Getting Kepler wrong

In recent times a bit of philosophy of science bun fight took place on the Intertubes. It started off in the New York Times with an opinion piece by James Blachowicz entitled, There is no Scientific Method. The title is actually a misnomer, as what Blachowicz actually argues is that the problem solving procedure usually called the scientific method is not unique to science. I’m not going to discuss it here but it is hardly an original theory, in fact I’ve argued something very similar myself in the past. I will, however, say that I don’t think that Blachowicz argues his case very well. Above all I think that his final three paragraphs in which he explains why, if the method is not exclusive to science, science is different to other form of knowledge are pretty crappy and largely wrong. Someone who also disparages those final three paragraphs in physics blogger Chad Orzel, who has written a pretty nifty book about the scientific method himself.[1] Chad wrote a post on his Uncertain Principles blog entitled, Why Physicists Disparage Philosophers, In Three Paragraphs, which if your read or have already read Blachowicz’s opinion piece you should definitely also read.

Chad was not the only physicist who weighed in on Blachowicz’s opinion piece with Ethan Siegel posting on his Forbes blog, Starts with a Bang, a rejoinder entitled Yes, New York Times, There Is a Scientific Method. In his piece Blachowicz illustrates his interpretation of the use of the scientific method in actual science with a brief discussion of Kepler’s search for the shape of the orbit of Mars using Tycho Brahe’s observational data as extensively described by Kepler in his Astronomia Nova in 1609. Here are the actual paragraphs from Blachowicz:

Now compare this with a scientific example: Johannes Kepler’s discovery that the orbit of Mars is an ellipse.

In this case, the actual meaning of courage (what a definition is designed to define) corresponds with the actual observations that Kepler sought to explain — that is, the data regarding the orbit of Mars. In the case of definition, we compare the literal meaning of a proposed definition with the actual meaning we want to define. In Kepler’s case, he needed to compare the predicted observations from a proposed explanatory hypothesis with the actual observations he wanted to explain.

 Early on, Kepler determined that the orbit of Mars was not a circle (the default perfect shape of the planetary spheres, an idea inherited from the Greeks). There is a very simple equation for a circle, but the first noncircular shape Kepler entertained as a replacement was an oval. Despite our use of the word “oval” as sometimes synonymous with ellipse, Kepler understood it as egg-shaped (in the asymmetrical chicken-egg way). Maybe he thought the orbit had to be lopsided (rather than symmetrical) because he knew the Sun was not at the center of the oval. Unfortunately, there is no simple equation for such an oval (although there is one for an ellipse).

When a scientist tests a hypothesis and finds that its predictions do not quite match available observations, there is always the option of forcing the hypothesis to fit the data. One can resort to curve-fitting, in which a hypothesis is patched together from different independent pieces, each piece more or less fitting a different part of the data. A tailor for whom fit is everything and style is nothing can make me a suit that will fit like a glove — but as a patchwork with odd random seams everywhere, it will also not look very much like a suit.

The lesson is that it is not just the observed facts that drive a scientist’s theorizing. A scientist would, presumably, no more be caught in a patchwork hypothesis than in a patchwork suit. Science education, however, has persistently relied more on empirical fit as its trump card, perhaps partly to separate science from those dangerous seat-of-the-pants theorizings (including philosophy) that pretend to find their way apart from such evidence.

Kepler could have hammered out a patchwork equation that would have represented the oval orbit of Mars. It would have fit the facts better than the earlier circle hypothesis. But it would have failed to meet the second criterion that all such explanation requires: that it be simple, with a single explanatory principle devoid of tacked-on ad hoc exceptions, analogous to the case of courage as acting in the face of great fear, except for running away, tying one’s shoelace and yelling profanities.

 It is here that Ethan launches his attack accusing Blachowicz of not having dug deep enough and of misrepresenting what Kepler actually did. After posting a picture of Kepler’s wonderful 3D model of his Platonic cosmos:


Ethan posted the following:

Kepler’s original model, above, was the Mysterium Cosmographicum, where he detailed his outstandingly creative theory for what determined the planetary orbits. In 1596, he published the idea that there were a series of invisible Platonic solids, with the planetary orbits residing on the inscribed and circumscribed spheres. This model would predict their orbits, their relative distances, and — if it were right — would match the outstanding data taken by Tycho Brahe over many decades.

