TERRESTRIAL MAGNETISM. 183 



parisou of our survey with those of our predecessors fifty and thirty 

 years ago, by M. Moureaux's results in France, and by Captain Creak's 

 collation of previous observations. Yet, in spite of this, Stonyhurst, 

 which is some 200 miles north of Greenwich and Kew, and should there- 

 fore outrun them, sometimes lags behind and then makes up for lost 

 time by prodigious bounds. Between 1882 and 1886 the total secular 

 change of declination at Stonyhurst was about 3.5 minutes less than 

 that at Greenwich and Kew, whereas in the two years, 1890-92, it 

 reached at Stonyhurst the enormous amount of 28 minutes, just doub- 

 ling the corresj)ondiug alteration registered in the same time at Kew. 

 If these fluctuations are caused by the instruments or methods of re- 

 duction, my argument in favor of frequent comparisons and uniform 

 treatment would be much strengthened, but, apart from the inherent 

 improbability of such large differences being due to the methods of 

 observation, the i^robability of their physical reality is increased by 

 the work of the magnetic survey. 



The large number of observations at our disposal has enabled us to 

 calculate the secular change in a new way, by taking the means of 

 observations made about five years apart at numerous though not iden- 

 tical stations scattered over districts about 150 miles square. The 

 result thus obtained should be free from mere local variations, but as 

 calculated for the southeast of England for the five years 1886-91, it 

 diff"ers by nearly 5 minutes from the change actually observed at Kew. 



We have also determined the secular change at twenty-five stations 

 by double sets of observations made as nearly as possible on the same 

 spot at intervals of several years. The results must be interpreted 

 with caution. In districts such as Scotland, where strong local dis- 

 turbances are frequent, a change of a few yards in the position of the 

 observer might introduce errors far larger than the fluctuations of 

 secular change. But when all such changes are eliminated, when all 

 allowance is made for the possible inaccuracy of field observations, 

 there are outstanding variations which can hardly be due to anything 

 but a real difference in the rate of change of the magnetic elements. 



A single example will suffice. St. Leonards and Tunbridge Wells 

 are about 30 miles apart. Both are situated on the Hastings Sand for- 

 mation, and on good nonmagnetic observing ground. At them, as at 

 the stations immediately around them — Lewes, Eastbourne, Appledore, 

 Etchingham, Heathfield, and Maidstone — the local disturbing forces 

 are very small. All these places lie within a district about 40 miles 

 square, at no point of which has the magnet been found to deviate by 

 5 minutes from the true magnetic meridian. ISTo region could be more 

 favorably situated for the determination of the secular change, yet 

 according to our observations the alteration in the declination at St. 

 Leonards in six years was practically equal to that at Tunbridge Wells 

 in five. It is difficult to assign so great a variation to an accumulation 

 of errors, and this is only one among several instances of the same 

 kind which might be quoted. 



