14$ M- W. 111.. \CK.M.\.\. 



ings, although in the preparations there is no difficulty in recog- 

 nizing the true shape of the chromosome, the apparent difference 

 in shape being due to the view-point from which it is seen. 



The five chromosomes of type B form a graded series as regards 

 size, just as with those of type A. The largest one is very per- 

 ceptibly greater than the smallest, and the intermediate ones 

 differ in size to such an extent that there is usually little difficult)" 

 in assigning them to their proper place in the series. No indi- 

 vidual of this type shows any constant precocity or tardiness in 

 division, although in some cells one or more of them are farther 

 along than the others (Figs. 6, 7, 17, J). 



The rod-shaped tetrads (type C) are five in number in Scolo- 

 pendra heros. They show the same constancy in size relation as 

 do the other types, and may be readily arranged in a graded series- 

 Usually one or more of this type of chromosome show the com- 

 ponent parts overlapping each other at an angle or partially 

 wrapped around each other, indicating that they arise from the 

 twisted threads often seen in the prophase and already described. 

 However, these are not constant in occurrence and this condition 

 seems to depend largely upon chance. 



A fact which has proved rather puzzling was brought out by a 

 careful study of the various chromosomes of this type. \Yhile 

 it cannot be doubted that four of the rod-shaped chromosomes 

 divide longitudinally in the first spermatocyte division, the fifth 

 tetrad of this shape apparently divides transversely. In all cases 

 in which this element is well advanced in the metaphase there is 

 a very evident constriction at its middle point, and in some cases 

 this amounts to a nearly complete interruption ol the chromatic 

 material. This is especially evident in Figs. 3, 4, 10, chromosome 

 N. Indeed, it seems hardly possible to escape the conclusion 

 that at the same time the other fifteen bivalent chromosomes are 

 undergoing longitudinal division, this one element is divided 

 reductionally. 



This, however, is no less to be expected than is the behavior of 

 the accessory chromosome in this same division. It differs from 

 the other chromosomes in being univalent (i. e., it has no synaptic 

 mate), while the rest are bivalent. Alter the formation of the 

 spindle it lies among the other chromosomes and is scarcely dis- 

 tinguishable from the rod-shaped ones aside from the fact that 



