26 MR. J. H. POYNTING ON THE 



proportional error in the observed time made a large error 

 in the value of M^I^; and the following method, that 

 usually adopted in magnetic observations, was employed in 

 preference. A stirrup was suspended by a platinum wire, 

 and its time of vibration [t,) against the force of torsion 

 {fi) of the wire was observed. The moment of inertia of 

 the stirrup being S, we have 



a_'7r b 

 t, — 



The time of vibration (t^) was then observed when a cylin- 

 drical brass bar of known moment of inertia (B) was in- 

 serted in the stirrup. We now have 



The bar was then removed and the balance-beam inserted 

 in its place ; and the time of vibration (^j) gives 



/-=-(s+Mr). 



* /u, 

 From these three equations, eliminating S and /a, we obtain 



Now B^ was calculated from the weight and dimensions 

 of the bar to be 6332-83 (in centimetres and grammes). 

 The observed times were ti = S'(>79'2.% ^i = 4*495', t^ = 

 7•I483^ From these values we find 



M^r=3565i-6* 



* To this a small correction skould be added if the adjusting-bob is not in 

 its lowest position. This amounts to 7*6 for each tui'n of the screw, and may 

 therefore in general be neglected. 



