The Science of Astrology

Michel Gauquelin was born in Paris on 13 November 1928. As a schoolboy he was so taken with astrology that his classmates called him "Nostradamus" (as Geoffrey Dean, the best-known academic critic of astrology, recounts from Gauquelin's own books). With the confidence of a young scientist trained in statistics, he set out to put the claims of the astrologers to the test of large numbers. He began with the published statistical work of the French and Swiss astrologers of the generation before him, Paul Choisnard, Karl Ernst Krafft and Léon Lasson, and one by one their results failed under his checking. He then tested the signs, the aspects and the transits as the astrologers of his day used them, in large samples and as single factors. He found nothing in them that chance could not explain. He was writing a book to demolish astrology.

Something appeared in his tables that he had not been looking for. The positions of certain planets in their daily circle, rising over the eastern horizon or climbing to the highest point of the sky, were not spread at the births of eminent professionals as chance spreads them. In his first book, L'Influence des astres, published in Paris in 1955, he wrote that he had been "sincerely persuaded" that his work "would have nothing to offer except a critique of astrological doctrine … But in the course of our work we were confronted by results so remarkable that scientific rigour obliged us to pursue and extend them" (p. 12, as translated by Dean). One of my favorite moments in the modern history of our art is a young man with every reason to close the door on astrology, finding that the data would not let him. It is the experience Kepler confessed three and a half centuries earlier, a most unfailing experience that "instructed and compelled my unwilling belief" (Chapter 5, §5.7).

The popular retelling makes Gauquelin a determined enemy of astrology, converted against his will by his own data. He was a former enthusiast who had become a rigorous tester, and he tested everything the same way. The signs, the aspects, the transits and the published statistics of his predecessors all failed. Only the planetary effects survived. He did not work alone. His wife, Françoise Schneider-Gauquelin (1929–2007), shared the labor of the collection for decades. Between them they gathered from the civil registers of France and four neighboring countries the hours of birth of tens of thousands of people. The archive they left is now held openly by CURA as the Archives Gauquelin, about 149,000 birth and death records in all, including the large series they gathered to test planetary heredity. Within it stand about 16,000 eminent professionals of every kind: physicians and scientists, champions and soldiers, actors, politicians, writers, painters and musicians. We have downloaded the professional volumes ourselves and recomputed the central results from the raw dates, times and places. What follows uses the Gauquelins' own figures and our independent check of them.

Study card — The Gauquelin archive

21.1 The Method of Sectors

Every planet, like our Sun, rises in the east, climbs to its highest point in the south, sets in the west and passes beneath the Earth to its lowest point before rising again. This daily round is the diurnal circle, produced by the turning of the Earth. At any place and moment each planet stands somewhere on it: just risen, high in the sky, setting or below the horizon. Astrologers have always read this circle. The houses of a horoscope are its divisions, and the ascendant and the midheaven, the rising point and the culminating point, are its hinges. Gauquelin measured a planet's place on this circle in a way that respects the planet's own motion. The same measure could then be applied to every birth, in every season and at every latitude.

He divided each planet's diurnal circle into twelve sectors, or into thirty-six for finer work, and he did it by the planet's own time above and below the horizon. The time a planet spends between rising and setting, its diurnal semi-arc as the astronomers say, varies with the planet's declination and with the latitude of the birth. A planet may therefore be above the horizon for ten hours on one day and fourteen on another. Gauquelin cut the planet's time above the horizon into six equal parts and its time below into six more. He numbered them in the direction of the daily motion, beginning at the rising. Sectors 1 to 3 run from rising to upper culmination and sectors 4 to 6 from culmination to setting. Sectors 7 to 12 carry the planet through the hours below the Earth back to the east.

The finding lies in two of these twelve sectors. Gauquelin found that eminent members of certain professions were born with particular planets in sector 1, just after the planet had risen, or in sector 4, just after it had passed its highest point, more often than chance allows. These two became the key sectors of all later work. Two sectors of twelve suggest a chance rate of one in six, 16.7 percent, and the true chance rate is a little higher. Births are not spread evenly through the hours of the day or the months of the year, and each planet moves at its own pace through the zodiac. Gauquelin worked out the expected rate from the births themselves, by hour, by season and by place. For Mars among the champions he found it to be about 17.2 percent (17.17 percent in the figure Dennis Rawlins reported from Gauquelin's work).

