The mirror that dims by aiming its silver at the roof
Assumes: The bend at the boundary, and what it is really about · The angle at which reflection picks a side
Most cars made since the 1950s have the same small lever under the interior rear-view mirror. Flipped at night, it turns a mirror bright enough to dazzle into one dim enough to look at, while still showing the road behind. Many drivers assume the mirror has a second, tinted face that swings into place, or a filter. It has neither. The whole mechanism is a single piece of glass, slightly thicker at one edge than the other, silvered on the back, and the lever does nothing except tilt it by a few degrees.
The trick depends on a fact about clear glass that is usually beneath notice — that its bare surface reflects four per cent of the light falling on it — and on making that four per cent leave in a different direction from the reflection off the silver behind it. The second part is geometry and the first is electromagnetism, and the brightness the driver sees at night is set by the refractive index of the glass and by nothing a manufacturer can adjust.
Four per cent from bare glass
Any surface between two transparent materials of different refractive index reflects some light. The angle at which reflection picks a side followed how much, using the boundary conditions that the electric and magnetic fields must satisfy at the surface, and found the answer depends on the angle and on the polarisation. Straight on, both polarisations reflect alike, and the fraction is
which for ordinary window glass, with = 1.52, is 4.26 per cent.
A driver looking into a rear-view mirror sees light that arrives at it at angles of perhaps ten to twenty degrees, and over that range the reflectance hardly changes: it is 4.41 per cent at thirty degrees, and rises steeply only past sixty, the grazing reflection that turns a lake into a mirror at sunset. One polarisation reflects more than the other even here, but headlight beams are unpolarised and the average is what matters. So the dim image the driver gets in the night position has a brightness fixed by one number, the glass’s index, and it is the same in every prism mirror made of ordinary glass.
The dependence on the index is steep enough to matter for a designer who wanted a different night brightness. Glass of index 1.45, like fused silica, reflects 3.4 per cent; a dense flint of index 1.8 reflects 8.2 per cent, twice as much; a plastic of index 1.49, like acrylic, 3.9. A prism mirror made of flint would dim only tenfold, and one of a low-index plastic about twenty-five-fold. The choice of ordinary soda-lime glass, made for cost and durability, therefore fixed the dimming at nineteen, and a reflection measured in a darkened room off the front of any such mirror gives back its index, as a well-known laboratory exercise does with a glass block and a light meter.
Four per cent is not small enough to be invisible. It is the faint image of a room seen in a window at night, and it is why spectacle lenses and camera optics carry coatings: the layer that makes a reflection vanish followed how a film a quarter of a wavelength thick cancels the reflection by interference. A prism mirror has to do the opposite of what a coated lens does. It needs its front surface to reflect, because that reflection is its night image, and an anti-reflection coating on its front would ruin it.
Why the silver is behind the glass
Most household mirrors are silvered on the back, and for good reasons that have nothing to do with night driving: the glass protects a soft, tarnishing layer of silver from fingers, cleaning cloths and air, and the silver is sealed behind it with paint. The price is that light reaches the silver only after crossing the glass, and the front surface takes its four per cent on the way in. In an ordinary flat mirror the two reflections leave parallel and almost on top of each other, separated by twice the glass’s thickness, and the faint one is lost under the bright one; only a candle flame held close shows the double image. Instruments that cannot tolerate it — telescopes, laser optics, the mirrors in a camera’s viewfinder — are silvered on the front instead and handled with care.
The prism mirror takes the defect of a back-silvered mirror and turns it into the feature. It needs both reflections, one strong and one weak, from a single piece of glass, and it needs them to be separable. A flat back-silvered mirror has the two reflections but cannot separate them. A wedge separates them by an angle, and the rest is aiming.
