336 
ME. W. CROOKES ON REPULSION RESULTING- FROM RADIATION. 
off the scale. The diagram fig, 4, Plate 35, shows graphically the above series. 
The isolated dots show the experimental observations, whilst the continuous lines show 
the curves which the observations ought to have taken according to the law of inverse 
squares. The agreement is sufficiently close to prove that the force of radiation varies 
inversely with the square of the distance of the source. The discrepancies, especially at 
the greater distances, are considerable. Some are doubtless due to irregular torsion of 
the silk fibre, to interfering heat which penetrated the screens, to some of the observa- 
tions following too closely those preceding them, but chiefly to the irregular burning of 
the standard candle. Indeed this cause alone is sufficient to account for all the discre- 
pancies between theory and experiment ; and it would appear to be very active in some 
cases, as a reference to the diagram will show. The observations with the glass plate in 
front followed immediately after the corresponding observation with the naked flame, so 
that if the candle were burning irregularly in one instance it would probably be burning 
irregularly in the other. The diagram shows the observations to agree pretty well with 
theory from 6 feet off to 9 -5 feet. At 10 feet the action of the naked flame is less than 
it ought to be, and this is repeated when the sheet of glass is interposed. At 14 feet 
off the naked-flame observation shows more action than is required by theory, and the 
observation behind the glass plate is also in excess. The dots at 16, 17, 18’5, 22, 24, 
and 25 feet follow the same rule ; where one is in excess or deficient the corresponding 
one repeats the error. At 26 feet off the candle must have been burning with extra 
brilliancy, for the action on the pith is as strong as it was when the candle was only 
23 feet off; and almost identically the same thing is noticed when the plate of glass is 
interposed and the corresponding observation taken ; at 33 and 34 feet off the same dis- 
crepancies occur. Altogether I consider that the comparison of these curves shows that 
unequal burning of the candle must be credited with most of the discrepancies. 
137. This apparatus was now placed so that I could put a candle or other source of 
light on each side of it ; and its sensitiveness was greatly diminished by lowering the 
controlling magnet. A thin glass screen was placed on each side of the black velvet 
box containing the bar-apparatus. The scale, divided into millimetres, was placed 
5 feet 6 inches from the bar. No screen was put in front of the white half of the pith 
bar, the movement, under the influence of radiation, being a differential one, due to the 
superior sensitiveness of the black over the white surface : — 
1 candle, 48 inches from bar, deflected the luminous index 
2 candles, 48 „ „ „ „ 
2 „ 72 
1 candle, 72 
1 „ 36 
3 candles, 72 
90 millims. 
177 „ 
98 „ 
50 „ 
180 „ 
160 „ 
These results are sufficiently close to theory for the differences to be accounted for 
by variations in the light of the candle used. 
