GLANDULAR CELLS OF THE FROG'S PANCREAS 371 



ously passes over to a cord of the nuclear network, which, in 

 turn, is attached either to the nucleolar corpuscles or to the nu- 

 cleolus (figs. 44 to 55, 71). In addition to those shown on the 

 plates accompanying this paper, many other instances of such 

 connection could be observed. Preparations fixed in formalin 

 or Meves' fluid, and stained with iron-haematoxylin are most 

 favorable for the study, since both the mitochondria and the 

 cords of the nuclear network are deeply stained and the continu- 

 ity of the two can easily be followed. Since this continuity can- 

 not be established with certainty in a surface view of the nucleus, 

 we must make cross-sections. Although it cannot be said that 

 the connection can always be detected in such sections, it is, 

 however, not rare that there are seen two or three connections 

 in one and the same nucleus (figs. 48, 52). 



From the above observations the conclusion seems justified 

 that the chondriocontes of the pancreatic cell are derived from 

 the nucleus, by a certain nuclear substance passing out through 

 the nuclear membrane into the cytoplasm. The opposite condi- 

 tion is not conceivable, for, there can be little doubt that the 

 chondriocontes, as will be afterwards described, are used up in the 

 formation of zymogen granules. Several objections might be 

 raised to the preceding considerations. First, if the chondrio- 

 contes pass over the nucleus, an appearance of a connection of 

 the former with the nuclear network might be produced when 

 viewed from the surface. I exclude such uncertain cases from 

 consideration, drawing my conclusion only from the observa- 

 tions made upon the cross-sections of the nucleus. Secondly, 

 mitochondrial filaments might be brought in contact with the 

 nuclear membrane accidentally or be superposed upon the nu- 

 clear area, by the action of the knife in cutting. I do not, of 

 course, profess to draw the above conclusion from such an 

 observation; the picture of the chondriocontes passing over to 

 the cords of the nuclear network must be continuous in order 

 to prove that the connection actually exists. In addition to the 

 direct evidence mentioned above of the nuclear derivation of 

 the mitochondrial filaments, there is another, indirect: that 

 mitochondria are never lacking near the nucleus, even when 



