According to Krall their progress was, as in the case of Hans, astonishingly fast. Moreover, the two horses displayed different aptitudes. Muhamed was much quicker at arithmetic. Instruction began with this subject with an hour and a half to two hours of " study " a day for each animal. At the end of three days they could recognize the first numbers, 1, 2. 3 written upon the board, and would touch with the muzzle the one called for. At the end of ten days Muhamed could count up to 4. Several days later the tens were explained to them and they were taught to designate them with the left foot, the right being used for the units. On the 14th 304 THE LIMITS OF TRAINING IN ANIMALS of November, 1908 (the horses were purchased on November 1st, the first lesson beginning on the 2nd), Muhamed performed correctly such small additions as 1 -|- 3, 2 -|- 5, etc., and also subtractions such as 9 — 3. He passed to multiplication and division on November 18th, to fractions and the addition of fractions on the 21st. In December he was taught French and replied equally well to ques- tions asked in French or in German. In the following May Muhamed could extract square roots and cube roots, and could do sums like the following: 3 3 In February, 1909, spelling lessons commenced. Spelling was made possible by means of a conventional alphabet (where each letter or diphthong is represented by a number between 11 and QQ, e.g., 11 would represent e, 22, h, etc.) as follows: etc.
30 The horse spells by tapping with his foot (the tens are tapped with the left foot, while the units are tapped with the right foot) the number corresponding to the desired letter. By this method Zarif, at the end of four months, could spell, in his own fashion, words spoken to him which he had never seen written. Below are presented some of the different spelling of " Pferd " as spelled by Muhamed and Zarif: Muhamed: bfert. bfrt, fard. fart. fert. frt, faart, faerd, faert, farb, f pferd, frrt. pilrd. pfart ppverd, pfer, pferd, tfert, fed; Zarif: bferd, fared, fferwt, fvert, pfrde, sdfert, pfert, bffet, fdaerp, etc.; Zucker was spelled zkr, zukr, ziiqr, zuqr. czukr, sucr, szukr, zuakr, cukr. zucher zuker. zucher, suqker. etc.; Muhamed spelled his name, maemuaamt, muahmet, muamiit, etc.; Zarif spelled his, tsarem, THE HORSES OF ELBERFELD 305 zarif, sfrai, zuarif, zuriif. Finally Mr. Krall, noticing that the horses tapped " out of lessons," took down their tappings to see if he could make anything out of them. Sure enough! While the discourse was at times as incomprehensible as the first babblings of an infant, nevertheless there were fragments which could be understood. One day Krall told Muhamed that he was going to give him carrots: " fiinuf " (i.e., fiinf, 5) replied Muhamed spon- taneously. At another time he replied: " iohn, hf r, gbn " ( iohann haf er geben, John gives oats ). Soon he detected genuine conversa- tion between the two pupils. One morning when Zarif had shown himself lazy his companion was asked: " Warum war Zarif nicht lieb?" And Muhamed answered: " weil vaul isd " (because he is lazy). "And why is he lazy?" "Weil r sagt begin will nijd wisn " (because he said at the beginning that he didn't want to know anything).
Notwithstanding the lesson taught by Hans, Krall's ex- aggerated report was received favorably by many distin- guished men. Haeckel wrote to Krall: ' ' Your careful and critical experiments show in a convincing fashion the existence of reason in the animal, which I have never doubted." On the other hand, Ettlinger attacked Krall's experiments with a great deal of vigor. Professor Dexler, in a strongly worded article, declares that " Krall's book is a vile blot on our contemporary literature. Born in the poisoned atmosphere of humbug and of trickery, it is a monument raised to the cult of the beast." Argumenta- tion, however, is not the method to settle "behavior ques- tions. Many tests have been made upon the horses but none by thoroughly trained investigators of behavior. The two reports by Claparede are probably the most reliable we have at present. The first was made in August, 1912; the second in April, 1913. The first set of tests is not really scientific in character since Krall himself acted mainly as questioner, and since even when Claparede acted as questioner adequate controls were not introduced. Claparede 's second set of tests were undertaken in a much more critical spirit.
