34 CALIFORNIA ACADEMY OF SCIENCES. [3D Ser., 



ent they form a triangle (figs. 52, 58^) with a position rela- 

 tive to each other very much like that of the centrosomes 

 of the lymphocytes, according to the diagrams of Heiden- 

 hain (see his " Neue Untersuchungen," figs. 3, 25, etc.), or 

 they may simply lie in a half circle, as in figs. 52 and 65. 

 In one plasmocyte I found four centrosomes lying in a square 

 (fig. 60), connected by a film of somosphere. In others 

 the arrangement was less regular and the centrosomes were 

 placed at different depths, one above the other. At no time 

 have I observed more than four centrosomes in the same 

 microcentrum. The somosphere is either diffuse, spherical, 

 crescent-shaped or ringlike. The diffuse and spherical 

 somospheres are always homogenous or very finely granu- 

 lated, while the crescent-shaped or ringlike somosphere is 

 seen to more or less perfectly enclose one or more highly 

 refractive yellowish bodies. When the somosphere is ring- 

 like it is always found to be wider at one point from which 

 it tapers in both directions towards the opposite sides. In 

 the thickened and crescent-shaped part are found the minute 

 centrosomes. When more than one is present they are al- 

 ways found close together and frequently so approximated 

 that only the most delicate manipulation of the light will 

 show them to be separate from each other and from the 

 somosphere. 



Between the crescent -shaped and the ringlike somo- 

 sphere there are numerous intermediate links, the extreme 

 forms being of equal frequency. These forms are undoubt- 

 edly due to the enclosures mentioned above. These are of 

 a rather solid nature, and being always round they cause 

 the somosphere to assume the shape of a crescent or ring; 

 the former if the sphere is small and cannot compass the 

 globule; the latter if it is larger and can extend all around 

 it. That these globules constitute a food supply, perhaps 

 derived from the granosphere, will presently be mentioned. 

 (See figs. 81, 83, 88.) In corroboration of this is the fact 

 that when the somosphere is crescent or ring-shaped the 

 centrosomes appear in greater activity, actually in the process 

 of budding (fig. 83). By budding I do not necessarily im- 



