ring form can- 

 not br^a^^^ out since it fits closely to the axial filament upon 

 which it appears more as a bead upon a thread than as a loosely 

 S annulus. The migration is quite rapid apparently, as but 



