10 



DISCUSSION OF THE HORIZONTAL COMPONENT 



The result from this combination + 1.3 confirms the preceding value, the result, 

 according to weight or + 1.5 scale divisions or 5=0.0000548 in parts of the hori- 

 zontal force has, therefore, been adopted in the reduction of the bifilar readings to 

 a standard temperature, for which + 63.0 Fahr. has been determined upon as the 

 mean temperature of the magnetic bar during the five years series of observations. 



The difference in the resulting value for q, when obtained from deflections or 

 vibrations, and from combinations of the bifilar readings themselves, has been re- 

 marked before, and no satisfactory explanation has as yet been given of it. Thus, 

 for instance, at Toronto, the two respective values were 2.69 and 1.63 scale divi- 

 sions, as shown in General Sabine's remarks (Vol. III.) The existence of a similar 

 discrepancy in the case of the Makerstoun bifilar has been detected by Mr. Broun. 

 Whatever may be the cause of the difference, there can be no hesitation in saying 

 that the result derived from the bifilar observations themselves is the one to be pre- 

 ferred. At St. Helena (Vol. II., London, 1860), the two values were 1.45 and 0.98, 

 the half yearly comparisons at this station even show a less value, viz., 0.88 scale 

 divisions; 0.98 (for convenience 1.0) was adopted hi the reduction. Dr. Lamont, 

 in his Handbook of Terrestrial Magnetism (p. 206, edition of 1849), says: " It de- 

 serves to be remarked that the value obtained by comparing monthly mean readings 

 of the bifilar at high and low temperatures is smaller than that obtained by direct 

 observation." 



In the present discussion the value ? = _! 



k 0.0000365 



1.5 has been adopted. At 



Toronto this va!uo was? , _ 1.63, and at St. Ho.cna f-^ - 1.0. 



It will be seen from these values that the Philadelphia bifilar magnetometer was 

 very sensitive; its scale value in parts of the horizontal force is but four-tenths of 

 the Toronto value, and only two-tenths of that of the St. Helena instrument. 



In the computations which follow the tenths of scale readings have been omitted 

 (keeping only the nearest unit) as contributing nothing to the accuracy of the results, 

 and merely increasing the labor of reduction. The uncertainty in the readings 

 arising from the uncertainty in the value of q probably affects the units, and the 

 same may be said of the declination changes, so that in extreme (individual) cases 

 the next higher figure may be affected. 



