The Gauquelin work
Abstract -- Many astrologers see Gauquelin planetary effects as the best factual evidence not only for astrology but also for the limitations of science. Scientists tend to be divided, seeing planetary effects as variously explicable, inexplicable, real, or an artifact. But the real challenge is not so much planetary effects as their associated puzzles for both astrology and science. The puzzles seem totally and utterly baffling. Can artifacts explain them? Here an artifact is something other than astrology that would result in Gauquelin planetary effects. Planetary effects have tiny effect sizes, and it does not need much of an artifact to produce a tiny effect size. Three types of artifact are identified -- (1) artifacts of natural cycles such as astronomy (Mars spends more time conjunct Sun than opposite Sun) and demography (birth rate varies with time and place); (2) artifacts of procedure such as selection bias (use only data that work) and improper statistics (shoot enough arrows and some are bound to hit); and (3) artifacts of social behaviour such as faking by parents (adjust birth data to match a desired astrology) and self-attribution by subjects (adjust behaviour to match astrology). Artifacts type (1) and (2) do not explain most planetary puzzles and planetary effects persist when the artifacts are controlled. So these artifacts remain artifacts, not an explanation of planetary effects. Artifacts type (3) easily explain every puzzle including the apparent disappearance of heredity effects. The probability of this occurring by chance is negligible. Nevertheless social artifacts are still artifacts, not an explanation of planetary effects. They become an explanation only if planetary effects disappear when the artifacts are controlled, as when births are reported by doctors instead of by parents, and the child is ignorant of its birth planets. As yet nobody knows if they do or not, although skeptics might argue that Gauquelin's failure to find planetary and heredity effects in recent births has already put this point to the test. Social artifacts clearly need to be controlled before researchers can proceed further. Includes sources for readers wanting more information.
Part 1 gave a concise illustrated history of the work and findings of the Gauquelins based on their many books and articles. Part 2 takes up the crucial topics of artifacts and how the baffling puzzles created by Gauquelin planetary effects might be explained. As before, "Gauquelin" means Michel Gauquelin (1928-1991) and "the Gauquelins" means Michel and Francoise (1929-2007), his Swiss-born wife and co-worker until 1985. Where source details are incomplete the details will be found in Part 1.
Gauquelin's work simultaneously discredited astrology and replaced it with weak planetary effects of no practical use. Despite their tiny effect sizes, planetary effects attracted a huge controversy about whether they were real and (if real) what caused them. In effect there was total disagreement on whether planets could affect people. But planetary effects create far more intriguing puzzles than this, and it is these puzzles that are the real challenge of the Gauquelin work and the real legacy of his astonishing labour. Which is why explanations that address only planetary effects and not the puzzles, such as pineal chemistry (Frank McGillion, Blinded by Starlight, Xlibris 2002), reincarnation (Peter Roberts, Message of Astrology, Aquarian 1990), and magnetospheric resonance (Percy Seymour, Scientific Proof of Astrology, Quantum 2004), are premature and incomplete. In what follows I look at artifacts in the Gauquelin work and their relevance to the baffling puzzles. First some opinions.
Opinions by astrologers
In short, these astrologers hold that Gauquelin planetary effects are a primary truth, a major scientific achievement, without mechanistic explanation, uncommon, conclusive, sound, strong, revolutionary, challenging to the prevailing world-view, and stubbornly not ignorable.
Opinions by scientists
In short, scientists have been divided. In their view Gauquelin planetary effects are variously inexplicable, worthy of serious attention, replicable, questionable, significant, implausible, unacceptable, real, unworthy of serious attention, and explicable in causal terms. Against these conflicting opinions we might bear in mind that our knowledge of planetary effects is not exhaustive, so we cannot know what fresh investigations might bring, especially when few of the above scientists have actually made fresh investigations other than from their armchair. And fresh investigations take time.
For example the Cottingley fairies stood for over 60 years before they crumbled. It took 40 years for the Fox sisters to admit their communications with the spirit world were a schoolgirl prank. Both Piltdown Man and Soal's "watertight evidence" for psi were found to be fraudulent only some 40 years later. It took nearly 60 years to learn that the famous 1919 spirit photograph taken by Dr d'Aute Hooper, and the famous 1934 photograph of the Loch Ness monster, were fakes. It has taken more than 60 years to discover that the sheep-goat and decline effects usually seen as support for ESP in card guessing are due to the effect of personality variables on sequential guessing behaviour that either tends to match (sheep) or not match (goats) the limited probabilistic runs possible in a deck of 25 Zener cards. Even when more straightforward claims attract wide investigation, sorting them out typically takes around ten years, whether positive (eg bacterial causes of ulcers) or negative (eg polywater). In what follows we will see how the above opinions about planetary effects, whether from astrologers or scientists, might be premature.
Concern for artifacts
In his article in Journal of the American Statistical Association 98, 697-698, 1987, the statistician I.J.Good describes how artifacts might play a role in the Gauquelin results as follows (in fact his idea is now known to be invalid, but it illustrates the point): "There is a correlation between month of birth and social class at least in Britain (Smithers 1984), and the top professionals might tend to have wealthy parents. If there is also a correlation between wealth of parents and reported hour of birth and a further correlation between hour of day and whether a planet is in key sectors, then one has the beginning of an explanation of a spurious correlation [ie] artifact" (p.698).
Now for a crucial point. Gauquelin planetary effects have tiny effect sizes, and it does not need much of an artifact to produce a tiny effect size. But the sample sizes required to reliably measure such tiny effect sizes are often larger than even the Gauquelin data can provide. For example the typical Gauquelin sample size of N=2000 will reliably detect an effect size of around 0.07 (nearly twice the typical effect size), but to reliably detect 0.02 (half the typical effect size) requires N=25,000. In effect we tend to be faced with detecting whispers above the roar of city traffic. If we are to proceed at all, we must attend less to individual outcomes and more to the overall picture. What matters is the consistent explanation of planetary puzzles, and whether planetary effects persist when artifacts are controlled.
