66 bulletin: museum of compakative zoology. 



This arrangement is the one most commonly met with, occurring, as it 

 does, in a large proportion of the cells at this and later stages. It is evi- 

 dent that sections cut in planes other than that of the three parts of the 

 centrosome might give rise to different appearances. In many cells the 

 plane of the section is perpendicular, or nearly so, to the plane just de- 

 fined, so that in focussing one encounters, in succession, a centrosome, 

 the proximal end of the axial filament, and the second centrosome. 

 Figure 86 represents one of the very few cases in which the centrosomes 

 bear a different relation to the filament. Here the joining line is nearly 

 parallel instead of perpendicular to the filament. It is probable that in 

 these cases the centrosomes assume the usual relation later. 



In a number of cases I have observed very fine granular threads (Figs. 

 87, 90) that arise from the two lateral centrosomes and seem to join the 

 axial filament at an acute angle at some distance from its proximal end. 

 This condition is to be correlated, I think, with the disposition of the 

 centrosomes and their relation and connection with the axial filament 

 observed at much later periods of transformation. 



In Figure 87 are shown three centrosomes distinct from the axial fila- 

 ment. Two of these occupy the usual position, while the third lies at an 

 equal distance from the proximal end of the filament in the direction of 

 its prolongation. All are connected with the filament by fine granular 

 fibres as shown in the drawing. At the end of the filament proper there 

 is a slight enlargement indicating the presence of a fourth part of the 

 centrosome. In Figure 89 the centrosome fragments are still more 

 numerous. Here there are six entirely free from the filament, three 

 arranged as in Figure 87, two others forming a pair distal to the normal 

 pair, and the sixth still more distal in position. These together with 

 the knob-like end of the axial filament make the total number seven, the 

 largest I have found. 



As to the significance of this multiplication of centrosomes, I can say 

 nothing, except that in later stages only three are present, the two lateral 

 bodies and the central one which forms the enlargement at the end of the 

 filament. It should be observed, however, that when a larger number of 

 centi-osomes are present they are smaller, as will be readily seen by a com- 

 parison of the figures. This is especially true of the central one, which 

 usually is quite conspicuous, but in Figures 87 and 89 is much smaller. 



The elongation of the axial filament proceeds until the distal end has 

 reached the vicinity of the cell membrane (Fig. 91), when the lengthen- 

 ing of the cell itself begins. When it first appears, as we have already 

 noted, the axial filament is usually tangential to the surface of the 



