The ultraviolet catastrophe

Classical physics, applied carefully and without error, predicts that every warm object in the universe emits infinite energy. The oven in your kitchen should have killed you.

1859 – 190517 min

By the end you should be able to

Say what a blackbody is and why its spectrum was such a good test of physics
Explain where the classical prediction comes from and why it diverges
Judge how much of a "crisis" this actually was at the time

Heat something up and it glows. A stove element goes dull red, then orange. A tungsten filament goes yellow-white. The Sun is white; the star Rigel is blue. Hotter means bluer, and the relationship is so dependable that you can measure the temperature of a furnace, a star, or a piece of steel without touching it — from its colour alone. Blacksmiths were doing this by eye for centuries before anyone could explain it. A regularity that reliable demands an explanation, and by the end of the nineteenth century producing one was among the outstanding problems in physics. It was also commercially urgent: the German national standards laboratory was funding the measurements because the electric lighting industry needed to know how to get the most visible light out of a hot filament.

What a blackbody is, and how you build one

To make the problem clean you want an object that absorbs everything falling on it and reflects nothing — so its glow depends only on its temperature, not on what it is made of. That is a blackbody. It sounds like an unbuildable idealisation. There is a lovely trick: take a hollow box, heat it, and drill a very small hole. Light entering the hole rattles around inside and has almost no chance of finding its way back out. The hole is therefore almost perfectly black — and what streams out of it is pure thermal radiation, with no trace of the box's material. Kirchhoff proved in 1859 that this spectrum must be universal: identical for every material at a given temperature. That is what made it such a formidable test. A universal curve depends on nothing but temperature and the constants of nature, so any theory of matter and light has to produce exactly it, with nowhere to hide.

Move the temperature, then switch on the classical prediction and watch where the two curves part company.

Loading the spectrum…

The measured spectrum (green) against what classical physics requires (red). Both come from the formulas, not from artistic licence — and the y-axis is scaled to the Planck peak, so the classical curve genuinely leaves the top of the frame. At long wavelengths the two are indistinguishable.

Take the consequence seriously for a moment, because it is genuinely absurd. Every object above absolute zero radiates. If the classical result were right, every one of them would radiate infinite energy, concentrated at the short-wavelength end. Your body would be emitting unbounded ultraviolet. So would a cup of tea, a rock, the walls of the room. Opening an oven door would sterilise the building. Thermal equilibrium would be impossible, because no finite amount of energy could ever fill the infinitely many high-frequency modes waiting to be filled. Paul Ehrenfest named it the ultraviolet catastrophe in 1911, and the name is exact: the disaster is at the violet end, and it is a catastrophe.

You might think

The classical calculation must have contained an error, which Planck found and fixed.

Actually

There is no error in it. Nobody dropped a factor. The mode counting is straightforward geometry and is still used unchanged today; the equipartition theorem was derived from the foundations of statistical mechanics by Maxwell and Boltzmann and was among the best-established results in physics. Rayleigh and Jeans did the calculation correctly. That is precisely what makes it important. When two secure premises combine validly to give an impossible answer, you have not found a slip — you have found that one of your foundations is false. The classical derivation is not a strawman to be knocked down; it is a proof by contradiction that classical physics cannot be complete.

Now a correction to the story as it is usually told, because the tidy version is not what happened. The standard telling is: physics faced the ultraviolet catastrophe, and Planck invented the quantum to escape it. The dates do not support it. Planck published his law in October 1900. Rayleigh's classical calculation appeared in June of the same year — and was incomplete; Jeans did not supply the correct coefficient until 1905. The divergence was not clearly stated until after the solution existed, and the phrase "ultraviolet catastrophe" was not coined until 1911, eleven years later. Worse for the legend: Planck was not trying to fix a divergence at all. He was working on entropy and the second law of thermodynamics, attempting to derive Wien's empirical formula from first principles — Wien's law fitted the short-wavelength data well and failed at long wavelengths, which was the discrepancy actually bothering him. New measurements in the far infrared, made a few streets away in Berlin, had just shown Wien's law failing there. He was fixing the other end of the curve.

