— 9 — 



in a telescope mounted upon the graduated arc. To take account of possible 

 heterogeneity of distribution, the magnet was inverted and reversed in each of 

 its positions as determined by the value of r ; so that for any one numerical 

 value of )• there were eight magaetometer readings taken, four with the magnet 

 to the east, and four with it to the west of the magnetometer. 



CORRECTIONS. 



Corrections were applied for temperature, arc of oscillation, torsion of 

 the suspending fibre, and induction on the magnet. 



Temperature correction was applied to the moment of inertia by assuming 

 the coefficient of expansion for steel to be 0.000011 per degree C. In the experi- 

 ment for determining -tt a like correction was applied to the scale reading of the 



distance of the magnet from the magnetometer, the coefficient of expansion for 

 brass being taken as 0.000019. 



The arc of oscillation was measured by means of the image of a scale 

 reflected from the polished end of the magnet. The scale was so graduated as 

 to give arcs in radians by direct reading. 



The torsion of the suspending fibre was determined by turning the 

 torsion head attached to the tapper end of the fibre through five complete 

 revolutions and observing the deflection thereby produced on the reflected 

 image. This gave torsion in terms of the product MH for the magnet used, 

 and the correction was applied accordingly. The mirror magnetometer, being 

 suspended by a spider thread, which proved to have a very small torsional 

 rigidity, was not corrected for torsion. 



The chronometer rate as determined for the pendulum experiments was 

 found to be outside the errors of experiment : and the times of a single 

 oscillation were reduced from sidereal to mean solar seconds. 



To correct the result for the induced magnetism on each of the bar 

 magnets, the induction for a given value of field was determined in the 

 following way. The bar magnet was placed in the same position as 



it was in determining ^ and a solenoid which was about twice as long as the 



H 

 bar magnet was slid over it and the V-groove on which it lay. A known 



current was passed through the solenoid, and the magnetometer reading was 



taken. Thus the field inside the solenoid was known; and 5M the increment 



of the moment of bar magnet could bo calculated. The curve obtained by 



plotting the increments of moment agaijist the field in the solenoid was very 



nearly straight and was quite the same whether the field was increasing or 



decreasing. The maximum field used was 0.6 (C. G. S. unit). From these data 



the induction effect was computed, and the corresponding correction applied to 



the values of H both at 'J'okio and Ogasawarajima. 



The following are the values of II thus determined. 



