ox THE EARTHQUAKE AND VOLCANIC THENOMENA OF JAPAN. 87 



as they require attention at intervals of from 8 to 1 2 hours they are being 

 replaced by lamps such as are used in magnetic observations burning 

 benzine. 



Every day from 12 noon to 1 p.m. the lamps are removed and a reading 

 is taken, so that time intervals are marked on the photographs and scale 

 values are obtained. 



It does not seem necessary that the boom should be made of alumi- 

 nium, as I obtain what appear to be equally satisfactory records witli booms 

 of brass or even wood. The most delicate pendulum I have has a booui made 

 of varnished bamboo with brass fittings. It is about 5 feet in length, and 

 when last rated had a period of 55 seconds. I say last rated because I 

 find that this pendulum, like all others I work with, changes its period, and 

 therefore its sensitiveness, from week to week. I notice tliat this source of 

 error when computing results is also found in the infinitely better con- 

 structed and Ijetter installed apparatus used by Dr. von Rebeur-Paschwitz. 



When the pendulum has its 55-second period one millimetre deflec- 

 tion on the photographic plate is equivalent to a tilt of 0'08 second of arc. 

 With this degree of sensitiveness a 141b. weight placed on the column, 

 which is old and massive, at a distance of 2 feet from the instrument 

 causes a deflection of 0'5 mm. My weight on the floor ut the outer end of 

 the boom produces no visible eflfect. 



In this condition the pendulum is, however, often too sensitive, as it 

 will, from time to time, wander an inch or so to the right or left of its 

 mean position, and the spot of light fall outside the film. A sensitiveness 

 of 1 mm. motion per 0"'5 arc is usually quite sufiicient, and I do not think 

 that apparatus like those of Wolf, d'Abbadie, Darwin, or von Rebeur- 

 Paschwitz capable of recording tilting of from -j^^ to ^^ of a second 

 could be used on the alluvium of Tokio even when installed on a con- 

 crete bed underground. 



Such apparatus might, however, be used on the solid rock which crops 

 out round the Tokio j^lain. 



An attempt to test the accuracy of one of the horizontal pendulums 

 was made by placing it on an ii'on plate resting on a plank 18 in. broad, 

 1| in. thick, which in turn rested on supports near its end 6 feet apart. 

 It was then adjusted, so that trials with the test screw indicated that turns 

 of 10° gave an average deflection of 11' 5 mm. 



Side by side with the pendulum a transit instrument having a good 

 telescope was placed, and this read on a scale fixed on a brick wall at a 

 distance of 720 feet. The supporting plank was then loaded at its middle 

 until the telescope showed a deflection of 14 in. on the scale and the 

 pointer of the pendulum moved 93 mm. From this it seems that the 

 pendulum indicated a tilt of 1 in 562, while the angular tilt of the 

 telescope was 1 in 616. 



These are the means of a series of experiments, and assuming that the 

 readings tlirough the telescope were correct, then the pendulum indications 

 are about 10 per cent, below their true values. On the other hand, 

 assuming that the readings through the telescope were one inch too small, 

 and it was difficult to read within that quantity, then tlie pendulum 

 indications are 2-3 per cent, short of their true value. A great source of 

 error no doirbt resides in the test screw of the pendulum. 



The instrument described will be recognised by those engaged in 

 similar in^'estigations as coarse in construction, roughly approximate in 

 its records, and because it is large as being in all probability subject to 