Why the wrong story survives

The catastrophe-then-rescue version persists because it is a better narrative: a problem, a crisis, a hero, a solution. The real sequence was solution first, problem articulated afterwards, and then roughly a decade before anyone — chiefly Einstein and Ehrenfest — grasped how much the solution actually cost. Planck himself spent much of that decade trying to get rid of his own idea. This is worth noticing as a general matter. Physics is written up backwards. Papers present a clean logical route from problem to answer, and textbooks tidy further. The actual work is messier, luckier and slower, and the person who finds the answer frequently does not understand what they have found. Being right and knowing why you are right are different achievements, sometimes separated by years.

None of which makes the catastrophe unimportant. It matters because it locates the failure precisely. Look again at the long-wavelength end of the sim, where the classical curve lies exactly on top of the measured one. That agreement is not luck. Classical physics is not approximately right there — it is right. And any successor theory is obliged to reproduce it, which is a severe constraint on what the successor can be. So the useful question is not "why is classical physics wrong?" but "where, exactly, does it stop working, and what changes there?" The answer: it fails at short wavelengths. Short wavelength means high frequency. And — though nobody could yet see why this was the relevant variable — high frequency turns out to mean large energy per oscillation. That is the crack. Everything in the next four acts comes out of it.

The experiment

Measuring the cavity spectrum into the far infrared

Otto Lummer and Ernst Pringsheim; Heinrich Rubens and Ferdinand Kurlbaum · 1898 – 1900 · Physikalisch-Technische Reichsanstalt, Berlin

The question: What is the actual shape of the blackbody spectrum — in particular at long wavelengths, where Wien’s successful formula had never been properly tested?
The apparatus: Electrically heated cavities of platinum and porcelain with small apertures, held at temperatures up to about 1800 K and measured with thermopiles and bolometers. Reaching the far infrared required the Reststrahlen ("residual rays") method: repeated reflection from rock-salt, fluorite and sylvine crystals, each reflecting strongly only in a narrow band, which isolated wavelengths out to 50 μm without a usable prism or grating.

Theory predicted

Wien’s 1896 law had matched every measurement so far and was widely believed to be exact. It predicts the intensity falling away exponentially at long wavelengths.

They measured

At long wavelengths the intensity was systematically higher than Wien’s law allowed, and it approached proportionality with temperature rather than Wien’s exponential form. Rubens and Kurlbaum confirmed this decisively in the autumn of 1900.

How sure could they be? A few per cent — but the deviation from Wien grew steadily with wavelength rather than scattering randomly, which is what made it convincing rather than dismissible as error.

Why it mattered

This is the measurement that actually prompted Planck, and it is the reason the popular story is wrong. Rubens visited Planck at home on 7 October 1900 and told him of the long-wavelength results; Planck produced his interpolation formula that same evening and presented it to the German Physical Society eleven days later. The crisis he was responding to was at the infrared end, not the ultraviolet one.

1859Kirchhoff proves the blackbody spectrum must be universal, and offers a prize for finding it.
1879Stefan finds the total radiation goes as T⁴; Boltzmann derives it thermodynamically in 1884.
1893Wien’s displacement law: the peak shifts as 1/T.
1896Wien’s distribution law fits the available data, and is believed exact.
1900Rubens and Kurlbaum measure the far infrared and find Wien fails there. Rayleigh publishes the classical derivation. Planck presents his law in October.
1905Jeans corrects Rayleigh’s coefficient, completing the classical prediction — five years after it had been superseded.
1911Ehrenfest names it the ultraviolet catastrophe.

Problem

Take the temperature of a star

Betelgeuse is visibly red. Its spectrum peaks at about 805 nanometres, just past the red end of what your eye can see. What is its surface temperature?

Peak wavelength: λ_max = 805 nm
Wien displacement constant: b = 2.898 × 10⁻³ m·K

Where this leaves us

Something has to give, and both candidates are unthinkable. Either the mode counting is wrong — which would mean Maxwell's electromagnetism is wrong, the theory that had just unified light, electricity and magnetism and was the crowning achievement of Act 1. Or equipartition is wrong — which would mean statistical mechanics is wrong, and with it the atomic theory of heat. In October 1900, Max Planck — a deeply conservative physicist of forty-two who had built his career on the absoluteness of the second law of thermodynamics, and who had once been advised not to go into physics because everything was already known — guessed a formula that fitted the data across the whole range. Then he spent eight weeks trying to work out what it meant. What he found appalled him, and he spent the next decade trying to undo it.