100 itEPORi— 188i. 



spike at the bottom of the pendulum tlirougli O'OOOl mln. or o^sVwTr^h of 

 an inch. The same displacement would be produced by an alteration in 

 the direction of gravity with reference to the earth's surface by jV^h of a 

 second of arc. 



A rough computation showed that the to and fro motion of the 

 pendulum in the N.S. azimuth, due to lunar attraction, should, if the 

 earth be rigid, be the same as that produced by 2f turns of the micro- 

 meter screw. 



We now return to the other arrangements made in re-erecting the 

 instrument. 



A new mirror, silver-ed on the face, was used, and was hung in a 

 slightly different manner. 



The fluid in which the pendulum was himg was spirits and water. 

 The physical properties of such a mixture will be referred to later. In 

 order to avoid air-bubbles we boiled 3| gallons of spirits and water 

 for three hours in vacuo, and the result appeared satisfactory in that 

 respect. 



After the mirror was hung, the plate-glass front to the box was fixed 

 and the vessel was filled by the tap in the back of the box. The disturber 

 was not introduced until afterwards, and we then found that the pen- 

 dulum I'esponded properly to the disturbance. 



As the heat of a lamp in the neighbourhood of the piers exercised a 

 large disturbance, we changed the method of observing, and read the 

 reflection of a scale with a telescope. The scale was a levelling staff 

 divided into feet, and tenths and hundredths of a foot, laid horizontally 

 at 15 feet fx-om the piers, with the telescope immediately over it. 



Since the amount of fluid through which the light had to pass was 

 considerable, we wei'e forced to place a gas- flame immediately in front of 

 the scale ; but the gas was only kept alight long enough to take a 

 reading. 



After sensitising the instrument we found that the incessant dance 

 of the image of the scale was markedly less than when the pendulum 

 was hung in water. A touch with a finger on either pier produced 

 deflection by bending the piers, and the instrument responded to the 

 disturber. 



The vessel had been filled with fluid for some days, and we had 

 just begun a series of readings, when the plate-glass front again cracked 

 quite across without any previous warning. Thus ended our second 

 attempt. 



In the third experiment (July and August, 1880) the arrangements 

 were so nearly the same as those just described that we need not refer to 

 them. The packing for the plate-glass front was fonned of red lead, 

 and this proved perfectly successful, whereas the indiarubber pack- 

 ing had twice failed. As we were troubled by invisible leakage and 

 by the evaporation of the fluid, we arranged an inverted bottle, so as 

 always to keep the chimney full. We thought that when the T-piece at 

 the end of the shaft became exposed to the air, the pendulum became 

 much more unsteady, but we now think it at least possible that there 

 was merely a period of real terrestrial disturbance. 



From August 10 to 14 we took a series of observations from early 

 morning until late at night. We noted the same sort of diurnal oscilla- 

 tory motion as before, but the outline of the curve was far less regular. 

 This, we think, may perhaps be explained by the necessity we were under 



