DAVIS: SPERMATOGENESIS. 81 
this appearance to be deceptive, and due to the fact that the loops at 
this stage have a great length, which, in some cases at least, is prob- 
ably several times the diameter of the nucleus. ‘This results in their 
being more or less bent upon themselves, so that a single loop may 
extend across the nucleus several times. During the latter part of 
this stage the loops become much shorter, while at the same time the 
nucleus increases markedly in size, so that in most cases the arms of 
the loops are not much longer than the diameter of the nucleus. 
I have been unable to count at this stage the loops in Dissosteira 
but in Hippiscus, where the material is more favorable, the number 
seems to be about one-half that of the spermatogonial autosomes. 
It is impossible to count the loops with perfect accuracy, but I am 
convinced that the number is much less than that of the autosomes in 
the spermatogonia. It thus appears probable that each loop represents 
two univalent autosomes of the spermatogonia which have become 
joined end to end. ‘This interpretation is strongly supported by stage 
b, where, as we have seen, there is good reason to believe that each 
autosome is converted into a single chromatin thread. If, then, we 
imagine these threads to become united in pairs at one end, while the 
other end becomes attached to the nuclear membrane, we should have 
the arrangement of threads seen in stage d. ‘This I believe is the 
manner in which the loops are formed, although it is obviously im- 
possible to demonstrate it, since in the earlier stages the threads are so 
closely crowded and pursue such tortuous courses that it is impossible 
to follow one for any great distance. ‘The structure of the spireme- 
thread during stage d is practically the same as in the preceding stage, 
being made up of a single row of chromatin granules imbedded in a 
linin thread. 
The loops of the spireme are usually attached to the nuclear mem- 
brane at a point adjacent to the greater mass of cytoplasm and to 
the interzonal body. In other words, the attachment is at the distal © 
pole of the nucleus, i. e. at the pole nearest the plane of the last division. 
However, in some cases the point of attachment may be at first as 
much as 90°, or even more, from the distal pole of the cell, but during 
the later stages of the growth period it is almost invariably at this 
pole. Obviously, this implies in some cases a considerable revolution 
of the nucleus, but I do not believe such a revolution can be considered 
general, since in the majority of spermatocytes the loops are attached 
at the distal pole when this polar arrangement first becomes evident. 
It is only fair to say, however, that at this stage it is often difficult to 
distinguish the distal pole on account of the small amount of cytoplasm. 
