191*] Snook-Long: Parasynapsis in Ancides luguhris 517 



of nuclei usually found in this condition. If the number of 

 amphitene cells can be taken as an indication of the duration of 

 the period, it must be considered as relatively short, because the 

 cells occupy a much smaller proportion of the testis than either 

 the spermatogonia or tlie cells that have advanced to the next 

 stage. 



Posterior to the region in the testis last described are imclei 

 containing horseshoe-shaped loops, with their free ends turned 

 towards the sphere and their bends in the distal half. They appear 

 somewhat granular (pi. 25, fig. 8 find pi. 26. fig. 10). These loops 

 have been described in a number of urodeles and would appear 

 to be very characteristic of spermatogenesis in salamanders. In 

 thickness the.y exceed somewhat the thick threads formed by the 

 union of two fine diverging ones. No trace of the latter can now 

 be seen. Cross-sections through the proximal portion of the 

 nucleus show clearly twenty-eight cut ends, demonstrating the 

 number of loops to be fourteen (pi. 26, fig. 11). Polar views 

 show the presence of twenty-eight converging threads with their 

 free ends crowded together in the immediate vicinity of the 

 sphere and very near to the nuclear membrane (pi. 26, fig. 9). 



It will be remembered that this polar orientation is also a 

 characteristic of the preceding stage, a lateral view showing the 

 same arrangement of the thick threads forming the stem of the 

 Y-flgiire as is here presented by the ends of the loops. Thi.s 

 resemblance is very marked, indeed, and together with the evi- 

 dence afforded by seriation leads to the conclusion that the loop 

 is derived from two fine threads which have now completely 

 disappeared as separate threads and have fused into one. 



Up to this point the cells have been gradually increasing in 

 volume, a fact which makes seriation more certain; but during 

 the stage of well-developed loops relatively little growth takes 

 place. As the cells in this condition usually occupy a rather 

 large portion of the testis, it seems probable that this period lasts 

 for a much longer time than the amphitene stage. 



Later all evidences of polarization are lost and the chromo- 

 somes end without apparent relation to the position of the centro- 

 somes or to each other. At the same time a longitudinal split 

 develops, dividing each loop more or less completely into two 