That expected rate is the line against which we measure every result, and his opponents tested it themselves. Marvin Zelen, the Harvard statistician, with Paul Kurtz and George Abell, gathered a control group of 16,756 ordinary people born in the same towns within three days of a sample of champions. The ordinary births came out at 16.4 percent, just where Gauquelin had always said ordinary people would fall (Chapter 24, where the study card for this test stands). In 1983 the same opponents wrote that "Gauquelin was correct in his prediction that 17 percent of ordinary people would be born with Mars in key sectors" (Abell, Kurtz and Zelen, Skeptical Inquirer 7(3), p. 79). In the same article they declined to call the effect confirmed until it had been replicated independently (p. 82). Both sides accepted the base rate in print.

The hour of birth in the archive is the hour written into the civil register of the town where the child was born. A clerk entered it at the declaration of the birth, decades before anyone could know whether the child would become a champion, a general or a member of the Académie. No astrologer chose those hours, and no researcher could change them, since the registers stand in the town halls of Europe for anyone to check. In the nineteenth century the hour was reported to the clerk by the family, as Dean has stressed. Whether that reporting could have carried a bias is the main question a critic has put to these data. We take it up in the next chapter (Chapter 22, §22.5).

When we recomputed the archive we set our own expectation inside each sample. We kept every person's date and place of birth and gave them the hour of birth of another member of the same sample, chosen at random. We repeated the shuffle ten to twenty times. This keeps everything a skeptic could suspect of producing an artifact: the sample's own spread of birth hours, its seasons, its places and the astronomical cycles of its years. It destroys only one thing, the link between a particular person's moment of birth and the planet. The method is the counter-experiment the Belgian skeptics of the Comité Para themselves used. For Mars among the champions it gives an expectation of 17.2 to 17.5 percent, slightly above Gauquelin's own, so that our figures are, if anything, harder on the effect than his. A second control shifts each birth by two to ten years while keeping the month, day, hour and place. It gives 16.7 percent for the same sample, and we report the stricter of the two throughout.

The traditional houses are counted against the daily motion from the ascendant. A planet that has just risen has therefore passed above the ascendant into the twelfth house, and a planet that has just culminated has passed the midheaven into the ninth. Gauquelin's key sectors fall on the cadent sides of the angles, in the houses the tradition counts weakest. Ptolemy placed the strength of a planet at the rising and the culmination themselves, "in the orient and at the horoscope" and at the "culminations", and he made the planet "at mid-heaven" a lord of action (Tetrabiblos III.13 and IV.4, Robbins). Our tradition was right that a planet near rising or culmination is strong. It placed that strength a little earlier than the data place it, on the angular side of the line where the data find it on the cadent side. That is a partial contradiction of the tradition, recorded as a partial in our scorecard (Chapter 6, §6.6).

A parent shifting a reported birth time toward an astrologically favorable hour, or an astrologer selecting births to fit a theory, would have shifted them toward the angles the books praised, the ascendant and the midheaven themselves. The data instead peak just past those angles, in zones the books of Gauquelin's day called weak.

21.2 Discovery and Replication

The French data of 1955 were the discovery. In L'Influence des astres Gauquelin published the birth data of some 5,700 French professionals, as Dean counts them, and reported surpluses of Mars, Jupiter and Saturn in the key sectors of particular professions. In our view the physicians gave the first and cleanest result. In 576 members of the Académie de médecine, the most distinguished physicians of France, Mars and Saturn stood in the key sectors more often than chance allows. Gauquelin then gathered a new sample of 508 physicians and ran the test again. He found the same result, as Kenneth Irving reports it. That was the first replication, inside France and inside a single profession.

