ON COMPARING AND KEDUCING MAGNETIC OBSERVATIONS. 2& 



storms in Great Britain and those of declination disturbances at the Kew 

 Observatory. As we continued this investigation we came to the conclu- 

 sion that the best method of procedure would be to compare together 

 what may be termed wind-weather and declination-disturbance-weather, 

 in order to see if there is any apparent connexion between them ; our 

 hope of a positive result being strengthened by the belief that there is 

 accumulating evidence in favour of a connexion of some kind between 

 the convection currents of the earth and the oscillations of terrestrial 

 magnetism. 



We shall, therefore, begin by defining precisely what we mean by 

 wind weather and by disturbance weather. We have obtained, through 

 the kindness of the Kew Committee, records of the total amount in miles 

 gone over by the wind at Kew for each day of the years 1858-73 (sixteen 

 years in all), and we have likewise obtained from the same source daily 

 aggregates of the disturbance of magnetic declination at Kew separated by 

 Sabine's method. 



To begin with the wind values, we have first of all smoothed these 

 down into daily averages of three days. Let us call this Table A. 



We have next obtained a Table B, where each day's value is the 

 average of 25 days of Table A, all being properly placed as regards 

 dates. 



Next, taking the difference between the entries of Tables A and B, we 

 obtain a series representing departures from the mean — -^^Zms when in 

 excess, and minus when in deficiency — which may be taken to represent 

 tvind weather. The declination aggregate daily disturbance numbers (for 

 which the nnit is -jJ^ of an inch measui'ed on the curve) have been 

 treated in exactly the same way as the wind numbers, and the differences 

 finally obtained have been taken to represent disturbance weather. The 

 values representing wind weather have then been formed into series of 

 twelve terms, each so chosen that maximum wind values come too-ether at 

 the middle of each series. The yearly sums of these series, as well as four- 

 yearly sums and total sum for 16 years, are exhibited in Table I^. 



The disturbance weather values have then been arranged into series of 

 twelve terms each, so that each entry is two days previous in date to the 

 corresponding entry of Table I^. The yearly, four-yearly, and total sums of 

 these series are given in Table I,,. 



The values representing wind weather have next been formed into 

 series of twelve terms each, so chosen that minimum wind values come 

 together at the middle of each series. The yearly, four-yearly, and total 

 sums of these series are given in Table 11,. 



Finally, the disturbance-weather values have been arranged into series 

 of twelve turns each, so that each entry is two days prior in date to the 

 corresponding entry in Table 11^, and the sums of these are entered in 

 Table 11^. 



The general results of the comparison for the sixteen years are shown 

 in the accompanying table, and if averages be taken from these for the 

 three minimum sunspot years 1865-6-7, and for the three maximum years 

 on either side thereof, we get 



1858-59-60 

 1865-60-67 

 1869-70-71 



