50 Br. C. Davison on Earthquake- Sounds. 



of the rocks in which the earthquake originates or of those 

 on which the observers are situated. Mr. C. A. Stevenson 

 remarks that in the Scottish earthquake of November 28, 

 1880, most of the observers near the epicentre who heard the 

 sound were stationed on hard dense rocks with little or no 

 soil near them*. But the records of this earthquake are too 

 few to establish a general law. 



Prof. Milne's discussion of the sound-phenomena of the 

 Japanese earthquakes is founded on a very large number of 

 records. In the whole country he indicates fifteen districts 

 in which earthquakes are specially numerous, and examines 

 the sound-phenomena with respect to its geological structure. 

 The conclusion at which he arrives is that " generally sound 

 is heard in rocky mountainous districts, while in the alluvial 

 plains it is but rarely observed "f. 



The relative, rather than the absolute, frequency of sound- 

 accoinpanied earthquakes being, however, the more correct 

 test, I have calculated the percentage of the earthquakes in 

 each district that were attended by sound and that originated 

 under land. They are as follows: — 21'6, 18'6, 1'8, 18*7, 

 30-0, 29-9, 23-1, 34-5, 26-7, 33'3, 8-0, 50-9, 42'2, 16-9, and 

 50*0. Prof. Milne's general conclusion is in part supported, 

 and in part contradicted, by this more definite evidence. On 

 the one hand, we have the comparatively high percentages of 

 422 and 34*5 in the mountainous districts 13 and 8 respec- 

 tively. On the other, we have a percentage of 18' 7 in dis- 

 trict 4, where " the coast is rocky, built up of palaeozoic 

 strata'''; while in the mountainous district 9 the percentage 

 (26'7) is practically equal to the average for the whole 

 country (26*5). The high percentage of 50 in district 15, 

 which consists " of a sandy coast bordering a plain," is not 

 very reliable, as it rests on no more than eight earthquakes. 



Possibly the districts are too large for the deduction of 

 general laws. But, if we take only the rectangles, whose 

 areas vary from 114"3 square miles in the south of the country 

 to 92'8 square miles in the north, we find great variety of 

 audibility-percentage within the limits of a district. For 

 instance, in four consecutive rectangles J, bounded by the 

 same parallels, in the Mino-Owari plain, the percentages are 

 14-3, 41-0, 12'2, and 23*5. Immediately to the south of the 

 first two of these rectangles are others § in which the per- 

 centages are 77*4 and 11*5. 



If, however, the audibility of an earthquake depended only 



* Edinburgh Roy. Soc. Trans, vol. xi. 1881, pp. 184, 185. 



t Japan Seismol. Journ. vol. iv. 1895, pp. xviii-xix. 



t 1401-1404, district 7. § 1456 and 1457, district 7. 



