d44 Messrs. Simmance and Abady on the 
longer the same. It is quite clear from this (and also we 
may gather the cause from Fechner’s law and Von Helm- 
holtz) that the retinal sensation of brightness does not 
increase according to the same law for all colours. It is 
obvious, therefore, that no unaided eye can judge accurately 
in photometrical comparisons so long as the disturbing effect 
of colour is present, unless the distances between the re- 
eciving surfaces and the lights are arranged according to 
some rule which must take the place of that of inverse 
squares. ‘'o put this in another way, it follows that even 
though an expert photometrist can with a Bunsen or Lummer- 
Brodhun apparatus arrive at what he is certain is (and what 
is no doubt) a fair balance of the two tints he is comparing, 
yet grave doubts must exist as to whether the relative candle- 
powers of the two sources can be taken as being indicated by 
this balance. The Purkinje effect is purely a colour effect 
and colour does not enter into consideration in this “ Flicker ” 
photometer. 
No one would dream of considering a colour-blind person 
to be affected by the Purkinje law, and yet, as has been 
shown, such a person can read accurately with this photo- 
meter. The experiments made to prove the Purkinje pheno- 
menon have all been carried out with coloured surfaces, 
never, so far as is generally known, with a scintillation pho- 
tometer, but conclusive trials have shown that with this 
instrument as now shown, no alterations in distance disturb 
the true readings. A red light compared with a green at 
3 metres gives the same comparative intensities to each as at 
2 metres, and an arc-light tested against a 10-candle pentane 
lamp at 40 feet indicates the same candle-power as when the 
distances are reduced to 15 feet. Thus the question of colour 
does not interfere with the results. 
The accuracy of the photometer for coloured lights is 
confirmed by the following experiments. 
Two standard lamps of exactly the same power were used, 
one at each end of the photometer-bar. A coloured glass 
screen was interposed, to intercept the rays from one light, 
which was then measured against the unscreened light. This 
gave value for one coloured screen. A second colour was 
then substituted for the first and its value measured against 
the same unscreened light. Thus the following values were 
obtained :— 
(a) Unscreened lights. 
(b) Candle-power of one light screened with one colour. 
(c) Candle-power of one light screened with the other 
colour. 
