No. 2.] EPITHELIUM OF DRAGONFLY NYMPHS. 105 



lowed by paracarmine for staining in toto, or by haematoxylin 

 (Gage's) and eosin for staining on the slide. For differentiat- 

 ing functional from nonfunctional parts, Hermann's fluid was 

 used for fixing without subsequent staining. Numerous prep- 

 arations were made from each of the three species named, at 

 frequent intervals from one to eight hours after feeding. Serial 

 sections of the whole of each preparation were mounted and 

 studied. 



I. THE ACCUMULATION OF THE DIGESTIVE SECRETION. 



A well-known type of digestive epithelium in insects consists 

 of more or less elongated cylindric cells which rest upon the 

 membrana propria and bear upon their free inner ends a strongly 

 refractive striated border ; interspersed among these cylindric 

 cells are little groups or nidi of small, roundish cells close to 

 the membrana propria, having no communication with the 

 digestive cavity. The cockroach and other Orthoptera beauti- 

 fully exemplify this type. The epithelium of dragonfly nymphs 

 differs from it only by the closer approximation of the nidi and 

 the crowding and elongation of the cells between them. Fig. 2 

 represents the epithelium in its normal resting condition. The 

 digestion of a meal has recently been performed, and the cells 

 have resumed their habitual aspect. The cylindric cells clearly 

 reach the basement membrane. The cells of the nidus are 

 few. The striated border is complete and undulating. 1 



Fig. 3 shows the condition found in a nymph after fasting 

 two weeks. Midway between the nidi the cells have become 

 very much crowded together, their nuclei are flattened as if by 

 pressure, they rise in prominent elevations with very turgid 

 summits from which the striated border has disappeared. The 

 cells of the nidus have also increased in number. An accumu- 

 lation of granular secretion in the cells between the nidi and 

 their protrusion above the general level are very evident. 



I believe that all these cells retain their connection with the 

 basement membrane, although I was unable to trace them to 



1 Compare the single figure by Sadones for Libellula depressa L., loc. cit., 

 PI. II, Fig. 1 6. 



