364 Messrs. Milne and Gray's Experiments 



half-turn of the pulley causes the ring to be pushed out 

 towards the clock and then to be withdrawn. 



The instruments which we have now described, and many 

 of them variously modified, have all been more or less tested 

 by subjecting them to a series of actual earthquakes. From 

 our experiments, it would seem that the choice of an instru- 

 ment must depend somewhat on the kind of record which is 

 sought for. 



For a direct single record, either the pendulum with vertical 

 index, or the double-bracket instrument appears to be the best. 

 It is probable that two conical pendulums arranged to take up 

 two rectangular components, and, again, by means of connect- 

 ing-rods, to communicate these components to the same ver- 

 tical lever, may act well ; but this has not been tried. For 

 component records, a pendulum with two indices, brackets, 

 conical pendulums, and cylinders seem all to act well; but 

 the conical-pendulum instrument is probably the best. 



For large earthquakes it is probable (although we have had 

 no practical experience) that the bracket and pendulum instru- 

 ments would prove useless. For a component of the record, 

 either a simple hollow cylinder of considerable mass and rest- 

 ing with its axis horizontal on a level plane, or a three-wheeled 

 carriage similarly supported, will be found good. 



Part IL 

 Earthquake Motion. 



The subject of earthquake motion may be treated from 

 three points of view: — 1st, theoretically; 2nd, by the exami- 

 nation of artificially produced vibrations ; and 3rd, by the 

 instrumental observation of the vibrations we feel in actual 

 earthquakes. 



(a) Theoretically. — The probable nature of earthquake 

 movement may be gathered from the writings of Hopkins, 

 Mallet, &o., and from treatises on elasticity by several mathe- 

 matical writers. From these we conclude that there ought to 

 be at least two sets of vibrations, known respectively as normal 

 and transverse vibrations. The normal vibrations travel more 

 quickly than the transverse vibrations, to which they are at 

 right angles. According to Poisson, these velocities would 

 be v3 : 1. Poisson's theory, however, has been shown to be 

 at fault in this respect; and we have to rely upon direct expe- 

 riments on the elasticity-moduli of the materials for correct 

 theoretical estimates of this ratio. The following, according to 

 such experiments made by us (an account of which we hope 

 shortly to publish), is the value of this ratio for the kinds of 

 rock mentioned in the table, the actual velocities being also 

 given : — 



