1909.] OUTLOOK OF SEISMIC GEOLOGY. 263 



creasing values as they recede from it, one who has not compared 

 the individual data will scarcely believe. A local intensity which is 

 too large can be explained either by a soft or a wet basement, by an 

 earthquake " bridge " or by probable error of observation ; while 

 one too small may be explained by an earthquake " shadow," by an 

 interference of waves, etc. Many curiously anomalous data not 

 possible of explanation on any of these grounds may be dismissed 

 as " earthquake freaks." 



As regards time of arrival of shocks, " too early " or " too late" 

 data have not uncommonly been included among those which seemed 

 a priori the most reliable. Especially good examples of such data 

 are furnished by the studies of the Agram earthquake of 1880, the 

 Andalusian earthquake of 1885, the Charleston earthquake of 1886, 

 and the Indian earthquake of 1887. Out of 260 time data collected 

 by Dutton in connection with the Charleston earthquake, 47 were 

 rejected as " too early." 



To average the determinations of an unvarying value in order to 

 eliminate the errors of observation and experiment, is indication of 

 a desire to secure accuracy which must be commended as eminently 

 scientific in its nature ; but to average the values of a property the 

 distribution of which either in space or in time is likely to be sig- 

 nificant, is, on the contrary, one of the most pernicious, as it is one 

 of the most common and unconscious methods. Such a practice is 

 often condoned on the ground that the data may otherwise appear 

 to possess an accuracy beyond what they really have ; forgetting, what 

 is far more important, that through the averaging process the data 

 lose their most significant characters. Now that so many sciences 

 are entering upon their quantitative stages it is important that this 

 method be corrected. 



A companion fallacy to the supposed necessity for averaging 

 data of different values is that nature in all its moods has avoided 

 angles and straight elements in favor of the curving outline, and 

 that in consequence results are incorrect in proportion as they bring 

 out strong accent, or definiteness of character, or exhibit straight- 

 ness of contour. In no field, perhaps, has this fault been more often 

 committed than in topographic mapping, where it has been encour- 

 aged as tending toward accuracy. A new era is dawning, however^ 



