What Popper Actually Said About Astrology
Karl Popper’s falsifiability criterion is often summarized in a way that makes it sound more sweeping than his actual argument. The common version: “astrology is unfalsifiable, therefore unscientific, full stop.” Popper’s actual discussion, in Conjectures and Refutations and elsewhere, was more specific than this.
Popper’s concern wasn’t that astrology makes zero testable claims. It was that astrology, as typically practiced, is constructed and used in a way that insulates it from being tested in any way that could threaten the practitioner’s confidence in the system. He drew an explicit contrast with his preferred example of genuinely scientific theorizing — Einstein’s general relativity, which made a specific, risky, quantitatively precise prediction (the bending of starlight around the sun, measurable during a solar eclipse) that could have come out wrong, and whose confirmation in 1919 was therefore genuinely informative, because failure had been a real possibility going in.
Astrology, on Popper’s reading, doesn’t typically generate predictions of this kind — specific, quantitative, risky, stated in advance of being checked. It generates characterizations broad and flexible enough that almost any subsequent observation can be read as consistent with them. This is a claim about astrology’s typical interpretive practice, not a claim that no statement derived from a birth chart could, in principle, be specific enough to fail a real test.
Sorting Astrological Claims by Falsifiability
Applying Popper’s criterion carefully, rather than as a blanket dismissal, requires sorting astrology’s claims into categories that turn out to have quite different falsifiability profiles — a distinction that gets lost when “astrology” is treated as a single undifferentiated claim.
Sun-sign personality claims are, in fact, straightforwardly falsifiable, and have been falsified, repeatedly, in exactly the way Popper’s criterion would want: take a large sample, measure personality using a validated instrument, check for correlation with sun sign, see if the correlation exceeds chance. This has been done many times, discussed at length elsewhere in this series, and the consistent finding is no detectable effect. This is Popper’s criterion working exactly as intended — a clear, falsifiable claim, tested, and falsified.
Gauquelin-style planetary position claims — discussed in the companion piece on meta-analysis — are also genuinely falsifiable in Popper’s sense: a specific, quantitative, statistical claim about birth-time clustering relative to a person’s profession, checkable against birth records and professional achievement data. And, unusually within this field, these claims have been tested, replicated by researchers including skeptics, and have not been straightforwardly falsified — producing the contested, weak-but-significant result discussed in detail elsewhere. Whatever one concludes about what this result means, it’s a case where astrology generated a falsifiable claim and the claim survived testing better than Popper’s framing would predict, at least so far.
Individual chart reading statements — the actual content of a typical consultation, “you tend to internalize conflict rather than expressing it outwardly” — are where Popper’s concern bites hardest, and where the base-rate-neglect and Barnum-effect mechanisms discussed elsewhere in this series are most directly relevant. These statements are usually not falsifiable in any meaningful sense, not because chart-derived statements are inherently unfalsifiable, but because they’re typically constructed (whether deliberately or through the selective survival of effective interpretive language over centuries) to be broadly true of almost anyone, which makes them compatible with virtually any subsequent observation about the person.
The Falsifiable Core, the Unfalsifiable Shell
This sorting reveals something more precise than “astrology is unfalsifiable”: astrology contains a falsifiable empirical core (specific, testable claims like sun-sign-personality correlation or Gauquelin-style birth-time clustering) wrapped in a much larger, practically dominant shell of interpretive practice that isn’t constructed to be falsifiable, and largely isn’t tested as such by practitioners or clients in everyday use.
This distinction matters because it changes what the falsifiability critique is actually accusing astrology of. It’s not that astrology can’t generate testable predictions — it has, several times, and those predictions have been tested with real rigor. It’s that the practice as actually used — the daily horoscope, the consultation, the chart reading a person receives and evaluates for personal accuracy — operates almost entirely within the unfalsifiable shell, insulated from the more rigorous testing that the field’s falsifiable core has occasionally been subjected to by researchers.
