ME. W. CEOOKES ON EEPULSION EESULTING FEOM EADIATION. 
373 
214. The first operation was to ascertain the number of degrees of torsion which 
were equal to the of a grain weight. The apparatus being well exhausted *, 
the index ray of light being at zero, and the pointer of the torsion-handle standing at 
0° on the divided circle, the iron weight was picked up by a magnet, and placed exactly 
on the centre of the pith surface. The end of the beam instantly fell down, and I had 
to turn the torsion-handle 27 complete revolutions and 353° in addition, or 10073°, before 
the beam became horizontal and the index ray again stood at zerof. The downward 
pressure of of a grain was therefore equivalent to the force of torsion of the 
glass thread when twisted through 10073°. 
215. My next operation was to find out the degree of delicacy of the balance. The 
law of torsion is that the force with which the glass fibre tends to untwist itself is 
directly proportional to the number of degrees through which it has been twisted J . I 
can therefore find out the smallest amount of weight which the balance will indicate by 
ascertaining what is the smallest angular movement of the torsion-handle which will 
cause an appreciable movement of the index ray of light. 
I found that a movement of three degrees of torsion sent the index ray a considerable 
distance. One degree of torsion gave a very decided movement, whilst half a degree 
displaced the index ray to a sufficient extent to be easily seen. I believe less than half 
a degree could have been detected had the scale been further off ; but not to run any 
risk of over-estimating the delicacy of the balance, I will take one degree of torsion as 
its ordinary working limit. 
To what fraction of a grain will this torsion of one degree be equivalent X This is 
easily calculated. A twist through 10,073° balances the of a grain; a twist of 
10,074° overbalances it. 
10,073° : 0-01 grain : : 10074° : 0-01000099 grain ; 
.-. 10,074° — 10, 073°=0 01000099 grain — 0 - 01 grain, 
.'. 1° —0-00000099 grain. 
The balance will therefore turn to the Too, 000,000 of a grain. 
Divide a grain weight into a million parts, place one of them on the pan of my 
balance, and the beam is instantly depressed ! 
216. The balance was exhausted up to the highest point, the index ray of light was 
projected on to the scale, and the apparatus was kept in darkness, well protected from 
external influences by screens around it. The index ray soon became stationary ; it 
was then brought to zero, and the pointer of the divided circle set to 0°, the counter 
also being at 0. 
A standard candle was then adjusted 10 inches from the centre of the mirror, and 
* This preliminary testing the value of the ^ grain weight can he equally well performed in air. 
+ This is the mean of several concordant experiments. 
t Biot ; Traite de Physique, tom. i. p. 486 ; Eixchte, Phil. Trans, vol. 120. p. 215. 
MDCCCLXXVI. 3 p 
