ON SEISMOLOGICAL INVESTIGATIONS. 63 
an earthquake in Europe they are distributed in various azimuths 
(not quite a complete circuit even then), but for distant quakes they 
cluster in the same azimuth and give no material for discussion in 
azimuth (see Section VIII.). The Milne stations, however, especially 
those in Australia, can supply this information. 
It is clear, then, that the usefulness of the Milne instruments is 
by no means at an end, as the perfection of modern seismographs 
(especially the Galitzin instrument) might at first suggest. And it 
should not be difficult to. extend it considerably. 
(a) It can be damped effectually. Mr. J. J. Shaw, of West Brom- 
wich, has done this electro-magnetically with an aluminium plate 
in place of the Galitzin copper plate, which is too heavy for the light 
Milne boom. At present, however, he has not obtained simultaneously 
sufficient magnification to give the damping effect: damping is 
chiefly of use for following the movements of the long waves, and the 
scale should be big enough to show them clearly. Mr. Shaw is still 
at work on the instrument, and hopes to obtain the requisite magnifi- 
cation. 
(b) There should be little difficulty in increasing the magnification 
moderately both in movement and in time scale, though it may 
not be easy to settle which is the very best way of doing it. The 
experiments being made by various observers should at least give us 
a feasible plan. 
(c) Meanwhile if special attention is paid to getting good time 
determinations, and if the films are carefully read with a lens, the times 
of arrival of P and S for Milne stations should enable us to correct 
the tables for considerable distances from the epicentre where the 
European stations all agree and are all in error owing to their con- 
gestion in azimuth. (See next Section.) 
VIII.—Correction of the Tables for P and S. 
Recurring to the discussion of Section IV., it was shown that the tables 
for both P and S were sensibly in error, and the question arises how far 
they can be corrected. The main facts are these :— 
(a) The tables for small values of A are sensibly correct. This is 
shown by the agreement of determinations of epicentres from Pulkovo 
and Eskdalemuir, quoted by G. W. Walker in his monograph (p. 65). 
From each station the azimuth a and the distance A can be determined ; 
and from the two azimuths a and a, the epicentre can be determined 
without reference to A at all.1_ This is a modern advance, the importance 
of which is not easily over-estimated. If then the values of A determined 
from the P and S tables agree (to a fraction of a degree) with those found 
from the azimuths, the tables must be fairly correct. The value of A 
is about 20°. 
(3) But this single example may give quite a wrong impression of 
the accuracy with which an epicentre is at present determined. At 
greater distances we gradually lose the accordance between these stations. 
Thus, on January 4, 1912, Pulkovo gives 175° E., 49°-5 N., and Eskdale- 
1 See letter of Galitzin and Walker-in Nature for September 5; 1912, 
