482 JEAN REDMAN OLIVER 



tion with the cells of Sertoli, burying their heads in a broad pro- 

 longation of the cytoplasm of these cells, and thus becoming 

 closely packed together. The cell membrane can be traced for- 

 ward over the nucleus in many cases, however, and undoubtedly 

 persists in the adult sperm, forming the semi-permeable limiting 

 membrane described by Koltzoff ('09) as occurring in the sper- 

 matozoa of various species of animals. Over the surface of the 

 future connecting piece this cell membrane is not to be con- 

 founded with the much more strongly differentiated walls of the 

 caudal tube, which stand out in optical section as two curved 

 lines approaching each other distally, and sharply set off from 

 the surrounding cytoplasm. Within the tube the cytoplasm 

 tends to become denser and stains darker than before. At the 

 proximal end the line of attachment of the tube to the nuclear 

 membrane becomes marked by a circular ring-like thickening 

 which stands out clearly, and in optical section as shown in 

 figure 27, presents the appearance of a spherical knob at the 

 junction of tube wall and nucleus. In cross section (fig. 26) 

 the thickened wall of the caudal tube presents an elliptical out- 

 line, near the center of which may be seen the centrosome or 

 the caudal filament, depending upon the location of the section. 

 At about the beginning of the migration of the annulus a marked 

 change is observed in the caudal tube. It becomes detached 

 from its original line of union around the distal end of the nucleus, 

 and undergoes a continuous shrinkage, so that it finally lies 

 below and apparently free from the nucleus (figs. 29 to 32). 

 Another result of this shrinkage is a lateral displacement of the 

 caudal tube, apparently due to the projection of the rodlet against 

 the wall of this structure. When the shrinkage occurs this wall 

 is apparently prevented from approaching the tail filament, while 

 the opposite wall, having no such resistance, approaches the tail 

 filament freely until it nearly reaches it (fig. 32). Shortly after 

 this process of shrinkage is completed the rodlet suddenly dis- 

 appears. What its fate may be is not clear, but it is clearly 

 evident that it takes no part in the formation of the spiral fila- 

 ment surrounding the tail thread, as described by Schoenfeld ('00) 

 for the bull, and by Van Moll^ ('06) for the squirrel. 