The Gauquelins and their critics were generally concerned about two types of artifact, namely (1) artifacts of natural cycles such as astronomic effects (Mars spends more time conjunct Sun than opposite Sun) and demographic effects (birth rate varies with time and place); and (2) artifacts of procedure such as selection bias (use only data that work) and improper statistics (shoot enough arrows and some are bound to hit). Today we can add (3) artifacts of social behaviour such as faking by parents (adjust birth data to match a desired astrology) and self-attribution by subjects (adjust behaviour to match astrology). The point is, each artifact has the potential to produce astrology-like effects that have nothing to do with astrology. These artifacts are explored in turn below.
Artifacts of natural cycles (astronomy, demography)
The Gauquelins had accommodated these points by an elegant combinatorial approach described in their 1956 Methodes book, summarised later in APP 3(1), 20-23, 1985, which uses differences in the sector positions of Sun and planet to simultaneously correct for astronomic and demographic effects in the actual sample. But critics took no notice. Furthermore, when Gauquelin's first book appeared in 1955, he had sent a copy with an invitation to verify its results to astronomers such as Paul Couderc and Bart Bok (both highly critical of astrology) and to the Belgian Comite Para (short for the Comite belge pour l'Investigation scientifique des Phenomenes reputes paranormaux), which had been founded in 1949 to investigate paranormal claims. But there was no reply.
When Gauquelin's second book appeared in 1960, he again approached the Comite Para, but they declined on the grounds that new birth data would be hard to find. In 1966 he tried again, pointing out that new data should not be hard to find in sport, simply because hundreds of sportsmen had risen to prominence since his own data collections and were now listed in biographical dictionaries. This time the Comite agreed to make an independent test. Two years later they sent him their initial results for 535 sports champions, which as predicted by Gauquelin showed a strong planetary effect (the planet was Mars, which is why it was later called the Mars effect). Eight years later the Comite published their official report in their journal Nouvelle Breves 43, 327-343, 1976. It said they were unable to find any error in the expected frequencies (their discussion occupied 11 pages), but on the suspicion that one existed they claimed the Mars effect was not proven.
The Comite Para report failed to mention they had run additional tests known as counter-experiments in which times of birth were shifted from one champion to the next for a total of ten tests. If the suspected error existed then all ten tests should show a Mars effect. But they did not. The suspected error did not exist. This result was suppressed by the Comite for six more years until 1982, when Gauquelin forced their hand by publishing the details in Zetetic Scholar (10:70). But the Comite continued to argue that the expectancies had not been properly established, therefore (despite its own evidence to the contrary) the Mars effect was unproven (Zetetic Scholar 10:66). That is, it was due to an unknown artifact.
But interest in the Mars effect was growing. In 1976 Marvin Zelen, professor of biostatistics at Harvard University, pointed out that non-champions born on the same day as champions would show a Mars effect if the effect really was an artifact of astronomy and demography. Gauquelin duly collected 16,756 non-champions born within three days and in the same place as 303 champions picked at random from the Gauquelin and Comite Para data (a minimum of 300 was required for the test to be statistically meaningful) and mailed off to Zelen the 16,756 replies from registry offices.
The analysis was performed by Zelen assisted by philosopher Paul Kurtz and astronomer George Abell, both strong disbelievers in the Mars effect. To their dismay the result clearly supported Gauquelin. It also confirmed that the Comite Para's unspecified error did not exist. In 1977 the Zelen team published the result and attempted to explain it away by pointing to fluctuations in the Mars effect that appeared when the 303 champions were sub-divided, which violated their own requirement that the sample be not less than 300, and was to be expected on purely statistical grounds anyway. Six years later, after pressure from their own critics, they eventually admitted that "Gauquelin adequately allowed for demographic and astronomical factors in predicting the expected distribution of Mars sectors for birth times in the general population" (Skeptical Inquirer Spring 1983:77-82). Gauquelin had been right all along.
Artifacts of procedure (bias, data selection, fraud)
(He also had the computer calculate which kind of sector gave the best results -- those based on the actual rise and set of each planet, which was the approach described in his popular books, or those based on a modified Placidus table of houses given in Methodes, which was the approach he actually used in practice, see Neo-Astrology 1991:111. The results showed that neither was consistently better than the other. Varying the sector size in steps of half a 36-sector showed that the key sector size in the 36-sector division gave the best results.)
Gauquelin was also aware of the problems of data selection and fraud, which is why he was careful always to use the entire contents of a biographical dictionary, and to publish in both English and French the data for every birth and every experiment. Sometimes the dictionary was in error or the registry office did not have the data, so gaps did exist. But all the data used was there in black and white for anyone to check. He also invited anyone to check his records, including the many thousands of letters received from registry offices, all of which were meticulously organised and freely accessible. In terms of openness he set standards well above those of his opponents.
Following the Comite Para scandal, independent skeptic tests of the Mars effect were made on 408 American data by CSICOP (begun 1977 published 1979) and on 1066 French data by CFEPP (begun 1982 published 1996). Ironically these two tests were no less scandalous than the Comite Para test and merely inflated the controversy they were designed to allay. For example the CFEPP and Gauquelin signed and published (in Science & Vie October 1982) an agreement on procedure that, among other things, said the CFEPP would keep Gauquelin informed of progress and that every step should receive his approval before proceeding. But over the years, despite his many queries, they kept silent. They broke the agreement.
In summary, the three skeptic groups took every opportunity to plant doubt, to ignore Gauquelin's efforts to control artifacts, and to question his integrity, while neglecting to observe agreed procedures. In general they behaved less like scientists and more like a secret society with an agenda. In 1982, after a careful evaluation of the arguments on both sides, the psychologists Hans Eysenck and David Nias (in Astrology Science or Superstition) concluded against the skeptic groups as follows:"far from showing the impartiality popularly associated with science, critics have gone out of their way to demonstrate bias, prejudice and hostility. ... We have come to the definite conclusion that the critics often behaved in an irrational and scientifically unusual manner, violating principles they themselves have laid down, failing to adhere to their own rules, failing to consult the Gauquelins on details of tests to be carried out, or failing to inform them of vital points of the results. We have not found any similar misdemeanour on the part of the Gauquelins, who seem to have behaved throughout in a calm, rational and scientifically acceptable manner, meeting criticism by appropriate re-analysis of the data, by the collection of new data, however laborious the process might have been, and by rational argument. We do not feel that the 'scientific' community emerges with any great credit from these encounters" (p.202).
