the Magnetic Effect of Electric Convection. 451 



allow of the measurement of magnetic fields as low as 10 ""^ 

 c.G.s. — the order of the field in most convection experi- 

 ments. Further, from the nature of its construction, a 

 sensitive astatic system always tends to drift in a certain 

 direction, due to the slow and unequal demagnetization of 

 the needles which compose it. Another cause of instability 

 is that the direction assumed by such a system is determined 

 by the difference between the earth's field and that of the 

 compensating magnets : consequently any slight variation in 

 either of these fields will produce a considerable change in 

 the i)osition of equilibrium of the system. In particular, if 

 the field to be studied is produced by oscillatory phenomena 

 more or less dam]:)ed, astatic systems become practically use- 

 less, since the demagnetizations which accompany these 

 damped oscillations not only afiect the system itself, but also 

 the directing magnets. Finally, in the case where the field 

 to be studied is very feeble, it seems advantageous to increase 

 the effect by using strong magnetic poles ; but it is well 

 known that the sensibility of an astatic system is independent 

 of the moments of the magnets which form it. 



All these inconveniences can be avoided, without decreasing 

 the sensibility, by employing an extremely simple system. 

 The system consists essentially of a light horizontal beam 

 carrving at one end a vertical mag-net and at the other a nou- 

 magnetic counterweight of brass. The whole is supported 

 by a long fine metallic wire fixed to the centre of the beam. 

 If the magnetic axis of the magnet is exactlv vertical, the 

 couple to which this magnet is submitted due to the earth's 

 field will produce no action on the torsion wire. The beam 

 will therefore take up a position of equilibrium due only to 

 the torsion of the wire which supports it. To adjust the 

 system to this condition, the magnet is first replaced by a 

 non-magnetic needle of the same form and weight, and the 

 period of oscillation of the system determined. The magnet 

 is then replaced. In general the period of oscillation becomes 

 nmch shorter, and the beam takes up a diff'erent position of 

 equilibrium. However, by adjusting the counterweight an 

 exact vertical ity of the magnetic axis of the magnet can soon 

 be obtained, which is indicated by the system again taking 

 the same period of oscillation as when the magnet was 

 replaced by the non-magnetic needle. A magnetic system is 

 then realized which is directed only by the torsion of the wire. 



With systems of this kind we have obtained very great 

 sensibilities. For example, for a magnet-pole of 20 c.G.s., 

 supported by a silver wire 95 cm. long and *025 mm. in 

 diameter, we obtained a deflexion of 34 mm. on a scale 

 2 metres distant for a variation of the field of 10~^ C.G.s. The 



2 G2 



