the series, counted. A crab going in and coming out again was counted as in, but in every case by far the greater number that had entered remained behind the screen as long as the food was there. Two stimuli to two different senses, taste and vision were thus simultaneously and contiguously presented; and if this led to an association between the two "constructs," "food" and "screen-darkness," it would necessitate the overcoming of the natural positive heliotropism manifested at the start, and by contrast on this background the association would stand out more prominently. After the crabs were fed each

1 This long time was granted in order, if possible, that the crabs might ultimately find their way in as a result, perhaps, of merely wandering around in their evident endeavors to localize the source of the taste stimulus, which in every case caused considerable agitation. The varying length of time used throughout the series was a matter of best adapting the means to the end. Thus it was quite justifiable to shorten the time later to 3' if the majority of the crabs went through the openings in that time, although to do this would give manifestly more favorable results than to wait longer.

2 On this 5th. as well as on each successive day up to the 9th. it was noticed that as soon as the screen was put in, and the fish was placed behind it, the crabs were much agitated and some started for the openings.

3 The three that did not go in on the previous day had been removed for the purpose of determining if possible if there was already a certain permanency of habit among the twenty-seven entering; the result was confirmatory. The three were then returned.

One had died.

day under these conditions, fish and screen were removed, and the latter was carefully washed with running sea water, as it was also each time just before using. After a few days of this treatment immediately upon the insertion of the screen the

crabs became most agitated, some hurrying and scurrying about, others making almost directly for one of the openings. A preference for the right-hand one seemed to exist and this may be connected with the right-handed asymmetry. No attempt however to investigate this matter systematically was made, and there were no "landmarks" in the aquarium, such as stones, etc., whereby a path could be learned. The preceding tabulated results show, (1) the total number each day, (2) the number going into the dark within (3) a certain time, when fed under the condition named, (4) the per cent. entering, and (5) for comparative purposes, the per cent. entering in one minute.

At the end of the 5′ the fish was put behind the screen in order that the association might not be broken up by the crabs entering and finding nothing there. When this was done all but one of the eight outside, four of which had gone in and come out again, entered within 2'. This was done each successive day.

2 Then fed, and in 1' all but one had entered. This shows again the constant improvement, the learning taking place in a surprisingly short period. 3 In one minute after feeding all of the seven outside had entered. After feeding all entered in 11⁄2'.

5 After feeding all but one entered in 2'.

Accordingly, after the 8th. day, evidence having been thus secured that an association between the two "constructs," food and sense-darkness, had been established, its efficiency was

further tested by simply putting in the screen with no fish be. hind it, and the record of the crabs entering was taken in the same way as before. The results are tabulated above.

These results, if confirmed by control experiments, must, we think, be accepted as showing conclusively that the Hermit crab of the species longicarpus, firstly, forms an association between two sense-constructs, which, secondly, can be interpreted as showing that the crab, subsequently, when only one stimulus is presented, reproduces an image of the other. The same reaction, entering the dark, which previously demanded two stimuli, is later secured with only one stimulus; the other therefore must either be excited or reproduced. We may say perhaps that if, when only the screen is put in as in the second series, only visual perception or recognition takes place, then there is no reason why the crabs should not remain where they are, in the light, which is their natural preference. The screen, which they now recognize, has however through association come to ' mean for them other than something to be avoided; it means "food," and this meaning is present when the food is not. The difference between YERKES' experiments and these consists, therefore, in this, that YERKES' crabs "acquired the habit" of going by a correct path from a place disliked to a place liked; the Hermits on the other hand go from a place liked to a place naturally disliked, but "artificially" liked because of food either there or may we say "expected" to be there. This must mean that an associative element at first external, i. e., physical, but now no longer that, is, nevertheless, now present as internal, and its internal presence must be due to either an excitation or a reproduction by the other stimulus. If the latter, then the Hermit may be said to remember vaguely, i. e., to reconstruct.

These conclusions are strengthened by the following control experiments:

Lot 4.

EXPERIMENT III.

Aug. 20th. Forty crabs in a similar aquarium; the same screen was used, carefully washed each time. The previous procedure was reversed here, by placing the crabs be

hind the inserted screen first, and feeding in the light.

From

the data already obtained it would be expected that in this a greater per cent. of crabs would from the start make their exit than had in the first experiment made their entrance. This proved to be the case as the following record shows:

Comparison of this with Experiment I shows that not until the 5th. day was as high a ratio approximately obtained. In both cases of course the crabs are guided by taste; the two experiments together, therefore, serve to demonstrate the natural preference for the light, and the quicker results obtained when this works with than when against the feeding instinct.

EXPERIMENT IV.

Lot 3. This same preference was demonstrated by a second control experiment. Twenty-four crabs were placed in a large rectangular aquarium directly next to a window with a northeast exposure. Half of the aquarium was darkened by holding down with a stone a box with one end taken out. Sand was placed over the entire bottom.

Twenty observations failed to discover a single crab inside the "dark line" from edge to edge of the box. The crabs, therefore, seem to prefer the light to the darkness, at least in the aquarium.

EXPERIMENT V.

Lot 6. A third very similar control experiment was also confirmatory. Thirty crabs, the aquarium jar and screen exactly as in Experiment I were employed. The question might arise, would not the crabs possibly "learn" to go behind the screen when this was inserted even though no fish was there, the necessary feeding being done at other times. Should they do this the conclusions from Experiment I would be quite

Ten

worthless. The results were that they would not enter. observations were made; in seven of these cases no crab (within 5') went further than to the edge of the entrance, which had in no experiment been counted as "in"; of the three other cases, within the period of 5 minutes, 5, 8 and 5, respectively entered, but there was no evidence of progress.

ADDITIONAL EXPERIMENTS

Maze Experiment. Can the Hermit learn to go through a maze for its food? A simple maze was constructed of thin boards in such a way that it was necessary for the crabs, 16 being used, to enter at one end, go through to the other, enter the second passage, go back to the other end and enter the food compartment, The crabs were first given an opportunity to "become acquainted" with the maze by leaving it in the aquarium about two hours. The next day this was repeated with the result given in the first line of the table. are as follows:

The results

This experiment was not conducted as carefully as it would be possible to make it; the crabs could not be watched all of the time, but only occasional observations and a record, as the above, at a stated time, made. The results showed, however, though perhaps not as clearly as might be desirable, that an improvement had taken place, and they have a value, as being in general confirmatory of the first experiment.

Conclusions.

The Hermit Crab, Eupagurus longicarpns, is capable of p.ofiting by experience, in a rather short time, by associating