The special subject concerned is Arnold van der Kammen (A.K.), who has reported scoring rates between 62% and 76% (95% confidence interval) in a self-administered binary forced choice card color guessing test. This test was suggested in a correspondence course on parapsychology in which the subject enrolled in 1993. High scoring did not decline after the teacher (one of the authors, J.G.) stressed the importance of measures to prevent sensory leakage. A remarkable free-response hit of A.K. in an informal experimentstimulated us to set-up a formal experiment. Appreciating the relevance of the home context in which these strong results were obtained, the following procedure was designed to leave intact the circumstances in which A.K. performed well. The few precautions that were taken, were justified by telling A.K. that these would protect him from accusations afterwards (in case of significant results!).


Each of 4 sets of 100 new playing cards, 50 red and 50 black, were randomised by shuffling numbers from 1 to 100. Even numbers resulted in red and odd numbers in black cards. The shuffling occurred on the basis of the output of a hardware Random Number Generator. The 4 sets, named A, B, C and, D, were prepared and split up in 10 packs of 10 cards each by an assistant not aware of the intended use of the cards. The 40 resulting packs of 10 cards were put into opaque bags. Next, the 40 bags were sealed and were put into 40 Loksure security bags (a Loksure security bag can be identified by a unique number and cannot be opened without destroying it - Delanoy et al, 1993). The experimenter (J.G.) delivered the packs at A.K.'s home. In a written instruction handed to him it was stressed that he could do the series at his leisure and it was suggested that he should keep a diary of his proceedings.

He was informed that he had to try to guess the colors of each card in a pack down through, while holding the pack with its series number up. Unknowingly to A.K., the cards from set 4 (D) were in fact kept at the Parapsychology Institute. Instead, A.K. received packs with dummy-cards (100 jokers).

He was provided with a scoring form similar to the self-devised form he had used in the correspondence course. It was explained to him that, due to a theory-relevant element of the experiment, he would receive feedback later on three of the four packs (of course he was kept blind as to the identity of the set that he would not receive feedback on).

Apart from the controlled sets (A, B, C, D), A.K. received another 400 cards in 40 opaque sealed bags which were not put in Loksure bags. He was allowed to open these series (the sets a, b, c, d, each also containing 10 packs of 10 cards) immediately after he had finished the 100 calls for a set. These uncontrolled series served as a sort of reassurance for A.K., since here he would receive the immediate feedback he was familiar with during his former self-administered series (feedback of the A, B, C, D series was expected to take days or even weeks because the scoreforms and packs were collected after ALL series were completed).

Therefore, before starting a run of 10 calls, A.K. had the opportunity to choose between direct feedback sets (a, b, c, d) in sealed bags and delayed or no feedback sets (A, B, C, D) in Loksure bags.

For the direct feedback series, there was no control whatsoever against cheating. If he chose to do so, the subject could first open the bags and subsequently enter his guesses on the scoring sheet according to the card order he found in the packs.

After completion of the eight series, we collected the bags and scoring lists. The following day the identity numbers on the Loksure bags were checked and subsequently we took the sealed envelopes out of the Loksure bags. The closed envelopes with the scoring lists were then returned to A.K.. The subject was the first to observe feedback on his calls because the experimenters delayed checking scoring lists against the card orders until the subject phoned that he had finished his checking .

Design and hypotheses

Before returning the packs to the subject for feedback, set B was reshuffled and the dummy cards of set D were replaced with the real cards kept at the Institute. Set A was left intact, while set C was not returned at all (as pointed out above, A.K. was informed beforehand that he would not receive feedback on one of the sets). Therefore, we have the following four conditions:

Table 1. Experimental conditions
A correct feedback
B incorrect feedback
C no feedback
D correct feedback (real pack D kept at a distance during call-phase)

Possible theoretical explanations and corresponding hypothetical implications of this study were announced in advance (in Bierman & Gerding, 1992):

  1. The traditional (second sight, third eye or sixth sense) psi model will be confirmed by significant hitting on all sets A, B, C and D. Hitting on A and C and not on B and D will fit the same model, however in this case fraudulent handling of the Loksure bags by sophisticated methods can not be excluded.
  2. Significant hitting on A and D and not on B and C would conform to Observational Theoretical models (e.g. Millar, 1978), which stress the relevance of correct feedback. In this case a theoretical relevant result will have been obtained, while deception is quite improbable. Strong support for the Observational Theoretical models will be obtained if A.K.'s B-list is significantly correlated to the incorrect feedback.
  3. Hitting on A, B (B is the real set during guessing, on which incorrect feedback was given) and C, and not on D (the joker set) could suggest cheating.