STUDIES ON INSECT SPERMATOGENESIS. 399 



The results obtained do not appear to be susceptible of any logical 

 arrangement, for between species of the same genus, e.g., Thyanta, 

 great differences may occur. On the other hand, in Euschistvs the 

 differences seem to be similar (though perhaps varying somewhat in 

 amount) throughout the series of species examined. It is evident, 

 therefore, that no obvious and constant correlation exists between the 

 occurrence of polymegalous spermatocytes and generic relationships, 

 in which respect these cellular differences parallel the similar lack of 

 correlation between chromosome number and grouping, and external 

 characters which has now been observed in a number of cases. Never- 

 theless, it is a striking fact that among insects these particular differ- 

 ences occur only in this single family, so far as I know, and whatever 

 the cause of their origin may have been, it was doubtless the same for 

 all the cases. Possibly we have to deal with a case of parallel muta- 

 tion. In line with this apparently capricious occurrence of polymegaly 

 is the absence of any conspicuous regularity in the particular lobes 

 affected. On the other hand, there is a curious tendency for the lobes 

 with large cells to occur in pairs, the components of which are separated 

 by a third lobe which often contains the smallest generation of cells; 

 and, further, these large-celled lobes tend to be located on the side of 

 the testis opposite that from which the efferent duct takes it origin. 

 With respect to their comparative morphology, I have noted that the 

 large-celled lobes are frequently narrower than the others (see Text- 

 figs. IB and C), while the lobes with unusually small cells are occa- 

 sionally of exceptional size. There is, therefore, no direct relation 

 between the sizes of the lobes and the cells they contain. 



One of the most interesting points brought out by this sin-vey is 

 that the degree of size difference may vary greatly in different genera, 

 so that a whole series might be arranged running in graded order from 

 the complete absence of this phenomenon, to very exceptional cases 

 of size difference. Thus in Banasa, Acrosternwn, and other genera 

 there seems to be no visible polymegaly; in Murgantia (Text-figs. 

 2-1) the difference, though small, is usually readily distinguishable, 

 but by no means striking; in Etischistus (Text-figs. 2-II) and 

 Brochymena the difference is conspicuous; while, finally, in Arvelius 

 (Text-figs. 2-III) it is truly extraordinary, the whole testis being 

 dominated by the generations of large spermatocytes beside which the 

 normal and smallest generations are completely dwarfed. This last 

 mentioned genus deserves special comment for in Arvelius the phe- 

 nomenon of polymegaly reaches its greatest development. In fact, the 

 volume of the large primary spermatocytes is something like eight 



