723 



Here it appeared that there was no question of a constant; the 

 calculated constant decreased, but so regularly and (after elimination 

 of all possible disturbinij; influences) so very concordantly, that we 

 were able to conclude that this decrease might be traced to a very 

 definite cause. 



The subjoined table shows one of the many measurements. 



Isovaleric anhydride at 25"' .0. 



When the constants obtained in the different measurements were 

 plotted against the time, we could draw through the points thus 

 obtained smooth curves which either coincided or ran completely 

 parallel '), a sign that not only were we not dealing with experi- 

 mental errors, but that the fall must be attributed to a disturbing 

 reaction and, looking at the regularity, to a follow-reaction. 



1) The latter, because the readings did not always commence exactly at the 

 same moment after the anhydride had been dissolved, as the filtration sometimes 

 took a little longer and because the temperature equilibrium in the resistance 

 vessel was not always attained in the same time. 



