300 



Vartation of the cgmpasS. 



Remarks, &c. at the Royal Observatory at Paris, between the years 1783 

 and 1788, but the effect was considerably greater in his ob- 

 servations, than in those menlioned above ; his results how- 

 ever were, in my opinion, drawn from too few observations, 

 being from only 3 days observations about the times of the 

 equinoxes and solstices, which differ considerably among them- 

 selves; and experience teaches us, that magnetical obser- 

 vations, made for a period so limited, are not sufficient for 

 minute purposes : I have, therefore, in the results here given, 

 taken the mean of the observations made during the whole 

 month in which the equinoxes and solstices fall, which appear 

 to me likely to furnish results more satisfactory; and all the 

 foregoing observations are to be considered as the results or 

 mean of a great many, by way of arriving at greater accu- 

 racy than could be obtained without; this, however, was 

 found to be more necessary at some times than at others; 

 sometimes the needle would be extremely consistent with 

 itself, so as to return exactly to the same point, ho\Vever 

 Unsteadiness often it might have been drawn aside ; at other times it varied 

 of the needle. 2 or 3', sometimes 8, 10', or even more: this uncertainly in 

 the needle arises principally, I believe, from changes "tn the 

 atmosphere, for a change of wind, from any quarter to ano- 

 ther, almost always produced a change in the needle from 

 steady to unsteady, and vice versa, but it was generally more 

 unsteady with an easterly wind than when it blew from any 

 other quarter, and most steady when the wind was south or 

 south-westerly. An Aurora Boreaiis always produced consi- 

 derable agitation of the needle. 



It has been mentioned in this Paper, that the annual in- 

 crease of variation was found about the year 1775 to be nearly 

 10'; and was considered at that time to be gradually dimi- 

 nishing ; but it is remarkable that this rate of increase ap- 

 pears, from the annexed Table, to be nearly the same at 

 which it has been found to move between all the different 

 periods in the said Table, from 1.5S0 to 1787, a period of 

 more than 200 years, excepting between the years 1692 and 

 1723: the observations of Halley in 1692, and Mr. Graham 

 in 1723, make the annual increase 16'; to what this diffe- 

 rence could be owing I am at a loss to account : on referring 

 . to observations made at Paris for those two years, the annual 

 increase is H'; subsequent observations made by Mr. Gra- 

 ham 



Annual in- 

 crease of the 

 variation. 



