DAVIS: SPERMATOGENESIS. 83 



more marked than at any other time. Figure 33 shows a spireme as 

 seen from the pole. It is evident that the loops all tend to converge 

 at a single point, where they are attached to the nuclear membrane. 

 In reality, the ends of all the loops do not come together at a com- 

 mon point, but some unite with each other a short distance from the 

 point of attachment, so that the threads often appear to branch. It is 

 almost impossible to make out the exact arrangement of the threads 

 near the point of attachment, since they are so closely crowded to- 

 gether (often overlying one another), and also since in this region the 

 chromatin granules are usually much larger, and distributed along 

 the linin thread at much longer intervals than elsewhere. In such 

 cases it is often very difficult to follow the lightly staining linin thread. 



The next stage (/) marks the end of the growth period (Plate 3, 

 Fig. 46). The spermatocytes have now reached their greatest size, 

 both as regards the nucleus and the cytoplasm, while the chromatin 

 has become more diffuse than at any other stage. Judging from the 

 number of cells found in this condition, this is the longest of the 

 spermatocyte stages. 



The transition from stage e to stage /is gradual and consists chiefly 

 of an opening out of the loops, so that most of them come to lie close 

 under the nuclear membrane. The result is that in most cases the 

 polarity of the spireme seems to have been lost and the whole appear- 

 ance suggests that there is a continuous spireme extending around the 

 periphery of the nucleus. Farmer and Moore (:05) have, indeed, held 

 that such is the case in Periplaneta, but in the Orthoptera which I 

 have examined I do not believe there is a continuous spireme at any 

 time. In fact, during stage /, when in most cases there appears to be 

 no trace of the former polarity, the loops for a considerable time at 

 least retain their original attachment to the nuclear membrane. In 

 nuclei which are favorably oriented so as to afford a view of the distal 

 pole, the ends of the loops can still be seen to be attached to the nuclear 

 membrane as in the preceding stages,^ but owing to the tortuous course 

 of the threads around the periphery of the nucleus, all trace of the 

 radiate arrangement is lost except in the immediate vicinity of the 

 distal pole. In some species of Acrididae, where the polarity is much 

 more marked, it is plainly distinguishable even at this stage. 



Concurrently with the opening out of the spireme loops the longi- 

 tudinal s])lit becomes indistinct, and the thread seems to stain much 

 less deeply. This appears to be due chiefly to the chromatin granules 

 becoming broken up into much finer ]iarticles, which are irregularly 

 distributed along the linin thread. At the same time the amount of 



