This paper concerns astrology, research and philosophy of science. It argues that, in the final analysis, they are inseparable: astrology is in urgent need of good research, and that in turn requires some grounding in philosophy of science. Different variants of the latter are discussed, and one – the “methodology of scientific research programs” – is used to assess the current state of astrology. On the basis of this assessment, advice is offered concerning some present and future research.

Much astrological research can be faulted because of its neglect of proper methodological safeguards. This article illustrates a need to ascertain in a very precise manner the amount of knowledge of astrology possessed by subjects, as this may inadvertently influence their filling in of questionnaires, personality descriptions, etc. It is shown that apparently very successful studies can be reduced to non-significance by controlling of this variable.

This is a report on a study undertaken in the Southern Hemisphere which looked at personality in relation to date of birth in a predominantly female sample population. The findings are consistent with those of similar studies carried out in the Northern Hemisphere. The author states that prior knowledge of astrology could account for most but not all of the findings of these studies and suggests that media exposure to astrology could be a significant and as yet unexplored area of self-perception.

An analysis of Gauquelin’s professional data revealed significant gender effects: female professionals (actors, athletes, writers) outdo their male colleagues regarding birth frequencies in sensitive sectors for Mars ( p = .01) and, marginally, for the Moon (p = .09). The significance level for the two planets combined is p = .006. In addition, G-sector births of women were generally more frequent than those of men when summed across all five Gauquelin planets ( p = .01). Eleven issues pertinent to these results are discussed. It is concluded that any theory that might be suggested to explain Gauquelin’s planetary effect for professions will also have to account for planetary effects associated with gender.

Harmonic analysis of the Gauquelin planetary distributions for professional groups is used to show that the phase angles of individual harmonics for each planet are closely clustered. This is used to demonstrate that any planetary “influence” is travel ling at the velocity of light. Other supporting evidence is offered and the supposition advance that the observed associations occur because of awareness by the developing embryo or fetus

The authors look, in some depth, at both the development and the publishing history of Jung’s concept of synchronicity in general and Jung’s astrological experiment in particular. This paper addresses possible serious misunderstandings which appear to have arisen regarding the concept of synchronicity and Jung’s astrological experiment and how this might account for the apparent relationship between planetary movements and mundane events.

The validity of E (extraversion) and N (emotionality) for testing astrology is reviewed in detail (60 references) and is found to be adequate in every respect. The conclusions of Parts 1 and 2 are confirmed after the most critical scrutiny, comparison with other studies, and investigation of issues raised by readers. Much previously unpublished work is presented including a survey of the perceived discernibility of human characteristics in birth charts, Pierce’s factor analysis of astrological symbolism, a meta-analysis of 14 studies of astrologer agreement, the distribution of variance between astrologers and charts, a meta-analysis of 31 Vernon Clark results, a new Vernon Clark test of everyday topics, a comparison of the reliability and validity of astrology with that of other diagnostic tests, and an extension of Startup’s power analysis of astrological effects. The picture that emerges consistently from all of this work is that astrology as applied by astrologers using the whole chart has negligible reliability and validity. Aggregating astrologers improved the reliability as expected but had no effect on the hit rate, which remained close to chance level. The upper bound to the variance accounted for by astrological effects is orders of magnitude smaller than that required to meet the claims of astrologers.

The Gauquelin effect (G%) should increase with increased birth time precision. The data, however, do not bear this out, on the contrary, G% even tends to diminish with better birth time recordings, at least from AD 1880 onwards. The expected positive effect might have been overridden by psychological/sociological variables depressing G% and enhancing birth record precision at the same time. Until now, however, such intervening variables remain enigmatic.

Gauquelin & Tracz (1991) rejected Ertel’s (1990) and Rodenbeck’s (1990) conclusion that the Gauquelin character trait hypothesis (CTH) does not hold. They repeated our former biographical trait counts and reported results supporting CTH. The contradiction was explained in terms of sloppy vs. careful trait extraction: our students’ extractions were deemed poor, whereas their students’ extractions were taken as reliable. A close look at procedures and output of extractions, however, indicates that our students had worked at least as carefully as theirs. Evidence shows, instead, that Gauquelin’s own care left something to be desired in that he selected duplicates of traits extracted by his helpers excluding from analysis all non-duplicates. Since he was generally aware of planetary positions he was liable to biased selections and deletions. Our conclusion is backed by results from three statistical comparison on Fresno/Gottingen data. Gauquelin & Tracz’s study therefore does not invalidate our contention that the character trait hypothesis is wrong. Since Gauquelin’s basic planetary effect has continuously gained support, failures with CTH are a challenge to find an explanation for the persistent main effect in entirely different terms.

If the stars incline but do not compel, then astrological outcomes are innately probabilistic. Therefore only probabilistic models are appropriate but hitherto their use has been neglected. This is unfortunate because they promise to revolutionise the everyday practice of working astrologers regardless of whether astrology is used as a language or as an information source. Probabilistic models examine each chart factor (without limit on number or complexity) using modern computer techniques of item response theory, and then produce the best synthesis by maximum likelihood estimation. They do what an astrologer does but include what no astrologer can do, namely address astrology’s inherent uncertainty directly. In effect, they allow astrologers and techniques to be separated from clients, thus delivering astrology freed from the quirks of a particular approach and armed with an objective evaluation of predictive reliability, meaning an end to hedging bets with ifs and buts. Probabilistic models require the use of a computer but are otherwise easy to apply. They are perhaps the most obvious way of improving the application of astrology to individuals.