XVI 



Introduction. 



16. The Suspension Thread and its Torsion Force. 



The errors due to the torsion of the suspension thread are produced in two 

 ways. First, by the magnet moving out of the plane of detorsion ; secondly, by 

 the variation of this plane, due generally to the varying humidity of the atmosphere.* 

 The greatest change of declination from the mean has been within 2° ; the greatest 

 correction for the torsion from this change would be within 8'''. The approximation 

 of the line of detorsion to the magnetic meridian is probably seldom within 2° ; it is 

 found on trial to have deviated from 5° to 40° from the magnetic meridian, and is so 

 variable as to swallow up completely the changes from varying declinations. 



If the magnet be deflected u' by turning the arms of the torsion circle u/, the 

 torsion is w^—u', the ratio 



coefficient of torsion force 



ro' — v! magnetic force x moment of free magnetism 



H 

 F 



is the quantity by which the deviations of the magnet from the plane of detorsion 

 should be multiplied, to obtain the decrements due to torsion force. If n be the 



1 + — W = the true deviation. 



H 



The observations for the values of — are given below ; no use has been made 

 of them for this correction, from the reason given above. 



Table 6. — ^Values of 



H 





!<;=90° 





Period to which the factor 



Date of the 

 Observation. 





H 

 F 



H ,. 

 1 + —applies. 



u 



1841. 









July 8 



4-63 



0-00086 



July 8—20 



Jtdy 17 



4-63 



000086 





July 30 



5-40 



000100 



July 26— Dec. 4 



Dec. 28 



4-91 



0-00091 



Dec. 22 — Jan. 19 



1842. 









March 3 



5-36 



000099 



Jan. 21 — May 24 



June 6 



4-74 



0-00088 



May 24 — June 6 



June 1 8 



4-00 



0-00075 



June 7— June 26 



June 21 



4-29 



0-00080 



June 7 — June 26 



June 27 



4-14 



0-00077 



June 27— July 18 



Aug. 22 



8-73 



0-00162 



July 20— Dec. 31 



17. The second and most important error due to the torsion force, is from the 

 varying plane of detorsion. Unless when the period and extent of change is known, 



* A thick cotton cover was put over the whole Declinometer in January 1844, which seems to 

 remedy this very much. 



