TRANSACTIONS OF SECTION A. 435 
tangential stress on the plane of incidence Ksin 6cos@=4 Esin20. If » of the 
incident beam is reflected, the normal pressure is (1+)Ecos?0@ and the 
tangential stress is et TE sin 26.1. When there is absorption, the tangential 
stress has a maximum value at 45° if » is constant. When there is no absorption 
the tangential stress disappears. 
The tangential stress is much more easily detected than the normal pressure, 
for the action of the gas surrounding the surface is normal to it, and is with 
difficulty disentangled from the normal light pressure. But the gas action is at 
right angles to the tangential stress, and it is merely necessary to arrange a 
surface, free to move in its own plane, to eliminate the action of the normal forces 
and to reveal the tangential stress. 
With the assistance of my colleague, Dr. Guy Barlow, to whom I am much 
indebted for help in the work, I have made the following experiment to show 
the existence of the stress. 
Two circular glass discs, each 2°75 sq. cm. area, were fixed at the ends of a 
horizontal light glass rod 5 cm. long, the discs being perpendicular to the rod and 
fixed to it at their highest points. One of the discs was lampblacked and the 
other silvered. The rod was placed in a light wire cradle and suspended by a 
fine quartz fibre about 25 cm. long in a brass case with glazed sides. On the 
cradle was a mirror, by which deflections could be observed with a telescope on a 
millimetre scale 1‘8 metres.distant. The moment of inertia of the system was 
2°35 gm. cm’, and the time of vibration was 146 seconds. A deflection of 
1 scale division therefore corresponded to a tangential force on a disc of about 
one two-millionth of a degree-—more exactly ‘483 x 10-5, 
The air was pumped from the case till the pressure was less than 1 cm, of 
mercury. At this pressure the irregularity of the disturbances, due to the 
residual gas, is very greatly reduced. A parallel beam of light from a Nernst 
lamp was then directed so as to be incident obliquely on the lampblacked disc. 
From the arrangement of the discs it is obvious that a uniformly distributed 
normal force would have no moment tending to twist the system, while a 
tangential force would have a moment and would twistit. In all cases the disc 
moved away from the source of light. The deflection was a maximum when the 
incidence was not very far from 45° and fell off on each side of the maximum 
value. As there are various sources of error not yet removed, we have not made 
a complete series of measurements, but have only made sure that the effect is of 
the order to be expected from the theory, by finding the deflection for an 
angle of 45°. 
The beam from the Nernst lamp, when incident at 45°, turned the rod through 
16°5 scale divisions. Assuming total absorption, the tangential force should be 
$ Esin 26 x area of disc = 3 E x 2°75, 
Equating to the value of the force given by the deflection, viz. 0-483 x 10-5 
x 16°35, we have H=5'8 x 10-6. 
The same beam was then directed on to a small lampblacked silver disc of 
known heat capacity, through a glass plate of thickness equal to that of the side 
of the case. ‘The initial rise of temperature per second was measured by a 
thermo-junction of constantan wire soldered to the disc. The energy density of 
_ the stream was thus found to be EH =6'5 x 10-°, + 
The agreement of the two values is quite as close as could be expected in so 
rough a determination. 
When the beam was directed on to the silver dise at the other end of the 
torsion rod the deflection was much less, as was to be expected. 
We have also made some qualitative experiments with a blackened glass 
cylinder—a ring cut from a test tube—suspended by a quartz fibre with its axes 
vertical. When a beam fell on this in any direction, not along a diameter, there 
was always a twist. in the direction corresponding to the tangential stress, 
1 These expressions are given in ‘ Radiation in the Solar System,’ Phil. Zrans. A 
vol. 202, p. 539. : : 
Lf 
Los 
nw 
