14 Ml*. A. Campbell, on the Magnetic Flaxes in 



to 15 per cent, too small ; but it proved a very convenient way 

 of comparing the flux-densities at various positions. It was 

 thus found that when the main current (full load) was switched 

 on in additioLi to the shunt-current, the flux-density under 

 the middle pole was nearly doubled. The most curious fact 

 brought to light by the method was that across the air-gap of 

 the permanent magnet a considerable alternating flux is in- 

 duced, the value of B at that position being actually about 

 5 of that between the poles of the shunt-magnet with the 

 shunt-current alone. The permanent magnet forms a kind of 

 secondary magnetic circuit directing around itself the alter- 

 nating eddy-currents in the disk. Whether this has any 

 sensible demagnetizing effect upon the permanent magnet the 

 writer has not determined. 



The fhixes were measured more exactly by the method of 

 the search-coil and thermopile described above. With shunt 

 alone the root of mean square B was about 50 in the air-gap 

 and 800 in the iron core just above the shunt-bobbin. The 

 shunt-current at 100 volts (86 ~ per sec.) was found to be 

 0'031 ampere. If the current followed a sine curve its maxi- 

 mum value would be 0*044. When a direct current of this 

 value was tried the fluxes produced were much larger than those 

 with the (supposed) equivalent alternating current. This is 

 partly due to the fact that the shunt-current does not follow 

 the sine law, but is no doubt also due to the existence of eddy- 

 currents in the iron core. That these currents even in the 

 disk spread the flux anyd reduce the flux-density in the air-gap 

 was shown qualitatively by placing a search-coil, connected 

 with a telephone, in the air-gap of a ring electromagnet 

 excited by alternating current. The sound in the telephone 

 was lessened when a copper disk was held near the coil in the 

 air-gap. The search-coil and thermopile method showed that 

 the flux-density had been reduced by 8 per cent. 



To find how the core-flux varied with alteration of the 

 voltage on the shunt-coil, by the same method the B just 

 above the shunt-bobbin was measured for a number of voltages 

 from 20 up to 100 volts. It will be seen from the curve in 

 fig. 7 that B is very neaily proportional to the potential- 

 ditference. In practice it is found that the speed of rotation 

 is very nearly proportional to the voltage. 



Power spent in Meter. — To approximate to the amount of 

 power spent by reason of hysteresis when the shunt-current 

 alone is on, the iron was carried through a cycle by means of 

 direct current of such amount as to give a maximum B nearly 

 corresponding to that given with alternating current at 

 100 volts. The curve obtained is shown in fig. 8. 



