38 6 
MR. GEORGE W. WALKER ON 
reduced to | original size. The lettering of the reflexion is the Pulkovo judgment, 
and I have added arrows to show the theoretical times. 
We note how sharp P is in fig. 4 as compared with fig. 5. PR 4 is a quite clear 
feature in both, although it arrives earlier than the theoretical time by 30 to 
40 seconds. In fig. 4 PR 4 is still sharp in the horizontal components but smoother 
in the vertical. It is distinctly smaller than P and clearly of opposite sign. In 
fig. 5 PR X is smooth in all components, and only slightly less than P. In this case, 
also, it is of opposite sign to P. In fig. 4 PR 2 arrives about the theoretical time. 
It is smoother than P in all components, about the same magnitude as PR X , and of 
the same sign as P. In fig. 5 PR.> is apparently early, is about the same magnitude 
as PR 1; and I think of the same sign as PR 4 . In fig. 4 PR 3 is about right as 
regards time, but is a somewhat vague movement, and PR 4 is absent. In fig. 5 
PR 3 and PR 4 marked as earlier than the theoretical time are, in my opinion, too 
insignificant to merit consideration. Thus theory agrees in some respects and differs 
in others. 
The appearance of P and its reflexions in fig. 5 is strongly suggestive of a 
limited train following an impulse, and the same may be said of PR 2 and PR 3 in 
fig. 4. 
As we have shown, such an effect (established by Lamb for water waves) might be 
produced by dispersion or reflexion from a variable layer, my own view being that 
dispersion is not of much importance in seismology. On either hypothesis, however, 
a sharp impulse may on reflexion give rise to a limited train of periodic waves of 
distinctly larger amplitude than the primary impulse without violating the principle 
of energy. Everything depends on the sharpness with which the impulse rises to its 
maximum (e.g., 1 second) and the number and period of the waves set up. Such an 
effect may be a contributing factor in the apparent magnitude of the reflexions as 
compared with P. 
The discrepancy in time between PR 4 and the theoretical value obtained from the 
time curve for P is not a mere accident in the two cases shown. A large number of 
others will be found in the Pulkovo ‘ Bulletins/ From an inspection of these it 
appears on average that at 6000 km. PR 4 is some 10 seconds early, at 8000 km. 
about 30 seconds, at 10,000 km. it is correct, while at 12,000 km. it is about 
10 seconds late. 
In conjunction with Galitzin’s observed values of the emergence angle e, we get 
confirmation that the primary curve for P, as given by Zoppbitz, requires revision. 
The change required is a depression of the ordinates of the time curve extending 
from 2000 km. to 6000 km. I mention this because, when the time curves are 
revised, I think it would be extremely unwise to make any alteration without 
considering the valuable information revealed by the times for the reflected waves 
and the apparent angle of emergence e by direct measurement. The last members of 
the P reflexions, although I do not think they can really acquire any importance, 
