QUANTITATIVE ESTIMATES OF SENSORY EVENTS 313 



menal relations observed in this type of experiment or by what right we 

 assumed that it is differences of sensation intensity which determine the 

 grading, we find that there is no justification whatever in either case. 



We have here an example of the mental suggestion produced by the 

 repeated use of certain words in a loose and inaccurate manner. The 

 instructions given to observers invariably beg the question of the nature of 

 the operation to be performed. These instructions naturally vary, but they 

 are certain to contain such phrases as ' bisect the interval,' or ' adjust . . . 

 half-way between,' or ' adjust C to be as much brighter than B as S is brighter 

 than A,' and so on. In all descriptions of the experiment which I have seen, 

 the process is described in terms of this kind which have a definite significance 

 only when applied to magnitudes which we already know how to measure. 

 The observer is in fact instructed to perform a definite quantitative opera- 

 tion, but is fiot told hoiv to do it. He does the only thing he can do ; that is, 

 adjust the stimuli until he detects some unique character in the relation 

 exhibited by each pair. There is nothing in the operations he actually 

 performs (adjustments of lamps or of sound-emitters, etc.) to tell us what 

 kind of relation it is which he arrives at, and there are no a priori grounds for 

 assuming that differences of sensation intensity, or equality, either of such 

 differences or of any other magnitude, enter directly into that relation. 

 Nevertheless the terms of the usual instructions, by their quantitative 

 implications, suggest the idea of equality and difference as if the operation 

 were exactly analogous to marking off a metre stick into a series of equal 

 parts. If we are instiaicted to bisect a metre stick, or to arrange three 

 points A, B and C, so that C is as much higher than B as JB is higher than A, 

 the instructions have a definite meaning in virtue of the phenomenal rela- 

 tions and operations defining length as a measurable magnitude ; and we 

 can obey the instructions by performing operations in accordance with 

 these definitions ; but if we had not already defined practical criteria for 

 equality and addition of lengths there would be no operation corresponding 

 to ' bisection,' or to the establishment of any other quantitative relation, for 

 lengths. This is the position in the present experiments. Though we 

 have a criterion of equality for sensation intensities we have no operation 

 of addition, and until we have there is no meaning in associating any quanti- 

 tative relation, equality or any other, with differences of sensation intensity. 



It should be clear, therefore, that the description of this experiment 

 should be couched in language which makes no assumptions about the nature 

 of the relation which the observer is to establish, and that we should avoid 

 all terms like bisect, e^waZ-appearing, sense-distance or others which suggest, 

 by their association with the properties of measurable magnitudes, an 

 unjustifiable interpretation of the operations performed in the experiment. 



Let us attempt to describe the experiment in terms of the operations 

 performed. 



We set up three ^ similar objects of the type which are perceptible by the 

 sense — vision, hearing or whatever it may be — for which the experiment 

 is to be made. Object is here used in the most general sense to denote 

 anything external to ourselves which is perceptible in virtue of the fact 

 that it is the origin or apparent origin of a stimulus affecting one of our 

 senses. In the case of vision the objects may be lamps, or self-luminous 

 surfaces, or surfaces seen by reflected light. It is necessary to emphasise 

 the fact that the first requisite of the experiment is a set of perceptible 

 objects. If we keep discussing stimuli without reference to their origin we 

 may easily lose sight of the fact that all stimuli reach us from our environ- 



' Three or more, but three is enough to illustrate the principles involved. 



