414 Dr. CO. Davison—British Earthquakes. 
observed only near the epicentre, so that, at a distance from that 
region, the shock consists of two entirely detached parts. The 
average duration of the interval of rest is 2°7 seconds, and that of 
the whole shock about six or eight seconds, rising, as in the Hereford 
earthquake of 1896, to 9°3 seconds. It is now known‘ that, in these 
cases, the double shock is caused by impulses in two distinct foci, the 
impulses being either simultaneous or separated by an interval which 
is often less than the time taken by the earth-waves to travel from 
one focus to the other. Of the 250 earthquakes, at least eleven (or 
about 4 per cent.) belong to this class of twin earthquakes. 
SounpD-PHENOMENA. 
The sound which accompanies an earthquake is a deep rumbling 
sound, so low that it is inaudible to some persons, while to others 
in the same place it appears louder than any thunder. That the 
inaudibility is not due to inattention is clear from the fact that, in 
the Inverness earthquake of 1901 (which occurred at 1.24 a.m.), 
86 per cent. of the obseryers who were awake, and 84 per cent. of 
those who were asleep, heard the sound which preceded the shock. 
For the Doncaster earthquake of 1905 (which occurred at 1.37 a.m.) 
the corresponding figures are 93 and 91 per cent. 
Of the 250 earthquakes, 197 (or 79 per cent.) were certainly 
accompanied by sound. In 4 cases, the observers state that they 
heard no sound; in 49 cases, no reference is made to the sound- 
phenomena. Of these 53 earthquakes, 28 occurred in the Ochil 
district, 10 in Glen Garry, 11 were after-shocks of strong earthquakes, 
and 4 were slight earthquakes in various places. But all, without 
exception, were feeble tremors, the number of observers was small, 
and there can be little doubt that, with a larger number, there would 
have been some who would have heard and recorded the sound. It 
is probable, therefore, that all British earthquakes are attended by 
audible vibrations. 
In strong earthquakes the sound was heard on an average by 
83 per cent. of the observers, in moderate earthquakes by 98 per 
cent., and in slight earthquakes by 97 per cent. This difference is 
partly due to the fact that, in slight earthquakes, the sound is 
generally a much more prominent feature than the shock, and partly 
to the variable size of the sound-area. In slight earthquakes, the 
sound-area either coincides with or overlaps the disturbed area; in 
moderate earthquakes, the two areas are approximately coincident; 
in strong earthquakes, the sound-area falls short of the boundary of 
the disturbed area, the sound-area varying from 43 per cent. of the 
disturbed area in the Derby earthquake of 1904 to 82 per cent. in 
the Inverness earthquake of 1901. On an average, the sound-area 
is 64 per cent., or roughly two-thirds, of the disturbed area. 
In moderate and slight earthquakes, the percentage of audibility 
is practically uniform throughout the area affected. In strong earth- 
quakes, with a large disturbed area, the decline in audibility as the 
1 Quart, Journ. Geol. Soc., vol. lxi, 1905, pp. 18-83. 
