CYTOPLASMIC INCLUSIONS 117 



of flagellates from termites. In normal Pseudodevescovina, large num- 

 bers of mitochondria can be demonstrated by the Flemming-Regaud 

 method but, after feeding for three days on filter paper soaked in one- 

 percent Janus green B, no mitochondria can be demonstrated. Addi- 

 tional examples of the lack of mitochondria are furnished by studies of the 

 whole life cycle of certain Protozoa. Horning (1929) was able to demon- 

 strate mitochondria in most stages of Monocystis (Figs. 19, 20), but 

 found that these granules disappear in the sporozoite stage (Fig. 21). 

 Beers (1935) and MacLennan (1936) also found that mitochondria 

 disappear in the later encysted stages of ciliates, although the same 

 methods give positive results in other stages of the cycle. These observa- 

 tions under both normal and experimental conditions demonstrate that 

 mitochondria are neither universal nor permanent cytoplasmic constitu- 

 ents. 



The crucial point in the classification of mitochondria as autonomous 

 organelles is whether they always arise from preexisting mitochondria. 

 The occurrence of dumb-bell-shaped mitochondria and other possible 

 division stages have been found so often in fixed and stained preparations 

 that it is unnecessary to quote this evidence here. The observations on 

 living material are few and perhaps the clearest is that of Horning 

 (1926), who reported division stages of mitochondria in a living 

 heterotrich and was able to confirm the descriptions based on fixed 

 material. Thus division is a factor in the increase in numbers of mito- 

 chondria, but returning again to the studies of the whole life cycle, we 

 find that division is not the only method, since mitochondria must be 

 formed de novo (Figs. 18-21) in those species in which the mito- 

 chondria have disappeared during the quiescent phases of the life 

 cycle. Mitochondria are not self-perpetuating organelles, but are differ- 

 entiations which may endure for a longer or a shorter period during 

 the cycle of the cell. 



In the cases considered above, mitochondria are cytoplasmic in their 

 origin, but this is by no means the only possibility. Joyet-Lavergne 

 (1926) states that the group of mitochondria attached to protein granules 

 (Fig. 27) are derived from the nucleus along with the protein reserves 

 (for discussion of this point, see p. 163), but Daniels (1938) in the 

 same or related species could find no mitochondria attached to the 

 protein granules. Calkins (1930) found in Uroleptus one set of cyto- 



