302 CHATLES DAVISON: 
the displacement, rather than its effect, is the more important 
subject for investigation. 
For determining the position of the epicenter, three methods 
have been employed, depending respectively on observations of 
the direction, time of occurrence, and intensity of the shock. 
The first method was suggested by Mallet, and used by him 
in studying the Neapolitan earthquake of 1857, and by a few 
other seismologists who have followed in his steps., Later on, the 
method fell into disrepute; and, so far as it depends on individual 
observations of the projection or fall of bodies, etc., it must, I 
think, be regarded as unreliable. But, if the number of observa- 
tions be large, the average of all the records in one place may 
give a close approximation to the true direction. This was first 
shown by Professor Omori for the Tokio earthquake of 1894. In 
that city, the earthquake, besides a number of minor vibrations, 
consisted of a single great oscillation, the maximum displacement 
in which wase73 42 inv therdixection Wa Zone S-sand Om Ne 
Many columns and monuments were overthrown, and especially 
a large number of ‘Ishidoro” or stone lamp-stands placed in 
gardens. Professor Omori measured the directions in which 245 
bodies fell in different parts of the city, 144 being ‘‘ Ishidoro” 
with circular bases. The directions are extremely varied, and at 
first sight appear to be subject to no law, but the mean direction 
given by all the observations is W. lo}, Se and) 3 woeeNeoeelitnwe 
take only the 144 ‘‘Ishidoro” with circular bases, and regard all 
the determinations of the direction as of equal value, we find the 
mean direction to agree exactly with that given by the seismo- 
graphic record, and to have a probable error of less than two 
de gneesiaas 
The study of the Hereford earthquake of 1896 led to a some- 
what similar result. In this case, observations of overturned 
bodies were not available, and the estimates of the direction were 
all made from personal impressions. They are extremely rough, 
few of the observers referring to more than the eight principal 
points of the compass. Moreover, as a general rule, the apparent 
t Bull. della Soc. Sismol. Ital., Vol. II, 1896, pp. 180-188. 
