The Shape of LightAn interactive exhibition about light

The third room

Inside the lamp

The first two rooms followed light outward: through instruments to the eye, then past the edges of sight. This room turns round and walks back to the lamp. Why does a hot thing glow? Why does each element shine in colours of its own? And how can one thing spread like a wave and arrive like a grain?

The third room · Inside the lampChapter 13 of 21

Computed: the colour of a glowing body from 800 to 12,000 kelvin. The brightness is compressed to fit a screen. The hot end really outshines the cool end more than a million million times.

Heat a piece of iron and it glows dull red, then orange, then a yellow that is nearly white. For most of history every lamp was something made hot. In 1900 the exact shape of that glow broke the physics of the century that had just ended.

Chapter 13 · Inside the lamp

The Glow

Why hot things shine, and why with the colours they do. The first lamp was a fire. What decides its light?

Look first

The oldest lamp

Anything warm gives off light. You do: a body at 310 kelvin glows most strongly near 9 micrometres, far into the infrared, which is how the snake in chapter 8 finds a mouse in the dark. Make a thing hotter and two changes come together. It gives off far more light, and the light shifts toward shorter waves, until the short end of it crosses into the narrow band you can see.

A lamp filament is a wire kept at about 2,800 kelvin. Below is everything it gives off, with the visible band marked. Only about 6% of it is light you can see. The rest is infrared, which is why an old-fashioned bulb is too hot to touch, and why Herschel’s thermometers in chapter 7 went on rising past the red.

computed
A black body at 2,800 K. The curve peaks at 1,035 nm, past the red. The coloured band is all the eye can use.

Commit to a prediction

Hotter than white

Watch iron heat in a forge. It glows dull red at about 1,000 kelvin, orange near 2,000, and by 3,000 a yellow-white. The surface of the Sun, at 5,772 kelvin, is the white we call daylight. All this time the peak of the curve has been sliding through the spectrum, from the infrared toward the blue. At the Sun’s temperature it sits in the green.

Keep heating. The peak moves on through the blue and the violet, and out into the ultraviolet.

What colour does a body glow at 10,000 kelvin, and at 40,000?

Why the first guesses made sense

A curve, not a colour

If you chose the rainbow, you followed the peak, and the peak does move just as you thought: its wavelength is inversely proportional to the temperature. What the peak cannot tell you is how broad the curve is. The whole visible band is a small piece of it, and all the eye reports is the tilt of that piece: more red than blue, level, or more blue than red. Level is what we call white.

If you expected the glow to fade, you pictured the curve sliding sideways, like a hill moving out of a window. It does not slide. It grows. Every wavelength brightens as the temperature rises, and the short ones brighten fastest.

Why the curve has that shape

The curve that should not exist

A hump that falls away on both sides looks unremarkable. In 1900 it was a scandal, because the physics of the day could explain only one side of it.

Think of the light inside a hot oven as waves trapped between the walls, like the notes of an organ pipe. Shorter waves fit in more ways: halve the wavelength, and sixteen times as many notes are available in each band. The best-tested rule of heat said that every note, whatever its pitch, takes the same average share of the energy. So the glow should climb without limit toward short wavelengths: more in the blue than the red, more in the ultraviolet than the blue, more again in X-rays. Every fire should pour out its heat as a flash of the hardest light.

It does not. At 1,500 kelvin the rule’s prediction for green light is about 2 million times too bright.

calculated
A body at 1,500 K. Solid: what is measured, and what Planck’s law gives. Dashed: what the physics of 1900 required. Both scales are logarithmic.

Max Planck found the cure in the last weeks of 1900, and disliked it. Suppose a note of frequency f cannot hold just any amount of energy, only whole lumps of size hf, where h is a new constant of nature. Low notes have small lumps, and take their fair share as before. High notes have lumps that cost far more than the heat can usually pay, so most of them sit empty.

At 3,000 kelvin the typical thermal energy is 0.26 electron volts. One lump of violet light costs 3.1, twelve times as much, and the chance of finding it paid is about 6 in a million. That is the fall on the short side of every glow. It is also the first place the grain of light, which you met in chapters 5 and 12, ever showed itself.

The bench

A body you can heat

One slider sets the temperature. The curve is drawn to its own peak each time, so that its shape can be seen; the readout says how much brighter or dimmer it really is.

The glow

Things to try

  1. Start at your own temperature and heat slowly. At what temperature does a millionth of the glow first become visible?
  2. Find the temperature that makes the most visible light for each watt. Which preset is nearest?
  3. Switch on what 1900 required. Where does it agree with the real curve, and where does it part?
its colour

What you can now say

Anything warm glows, with a spectrum set by its temperature alone: hotter means far brighter and bluer, from dull red through white to pale blue, and never green. The curve falls away at short wavelengths because light is given out in lumps, and the costly ones are rarely paid.

The lamp filament, the Sun, a forge and your own skin all follow the one curve. Only the temperature differs.