A wedge parts the two reflections
If the glass were a flat plate, the reflections from its two surfaces would leave parallel, and the driver would see both at once, the faint one superimposed on the bright one a few millimetres to the side. Making the glass a wedge separates them in angle. Light reflected from the front surface leaves as from a plane mirror. Light that enters the glass bends towards the normal at the front surface, by the rule the bend at the boundary derived, reflects from the silvered back, which is tilted by the wedge angle, and bends again on the way out. The back reflection is turned by twice the wedge angle inside the glass, and refraction at the exit multiplies that by about the index.
The two reflections leave about apart, roughly three times the wedge angle: a four-degree wedge, a figure of the right size for these mirrors, parts them by 12.7 degrees when traced exactly. That is far wider than the driver’s view through the mirror, which spans only a few degrees as seen from the driving seat, so when one image fills the mirror the other is aimed somewhere else entirely.
The lever swaps which one the driver sees
Tilting a mirror by an angle turns its reflection by twice that angle, a fact any child with a mirror and a sunbeam has used. Both of the wedge’s reflections turn together, because they come from the same piece of glass, so they stay 12.7 degrees apart while the pair swings.
In the day position the mirror is set, by the driver, so that the silvered reflection shows the rear window. The front-surface reflection then points 12.7 degrees away — for a mirror whose thick edge is at the bottom, at the car’s ceiling or the rear parcel shelf, depending on how it is built — and it is faint and shows something dull. Flipping the lever tilts the whole mirror by half the separation, 6.3 degrees, and the roles exchange: now the front surface’s reflection shows the rear window, and the silver’s strong reflection is aimed at the roof lining. The driver still sees the road behind, at four per cent, and what the silver now shows is a dark grey headliner, whose own faint image is mixed into the view at a brightness nobody notices.
The geometry has a convenience that is easy to miss. Because both reflections turn together when the mirror is moved, the driver’s adjustment in the day position sets the night position as well: the lever always tilts by the same 6.3 degrees, and whatever the silvered reflection was aimed at by day, the front reflection is aimed at by night. No second adjustment is needed, for a tall driver or a short one, and the lever has only two stops.
That is the point of the design. A tinted mirror would dim everything, including the dark road. The prism mirror dims the view by aiming its bright reflection at the darkest thing in the car, and it can do so only because the darkest thing in the car is a dozen degrees away from the rear window.
Twenty times dimmer, and the ghosts by day
The silvered reflection has to cross the front surface twice, losing 4.26 per cent each way, and the silver itself returns about ninety per cent of what reaches it through the glass; together the day image carries about 82.5 per cent of the light. The night image carries 4.26 per cent.
The ratio is nineteen, a little over four photographic stops. Headlights a few car-lengths behind are among the brightest things a night-adapted eye meets, and cutting them by a factor of nineteen turns a dazzle into something that can be looked at without losing the eye’s adaptation to the dark road ahead. A dimmer night image would lose the darker parts of the scene behind; a brighter one would not tame the glare. Whether a factor of twenty is the best choice is a question about the eye, which the flash the eye counts found can register a handful of photons when fully dark-adapted, and the mirror does not get to choose it: the factor is set by the glass’s index.
By day there are ghosts. The front surface’s four per cent goes 12.7 degrees off, and a further 3.16 per cent comes from light that reflects off the silver, back off the inside of the front surface, and off the silver again before leaving, turned by twice as much. Against a daylit scene they are lost, because they show other parts of the car’s interior at a twentieth of the brightness. Against a dark scene with one bright point in it — a following car’s headlight on a dusky evening — they appear as faint extra headlights in odd places, which is often the first time a driver notices that the mirror is not quite a mirror.
The ghosts form a series, and the series is easy to sum. Each further pass inside the glass reflects once more off the silver, at ninety per cent, and once more off the inside of the front surface, at 4.26 per cent, so each internal ghost is about twenty-six times fainter than the one before it and turned a further 12.7 degrees: the first internal ghost carries 3.16 per cent of the light, a twenty-sixth of the main image, and the next 0.12 per cent. With the front-surface ghost at a nineteenth, a single bright point behind the car makes a short row of images strung out along the wedge’s direction, of which only the first two are ever seen, and all of them together hold about nine per cent of the main image’s light. The same series exists in the night position, but there the ghosts are aimed at the roof along with the silver’s main reflection, and they vanish from the view.
