﻿the Measurement of Light. 589 



think it is most highly inconvenient to denote the truly 

 fundamental magnitude by a name (flux density) so highly 

 suggestive of a derived magnitude. And we have preferred 

 4> to C in order to indicate the close connexion between 

 intensity and flux. 



But Ave have more serious differences with the American 

 Committee. On p. 282 a lambert is defined as the bright- 

 ness of a substance emitting one lumen per square centimetre 

 of projected area in the direction considered. But to speak 

 of a lumen emitted in a direction is to talk nonsense ; the 

 flux emitted within an infinitesimal angle is always infini- 

 tesimal. In place of "one lumen" should be substituted 

 " one lumen per steradian." Since on p. 231 brightness is 

 defined as flux per steradian per square centimetre, it is 

 probable that the omission of the words " per steradian " 

 is a mere oversight — the more unfortunate because it is 

 twice repeated. 



But the statement, also on p, 232, that "one candle per 

 square centimetre equals 3*1416 lamberts " is less easily 

 comprehensible. In the first place, lamberts have only been 

 defined for surfaces which obey Lambert's law ; for with 

 other surfaces the lumens per steradian per unit of projected 

 area will vary with the direction considered. Since a surface 

 made up of " candles per square centimetre " does not obey 

 that law, it is meaningless to attribute to it any number of 

 lamberts. To remove this objection two alterations can be 

 made: — (1) For "candles" can be substituted "candle- 

 power " ; for candle-power is often used as a synonym for 

 what we and the American committee call intensity "*. 

 (2) Some specification can be added concerning the direction 

 in which the candle-power or intensity is to be measured. 

 For (2) it is most natural to take the direction normal to the 

 surface ; but if we take this direction, a surface of one square 

 centimetre emitting an intensity of one candle-power {i.e., 

 one lumen per steradian) has a brightness of 1 lambert, not 

 7r lamberts as the American committee affirm. We might 

 also take average intensity over a hemisphere based on the 

 surface. If the surface is plane and obeys Lambert's law 

 the average intensity over this hemisphere is \ candle-power 

 per square centimetre, and a surface emitting 1 mean 

 candle-power per square centimetre would be equal to 



* E. g., average candle-power or horizontal candle-power means 

 average or horizontal intensity. The term " candles " is often here, too, 

 used in place of candle-power. We believe that this confusion between 

 a thing and its properties is one of the main sources of the obscurity of 

 photometric definitions. 