A discovery made in one country's records could still be an accident of that country, of its registry customs, its clerks or its population. So in the years after 1955 he went abroad. He gathered about 15,000 new births of eminent professionals from Germany, Italy, Belgium and the Netherlands, together with about 2,000 more from France. In 1960 he published, in Les Hommes et les astres, the result of putting his French findings to that new data. He had thirteen planet-and-profession findings from France to test, and he tested each one in the new samples. Dean lists them by profession: three planets for the physicians and scientists (Mars, Jupiter and Saturn), two for the champions (the Moon and Mars), three for the military men (the Moon, Mars and Jupiter), two for the politicians (the Moon and Jupiter), one for the actors (Jupiter) and two for the painters (Mars and Saturn). Those findings came from the French births, made before the foreign births were gathered. The test of 1960 was therefore a prediction checked against records its author had not yet seen when he made it. Beside them he set the same tests run on non-eminent members of the same professions. Dean, the best-known academic critic of astrology, reproduces the whole tally from p. 161 of the 1960 book.

Table 21.1 — The 1960 replication: thirteen planet–profession findings, sorted by how rarely chance would produce each (Gauquelin 1960, p. 161, as reproduced by Dean)

The column of other countries comes first. These are the German, Italian, Belgian and Dutch births that Gauquelin had not seen when he made his French findings, and so they are a true test of them. Of the thirteen findings, eight reached better than 1 in 100 in this new data, where chance would give 0.13, a little more than one eighth of a single result. Chance would be expected to produce one result as strong as these in about every eighth complete replication of the whole thirteen-test study. The other countries produced eight such results in a single replication, a pattern that chance would essentially never produce. We print no exact chance for it, because some of the thirteen tests share samples and are not fully independent. The combination of these samples is given below (§21.5). Across all the data together, France and abroad, eleven of the thirteen findings replicated at better than 1 in 100, and seven of them at better than 1 in 1,000.

The column of the non-eminent holds the main control in Gauquelin's work. These were members of the same professions, physicians, soldiers, actors and the rest, who had not reached the rank of eminence his main samples demanded. They were born in the same countries and the same era and entered in the same kinds of register. Ten of their thirteen results were no better than chance gives at 1 in 10, very nearly the 11.7 that chance would give. The table prints the three that did not: one between 1 in 20 and 1 in 10, one between 1 in 100 and 1 in 20 and one going beyond 1 in 1,000. Whatever the planetary effect is, it is absent in the ordinary members of a profession and present in its most distinguished members. That is the eminence effect in embryo, which Ertel later measured with an independent scale of fame (Chapter 22). It is fatal to any artifact that would act on every birth alike. A quirk of the astronomy, a habit of the registrars or a custom in the hours at which children were born or recorded would mark the ordinary physician as surely as the academician, and the data mark the academician.

Study card — The thirteen findings put to new countries. Michel Gauquelin, France, 1960 (Les Hommes et les astres, p. 161; the tally is reproduced by Geoffrey Dean, the best-known academic critic of astrology); he used about 15,000 new births of eminent professionals from Germany, Italy, Belgium and the Netherlands, copied from civil registers of the kind described above; thirteen findings made on French births before the foreign births were gathered, each tested on the new births, with a mean sample of 1,416 eminent and 1,639 non-eminent professionals; eight of the thirteen reached better than 1 in 100 abroad, where chance gives 0.13 of one, and eleven did across all the data, while ten of thirteen among the non-eminent stayed at chance; no single chance figure is printed because the thirteen tests share samples (the measured replication is carded in §21.5).

Gauquelin summed up the reach of his replication in 1960 in words Dean has preserved for us: "Our research has given the same fundamental results across five different countries, for births during 1800–1930, in latitudes of 37 to 55 degrees spanning more than 20 degrees of longitude, despite frontiers, differences in language, institutions, mentality, or history" (p. 165, as translated by Dean). It held across more than a century of births, from the age of Napoleon to the years after the First World War.

