Propagation of the Charleston Earthquake. 13 



In the above table the comparison at Cincinnati takes account 

 only of a single clock, whose error happened to be known ex- 

 actly. The time of beginning in that city is also known with 

 exceptional certainty and accuracy. It will not differ more 

 than eight or ten seconds from 9h. 16m. (Cincinnati local 

 mean time or 9h. 53m. 41s.). If we consider Cincinnati and 

 suburban towns within fifteen miles of the city which are 

 supplied with local time from the Cincinnati observatory, we 

 have no less than twenty-two time reports, of which nine are 

 stopped clocks. Two personal observations giving 9:15 local 

 have been rejected because they are multiples of five. One 

 report giving 9:17:45 has been rejected because its author, be- 

 sides indicating that it refers to an advanced phase, throws 

 doubt on his own observation. Of the remaining ten personal 

 observations one gives 9:15:40, eight give 9:16, and one gives 

 9:16:30. Of the stopped clocks, three were in the central of- 

 fice of the Western Union Telegraph Co. They kept standard 

 time and were read only to the nearest minute. All three are 

 reported to have stopped at 9:54. The clock in the fire tower 

 is the one whose error was known. Its corrected reading was 

 9:16:40. The remaining clocks gave (9:15), (9:16), (9:17), (9: 

 17:20), and (9:19). Four of the latter were from the suburban 

 town of Lockland. Reducing to standard time and taking their 

 mean, the ratio of the time-interval by stopped clocks to that 

 by personal observation is 1.26, a result identical with that de- 

 rived from the clock in the fire tower alone and nearly the 

 same as that in the table. There is reason to believe, however, 

 that this ratio is a little too great for the mean of stopped 

 clocks throughout the entire country, and especially so for 

 those of very distant localities ; for if the ratio were uniform, 

 the absolute differences between the two kinds of data would 

 be very wide in remote regions and small near the centrum. 

 This is not the case. The absolute differences at very remote 

 localities are very little, if any, greater than those at the middle 

 distances. This difficulty prevents us from assigning any 

 specific value to the correction and from determining its prob- 

 able error. Nevertheless the comparisons just made indicate 

 that the systematic error is probably of such magnitude that, if 

 due allowance were made for it, the corrected result for the 

 stopped clocks would not differ much from those of the pre- 

 ceding groups. While this group furnishes evidence which 

 strongly supports the approximate correctness of the results of 

 the other three it cannot be a source of greater precision nor 

 can it furnish the means of reducing the final probable error. 



