Fig. III. In this figure the screen of the upper drum is arranged to show but a single syllable, an arrangement used in Experiments B and C below. In Experiment A it was open full width horizontally and showed three syllables at a time. The treffer drum is seen below behind the black cardboard screen.

For observer S the a series has a slight advantage over the b series, but the P. E. is so large that its advantage is quite uncertain. The h series is inferior to the g series. Taking account of subjective conditions we find that the a series was on the average learned under more favorable circum

The average of these results gives the modified average. Stated in tabular form the results are as follows: The figures in parentheses stand for the averages reckoned from perfect syllables only, and those standing free, for the modified averages.

.19

±.15

H

stances than the b series, and the g series than the h series; i. e., there were more "mnemonics," the syllables were easier, or attentive conditions were better. For E the a series is poorer on the average than the b series, but the h series is better than the g series, if only perfect syllables are considered; if the modified averages are compared the two are equal. Here subjective conditions are slightly in favor of the g series. We may therefore infer from the results of both observers that the mere fact of having been shown as a side series does not favor that series when that series itself is to be learned at least not to a degree sufficient to be detected by this method of experimentation.

If we arrange the averages of each series according to the position it occupied in the day's programme we have the following table:

The figures on the first line, reading across, denote the place in the day's work; the first column gives the name of the series. The partial successes are included in the figures used for this table. These figures show that when series of a certain denomination come early, they almost without exception show better results than when they occupy a later place in the experiment, and there is also a tendency to a general decrease from the first to fifth place as the X series and the averages show. We find a slight exception for S, where there is a rise for the third series of the set after a low score for the second series. Subjective conditions probably account for the increase here. What we had therefore in Table I is probably only the result of this general tendency.

Experiment B was next undertaken. The problem was to find whether a side series actually read a single time with full attention would be learned more readily for that fact, if between the reading and the learning, another series were learned. The apparatus and conditions of the experiment were those for experiment A with the following changes: Two pieces of black cardboard were made to fit in a groove under the slit in the screen. One was a straight piece which was just long enough to cover the middle and one side syllable, letting the other side syllable show. This could be slipped to the right or the left, exposing the syllables of whichever side was desired. A second piece of cardboard had a square hole cut in the middle, so that when it was slipped into the groove, only the middle one

of the three series appeared. One or the other of these pieces of cardboard was kept before the slit all the time, so that only one syllable ever appeared to the subject at a time, and that a syllable of the series being read or "learned." The cardboards were manipulated by the observer. The procedure was as follows: One series was read a single time by the observer, the shutter was closed by the experimenter and the cardboard changed or shifted so that the series to be "learned" would show; the experimenter lowered the shutter at the end of the one intervening revolution, and the observer then began the reading of the series to be "learned." As before, ten minutes intervened for rest and introspections between the beginning of successive series. The programme in outline is as follows,— Read a once, and wait 1 revolution of the drum "Learn" b, i. e., read b 12 times

Read i beginning 10 min. after learning h; wait I turn of the drum "Learn" j, i. e., read 12 times

a

beginning 10 min. after beginning j. The a series was always written at the left of one strip, the b series in the middle, the g series in the middle and h on the right of a second strip, the i and j series on the left and right, respectively, of a third strip of paper. No extra series were used. The g and h series, which were learned without having been previously read were used as a check on the results of the other four series.

The experiment extended from April 14 through April 24, omitting the intervening Sunday, making ten days of experimental work. Twelve repetitions were used for each series except for S, for whom the number of repetitions was reduced to ten on the seventh day, because he was frequently getting more than half the syllables right. The results are given in the following table:

TABLE III

Obs. S 4.7
.36 4.6 .19 4.5
4.7 .36 4.3
.27 4.1

Series b (P.E.) a (P.E.) g (P.E.) h (P.E.) j (P.E.) i (P.E.)

E 3.7 .15 3.2 .24 3.3
3.5 .23 3.0 .23 3.0 .17 3.1 .21 2.8

.30 3.4 .17 3.2

.28 3.4 .32 .28 3.3 .37

The second line of averages, reading across, are those for perfect syllables only.

If a single reading of the series before learning a second series has helped the first series when it was learned ten minutes later, the a series will be better than the b and the i than the j. But considering the large P. E. neither series for E is helped by its reading, nor the a series for S. But for the latter the i series shows a marked superiority to the j series. This is in part explained by the introspective accounts which show that this series was favored by slightly better conditions, as ease of syllables and attention paid while learning the series. The superiority of the g over the h series is also to be explained in the same way.