ON SEISMOLOGICAL INVESTIGATIONS. 



71 



variants are sufficiently numerous to render a larger correlation non- 

 emergent. . 



« A tabulated list was then made of the stations which stood out, on 

 the curves plotted originally, as abnormal— i.e., those that stood at the 

 head of very steep gradients when the points representing the observations 

 were joined ; such observations represented in numbers 2 to 4 a: A list 

 of these showed that the results from six stations were consistently abnor- 

 mal at whatever distance from the origin they might be situated in different 

 earthquakes. These stations were Irkutsk, Taschkent, Dorpat, Niko- 

 laiev, Potsdam, and Hamburg, and they were accordingly deleted as 

 representing a negligible variant. The curves were not much improved 

 by this treatment, however, and so the correlation was re-calculated. 

 This time thirty-one earthquakes were taken from Liste A in the 

 Katalog, six from " Earthquakes and other Earth Movements recorded in 

 the Antarctic Regions, 1902-3 " (Royal Society— in press), and the Great 

 Indian Earthquake of 1905 from Part II. of the report (Earthquake 

 Investigation Committee, Tokyo, 1907). These gave 1,029 observations 

 in all, and the correlation came out as -10, with the probable error ± -02, 

 a more significant correlation.' 



The next step in the investigations was to plot the duration of a given 

 earthquake as recorded at stations situated at different distances from an 

 origin upon squared paper. This was done for twenty-two disturbances, 

 the distances of stations from origins varying from between 20° and 160°. 

 The resulting curves or figures were, as might be expected, serrated in 

 appearance (see dotted line, tig. 3), but still the greater number of them 

 suc^sested that duration increased as the distance from the origin increased. 



Fig. 3. 



Mins 

 200 



150 



100 



5 



1 2 3 4 5 6 7 6 9 10 H 12 13 14 150° 



Degrees in tens. 



No. I, March 28, 1903. Shide List, No. 601. 

 Origin Banda. Nineteen observations. 



When a number of stations at approximately the same distance from an 

 origin were taken in groups and a mean time for each group was plotted 

 against a mean distance an increase of duration with distance became 

 more pronounced. The general trend or approximate curves for the first 

 twelve of these figures is shown in fig. 4. The meaning of line No. I., 

 for example, is that this earthquake at a distance of 25° from its origin 

 had a duration of 135 minutes, while at 130° distant the duration was 

 about 178 minutes. It is identical with the thick line shown in fig. 3. 

 These durations are those indicated on seismograms jjIus the time taken 

 for preliminary tremors to travel from an origin to the observing stations. 

 Whether this correction is or is not necessary, because it is small, its 

 effect upon the general result is small also. For four of the earthquakes, 



