SURFACE REFLEXION OF EARTHQUAKE WAVES. 
385 
Only the first four reflexions are computed, and even this is rather far for the 
short distances. The time curve is not sufficiently known for short distances and, 
moreover, the effect of finite depth of focus would come in. 
We first note that successive reflexions differ in sign. The table shows that for 
distances up to 3000 km. the reflected effects are small in comparison with P, but as 
we pass to 6000 km. PR,! becomes about two-thirds of P in the horizontal component 
and about one-third in the vertical component. As we pass to 12,000 km. PR X 
actually exceeds P in the horizontal component and is only somewhat less in the 
vertical component. Moreover, the second reflection, and even the third, are not 
insignificant. 
The table also shows that when the reflected wave has angle e about 20° there is 
a large drop in the magnitude. As in the case of 2000 km., the successive reflexions 
show a tendency to increase for a certain range, although they must finally vanish. 
In this connexion I may refer to a point which I have elsewhere proved, viz., that 
longitudinal disturbance cannot be propagated along a plane surface. Prof. Turner, 
‘ B.A. Report,’ 1915, remarks that this is in apparent conflict with Knott’s result, 
that there is complete reflexion when e = 0. There is, however, no conflict, for 
although the 'reflected amplitude is then exactly equal to the incident amplitude, 
there is a change of sign, so that the two disturbances exactly annul and there is no 
resultant motion at all. 
We turn now to the experimental evidence. Reflected waves have never been 
recognised as distinct features for A less than 3000 km. I may, however, draw 
attention to two records (‘Modern Seismology’) of Turkish earthquakes A about 
2500. km. and a Pulkovo record of an Icelandic earthquake A about 2500 km. 
(July 26, 1913). These all show somewhat similar disturbance within the 40 seconds 
succeeding P, and raise the question • whether the whole series of reflexions which 
must occur within this time could have combined to affect the seismogram 
measurably. 
When we reach A from 6000 km. to 9000 km., PR X becomes a marked feature 
comparable with P itself, and PR 2 only slightly less so, but PR S and the succeeding 
reflexions are not clearly made out. Specimen records, figs. 4 and 5 (Plates 3 and 4), 
to which we shall return, exhibit this. Passing to A from 10,000 km. to 12,000 km. 
we can usually trace the reflexions a little further, and PR! may be two or even three 
times the magnitude of P. 
Thus the numbers in Table IV. are in remarkably good general agreement with 
observation, showing how PR X acquires increasing importance as A increases. But 
although the table goes in the right direction it does not go far enough. It removes, 
however, the main difficulty in understanding the variations, and suggests the 
necessity of investigating reflexion from a variable layer instead of an ideal surface, 
and also of considering the effect of spherical divergence. The traces, figs. 4 and 5, 
are photographically reproduced from those issued by Galitzin in 1914, but are 
