flll/t 



1884 



NA TURE 



clamp at e, by which the pressure of the tracer on the 

 glass plate may be reduced to an amount just sufficient to 

 scratch off a thin coating of lamp-black with which the 

 glass is covered. In Fig. 2 the two pendulums are seen 

 in plan, with their tracing pointers touching the glass 

 plate g at different distances from its centre. The plate 

 and pendulums are mounted on a -single base, which is 

 very rigidly secured to the top of a broad post, stuck 

 firmly in the earth and projecting only a few inches above 

 the surface. Continuous rotation is communicated to the 

 plate by a friction-roller, k, held in a slot guide and con- 

 nected by a universal joint to one of the arbors of a clock, 

 which is wound up once a day. Government by an 

 escapement being out of the question, the clock is con- 

 trolled by a fluid-friction governor connected to the wheel 

 train, also by friction gear, as shown in Fig. 3. The balls 

 are four in number to prevent disturbance of them by an 



Fig. 4 



; 'V&v/W 



earthquake. The vanes dip into oil, and are drawn back 

 by two springs which tie them to the spindle. 



When the earth shakes, the axis, d, of each bob remains 

 sensibly at rest as regards components of motion perpen- 

 dicular to the corresponding pendulum, and the tracing 

 point is therefore displaced over the glass plate, in the 

 direction of the plate's radius, through a distance which 

 in this case is four times the motion of the earth. So 

 long as no earthquake occurs each pointer traces over and 

 over again a single circle on the plate. The circle fre- 

 quently tends to widen inconveniently, especially if the 

 pendulum is very nearly astatic. This is in part at least 

 due to such changes of the vertical as have been observed 

 by d'Abbadie, Plantamour, G. H. Darwin, and others. 

 The plate consequently requires frequent attention, and 

 where that cannot be given, an electric starting arrange- 



ment is to be preferred. When an earthquake has oc- 

 curred, the plate is removed, varnished, and photographed 

 by using it as a " negative.'' 



The bob of each pendulum may of course be rigidly 

 attached to instead of pivoted on the pendulum frame. 

 In that case the centre of percussion of the frame and bob 

 together (which will then be a little farther from the sup- 

 port than the centre of the bob) will be the steady-point. 

 The writer, however, prefers the arrangement described 

 above, which gives great compactness and a maximum of 

 effective inertia, and which has the advantage of making 

 the position of the steady-point at once determinate. 



It would take too much space to describe or even to 

 enumerate the many other devices which have been 

 suggested to secure a steady-point by various methods of 

 astatic support, 1 leaving one, or in some cases two, 

 degrees of freedom to move horizontally. The horizontal 



;HiL_. 



h=^ 



pendulum has been modified by substituting a flexible 

 wire and spring for its rigid pivots, thereby avoiding all 

 but molecular friction at the axis of support. Spheres 

 and cylinders, free to roll on plane or curved surfaces with 

 or without a slab above them, have been tried, but their 

 friction is excessive. The approximate straight-line 

 motions of Watt and of Tchebicheff have been pressed 

 into the service as means of suspending a mass with 

 freedom to move in a horizontal path. The common or 

 vertical pendulum, an old favourite with seismologists, has 

 suffered many transformations in the effort to reduce its 

 stability, which is preposterously great unless we make 

 the pendulum very long. A 20-foot pendulum consisting 

 of a cast-iron ring weighing half a hundredweight, hung 



1 See papers by Gray, Milne, the writer, and others in the Transactions 

 of the Seismological Society of Japan, vols. i. to vi. ; or a memoir on 

 "Earthquake Measurement," t-ublished a year ago by the University il 

 Tokio. 



