differs in any essential respect from that in the earlier treatises. Some parts of the older exposition are omitted and others condensed, to adapt them to an introductory work, while yet others are given a fuller and clearer statement (e. g., the doctrine of the feeling of activity, p. 189); but the theory itself stands as it did before.

CHAPTER III.

EXPERIMENTS.

SECTION 1. THE PROBLEM AND THE METHOD.

The purpose of our investigation was to obtain definite and, if possible, quantitative results as regards one feature of the phenomena of association and apperception-to determine the amount of change which might be made in an object ordinarily perceived or assimilated in a certain way without change in the character of the resultant perception or assimilation. The original impetus to it was given by the observation of some peculiar mistakes occurring in the experiences of every-day life. One of the first instances, and one which was typical of several that fell into the same class, consisted in the interpretation of the letters MA I L, roughly printed by hand, as a number. I was trying to find by its number (1423) a house which I had visited once before, but which had been so far forgotten that I could no longer be sure of it from general appearance alone. I noticed a house similar in aspect and situation to that sought, and went up to it. As I approached, I read the number 1427 on the mail box. I studied this a moment, and was so fully satisfied that I had been deceived by the general appearance that I turned and went back from the door. As I reached the street, I looked again at the house, which seemed strangely familiar, and then noticed the proper number, 1423, over the entrance. On looking again, I saw that the mail box was marked MAIL in rough letters.

Other mistakes of the same kind were made in different fields. I had myself, as I suppose most have, several times mistaken one man for another whom I strongly expected to see. Instances of striking errors in determining the nature of odors, while the subject was under the influence of strong suggestion, had also come under my notice. Many cases of similar nature are recorded in the literature. Wundt, e. g., gives an example of the kind in his reading of the word TUCHHALLE, printed in gilt letters on a black ground. He could read it either correctly or as a mere jumble of letters, according as he expected the lettering to be gilt on black or black on gilt.1

'Phil. Studien, VIII, p. 337.

It is evident that in all these instances there are two influ

ences at work. A group of immediate, peripherally excited sensations is supplemented or changed by certain other, centrally excited sensations. In Herbart's language, the apperceiving mass works a certain alteration in the apperceived elements. In Wundt's terminology, the new elements call up and simultaneously associate with certain old elements to form a resultant new idea. [On either description the facts are the same: that, under certain conditions, a stimulus gives rise to an idea in some respects different from that which it would, of itself, usually and normally arouse. Our first aim, therefore, was to find a means of measuring the amount of change producible by the central supplementing. This must be a function of two factors: the intensity of the external stimulus which is at work, and the disposition of the mind at the moment, the mental "trend" at the time of its presentation. The stimulus intensity can be approximately measured, and the change noted which the corresponding sensation undergoes with change of that tensity. The mental disposition, on the other hand, lies far remote from direct measurement, but can, under certain conditions, be artificially varied. We can then measure the increase or decrease in the amount and kind of alteration which the sensation undergoes with each variation.

Numerous preliminary experiments, extending over half a year, were made in order to decide upon a form of stimulus which would be easily supplemented,' and at the same time be capable of regular and measurable change. We first tried pictures, colors and geometrical outlines; but none of these gave satisfactory results. Pictures afford no adequate means of measuring the amount of the departure from the usual combinations; they can hardly be broken up into separate units. Moreover, the perception of a picture is normally, in a greater or less degree, a case of associative completion; so that it would be very difficult to ascertain the value of the difference between the amount of completion necessary when the part completed represented the object well or ill. Colors, on the other hand, are so simple that they cannot readily be completed;' unless form is included, they carry with them very few natural suggestions of objects which they might represent. Geometrical forms

themselves are open to both the above objections.

We finally settled upon the use of typewritten words. Although it is not entirely satisfactory to regard the letter as the unit from which the word is constructed, that assumption was, on the whole, the most satisfactory that could be thought of or experimentally obtained. Besides the advan

tage that a word may be roughly regarded as made up of letters as units, it carries with it as a whole a certain suggestion of what its parts shall be. The value of a letter may be expected to vary with its position in the word, with its form (long or short, round or square, usual or unusual), and with the smaller wholes (long between two shorts, etc.) of which it forms a part. All these variations, however, lend themselves to regulation, and in some degree to measurement.

At this point our problem divides, then, into two distinct problems. The first is to determine the variation in the value of the unit of stimulus under changing circumstances, i. e., to determine the value of the sensational factors which are concerned in reading. The other, related to the former as end to means, is to test the conditions under which completion takes place; to learn the ways in which these conditions acted, and to discover, so far as possible, the relative value of each factor in the total experience.