Ultimately the documentation involved in the three skeptic scandals occupied an incredible ten feet of Gauquelin's shelf space (Neo-Astrology 1991:36). Details can be found in Gauquelin's Truth about Astrology (1983:97-114), and at length in Ertel and Irving's Tenacious Mars Effect (1996), which includes a 26-page chronology by Jim Lippard of events and publications up to 1995.
But ultimately the skeptic tests are of no consequence, simply because (1) they proceed as if planetary effects other than the Mars effect did not exist, and (2) there are now enough independent replications to show that Gauquelin's findings are much as he specified them more than 40 years ago in his first two books. In short, planetary effect sizes remain tiny but they cannot be explained by artifacts of astronomy, demography, bias, data selection, or fraud. The puzzles remain. Now for artifacts of social behaviour.
Artifacts of social behaviour (faking, self-attribution)
To set the context you should be aware that the Gauquelin births are not recent (85% of professionals and 65% of parents were born more than 100 years ago), so modern lifestyles and ways of thinking do not apply. Most of the Gauquelin births occurred during the nineteenth and early twentieth centuries in Western Europe, when living conditions were very different from those today, and when early world views still survived, see Figure 1. It was a time when births were reported verbally to the registry office by the parents, when occupations and eminence tended to run in families, and when some popular almanacs gave the rising, setting and sometimes culminating times of the visible planets. Notice how the last immediately suggests a provocative match to the Gauquelin findings.
The Gauquelins' Profession-Heredity. Results of Series A & B (LERRCP, Paris 1972) is a thick US-quarto-size book of 211 pages in their own typing. Parallel texts in English and French give full details of their planetary and heredity results up to that time. In 1979 in New York you could buy a copy for $US12. Page 50 gives the numbers of eminent professionals born in each hour for each of ten professions. Inspection reveals that the proportion born at midnight varies greatly from one profession to the next. Gauquelin never mentioned this variation, so could it mean something? To find out, I plotted the largest planetary effect size for each professional group against the proportion born at midnight. Here the effect size is given by (O-E)/(N-E), technically known as Cohen's kappa, where O and E are the Observed and Expected frequencies in key sectors and N is the sample size. Thus for 1095 scientists the largest planetary effect size is for Jupiter and is (148-180.4)/(1095-180.4) = -0.035, which is negative because in this case Jupiter shows a deficit in key sectors. Here only magnitude is relevant, so I ignored sign and took the effect size as 0.035.
Since there is no reason to suppose that midnight has anything to do with planetary effects, I expected the plot to look like nothing in particular. In fact it was extraordinary, see Figure 2. There are ten professionals groups plus Mm = military musicians:
Figure 2 shows that planetary effect sizes tend to increase as the proportion born at midnight decreases. Despite the statistical uncertainty in measuring such tiny effect sizes, the correlation is remarkably strong and highly significant. Gauquelin never mentioned this correlation, yet it had been in his data all along. Could there be something special about midnight? To find out, I plotted the number of professional births in each hour as in Figure 3.
Figure 3 shows that midnight tends to be dramatically avoided. In fact such avoidance appears in many sets of older birth records, so at first sight there is nothing special about the Gauquelin data. The usual explanation is that registrars rounded times away from midnight to avoid ambiguity in their records (days begin and end at midnight, so to which end of which day does a given midnight belong?). But Figure 3 shows that births at 1-59 minutes past midnight also tend to be avoided, even though a similar ambiguity does not apply. Which leads to an interesting question: Could an additional factor be involved in this avoidance of the whole midnight hour? For clues I looked at the number of professionals born on each day, and found more surprises as shown in Figure 4.
Figure 4 shows that, compared to the day before and the day after, more professionals are born on days that European folklore says are desirable such as the 7th, and fewer on undesirable days such as the 13th. The difference between the target day and the adjacent days averages 12% of their mean births per day, so in broad terms a 12% peak will be produced if 4% of day-before births and 4% of day-after births are moved to the target day, or 1 in 25 of the births on those days. Similarly a 12% trough will be produced if 4% of target births are moved to the day before and 4% to the day after, or 2 in 25 of the births on that day. This might suggest there was more pressure to avoid an undesirable day than to select a desirable day. Or it might merely reflect the arithmetic -- it was easier to avoid an undesirable day (choice of two alternatives) than select a desirable day (choice of one alternative).
In such work it is of course important to avoid using occasions where folklore is inconsistent, for example midday was held to be desirable in Quercy but undesirable in Alsace (Gelis, History of Childbirth 1991:198). Here we know that the days shown in Figure 4 are consistent because on average each result is in the same direction for 9 out of 10 professions, never less than 8 out of 10, whereas if there was no consistency it would average 5 out of 10 purely by chance. The binomial p for each result ranges from 0.04 (8 out of 10) to 0.001 (10 out of 10), so they show the sought-for consistency mentioned earlier.
The above results are supported by Suitbert Ertel (Correlation 20(1), 30-36, 2001) who found that 884 French priests and 1506 Belgium Benedictine monks showed a 39% excess of births on Christian feast days. The difference is large and in the expected direction, but was dismissed by Ertel because it failed to reach significance, even though it was unrealistic to expect significance when the sample size on feast days was so small. His dismissal is a good example of how failing to consider consistency can lead us astray.
The results of Figure 4 are clearly not due to chance. They show that parents were faking (ie adjusting) birth data to suit prevailing beliefs, which in those days was easy. Thus in France after the French Revolution of 1792, and subsequently in the other Gauquelin countries, the father of each new child became legally responsible for registering its birth date and time at the nearest town hall. The father had to be accompanied by two friends to confirm that the child was his, and by the child itself. There was no medical certificate as is routine today, and most likely the two friends would not have witnessed the actual birth, so the father could easily adjust his report without detection.
In the case of witching days (witches sabbats) such as the dreaded Walpurgisnacht (30 April) the avoidance was understandable given the massive witch hunts that for three centuries had terrorized Western Europe. Furthermore, because the midnight hour was universally the witching hour when witches were proverbially active (for example in Shakespeare's Hamlet it was "the very witching time of night, when churchyards yawn and hell itself breathes out contagion to this world"), some people might have avoided reporting a midnight birth because it was connected with witches. This of course would be in addition to the registrar's efforts to avoid ambiguity.
