ON EARTH TREMORS. 299 



wires supporting one side of the mirror by two converging wires, and 

 thus to allow it to move only in an east and west direction, and so 

 eliminate this error. It would, however, have the drawback of compli- 

 cating the instrument. 



If d be the vertical distance between the points of support of the 

 silver wire, it is obvious that the circular measure of one second is cOjd, 

 and therefore c=:f?/2940. In the present instrument d is about 12 inches, 

 so that the horizontal distance between the points of support is only 

 0-004 inch. 



On the first three days of August, the scale- value for a tilt of two 

 seconds was redetermined in the manner described above. The average 

 scale- values of six pairs of east and west tilts on each evening were 



6-39±03, 6-06±-06, and 6-14±-08 inches, 



and the average of the eighteen pairs was 6-20±04 inches. Between 

 the end of May and the beginning of August, the scale- value for one second 

 changed from 3"44 to 3'10 inches, and there must, therefore, have been a. 

 tilt towards the north of about seven seconds. 



Sensitiveness of the Pendulum. — Except on two occasions, which will 

 be referred to below, the image has always been perfectly steady, and 

 chiefly to this is due the extreme delicacy of the pendulum. If the gas- 

 jet be displaced by y-^Yfths of an inch, the image and the vertical cross- 

 wire in the telescope are clearly separated, and a displacement of y^g^th 

 of an inch of the gas-jet can be easily detected. Thus the scale-value of 

 one second being 3'44 inches, it follows that we can observe a tilt of the 

 ground of less than y^th of a second ; ' and since the pendulum might 

 be rendered more sensitive, and the optical arrangements improved, it 

 would not be impossible to detect a tilt of tqVo^'^ ^^ * second, or even 

 less than this if desired. 



Ohservations with the Pendulum. — The principal objects of the experi- 

 ments being to test the working of the instrument, observations were 

 made when convenient, and not at regular intervals. They were made 

 frequently enough, however, to give some idea of the nature of the daily 

 motion. At first the movement indicated appeared to be always to the 

 west, and the observations, though not generally continued later than 

 10 P.M., showed no trace of a return movement to the east. This westerly 

 tilting, of at first about a second a day, appeared to be chiefly due to the 

 settling of the foundation, and by the end of June became nearly imper- 

 ceptible. Towards the end of May and the beginning of June, observa- 

 tions on the daily movement were made for about two weeks consecutively. 

 The direction of the movement changed from east to west at from 7 to 

 9 A.M., and from west to east at from 10 to 12 p.m., but these epochs were 

 altered when the continued westerly tilting ceased. During the middle 

 of August, the easterly movement lasted till noon or about 2 or 3 p.m. 

 Subsequent experiments, described below, seem to show that this daily 

 movement may be, in part at least, due to the action of convection currents 

 in the oil surrounding the mirror. 



The pendulum in the Cavendish Laboratory having been extremely 

 sensitive to slight changes of pressure on the floor of the room, it seemed 

 desirable to make some similar experiments with the present instrument. 

 I stood for one minute about a foot and a half west of the pendulum, and, 



' An isosceles triangle with a base one inch long and each of the equal sides 1,000' 

 miles would have a vertical angle of about 3^th of a second. 



