ox SEISMOLOGICAL INVESTIGATIONS. 69 



the P waves are many times reflected they must travel so close to the 

 surface as to approximate to this impossible type. On the other hand, 

 C. G. Knott (' Physics of Earthquake Phenomena,' chap, x.) gives, from 

 theory, a total reflection of longitudinal waves at grazing incidence. 

 The theoretical difliculty may disappear on scrutiny ; if not, some other 

 explanation must, of course, be found. But that now offered seems to 

 fit the facts mentioned below, and is therefore put forward for considera- 

 tion. It amounts to suggesting waves (Y) travelling close to the spherical 

 surface with practically the initial velocity of P, which we take as 14-9 s. 

 per degree. The times would thus compare with those of S as below : 



Mr. J. J. Shaw suggested the name ' polychord ' for the phenomenon 

 considered ; and the letter Y from this term may be used to designate 

 it. It will be seen that Y"" crosses S near A=110°. But in the last Report 

 certain corrections were suggested to the printed tables. It is not yet 

 advisable to alter the figures used, as discussion is proceeding ; but we 

 may indicate in brackets the result of the corrections suggested, which 

 are as below : 



A= 15° 25° 35° 45° 55° 65° 75° 85° 95° 105° 



*• s. s. s. s. s. s. s. s. s. 



Correction P= 0.0 Q — i _3 _8 — 15 —24 



Correction S= +5 —4 —8 —11 —14 —17 —24 —35 —50 



The comparison of Y with S would, with these corrections for S, stand 

 as follows : 



A= 60° 70° ^ 80° 90° 100° 110° 



Y= 894 1043 1192 1341 1490 1639 



(S)= 1090 1210 1323 1424 1514 1588 



Y-(S)=(-196) (-167) (-131) (-83) (-24) (+51) 



The point of crossing is thus shifted about 8° nearer the epicentre. 



The plan of giving corrected figures in brackets will be followed below. 

 The example which first suggested this hypothesis was the Eskdalemuir 

 trace for the earthquake of November 24, 1914, to which attention was 

 drawn by Mr. J. J. Shaw. The lettering on the rough trace shown in 

 the illustration is his. See also Milne-Shaw record Plate II. At the time 

 of examination of this trace no other material was available, and a brief 

 summary may first be given of the argument as it then stood. 



_ It is natural to identify a with P and y with S. The distance of the 

 epicentre conesponding to 



S-P=12 h. 16 m. 55 8.-12 h. fi m. 43 s.=612 s. 



would then be A=.81-3° (84-1 corrected tables). The time for PR, 

 would then be 12 h. 10 m. 16 s. (12 h. 10 m. 30 s.). But PR, can almost 

 certainly be identified with yS at 12 h. 10 m. 54 s.*— much later than 

 either the uncorrected or corrected time. Similarly, the time for SR, 

 does not fit the trace at all. 



If, however, we identify S with S at 12 h. 17 m. 45 s. we then have 

 * This is the reading published by Eskdalemuir. 