Such fakings or adjustments (which in some early articles I called tamperings) are clearly a potential artifact in anything such as Gauquelin planetary effects that depends on birth records. But to be an actual artifact they need to be related in some way to planetary effects. Yes, parents are faking birth data, but do planets come into it? To find out we can apply tests based on the following consideration, which was kindly suggested to me by Suitbert Ertel:
Faking will be minimum on avoided occasions and maximum on preferred occasions, simply because there is no point in faking an occasion that nobody wants, and every point in faking an occasion that everybody wants. Avoided occasions are undesirable days such as the 13th, and avoided times such as midnight. Preferred occasions are desirable days such as the 7th, and adjusted times such as rounded birth times. In other words, if faking is related to planetary effects, we expect planetary effects to be smaller for births on undesirable days, for births at midnight, and for births with unrounded (ie unadjusted) birth times; and larger for births on desirable days, for births at hours other than midnight, and for births with rounded birth times. This expectation is tested in Figures 5 and 6. Here the planet or planets for each professional group are those showing the highest positive or negative effect sizes.
Figures 5 and 6 shows that in every case our expectation is confirmed. As faking and adjusting increase, planetary effects (whether positive or negative) get stronger. The results around midnight in Figure 6 are less reliable than the others due to small sample sizes, nevertheless the reversal in profile between positive and negative effects is striking evidence that midnight is related to planetary effects. We can check this by comparing professionals born with or without their relevant planet in key sectors (the former show planetary effects, the latter do not), so if planetary effects are related to midnight avoidance the former should have fewer midnight births (ie fewer unfaked births) than the latter. And they do: 49/3892 or 1.26% vs 185/12050 or 1.54%. In other words the answer to our key question is yes, planets do come into it. So we can conclude that an actual social artifact (not just a potential one) exists in Gauquelin planetary effects. It exists due to belief in the popular astrology of the time. More on the actual belief later.
Do not imagine that implausible levels of faking are needed to generate the Gauquelin results. In fact to generate the mean observed effect sizes given in Part 1 (0.043 for planetary effects or 0.016 for heredity effects) only 1 in 30 or 1 in 60 births need faking. Both levels are notably less than the average of 1 in 15 actually present in the day-faking of Figure 4, and are certainly no higher than today's level of faking among Asian parents (see next section). So they are not implausible. The same economy is true of Gauquelin's inadvertent bias in reading Placidus tables, where only 1 biassed reading in 750 is enough to create the observed small decrease in effect size between the hand-calculated and computer-calculated results. Of course, if faking occurs then the time reported by parents will differ from the actual time. This point is explored next.
Direct evidence for faking
The Gauquelin professionals are older and less urban than these Paris births, so for them the extent of differences could well be higher. Francoise Gauquelin considers differences between hospital and registry office to be inevitable when the registry office is receiving a birth report from a "declarant who had not even witnessed it" (APP 9(1), 34, 1993). Nevertheless the differences do suggest a lack of concern about accuracy that faking parents could use to advantage. Furthermore, if 1 in 7 of her fairly modern cases can involve an adjustment of 1 to 3 hours, or roughly one sector, then the 1 in 30 required to explain planetary effects for eminent professionals, or the 1 in 60 required to explain heredity effects, are not inconceivable.
Indeed, even today parents in the East still fake birth data due to belief in popular astrology. For example according to Chinese astrology women born in a fire-horse year will have unhappy marriages and are likely to murder their husbands. In the most recent fire-horse year (1966) Japanese births were a staggering 25% less than in adjoining years (Kaku, Annals of Human Biology 2, 111-115, 1975). Conversely an earth-dragon year is said to be auspicious for marriage, birth, and new business ventures. In the most recent earth-dragon year (1988) Chinese births in Malaysia showed an equally staggering 24.0% rise over the previous (earth-rabbit) year, while in Singapore the rise was 25.9% There was no effect among the non-Chinese population, which confirms its astrological origin (Goodkind, Population and Development Review 17, 663-686, 1991).
The above effects seems large because they relate to only a small part of the sample, namely those giving birth in that particular year. In terms of the whole population the corresponding effect sizes are about 0.04, closely similar to planetary effect sizes, which adds to the plausibility of a link between social artifacts and planetary effects. Indeed, when the event date is easier for people to select, the effect size can be much higher. For example in Vietnam the (50% of) years considered to be desirable by Chinese astrology attract 57.6% of marriages vs 50% expected, an effect size of 0.15 (Thang and Swenson, Journal of Biosocial Science 28, 367-377, 1996). This tells us that the power of astrological beliefs should not be underestimated. Further examples of faking are given later under "Why fake?".
The Gauquelins and social artifacts
"At the time of our previous work [with eminent professionals], it was objected several times to us that registry office information was not reliable enough for our statistical studies. There could be a family member whose birth was declared in a false way, either for reasons of convenience (children born the evening of December 31 are sometimes declared born the morning of January 1), or of lateness (if the birth is not declared within three days then an interminable procedure must be followed unless the registry official shifts the birth to bring it within the deadline), or of superstition (certain parents feel reluctant to declare a birth as having taken place on Friday, or on the 13th, or worse still on Friday the 13th! They try to avoid bad fate by advancing or moving back the date of birth). Such cases occur, of course. But we are surprised at the importance that some people attach to a false declaration when a hundred others are perfectly correct. One cannot prevent some errors being introduced into very large sets of data. What matters is whether the errors are numerous enough or large enough to cause fallacious results. Nowadays false declarations made at the registry office are extremely rare. For a rather long time already [he does not say how long], the doctor who notes the birth writes a certificate that must be presented to the registry office for recording the birth. Thus the family almost never has the possibility of changing the birth hour or date. Also the registry official would be naive to make concessions for lateness or superstition" (pp.170-171).
