Preface In February 1996, publishing house PROMETHEUS released "The Mars effect. A French test of over 1,000 sports champions" by Claude Benski et al. including a chapter by Dr. J. W. Nienhuys "Comment on the CFEPP report" in which the author severely criticizes my studies on the Mars effect and its positive conclusion. The article below is my rejoinder. In the first place I am addressing readers who are familiar with Dr. Nienhuys' critique. A formal scientific style in the present piece has not been maintained throughout considering the somewhat less formal environment of its distribution: The Internet. A full account of my reanalyses of the CFEPP's and other skeptics' studies on the Mars effect has just been published by URANIA TRUST, London: "The Tenacious Mars Effect" by Suitbert Ertel and Kenneth Irving Foreword by Jim Lippard Goettingen, April 2, 1996 Suitbert Ertel --------------------------------------------- ######################################################################## On Dr. Nienhuys' stamping the evidence for a Mars effect based on the CFEPP's athletes data. ######################################################################## Suitbert Ertel Abstract ######## According to Ertel (1994/95) and Ertel & Irving (1996) the third attempt of organized skeptics to disprove the reality of Gauquelin's Mars effect has failed. The French CFEPP, responsible for the last trial, however, alleges that it was successful (Benski et al., 1996). Nienhuys, in Benski et al, acclaims their negative result and attempts to discredit my counterevidence. In the present rejoinder I maintain that the CFEPP's crucial expectancy calculation is wrong, that their selection of athletes was not done blind, that selections were biased, and that the researchers avoided to use fairer test conditions (optimal Mars sector division). Regarding my own analysis of the same data, I maintain that my data base was correct, that the Mars effect increases with sports success (eminence), and that an eminence correlation could even have been discovered by the CFEPP themselves. After summarizing the many flaws in the CFEPP's study - the details of which are based on Nienhuys' own scrutiny - and in view of their refusal to correct the deficiencies which Gauquelin had already drawn to their attention, it is concluded that Benski et al's results and Nienhuys' defense of them must be strongly rejected. A new, straightforward procedure of analyzing the CFEPP's data (successive birth time shifts) confirms my former evidence for the Mars effect, making it verifiable by mere visual inspection of a Figure showing the genuine birth percentage of athletes in Mars G-sectors as a peak among time-shifted controls. Introduction ############ The French skeptics (CFEPP), after fourteen years of testing the claim of Gauquelin's Mars effect with sports champions, concluded in Benski et al. (1996): "[The results] show no evidence for the influence of Mars on the birth of athletes." In response to the CFEPP's negative conclusion, acclaimed by Dr. Nienhuys in his chapter to Benski's book, I recently introduced a more straightforward test for planetary effects and applied it to the CFEPP's sports champions (to be published in the next issue of CORRELATION): Birth hours are shifted stepwise by one hour forward (1, 2...11 hours) and one hour backward (-1,-2...-11 hours). For each of the 22 new samples of champions with shifted birth times the percentage of births in Gauquelin Mars sectors (G%) are calculated. Result: G% of the original data (not shifted) is largest and drops both with forward and backward shifts. (See Figure 1) The observed G% with genuine birth times thus outdoes 22 controls. The presence of the Mars effect in the CFEPP's data has thus been made visible, statistical tests are hardly required, and previous issues of the controversy, such as eminence and expectancy, are no longer crucial. Figure 1 Mars G percentages for genuine birth times and for shifted controls (here without numerical precision) o Genuine birth time x shifted birth times G% (Mars percentage) I I I o I x I x I x x I x x chance x I x --- x x I x x expect x I x x x I I I I____________________________________________________ 08 07 06 05 04 03 02 01 00 01 02 03 04 05 06 07 08 size of birth time shift (hr) Nevertheless, Dr. Nienhuys' recent attempt in Benski's book to refute my earlier evidence deserves detailed comment. Less of a problem for the reader interested solely in the facts is Nienhuys' derogatory style, culminating in: "Ertel's [] analyses are rather like tea-leaf readings and he has not understood the value of the CFEPP's experiment at all." But even though probably not much approved of by most readers, such presumptuous ad hominem is likely to raise suspicion among some that Ertel's claims should not be too much trusted either. As distrust is merely uncertainty spiced with a negative flavour, it may be reduced by unflavoured factual information. Two Scyllas lured in my 1988 finding Strangely, on several occasions and again in his Benski chapter, Dr. Nienhuys quotes my 1988 result