Claparede's Examination: Results of the first set of observa- tions: Zarif was first brought in. Krall showed us (summar- ized from Claparede's report) that his pupil understood French. He wrote on the board (phonetically without the subleties of orthog- raphy) fat sero, (the horse made the sign of negation with his head, which signified zero in his language), then, Kompta dis (the 306 THE LIMITS OF TRAINING IN ANIMALS horse tapped once with his left foot, correct answer). Having written 34 he asked the horse to read the number both ways (reply 34, then 43, right), then to multiply the two figures (3 x 4) (reply correct), and then to square the larger of the two figures. That time Zarif made an error and replied 15 instead of 16. When asked to correct it he replied 15, then tapped 24. Zwei und zwanzig was written on the board. The horse failed twice to read this number. Then Krall wrote " adire zu elf" (add it to 11). He replied 33. The horse was rewarded for this answer. " Adire zu ein und dreizig " (reply correct). " Zu vangt troa " (phonetic for 23 ). This time the answer was 44 instead of 45. When told to correct it Zarif gave 54 (which is the reverse of 45) and on a third trial gave 45. Krall wrote on the board " dus " (12). The horse read 22. "Add 12 and 2." Reply 24, then 14. Krall wrote on the board y/ 25 X \/49 =. The horse replied 24. Krall said it was wrong. He repeated 24. Zarif showed signs of embarrassment (at least as interpreted by Krall); he dropped his head like a shamed school boy. Krall told him that several people were present, one of them a professor who was taking notes and putting down all the wrong answers. Then the horse began tapping of his own accord. Krall put down on the board the numbers he tapped: 14, 26, 23, 54, 13, 13, 32. Translated the corresponding word is schlprrd. What did it mean? Krall said he could not make it out. We quote Clapar&de: "But it was not hard for me to discover between these letters and the letters of my name a striking analogy. Krall said it was impossible that Zarif had attempted to spell my name for he did not even know it, but in the afternoon he told me that he remem- bered having mentioned my name before Zarif while talking to a third person. Was it my name which he voluntarily spelled, wishing to convey the information that he knows that it is I who is taking notes? Cruel enigma! " In order to end this seance which had lasted twenty minutes, Zarif was recalled to his problem (the multiplica- tion of square roots): he gave the correct answer, 35. Then Hanschen was brought in. He resembled an ass, but only externally! Indeed with astonishing quickness, considering the fact that he had been a pupil only a few months, he did a dozen correct additions of numbers of two figures written in a column on a little blackboard before him in an inclined position, for he was too little to use the board fastened on the wall. For the sum of 23 -]- 12 -)- 20 he an- swered first 65, then 45, then 55, which is correct. Strange to say he corrected himself. He hardly finished one sum before he started another. The problem 2 X 32 -}- 12 brought out a number of wrong answers. The order " zweimal drei und dreizig" (2 times 33) brought first 86, then 66. During Hanschen's exercises, which lasted twenty minutes, Krall stood completely behind the animal. He did not leave his position except to write the problems on the board, which was done in such a manner that no one near by could see it. The groom remained perfectly motionless the entire time. C. was beside him and watched him closely. He did not detect the least sign which could be suspected of being given in correlation with Hanschen's replies, who had his head lowered, with muzzle almost on the board.
At last Muhamed was brought in. He is the genius of the group. He is much more lively, much wider awake than Zarif. Krall wrote upon the board y'' 36 x v/49=z:. Muhamed first answered 52, then, when told that that was wrong, 42, which is correct, Krall then substituted the -|- sign for the X ^^S^ i" ^^^ problem. The answer was correct, 13. Krall then gave him the following problem: and went out of the room. Muhamed looked at the problem and immediately answered 115, which is wrong. (The hundreds are tapped with the right foot, but are given after the tens, which are tapped with the left foot. This distinguishes them from the units, which are also tapped with the right foot, but which precede the tens.) Krall then called to him from without to try it again. He tapped 25, then 125 (which is still wrong, the correct answer being 135). When an answer is wrong Krall prefers not to insist on it, and changes the problem, only to return to the original problem in a moment. Clapar^e was then asked to choose from a large list of squared numbers a number for Muhamed to extract the square root. He chose at random 64516, the root of which is 254, but Muhamed persisted in tapping numbers of only two figures, which made us think that he wanted to say something by means of the conventional alphabet, which contains only numbers of two figures. So these tappings were noted with the following results: rihgvgdhaig. It was incomprehensible. He was asked to repeat it. The second attempt was: rmigdjg. This was no clearer. But the similarity between the two series of letters was too striking to be entirely haphazard. But Krall could not imravel it. Muhamed was then asked how many syllables were in the word he wished to spell. He answered three. " Is the word finished? " " No, " he replied. " Un- fortunately we could get nothing more out of him." Krall then went back to the two problems which he had not solved and which were still written upon the board, and this time, with no difficulty, Muhamed replied 135 to the one and 254 to the other, which are correct. As a reward he was given three carrots: " How many carrots have you eaten? " he was asked. " Threej" he replied. To the card y/ 117649= Muhamed answered 4, then 13, then 346, then 347, then 343, which is right. To the card y/ 383161 =: he answered 58, 317, then, after two " bavures," 619 (right). By "bavures" are meant incomplete wrong answers given by the horse, — incomplete because Krall, seeing that the number of units is wrong, interrupts the horse without letting him go on to the tens and hundreds. Upon this last correct answer, the exhibition ended.