He then cites the 1959 study by Francoise that looked at 6372 birth records 1928-1957 from a large Paris hospital and 10,878 birth records 1860-1950 from registry offices near the hospital. Hospital birth times were recorded to the nearest five minutes but registry office times were sometimes rounded to the nearest hour or half hour. Nevertheless the birth time distributions were closely similar; repeat requests to registry offices showed that transcription errors were extremely rare; and (as already noted) comparison of about 100 registry office records 1928-1950 with the original hospital records showed that the average error in time was only one-eighth of the average time spent by a planet in a key sector. Furthermore her results were "based on the same documents as those used in the first heredity experiment. The validity of these [heredity] data is thus confirmed" (p.171).
He also notes: "The documents coming from provincial areas are in general less precise [than in Paris] and contain more birth times rounded to the hour. And often birth times of 0am [midnight] are reported as 11pm or 1am, but this does not make the record less usable statistically, provided the change ahead or behind is not too large" (p.171). This is because a planet generally spends 2-3 hours in a key sector. All of the above refers of course to his heredity studies.
For eminent professionals a similar discussion of social artifacts occurs in Les hommes et les astres (pp.169-171) but omits all mention of superstition, and even this reduced discussion is omitted from Gauquelin's later books in English. In other words Gauquelin was aware of the avoidance of midnight and of the 13th, and of superstitious effects in general, but felt they were quite unimportant. Ironically they now provide important clues about the strange behaviour of planetary and heredity effects. For example, as births are reported more and more by doctors and less and less by parents, there is a progressive reduction in the avoidance of midnight. I call this the "doctor effect", and its effect on births in Paris and in France as a whole is shown in Figure 7.
In Figure 7 the avoidance measure is derived from the mean count for 11pm and 1am minus the count for 0am, and is effectively 0% if 0am is not avoided and 100% if 0am is completely avoided. The Paris results are for the 24,948 family births gathered by the Gauquelins for their 1966 heredity experiment. The France results are for the 7952 French professionals gathered through 1960, with two additional data points for 1950 (French census N=834,893 cited by Francoise in her 1959 study p.684) and 1972 (a representative sample of 4330 natural French births from Journal of Interdisciplinary Cycle Research 10, 195-205, 1979).
Figure 7 shows that the level of midnight avoidance for Paris births starts to decrease around 1920 and tends to disappear after 1950. France as a whole follows about 10-15 years later, which suggests that the old reporting procedures tended to persist in provincial areas. For the Gauquelin professionals in other countries the decrease starts around the same time as in France, or a little later (it is hard to be sure due to small sample sizes in these later years).
Between 1800 and 1920 there is little change in midnight avoidance. But there is a four-fold increase (to 50%) in the proportion of birth times not rounded to the hour (Ertel, Correlation 9(1), 20, 1989), no doubt due to the increasing availability of accurate clocks. So the doctor effect seems unrelated to the precision of reporting. As we shall see, the doctor effect provides important clues about planetary connections. But first a look at Gauquelin's heredity findings.
The strongest heredity effect was for Venus, which echoed the Venus effects then recently discovered in eminent professionals, followed by Mars and the Moon, with Jupiter and Saturn equal last. This is almost the opposite of their importance for eminent professionals, and seems at first sight to be in keeping with what social effects might predict for families -- except the order of importance did not replicate in the two later experiments. As for faking, I found that the births show much the same faking as for professionals, except the preferred days are now family-related with a distinct gender influence. For example, children tended to be born on the same date, weekday or hour as their same-sex parent, and therefore on the same date, weekday or hour as their siblings, see Figure 8.
Figure 8 suggests that a small proportion of parents wanted more uniformity among their childrens' births than was allowed by nature. This may reflect the sort of traditions that led to particular weekdays being chosen for events such as weddings. Interestingly, among parents overall there was no obvious avoidance of witching days, at first sight explainable by the reduced rural emphasis in urban Paris. But closer inspection in terms of gender was revealing -- females avoided these days whereas males did not, which is consistent with authority figures in witch trials being entirely male and the accused being mostly female.
Gender influences also emerged in the heredity effects (ie planets in key sectors) passed on by parents to their children. Gauquelin had observed that the key sectors were generally interchangeable, that is, parents with a rising planet could have children with either a rising or culminating planet. In fact rise-rise and culmination-culmination links are slightly stronger than mixed links, which is compatible with the presence of social artifacts. Also compatible are the mean gender-specific effect sizes shown below, calculated as phi coefficients for 2x2 data. Here the sample size N's are large enough for the differences to be meaningful:
Father-Son 0.025 N=3888 Mother-Son 0.020 N=4455
The effect size between fathers and sons is six times larger than the effect size between fathers and daughters, which in those days made social sense. Or as Lady Catherine de Bourgh says in Jane Austen's Pride and Prejudice (1813), "daughters are never of much consequence to a father." But the differences tell us more than this. Gauquelin had seen parent-child effects as confirming the reality of planetary effects. He had found that fathers and mothers contributed roughly equal amounts of planetary effect to their children, which suggested a link with genetics. But he did not make the same-sex and opposite-sex comparisons shown above; or perhaps he made them but did not report them because their utility was obscured by his use of significance levels rather than effect sizes. As we have seen, the Gauquelins were aware of social effects but saw them as unimportant. However, genetics predicts equal contributions regardless of sex whereas social effects predict unequal contributions. So we have the classic situation of two rival hypotheses. In this case the winner is social effects.
Back to the doctor effect, the effect of births being reported more and more by doctors and less and less by parents. It seems reasonable to assume that opportunities for faking will follow the same pattern as the doctor effect in Figure 7. So if social artifacts are linked to heredity and planetary effects, both should tend to disappear in recent times as births are reported more and more by doctors and less and less by parents. Figure 9 shows how the heredity effect size does indeed decrease after the 1920s more or less in tune with the doctor effect shown in Figure 7.
Unlike the Paris-born families, the professionals were born across five countries, and here the doctor effect does not occur until after the 1930s. Alas, insufficient professionals were born in these years to allow a reliable test. But later data collected by Gauquelin showed that "the Mars effect in sports champions born after 1950 tends to disappear" (Truth about Astrology 1983:176), so he predicted "a disappearance of the Mars effect for athletes born in recent years" (Correlation 9(1), 28,1989), which of course is in tune with the doctor effect shown in Figure 7. More on this later in the sections starting with "Why fake?".
