243 



the stray flux f\ of the field magnet, in a manner sensibly proportional 

 to the increase of the ampere-turns of the field. If, therefore, the 

 field-magnet ampere-turns are increased by 



x/7 



in order to compensate for the counter-ampere-turns, CA T, of the 

 armature, the stray flux /i produced by the field magnet following 

 the circuit BADC would be increased by a quantity, 



o.27rKN 2 I d \ / ~2 o.znKNJdV^ (CAT] 



R ~ approximately =0.4-:^ - , 

 K f , K fl 



R l being the reluctance of the field magnet. 



This increase of the flux through the field magnet increases the 

 magnetic density in the latter and demands consequently a correction, 

 as I pointed out in 1899, without tracing it in detail. At that time 

 I conducted the inquiry simply as follows, supposing the role of the 

 field magnets to be sufficiently unimportant to permit approximate 

 correction being applied. 



Let BI be the flux density in the field magnet, corresponding to 

 the no-load E.M.F. e, that is to say, to the flux </> a +/i; and let us 



call Vi the Hopkinson coefficient ^-~ . The full-load induction 



<t>a 



will be 



approximately 



=1 i + 



and consequently the total ampere-turns will be increased by the 

 quantity by which the change from BI to B\ increases the ampere- 

 turns (which we shall call N\I\) specially absorbed by the reluc- 

 tance of the field magnets in the condition considered. The point 

 d' upon the curve will be in consequence displaced toward the left by 

 the quantity corresponding to this increase of ampere-turns. 



Starting from the point Q' duly corrected, it is sufficient to take 

 a length representing the ampere-turns equal to CA T of the armature 



