370 S. SAGUCHI 



in spite of their continual expenditure in the formation of zymo- 

 gen granules, as will be explained, they never disappear from the 

 cell. Hoven ('10, '12) believes to have found that the chon- 

 riocontes in the pancreas cell either multiply by longitudinal 

 splitting or by elongation of the remainder. So far as I have 

 been able to ascertain, there was no evidence of multiplication 

 by the splitting of the filament; the second assumption of the 

 author is more plausible, yet it is only a hypothesis. 



Observation. In my previous study of the amphibian epi- 

 dermis, I observed that in the physiological degeneration of the 

 glandular cells a certain substance in the nucleus passes, in the 

 form of filaments, into the surrounding cytoplasm, and that these 

 filaments cannot be distinguished by form and staining reactions 

 from the mitochondrial filaments present. I have, further, 

 noticed that, in the retarded formation of Leydig's granules, cer- 

 tain nuclear corpuscles stained deeply by iron-haematoxylin 

 may supplement the above granules. From these observations 

 I have been led to the view that certain constituents of the cyto- 

 plasm are in close relation to the nucleus, and that possibly those 

 which are necessary for cell function are located in mitochondria 

 and derived from the nucleus. In order to solve this question, 

 I made a large number of thin sections from variously fixed ma- 

 terials and searched for possible relation between mitochondria 

 and nuclear contents. It should be noted that this work is not 

 only a matter of difficulty, but also may give rise to misinterpre- 

 tation, so that great care should be taken in judging the findings. 



In close examination of preparations taken from variously fixed 

 materials, such as sublimate-acetic acid, formalin, sublimate- 

 formalin, Zenker's, Meves', Benda's, Champy's, Rubaschkin's, 

 Maximow's, etc., I have often met with a connection between a 

 mitochondrial filament and the nuclear network. In these prep- 

 arations, also, there are many cases where the one end of the chon- 

 drioconte appears to be attached to the outer surface of the 

 nuclear membrane. It would be rash, however, from this appear- 

 ance, to assume the close relation between the two. An actual 

 connection must be, and, in fact, is, formed in such a way that 

 one end of the chondrioconte present near the nucleus continu- 



