1921.] Angenheister.—A Study of Pacific Earthquakes. 221 
that the sea-bottom is continually sinking under the weight of sedi¬ 
mentary deposits. This may press the lighter magma away from the 
ocean to the continent, leaving the heavier and more viscous lava behind. 
In addition to this the damping due to the water of the ocean will be 
greater than that at the air-surface of the continent. 
In a later paper the author hopes to give a calculation of the influence 
of absorption coefficient of energy on the period, amplitude, and velocity 
of surface earthquake-waves, but one consideration will not be out of 
place here. The theoretical relation between vertical (Z) and horizontal 
(H) maximum displacements, i.e., A Z /A H , should be equal to 1*5 for both 
damped and undamped Eayleigh waves.* The observations show in nearly 
all cases values considerably smaller than 1*5, and we must seek another 
reason than damping for the smallness of this ratio in practice. Ray¬ 
leigh waves may be accompanied by L, waves — i.e., waves without 
vertical component — and thus we may obtain smaller values for A z /Ah 
according to the relative amount of L 1 vibration compared with that of 
the Rayleigh wave. 
Now, L tJ waves, being transverse waves, should not pass through a 
fluid, and thus we may expect that these waves should not be present 
in earthquake-wave vibrations that have passed through shallow under¬ 
ground pockets of residual magma, and in this case the value of A z /Ah 
should be 1-5. It is interesting, therefore, to find that the observations 
at Apia of earthquakes from the New Hebrides show this ratio equal to 
1*5, and thus we may assume between these islands the presence of such 
lava-residuals. This is not altogether against geological evidence. 
VII. Period of Surface Waves and Thickness of Upper Crust. 
Rosener has compared the period of the maximum • of the Rayleigh 
waves and Coda waves for difference epicentres, using the Gottingen records. 
He has shown that principally waves of 12- and 18-seconds period are con¬ 
tained in these waves. For the records of Siberian earthquakes he found 
an influence of the magnitude of the earthquake on the period of the 
waves. For strong quakes a period of 18 seconds is predominant, and for 
weaker quakes 12 seconds. For practically the same epicentral distance 
from Gottingen the Mexican quakes show a predominance of an 18-seconds 
period, while the Japanese quakes give 12 seconds. Thus the path of the 
wave under the Atlantic (18) and under the Asiatic continent (12) appears 
to have an influence. 
The observations of Rosener show that the periods of the long and 
Coda waves are not primarily dependent .on the conditions near the 
observing station— i.e., the ground on which the station lies. We may 
thus safely compare the periods of surface waves from the same quake at 
different stations. 
In Table IX a comparison is given of periods of the waves at Apia and 
Gottingen from the same earthquakes as were used by Rosener. The 
table shows that for Pacific earthquakes the Coda shows an average period 
of 11 seconds at Apia and 18 seconds at Gottingen, while for European 
earthquakes the periods are 18 seconds at Apia and 12 seconds at Gottingen. 
American quakes give 17 seconds at both stations. For Asiatic quakes, 
particularly those from Japan, the period is shorter at Gottingen (12) than 
at Apia (16 : 17). 
* G. W. Walker, Modern Seismology, p. 46, 1913. 
