﻿tlie Measurement of Light. % 587 



which, though not absolutely necessary to photometry, arc 

 useful in its more elaborate developments. It is doubtful 

 whether they should be included in any official statement 

 concerning photometric measurement ; for it is difficult to 

 describe precisely the circumstances in which they are 

 accurately true, and in the last resort measurements should 

 always be checked by the basic methods that have been 

 described so far. But three of them may be noted. The 

 first arises when reflecting or refracting surfaces are placed 

 between the P.S. and the source or (more often) the eye. 

 We have then to note that the line joining the P.S. to the 

 source or eye is to be taken as the optical path between 

 the two, and all statements concerning direction or distance 

 interpreted accordingly. The second arises- when a sphere- 

 photometer is used to measure average flux or average 

 candle-power. The law involved is then that the illumina- 

 tion of the P.S. used in the measurement is determined 

 wholly by the average intensity of the source in the 

 sphere. This proposition is never true universally, and only 

 experiment with each photometer can tell within what limits 

 it is true. The last is Talbot's law, employed in rotating 

 sector methods or when asymmetrical sources are rotated to 

 obtain average intensity. This law is apparently accurately 

 true, and might therefore be included on an equality with 

 the other laws of photometry. 



11. Having stated the facts, we may proceed to conven- 

 tions. We have to define the units of the various magnitudes. 

 Since we have one fundamental magnitude (illumination) 

 and three derived magnitudes (intensity, flux, brightness), 

 we shall need n units arbitrarily assigned to some specified 

 physical systems or substances and 4 — n formal constants for 

 the three numerical laws defining the derived magnitudes 

 (n<4). We also need units of the other fundamental (or quasi- 

 fundamental) magnitudes involved in the laws of derivation, 

 namely, distance, solid angle, and surface. For many reasons 

 it is convenient to make n=l. But it is important to observe 

 that it is not necessary to assign to the arbitrarily selected 

 system unit value of the fundamental magnitude; we can 

 assign to it unit value of any of the connected magnitudes 

 and define the unit value of the fundamental magnitude as if 

 it were measured as quasi-derived. 



This procedure is actually the most convenient ; we assign 

 arbitrarily a value of intensity and not of illumination. We 

 describe a physical system which we call a standard candle. 

 We then assert that the intensity of the light emitted by it 