Our own recomputation shows the same travel in a single profession. Among the 1,093 French champions of the archive, Mars stands in the key sectors at 21.9 percent of births against an expectation of 17.2 percent. Among the 994 champions from other countries it stands at 20.1 percent against the same expectation. The French figure would arise by chance alone about 0.004 percent of the time, 1 in 23,000: strong. The figure from other countries would arise by chance alone about 0.8 percent of the time, 1 in 125: significant by the usual scientific line. It is a smaller excess in a smaller sample and in the same direction. As we teach our students, independent replications of a real effect tend to come in a little lower than the discovery. Dean's own tally of the later literature shows this pattern (Chapter 22, §22.4). (Evidence table A.21.)

21.3 Mars and the Champions

Of all the planet-and-profession findings, one became the emblem of the whole. Sports champions, the great cyclists, boxers, footballers, runners and fencers of Europe, are born with Mars just past its rising or just past its upper culmination more often than chance allows. The result has been tested, attacked and tested again for seventy years, and it still stands (Chapters 22 and 24).

Scientists ask how often chance alone would produce a result this strong, and the answer comes as a percentage and as odds. Science itself calls a result significant when luck alone would give it about 1 in 20 times or less. Medicine, psychology and nearly every science you have heard of publish and act on results at that line. We hold our results to the same line and then keep climbing, to 1 in 100, to 1 in 1,000, to 1 in 100,000 and on to the physicists' five sigma, their standard for announcing a discovery. As we count it, five sigma means a result that chance alone would produce about 1 in 1.7 million times, whichever direction it points, and 1 in 3.5 million if only one direction counts. A single result near 1 in 20 is a lead, and many results that agree are the evidence. Each result below carries its place on that ladder, and the Reading Guide gives the ladder.

The classic sample is Gauquelin's 2,088 European champions, published in 1970 as the first volume of his series of birth data. The figures were stated by an opponent. Dennis Rawlins, an astronomer on the first council of the skeptical committee that would later test the Mars effect itself, wrote in the Skeptical Inquirer that "in his analysis of 2,088 European sports champions, Michel Gauquelin found that Mars appeared in a specified pair of his 12 celestial 'sectors' (1 and 4) at the birth of about 22 percent (more exactly, 21.65 percent) of the sample, instead of the theoretical expectation of 17 percent (about 2 in 12; more exactly, 17.17 percent). The odds against this occurring by chance are some millions to one" (Rawlins 1979–80, Skeptical Inquirer 4(2), p. 26). In the same issue the three authors of the American test wrote that "in Gauquelin's sample of 2,088 champions, 21.6 percent are in the key sectors", and they put the true share at 21.6 percent give or take 1.8, so that the whole of their range lies above the expected 17.17 (Kurtz, Zelen and Abell 1979–80, p. 24).

From Rawlins's figure the number of champions follows, since 21.65 percent of 2,088 is 452. Neither source prints the integer, so we give it as derived: 452 champions born with Mars in the key sectors where chance would place about 358. We worked out the exact probability ourselves from those figures. A surplus this large would arise by chance alone about 0.000008 percent of the time, 1 in 12 million, which is "some millions to one" as Rawlins said. That is past five sigma.

Study card — Mars and the champions, the classic sample

The later collections brought the published athletes to 2,888, and among them 630 are born with Mars in the key sectors, 21.8 percent against 17.2 percent (Ertel and Irving's figures). We worked out the odds ourselves from the published counts. Chance alone would produce that excess about once in nine billion trials.

Study card — All the published athletes. Suitbert Ertel, professor of psychology at Göttingen, and Kenneth Irving, 1996 and after, both sympathetic to the effect, took Gauquelin's published athletes from all his volumes, 2,888 in number, and found 630 of them (21.8 percent) born with Mars in the key sectors against 17.2 percent expected. Chance alone would give that about 1 in 9 billion times, Rung 5, Beyond the discovery standard. This is the published, selected set, and Chapter 22 cards the complete pool.