showing that Gauquelin's selections of sports champions were biased. I had found that Gauquelin tended not to separate low eminence champions from his experimental sample when they had Mars in a G-sector. Nevertheless, Dr. Nienhuys' reference to this finding seems puzzling, because the finding itself does not invalidate the Gauquelin claim, but instead actually supports it. How can this be? On the one hand, as we might expect, Gauquelin's bias raised the champions' overall G% ("Wow: Gauquelin's Mars G% surplus is due to biased selections!"). However, at the same time this bias also adversely affected the eminence correlation, which is based on G% differences between low and high eminent athletes, and this was in fact greatly diminished ("Gosh: The eminence correlation cannot be due to Gauquelin's biased selections"). (See Figure 2). By my reuniting Gauquelin's published (high G%) with his unpublished (low G%) sports champions, the overall G% level went down, but at the same time the eminence correlation emerged more clearly, i.e. G% differences between low and high eminent athletes became considerably greater - a seeming paradox: Gauquelin's Mars effect claim was thus supported by mending his biased sample. Figure 2 Effects of Gauquelin's biased selections (exaggerated for didactic reasons) o Gauquelin's published sample x Gauquelin's published and unpublished sample reunited G% (Mars percentage) I I o o o I o o x I x I x I x I x I- - - - - - - - chance expectancy I I_______________________________ 1 2 3 4 5 eminence ranks (5 highest rank) In view of this result and in view of subsequent results obtained from the CFEPP's data using the same eminence measure, Dr. Nienhuys reacts to this measure as the ancients reacted to Scylla, the attractive nymph who later became a monster. How attractive was my early discovery of Gauquelin's bias (which caused G% to be illegitimately boosted), and how monstrous is my later finding, obtained by the same procedure, that the CFEPP's data selection was also biased (causing G% to be illegitimately suppressed). Apparently, Dr. Nienhuys solves this problem by a comforting doublethink which allows him not to notice that by discrediting my eminence procedure and its more recent undesirable result he would logically have to renounce its former desirable result. Dr. Nienhuys' objections to my critique of the CFEPP's method ################################################################ My critique of the CFEPP's method was threefold. . Ertel: The CFEPP's expectancy of 18.2% is definitely wrong (disfavoring the Mars effect). Dr. Nienhuys rejects my contention that the CFEPP's expectancy (18.2%), contrasting with my own (17.5%) is wrong, because Ertel "made easily detectable errors himself", referring to an unrelated and inconsequential oversight in my 1988 paper. However, as a check I asked Mark Pottenger to calculate expectancies for the CFEPP's data, which he kindly did. He used a commercially available program of his own creation which can figure expected sector frequencies for a database using Gauquelin's classical procedure. The algorithm differs from those used by myself and the CFEPP. Recall that these methods of Gauquelin's were shown to be correct by CSICOP's Zelen test in 1977 and that they were even endorsed as sound by CSICOP's own research team (KZA 1983), as well as others. Mark Pottenger's printout showed an expected frequency for Mars of 17.5%, which confirms my earlier expectancy exactly, despite the differences between our methods of calculation. My message to the CFEPP, to Nienhuys and to Kurtz (publisher of the CFEPP report) that according to my own and Pottenger's results the CFEPP's G%(exp) turned out to be wrong was ignored. Their book was published without corrections. A requested hardcopy output of the CFEPP calculations which I could use to check the calculations was not provided. The CFEPP error was not only easily detectable, it was (unlike my own errors, with which Nienhuys distracts the reader) central to the issue, and it was, after all, easily correctable. Both Kurtz and Nienhuys owe purchasers of the CFEPP report an answer as to why this was not done, given the ample warnings. . Ertel: The CFEPP's athlete selections are biased (disfavoring the Mars effect) Dr. Nienhuys replies that "the mediocre athletes ... are just the 'lesser great'" which is a misleading belittlement of the highly significant differences regarding citation counts between athletes accepted by the CFEPP (and not accepted by Gauquelin) and athletes not accepted by the CFEPP (but accepted by Gauquelin). He then objects that Ertel "gives not even one credible example" of an athlete who did not belong in the sample. I didn't do this because anecdotal examples have no bearing on the citation counts needed for a credible conclusion. Then he objects that I did not use the reference books Le Roy and l'Athlege to assess the athletes' eminence. It should be evident, however, that the occurrence of the athletes' names in a book from which the names are taken does not provide independent eminence information. As