January 22, 1909] 



SCIENCE 



127 



the others severally after 38 and 53 days." 

 The Yakutat series began with a strong 

 shock September 3, 1899, "and there were 

 shocks at intervals Tintil September 10, 

 when, at 9 :20 a.m., they began to be alarm- 

 ing. There were fifty-two shocks, culmi- 

 nating in one of great violence at 3 p.m. 

 . . . There was another violent earthquake 

 September 15 and other shocks until Sep- 

 tember 20."^ 



In view of these facts the promptings of 

 terror when a great shock comes may well 

 be seconded by the admonitions of wisdom, 

 for even though it be probable that the 

 worst is over, a substantial possibility re- 

 mains that the worst is still to come. 

 American experience suggests that as often 

 as one time in eight a powerful shock, in- 

 stead of being the climax of the earth- 

 quake, may be only the forerunner of the 

 climax; and when life and limb are at 

 stake the odds of seven to one in favor of 

 safety form but a slender basis for mental 

 serenity. 



Turning now from the statistics' of se- 

 quence to the question of underlying 

 causes, I wish to present a conception of 

 earthquake mechanism which has developed 

 gradually during the study of California 

 phenomena. The block movements associ- 

 ated with earthquakes in California are 

 dominantly horizontal, and the fault 

 planes along which the blocks slide are 

 vertical. For this reason my mental pic- 

 ture of the system of faults (habitually 

 drawn with two dimensions only) is a 

 map instead of a section on a vertical 

 plane. Imagine a large tract of the 

 earth's crust superficially divided by faults 

 into acute-angled blocks, which have a pre- 

 vailing trend in one direction. In compo- 

 sition the blocks are heterogemeous, im- 



"MS. of report on New Madrid earthquake by 

 M. L. Fuller. 



' " Recent Changes of Level in the Yakutat Bay 

 Region," by R. S. Tarr and Lawrence Martin, 

 Bulletin Geol. Soc. Amer., Vol. XVII., p. 31. 



eluding stratified, metamorphic and igneous 

 rocks, with complicated structure. The 

 fault surfaces are not mathematically 

 plane, but gently undulate, so that move- 

 ment among the blocks involves more or 

 less distortion of the blocks themselves. 

 Imagine also that the tract is subject to 

 external horizontal force of such nature 

 as to induce internal shearing strains and 

 the associated shearing stresses; and that 

 the application of external force is con- 

 tinuous, making the internal stress cumu- 

 lative. The internal stress is not uni- 

 formly distributed, because the more 

 plastic rocks relieve the strain by flowage. 

 When the stress along some part of a fault 

 surface becomes greater than the adhesive 

 force a slip occurs. When the stress 

 within an elastic rock becomes greater than 

 the shearing strength fracture takes place. 

 In either case there is an instantaneous 

 redistribution of stress. Relief of stress 

 in the rock adjacent to the rupture is ac- 

 companied by increase of stress about the 

 edges of the surface of parting, with the 

 result that the area of the parting grows; 

 and the growth is continued until regions 

 of small stress are reached. The magni- 

 tude of the resulting earthquake depends 

 chiefly on the quality of energy released 

 by the relief of accumulated strain and 

 stress. 



If the quantities are large there are im- 

 portant after-effects. The discharge of 

 strain causes a new arrangement of strains 

 and stresses through a large tract; this 

 leads to flowage and the local concentra- 

 tion of stress, especially in the more 

 elastic rock; and this in turn causes frac- 

 tures, of which the surface manifestations 

 are after-shocks. 



When finally equilibrium is restored, 

 and the train of after-shocks is complete, 

 the system of stresses, not only in the im- 

 mediate neighborhood of the fault, but 

 throughout an extensive tract, is mater- 



