368 HARVEY M. SMITH. 



understand the meaning of the abnormally low carbon dioxide 

 production. All the individuals in the low group except No. 14 

 were used for a number of trials; 14 died after 105 days of starva- 

 tion when only one trial had been made. Xos. 2 and 8 are 

 females; 6 and 14 males. Females usually run lower than 

 males (see below) but this explains only two of the cases. The 

 detailed record of No. 6 shows a low initial period, a fluctuating 

 intermediate period, and a very high final one as starvation 

 progressed. The final period was characterized by a rapid loss 

 of weight. 



In the other species studied the same inverse ratio between 

 body weight and metabolism is evident. \r\Amphiuma the result 

 is strikingly higher for the smallest animal, and unusually con- 

 stant for the three larger specimens. In the frogs the same 

 trend may be observed, but it is not as much in evidence, probably 

 because of the preponderance of other complicating factors. 



These observations suggest that weight is not a thoroughly 

 satisfactory basis for computing basal metabolism in these ani- 

 mals. The ideal basis on which to make such computations 

 would be mass of respiring tissue in the body. In Amphibia 

 this mass would be less in proportion to the total body weight 

 in Salientia than in Caudata, due to the greater mass of bone in 

 the former. This proportion would also be smaller in large 

 animals than in small individuals of the same species, because of 

 the increased ossification and connective tissue in the former. 

 In man it has been found that the body surface is the more 

 reliable criterion, and elaborate formula? have been worked out 

 for computing this surface from the weight and height. In 

 Amphibia such formula? would be useless, on account of the great 

 variation in shape which is found between the various species. 

 It appears that the most hopeful method of eliminating weight 

 variations in comparing different species as to carbon dioxide 

 output is to choose animals from each species which have ap- 

 proximately equal weights. In the final section of this paper an 

 attempt is made to do this. 



2. Starvation is a factor leading to important variations in 

 the results, especially with Necturus. The rate of carbon dioxide 

 elimination increases as starvation progresses. This fact has 

 already been indicated by the observations on Necturi with 