The remaining seanees were equally productive of star- tling results. It is unnecessary to present them in detail. From such a set of tests absolutely nothing can be con- cluded, and yet Claparede, being much under the influence of his emotions, devotes many pages to speculations con- cerning the analysis of his observations. The results of the 308 THE LIMITS OF TRAINING IN ANIMALS tests made in April of the next year show that Claparede 's attitude has become scientific and critical. We give them in some detail.
Later Tests by Clapaeede: On the 26th, 27th, and 28th of March, 1913, Claparede paid another visit to the horses, accompanied by Dr. J. de Modzelewski and Dr. Weber-Bauler, both of Geneva. Mr. Krall was not at Elberfeld during the first series of tests. We give below a rather careful summary of the tests on the various horses: Tests on Moening of the 26th of March. — Hanschen: During the tests on this horse Claparede, with the two gentlemen named above were present. The horse was led in by the groom, Albert, who remained in the room. The pony was shown two cards desig- nating the number 53. He responded correctly. The digits were next transposed, making the number 35. The response was correct. Every one then left the room exept Claparede. Having shuflfled a series of cards on which numbers were printed he drew out one at random and placed it under the eyes of the horse without glancing at it himself ( procedure " without knowledge " ). The horse tapped 42, which was wrong, the correct answer being 45. Twice more this type of test was repeated. The horse tapped 73 for 25; 53 for 84. We have here three wrong answers when the observer tested by the procedure "without knowledge." For the sake of comparison he repeated the test, first glancing at the cards (procedure "with knowledge " ). The results were no better: 52 was given for 33. When asked to repeat it he gave 31. The experimenter wrote down 63; the response was 16. He then wrote 43 on the board: the answers were in order, 42, 52, 53, 64. Calculation was next tried: 34-f-25 was written on the board: the replies in order were, 22, 42, 73, 66. The horse was next tested with 24 -f- 12: responses in order were 37, 46, 36, the latter being correct. At this moment the assistant came back. He wrote on the board 34 -f 25. The horse returned 73. After this there followed a long series of tests the answers to which were mainly wrong, although now and then a correct one would be given. Under such conditions it seemed useless to give tests where the experimenter was in ignorance of the answers. The horse having tired, he was removed.
Beeto: The blind stallion Berto was next tested. The animal had double congenital cataract. The horse had learned speedily by the contact method of training employed with blind human beings. It must be remembered that the type of response called for here is much simpler than that demanded by the other horses, at least so far as the " reading " of numbers presented through contact is concerned. At the end of five lessons, the animal, according to his trainers, could add simple numbers, such as 5 -]- 1 and 5 -f- 2. When the horse was brought in Claparede verified the state of the animal's vision. The horse was completely blind. ("II ne pent y avoir aucun doute a cet egard: I'animal n'apercoit pas une carrotte que Ton place sous y eux." It is hoped that Claparede does not mean seriously to affirm that such a test is really crucial. The testing of complete blindness in an animal is very difficult. It is LATER TESTS BY CLAPAR^DE 309 doubtful if any physician can determine this surely except in cases where complete degeneration of the retina is revealed by skioscopic methods.) Berto was quite restless and in order to soothe him the groom held the halter. The groom retired during the reply of the horse. In certain tests the groom was not allowed to hold the halter even while the question was asked. The groom first asked the horse to tap out the following numbers: 2, 14, 42, 33, 40, 2, and to add 2 and 3 and to multiply 5x3. Only a part of the answers to these questions was right (see appended table). The horse was able to give a number traced upon his rump. In view of the blindness of the horse, procedure " without knowledge " could not easily be undertaken.
