THE MECHANICAL ACTION OF LIGHT. 267 



pith equally in opposite directions, and the luminous index remains 

 at zero. When, however, I cut one candle off, the candle on the op- 

 posite side exerts its full influence, and the index flies to one end of 

 the scale. I cut the other one off and obscure the first, and the spot 

 of light flies to the other side. I obscure them both, and the index 

 comes quickly to zero. I remove the screens simultaneously, and the 

 index does not move. 



I will retain one candle 12 inches off, and put two candles on the 

 other side 17 inches off. On removing the screens you see the index 

 does not move from zero. Now the square of 12 is 144, and the 

 square of 17 is 289. Twice 144 is 288. The light of these candles, 

 therefore, is as 288 to 289. They therefore balance each other as 

 nearly as possible. Similarly I can balance a gaslight against a can- 

 dle. I have a small gas-burner here, which I place 28 inches off on 

 one side, and you see it balances the candle 12 inches off. These ex- 

 periments show how conveniently and accurately this instrument can 

 be used as a photometer. By balancing a standard candle on one side 

 against any source of light on the other, the value of the latter in 

 terms of a candle is readily shown ; thus in the last experiment the 

 standard candle 12 inches off is balanced by a gas-flame 28 inches 

 off. The lights are, therefore, in the proportion of 12 2 to 28 2 , or as 

 1 to 5.4. The gas-burner is, therefore, equal to about five and a half 

 candles. 



In practical work on photometry it is often required to ascertain 

 the value of gas. Gas is spoken of commercially as of so many can- 

 dle-power. There is a certain " standard " candle which is supposed 

 to be made invariable by act of Parliament. I have worked a great 

 deal with these standard candles, and I find them to be among the 

 most variable things in the world. They never burn with the same 

 luminosity from one hour to the other, and no two candles are alike. 

 I can now, however, easily get over this difficulty. I place a " stand- 

 ard " candle at such a distance from the apparatus that it gives a 

 deflection of 100 on the scale. If it is poorer than the standard, I 

 bring it nearer; if better, I put it farther off. Indeed, any candle 

 may be taken ; and if it be placed at such a distance from the appa- 

 ratus that it will give a uniform deflection, say, of 100 divisions, the 

 standard can be reproduced at any subsequent time ; and the burning 

 of the candle may be tested during the photometric experiments by 

 taking the deflection it causes from time to time, and altering its dis- 

 tance, if needed, to keep the deflection at 100 divisions. The gaslight 

 to be tested is placed at such a distance on the opposite side of the 

 pith-bar that it exactly balances the candle. Then, by squaring the 

 distances, I get the exact proportion between the gas and the candle. 



Before this instrument can be used as a photometer or light-meas- 

 urer, means must be taken to cut off from it all those rays coming 

 from the candle or gas which are not actually luminous. A reference 



