circumstances, too, the extremely low intervals perceptible are not surprising. The usual method was to make a decision on the first pair, and then see if the others confirmed it. In the two-second series rhythm was generally noticeable and helpful; the one-second rate was most agreeable and pleasant; the half second very lively; the four-second "deathly slow" and " nervous." It should be mentioned that at the fastest rate the apparatus did not always function perfectly, occasionally skipping a click or flash. Subjective control and introspective analysis of method seemed also quite difficult at this rate, and here S. reported that the clicks and flashes failed to combine, but formed independent series.

To recapitulate briefly, this study has shown:

1. That the flash-click order can be recognized when the interval is shorter than that required for the click-flash order. 2. That this holds true for a series of pairs of stimuli as well as for a single pair.

3. That the serial repetition of the pairs materially reduces the time interval necessary for right judgment.

4. The cause of this seems to be a retardation of the click due to greater attention-claiming quality attaching to the flash.

XIII. THE TIME REQUIRED FOR RECOGNITION.

By F. W. Colegrove.

The method employed in the following rough study was extremely simple. Sixty-eight pictures, three to four inches in length and two to three inches inches in height, were cut from an old magazine and pasted upon cards. These were inserted, one at a time, in the clips of the Cattell Fall-chronometer and exposed by the sudden falling of the screen. At the instant of exposure, the falling screen released one pendulum of an electrical vernier chronoscope, the other being released by the subject as soon as he was able to decide whether he had seen the picture before or not.1 If the picture was recognized, the subject reacted with his right hand; if unrecognized, with his left. Five or six reactions to the letters R (right) and L (left) were taken before and after each sitting, and the discrimination times thus found furnish both a control of the other experiments and a means of finding the pure recognition time free of all peripheral processes.

1 For the mode of operating the vernier chronoscope, see this Journal, Vol. IX, 191-7, Jan., 1898.

In the tables below, however, these simple discrimination times. have not been deducted, but, on the contrary, the full time of response has been retained.

Five subjects were tested, all of whom had had some laboratory experience and two of whom had had a good deal. Five pictures, numbers 1, 2, 6, 44 and 68, were shown each subject before beginning and he became familiar with them. He also saw them again before each sitting. In what follows they are termed the "well-known pictures." On the first day of experimenting these were shown in irregular order with other pictures from the series. On the second day both the "wellknown pictures" and the new ones of the first day could be drawn upon as known pictures to mix with a second group of unknown pictures; and on the third day the pictures of both the first and second days, and so on.

A considerable mass of records was thus obtained, both for the time required for recognizing the "well-known pictures,' and for the time required for other pictures after one, two, three, four or more exhibitions. It is hardly necessary to mention that the first recognition, except in cases of mistaken reactions, occurs on the second exhibition and so on. The results for the earlier and later recognitions of the well-known pictures are given in Table I. In forming this table, the series of recog

TABLE I.

Showing Times for Signaling the Recognition of the Well-known Pictures; Times in 0.001 Seconds.1

The times are given in the usual unit for convenience of the reader, though, as the unit of the chronoscope itself was 0.02, no significance is attached to the third figure of the results. It might be expected that with the method of division described in the text the number of

nitions for each picture was divided in the middle, if the number of recognitions was even, and the first part taken for the column of earlier recognitions, the second for that of later recognitions. If the number was odd and the full series could not be evenly divided, the middle term was discarded and the remaining parts treated as if the series had been even.

These figures show that the full time of signaling the recognition of a well-known picture lies somewhere between 424 to 615, and that it is shorter in the second half, where the familiarity was greater. This quickening may be due in part to increased skill in reacting. Two of the five subjects show a similar gain in reacting to the letters R and L, and with one subject, Q, the difference is more than that between the early and late trials in Table I, making the pure recognition times respectively 68 and 89. But it must be due chiefly to increasing familiarity with the pictures. Four of the five subjects show the same relation in the pure recognition times as in the table. The average pure times, found by subtracting 316 and 311 from 527 and 459 respectively, are 211 and 148.

The same thing is shown, though somewhat irregularly, when the successive recognitions of other pictures are examined, as in Table II.

How this increased speed of recognition should be regarded, whether as a hastening of the recognition process or as a gradual change in the character of that process from one which is more or less conscious toward one which is wholly automatic, or as involving both tendencies, is, unfortunately, not shown by the data at hand.

Beside this general question there are several of a subordinate interest, namely: Is there any difference in quickness of response when a picture is signaled as unrecognized? Is the quickness of response different when errors are made, i. e., when a known picture is signaled as unknown, or vice versa? Is there any difference in the quickness with which different pictures are recognized? Such data as the experiments have furnished upon these points are gathered in the following paragraphs.

In Table III the time for the first recognitions has been taken from Table II, instead of the average time of all recognitions, as corresponding more nearly with the condition present when the pictures (before unknown) are signaled as unrecognized. It will be observed that three subjects (W, S and K) take longer to determine and signal a recognized picture than an unrecognized one; and two, Y and Q, take longer for the unrecogobservations would be the same for the same subject in both early and later recognitions, and such would be the case except for differences introduced by failures in the functioning of the apparatus, and by erro. neous reactions on the part of the subjects.

TABLE III.

Showing Comparative Quickness in Signaling Recognized and Unrecognized Pictures.

*One record, nearly four times as large as that next it in size, was omitted in making this average.

nized. This appears to be due to a difference in mental attitude, which will perhaps be clearer after a consideration of the results where errors were made. Y, Q and K show the same tendencies in Table IV as in Table III; the times of W when in error are practically the same without regard to the nature of the error; while for S the relation of Table III is reversed. The small number of cases and the large M. V. make it seem likely that this difference in the case of S is accidental, and examination of the separate determinations confirms that opinion.