52 Mr. D. Thoday. Experimental Researches on [June 11 



The respective errors in the gain per hour per square decimetre are as 

 follows : — 



Dxiration of experiment. 



Error, assuming slirinkage 

 to the extent of 



1 "1 per cent. 



3 per cent. 





milligrammes. 

 2-9 

 4-8 



milligrammes. 

 4-6 

 7-6 



3 , 





Brown and Escombe's results, with which these estimates are to be 

 compared, may now be given : — * 





Increase in milligrammes per square decimetre per hour. 



Found by dry 

 weight method. 



Calculated from 

 intake of CO2. 



Difference. 





9-8 



1-8 



+ 8-0 



„ 2 



7-1 



1 -8 



+ 5-3 



3 



2-6 



2-9 



-0-3 



4 



7-2 



2-9 



+ 4-3 



5 



Loss of weight. 



3-0 



Greater than — 3 '0 



Since the greatest discrepancy between Brown and Escombe's gasometric 

 and dry weight results is 8 milligrammes per square decimetre per hour, it 

 is probable that the very large and very variable errors found hy them in 

 the dry weight viethod are all explainable as due to asrjmnietry and shrinkage. 

 These will explain negative errors as well as positive : for asymmetry may 

 Jiave a positive or negative effect on the result ; and, as their own experiment 

 shows, changes of area may bo positive as well as negative, although the fact 

 that they are in general more often negative accounts for the usual positive 

 direction of the resultant error. 



Note.— One other fact may be adduced in support of the exi)lauatiou of these large 

 errors by the occurrence of considerable shrinkage. In Table VI 1+ they give tliree 

 experiments with Caialpu hignonioi.des, lasting respectively 7, 5, and 'A^ hours. 



The results per square decimetre jier hour wore as follows : — 



In 7 hours 7-9 inilligramnieB 



5 „ 13-7 



3i „ 16 -6 



* J.on. cil., p. 58, Table VIII. 

 t Loc. <;ii., p. 0(5, 



