>2 4 



NA TURE 



[December 16, 1922 



Physiological Aspects of Physical Measurement. 1 

 By Sir John Herbert Parsons, C.B.E., F.R.S. 



PHYSICISTS too often forget that the basis of 

 *■ physii al measurements is biological, for the so- 

 called " outer world " only exists for us by virtue of 

 ns it arouses in our bodies Physical 

 ; open to the errors of all human 

 ions, and these vary in degree according to 

 the type of observation. In all cases the ob 

 is the formation of a judgment, based on the sensa- 

 tions derived from the stimulation of a sensory 

 organ. Physiological experiments show that stimula- 

 tion of some sensory organs gives more sharply 

 defined responses than others. Thus, the responses 

 to smell and taste are crude and vague ; tin >se to 

 moderate cutaneous stimuli — touch and temperature 

 — much better denned ; those to auditory stimuli, still 

 be1 ter, and those to visual best of all. 



in among the varieties of a given type of 

 sens ition various degrees of definition are met. Thus 

 pain, though cutaneous, is crude like smell and taste; 

 in vision, form sense is much more accuratelv defined 

 than colour sense. Definition, indeed, varies as the 

 biological differentiation of the sense organ. 



Now, the most highly differentiated sensory organ 

 is the- eye, and the fovea is its most highly differenti- 

 ated part. Experiments show that the greatest dis- 

 crimination is met with in foveal stimuli. The highest 

 degree of sensory discrimination is the appreciation 

 of continuity or lack of exact continuity in two 

 straight lines set end to end, as in the vernier. This 

 m. 1 j be railed /ni, m identity, and it is noteworthy 

 that it has been adopted empirically by physicists in 

 the vernier, balance, and other instruments. Physi- 

 cists have been very ingenious in applying this criterion 

 to otherwise apparently unsuitable measurements, as, 

 for example, the measurement of temperature. But 

 there are many physical measurements to which it 

 cannot be applied, or at any rate has not been applied. 

 Photometry is an example. Here we are measuring 

 the brightness of two lights. By various devices the 

 principle of identity or equality of sensations is made 

 use of — thus utilising the only accurate psvchological 

 comparison — but the quality of the sensation to be 

 adjudicated upon does not admit of the accural y of 

 linear identity. Even in homochromatic photometry 

 we are comparing the brightnesses of two illuminated 

 areas. As is well known, these areas react upon each 

 other physiologically — by the process of induction or 

 simultaneous contrast. Moreover, the judgment is 

 affected by the previous stimulation of the retinal 

 areas concerned (successive contrast). It is further 

 \ it 1,1 led by variations in adaptation. 



Still more open to error are the comparisons of 

 brightness of different coloured lights, heterochro- 

 matic photometry. Here the difference in colour acts 

 as a very disturbing element. Yet by practice it is 

 Main almost as accurate results as in 

 homochromatic photometry. But how can we judge 

 of the accuracy of these determinations ? In this 

 particular instance we can have recourse to the fact 

 that the critical frequency of flicker depends upon 

 brightness and follows a definite mathematical law. 

 The eye is extremely sensitive to flicker, so that the 

 disappearance of flicker affords a very sensitive 

 criterion. It has been found that the results obtained 

 by the flicker photometer confirm the results obtained 

 by the best so-called " equality of brightness " 

 'lions. 



Xo matter how delicate the criterion there are still 

 errors of observation due to imperfections of a bio- 



1. presidential address to the Illuminating Engineering Society, 

 on May 25. 



NO. 2772, VOL. I IO] 



logical nature common to all human observers and 

 also to the so-called " personal equation " of the 

 given observer. How are these to be eliminated ? 

 Recourse is had to mathematical theory. The basis 

 ( if the theory of error, which is a branch of the theory 

 of probability, is that small errors will be more fre- 

 quent than large ones, very large ones will be prac- 

 tically absent, and the mean is the result of the 

 mutual destruction or compensation of many small 

 sources of error acting in opposite directions 



The kinetic theory of gases is built entirely upon 

 this statistical foundation, and its success in explaining 

 the physical properties of gases is strong evidence in 

 favour of the statistical theory. There are several 

 mathematical " averages or means," and much de- 

 pends upon the choice of the suitable " means," 

 winch itself depends upon the frequency distribution 

 of the observations. Graphic methods of eliminating 

 errors are constantly used by physicists. One of the 

 commonest is the method of interpolation, and the 

 smoothing of the curves. 



An interesting example of the opposite aspect of 

 averages is the modern view of atomic weights. 

 These are some of the most accurate physical measure- 

 ments ever made and have been corrected by the 

 best statistical methods. Many of them approximate 

 nearly to whole numbers and there are many theo- 

 retical reasons for believing that they are whole 

 numbers. Recent investigations, chiefly by Aston, 

 have shown that the atomic weights hitherto obtained 

 ue themselves averages : that there are many so- 

 called " isotopes," having almost if not quite identical 

 chemical properties, but differing from each other in 

 the number of their electrons and also in their true 

 atomic weights, which are invariably integers. 



I hope that this philosophical parenthesis suffices 

 to show that even in the matter of physical measure- 

 ments the physiological aspects of the subject must 

 perforce be taken into account. But in dealing with 

 illumination we are dealing not only with foveal 

 vision, but also with peripheral vision and alterations 

 of sensitiveness of the eye under different conditions 

 of stimulation. It is well known that the foveal 

 region of all parts of the field of vision alters least 

 in sensitiveness under different intensities of illumina- 

 tion. It is, therefore, relatively stable, and observa- 

 tions founded on criteria derived from central vision 

 are proportionately trustworthy. It is quite other- 

 is ise with the other parts of the field of vision. Here 

 itiveness of the retina increases enormously 

 with diminution of the intensity of stimulation. This 

 function of retinal adaptation, which is of such 

 tremendous practical importance in the life of the 

 individual and indeed of the species, interferes very 

 seriously with the accuracy of scientific investigations. 

 I 'hysicists have been led astray by ignoring it, as, 

 for example, in the so-called "deviations from 

 Newton's law of colour mixtures " described by 

 Ki mig. 



Physicists, indeed, are so accustomed to deal with 

 measurements of the highest order of accuracy, 

 founded upon what I have called " linear identity " 

 observations, that they succumb to two errors: 

 (1) that of regarding these observations as of the 

 supreme validity of mathematical abstractions ; (2) 

 that of regarding other observations, to which the 

 " linear identity " criterion is inapplicable, as of far 

 greater accuracy than is in fact the case. When the 

 mistakes arising from these errors are too patent to 

 lie ignored, physicists are apt to exhibit an un- 

 warranted impatience with the shifting sands of 