But beginning in the early 1600s, when Kepler had access to the full suite of Brahe’s data, he found that it didn’t match his model. His other efforts at models, including oval-shaped orbits, failed as well. The thing is, Kepler didn’t just say, “oh well, it didn’t match,” to some arbitrary degree of precision. He had the previous best scientific model — Ptolemy’s geocentric model with epicycles, equants and deferents — to compare it to. In science, if you want your new idea to supersede the old model, it has to prove itself to be superior through experiments and observations. That’s what makes it science. And that’s why the ellipses succeeded, because they gave better, more accurate prediction than all the models that came before, including Ptolemy’s, Copernicus’, Brahe’s and even Kepler’s own earlier models.

Unfortunately Ethan has hoisted himself with his own petard. He has not dug deep enough and what he presents here is presentist interpretation of what Kepler actually thought and did over a period of around thirty years. I will explain.

At the various stages of Kepler’s development that Ethan sketches Kepler is dealing with and providing answers for different non-exclusive question, which don’t replace each other sequentially.

At the beginning Kepler was looking for an answer to the question, why there are only six planets? In the Copernican system the seven planets of the Greek’s had been reduced to six as the Earth and the Sun exchanged places and the Moon became the Earth’s satellite (a word that Kepler would coin later with reference to the newly discovered moons of Jupiter). This metaphysical question seems rather strange to us today but it fitted into Kepler’s metaphysics. Kepler was deeply religious and his God was a rational, logical creator of a mathematical (read geometrical) cosmos. Kepler’s cosmos was also finite, so there were and could only be six planets. He was later mortified when Galileo announced the discovery of four new celestial bodies and infinitely relieved when there turned out to be satellites and not planets. Kepler’s answer to his question was the model shown above with the spheres of the six planets inscribing and circumscribing the five regular Platonic solids. There are, and can only be, only five regular Platonic solids therefore there can only be six planets, Q.E.D. Using the available data on the size of the planetary orbits Kepler turned his vision into a mathematical model of the cosmos and discovered that it fit roughly but not accurately enough. His passion for precision and accuracy was a major driving force throughout Kepler’s scientific career. Kepler was aware that Tycho had been collecting new more accurate astronomical data for thirty years and this was one of his major reasons for wanting to work with Tycho in Prague; the other reason was that Kepler, as a Protestant who refused to convert to Catholicism, was being expelled from Graz and desperately needed a new job.

In Prague Tycho, who thought he had been plagiarised by Ursus, was not prepared to hand over his precious data to a comparative stranger and instead gave Kepler a couple of commissions. The first was to write an account of Tycho’s dispute with Ursus, which Kepler did producing a classic in the history and philosophy of science, which unfortunately was not published at the time. Kepler second task was to determine the orbit of Mars based on Tycho’s observational data. At this time, this had nothing to do with his previous work in the Mysterium Cosmographicum. Famously, what Kepler thought would be a simple mathematical exercise taking a couple of weeks turned into a six year battle to tame the god of war, published in all its gory detail in his Astronomia nova in 1609. Having at some point abandoned the traditional circular orbits Kepler hit upon his oval, meaning egg shaped rather than elliptical, orbit and calculated it using Tycho’s data. His calculations displayed eight arc minutes of error in places, that’s eight sixtieths of one degree, a level of accuracy way above anything that either Ptolemaeus or Copernicus had ever produced. He had superseded the old model easily to quote Ethan, however eight arc minutes of error was an affront to Kepler’s love of accuracy and in his opinion an insult to Tycho’s observational accuracy, so it was back to the drawing board. In his further efforts Kepler finally discovered his first two laws of planetary motion and his elliptical orbits[2]. This set of answers were however to a different set of questions to those in the Mysterium Cosmographicum and in no way were considered to replace them.

Throughout his life Kepler remained convinced that his Platonic model just required fine-tuning, which he meant quite literally. Already in the Mysterium Cosmographicum he muses about the Pythagorean music of the spheres and his magnum opus, the Harmonices Mundi published in 1619, is a truly amazing conglomeration of plane and spherical geometry, music theory, astrology and astronomy containing many gems but most famous for his third law of planetary motion, the harmonic law. Throughout all of this work the Platonic solids model of the Mysterium Cosmographicum remained Kepler’s vision of the cosmos and in 1621 he published a revised and extended version of his first book confirming his belief in it. It is this combination of, from our point of view, weird Renaissance heuristics, Platonic solids, harmony of the spheres, combined with the high level highly accurate modern science that it generated, the laws of planetary motion etc., that led Arthur Koestler to title his biography of Kepler, The Watershed. He saw Kepler as straddling the watershed between the Middle Ages and the Early Modern Period with one foot planted firmly in the past and the other striding determinedly into the future. The inherently contradictory duality is what leads presentists such as Ethan to misunderstand and misrepresent Kepler. He didn’t replace his metaphysical Platonic solids model of the cosmos with his mathematical elliptical model of the planetary orbits but considered them as equal parts of his whole astronomical/cosmological vision. We do not have Ethan’s Whig march of progress of one model replacing another but rather a Renaissance concept of the cosmos that can only be considered on its own terms and simply doesn’t make sense if we try to interpret it from our own modern perspective.

