1894] on the Newtonian Constant of Gravitation. 371 



best sets of four points of rest obtained from six elongations. They 

 were : — 



Worst set 

 + position 



24493 

 24493 

 24493-5 

 24492 

 (24491-7)* 



24492-9 mean. 



Best set 

 — position. 



20795-4 

 20795-7 

 20795-5 

 20795-5 



20795-5 mean. 



Taking all the mean points of rest, as determined above, in groups 

 of three to eliminate slow shifting, if any, of the points of rest, the 

 series of deflections were : — 



3696-5 



3696-6 



3696-0 



3696-8 



(Interval of one hour, in which oscillations of large amplitude were 

 observed for period.) 



(3697-7) 



3696-0 



Immediately after the oscillations of large amplitude, which in 

 this case at the end were rather badly disturbed by trains or other- 

 wise, a rather different deflection was observed, but not seriously 

 different. As examination of the figures shows only one anomalous 

 point of rest immediately after the large amplitude disturbance, I feel 

 justified in rejecting the only discordant figure, and in taking the 

 mean of the rest as the true deflection. The unit in this case is 

 1/10 division. It corresponds to an angular movement of 1/280,000, 

 i. e. about three-quarters of a second of arc. Now a calculation of 

 the angular twist due to a rotation of the air based upon the period, 

 the moment of inertia, and the logarithmic decrement, shows that if 

 the air in the tube were made to whirl round at the rate of one turn 

 in six weeks, so that the air would blow past the gold balls at the 

 rate of one inch in a fortnight, the deflection produced would be 1/10 

 division, an amount which is greater than the uncertainty of the 

 deflection on this particular night. Again, an examination of the 

 points of rest through the night in the positive and in the negative 

 positions shows a very small steady creep, the same in each case. 

 Creepage of this sort has been, I believe, mentioned as a defect of 

 quartz fibres. When it gives trouble it is due to draughts, as already 

 explained, or to imperfect attachment of the fibres.* In the present 

 instance the creepage observed corresponds to a surface rate of 

 movement on the fibre of a millionth of an inch a month. 



* Disturbed by trains. 



t See my paper on 'Attachment of Quartz Fibre,' Phil. Mag. May 1894. 



