QUANTITATIVE ESTIMATES OF SENSORY EVENTS 307 



assigned the value unity. The essential point is that whatever apparatus 

 we use in an attempt to measure radiation intensity we in fact measure a 

 magnitude of which radiation intensity is only one factor, the other factor 

 being a property of the particular kind of physical system on which radiation 

 produces the detectable effect utilised in our measuring apparatus. 



In the case of a thermopile or similar instrument this second factor is 

 practically independent of wavelength, and for practical purposes we treat 

 the magnitude as though it were f\ alone. With detectors of the photo- 

 electric or photo-chemical type, the instrumental factor varies considerably 

 with wavelength, butanysuch detector furnishes a practical criterion of equality 

 for a magnitude E\ a\ where ok is its sensitivity factor at wavelength X. 



Now this is exactly what the human eye does in photometric measure- 

 ments involving differences in spectral composition. The part of the eye 

 which interacts with radiation is, of course, a physical instrument which 

 converts some of the radiant energy into other forms, photo-chemical and 

 photo-electric effects being produced which in turn cause stimulation of 

 the optic nerve. We are not at this stage of events concerned with the 

 physics or physiology of the nerve system or with how the physical pheno- 

 mena occurring in the nerves ultimately produce the sensation of light. 

 The peripheral organ is simply a physical detector of radiation, and when 

 used in conjunction with a photometer provides a criterion of equality for 

 a magnitude Ek a\ where a\ has the same kind of significance as in the other 

 cases referred to above. 



Now in the case of the eye CTa is the efficiency of radiation of wavelength X 

 in stimulating the sensation of brightness, so the magnitude E\ c!\ has the 

 dimensions of sensation intensity, as of course it must have since it is 

 equality of sensation intensity which provides our criterion of equality for 

 E\ CTx. If we could measure E\ c!\ we should be measuring sensation 

 intensities ; but while we have a practical criterion of equality for this 

 magnitude we have no operation of addition. All the means employed to 

 vary the intensity of the light reaching the eye from the photometer are 

 applied solely to the beams of radiation and depend ultimately on the 

 operation of addition applicable to E\ alone. There is no operation we 

 can perform which would correspond to the addition of quantities of the 

 product E\ C7\. It is therefore impossible to deduce any quantitative 

 information about unequal intensities of sensation from photometric 

 measurements such as those embodied in the visibility curve. We can 

 predict that certain relative quantities of energy of different wavelengths 

 will produce equal sensation intensities under prescribed conditions, but 

 that is all, and the whole science of heterochromatic photometry is devoted 

 to establishing this kind of equivalence between stimuli of different physical 

 qualities. The variation of relative sensitivity with wavelength as exhibited 

 by the visibility curve is of course an important constituent of ' sensory 

 events ' and can, as we have seen, be quantitatively described by the results 

 of measurements. 



The kind of measurement denoted by the term ' Colorimetry ' involves 

 only an extension of the same principle. 



To discuss colorimetry would carry us beyond the limits of available 

 space. Suffice it to say that in all such measurements the only criterion 

 provided by the eye is one of equality, in this case not only equality in the 

 one attribute of brightness but simultaneous equality in colour and bright- 

 ness. As in ordinary photometry all quantitative operations performed in 

 the measurements are performed on beams of radiation, and do not provide 

 any experimental operation which can be identified with addition of sensa- 

 tions. The results of colorimetric measurements can therefore only be 



