362 HARVKY M. SMITH. 



respiring animal in the jar. This factor has already been shown 

 to be overcompensated by the loss resulting from incomplete 

 absorption of the carbon dioxide by the barium hydroxide. 

 Inasmuch as the first two factors are important, they would tend 

 to neutralize this loss. It is therefore concluded that the method 

 is sufficiently accurate to allow for making comparative carbon 

 dioxide output determinations on the animals used. 



Indirect evidence that the factors discussed above are not 

 important in producing error is obtained by comparing the 

 results obtained on different animals. Errors caused by the 

 passing through of excess carbon dioxide would tend to increase 

 the apparent result for small animals more than for larger ones, 

 while for small animals the error resulting from the loss of carbon 

 dioxide that was not absorbed would tend to be minimized. 

 With these ideas in mind, if one looks at the results for Rana 

 palustris (Table I.) it is apparent that Nos. 6-17 have within 

 the limits of variation the same carbon dioxide output result 

 as No. 5, although the latter has twelve times the chance of 

 being thrown off by the errors as have the former. Again Bufo 

 americanus shows only slightly higher results for a 1.9 gm. 

 individual as for an n gram one, and Rana clamitans shows a 

 lower result for the lighter animal. 



V. COMPLICATING FACTORS. 



A. Cell Size. 



i. If the figures presented in this paper are compared \vith 

 those obtained by other workers, it will be found that they run 

 decidedly low. Measurements are presented for the same species 

 only in the case of Amphiuma and Cryptobranchus, but the indica- 

 tions are in the same direction for the frogs and toads, in which 

 different species have been used here than those used by other 

 workers. There are two possible causes for this difference; first 

 that the writer has obtained shrinkage of the corpuscles, as 

 previously pointed out, and second that the measuring technique 

 used is faulty. Several attempts have been made to check the 

 latter point. The scale used has been repeatedly compared with 

 a Zeiss stage micrometer, and found to be accurate. By means 

 of an ocular filar micrometer the widths of the different 10 micron 

 divisions of the Xeiss stage micrometer have been measured. 