This is, in fact, closer to what Lakatos’s “research programme” framework (discussed in the companion piece on the demarcation problem) captures more precisely than Popper’s original criterion does: a “hard core” of underlying claims (celestial positions matter) surrounded by a “protective belt” of interpretive flexibility that absorbs disconfirming evidence without ever exposing the hard core to a real test in ordinary practice. The falsifiable core exists. It’s rarely what’s actually operating when someone reads their horoscope.
What Happens When the Falsifiable Core Fails
The most philosophically revealing moments in this story are the ones where astrology’s falsifiable core was tested and did fail — because the response to that failure is itself diagnostic, in exactly the way Lakatos’s framework anticipates.
Shawn Carlson’s 1985 study, discussed at length elsewhere in this series, generated a clear, falsifiable prediction (professional astrologers should be able to match natal charts to corresponding personality profiles at above-chance rates) and tested it under double-blind conditions. The result, on the primary measure, was a clear failure of the prediction. The response from the astrological community, broadly, was not to revise core claims about celestial influence on personality. It was closer to the pattern Lakatos describes as degenerating: criticism of Carlson’s specific methodology (some of it reasonable — the 2023 reanalysis discussed elsewhere found legitimate grounds for concern about how the study’s borderline results were reported), but with no subsequent generation of new, independently testable predictions that succeeded where Carlson’s test had failed. The protective belt absorbed the failure. The hard core went untouched, and no new successful predictions emerged from the encounter.
This is precisely the pattern Popper himself worried about, even if his original “single risky prediction” framing didn’t fully capture why the pattern is diagnostic. The problem isn’t that the prediction failed once. Real, legitimately scientific theories fail specific predictions sometimes, and survive by genuinely revising in response. The problem is the characteristic response to failure — explaining it away through methodology criticism and auxiliary hypotheses, without that process generating new successes that wouldn’t otherwise have been expected.
What a More Popperian Astrology Would Look Like
It’s worth asking, in the spirit of taking the philosophical exercise seriously rather than just using it to dismiss astrology, what a version of astrological practice that actually satisfied Popper’s criterion more fully would look like.
It would generate specific, quantitative, advance-stated predictions — not “you may experience some tension in relationships this month” but something with a defined, falsifiable shape, checked against outcomes with the same rigor Gauquelin’s planetary-position claims have received. It would track and publicly report its failures with the same prominence as its successes, rather than relying on the hindsight-bias and selective-memory dynamics discussed elsewhere in this series. And when predictions failed, it would treat the failure as informative about the underlying framework, revising the framework in response, rather than treating each failure as a special case requiring an ad hoc explanation that leaves the framework itself untouched.
This is, notably, close to the description of what does happen in the small corner of astrological research that has engaged seriously with empirical testing — the Gauquelin tradition specifically, and the ongoing methodological dispute over how to interpret its results, discussed in the companion piece on meta-analysis. That corner of the field looks considerably more “Popperian,” in the sense of generating falsifiable claims and taking their testing seriously, than the daily-horoscope-and-consultation practice that constitutes the overwhelming majority of how astrology is actually used and experienced.
The Honest Verdict
Applying Popper’s test carefully to astrology, rather than using it as a one-line dismissal, produces a more specific and more useful verdict than “unfalsifiable, therefore pseudoscience.” Astrology is not uniformly unfalsifiable — it has generated genuinely falsifiable claims, some of which have failed clearly (sun-sign personality correlation) and at least one of which has produced a genuinely contested result that hasn’t failed cleanly (the Gauquelin findings). What’s unfalsifiable, and what dominates the practice as most people actually encounter it, is the interpretive shell that surrounds this core — constructed, through centuries of evolution toward whatever felt compelling to clients, in ways that resist the kind of test Popper had in mind.
The falsifiability criterion doesn’t settle whether any version of astrology is true. What it settles, more precisely than the blanket version of the critique usually allows, is which parts of the practice have actually been exposed to the kind of risk that would make their survival meaningful — and which parts never were, by design or by accident, regardless of what happens when you check them against the world.