Mechanical Action of Light. 
253 
1876.] 
ceiling. I now turn the torsion handle, and in much less 
time than in the former case the ray is brought back to zero. 
On looking at the counter I find it registers four revolutions, 
and the index points to 188 degrees, making altogether 
360° x 4 + 188 = 1628°, through which the torsion fibre has to 
be twisted to balance the light of the candle. 
It is an easy calculation to convert this into parts of a 
grain weight ; 10,021 torsion degrees representing o*oi grain, 
1628 torsion degrees represent 0*001624 grain. 
10,021° : o*oi grain : : 1628° : 0*001624 grain. 
1 
The radiation of a candle 6 inches off, therefore, weighs or 
presses the two square inches of blackened pith with a 
weight of 0*001624 grain. In my own laboratory, working 
with this torsion balance, I found that a candle 6 inches 
off gave a pressure of 0*001772 grain. The difference 
is only 0*000148 grain, and is fairly within the allowable 
limits of a ledture experiment. But this balance is capable 
of weighing to far greater accuracy than that. You have 
seen that a torsion of 10,021° balanced the hundredth 
of a grain. If I give the fibre 1 degree more twist the 
weight is over-balanced, as shown by the movement of the 
index-ray on the ceiling. Now 1 degree of torsion is about 
the 1-10, oooth part of the whole torsion required by the 
i-iooth grain. It represents therefore the 1-10, oooth part 
of the i-iooth, or the millionth part of a grain. 
Divide a grain weight into a million parts, place one of 
them on the pan of the balance, and the beam will be in- 
stantly depressed ! 
Weighed in this balance the mechanical force of a candle 
12 inches off was found to be 0*000444 grain ; of a candle 
6 inches off, 0*001772 grain. At half the distance the weight 
of radiation should be four times, or 0*001776 grain ; the 
difference between theory and experiment being only four- 
millionths of a grain is a sufficient proof that the indications 
of this instrument, like those of the apparatus previously 
described, follow the law of inverse squares. An exam- 
ination of the differences between the separate obser- 
vations and the mean shows that my estimate of the sensi- 
tiveness of this balance is not excessive, and that in practice 
it will safely indicate the millionth of a grain. 
I have only had one opportunity of getting an observation 
of the weight of sunlight : it was taken on December 13th, 
but the sun was so obscured by thin clouds and haze that it 
was only equal to 10*2 candles 6 inches off. Calculating from 