early as 1988 I had replaced the method of selecting athletes after inspection of their sports careers (the traditional Gauquelin method, refined by the CFEPP). More on that below. . Ertel: The CFEPP should not have used the 12-sector division (disfavoring the Mars effect). A test of the Mars effect, if based on the "extended" G-zone (36-sector division), is fairer than if based on the narrow zone (12-sector division). Dr. Nienhuys disagrees on this point since "it is ... wrong to change the objective of an investigation halfway into it, especially if this objective was explicitly stated in the protocol". Here several comments are suggested: (1) If the Gauquelin Mars effect does not exist, it can neither become manifest with narrow nor with extended sector definition. (2) Gauquelin explicitly proposed using the extended-zone definition for studies of planet-birth relationships prior to the CFEPP's data collection. (3) The CFEPP readily and very clearly breached the protocol in several crucial ways, e.g. by not referring data selection to a neutral third party, and by collecting birth times from Town Halls without first asking Gauquelin to approve of their sample. This shows that the CFEPP's and Nienhuys' sticking to the letter of the 1982 Protocol was a pretense - attending to the Protocol was good for them only insofar as it disfavoured a Mars effect. Dr. Nienhuys' objections to my own way of analyzing the CFEPP's data #################################################################### . Nienhuys: "Ertel's database is not correct" N's critique: "30 Le Roy athletes are not marked as such, 15 athletes are marked Le Roy, but they should not". According to Dr. Nienhuys' e-mail message of 25 Nov 1994 to me, Le Roy code misses were 26, not 30, and he reported only 4 wrongly coded athletes, not 15. My assistant checked the Le Roy codes and found that 15, not 26 and not 30 codes were missing, and only one, not 4, nor 15 were erroneously coded. These inadvertancies, by the way, are as inconsequential as erroneous gender codes. Not 12 (Nienhuys), but 8 gender codes had been mixed up (my assistant). Nienhuys: "[Ertel] had managed to obtain parts of some office copy of the June 1990 report on which Gauquelin's proposals were marked in handwriting". We had received this material from Dr. Benski without being told why he had added five names. So we took it for granted that Dr. Benski wanted to have them added. Nienhuys did not address the main question: Do Mars sector frequencies obtained from my database (Mars sectors had been calculated by a different computer program) differ from frequencies reported by the CFEPP? Rob Nanninga (e-mail message of Nov. 19, 1994) kindly compared the CFEPP's and my Mars G-sector frequencies. He found only two G-sector cases more in my database as compared to the CFEPP's database, which is apparently due to slightly different longitude/latitude data for birth places. My G% level is 0.02% larger than the CFEPP's level. Conclusion: Incorrectness in my CFEPP database is negligible and inconsequential, just as ordinary typos in an article are inconsequential for its content. . Nienhuys: "Ertel's eminence counts using the standard sources do not show any trend" Dr. Nienhuys' Table 9 "Citation counts for 933 athletes" is not correct. First, he omits 1,066-933 = 133 athletes, and he does not say why. Second, he deliberately added Le Roy to the reference books, one of the two sources from which the names for this study was taken. As was indicated above, sources from which names are drawn do not add eminence information. In the present case this informatiom is even distorted since Dr. Nienhuys, after deciding to utilize one of the two sources on which the sample is based (Le Roy), should at least not have missed to also consult the second source (l'Athlege) for the same purpose. He did not, so Le Roy athletes had the advantage of an eminence score over L'Athlege athletes, and athletes who had entries in both books received only one instead of two citation scores in Nienhuys's count. Correct eminence counts for all 1,066 CFEPP athletes are given in The Tenacious Mars Effect p. SE-14. A test of monotonic eminence trend for the CFEPP's sample proved to be significant (p= .05, Kendall). There is some unexplainable variation in this curve with eminence grades 2 and 3, but the expected significance of the entire trend was nevertheless obtained. Conclusion: My eminence test was correct, and the existing trend was blurred only by Dr. Nienhuys' inconsequential decisions. . Nienhuys: "Ertel's eminence indicators based on the CFEPP's own coding are deficient Specifically, Ertel "is confused about the meaning of "M", "O", and "I" and apparently has not understood the criteria, even though he had at least three years to study the French text." In Ertel (1994, CORRELATION) the codes are given as used by Benski et al. (1996): M=World championship or record (mondiale). O=Olympic record. I=International match participation. I don't know what Dr. Nienhuys is talking about. He says that I used "CFEPP's uncorrected selection code". The