ON THE MEASUREMENT OF THE LUNAR DISTURBANCE OF GRAVITY. 101 



of leaving the doors open a good deal, in order to permit the cord to pass 

 by which Lord Rayleifjh was spinning the British Association coil. 



Notwithstanding that the weather was sultry the warping of the 

 stone coliimus mast have been very slight, for a thermometer hung close 

 to the pier scarcely showed a degree of change between the day and 

 night, and the difference of temperature of the IST. and S. faces must have 

 been a very small fraction of a degree. At that time, however, we still 

 thought that the whole of the diurnal oscillation was due to the warping 

 of the columns. 



We next tried a series of experiments to test the sensitiveness of the 

 instrument. 



As above remarked the image of the scale was continually in motion, 

 and moreover the mean reading was always shifting in either one direc- 

 tion or the other. At any one time it was possible to take a reading to 

 within -iVth of a foot with certainty, and to make an estimate of the 

 -p\^th of a foot, but the numbers given below are necessarily to be 

 regarded as very rough approximations. 



As above stated, the gallows faced about to the S.E., and we may 

 describe the two square piers as the E. and W. piers, and the edges 

 of each pier by the points of the compass towards which they are 

 directed. 



On August 14, 1880, my bi'other stood on a plank supported by the 

 pavement of the room close to the S.W. edge of the W. pier, and, light- 

 ing a spirit lamp, held the flame for ten seconds within an inch or two of 

 this edge of the pier. The effect was certainly produced of making the 

 pendulum-bob move northwards, but as such an effect is fused in the 

 diurnal change then sroing: on, the amount of effect was uncertain. He 

 then stood similarly near the N.E. edge of the E. pier, and held the 

 spirit flame actually licking the edge of the stone during one minute. 

 The effect should now be opposed to the diurnal change, and it was so. 

 Before the exposure to heat was over the reading had decreased '15 feet, 

 and after the heat was withdrawn the recovery began to take place 

 almost immediately. "VVe concluded afterwards that the effect was equi- 

 valent to a change of horizon of about 0"'15. 



When the flame was held near but not touching the lintel for thirty 

 seconds, the effect was obvious but scarcely measurable, even in round 

 numbers, on account of the unsteadiness of the image. 



When a heated lump of brass was pushed under the iron box no 

 effect whatever was perceived, and even when a spirit flame was held 

 so as to lick one side of the iron box during thirty seconds, M^e could not 

 be sure that there was any effect. We had expected a violent disturb- 

 ance, but these experiments seemed to show that convection currents in 

 the fluid produce remarkably little effect. 



When a pull of 300 grammes was delivered on to the centre of the 

 lintel in a southward direction, we determined by several trials that the 

 displacement of the reading was about '.SO feet, which may be equal to 

 about 0"'.3 change of horizon. 



Two-thirds of a watering-can of water was poured into the ditch at 

 the back of the pier. In this expei'iment the swelling of the ground 

 should have an effect antagonistic to that produced by the cooling of the 

 back face of the pier, and also to the diurnal changes then going on. 

 The swelling of the ground certainly tilted the pier over, so that the 

 reading was altered by '10 foot. A further dose of -^^'ater seean.ed to have 



