102 IIKPORT— 1881. 



the same effect, and it took more than an honi* for the piers to regain their 

 former position. As the normal diurnal change was going on simulta- 

 neously, we do not know the length of time during which the water con- 

 tinued to produce an effect. 



On August 15 we tried a series of experiments with the disturber. 

 When the disturber was displaced on its rails, the penduluna took a very 

 perceptible time to take up its new position, on account of the A^scosity 

 of the fluid in which it was immersed. 



The diurnal changes Avhich were going on pi'evented the readings 

 from being very accordant amongst themselves, but we concluded that 

 twenty-five turns of the screw gave between '4 foot and '3 foot alteration 

 in the reading on the scale. From tlie masses and dimensions of the 

 pendulum and disturber, we concluded that 1 foot of our scale corre- 

 sponded with about 1" change in horizon. Taking into account the 

 length of the pendulum, it appeared that 1 foot of our scale corresponded 

 with y-4:\jyth of a mm. displacement of the spike at the bottom of the 

 pendulum. Now as a tenth of a foot of alteration of reading could be 

 perceived with certainty, it followed that when the pendulum point 

 moved through T^^-fnT^li of ^ mm. we could certainly perceive it. 



During the first ten days the mean of the diurnal readings gradually 

 increased, showing that the pendulum was moving northwards, until the 

 reading had actually shifted 8 feet on the scale. It then became neces- 

 sary to shift the scale. Between August 23 and 25 the reading had 

 changed another foot. We then left Cambridge. On returning in 

 October we found that this change had continued. The mixTor had, 

 however, become tarnished, and it was no longer possible to take a 

 reading, although one could just see a gas flame by reflection from the 

 mirror. 



Whilst erecting the pendulum we had to stand on, and in front of, 

 the piers, and to put them under various kinds of stress, and we always 

 found that after such stress some sort of apparently abnormal changes in 

 the piers continued for three or four hours afterwards. 



We were at that time at a loss to understand the reason of this long- 

 continued change in the mean position of the pendulum, and were reluc- 

 tant to believe that it indicated any real change of horizon of the whole 

 soil ; but after having read the papers of MM. d'Abbadie and Planta- 

 mour, we now believe that such a real change was taking place. 



By this course of experiments it appeared that an instrument of the 

 kind desci'ibed may be brought to almost any degree of sensitiveness. 

 We had seen, however, that a stone support is unfavourable, because 

 the bad conductivity of stone prevents a rapid equalisation of tempera- 

 ture between different parts, and even small inequalities of temperature 

 produce considerable warping of the stone piers. But it now seems 

 probable that we exaggerated the amount of disturbance which may 

 arise from this cause. 



A cellar would undoubtedly be the best site for such an experiment, 

 but unfortunately there is no such place available in the Cavendish 

 Laboratory. Lord Rayleigh, however, placed the ' balance room ' at 

 our disposal, and this room has a northerly aspect. There are two 

 windows in it, high up on the north wall, and these we keep boarded up. 

 The arrangements which we now intended to make were that the 

 pendulum and mirror should be hung in a very confined space, and 

 should be immersed in fluid of considerable viscosity. Tiie boundaiy of 