Those are Gauquelin's counts. We reproduced them from the raw data. Using the CURA edition of the first volume, we computed for every champion the position of Mars in its daily circle from the published date, hour and place of birth. We took the planet's apparent position, its semi-arc and the refraction at the horizon into account. Among 2,087 champions (one duplicate removed), 439 are born with Mars in the key sectors, 21.0 percent, against an expectation of 17.2 percent from our time-shuffle control. Chance alone would produce this about 0.0003 percent of the time, 1 in 300,000. Against the year-shift control of 16.7 percent the chance is about 0.000014 percent, 1 in 7 million. Our 21.0 percent is a little below the 21.65 percent of Gauquelin's own count. The likely reasons are the corrections of the later edition, our use of the regional capitals the archive gives as places of birth and small differences of sector convention. Study card — Our recomputation of the champions

Across the twelve sectors our recomputation counts 229, 187, 164, 210, 158, 141, 158, 178, 178, 160, 162 and 162 champions, against an average of about 174 per sector. The two highest counts in the whole circle are sector 1, just after Mars has risen, at 229, and sector 4, just after it has culminated, at 210. The rising sector stands highest of all, as Ertel's plots of the larger pool show. Nothing else in the circle comes near them, and the peaks fall where Gauquelin's French discovery said they would fall.

Table 21.2 — Mars at the births of sports champions

We state these chances for the direction the French discovery fixed in advance, an excess of Mars in the key sectors, and every later sample tested that direction (Chapter 7, §7.2). If the direction is not counted as fixed, each chance roughly doubles. At these magnitudes the doubling changes the strength little, so we read each strength from the doubled figure. The champions' Mars stands at some millions to one in Gauquelin's classic sample and at billions to one across all his published athletes. It still stands at about 1 in 14,000 when every athlete he had set aside is put back beside the ones he published (Chapter 22, §22.1): strong. The technical figures are in Evidence table A.21 at the end of the book.

21.4 The Planet That Matches the Domain, and the Planets That Did Not

Gauquelin's discovery concerns one planet in particular, the one whose meaning matches the domain, while the others stay silent. In our recomputation of the 2,087 champions we computed the key-sector rate for every planet the archive allows, each against its own time-shuffle expectation. Mars stands at 21.0 percent against 17.2. The Moon stands at 16.5 percent against 16.5, at chance. Jupiter stands at 17.7 percent against 16.9, Saturn at 16.2 against 16.0 and Venus at 17.6 against 17.2, each within the ordinary scatter of chance. Five bodies were measured at the same births, by the same method, from the same hours and places. One of them departs from chance, the planet that every astrologer from Babylon onward has given to contest, to strength and to the will to win.

That is specificity. An artifact of the astronomy, a bias in the hours of birth or a quirk of the method would act on every planet whose daily round passes through the same hours. It would show up in the Moon and Venus and Jupiter as well as in Mars. Our time-shuffle control already removes every such artifact by construction, and the other planets confirm it independently by sitting at their expectations. Whatever marks the champions marks them through Mars and through no other body in the sky at their birth. The Moon was among Gauquelin's thirteen French observations for the champions (§21.2), and in the archive as a whole it sits at its expectation, so the archive does not return that French observation and we count it as a miss.

The same specificity runs through the whole matrix of professions. Gauquelin specified, from his French data, a characteristic planet for each of the great professions he studied, and in some cases a planet that should be low. When we recompute the archive with our conservative controls, the predicted pattern appears. Saturn rises at the births of physicians and scientists, and most strongly among the members of the Académie de médecine, the most eminent of them all. Jupiter rises at the births of military officers, actors and politicians. Mars rises among the military officers and among the physicians and scientists. And the Moon, which Gauquelin linked to the writers only in the work of 1960, rises at the births of writers. The deficits he predicted appear in their places too, Saturn and Mars low among the painters and Jupiter low among the scientists.

Table 21.3 — The planet–profession matrix, recomputed from the CURA archive (key-sector rates against time-shuffle expectations, chances in the predicted direction, our re-analysis)

The odds beside each row are the cautious ones, for the exact figures at these sample sizes run somewhat higher than the rough ones, and we print the lower of the two. The writers' Saturn deficit was computed in two separate shuffle runs whose expectations differ slightly (16.3 and 17.0 percent). We print the whole range, from a modest tendency at the weaker figure to a clear deficit at the stronger. Saturn for the scientists taken alone is weak in this edition of the archive, 18.1 percent against 16.4. Chance alone would produce it about 6.5 percent of the time, 1 in 15, which is not significant by the usual line. The Saturn surplus is concentrated in the physicians and above all in the academicians. That concentration shows the eminence effect, since the academicians are the most eminent group the archive holds. The same concentration is measured directly, on an independent scale of fame, among the champions (Chapter 22, §22.2).

