64 MODERN SEISMOLOGY 
S-P is known, and the corresponding distance 4 of the epi- 
centre from the station is determined. The result is free from 
any absolute error of time at the station. In many cases, 
however, P is so small that its incidence cannot be accurately 
assigned, and then one may get an estimate of the distance 
from S-PR, or L-S, but these are much less accurate and 
ought only to be used as a check. 
When JZ is determined thus for three suitably selected 
stations the position of the epicentre is determined uniquely 
as the common point of intersection of three small circles on 
the sphere. Needless to say the circles do not precisely inter- 
sect at a point in practice, so that the epicentre is given only 
within certain limits. The co-ordinates of latitude and longi- 
tude may of course be obtained by computation or graphically 
on a stereographic projection. 
It was pointed out by Galitzin that if the first impulse 
represents the arrival of a longitudinal effect in the plane 
containing epicentre, station, and Earth’s centre, the ratio of 
the magnitudes of the displacements to north and to east 
must give the tangent of the azimuth of this plane, so that the 
distance and direction of the epicentre can be determined by 
observations at a single station. This principle has been sub- 
jected to rigorous examination first at Pulkowa and later at 
Eskdalemuir, and the results show quite conclusively that, pro- 
vided the first impulse is sufficiently clear and large, the epi- 
centre can be determined in this way with great accuracy. 
There is a possible ambiguity of 180° in the azimuth deter- 
mined in this way from the horizontal seismograms alone, for 
the first impulse may be a condensation or a rarefaction. The 
vertical component seismograph, however, removes the ambig- 
-uity, for if the impulse is a condensation the corresponding 
vertical movement is up, while for a rarefaction the vertical 
movement is down. There are indications that the first im- 
pulse may appear as a rarefaction at one station and as a con- 
densation at another. This might be expected on Stokes’ 
dynamical theory of diffraction, and if it proves correct, it 
suggests a means of finding the axis of the primary impulse; 
and this would be a valuable addition to seismological know- 