Figure 10 (left) shows that Ertel's students identified three broad clusters, which he labelled Fiction, Practice, and Knowledge, which I have re-labelled Art (musicians, painters, actors, writers), Action (sports champions, journalists, politicians, military men), and Science (physicians, scientists). At right is the cluster analysis using planetary effect sizes for the five Gauquelin planets; it shows the same three broad clusters, the only difference being that students located actors as shown. Even random data will give a tree, and for effect sizes as tiny as these there will be statistical uncertainties anyway, so what matters is the number and content of the broad clusters, not the order of components within them. In each case the Action and Science clusters eventually join, leaving Art notably independent from the rest, which independence is in agreement with the results of research for the Strong Vocational Interest Blank described in Appendix 2 of Part 1. Ertel concluded that planetary effects are not haphazard but conform to human views.
In our own test we want to know if the student's verdict matches the verdict produced by social adjustments. The results of the relevant cluster analyses are shown in Figure 11.
Figure 11 (left) shows the same student clusters as in Figure 10. At right is the cluster analysis of each professional group's preferences for 55 desirable days each year taken from Figure 4, namely Christian feast days, 1st-3rd-7th of month, equinoxes and solstices. If the faking to achieve desirable birthdays is unrelated to profession, we expect to see clusters at right that are haphazard in number and content. In fact they form the same three broad clusters as at left, the only difference being that students located journalists as shown. And Action and Science eventually join as before.
More to the point, the desirable-day clusters in Figure 11 are similar to the planetary clusters in Figure 10. Furthermore, cluster analysis of the avoidance of undesirable days gives the same three broad clusters and much the same result. We therefore conclude that planetary effects are not haphazard but conform to human social adjustments. Which confirms the earlier conclusion that an actual social artifact exists in Gauquelin planetary effects.
Not every almanac gave such information but those that did were enough for the purpose. Thus of the 56 European, UK and US 19th century almanacs I was able to examine, mostly one issue per title, almost all showed the sign the Moon was in for bloodletting, taking medicine etc, and most showed aspects between the Moon and planets for weather forecasting. All showed the rising and setting times of the Sun and Moon. More than one third listed the rising, setting, and sometimes culmination times of the visible planets, and another third gave information on planetary visibility during the year. The rest gave no direct planetary information, but positions could of course be inferred from the daily listing of Moon-planet aspects, especially conjunctions.
But how could parents know how to fake, given that astrology says that key sectors (= cadent houses) are weak, not strong? This too used to be a standard argument against social artifacts. To solve this one you need to look at what everyone was reading, namely almanacs, not at what few people were reading, namely astrology books (in Europe serious astrology had been effectively dead since 1700).
The most popular European almanac ever was Le Compost, short for Le Grand Calendrier et Compost des Bergers (The Big Calendar and Compilation of Shepherds), generally running to more than 150 quarto pages of which one-fifth was devoted to popular astrology. Here shepherd means the Good Shepherd and every man who works. It was first published in France in 1491, and other than the calendar part it was being printed largely unchanged more than 300 years later with equally long-lasting translations in other countries. In her survey of French almanacs, Genevieve Bolleme says "Le Compost was not only the model of almanacs but also the best expression of the popular mentality ... the level that the other [almanac] texts tried to approach" (Les almanachs populaires aux XVIIe et XVIIIe siecles, Mouton, Paris 1969:41). Relevant excerpts from the first edition (which are repeated in later editions) are shown in Figure 13.
Figure 13 shows how closely Le Compost matches Gauquelin's planetary findings. What people were reading closely matched what Gauquelin observed. Even so, they did not need Le Compost to prompt their interest in rising and culmination. Rising echoed the rising of the sun, hence earthly and heavenly greatness (light is a frequent biblical metaphor for Jesus). Culmination signified something at its highest power, just as "the culmination of our efforts" does today. A planet exactly on the horizon was generally invisible and was therefore without power. Thus in 1657 Placidus (of house division fame) argued that "the stars, where they do not rise, are inactive" (Primum Mobile 1.4, Cooper translation). And just as moving past the rising point brought planets into prominence, so did moving past the culminating point. Conflicts of analogy would be out.
As for the alleged planetary influences, Gauquelin shows in Neo-Astrology how they have stayed much the same from Babylonian times, a quite remarkable longevity. Each is a dramatic vignette of easily recognisable vice or virtue, so it is unsurprising that planetary meanings (but not sign meanings) have passed into everyday speech, as in solitary, lunatic, mercurial, venereal, martial, jovial, and saturnine. According to the Oxford English Dictionary, these words all date from 1500 or earlier, at least in English. The same vignettes appear in Le Compost. For example the man born under Mars shall be "Du tout s'adonne a faire guerre ou un grand chemineur par terre" (always addicted to battles or a great goer by the earth), which would seem to make it the planet of choice for a family with traditions in military service or competitive sport.
Recall that Gauquelin had found Mars prominent at the birth of military men and sports champions, Jupiter prominent for actors, Saturn for scientists, and the Moon for creative writers. He had also noticed that most military men and sports champions are martial, most actors are jovial, most scientists are saturnine, and most creative writers are moonish (his word for day-dreaming). All of which led him to conclude that planetary influences were real. As mentioned earlier, he was aware of superstitious beliefs but felt they would merely blur planetary effects rather than create them. He never mentioned 19th century almanacs during our many discussions, nor do I remember seeing any in his library. They are mentioned briefly in some of his books, for example in Astrology and Science he notes how the invention of printing allowed "a popular and practical form" of astrology to be "spread in the form of almanacs", which he then dismisses as "produced by people who knew little for people who knew nothing" (1970:109). In Neo-Astrology he mentions almanacs only as a survival channel for popular astrology during the 18th century (1991:112).
Similarly, if we can fake an auspicious birth date, or a planetary indication of greatness in a chosen occupation, it could have useful consequences. Being suitably destined in the eyes of the child and others has advantages. Our own beliefs do not come into it. The same motivation exists today when hotels omit 13 from floor and room numbers lest their occupancy be affected (the same applies in Japan where the unlucky number is 4 rather than 13), and when psychologists control for the expectations of experimenters.
