28o REPORTS ON THE STATE OF SCIENCE, ETC. 



as the necessary conditions on which depended the measurability of 

 sensation, so that this measurability stood or fell with the constancy of the 

 ' Weber-Fechner fraction.' 



This approach to the relation between stimulus and sensation has now 

 been almost wholly superseded by other methods, though interest in the 

 constancy of the difference threshold has not entirely died down. It is, 

 of course, difficult to date this change of attitude with any degree of 

 accuracy, but a useful turning-point may be arbitrarily fixed by the 

 publication in 1920 of a review by Marx (35) of work on die Unterschieds- 

 schwelle bei Schallempfindungen. Before this date work on the stimulus- 

 sensation relation for sound was done exclusively by means of what we 

 may term the ' Weber-Fechner approach,' the value of the just noticeable 

 difference was usually found to be fairly constant, and the sounds studied 

 were for the most part unpitched sounds (usually described as ' noises '). 

 After 1920 developments in electrical apparatus made work with tones 

 progressively easier and more accurate, the difference threshold was found 

 to be much less constant than had previously been supposed, and new 

 methods of investigation were evolved. 



These new methods may be classified under two main heads : I. Attempts 

 to assign numerical values to actual sensations, or to determine values of a 

 stimulus whose subjective effect should bear a given numerical relation to 

 the subjective effect of another value of the same stimulus ; II. Attempts 

 to discover the relationship between measurable physiological events and 

 the physical values of the stimulus. When the former can be shown to 

 bear the same or a similar functional relationship to the stimulus as the 

 subjective effects noted above, the hypothesis is commonly advanced that 

 loudness is determined by some such variable as the rate of change of the 

 physiological process or the number of nerve units activated. These are 

 not discussed in the present summary, since their relevance to the measura- 

 bility of sensation depends entirely on whether one is prepared to accept as 

 valid the hypothesis mentioned. 



In addition, the period under discussion has seen the development of 

 noise-analysis methods, by which a complex sound can be reduced to a 

 ' frequency spectrum,' and a better understanding obtained of the apparent 

 anomalies of masking. 



Table I contains a detailed summary of the principal work on Weber's 

 law in its application to the intensity of sound ; the following descriptive 

 notes may be of interest as giving a better indication of the actual experi- 

 mental procedures employed. They also contain mention of a few researches 

 not included in the table. 



The material may be divided into five groups as follows : (i) the earliest 

 work, 1856-79 ; (ii) work reported in a series of articles in Wundt's 

 Philosophische Studien, 1883-1900, and performed for the most part in 

 Wundt's Leipzig laboratory ; (iii) early work on tones, 1 888-1 905 ; (iv) 

 later work on tones, using electrical apparatus, 1922-35 ; (v) miscellaneous 

 work with a variety of instruments, and usually with some other end in 

 view than a simple examination of the truth of Weber's law, 1930-37. The 

 succeeding sections are numbered in accordance with this classification. 



(i) The experiments of the first group gave inconclusive results, and are 

 of interest chiefly for historical reasons. In particular, the work of Renz 

 and Wolf (44) is significant in that it is ' pre-Fechner,' and seems to have 

 anticipated the standardisation of the psychological methods. Renz and 

 Wolf, medical undergraduates at Tubingen, experimented with a watch, 

 intensity being measured in terms of distance. Various devices used to 

 eliminate accidental errors showed a nice balance between the requirements 