For the determined reader: the laws, the numbers, and the limits

Three laws for one curve

The brightness of a black body at temperature T, per unit of wavelength λ, is given by Planck’s law:

B = (2hc² / λ5) / (ehc/λkT − 1)

Here h is Planck’s constant, c the speed of light and k Boltzmann’s constant, which turns a temperature into an energy. When hc/λkT is small the law becomes B = 2ckT/λ4, the classical result of Rayleigh and Jeans. When it is large the law becomes Wien’s formula, with the 1 dropped. Two consequences follow from the full law. The peak lies at λ = b/T, with b = 2.898 mm·K. And the whole glow, added over all wavelengths and directions, is σT4 from each square metre, with σ = 5.670 × 10−8 W m−2 K−4: double the temperature, and the glow grows sixteenfold.

BodyTemperaturePeakFrom each cm²Visible shareLumens per watt
You310 K
A forge1,000 K
A filament2,800 K
The Sun5,772 K
Sirius9,940 K

Colour, and light for eyes

Colours are worked out with the CIE 1931 standard observer, the agreed description of an average eye, and shown as a screen can show them. A lumen is a unit of light weighted by how strongly the eye responds to each wavelength; green light at 555 nm gives 683 lumens for each watt. A black body is at its best as a lamp near 6,600 kelvin, where it gives 95 lumens for each watt it radiates. A filament at 2,800 K gives 15.

Where this model stops

Real surfaces are not black. Each gives off a fixed fraction of the ideal glow at every wavelength, the same fraction of light that it would absorb, and that fraction varies with the material, the wavelength and the finish. Tungsten, like every real material, gives less than the ideal glow. A flame, a thin gas, or anything light can pass straight through glows much less like the curve, as the next chapter shows. And the Sun is not at one temperature: its light comes from layers at different depths, and carries thousands of dark lines.

A case you have not seen

A cup in a kiln

A potter’s kiln is a closed box of firebrick. It is held at 1,300 kelvin until everything inside is at that one temperature. Inside stand two cups, one glazed white and one glazed black. You look in through a small hole. Nothing lights the inside but its own glow.

What do you see?

One more case

The glow of the sky

Nothing you have met in this chapter is a perfect black body. A filament, the surface of the Sun and a kiln with a hole in it all fall short of the ideal curve somewhere.

The whole sky glows faintly in microwaves: the leftover heat of the young universe, stretched to long wavelengths as space expanded. The COBE satellite, launched in November 1989, carried an instrument that measured the spectrum of that glow at 43 frequencies, from above the air.

How closely does the sky follow Planck’s curve, if the temperature is the one number you may adjust?

An act of despair

By 1860 Kirchhoff had set the problem. The glow of a black body, he argued, is one universal curve, and he called finding it a problem of the highest importance. For forty years nobody could.

The measurements that settled it were made in Berlin, at the Imperial Physical Technical Institute. The institute had been asked a practical question, whether gas or electricity would light the city’s streets more cheaply, and to answer it needed a standard of brightness. In 1895 Otto Lummer and Wilhelm Wien built the first oven with a small hole made for the purpose. Lummer, Ernst Pringsheim, Heinrich Rubens and Ferdinand Kurlbaum then followed the glow further into the infrared than anyone had. By the autumn of 1900 their curves no longer agreed with Wien’s formula, the best of the day, which Max Planck had endorsed.

On 7 October 1900 Rubens and his wife came to dinner with the Plancks, and Rubens described his newest measurements. After the guests had left, Planck found a formula that fitted them. He presented it on 19 October. It fitted every measurement, and he did not know why. Two months later, on 14 December 1900, he presented a derivation. It worked only if energy came in lumps. In the paper that followed he wrote that the energy had to be taken “not as a continuous, infinitely divisible quantity, but as a discrete quantity composed of an integral number of finite equal parts”. The value he gave there for his new constant is within about 1% of today’s. Of those two months he said, in his Nobel lecture, that “after some weeks of the most strenuous work of my life, light came into the darkness”. He later called the step “an act of despair”, and at first he took the lumps for a purely formal assumption.

In 1905 Albert Einstein proposed that they are real, and that light itself is made of them. That is the grain you counted in chapter 5.

Where this explanation stops

Light that is not a glow

The curve describes light made by heat alone, in matter dense enough to absorb whatever falls on it. A great deal of light is not made that way. The screen you are reading is far brighter than its temperature could make it, and its light is not spread in a broad hump. Nor is all the light of a flame.

Sprinkle salt into a gas flame and it flares a single sharp yellow. No temperature gives a glow like that. Where does it come from?

End of chapter 13

Added to your notebook
Anything warm glows, with a spectrum set by its temperature alone: hotter means brighter and bluer, never green. The curve falls at short wavelengths because light comes in lumps, and the costly ones are rarely paid.

Salt in a flame burns one sharp yellow. Why that colour, and only that?

Chapter 14 · The Line