Night window reflections

Why does a window turn into a mirror after dark and back into clear glass by day, when the glass never changes?

At noon a window is a window: you look through it and see the street. At ten at night the same pane, untouched, shows you your own room and your own face. The glass did not change. Nothing was switched. So either the glass is doing something new after dark, or it was doing this all along and you simply could not see it. Which is it — and how would you tell?

Let us start with what happens when light meets glass at all. A surface between two materials that transmit light differently — air and glass — does not simply let light through. Part of the beam crosses, and part turns back. This is not a defect of cheap glass or a matter of dirt; it is what a boundary between two optical media does, and the fraction is fixed by how much the two media differ. For ordinary window glass meeting air, that fraction is about four per cent at each surface.

Notice the word each. A pane has two faces: the one nearest you and the one facing outside. So roughly four per cent of the light arriving from your room turns back at the inner face, and of the ninety-six per cent that crosses, about four per cent turns back again at the outer face and comes home through the glass a second time. Call it eight per cent altogether. Ninety-two per cent goes out into the night and is gone.

Now hold that fixed and ask what actually varies. Two things reach your eye from the window: the reflected light from your side, and the transmitted light from behind it. If reflection is always eight per cent, then the only variable is how much light there is on each side.

Put numbers to it. Outdoors at midday there might be tens of thousands of units of light; a lit room at night has a few hundred. By day, the window hands you nearly all of a very large number from outside and eight per cent of a small number from inside — the reflection is present, but it is a whisper against a shout, and your eye, which works on ratios rather than absolutes, never registers it. After dark the outdoor number collapses toward nothing, while the room's stays put. Now the eight per cent is the loudest thing coming off that pane. The mirror was always there. Only the competition left.

Does that survive a test? Turn the room lights off at night and press your face to the glass, cupping your hands: the reflection vanishes and you can see the garden. Nothing about the pane altered — you only removed the light on your side. Conversely, stand outside a lit shop at midday and the window shows you the street more than the goods, because now you are the bright side.

One more detail is worth chasing, because it is visible if you look. Since there are two reflecting faces, you get two reflections, offset by however far the light travelled sideways inside the glass. Look at the window straight on and they lie almost on top of each other and read as one image. Look along the pane at a shallow angle and the light's path inside the glass is long and slanted, so the two images pull apart — a lamp becomes a lamp and a fainter ghost beside it. Glancing angles do something else too: the reflected fraction is only four per cent near head-on, but it climbs steeply as the angle flattens, approaching total reflection at true grazing. That is why a corridor window seen edge-on can look silvered even in daylight, and why a wet road throws headlights back at you.

The instinct to explain this as a change in the glass runs deep, so it is worth being explicit: the pane's reflectance is a property of the boundary and the material, and it is the same at midnight as at noon. What changed is entirely on the two sides of it.

That reframes what "mirror" and "window" mean. They are not two states an object can be in; they are two ways the same object can be perceived, and which one you get is decided by a ratio you happen to be standing in. A so-called one-way mirror in an interview room is exactly this and nothing more — a partially reflecting coating with a bright room on one side and a dark one on the other. Turn the lights on in the dark room and the illusion collapses in both directions at once.

Two honest qualifications. The four per cent figure is for light arriving near head-on; it rises with angle, as above, so treat it as a floor rather than a constant. And modern windows complicate the arithmetic without changing the principle — double glazing has four glass-air faces instead of two, which is why you sometimes count several ghost images of a candle. The mechanism is the same; only the percentages move.

• Why the reflected fraction climbs toward total reflection at grazing angles, and how polarised sunglasses exploit it.
• How anti-reflective coatings cancel that four per cent by interference rather than by absorbing it.

Every term the collection defines is gathered in the glossary.