TIME OF EARTHQUAKE : RATE OF PROPAGATION. 283 



within a few seconds of error. 1 Unfortunately the human equation of 

 error, and also the individual clock errors spread over the hours elapsing 

 General unreliahil- since 4 p.m of the preceding day must have had 

 ky of the recorded an incalculably large influence on the registration 

 on the earthquake °* those times as recorded in the earthquake forms, 

 forms. for, an examination of the latter shows them to 



be inconsistent with one another. Leaving out of account all vaguely 

 expressed or guessed times, and those given without any accurate refer- 

 ence to the standard taken (of which the number is legion), and taking 

 for consideration only those substantiated with the statement that the 

 time-piece had been checked with telegraphed Madras time (or railway 

 time, which was then the same), it nevertheless is at once evident 

 that the great majority, even of these so-substantiated times, are 

 hopelessly unreliable. This is made plain from a mass of instances 

 that might be given showing that in the same town or locality or district 

 (if we are to credit the so-recorded times) the shock occurred at impos- 

 sibly varying times. 



A few examples will be sufficient to establish this : — 



Locality. Time, said to have been verified by 



telegraph from Madras, 





Kashmir. 





H. 



M. 



Dras . 







5—50 



Srinagar 



. 



. 



6- 



-10 



Poonch 



... 



. 



6- 



-20 



Astor-Gi] 



git road . 



• 



6- 



-30 



1 Mr. G. T. W. Olver, Superintendent, Indian Telegraphs, informs hie that atl5-55all 

 work on all lines is stopped, and beats are given for 4 minutes. When the actual time, 

 16 hours, is reached, the letter T (indicating " time ") is signalled from Madias to 

 certain maiu centres. At these the operators are all waiting, and at the signal they 

 immediately repeat "TV again to further centres, and so on. At the most a loss of 

 1 second with each repetition might be expected. 



