AT TWENTY-ONE STATIONS IN EUROPE. 77 



the loss of magnetism by the needle is so fully made out, that I believe the re- 

 sults to be as free from error, on this score, as if no loss had appeared from 

 observation. The time of oscillation of the bar (A) was observed at Philadel- 

 phia in October, 1835; again in September, 1836, just before I set out from 

 home; and in December, 1838, after my return: it was observed in the inter- 

 mediate time at London, in June, 1837, and again in August, 1838; at Paris 

 in August, 1837, and in July, 1838. A curve was traced on a large scale by 

 the results thus obtained, the ordinates representing the relative forces of the 

 needle corresponding to the intervals of time from October, 1835, measured by 

 the abscissae. A regular curve being traced, departed very little from the points 

 obtained by observation, the differences between the ordinates of the mean 

 curve and those given by the particular observations being, in terms of the ori- 

 ginal force of the needle, 0.000, — 0.0005, + 0.0030, -f- 0.0034, and — 0.0031. 

 As these individual results must be affected hj small errors of observation, there 

 can be no doubt of the satisfactory correction for loss of magnetism by using 

 the ordinates of the curve, and, accordingly, the correction thus obtained is ap- 

 plied to the results, and is entered in the tables. This needle shows a tendency 

 towards a permanent magnetic state, and its loss is less than half that of the 

 other. The diminution of force of the cylindrical needle, (B,) since Septem- 

 ber, 1835, has been nearly uniform, and, accordingly, the curve representing it 

 differs but little from a straight line. The observations used to trace this curve 

 were obtained at Philadelphia in September, 1835, in September 1836, and in 

 December, 1838; and in addition at the same times and places as stated in 

 reference to the other needle. Although the correction for loss of force is so 

 much greater for this needle than for the bar, there is no reason to suppose, 

 from a comparison of their results, that this correction is not quite as well as- 

 certained as the former. The differences between the observed losses of force 

 and those given by the ordinates of the curve are, 0.0000, 0.0000, + 0.0035, 

 + 0.0051, and — 0.0042. The time of oscillation of this needle was farther 

 satisfactorily observed at Florence, before passing into lower Italy, and again in 

 returning to upper Italy; but while the general accordance of the results was 

 such as to show that the force had undergone no irregular change which was ap- 

 preciable, the time which had elapsed between the two observations was too short 

 to justify their use in the numerical determination of the loss of magnetism by 

 the needle. I have had no cause to suspect irregular changes in either of the 

 VII. — u 