Indeed, even in modern data the effects of faking are often clearly visible. For example 101,971 respondents replying to an ad in five US women's magazines in 1985-86 showed 83% fewer births than expected on 1 January, most likely to gain tax advantages made possible by tax authorities making no cross check with birth certificates, and 27% more births than expected on 14 February (St Valentines Day). The largest deficit (5%) of any date was on the 13th (APP 10(1), 26-29, 1994). Individual cases are harder to find because faking birth data, like tax evasion, seems unlikely to be admitted in public or in autobiographies. Nevertheless the birth of eminent mystery writer Georges Simenon was reported as a day earlier than its actual Friday 13 February 1903 because his mother considered it to be "too hard a fate for her sweet newborn baby" (F Gauquelin APP 9(2), 29-33, 1993).
Other social artifacts
The first is self-attribution, seeing ourselves as having the qualities that something (here astrology) tells us to expect. Self-attribution would be relevant wherever almanacs and popular astrology were common, which seems to have been most places. The presence of faking in the family data (see the earlier section "Heredity effects") indicates that at least some people knew their own birth planets. If they also knew what the planets meant, which seems likely (otherwise why bother with birth planets?), they might adjust their behaviour (or at least their self-image) to suit, just as people today are known to do for sun signs. Self-attribution is not needed to explain eminence effects simply because eminence tended to run in families, so children would tend to have it already.
But could self-attribution generate a planetary effect size of 0.04? For sun signs the effect size vs personality is typically 0.09, and the effect size vs job choice is possibly around 0.02. So a role for self-attribution in planetary effects is not inconceivable. We could deny it only if planetary effects persisted when planetary meanings were not known, but such data is not available for the Gauquelin professionals as a whole, although hints may appear in a few individual biographies. In short, enough self-attribution could exist in planetary effects to be a serious artifact but we cannot know for sure.
The second potential social artifact might arise wherever social beliefs were strong. Thus if social beliefs were strong enough to persuade parents to fake, they might also be strong enough to favour children who by chance had the right birth data, for example by helping them take up the relevant profession. The effect might be tiny, but so are planetary effects. More importantly, the stronger the social beliefs, the more the faking, and also the greater the help for unfaked births. In short, the level of social artifacts in the Gauquelin data could exceed the level of visible faking, but again we cannot know for sure.
All we can say is that these other social artifacts are not implausible. If nothing else, they warn us not to dismiss any hints of faking in the Gauquelin data, for such hints (and there are plenty of them) could be the tip of an iceberg. Now for those baffling planetary puzzles.
Can planetary puzzles be explained by artifacts
Can artifact explain puzzle?
Eleven puzzles for science Cycles Proc Social behaviour
The above table shows how artifacts of natural cycles (astronomy and demography) are unable to explain most of the puzzles. Furthermore, when these artifacts are controlled, planetary effects do not disappear. So these artifacts are artifacts, not an explanation of planetary effects.
Artifacts of procedure (bias, data selection and fraud) might in principle explain almost every puzzle. But no matter how hard we try, it would seem impossible to pick or invent data that would simultaneously fit every puzzle all the time. It would be like trying to please all the people all the time. The diversity is simply too great. (This is not to say that such data could not exist, only that we could not directly pick or invent it, especially if like Gauquelin we had to calculate everything by hand.) Furthermore, when these artifacts are controlled, planetary effects do not disappear. So these artifacts are artifacts, not an explanation of planetary effects. The same applies to the theories of McGillion, Roberts, and Seymour -- they may in principle explain a broad as-above-so-below effect but they do not explain all the puzzles, so they cannot be an explanation of planetary effects.
The situation is rather different for artifacts of social behaviour (faking and self-attribution). As shown in the last column, these artifacts easily explain every puzzle simultaneously. For example, the supposed conflict between astrology and the Gauquelin findings is no more than the conflict between astrology and popular beliefs. Between popular beliefs and the Gauquelin findings there is only agreement. As for the items (1-11) referred from above:(1) Diurnal position was the popular belief. There is no effect for signs or aspects because in almanacs they were linked with seasons or weather. In any case the required adjustment (days or weeks) is too great to be feasible.
(2) The planetary link is with occupation and not character because that was the belief in those days. Planetary themes today tend to focus on character traits but in those days the focus was on occupation. First, it was a dominant part of family tradition. Thus a military family would want their new son to be a great soldier rather than, say, a stable extravert. Second, only occupation could feasibly be shown in woodcuts.
(3) Eminence is important because occupation effects will be strongest where family traditions are strongest, where the match between planets and occupation is closest, and where the need to be suitably destined is strongest, as in eminent families. The role of family traditions and eminence is confirmed by Francis Galton in his 1869 book Hereditary Genius, where he showed that great men produce more great sons than do average men. Based on the work of others, Gauquelin concluded the same: "In general terms, two-thirds of famous people originate from five percent of the population comprised primarily of the wealthiest and most intellectual people. A professional proclivity is transmitted from generation to generation" (Cosmic Influences 1976:60).
(4) Gauquelin's original finding was quite definite: more eminence, more planetary effect. When Ertel measured eminence by counting citations in reference books, the finding was generally confirmed. But in a few cases (eg Saturn for scientists) the effect first increased with increasing eminence and then decreased, even though other planets for the same profession showed no decrease -- a baffling puzzle indeed (NCGR Research Journal 27-31 Spring 1989). The Gauquelins noted that Ertel's samples tended to be small and inappropriate, and that this could explain the results (same issue 33-36). But so can social artifacts. Men of the very highest eminence, namely those who are geniuses, tend not to follow family traditions in the way that men of ordinary eminence do. For them the faked planet would tend to be the wrong planet, so links with the correct planet would tend to decrease, whereas links to planets that included the faked planet would show no decrease.
(5) There were numerous reasons why the Moon was more important for creative writers than Mercury, which is traditionally linked to communication rather than creativity. Thus the Moon rules feelings and watery ink, the planetary man in Le Compost shows it ruling the head (seat of the intellect), the Egyptian Moon god Thoth was the inventor of speech and writing, Aristotle said the brain was humid and therefore subject to the Moon (hence the term lunatic), and the Moon is highly visible whereas Mercury is usually invisible.
