ON SEISMOLOGICAL INVESTIGATIONS. 59 



investigating possible combinations of waves when tables are available; 

 but it is, of course, nothing more or less than the investigation of the 

 angles of emergence as sketched in Walker's * Seismology,' p. 54. 

 Attention is called to the matter here, firstly because it seems possi- 

 ble that the publication of Dr. Klotz's table for PS may lead to some 

 erroneous identifications, and secondly because the question is raised 

 below whether we can have more than one reflected P wave at the same 

 point. 



Fig. 3 will show what is involved in this query. From the epi- 

 centre E, let EA and EB be two neighbouring paths for the wave P. 



Fig. 3. 



Then by regular reflection PR will be received at R, equidistant with 

 E on the opposite side of the little reflecting portion AB. The con- 

 dition may be written either 



time along EA+AR = time along EB+BR 

 or angle of emergence at AB= angle of reflection. 



Now, can both these conditions be also fulfilled, still for P waves 

 only, at another point S? Reasons are given below for believing that 

 they can — i.e. that we can have 



time along EA + AS = time along EB + BS 



while as regards the second condition it is only necessary that the path 

 AS should touch the path AR at A, the curvature being clearly 

 different; and similarly BS touch BR at B. We proceed to examine 

 this evidence, which is based on the study of records at stations distant 

 more than 100° from the epicentre. 



IX. Tables for P and S at Distances exceeding 110°. 



At distances from the epicentre greater than 110°, the times 

 recorded for the arrival of P and S are such as cannot be reconciled 

 with adopted tables by any reasonable extrapolation, and to explain 

 the anomalies various hypotheses of discontinuity in the interior of 

 the Earth have been suggested. It is believed that these are unneces- 

 sary, and that the hypothesis outlined below will fit the facts. It calls 

 for a modification of existing tables between the origin and 40° dis- 

 tance; and, until it is disposed of in one way or the other, the improve- 

 ment of these adopted tables cannot be satisfactorily undertaken. 



For the present attention will for simplicity be confined to P, 

 though S is subject to similar treatment. 



The nature of the anomalies will be seen by consideration of the 

 following earthquake, where the recorded arrivals of P have been 

 divided into two groups. One group can be identified with PR, but 

 the other clearly cannot. For the times of PR. the times for half the 

 arc according to adopted tables have been simply doubled. There is 



