~ 
ON SEISMOLOGICAL INVESTIGATIONS. D 
for military purposes, was offered to Mr. Burgess, and the opportunity 
was taken to make some experiments. With Mr. Shaw’s help the Milne- 
Burgess machine was dismounted from its pier at Shide, and remounted 
in the dug-out some few miles away. At some considerable personal 
trouble and inconvenience Mr. Burgess kept the machine running for a 
few weeks. [It was necessary to keep a pump going to avoid having 
the dug-out flooded.] No very satisfactory results were obtained, the 
wandering of the trace being just as serious as at Shide, and the experi- 
ments came to a natural end when the dug-out was reclaimed for military 
purposes. The main conclusion which emerged from the experiments 
was that rain had more to do with the wandering of the trace than had 
temperature. After rain the lines became much congested: when the 
rain ceased they were unusually expanded, returning to the normal separa- 
tion when the weather had been fine for a few days. The inference 
seems to be that the wetting and drying of the ground caused slow move- 
ments in opposite directions—a conclusion in accord with previous ex- 
perience at Shide. The dug-out was ina clay soil, which doubtless 
emphasised the effects. 
Milne-Shaw Seismographs. 
Several of these machines are nearing completion; that intended 
for Oxford is finished, and will be erected as soon as possible. 
Suggested Corrections to Adopted Tables. 
The work of disentangling the corrections to the adopted tables from 
other errors is attended with considerable difficulties, but nothing is more 
important, if the phases are to be rightly identified ; and although pro- 
gress is slow, the ground is being steadily cleared. One set of difficulties 
arises from ignorance of the clock errors, which may be expected especially 
at outlying stations. Recent experience aroused the suspicion that 
these are often considerable, so that accurate intervals S—P are attended 
by unexpectedly large errors in S and P separately. The experience 
here referred to is derived from the re-reduction of a number of the best 
observed earthquakes after applying the corrections to tables deduced in 
the last report, and carefully correcting the position of epicentre. 
Hence attention was recalled specially to the investigation in the 
Introduction to ‘The Large Earthquakes of 1913,’ p. iii, which was con- 
. fined to the use of intervals S—P, free from clock errors. This investigation 
iP 
gives no information about § and P separately which will enable us to 
correct the tables, and was naturally followed by a more comprehensive 
analysis where S and P were kept separate, the results of which were 
given in the Twenty-second Report, Table IT. We may denote these two 
investigations by the symbols (1913) and (22) respectively. 
~ Now for the smaller values of A (1913) differs essentially from (22), 
as is seen from Table I. below, and we must determine which is nearer 
the truth. For this the following method suggested itself. 
Pairs of Stations on Opposite sides of the Epicentre. 
Any pair of observing stations on directly opposite sides of the epi- 
centre give a check on the values of S—P. Suppose for simplicity we 