(6) Venus showed the strongest positive effect in the first heredity experiment but was among the weakest or most negative in the second and third experiments. But the other Gauquelin planets were equally variable in heredity, for example Saturn was among the weakest in the first two and strongest in the third. Such extreme variability was not observed for the professionals but is to be expected for families, where (unlike the professionals) there is no fixed planetary theme, so the order of planets is likely to vary from one set of data to another.
(7) As noted in Part 1 Figure 10 (transit of Venus), size is not a factor in astrology. Nor is distance. What matters is mythology, which is why Venus is seen as pleasant and harmonious despite having a permanent weather forecast of 450C under 90 atmospheres of carbon dioxide and a fog of sulphuric acid. Nor is Venus seen as changeable despite varying in brightness. Visibility is not important in astrology today but it was in the days of Le Compost when (obviously) only the visible planets were known. Even after Uranus (1781) and Neptune (1846) were discovered, 19th century almanacs tended to mention only the visible planets. Furthermore the finite speed of light means that Jupiter or Saturn are not visible on the horizon until they are actually about 10 or 20 degrees above it, yet their peak sector positions do not differ by these amounts from those of the other Gauquelin planets. What matters is the apparent position of visible planets, not their actual position or some other physical variable.
(10) Why do planetary effects disappear when the birth is induced or surgically assisted? Gauquelin had noted how "the Mars effect in sports champions born after 1950 tends to disappear" (Truth about Astrology 1983:176). He attributed this to medical intervention, which had upset the natural timing of birth and therefore (in his view) the natural planetary links. But as shown in Figure 9, planetary links between parents and children also disappeared around 1940-1950 even though there was no evidence of medical intervention, for which Gauquelin had no explanation (Correlation 5(1), 36-37, 1985). However, the doctor effect (births being reported more and more by doctors and less and less by parents) meant a progressive end to faking other than self-attribution, especially where intervention occurred since this would nearly always be in a hospital. Which would explain the decline in planetary effects for both sports champions and families. After all, the idea that faking is being prevented is more plausible than the idea that all hospitals and all midwives are intervening medically in all births. Also, the re-establishment of astrology in the 1930s meant a gradual end to popular beliefs and their match with the Gauquelin findings whether by faking or self-attribution. Had Gauquelin used effect sizes instead of significance levels, it is conceivable that he would have reached the same conclusion.
(11) Gauquelin found that high levels of geomagnetic activity at birth had no effect on the planetary links with profession but seemed to increase the planetary links with parents. Suitbert Ertel used a more sensitive approach and found no geomagnetic effect for either profession or heredity (Correlation 9(1), 5-23, 1989). Nevertheless a geomagnetic effect is not incompatible with social artifacts, for example it could conceivably influence the inclination to fake in the same way that it is alleged to influence heart attacks, traffic accidents, migraine severity, and so on.
Readers of Correlation 19(2), 20(1), 20(2), 21(1), 21(2), 23(1) will be aware that, according to nearly 90 pages of dense argument by Suitbert Ertel, with nearly 50 more pages on his website http://www.psych.uni-goettingen.de/home/ertel/, social artifacts cannot possibly exist. Nevertheless the close match between social artifacts and every one of those 22 puzzles should be pause for thought. Indeed, if we assume that the probability of scoring a yes purely by chance is 50%, the probability of scoring 22 yesses purely by chance is 0.522 or p=0.000,0002. In reality the probability of scoring a yes purely by chance is likely to be much less than 50%, so the probability of scoring 22 yesses purely by chance is likely to be much less even than p=0.000,0002. After all, if it was actually easy to explain those 22 puzzles, someone would have done so long ago.
Social artifacts are not an explanation
Ironically if social artifacts do turn out to be an explanation, it follows that, contrary to what most critics thought, planetary effects in historical data are to be expected, so their absence would be more surprising than their presence. Finally a bonus. The existence of artifacts of social behaviour in the Gauquelin data makes it a valuable new resource for sociologists studying the nineteenth century. It seems unlikely that this data, which took such heroic effort to collect, will ever be equalled. Examples of how quite complex beliefs and family relationships can be explored are given by me in the reference cited below (2000:33-37).
AcknowledgementsThe idea of searching for social artifacts came after talks between Michel Gauquelin, Suitbert Ertel, Rudolf Smit and myself in Utrecht in January 1991, four months before Michel's tragic death. At the time I never imagined the work would take eight years to complete, but such is the challenge of Gauquelin puzzles. It required endless visits to libraries and accumulating over 10 kg of photocopies. Of the specialised libraries in London other than the British Library, those of the Folklore Society, the Warburg Institute, and the Urania Trust were key resources. Since then a further year in total has been spent tidying loose ends and performing further tests.
Thanks are due to Peter Niehenke for the donation of DAV data; to Colin Miles for the donation of Astrocalc software for calculating planetary positions; to Rudolf Smit for donating various Gauquelin data books; and to the late Michel Gauquelin for donating since 1975 copies of his findings as they were published, and for many discussions and unfailing assistance during the fifteen years of our acquaintance.
For helpful comments or for help with obtaining material, thanks are due to Susan Blackmore, Christopher French, Arthur Mather, David Nias, Mike O'Neill, Peter Roberts, Wayne Spencer, Mike Startup (all UK); to Barry Beyerstein, Ivan Kelly (Canada); and to Francoise Gauquelin, Jacques Halbronn, Wout Heukelom, Francesco Maiello, Arno Muller, Rudolf Smit, and Robert van Gent (Europe). Especial thanks are due to Suitbert Ertel (Germany) for providing data and reference material, for untiring help with data cleaning, for critical comments, and for suggesting additional tests.
Finally thanks are due to Wout Heukelom for inviting my English invasion of his Dutch journal Astrologie in Onderzoek, renamed for the occasion Astrology under Scrutiny, in which the full technical details of social artifacts were first published with 300 references in Volume 13:1-72, November 2000. Includes a critique by Suitbert Ertel 73-84 and my response 85-87. Contact this website for information on how to obtain copies including a four-page update published May 2002 covering further tests, errata, and 14 inadvertently omitted references. An abridged and more readable version of the above is in Skeptical Inquirer May/June 2002:33-38, with a challenge by Ertel and my response in Skeptical Inquirer Jan/Feb 2003:57-59,65.