I. The first part of the problem falls again into several minor problems. (1) The first which suggested itself was the determination of the value of each letter of the alphabet in calling up a word, and the opposition which its absence would offer to the proper completion of the whole. Some letters would have, presumably, more or stronger associations than others, and would consequently be more likely to reproduce the proper word, or to prevent its reproduction if they were misplaced or absent. Although our investigation extended over two full years, its results are too few to decide this point. They make it probable, however, that reproduction is very nearly a function of the ease with which we distinguish the various letters. The latter question has been carefully worked out by Professor Sanford.1 The second part of Professor Sanford's problem, closely connected with the first,-the determination of the letters which are most easily confused with one another, could also be solved in time by our method; but it was found more convenient for our immediate purpose to eliminate the errors resulting from these and similar factors by using the same words wherever the results of two series given under different conditions were compared.

(2) A second problem was to discover the value of the different positions of the letters in the word. It is evident at a glance that there will probably be some positions in a word where a change of the letters will be of more influence in preventing its correct completion than a change of the same letters in other positions. These have been worked out, and

1AM. JOUR. PSY., I, p. 403.

the relative values of the more important positions determined.

(3) Thirdly, our results throw light on the relative values of the different kinds of changes to which the letter may be subjected. We made use of three sorts of changes in the letters. We omitted the letter, substituted another for it, or printed an "x" over it to give a shapeless blur in place of the original. Our problem here was to discover the relative effect produced by omitting, substituting and blurring a letter. The last two of these methods were comparatively pure, but the first is complicated by the fact that omission not only removes one of the units, but also shortens the whole word, and brings other differently shaped units into juxtaposition. (4) In a more general way we gathered information as regards the value of the length of the word in determining the character of its completion. This came mainly from scattered observations in the detailed records. At the same time, though we thus have indications of the importance of length, we cannot offer definite numerical results on the subject.

(5) It would have been interesting, further, to note the importance of the relative positions of the mutilations, where more than one were made, in determing the completion of the word which was shown. But as no more than two were used by us, except in eight instances, this problem also went beyond the scope of our work.

All of the points quoted may be decided by comparing the percentage of right and wron cases, . e., of cases in which the word is not completed with the cases in which it is completed. The percentages under two different conditions will give an approximate means comparing the effects of those conditions.

II. Under the second general problem-the determination of the strength of the suggestion which causes us to overlook mistakes in the word-we again have several partial problems. (1) The first factor to measure is the effect of the word as a whole in suggesting what its parts shall be. We can here count the number of right and wrong cases which, under normal circumstances, appear with a given amount of change. Of course, we must take the length of the word into consideration, and ascertain how the strength of suggestion is related to it. In a first series of experiments, therefore, we must leave the other subjective or centrally excited conditions of perception entirely to chance, and make no attempt to learn the conscious disposition or trend,-only occasionally noting what the nature of, these factors has been from the subject's report of his introspection. (2) In a second series

we varied, and in a measure regulated, the more remote factors by calling a word associated with that to be shown immediately before this was itself given. The procedure introduced a new factor, which tended in some degree to create an artificial mental disposition, and the effect of this could be measured in the change produced in the number of words completed. We owe this method of varying disposition to Professor Münsterberg.1 (3) We obtained some light on other points by classifying the effects of certain other factors which were brought out unexpectedly in the course of the experiments. The results are naturally much scattered, but are nevertheless, in certain cases, of even greater importance than the main line of investigation in making clear the efficiency of the different kinds of suggestion, and the nature of suggestion in general.

Our apparatus was similar to that employed, for another purpose, by Dr. E. W. Scripture. 2 It consisted in general of a lantern so arranged as to project an image upon a ground glass screen, with a delicate photographic shutter fixed in front of the lens to control the length of exposure. All the pieces were set up on a long, black table in a dark room.

In more detail: The lantern was of the ordinary type, provided with two double convex lenses. This was enclosed in a wooden case, with the necessary doors for adjusting the lamp and inserting slides, and an opening in front through which the cylindrical lens holder projected and which it fitted tightly. Immediately before the lens tube was placed the shutter, a Bausch & Lomb pneumatic photographic shutter. This was fitted into the front of a wooden box, which in turn was placed over the lens tube of the lantern, and supported by a standard. When all was ready for the experiment, and the doors closed, no light at all escaped from the lantern. A ground glass screen was securely screwed to the table 1.25 metres in front of the forward lens, and the subject was seated 2.50 metres in front of the screen, in a smaller dark chamber within the dark room. The words used were typewritten with extreme care, then photographed and printed on lantern slides. When thrown upon the screen the small letters had a height of approximately 2.3 centimetres. Constant illumination was obtained, during the course of the investigation, by keeping the height of flame of oil lamp constant. Later experiments have shown that the arc Jamp could be used to advantage in place of the oil lamp, but as the investigation was already under way before the arc light was in working order, the latter was not used. The method gave a

1

3

Beiträge, IV, pp. 20 ff. These experiments, undertaken to prove the similarity between centrally and peripherally excited sensations, were suggested in turn by some observations and experiments by R. Avenarius (Kritik der reinen Erfahrung, II, pp. 472, f.) in connection with the effects of disposition upon perception. 2 Phil. Stud., VII.

3

Typewritten, in order to secure greater equality of size for letters of different complexity than is afforded by print.