Reviews—Seismology in Japan. 517 
like those which are felt almost daily in Japan. In the following 
table are given the average ratio (mean of both components) of the 
amplitudes (A), periods (B), maximum velocities (C), and maximum 
accelerations (D), as observed on the free surface ground to those 
observed in the pit: 
A B C D 
Dlishijeanthquakesteeniens| sed cel eee coet Lee 1:0 13 1-6 
Severe earthquakes, large undulations ... ... 1:4 11 1:3 1:3 
Severe earthquakes, ripples ... ... ... ... 2°2 0°8 2°8 4-7 
These figures show that, for small earthquakes and for the principal 
undulations of severe earthquakes, there is not very much difference 
between the surface and underground observations; but that the 
difference is considerable for the ripples or small quick vibrations 
which precede and are superposed on the larger waves. In severe 
earthquakes these ripples are very prominent, and, being very quick 
in period though small in amplitude, their maximum accelerations, 
and consequently their overturning and fracturing effects, are very 
much greater than those of the principal waves. In the pit, these 
ripples are in part smoothed away, and their average maximum 
acceleration is only about one-fifth that of the ripples recorded on 
the surface. It thus appears that in. times of severe earthquakes, 
there should be less destructive action in deep pits than on the 
surface of the ground. 
(10) This paper deals chiefly with the construction of buildings 
in earthquake countries. Perhaps the most interesting part is that 
which relates to the possible issue of earthquake warnings. Previous 
attempts in this direction have failed, but Prof. Milne justly urges 
further inquiry before we relinquish what may seem an impossible 
task. The plan he suggests at all events appears worth trying. 
Starting with the hypothesis that “earthquakes are the result of 
faulting, and that faulting is due to a bending of the earth’s crust 
beyond its limits of elasticity,” he proposes that a very long water- 
level should be placed at right angles to an axis of elevation, or, 
better, two such levels at right angles to one another. It might 
thus be ascertained whether secular changes of level are measurable 
by such an apparatus, and, if so, whether they are in any way 
connected with the occurrence of earthquakes. 
(14) One of the subjects which have been studied most closely 
by Prof. Milne is that of earth-pulsations. Several instruments 
have been used by him, levels, tromometers of various lengths, 
pendulums, whose movements have been magnified in different ways 
and also made self-recording, and finally.a conical pendulum, of 
which a description will be found in the British Association Report 
for 1892 (pp. 107-8). In his earlier work, Prof. Milne discovered 
that a connection exists between earth-pulsations and local or distant 
winds. His later researches show that there is a still closer con- 
nection with the barometric gradient. With high gradients pulsations 
have always occurred, and with moderate ones they have generally 
been present. They are also most frequent during the winter season, 
when the barometric gradients are steepest. Prof. Milne considers 
that these pulsations are long flat waves of various amplitudes and 
