338 R. R. BENSLEY 



animals of the same age but of a different litter and much lower 

 in weight. The neutral red technique did not prove so satis- 

 factory in dogs as in the guinea pig, for the reasons that, although 

 the islet granules stained intensely in the dye, the large amount 

 of zymogen present and the greater size and solidity of the lobules 

 of the gland rendered study by the teasing method more difficult. 

 Moreover, in the dog I found it more difficult to secure a uniform 

 staining of the whole pancreas than in other animals. Later a 

 satisfactory method of counting for dogs was worked out, using 

 Ringer's solution as the solvent and janus green as the stain. 

 Teasing the pieces of the pancreas having proved unsatisfactory, 

 the method of cutting thick sections of the pancreas with a Walb 

 double-bladed knife having the blades separated 2 mm. was 

 adopted. This method, I believe, permits a fairly accurate esti- 

 mation of the islet content, if the pieces are collected and weighed 

 after the counting and a ratio established between number of 

 islets and unit weight. Unfortunately this method was devel- 

 oped too late in the work to permit of its application to more 

 than one series of dogs of the same age. This series (no. 2), 

 however, I believe, gives a fairly accurate idea of the frequency 

 of islets in the pancreases in question. I do not feel the same con- 

 fidence in the result's offered in series 1, which were counted by the 

 neutral red technique, as regards the actual content of islets, but 

 as the several pancreases in this list were counted under the same 

 conditions the relative error is probably about the same and the 

 series may be accepted as showing the relative numbers of islets 

 of Langerhans in the several pancreases, though probably much 

 below the actual numbers per milligram in each case. In any 

 case, there was no difficulty in seeing and counting the large islets 

 and if Vincent and Thompson's claims that the islets were rhuch 

 increased in size as well as number were true, the counts should 

 show this to be the case. As the period of Vincent and Thomp- 

 son's experiments was not known, the duration of the first series 

 of experiments was ten days, but in the second series, in view of 

 Laguesse's statement that five days was the optimum period for 

 producing the result, this period only was allowed to elapse from 

 the time of withdrawal of food to the time of counting the islets. 



