104 . bulletin: museuji of comparative zoology. 



In stage g (Fig. 83) several of the free loops are clearly much longer 

 than the others, but it is impossible to determine their number on 

 account of their great length and the crowded condition of the nuclear 

 elements. However, there is every reason to believe that their num- 

 ber is the same as that of the large tetrads which appear later, since at 

 a little earlier stage the number of polar loops in the nucleus is un- 

 doubtedly one half that of the spermatogonia! autosomes. In Steno- 

 bothrus I believe the evidence is well nigh conclusive that the large 

 tetrads are formed by the opposite arms of the loops approaching each 

 other, and not by an opening out of the longitudinal split. During 

 stage g the longitudinal split is never very wide and later (stage h), 

 when the definitive tetrads can first be distinguished, each arm has a 

 distinct longitudinal split. In fact there is no time, until the very 

 late prophase of the first maturation division, when the longitudinal 

 split is not plainly discernible. Although the structure of these tet- 

 rads is not well shown at this early stage on account of their large 

 size and the consequent bending, yet they are undoubtedly to be con- 

 sidered modifications of the crossed loops of Dissosteira, the differences 

 in shape being no doubt chiefly due to their much greater size. 



The autosomes rapidly shorten, thicken and become more com- 

 pact, so that the structure of the three larger elements can be easily 

 made out (Figs. 86, 87). They now show plainly their loop-like 

 structure, the arms sometimes being nearly parallel, but more often 

 twisted around each other; the free ends of such twisted autosomes 

 often come together (Figs. 86, 87; Plate 7, Figs. 183-185). In all 

 cases, however, when strongly decolorized each can be seen to be 

 longitudinally split, showing conclusively that the space between the 

 arms separates univalent autosomes and therefore is not the longitudi- 

 nal split. During the late prophase of the first maturation division 

 the large bivalent autosomes become arranged on the spindle with 

 their long axes at right angles to the spindle axis, while the spindle 

 fibers become attached, not at the ends of the arms as in Dissosteira, 

 but at a point some distance from the ends or the middle. This method 

 of attachment of the spindle fibers is of especial interest, since in the 

 spermatogonia in the case of one of the three pairs of large autosomes 

 the spindle fibers are attached at the center of the V-shaped element, 

 while in the case of the other two pairs they are attached at a point 

 nearer one end than the other. Owing to their more irregular shape 

 it is obviously impossible to determine with equal accuracy the attach- 

 ment of the spindle-fibers to the larger bivalent autosomes during 

 the first division, but after careful study I feel convinced that the 



