nati i. \ ae 
. 
202 RICHARD THRELFALL. 
The ballistic galvanometer was a fine instrument specially made 
for this kind of work. It was calibrated by turning over a large 
coil which was splendidly wound and which has been checked in 
many ways. I made use of the values for the vertical force 
obtained about nine months ago by Mr. Farr from a long series 
of experiments in my laboratory, under the best and most careful 
conditions and with the Kew apparatus. In all cases the induc- 
tion thrown on the galvanometer was checked by reversal of the 
galvanometer connections—except when experiment showed that 
nothing was gained by such reversals—the instrument being in 
another part of the laboratory to the magnetic system. In all 
cases large resistances from a box of coils constructed and cali- 
brated by myself from Cambridge standards, were inserted in the 
circuit so as to give the best range for the galvanometer. 
Results of eeperimenting.—The general result was that I got 
rather worse agreement than was noted by Bosanquet—especially 
at low inductions. I therefore set myself to find out the reason 
of this. Jam ashamed to say how long it took me to clear up the 
difficulty. I investigated the following possible causes : 
(i.) Imperfection of galvanometer law. This was got over 
by adjusting resistances till the same deflexion was 
obtained both on turning over the earth inductor and 
on magnetising the iron. 
(ii.) Effects due to residual state of the iron. This was got 
rid of by demagnetising the iron by an alternator and 
slide resistance, and observing magnetisations with 
the current in both directions, and also on reversal. By 
comparing the three sets of deflections I assured myself 
that the discrepancy was not due to any error of this 
kind. <A similar procedure when taking the tractions 
led to a like result. 
(iii.) By using induction coils of different diameters I assured 
myself that I was really measuring the operative induc- 
tions. | 
