FORCE OF THE WESTON NORMAL CELL IN SEMI-ABSOLUTE VOLTS. 
149 
order of nearly one half of 1 per cent.; (2) the difficulty of determining the mean radii 
of the coils, which were wound with silk-covered wire ; (3) the want of rigidity of the 
pulley arrangement for equalising the tensions of the suspending wires, and the 
imperfect elasticity of the control, which depended too much on torsion, and made it 
impossible to obtain readings consistent to 1 in 1000 for the deflections or the times 
of oscillation. These defects were so fatal to accurate work even of the order of 1 in 
10,000, which was all that it was originally contemplated, that it was found necessary 
to reconstruct the instrument entirely until nothing remained of the original except 
the frame, and even that required stiffening to a material extent. 
The arduous work of reconstructing the instrument was undertaken by 
Mr. R. O. King as part of his work during the tenure of an 1851 Exhibition Scholar¬ 
ship during the years 1897-8. The coils were re-wound with a double winding to give 
a perfect check on the insulation. The large coils were wound with a carefully 
measured length of hard rolled copper tape, which gave a very high order of accuracy 
in the determination of the mean radius—this device possesses the great advantage of 
avoiding the excessive refinement necessary in measuring the linear dimensions of the 
coils in the majority of other methods. The coils were also made reversible and inter¬ 
changeable to eliminate possible errors ofisymmetry, especially in the measurement of 
the distance between their planes. 
The dimensions of the suspended coils were determined by an electrical method of 
comparison with the fixed coils, which proved to be one of the most difficult measure¬ 
ments owing to extraneous magnetic disturbances. Mr. King’s results for this 
comparison showed extreme differences amounting to nearly 1 in 5000. This led me, 
in revising his work, to put the limit of accuracy at 1 in 10,000, although all the 
fundamental quantities could be measured to 1 in 100,000 or better. I have since 
observed that the largest discrepancy in this comparison was undoubtedly due to a slight 
failure of insulation duly noted by Mr. King at the time. Another series was inter¬ 
fered with by an inopportune magnetic storm. Mr. Shaw employing the identical 
tube made by Mr. King for supporting the small coils in a fixed position relative to 
the large coils, has succeeded after 15 years in recovering Mr. King’s value for the 
ratio of the currents required for balance to 1 in 30,000. I am now of opinion that I 
had greatly under-estimated the probable accuracy of this comparison, which Mr. Shaw 
appears to have carried out to an order of accuracy little if at all inferior to 
1 in 100,000. 
A weak point in many of the methods of absolute measurement is the relative 
smallness of the electrical force to be measured. The arrangement of the coils in 
parallel, rendered possible by rewinding the coils, permitted the attainment of a forty 
times larger force for the same current in the small coils, i.e., without introducing any 
unsteadiness or uncertainty of dimensions due to the heating effect of the current on 
the small coils. The relative importance of the gravity control in the bifilar suspension 
could thus be increased, with a corresponding improvement in the steadiness 