MuHAMED: The horse Muhamed was given (by cards) the number 35. The reply was incoherent. Several of them finished with the digit 5, e.g., 15, 25, etc. For 67 he returned 37, then correctly 67: for 48 a series of wrong responses with none correct. An exercise in spelling was next attempted with complete failure. There was incoherence all along the line. Under such conditions it was useless to attempt to make tests under rigorous conditions.
Afternoon of March 26th: — Simple exercises with numbers were continued but the results were deplorable. Zarif read very nicely the first number Claparede gave him, but after that one correct response nothing further could be obtained in the way of correct responses. Two incidental observations were made on the general behavior of the horses during the period of response. When one insisted that his reply was wrong, once, contrary to his habit, he began to watch the groom while tapping, but the response was wrong all the same. On another occasion the horse, which ordinarily tapped with machine-like precision, hesitated with right foot poised in air before beginning the reply. Muhamed was hardly more brilliant than in the morning. There were several good responses which were interesting in view of the fact that only Claparede and his colleagues were present. The horse read correctly and on the first trial the numbers: 5, 42, 49. To the addition 3 -)- 4 given on the card he replied 34. All then left the room except Claparede. Several numbers were then presented ( procedure " without knowledge " ): all responses were wrong. E.g., 46 was read 44; 43 was read 75; 7 was read 6. To 22 -[- 15 the horse replied 37, which was correct. After that response no further correct ones could be obtained regardless of the method of pro- cedure. Berto was then brought in. He gave a pretty series, Claparede doing the questioning. He responded correctly to 2x^1 and to 3x11- When asked to substract 3 from 7 he tapped 10. Hanschen was next brought in. He did not give a single correct response.
Afternoon of March 27th: At this session Krall, de Modzelew- ski, and Claparede were present. They examined successively Mu- hamed, Zarif, and Hanschen without getting any correct responses. Berto, the blind stallion, alone gave good replies. A recital of these would be without interest. The horses replied wrongly even to Krall.
March 28th, Morning and Afternoon: Claparede, de Weimar, and Krall were present. Several fruitless attempts were made to get 310 THE LIMITS OF TRAINING IN ANIMALS correct responses. Krall advanced as a possible explanation of the poorness of the results the fact that the horses were shedding.- Tables of results. — From cle Modzelewski 's note to Claparede's report, we copy the following table showing the number of correct and incorrect responses returned during the several tests: Total Number of Right and Wrong Responses March 20 Poor R ight but with help Good Hanschen, a.m.
42 5 5 P.M.
47 6 2 23 5 2 P.M.
9 2 0 P.M.
Responses to the Moke Difficult Problems (of the Above List ) Poor Right but with help Good Hanschen, 31 3 1 Berto, 5 2 6 Zarif, 1 1 1 Muhamed, 2 1 Total Number of Right and Wrong Responses March 21 Poor Right but with help Good Muhamed, 77 4 2 Zarif, 13 1 5 Berto. 26 8 9 Responses to the More Difficult Problems (of the Above List ) Poor Right but with help Good Muhamed, 26. 1 1 Zarif, 9 0 3 Berto, 33 4 1 Hanschen, 11 4 4 2Yet another gentleman only a few days later obtained very successful results. Krall then stated that Claparede's failures were due to the fact that he had not gained the confidence of the horses.
SmOIARY OF TESTS 311 As will be seen upon comparison with Pfungst's work on Hans, the percentage of correct responses in Krall's horses is about the same that he obtained by the procedure ' ' with- out knowledge." How shall we account for the 8-11^ of correct responses? Surely chance will account for a cer- tain percentage. It would be impossible to train horses in this way for months to respond to cards and figures upon the board without two things appearing: (1) Train- ing upon special numbers, cards (of square roots, etc.) establishes a set of special simple reactions. It is quite probable that many of the combinations (even those sug- gested by the outsider) are ones upon which the horse has been specially drilled. (2) If the horses really tap in groups (numbers?) spontaneously, by the theory of probabilities some of the " answers " must be right. What lend color to such a view are the statements of cle Modzelewski to the effect (a) that the very simplest problems are answered with no greater percentage of accuracy than the more difficult ones; and (5) that no matter how simple or how difficult the problem is, the horses start mechanically upon the answer; and (c), finally, the horses do not even look at the letters and figures upon the board. Furthermore, it lies well within the bounds of probability that there are cer- tain means of rapport between the horse and his questioner which may serve the horses of Elberfeld as " minimal movements " served Hans. We say this without prejudice and without any intention of intimating that fraudulent means are employed. This would account for the high ■percentage of accuracy which is sometimes obtained by Krall. (And since in many cases w^here positive results have been obtained in Krall's absence the grooms have not been excluded, a similar explanation is at hand to account for this fact.) On the whole, it seems not improbable that we are dealing here with responses w^hich do not rise above the level of those given by Hans.
