338 REPORTS ON THE STATE OF SCIENCE, ETC. 



for sensations is private. We cannot establish sensory equilibrium between 

 people as we can establish thermal equilibrium between bodies. Each 

 person would be a sensationmeter only capable of measuring its own sensa- 

 tion, like a thermometer only capable of measuring its own temperature. 

 It could provide no information about anything. 



However, we cannot even have whatever satisfaction there might be in 

 establishing this perfectly useless sensation scale. As we saw when con- 

 sidering temperature, in order that a magnitude may be defined in this way 

 it must be possible to postulate a one-one relation between the magnitude 

 to be defined and the measurable magnitude to which it is to be related by 

 the defining relation ; and that we are not free to make any such postulate 

 unless, from the nature of the case, it is certain on a priori grounds that the 

 defined magnitude can never enter experience except as ' the thing defined 

 by the adopted relation.' 



There is no one-one correspondence between sensation intensity and 

 stimulus intensity. Sensation intensity enters experience directly, in its 

 own right so to speak, and we know that, owing to adaptation, fatigue, or 

 various other causes, the same stimulus may evoke sensations of markedly 

 different intensity on different occasions. So we cannot define a scale of 

 sensation intensity as a postulated function of stimulus intensity even were 

 it of the slightest use to do so. The analogy with temperature advanced 

 by some psychologists is entirely fallacious. 



But what about the average sensation intensity corresponding to a 

 stimulus / ? May not this be treated as a B magnitude ? Over how long 

 a period are we to take this average ? Obviously if the whole of an observer's 

 life is to be included there will be a one-one correspondence between mean 

 5 and /, and we are free to postulate any law we like to define a relation 

 between them. But what have we achieved ? We cannot use this scale 

 to measure the individual 5 corresponding to / on any particular occasion. 

 It gives us no information other than that with which we started, that the 

 mean sensation intensity corresponding to any intensity of stimulus is just 

 whatever we have chosen to say it is. 



We must conclude therefore that sensation intensity is not measurable 

 either as an A magnitude or as a B magnitude. It is not measurable in any 

 sense of the term. 



IV. Notes on Mr. Guild's statement by members of the Committee. 

 A. By Dr. R. H. Thouless. 



(i) This account of what is meant by ' measurement ' is excellently clear. 

 I think ' measurement ' is primarily the physicist's term and I am willing 

 to accept what they say as to what the word means, and I do not think it 

 in any way restricts the possibility of quantitative experiment in psychology 

 if it is agreed that it is not ' measurement of sensation.' 



(2) The account of Fechner seems to make his account of the matter 

 much more clear and rational than it really was. Thus with reference to 

 ' Fechner's second principle ' (p. 309), I cannot find that Fechner formulated 

 any such principle in the Elemente. He assumed it, according to my reading, 

 without realising what he assumed, by stating Weber's Law in the form : 

 dy =K.d^/?>. 



(3) It is obvious that Fechner thought he could establish measurement 

 of sensation in a sense which is indefensible. I am not sure that Mr. Guild 

 disposes of a possible defence that what is really possible is a B measurement 

 of sensation (for a single individual, under specified conditions of stimulation, 



