342 ME. W. CEOOKES ON EEPULSION EESULTING EEOM EADIATION. 
The diagram Plate 36. fig. 7 shows the curve formed by these observations. 
The isolated dots show the experimental observations, and the continuous line gives the 
theoretical curve which ought to have been followed according to the law of inverse 
squares. They are sufficiently concordant to show that this is the law governing the 
movements of the radiometer. The diagram illustrating this was laid before the Royal 
Society on April 22, 1875; I therefore prefer to retain it rather than prepare another 
one with a more sensitive instrument. 
149. I next wished to ascertain if the speed of rotation would increase directly with 
the number of candles, the same distance off, shining on the instrument. 
The same radiometer that was used in the last experiment was placed in the centre 
of a circle, 2 feet diameter, having 24 standard candles arranged symmetrically round 
the circumference. All the candles were lighted at first, and the times of revolution 
taken as they were removed one by one. 
Number of candles 
burning 1 foot off. 
Number of seconds 
required for 
one revolution of 
[Radiometer. 
24 
6’4 mean. 
23 
7 
22 
7-5 „ 
21 
8-5 „ 
20 
9-5 „ 
19 
10 
18 
11 
17 
11-3 „ 
16 
12 
15 
13 
14 
13-3 „ 
13 
14’5 „ 
12 
16 „ 
11 
17 
10 
18-5 „ 
9 
19-7 „ 
8 
21 
7 
23-5 „ 
6 
28 
5 
35-5 „ 
4 
44-5 „ 
3 
59 
2 
92 
1 
180 „ 
The diagram shown in Plate 36. fig. 8 gives these observations, with the theoretical 
curve. Like the last one, this diagram was handed in to the Royal Society on April 
22nd, 1875. 
With a recently made instrument I should have been able to obtain better results. 
A radiometer now before me will revolve once in eight seconds to the light of a candle 
1 foot off, whilst 24 candles make it spin with such velocity as to become almost 
invisible. 
