314 Electrostatic and Electromagnetic Units of Electricity. 



These results can be separated according to the number of 

 discharges as follows: — 



1. 



2. 



3. 



4. 



5. 



300-59 



298-37 



295-73 



296-43 



296-50 



300-17 



298 61 



296-40 



297-24 



296-37 



296-72 



297-43 



298-75 



301-82 



297-38 



297-84 



297-78 



298-66 



295-02 



29687 



298-90 



30019 



296-75 



295-22 



296-31 



29857 











299-05 











30080 











296-56 











298-80 



298-48 



297-26 



297-15 



296-69 



In taking the mean I have ignored the difference in the 

 weights due to the number of observations, as the other errors 

 are so much greater than those due to estimating the swing 

 of the needle incorrectly. 



It will be seen that the series with one discharge is some- 

 what greater than those with a larger number. This may arise 

 from the uncertainty of the correction for the greater number 

 of discharges, and I think it is best to weight them inversely 

 as this number. As the first series has also nearly twice the 

 number of any other, I have weighted them as follows : — 



Weight. dxIO -8 . 



8 298-80 



4 298-48 



3 297-26 



2 297-15 



1 296-69 



Mean 298'15 



Or ^=29,815,000,000 cm. per second. 



It is impossible to estimate the weight of this determination. 

 It is slightly smaller than the velocity of light, but still 

 so near to it that the difference may well be due to errors of 

 experiment. 



Indeed, the difference amounts to a little more than half of 

 one per cent. 



It is seen that there is a systematic falling-ofF in the value 

 of the ratio. This is the reason of my delaying the publica- 

 tion for ten years. 



Had the correction, A, for the number of discharges been 

 omitted, this difference would have vanished ; but the cor- 

 rection seems perfectly certain, and I see no cause for 