full truth is that in 1993 I could not anticipate that Dr. Nienhuys would still discover quite a few coding errors in the CFEPP's data and that a corrected code was done only in 1995. He objects that I split the 116 cyclists into three eminence groups and "in other cases" I used a five-group division. Such differences are due to different sample sizes and associated differences of numbers among low-frequency, high-eminence subsamples. High eminence categories are only combined in order not to confuse the eye by random variation. Thus he refers to a subordinate technicality. Statistical trend-tests are in fact done with original (not combined) categories. What does Dr. Nienhuys mean by "There is no more talk of trends in the whole sample"? The reader cannot know, and I myself had to ponder. Nevertheless, Dr. Nienhuys conveys the impression that again Ertel had something to hide. Nothing of the sort occurred. I just picked the four largest subsamples in the CFEPP's individual sports categories, soccer, track and field, cycling, rugby, in order to point at the fact that even in the CFEPP's own data an eminence trend could have been discovered. Ertel's citation data, renounced by the CFEPP, wasn't even required. Nienhuys: "Le Roy is dropped as a source". Above I said that it is generally not sensible to use a reference source for both drawing a sample and counting citations. I made an exception to this rule only once, for the Gauquelin athletes', total of N=4,391, to which Le Roy contributed only 35%. The reason was that extracting for 65% athletes eminence information from this exceptionally good source, esp. regarding non-French champions, appeared advantageous so that the disadvantage of no or low eminence information regarding the 35% athletes extracted from Le Roy appeared tolerable. For my eminence study of 1988, I had checked whether the main result (significance of an eminence trend) was dependent on dispensing with Le Roy and found that the trend was significant irrespective of whether this source was used or not used. Back to the CFEPP's data: It would not have been admissible to use, as Nienhuys did, Le Roy for eminence counts alone. This source would have been admissible only together with L'Athlege, the second source from which the athletes were recruited, since athletes having entries in both reference books can be regarded as more important on the average than athletes occurring in either Le Roy or L'Athlege alone. After having obtained in the meantime L'Athlege codes from the CFEPP's printout (the book itself was not available to me) I ran another eminence test adding the two sources to those used earlier (based on the 36-ector system). The result: G% increases, as expected, with citation counts (see the following Table). Citations 1 2 3 4 5 5 and more Total 442 225 176 103 54 66 G-sector cases 102 60 44 26 15 19 ---------------------------------------------------------------------- Mars G% 23.1 26.7 25.0 25.2 27.8 28.8 --------------------------------------------------------------------- Conclusion: Dr. Nienhuys' criticisms have no bearing on the results of my analysis of the CFEPP data. Dr. Nienhuys' objections to my eminence procedure in general ############################################################ . "The eminence effect is quite weak". No doubt, in terms of effect size, the Gauquelin effect is weak, and so too is the eminence effect, but both are replicable. The question is not "is the effect strong or weak", but "does it or does it not exist". . "[The eminence effect] relies on the foreigners in Gauquelin's files". This observation is not correct, but it indicates what I expected (see the preceding paragraph): that Le Roy is a good eminence-discriminator for non-French athletes in the first place. . "[The eminence effect] relies ... on three sources that Gauquelin used as references ... So Ertel's effort of looking in 15 other sources is mainly a waste of effort." This needs clarification. The sample of sources utilized for citation counts (1) should not be determined post hoc (after extracting citations) and (2) should not be altered for subsequent planetary tests, except under specific and limited circumstances. In 1988, the sources for citation counts were determined after inspecting them on the shelves and prior to using them. A source was accepted if it covered either many sports categories or individual but otherwise neglected categories and if the athletes' names were easily retrievable by alphabetical lists or indexes. A sufficient number of names provided by a source was another, though subordinate, selection factor, since after all the numbers of hits could only be roughly predicted. So it happened that sources providing only small numbers of athletes were accepted for the standard reference sample. By applying the 18 standard sources to new samples, such as the CFEPP's, it may well occur that a source may only yield 1, 1, 2, 4, 7, 9 ... hits, (that's what we find at the lower end of the hit distribution). The most voluminous 5 or 6 sources would possibly have sufficed to obtain the