390 
MR. GEORGE W. WALKER OK 
I have enclosed in thick lines the region within which the reflexion is complex and the 
result probably untrustworthy. I have used the values of e obtained from the time 
curve, so that the table is only of value in showing the large changes that may be 
expected. My hope is that the apparent complexity of the observed results will no 
longer discourage the investigator, but, on the contrary, will yield to careful 
treatment and lead to results of great value in seismological theory. 
Case (l), fig. 5, is such a remarkably clear record and so free from ambiguity that 
I venture to add a discussion of the SR a shown in it. SR X is quite clear in the 
horizontal component, but practically negligible in the vertical component. We find, 
however, that SR-! is about 40 seconds earlier than the theoretical time calculated 
from the time curves. This is also what we found for PR!. Thus while the time 
interval S —P is not'sensibly in error, in my opinion at least, up to 10,000 km., 
I think that several independent lines of reasoning suggest that the time curves for 
P and S should be sensibly depressed from about 2500 km. up to 6000 km. 
The first sharp movement is entirely to North, and this movement attains its 
maximum before the East-west component begins to move. We thus enquire 
whether this movement in SRj is related quantitatively to any movement in S. 
Proceeding as formerly to ascertain the effects due to reflexion of waves which would 
otherwise produce the same results at the observation station, I have drawn in fig. 6 
the conclusions at which we should arrive. 
OSi and OS 2 represent the first two movements in S, and OS^, OS^ are the 
corresponding effects to be expected in SR. They are obtained as follows : Resolve 
OSx or OS 2 into its components in and at right angles to the azimuth plane OA. 