Study card — The planet-and-profession matrix (one card for the table; the strength of each row stands in its last column)

The academician who gave a life to research and to the austere discipline of medicine carries Saturn. The officer, the actor and the politician, whose lives are lived in public command and display, carry Jupiter, and the champion carries Mars. Each excess is a few percentage points, and each would arise by chance alone somewhere between once in a few hundred and once in half a million trials. And where the meaning does not belong the planet is silent, or low, as Jupiter is low among the scientists and Saturn and Mars among the painters.

A pattern produced by some general bias toward the key sectors could only add births to them. It could never take Saturn and Mars away from the painters at odds of about once in a thousand and once in five thousand. Gauquelin specified those deficits from his French work, and the archive returns them in their places. In our reading they show that the correspondence works by presence and by absence at once. The planet of austere discipline and the planet of combat are scarce at the births of those who gave their lives to color and form, just as Jupiter, the planet of public command, is scarce among the solitary laborers of the laboratory. We draw no doctrine from any single deficit, and we score none of them against the tradition, since Ptolemy wrote no deficits for these professions. The technical figures are in Evidence table A.21 at the end of the book.

21.4.1 What the Tradition Wrote, Row by Row

Gauquelin's predictions came from his own French data. The stronger test of our art is whether the planets the tradition itself wrote down for each profession, long before any of this counting, are the ones that appear. We built that scorecard in Part I from the gods whose visible bodies the planets are and from Ptolemy's Tetrabiblos, read in Robbins's translation. It uses his significations of character in III.13 ("Of the Quality of the Soul") and his assignments of occupation in IV.4 ("Of the Quality of Action") (Chapter 6, §6.6). Table 21.4 — The tradition's planets for Gauquelin's professions (our re-analysis of the CURA archive, with Ptolemy's Tetrabiblos in Robbins's translation)

Study card — The tradition's planets, read against the archive (a lighter card, since the table carries the figures and each row takes the strength of its own result above)

Two rows are hits on the occupation Ptolemy named, three are matches on a character word that we read after Gauquelin's result, and four are misses. The misses are plain nulls, the scatter of chance around an ordinary rate. Jupiter's signification is broad, covering leadership, honor and public life. We score Jupiter for the actors as a miss, since Ptolemy gave the theater to Mercury and Venus. Three of the strongest results fall on Ptolemy's significations of character in III.13: the warrior's spirit of Mars, the deep, austere, single-purposed labor of Saturn and the leadership of Jupiter. Gauquelin had found each planet before we laid the word beside it. On his occupations in IV.4 they would be partly different, since IV.4 gives the athlete's wreath to Venus with Jupiter.

Nabû was the scribe of the gods, and Ptolemy wrote "scribes" under Mercury in so many words. Among 1,351 writers of the archive Mercury stands at 16.2 percent against 15.4, a difference that chance alone would produce about once in every two or three trials. Mercury does not mark the writers in Gauquelin's data. The writers show the Moon instead, and strongly. We give no reason for the Moon after the fact (Chapter 6, §6.6). The Moon's place among the writers stands as Gauquelin's own finding. The same holds for Venus, which Ptolemy gave to the painters and, with Mercury, to the arts of the Muses. The painters show Venus at chance and the French musicians a positive tendency that does not reach significance, and both rows stand as misses.

The reasoning of the scorecard applies to them (Chapter 6, §6.6). A hypothesis that failed in its sharpest rows, the scribe's planet for the writers and the painter's planet for the painters, has been tested. Its hits stand at about once in nine billion trials for Mars among the champions, about once in half a million for Jupiter among the officers and about once in fifty thousand for Saturn among the academicians.