Why Is the Sky Blue?

Small things bounce blue light hardest

Junior level — plain language, no maths

Sunlight looks white, but it is every colour mixed together - you can prove that with a prism or a rainstorm. On its way down through the air, that light keeps bumping into molecules of nitrogen and oxygen, and each bump sends a little of it off in a new direction. That bouncing is called scattering.

Here is the crucial part: the molecules are much smaller than the waves of light, and when that is true, they bounce short waves far harder than long ones. Not a little harder - dramatically harder. Blue light has a wavelength about two thirds that of red, and it gets scattered about five times as strongly. So blue light gets thrown all over the sky while red light mostly carries straight on.

That gives you both halves of the picture at once. Look away from the Sun and you see scattered light, which is mostly blue - a blue sky. Look at the Sun low on the horizon and you are seeing the light that came straight through, with the blue scattered out of it along the way - so it looks orange or red. A sunset is a blue sky seen from the other side.

Now the question that catches everyone out. Violet has an even shorter wavelength than blue, so it scatters even more strongly. Why is the sky not violet? Two reasons, and both are about the receiver rather than the light: the Sun emits less violet than blue to begin with, and your eyes are far less sensitive to violet. Press the button and you can see exactly how little of it you are equipped to notice.

The inverse fourth power, optical depth and why sunsets are red

Student level — the core equations

When a particle is much smaller than the wavelength of the light hitting it, the scattering cross-section goes as \(\sigma \propto 1/\lambda^4\). That exponent does all the work. Violet at 400 nm against red at 700 nm gives \((700/400)^4 \approx 9.4\), so violet is scattered nearly ten times as efficiently. The physical origin is that the oscillating electric field drives a dipole in the molecule, the dipole re-radiates with an amplitude proportional to its acceleration, and acceleration brings two powers of frequency into the amplitude and therefore four into the intensity.

How much of the light survives the journey is governed by the optical depth \(\tau(\lambda)\) and the path length. Looking straight up you cross one air mass; with the Sun on the horizon the path is roughly 38 times longer. Transmission falls as \(e^{-\tau m}\), so a small optical depth for red and a large one for blue produce completely different survival rates once \(m\) gets large. That is the whole explanation of a sunset: at high air mass, virtually all the blue has been scattered out of the direct beam, leaving red and orange.

The colour you actually perceive is an integral, not a wavelength. The scattered spectrum is the solar spectrum multiplied by \(1 - e^{-\tau m}\), and to get a colour you must weight that by the three colour matching functions of human vision and convert the resulting tristimulus values to a display colour. The simulation here does exactly that - the sky colour on screen is computed from the physics rather than chosen, which is why it shifts correctly as you lower the Sun.

Haze changes the character of the effect. Aerosol and water droplets are comparable in size to the wavelength, putting you in the Mie regime where scattering is much less wavelength-dependent and strongly forward-directed. That is why a polluted or humid sky is whitish rather than deep blue, why clouds are white, and why the blue is most saturated at high altitude on a dry day.

Density fluctuations, polarisation, and the colour-science half of the answer

Scholar level — full mathematical depth

01Why a uniform gas would not scatter at all

The textbook picture of independent molecules each scattering on its own is a useful lie. In a perfectly uniform medium the scattered wavelets from different molecules interfere destructively in every direction except forward, and there would be no blue sky at all. What actually scatters is fluctuation: the statistical density variations of a gas in thermal equilibrium. Einstein and Smoluchowski put this on a proper footing, expressing the turbidity in terms of the isothermal compressibility, and the result reduces to the Rayleigh formula for an ideal gas. The correct statement is therefore that the sky is blue because air is grainy at the molecular scale, and the graininess is thermodynamic in origin.

02The full Rayleigh phase function and polarisation

The angular distribution for unpolarised incident light is \(\propto (1 + \cos^2\theta)\), symmetric forward and back with a minimum at 90°. At that same 90° the scattered light is almost completely linearly polarised, because the dipole cannot radiate along its own axis. Real air falls slightly short of complete polarisation, around 94%, because molecules are anisotropic - the King correction factor accounts for it. Multiple scattering and ground reflection reduce it further, which is why the neutral points of the sky, Arago and Babinet, exist at all and why their positions are a sensitive measure of atmospheric aerosol loading.

03The violet question, done properly

Three effects compound. The solar spectrum falls off below 450 nm and is further depleted by ozone Chappuis absorption and stratospheric attenuation. The human luminous efficiency function \(V(\lambda)\) is roughly 0.004 at 420 nm against 0.038 at 450 nm and 1.0 at 555 nm. And S-cones number about 5 to 10% of the cone population and are absent from the central fovea. Integrating the actual scattered spectrum against the colour matching functions gives a chromaticity near the blue-cyan boundary, not violet - so the sky's colour is a genuinely joint fact about atmospheric physics and human photoreceptor design, and an organism with different photopigments would report a different sky.

04Ozone, and why the twilight sky is blue at all

Near the horizon at sunset the direct beam is almost entirely stripped of short wavelengths, so naive Rayleigh theory predicts a washed-out sky after sunset. The deep blue of late twilight instead comes from the Chappuis band, a broad ozone absorption between roughly 500 and 700 nm. Light traversing long stratospheric paths loses its middle and long wavelengths to ozone, leaving blue. This is a nice case where the everyday explanation is incomplete in a way that only shows up at a specific time of day.

05What the same physics does elsewhere

The \(1/\lambda^4\) law is not atmospheric trivia. It sets the fundamental attenuation floor of optical fibre - Rayleigh scattering from frozen-in density fluctuations in the glass is exactly why telecom settled on 1550 nm, where the scattering loss has fallen far enough and infrared absorption has not yet risen. It also explains the blue of some bird feathers and of blue eyes, where structure rather than pigment produces the colour, and the reddening of distant starlight by interstellar dust, which follows a shallower power law because the grains are larger.