302 



NATURE 



[Jmy 28, 1887 



to 9h. 56m. 1 8s. for the beginning of the tremors, with a 

 probable error of fully one minute. This large probable 

 error is due to the very small scale upon which the mag- 

 netograph records time intervals (one-tenth of a milli- 

 metre corresponding to twenty seconds), and to want of 

 sharpness in the photographed traces. This time gives 

 2"66 miles per second, or 4250 metres, with a probable 

 error of one or two tenths of the amount. 



The clock in the Western Union Telegraph Office at 

 Pittsburgh (523 miles) was stopped at gh. 54m. 



From Cincinnati and suburban towns (500 miles) come 

 many reports. In this city local mean time is largely 

 used, owing to the fact that it is nearly midway between 

 the 75th and 90th meridians, where the only inconvenience 

 of standard time is at a maximum. The correction to 

 the 75th meridian is + 37m. 40s. The Western Union 

 Telegraph Office gives Qh. 54m. The lYmes-Star news- 

 paper gives, from the clock in its own office, 9h. i6m. 

 " exactly" (9h. 53m. 40s. standard) ; at the Commercial 

 Gazette office, 9h. 17m. 45s. local, 9h. 5Sm. 25s. standard 

 (probably noted after the shocks were over). At the fire 

 tower, after the principal shock, gh. i6m. 17s. was noted ; 

 clock error, twenty-three seconds slow, giving gh. 54m. 20s. 

 standard. Two other observers, noting by watches, give 

 9h. 1 6m. ; and one notes an advanced stage of the shocks 

 at 9h. 17m., but give no means of estimating their errors. 

 At Covington, Ky., across the Ohio River, I. J. Evans, 

 watchmaker and jeweller, reports his regulator clock 

 stopped at 9h. 17m. 20s., Cincinnati local mean time. 

 Phase of shock unknown. 



From Crawfordsville, Ind. (622 miles), E. C. Simpson, 

 C.E., reports through Prof. J. M. Coulter, of Wabash 

 College: " Suddenly felt my chair move, jumped up and 

 said, ' We are having an earthquake ' ; at once pulling out 

 my watch I found it was 8h. 54m. p.m. standard time 

 (Central)." Prof. Coulter adds that the watch was 

 exactly with railroad time as shown at the railroad 

 station, and also by the town clock. 



From Dyersburg, Tenn. (569 miles), Louis Hughes 

 writes : — " My time-piece was an English patent lever 

 watch of Chas. Taylor and Son, London, which from 

 business necessity I keep closely with railroad time at 

 the station, which receives the time at 10 o'clock every 

 morning. The railroad uses Central time. My first 

 thought was that the shaking was caused by the children 

 in the next room ; but in the next moment, recognizing the 

 peculiar sensation, I dropped the newspaper and observed 

 the time, which was probably four to six seconds after 

 8h. 54m,, and from that approximated it in even minutes." 

 Speed 3'25 miles, or 5230 metres. 



At Memphis, Tenn. (590 miles), the Signal Service 

 Observer reports a considerable number of stopped 

 clocks, one at gh. 54m. and the others at 9h. 55m. For 

 some unaccountable reason the seconds were not noted. 

 The phase is unknown. 



The foregoing comprise those time reports which seem 

 to justify the presumption that the errors do not exceed 

 one minute. There are others, which are obviously rude 

 approximations, giving exact hours, quarter-hours, or tens 

 of minutes. There are also some which look at first like 

 good observations, but which surely involve some large 

 unexplained error. 



As the discussion of the time data is now progressing, 

 no further comment will be offered here beyond the 

 remark that there can be no doubt that the speed of 

 propagation exceeded 3 miles, or 5000 metres, per 

 second. The only questions are how much this speed 

 was exceeded and whether the speed along any given 

 line was constant. As regards the latter question, the 

 data are not yet precise enough to justify an opinion. 

 This matter will be inquired into. 



The high rate of propagation will probably prove un- 

 expected to European seismologists. We propose, how- 

 ever, to follow it up with the suggestion that it is about 



the normal speed with which such waves ought to be 

 expected to travel, and that all determinations of the rate 

 of propagation in any former great earthquakes which 

 are much less than 5000 metres per second (for normal 

 waves at least) are probably erroneous in proportion as 

 they fall short of the Charleston earthquake. Finding as 

 the time reports accumulated that a speed in excess of 

 5000 metres was indicated, and this presumption having 

 become a conviction, we were led to inquire whether 

 there was not some speed deducible from the theory of 

 wave-motion in an elastic solid to which all great earth- 

 quakes ought to approximate. 



In a homogeneous and perfectly elastic solid, the rate 

 of propagation is, according to theory, dependent upon 

 two properties of the medium : elasticity and density. 

 There are two coefficients of elasticity in solid bodies, one 

 of which measures their resistance to changes of volume ; 

 the other, to changes of form. Absolute e.xperimental 

 determinations of the values of these coefficients have 

 never been made. If, however, we knew the ratios of 

 these coefficients in one substance to the homologous 

 coefficients in any other substance, and if we also knew 

 the rate of propagation in either of them, the rate in the 

 other would be at once deducible. The rate in steel bars 

 has been the subject of much experimentation, and is given 

 by Wertheim, whose researches have been as careful as 

 any, at 16,800 feet per second. But as the waves in a 

 steel bar are essentially waves of distortion, he multiplies 



this result by ^/ ^ or -- for the normal wave, giving a 



speed of.2i,ODO feet per second. The elastic modulus of 

 steel for engineering purposes is usually taken to be 

 29,000,000. The corresponding modulus for such rocks 

 as granite and basalt in a very compact state is about 

 8,000,000. If we may assume that these moduli are pro- 

 portional to the two elasticities of the two substances re- 

 spectively, we can compute the rate of propagation in 

 rock. This assumption may or may not be true ; but we 

 assume it to be so. Let V^ be the rate of propagation in 

 steel, and "VV the rate of propagation in rock, and let es 

 and e,- be their true compressional elasticities, and let 

 Ds and Dr be their respective densities. Our assump- 

 tion is that 2() -.Z : -.es : er from which we may form the 

 equation — 



V. ^ n 



Taking the density of steel at 7*84, and of deeply-buried 

 rocks in their most compact state at 2 '85 — 



——2 = 115 nearly. 



D, 



V,. 



\' 



/_29_ 



^^ 7-8^ 



Taking the rate of compressional waves in steel to be 

 6400 metres per second, gives 5570 metres for similar 

 waves in very compact and dense rock. The correspond- 

 ing rate for waves of distortion would be 4450 metres. 

 These results are so near to those deduced for the 

 Charleston earthquake that they seem to be worthy of 

 consideration. 



The experimental measurements of the rate of impulses 

 obtained by Milne and Fouque seem to us inapplicable. 

 The elasticity of the surface soil, we think, is no more to 

 be compared with that of the profound rocks which 

 transmit the great waves of an earthquake than the 

 elasticity of a heap of iron filings is to be compared with 

 that of an indefinitely extended mass of sohd steel. The 

 difference is toto coelo. But the rate of propagation is a 

 question of elasticity and density chiefly. The effect of 

 temperature we have not considered. Perhaps the most 

 striking experiment ever made with an artificial earth- 

 quake was at the Flood Rock explosion at Hell Gate, near 

 New York, where General Abbott found a speed of pro- 

 pagation approaching very closely to that of the Charleston 

 earthquake. 



