222 
The N.Z. Journal of Science and Technology. 
[Nov. 
We may conclude, therefore :— 
(1.) Epicentral distance is an important factor in determining the 
period of Coda waves, and also in a minor degree that of the main 
phase : nearer stations give shorter periods. It is probable that periods 
of 9-12 and 18 are generated, but that near the epicentre the shorter 
periods predominate, while for greater distances these short-period waves 
are damped o t in accordance with the well-known laws of damping. 
Thus for a distant station the larger period will predominate. The 
observations of Rosener mentioned above for Siberian earthquakes are 
not inconsistent with this view. It may be that for very strong earth¬ 
quakes a larger proportion of waves with long period are generated, 
as probably greater depths of earth-crust are affected. 
In the Calabrian earthquake of the 8th September, 1905, one of the 
largest recorded, the maximum of the main phase at Gottingen had a 
period of 9-12 seconds (neglecting the initial part of waves with rapidly 
decreasing period). The amplitude after 9 minutes fell to 10 per cent, of 
its maximum value. At Samoa the maximum part of the main phase had 
a period of 18-20 seconds, while 30 minutes later the amplitude and 
period were only slightly decreased. The Coda period at Samoa was 
18 seconds, and at Gottingen 12 seconds. It thus appears that the waves 
of 18-20-seconds period at Samoa are due to a natural vibration of some 
parts of the upper crust between Calabria and Samoa, which may store 
up energy and re-radiate it slowly. Such favourable conditions for 
origination of an 18-second period may not exist between Calabria and 
Gottingen. 
(2.) In addition to the influence of epicentral distance the nature of 
the path, whether under ocean or continent, appears to be an important 
factor. Thus the American quakes show the same period (18 seconds) 
at Apia as at Gottingen, the paths being under ocean in both cases and 
nearly equal. The Japanese quakes, on the other hand, give a shorter 
period (12 seconds) at Gottingen than at Apia (16-5 seconds). In the 
former case the path is under the Asiatic continent and in the latter under 
the deep Pacific. Also, the Coda waves of the Kamchatka earthquake 
(used above for velocity considerations) are shorter at Tiflis than at Samoa. 
If we assume, with Wiechert, that the L, waves and Coda waves contain 
natural vibration of an upper solid crust resting on a fluid lava, the 
thickness of the crust is one-half the wave-length of the fundamental natural 
vibration for a transverse 1 L l wave. Thus if D be the thickness of crust, 
T the period of vibration, and v the wave-velocity, then D 
If 
the values for v obtained in the preceding section be taken, i.e., v cont = 
3-87 km./sec., v ocenn = 4-68 km./sec., T may be best obtained from the 
Coda waves, which for the Kamchatka and Kurile earthquakes give T oc — 
18-5, and T c0 = 15-2. For Japanese quakes the values for Apia and 
Gottingen are 16-5 and 12-0. Taking the mean values T oc = 17-5 and 
T e0 = 14-5, we obtain the mean thickness of the earth-crust as 41 km. 
under the ocean and 28 km. under the Asiatic continent. 
Reasons have already been given in Section II for the assumption that 
for an epicentral distance of less than 6° the preliminary tremors for an 
earthquake under the Pacific Ocean travel the entire length of their path 
through the upper crust. The greatest depth which such waves reach in 
their path can be calculated to be 50 km. We may thus take 50 km. as 
an appropriate value of the thickness of the earth’s crust under deep ocean 
in confirmation of the value deduced above. 
