710 DABBYSHIRE [CHAP. 20 



in this connection was suggested by Macelwane (1946) and many other names, 

 including Ramirez (1940) and Gilmore (1946), were associated with this project. 

 The method used for storm tracking was a very simple one in principle — the tri- 

 partite method, which consists of three seismographs set at the corners of a 

 triangle and measuring the time difference of the arrival of an individual 

 identifiable wave on the three instruments. This method had been used pre- 

 viously to locate earthquakes and in the sound ranging of guns. It can be 

 shown that for a constant time difference between two seismographs, the locus 

 of the source is a hyperbola and with three instruments the source is at the 

 intersection of the hyperbolae. This method requires an identifiable wave, 

 which is not easy to obtain in practice with microseisms. Shaw (1922) showed 

 that there was no apparent similarity between seismograph traces taken more 

 than 10 miles apart and the side of the triangle was usually about 600 m in the 

 American project. The time differences were thus of the order of a fifth of a 

 second and so great precision was required in synchronizing the instruments. 

 Horizontal seismographs were used and these led to further complications as 

 the waves received would be a mixture of Rayleigh and Love waves and in- 

 dividual waves would be even more difficult to identify as the two kinds travel 

 at different speeds. Despite these difficulties, some success was achieved and 

 Gilmore (1946) has published results showing the successful tracking of hurri- 

 canes. The method was not consistently successful, however, and Donn and 

 Blaik (1953) concluded that the success gained was not commensurate with the 

 amount of labour and money spent. 



This indifferent success led Gilmore (1953) to a new empirical approach to 

 the problem. He argued that every storm giving rise to microseisms should 

 produce a definite ratio of intensity at two stations, dependent only on its own 

 position. Data collected over eight years were used to prepare micro-ratio 

 charts in which iso-ratio lines were drawn for pairs of stations. Some success 

 has been claimed for this method but the difficulties are still considerable as 

 seismograph characteristics have to be kept constant over a number of years. 

 It is also not at all clear that Gilmore's original assumption is valid, for if 

 microseisms are formed by the interference of trains of waves meeting head on 

 in areas which the storm has passed through, two storms with identical positions 

 of storm centre may have had entirely different previous histories so that the 

 regions of wave interference would be entirely different. Carder and Eppley 

 (1959) have checked this technique with several groups of storms at seven 

 microseism stations. It was found that, for storms in the same position, the 

 ratio of the amplitudes at two given stations could vary by a factor of eight. 

 On cutting the number of stations to three, a variation of three times was 

 found even when the work was limited to hurricanes occurring in 1954 and 1955 

 to reduce any variation in the seismograph characteristics. For these cases, 

 however, if an allowance is made for the path of the storm, assuming the 

 microseisms at a given time were generated not at the position of the storm 

 centre at that time but at its position 18 h or 24 h previously, so that waves 

 could have had time to interfere, the variation was reduced to 1.7 to 1. It was 