Peter. — The chimpanzee has always been a favorite sub- ject of eulogy. Very few of. the highly trained animals have come under observation where experimental condi- tions could be sufficiently controlled. Several experiments 312 THE LIMITS OF TRAINING IN ANIMALS upon imitation have already been noted on p. 282. Witmer has given a report of some observations made upon Peter in the Psychological Clinic. Peter's vaudeville perform- ance, which we have several times witnessed, is most im- pressive. He skates readily and with accuracy; drinks from bottle or glass; lights and smokes a cigarette (some- times after a good deal of " help " from the trainer); eats with a fork; and can '' ring for the waiter." Further- more, he has varied motor accomplishments ranging from the riding of a bicycle to the threading of a needle, and from locking and unlocking locks to tying and untjdng knots in a string.
When Peter entered the clinic Witmer's secretary bent down and he shook hands with her and then kissed the back of her hand. He then began to skate around the room in a lively manner and next examined a camera, attempting to turn the screws and actually manipulating the shutter. The trainer gave him a cork-tipped cigarette and offered him a box of matches. The cigarette was put into the mouth with the cork end first. He next took a match from the box held out to him and struck it on the side of the box. A different kind was offered him wrong end first. He at once reversed it and put the right end (tube) in his mouth. Peter was able to string beads. An incomplete string of beads was held out to him and then a bead w^as taken from a box and strung before his eyes. This was twice repeated. Then the box and beads were handed to the monkey. He promptly put a bead into his mouth, feeling it with his lips and tongue. He then took it from his mouth and strung it. One bead after another was taken from the box and strung; each was tested with the lips in order to get the hole in the middle in the proper position for stringing. When tested on the pegging board, after being shown how to put the pegs in, he put them in irregularly. The impulse to pull the pegs out was strong but was inhibited upon a word from the trainer. The pegs were given to him one by one. When tested on the lock which he used in his vaudeville performances he locked and unlocked it mechanically, A small padlock with a staple attachment was then given him. The staple had to be pulled entirely off the lock at the end opposite the hole. The key was a small one, difficult to insert and difficult to turn after it had been inserted. It was unlocked for him and the staple attachment taken out. The staple was put back and locked with key withdrawn. Peter reached for the lock, inserted the key, and unlocked it more rapidly than the experimenter. When told to put the staple back and unlock it, he inserted only one prong of the staple. He was told that it was wrong and had his ears boxed by his trainer. The lock was taken from him, the staple slowly pulled out and moved several inches away, then reinserted. This was repeated several times. When tried, he seized the lock, slowly and carefully pulled out the staple, then carefully reinserted it and shoved it THE BEHAVIOR OF PETER 313 home with a smack of his hand. A small box with a tiny keyhole was next given him. The key was on a ring with a large number of other different sized keys. It was the smallest one on the ring. When handed the ring, he tried first the largest key, then the next largest, and then the third largest, etc. The box was then opened for him and lumps of sugar shown on the inside. The small key was pointed out to him. The box was locked and the bunch of keys shaken up and again handed to him. He was unable to pick out the smallest key. It was picked out and handed to him. He then opened the box with the key and obtained a lump of sugar. He failed again on the second trial. A double-headed hammer and a piece of board on which were some nails and screws were then given to Peter. When the hammer was handed to him he proceeded to drive several nails into the board with the proper head of the ham- mer. He never mistook a screw for a nail. WTien given a screw instead of a nail, he stuck it into a hole in the board and at once selected a screwdriver, paying no attention whatever to the hammer lying on the table. The screw was long and wobbled as he turned it. He turned the screw always in the right direction. He used both