Black-body radiation

The hole is the black body, not the box. Kirchhoff's definition is a surface that absorbs everything falling on it. No material does that, but a small hole in a closed cavity does: light entering is reflected around inside until it is absorbed, and almost none finds its way out again. What leaves the hole is the radiation field in equilibrium with the walls, and by Kirchhoff's law — a good absorber at a wavelength is an equally good emitter at it — that field depends on temperature alone and not on what the walls are made of. What the traced rays are, and what they are not. They are the absorption argument, drawn: a ray enters through the hole, reflects off the walls where it actually meets them, and loses a fixed fraction at every bounce, so that after half a dozen reflections almost nothing is left to come back out. That is what makes the hole an ideal absorber, and by Kirchhoff's law an ideal emitter. They are not a picture of the radiation inside a cavity at equilibrium — that is a photon gas, with the walls emitting and absorbing everywhere at once, not a few beams in flight. The per-bounce loss drawn here is a round number chosen to make the fading visible in six bounces, not a measured reflectivity.

Named approximations. 1. The hole is small enough that letting radiation out does not disturb the equilibrium inside — true in the limit, never exactly. 2. The walls are opaque and all at one temperature. 3. Emissivity ε = 1. Real surfaces fall short: Lummer and Pringsheim measured a freely radiating soot surface at about 92% of black at 650 K, and the porcelain of their electrically heated body at about 90% at 373 K. 4. The bar is worse than either. Tungsten is not grey but selective — its emissivity falls with wavelength, so it emits relatively more in the visible than a black body at the same temperature. That is exactly why it is a lamp filament, and it means the bar's colour is not the black-body colour. Switch to the bar to see both.