eminence trend that was obtained from all 18 sources. The methodological rule, however, not to alter one's tools and rules of measurement, if not necessary, is mandatory. Dr. Nienhuys raised some more points. I regard them as frazzles and better stop here. I am rather confident that readers of his critique having been swaying to become suspicious of my studies will have regained trust. The present paper may demonstrate how easy it is to misdirect readers by exploiting their lack of familiarity with procedural details and how much effort is required to restore a framework for better judgments. I am quite willing to address, on other occasions, remaining problems and limitations etc. connected with citation counts and to improve the discriminate efficiency of this useful procedure. Final comments on the CFEPP's purported victory ############################################### In 1991, Dr. Benski informed Gauquelin that the CFEPP had completed their data collection. According to the 1982 protocol, Dr. Benski should have given Gauquelin the chance to approve of their sample in advance. If Gauquelin had asked me at that moment what to do I would have advised him to protest in public against this breach of the agreement and not to try to correct the CFEPP's biased sample himself. The question of whether and how it was to be corrected should have become a topic on an international forum. After nearly twenty-three years of such procedural diversions by three different skeptic committees, he was certainly entitled to such consideration. What Gauquelin actually did, unfortunately, played into the CFEPP's tendentious strategy. The CFEPP knew that their own sample was biased, because they had peeped at Mars sector positions of athletes that Gauquelin had obtained in earlier studies (information from Nienhuys; the overlap was 63%). They had readily accepted less successful sportsmen and missed many high achievers (see my eminence counts published in 1995/96). Gauquelin, after inspecting the CFEPP's sample, was fully justified to suggest the addition of many outstanding champions, many of them with Mars G-sector positions, and to exclude low success champions of which much less had Mars G-sector positions. To repeat once again, the protocol in fact required he be allowed to offer such comments before the actual birth time data was collected. Obviously frustrated at this treatment, Gauquelin suggested too many of these cases as someone who feared that the CFEPP would at best comply with only a small part of his suggestions. What the CFEPP eventually did was much more than what Gauquelin had feared: The CFEPP did not consider any of his suggested corrections, their sample was "frozen" (Nienhuys) and Gauquelin's suggested corrections were maliciously turned against him. Gauquelin's admittedly overdrawn and naive attempt to correct the severely biased CFEPP sample was trumpeted forth as clear evidence that Gauquelin's entire former research on planetary effects was biased. The effect of this propaganda is twofold, since not only does it seem to discredit all Gauquelin effects ever reported, but it also diverts attention from the CFEPP's study and the deficiencies to which critical readers should be directing their attention in the first place. How deficient was the CFEPP's data base? ---------------------------------------- As already indicated above, the CFEPP had not collected their data blind regarding Mars sector positions. This was urgent in regard to the fallacy of the preceding CSICOP experiment and it was a protocol prescription. Understandably, none of the CFEPP's own papers refers to their early Mars sector explorations. This fact alone would suffice to invalidate their entire results. In addition: - The CFEPP overlooked "ten per cent" (about 107 of 1,076) of their initial sample and did not send out birth time requests to Town Halls (Nienhuys). No reasons are given. - The CFEPP overlooked 71 athletes for which they had successfully requested birth times (Nienhuys). Nobody knows why this data was overlooked (observed G% 19.7%, expected 17.5%). - Five percent (about 53 of 1,066) of the CFEPP's selections are not in order regarding criteria of success (Nienhuys). - The CFEPP overlooked 46 Town Hall corrections of birth dates (Nienhuys), so Mars sector positions were calculated with 46 of 1,066 birth dates wrong. - The CFEPP excluded 13 French athletes not born in France (arbitrarily, as this was not prescribed by the protocol), 46% of them had Mars in G sectors, while chance expectancy was 17.5%. Mars sector positions of athletes not born in France were known from Gauquelin's published results. - The CFEPP excluded 13 Pelote Basque players (arbitrarily, as this was not prescribed by the protocol). 