24 Mr. E. H. Griffiths and Miss Marshall on the 



The method of removing and weighing the empty dropper 

 has already been indicated. 



17. Remarks on the Experiments. 



We give particulars of all our experiments with two excep- 

 tions. In No. XXI. a. the chronograph ceased to work during 

 a critical portion of the experiment, and we were thus unable 

 to determine £ E . In XXIV. a. Observer I. omitted to close the 

 entrance tap to the evaporating-flask before connecting with 

 the exhaust, thus the attempt to diminish the vacuum was a 

 failure and the experiment was relinquished after a minute or 

 so. In these two cases we at once performed other experi- 

 ments to replace the failures, but retained the numbers for 

 convenience of reference. 



We have, however, rejected, when drawing our final con- 

 clusions, several of the experiments whose details are given. 

 Such experiments are marked by a f in the Tables. We 

 have, in no case, rejected any experiment except as a conse- 

 quence of some note made during that experiment; that is, 

 before reducing the observations we had already decided as 

 to those which should be regarded as of little value. Thus 

 we have in no way been guided by the results. For example, 

 XX Y. differs from the mean of its group by a greater 

 quantity than either of the rejected experiments at that 

 temperature : however, as we have no note against it, we 

 are compelled to give it equal importance with any of the 

 others. A simpler plan would have been to reject entirely 

 the t experiments, but we think that a fairer idea of the 

 general accuracy of the work is given by including all those 

 that we completed. 



The reasons for the rejection of the f experiments are 

 given (as they appeared in our original notes) at the end of 

 each group. We have, however, in each case given the 

 mean of all in a footnote, and it will be seen that (except at 

 30°) our results would not have been appreciably affected by 

 the inclusion of those rejected ones. 



As a rule, we found that the thermal balance was most 

 easily maintained when the potential-difference was that of 

 4 cells, but this supply of beat would have been somewhat too 

 great at 20°, when we worked with 3 cells only. At all 

 temperatures above 20° we performed experiments with 3, 4, 



