So far we have spoken of the magnitude of the differential sensitivity. This may he ex pressed in two ways, absolutely and relatively. If. e.g. I can just distinguish a light of 50-candle power from one of 50.5-candle power. the value .5 gives a measure of the absolute, and the value .5/50 or 1/100 a measure of the relative differ ential sensitivity. But we can also measure the delicacy of the discriminative function. An ob server might. in one test, distinguish between lights of 50 and 50.2. while in the next succeed ing he .could distinguish only between 50 and 50.S. Here the magnitude of the differential sensitivity, as averaged from the two tests, is .5; hut the observer is evidently less delicate of discrimination than another who should give the values 50.4 and 50.G in successive trials. Not only the average. but the fluctuation of that average, must therefore be considered in a gen eral measure of discrimination.
(2) 'We turn to the question of the indirect differential sensitivity. And we must note. in the first place. that the importance of language as an instrument of psychology can hardly lie over estimated. The flexibility and ready variation of the spoken sentence and the permanence of the written record are very great advantages.
Still, Language is discrete. while the stream of consciousness is continuous: so that. at the best, we are seeking. so to say, to reproduce a fresco by a mosaie. .Moreover. the relation of the con tents of consciousness to the words that describe them is not constant : while some experiences (-all up habitual and familiar phrases, others baffle us, cause us to pick and choose our terms, and to halt for the right expressions. We can not, then. take it for granted that discrimination and the report of discrimination always run parallel, that the `good observer' is necessarily a 'good reporter.' The experimental psycholo gist must receive a special training in the use of words: lie must be taught to clothe his intro spection in language, a. lie is also taught to introspect : and the intercomparison of results w Inch the experimental method secures must ex tend Well a, to content,. recliner. 1.I. im nM di r l'sychophysik I Leipzig, 1,ss9) ; t;ramhiiye. (// r rhysiedoyischen Thychotoya•, (Leipzig, 1893) ; id., Oat/bus of i'syetiotoyy, trans. by Judd (London. 1898) ; thzt/ia.s of Psycho/09y 1S95).