SURFACE REFLEXION OF EARTHQUAKE WAVES. 
391 
The former component suffers no reduction by reflexion, but the latter is altered. The 
amount depends on what are the correct impingence angles for S and SR. There is 
room for considerable latitude in this, but I have taken provisionally 73° for S and 
57° for SR. With these values the component in the azimuth plane suffers reduction 
in the proportion 6'7 in H and 6'6 in V, and the sign is reversed. We thus get 
OSjRj, OSgRi for the expected horizontal movement, while the vertical movement 
is very small indeed. 
While the results are thus horizontal movements in the North-south component, 
they do not agree with the observed movement in SR. SjRj is of the opposite sign 
and S 2 Ri is too small. While the result is perhaps disappointing, I am still convinced 
that, reasoning conversely, the systematic analysis of the observed movement of the 
SR for a number of cases is certain to lead to valuable information as to the 
circumstances of reflexion, even if it involves a change of our ideas as to the full 
meaning of these reflexions. 
At considerable epicentral distances theory indicates that the general effect of 
reflexion is to suppress rapidly the component in the azimuth plane and make the 
SR n ’s more truly at right angles to the azimuth plane. But there is an important 
difference between these successive reflexions and the PR„’s. These tail off to 
nothing, while the transversal waves, as soon as the reflexion takes place at angles 
less than 56°, may be expected to undergo a change of type rather than of 
magnitude. Thus the tail of the transversal waves merges into the start of the 
long wave phase as observation undoubtedly shows to be the case. 
In my opinion the marked periodicity of the transverse reflexions indicates more 
clearly than do the longitudinal reflexions the necessity for examining the effects of 
reflexion from a variable layer of thickness comparable with the geological crust. 
The present discussion would be incomplete without some reference to waves that 
have been changed from transversal to longitudinal, or vice versa, on reflexion. These 
waves were called Wechseltvellen by Wiechert. The vibration must be in the 
azimuth plane, and they cannot show as a distinct phenomenon until the distance 
exceeds that for which the angle of impingence for transversal waves in the azimuth 
plane exceeds 56°. From the time curves this would be about 4000 km., but 
Galitzin’s direct observations of e would make the distance over 7000 km. 
These Wechselwellen must be made up of two parts : (l) transversal waves starting 
from the origin and reaching the station as longitudinal; (2) longitudinal waves 
starting from the origin and reaching the station as transversal. We should 
naturally seek to compare the first with the S waves and the second with the 
P waves, while there is no necessary connexion between the two except that they 
arrive together, and the vibration should be in the azimuth plane. It must, of 
course, be understood that in making any such comparisons we assume what is not 
. at all likely to be the case, viz., that waves starting from the origin are symmetrical 
about the origin. We are, in fact, simply estimating the changes due to reflexion 
3 f 2 
