250 REPORT— 1898. 



GO sees. From this it may be assumed that at all observatories, 

 whatever be the period of the magnets, each of them for a considerable 

 interval of time is subjected to identical or nearly identical periodic 

 movements of their supports. 



From this we should expect to find that large earthquakes would 

 disturb magnetometers at all stations. 



Sometimes the magnetic needles appear to be disturbed by the short- 

 period preliminary vibrations, and at other times by the succeeding long- 

 period earth waves ; and the question arises, whether the mechanical 

 movement these represent is likely to establish a rotational movement in 

 a suspended magnet. 



If we regard a magnet and its suspension as an ordinary pendulum, then 

 at all stations we should expect to find that the preliminary tremors of 

 an earthquake would establish a swing accompanied by more or less 

 rotation. When, however, we have rotational movements of magnetic 

 needles accompanying the larger eartli waves the explanation of this is 

 not so clear. The tilting which such waves represent may, as an illus- 

 tration, be taken at 10 sees, of arc. For such a tilt a magnet with a 

 suspension of 12 ins. would be displaced through a distance of about 

 one hundredth of a millimetre, and because the movement would be 

 extremely slow, taking from 5 to 10 sees, of time in 07ie direction, it is 

 likely that the magnet would closely follow its point of support. 



When movements of this character take place the resultant move- 

 ment recorded in the photogi'aphic film is a displacement having a 

 range of from 2 to 15 mm., indicating that the magnetic needle has 

 rotated through an arc of from 1 to 7 minutes. 



To determine whether tilting so slight and so slow results in so much 

 rotation is obviously a matter which without great difficulty may be 

 solved by experiment, 



The second assumptioji to account for the disturbance of magneto- 

 graphs at certain stations only, is the hypothesis that with regard to the 

 surface of the earth there is an unequal distribution of a subjacent 

 magnetic matei'ial the movements of which influence magnets in its 

 vicinity. 



On the surface these movements are apparently represented by waves 

 20 to 50 km. in length and 20 to 50 cm. in height. 



To explain the fact that magnetic storms and perturbations so often 

 precede large earthquakes and but seldom appear to precede small ones 

 (see Registers for Greenwich, Utrecht, Mauritius, Zikawei, &c.), we may 

 assume that the earthquake is preceded by chemical, physical, or mecha- 

 nical changes in the constitution of the materials where it originates. All 

 that we are certain about is that with many earthquakes there have been 

 enormous mechanical displacements of material sufficiently large to dis- 

 turb the Pacific Ocean for a period of twenty-four hours. 



Other earthquakes from submarine centres which have not disturbed 

 oceans, but have created equally large earth waves, indicate equally large 

 subterranean reliefs in stnain and material readjustments. 



These large earthquakes, originating beneath the bottom of the steeper 

 slopes of the earth's surface, suggest that at such places a secular flow in 

 subterranean material may be in progress, accelerations in which result 

 in violent shaking, which as it radiates is transformed into slow earth 

 waves. 



Near to the scene of such subterranean changes, prior to and at the 



