48 REPORTS ON THE STATE OF SCIm#NCE.—1914, 
In Table I. the differences are grouped under distances from epi- 
centre, all instruments other than Milne being grouped together. 
Errors greater than 6 m.—The five large positive residuals are as 
follows :— 
Instru- | rrors | Dist. fr 
Observatory ment | Date a 7 acne 
m. m. = 
FUPICSE auc) uancoe eee tts W 1913, Jan. 11 | +106 +7-4 95 
Tneshwceyy a1 ak Ww 1913, May 18 | +104 — | 96 
Triest . Se Dee W 1913, June 22. +103 — 83 
Czernowitz . : : Ma. | 1913, May 23  +10-1 +9-5 89 
Pompeii - . «| OA. | 1913, June 22 | 492 — | 88 
The difference between the times of arrival of P and S being near 
10m. it is possible that some of these are mistakes of P for S. But 
in the case of Czernowitz, a mistake of 10m. in both P and S seems 
probable. It is safer to omit these cases as anomalies than to attempt 
to correct them. 
Errors. <6 m. and >1 m.—With the exception of a couple near 
the epicentre these do not develop until near 90°. Between 90° and 
100°, however, they outnumber the normal errors given in the body of 
the table. They are doubtless due to the fact that a reflected wave 
has been mistaken for the direct wave. The fact that the first reflected 
effect PR is often more pronounced than P in the case of distant earth- 
quakes is duly noted in Walker’s monograph (p. 41); it may not, how- 
ever, be realised that it is so often mistaken for P in the published 
records of sensitive instruments. Beyond 100° from the epicentre 
no times for P were correctly given at all for the five earthquakes 
here examined. It is not intended to ignore the fact that these differ- 
ences will change with distance from epicentre, but for the present 
rough review we will neglect this change. The median is 3°75 m. 
or 3m. 45s. The mean of the differences from this is + 0°53m. But 
it does not seem clear that some of the differences which may be faulty 
P readings should be included. If these are excluded the median is’ 
3°8m.; the mean is 3°87m.; and the mean of the differences from 
the mean is + 0°35 m. 
Normal Errors.—Coming now to the main part of Table I., if we 
take the errors as they stand (assuming the time-table for P correct 
throughout) the mean of the 87 differences is —O'07m. or —4s. 
But there is a systematic run about the differences as may be seen 
from the following means for the separate columns :— 
0° — 40° — 80° — 90° — 100° 
m. m. m. m. 
+0-13 =-0-01 —0-11 —0:27 
The process adopted in the previous work does not justify any 
great. refinement of correction; but we may fairly correct the different 
columns by the quantities 
0° — 40° — 80° — 90° — 100° 
m. m. m. m. 
—0-1 0-0 +0-1 +0-3 