40% of them have Mars in G-sectors, something easily checked with the published Gauquelin data which the CFEPP consulted. - 13 cases, collected in 1982, were omitted in 1990 (Nienhuys). The CFEPP did not consider, for 11 athletes, time corrections in occupied zones during the war (Nienhuys). - 9 dates are wrong due to the CFEPP's not copying them from the registration forms correctly (Nienhuys). - For 5 omitted cases in 1994 (N=1,066), who were included in the 1990 sample (N=1,071), the CFEPP did not give any explanation whatsoever (Nienhuys). - 4 cases have wrong date/time combinations due to the CFEPP's negligence (Nienhuys). In sum, about 222 athletes were unjustifiably omitted, 53 cases should have been discarded (insufficient eminence), and copying and similar errors occurred for about 70 cases. It is a remarkable fact that these deficiencies of the data base were insufficient to make the CFEPP thoroughly correct their sample and to run a new analysis. But even more remarkable is the fact that despite all these maneuvers we found that the Mars effect was still significant. Who is responsible for the CFEPP's disaster? Dr. Nienhuys' position was strong, as he had immediate access to the CFEPP's data and he became familiar with it, even traveling to Paris in 1994 to check it. But Dr. Nienhuys was not independent. Even though he did not answer to the CFEPP, a group which apparently expired around 1992, he did in fact take his cues from the chairman of CSICOP, who has acted as promoter and pacemaker of the CFEPP study since 1982, and who has made the final decisions, including the decision to provide me with the CFEPP's final data as I did not yield to initial refusal. Obviously, Dr. Nienhuys noted many symptoms of the CFEPP's illness, but he was not free to diagnose nor to cure the patient. So in his Comments to Benski's book he scattered the symptoms around belittling them, appending "extras" for future studies, actually "regular" (not extra) data that the CFEPP should have analysed etc.. Fortunately, for slow and careful readers Dr. Nienhuys' scrutiny proves to be extremely efficient, and this should be greatly acknowledged. I am closing with two requests, addressed to CSICOP's Chairman: --------------------------------------------------------------- 1. I attempted to obtain from Dr. Benski, the author of the just published Prometheus book "The Mars effect", copies of documents showing the CFEPP's expectancy calculations ". My letter (with cc's to CSICOP's chairman and to Dr. Nienhuys) was not replied. This is inconsistent with Dr. Benski's promise in his book: "Whoever wishes to reproduce the calculations or verify the sample of athletes may contact us. We shall make all records and information available to interested researchers." Dr. Benski may delegate correspondence and postings to someone else. I would appreciate helpful intervention by CSICOP's Chairman. 2. I would also appreciate his support for my request to Dr. Benski, to refute, if possible, the following preliminary contentions by providing convincing documents: In 1990, prior to contacting Michel Gauquelin, the CFEPP did not only have collected their birth times, they had even completed all data analysis. In their methodological report "Verification de l'effet Mars" (1990) they withheld this information. Between 1990 (Verification...) and 1996 (Benski's book) nothing was changed, neither the CFEPP's sample (according to Nienhuys) nor its analysis. In other words, the CFEPP could have published their 1996 results five years earlier. Their waiting that long reduced the probability that an attentive public would protest against the CFEPP's stripping Gauquelin of his role as observer/supervisor and against their strategically exploiting his recommendations of sample improvement as a means of demolishing his reputation. The grounds for my suspicion: (1) In none of their publications did the CFEPP refer to their peeping at Mars sectors in 1982, which shows that they were able to suppress most important information. (2) The CFEPP's detailed account in their 1990 report of their method indicates that they commanded the logistic of data analysis in every respect. Can anyone imagine that they calculated Mars sector positions of randomized data sets only (that is what they admittedly did), postponing for 4-5 years calculation of Mars sectors of the original, unrandomized data set from which the long-expected final result was obtainable? But I am prepared to retract these propositions as soon as appropriate documents would prove that they don't hold. References Abell, George, Kurtz, Paul and Zelen, Marvin (1983). The Abell-Kurtz-Zelen 'Mars Effect' experiments: A reappraisal. The Skeptical Inquirer, Spring 1983, 77-82. Benski, C., Caudron, D., Galifret, Y., Krivine, J.-P., Pecker, J.-C., Rouze, M., & Schatzman, E. (1996). The "Mars effect": A French test of over 1000 sports champions. Amherst, N.Y.: Prometheus. CFEPP (1982, Octobre). L'effet Mars est-il r,el? Science et Vie (pp. 44, 46). CFEPP (1990). Verification de l'effet Mars. Etat de l'experience au 20 Juin 1990 (mimeographed). Ertel, S. (1998). Raising the hurdle for the athletes' Mars effect: Association covaries with eminence. Journal of Scientific Exploration 2 (1), 53-82. Ertel, S. (1995). Mars effect uncovered in French skeptics' data. Correlation 13(2), Winter 1994/95. Ertel, S. & Irving, K. (1996). The Tenacious Mars effect. London: Urania Trust.