Since I started writing this post there have been two further contributions to the debate that inspired it. On the bigthink Jag Bhalla interviews Rebecca Newberger Goldstein on the topic under the title, What’s Behind A Science vs. Philosophy Fight?

On The Multidisciplinarian, William Storage, in his The Myth of Scientific Method, takes apart Ethan’s (mis)use of Galileo in his contribution. This one is highly recommended


[1] Chad Orzel, Eureka! Discover Your Inner Scientist, Basic Books, New York, 2014

[2] As I’ve said more than once in the past the best account of Kepler’s Astronomia nova is James R. Voelkel, The Composition of Kepler’s Astronomia nova, Princeton University Press, 2001


Filed under History of Astronomy, Myths of Science

11 responses to “Getting Kepler wrong

  1. It’s too bad more people don’t read C.S. Peirce.

  2. Just for the record, did Kepler ever get over thinking of ellipses as a carload of dung?

    • Are you perchance referencing this?

      • Thanks for the reference. I was thinking that the famous carload of dung remark referred to ellipses rather than the much less mathematically elegant ovals. Of course ellipses still don’t comport with the greater symmetry of the nested Platonic solids; and Kepler would have needed additional cartloads of dung if Brahe’s tables were more accurate since the actual orbits of the planets aren’t exactly ellipses.

      • It is, in fact, one of the most fascinating facts of the history of astronomy that if Tycho’s observations had been more accurate, then Kepler would probably not have discovered his three laws. Those observations were just the right level of inaccurate to suggest the elliptical orbits.

  3. Once again Thony Christie shows how many of the arguments between scientists and philosophers suffer from an inadequate knowledge of history.

  4. “Those observations were just the right level of inaccurate to suggest the elliptical orbits.”

    Well, yes Tycho’s measurements were nowhere near accurate enough to detect the effect of gravitational perturbations on the planetary orbits.

    There were another couple of factors in Tycho’s measurements that need to be remembered. First, even 30 years of observations corresponded to just over one orbit of Saturn around the Sun; secondly, if Denmark’s weather is like the southern UK’s, he probably only had about 100 nights observing a year (even if he could make use of every available night). This is why Kepler needed to work on Mars’ orbit: he had a reasonable number of orbits (~15) combined with a an orbital eccentricity greater than all planets apart from Mercury (and Pluto) and an orbit that brought it closer to Earth than all but Venus.

    We also should not forget that Kepler did not have the mathematical tools developed by later mathematicians like Laplace and Gauss to calculate planetary orbits from observations.

  5. John Roth

    It’s probably fortunate that Galileo wasn’t able to follow up his observation of Neptune. That would have put a lot of stress on the notion that there were only six planets.

  6. Actually, the famous “8 minutes of an arc” had to do not with the egg-shaped orbit, but with the “vicarious hypothesis”, much earlier in the story; this was a classic eccentric circle with an equant, but violating Ptolemy’s bisection rule for where the equant had to be placed. The problem arose only when Kepler tried to account for both the latitudes and the longitudes of the Martian orbit with a single model. Ptolemy and his successors typically used one geometrical model for the latitudes, another for the longitudes, so I guess the “no ugly suit” analogy does apply to this aspect of Kepler’s work.

    It’s also a bit of an ironic (or presentist) twist that we associate the elliptical orbits so closely with the improved accuracy of Kepler’s Rudolphine Tables. In fact, other Keplerian innovations deserve the lion’s share of the credit: using the true sun instead of the mean sun throughout, the consequent stability of the orbital planes, and allowing the earth’s motion to be non-uniform in speed, i.e., treating it the same as the other planets. As Curtis Wilson says in his article “Predictive astronomy in the century after Kepler”:

    … the departure from circularity could be detected observationally only in the orbits of Mars and Mercury, where the eccentricity was exceptionally large; and even in these cases the choice between ellipse and other oval shapes was, as far as the observations could show, a matter for conjecture.

    The ellipse looms so large because (a) Kepler made a big deal of it, and (b) it played such a critical role in the sequel. But if we’re talking purely about model-observation agreement, at that time it was a relatively minor factor.

  7. Pingback: Whewell’s Gazette: Year 2, Vol. #51 | Whewell's Ghost

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s