VALUE OF THE BRITISH ASSOCIATION UNIT OF RESISTANCE. 
257 
error to have any value, but the distribution of errors round the mean is satisfactory, 
and the mean percentage error is very small, if we consider the complicated nature of 
the observations and the variation in the important conditions. 
We turn now to the measurement of the resistances of the coil used in Part IT. and 
their comparison with the B.A. standards. 
It will be remembered that in the comparison between the coil Flat of the B.A. 
units ’and the 1 ohm of the box used in Part I., the correction to be made for the 
copper pieces connecting the box to the wire bridge and the resistance of the plugs 
was thought to introduce some error, which, however, it was shown would not affect 
seriously the value of the B.A. unit. To reduce this error the 1-unit coil of the 
box was not compared directly with the B.A. standard. Lord RlAYleigh had had 
wound two 5-unit coils and one 10-unit, which had been carefully compared by him 
with the B.A. units. The comparison was repeated by one of us (R. T. G.), and the 
differences between the two results were found to be so small that we could use either 
value with all the accuracy required. 
Thus the values at L2,° were 
5 units 
10 units 
f 4-99392 
L 4*99376 
f 9-98360 
L 9-98393 
Lord Rayleigh. 
R. T. G. 
Lord Rayleigh. 
R. T. G. 
The second 5-unit coil was only measured by Lord Rayleigh and Professor 
Fleming ; it belonged to the latter, and had been taken away by him before our 
comparison was made. 
A third 5-unit, denoted afterwards by 5', however, had been constructed for the 
laboratory to replace it, and its value was found to be 5’00890 at 12° C., while 
Professor Fleming’s coil had a resistance of 5-02444. Lord Rayleigh’s value of the 
10 units was found by comparison with the 5 units + Fleming’s 5 units in series ; 
our value was obtained by comparison with 5 -\-5'. The close agreement between the 
two results is sufficient test of the accuracy of the comparisons. 
In determining the values of the resistances of the boxes, we started from these 
5 and 10-unit coils. Two boxes were used—one by Messrs. Elliott Brothers, No. 229, 
the other by Warden and Mltirhead, No. 202. The 10-unit coils in each of these 
boxes were compared with our 10-unit standard, using the modification of Carey- 
Foster’s method already employed to compare R and the resistance of the secondary 
circuit. Then 20 units in the box, made up in three different ways (viz. : by taking- 
out (a) plug 20 ; (b) plugs 10 and 10'; (c) plugs 1, 2, 2, 5, and 10), was compared 
with the 10-unit and two 5-unit standards in series. 
A large number of determinations were made both by R. T. G. and E. B. S. on 
different occasions. The various values obtained for the Elliott box, reduced to a 
MDCCCLX XXIII. 2 L 
