ON SEISMOLOGICAL INVESTIGATION. 63 



The assuruptioii that for large earthquakes, at least, the origin has 

 been at an epicentre rather than in a region at a certain depth below the 

 surface, is, so far as velocity determinations are concerned, of but small 

 importance. Although all stations have similar instruments, the recot-ds 

 from one or two of them indicate that their adjustment has not been 

 similar to that adopted at the remaining stations. Not only should each 

 instrument have a period of 15 seconds, but when its boom is deflected 

 7 or 8 mm. from its normal position, and then set free, it should take 

 7 or 8 minutes before returning to rest. If this latter condition has not 

 been observed, an instrument may not respond to the first preliminary 

 tremors, with the result that tJie time recorded for the commencement of 

 a given earthquake may be registered as one or two minutes after the 

 true time. 



Although errors of this order may aftect the results deduced from 

 observations within 20° of an earthquake origin, when we deal with paths 

 of greater length, and especially with large waves, the errors in the final 

 results are practically inappreciable. 



Another assumption made in connection with velocity determinations 

 is that the group of vibrations and waves as recorded at a distant station 

 extending between the first preliminary tremor and the first maximum — 

 which may extend over any interval up to 100 minutes — were all the 

 result of the principal movement or movements at the origin ; or, in other 

 words, they have the same initial times. To this assumption I do not 

 know of any serious objection. The fact that pronounced phases of move- 

 ment near to an origin are not only extended in time as they radiate, but 

 are also more or less equalised in their amplitude, frequently renders the 

 determination of corresponding points in seismograms obtained at different 

 stations more or less uncertain. This source of error is sometimes serious. 



4. Preliminm'y Tremors. 



In the compilation of the following table the only seismograms used 

 are those which show a distinct commencement. Each earthquake is 

 indicated by its British Association Register number, and the locality 

 from which it originated. Following this are the initial letters (see p. 88) 

 of the station or stations at which it was observed. The figures following 

 these initial letters give the number of minutes taken by the preliminary 

 tremors to reach these stations, and the number of degrees between the 

 stations and the earthquake origins. These figures are respectively placed 

 in positions corresponding to the numex'ators and denominators of fractions. 

 If an initial letter is followed by a zero for a numerator, this indicates 

 that all other time intervals are measured relatively to the observa- 

 tion made at the station represented by the initial letter. 



The fewness of these records chiefly arises from these facts : first, 

 they only refer to earthquakes with a known origin ; secondly, the 

 seismograms of small earthquakes recorded at distant stations do not 

 show the preliminary tremors corresponding to those given by large 

 earthquakes ; and lastly, in consequence of air tremors and other causes, 

 the earlier vibrations have in many instances been eclipsed or lost. Their 

 chief merit is that they give for several earthquakes records from point 

 to point, and that we have for the first time records relating to paths 

 which practically extend from an origin to its antipodes. 



