MU. C. V. BOYS OY THU UADIO-MICRO]\lKTER. 
175 
As has already been shown, this is not the best arrangement to make use of. Tliat 
is best in which the circiht is the shortest which will remain dead beat in the 
strongest magnetic field available. 
In the original instrument, of which figs. 2 and 3 show the pole pieces, a strong 
compound horseshoe magnet was used. The v/orking field was tested by replacing 
the active circuit by one composed of 50 turns of 1 sq. cm. each of the finest 
insulated coj^per wire. This was mounted so that it could be suddenly twisted 
through a definite angle by moving an arm between a pair of stops. The ends of the 
coil were connected with a ballistic galvanometer in a distant room, and the throw 
observed. The resistance of the whole circuit was measured. Then, in the place 
of the coil, a condenser of known capacity and a cell of known E.M.F. were 
arranged with a key so that a definite discharge of electricity could be sent 
through the galvanometer. The absolute value of that sent by each oscillation of 
the coil could thus be determined, and hence the value of the field found. In 
this way it was found that without the keeper a field of 1342 units existed in the 
working space. 
It appears, then, from the table on the previous page, that with such a field the 
circuit should be about 3 mm. long. I did not actually alter the shape of the circuit, 
but adjusted the field by means of a sliding armature, until the sensibility, or the 
resistance to the motion, produced a convenient result. As has been shown, this is 
not so good a plan as adapting the circuit to the strongest field that is available, 
though less is lost by reducing the field than might be expected, as will be explained 
later. 
If the breadth h that will give the greatest efficacy with respect to moment of 
inertia is required, there is no difficulty in finding the best a and I in terms of h, but 
the best h is involved in an expression of such complexity that it can only be found by 
arithmetical means. 
It is not worth while to give at length the table showing the successive values of 
the efficacy, as h varies ; it is sufficient to state that not only is the efficacy diminished 
by increasing the breadth beyond that of the active bar, but it is even increased as 
the breadth diminishes down to 2 mm., and probably far beyond. 
The conclusion, then, is obvious, that not only should the rectangle be as narrow as 
possible, but the junction should be arranged also in a correspondingly narrow form. 
Further, in consequence of the extreme narrowness of the circuit, the resistance and 
moment of inertia of the cross wire may practically be neglected in comparison with 
the now much greater length of the rectangle. I must here remark that I have thus 
been brought to the adoption of the excessively narrow form which M. d’Aesonval 
has used. I do not know whether he was aware that the narrow form is not only far 
quicker than a wide form, which is the reason he gives for adopting it, but that it is 
m addition, when a convenient period is arranged, also more sensitive. I certainly did 
not expect to find it so. 
