Mr. R. H. M. Bosanquet on Electromagnets. 533 



%. This was determined by the method given by Rowland 

 (Phil. Mag. xlvi. p. 140). The mean radius of the earth 

 induction-coil is 25*65 centim. It has 21 turns, 3 layers of 7. 

 The values of H and V employed were determined nearly 

 every day. Latterly H alone was employed. 



Earth's force. — The magnetic potential was taken as H x 

 length or V x length, as the case might be. The absolute 

 values of H and V disappear from p. 



Ampere-turns. — The currents were measured by two Helm- 

 holtz galvanometers, one having 18 coils and the other 2 coils ; 

 circumference 1 metre. The galvanometer-constant was kept 

 in the form CH, so that the values of H obtained daily were 

 conveniently employed. 



Magnetic resistance. — The magnetic potential was thrown 

 into the form PH tan 8. The induction also involved H. 

 The quotient of these quantities was therefore independent 

 of H. To secure identity of H in numerator and denomi- 

 nator, the tangent-galvanometer and earth induction-coil 

 were always adjusted on the same standard place ; being used 

 alternately, the one could be removed to make way for the 

 other. 



Moment. — The moment was obtained by observing the 

 deflection of a needle by the electromagnet placed at a known 

 distance E or W. The scale and mode of computation em- 

 ployed in the absolute determinations were used. The value 

 of H was always determined separately in the position of the 

 deflected needle, by observing the deflection produced by the 

 standard magnets, with which a complete determination had 

 been made in the standard position the same day. The moment 

 being known from the standard determination, H can be 

 obtained from the deflection only. 



These are all the observational data. With respect to /, I 

 define the foci of a bar as being such that, if F be their dis- 

 tance, F x lines of force = 47rM. This would be true if 

 these points were true poles, and the points of application of 

 the resultants of the lines of force. Then, to get a more 

 general datum, I put F/l=f y so that the numbers from differ- 

 ent bars may be directly comparable. 



To study the values I take §3 for abscissa, conveniently so 

 that 1000 is represented by 1 centim. Then any of the other 

 columns may be represented by ordinates. p and /are the 

 most important. In drawing the curves for p, I adopt a scale 

 of 10 centim. for 1 centim. of p*. It is easy to see how far 

 the curves for / would assume the same general character. 



* The woodcut is drawn to about half these dimensions. 