The day image has been through a prism
The wedge that parts the reflections is also a prism, and the silvered reflection has passed through it twice. Glass bends blue light more than red — the dispersion made of angles traced this in a prism — so the day image of a white point is spread into a short spectrum along the direction of the wedge.
For a four-degree wedge of crown glass the spread between blue and red is 4.4 minutes of arc. The eye resolves about one minute of arc in good light, so this is in principle visible, but it is visible only on a sharp, bright point: ordinary scenes have little contrast at that scale, and the colour fringe is lost in them. It is the reason the wedge is kept to a few degrees. A steeper wedge parts the reflections further, which is helpful, and smears colour proportionately, which is not; two glasses that cancel a derivative showed how lens designers remove such colour error by pairing glasses of different dispersion, and a cheap mirror cannot afford to. The night image, by contrast, is a pure reflection from a surface and has no colour error at all — the dim image is optically the better one.
The same contrast as a window at night
A sheet of window glass shows a room’s reflection at night and not by day, and the reason is the reason the prism mirror works. The window reflects four per cent of the room’s light in both cases. By day the light coming through from outside is far brighter than four per cent of the room, and the reflection is drowned; at night the outside is dark, and four per cent of a lit room is the brightest thing in the glass. Which image is seen is decided by which is brighter, not by which surface made it.
In the night position the driver receives two images at once from the mirror: four per cent of the road behind, through the front surface, and 82.5 per cent of the headliner, through the silver. The headliner is in shadow and dark grey, so its image is dimmer than the road’s image wherever there are headlights, and the road behind wins. Were the car’s interior lit — a courtesy light left on — the silvered image of the lit headliner would be bright enough to wash out the dim view of the road, and the night position would stop working, which drivers who have tried it with the dome light on will recognise.
A family of tricks with a second surface
The prism mirror belongs to a family of devices that work by giving light two surfaces and making them disagree. The rain sensor mounted behind the same mirror on many modern cars, which the windscreen that feels the rain described, depends on total internal reflection at the windscreen’s outer surface being spoiled by water. The rough surface that is a mirror at a glance found a surface’s mirror quality depending on angle, so that at grazing incidence a matt road shines. And the corner that sends light home uses three surfaces at right angles to send light straight back whatever its direction — the opposite intention to the prism mirror, which sends its strongest reflection deliberately where no one is looking.
Since the 1990s many cars have replaced the prism with an electrochromic mirror, in which a gel between two glass plates darkens when a sensor detects glare, and the reflectance varies continuously from about eighty per cent down to six or so. It needs power, a sensor and a control circuit to do what a wedge of glass and a two-position lever did with refraction alone, and it can dim the side mirrors too, which a prism cannot because a side mirror has no dark headliner to aim at.
What the trace leaves out
The traces here are in a plane, at one angle of incidence, with a wedge angle of four degrees chosen as illustrative: manufacturers’ wedge angles and glass are not published in a standard, and a few degrees is all that can be said. The silver’s reflectance through glass, ninety per cent, is a typical value for a protected silvered mirror and varies with the backing and its age; the day brightness scales with it while the night brightness does not. Absorption in the glass, a fraction of a per cent per pass for ordinary float glass, is ignored, though the corridor of mirrors that turns green found it colours light after many passes. And the brightness ratio is the ratio of reflectances; what a driver actually perceives depends on the eye’s adaptation and on scattered light inside the car, neither of which is in the arithmetic.
Still open: what level of dimming a driver needs
The prism mirror’s factor of nineteen was never chosen; it was inherited from the refractive index of glass. Electrochromic mirrors can choose, and how much a mirror should dim, and how fast, is a question about glare and dark adaptation rather than optics. Studies of disability glare from following headlights have grown more pressing as LED and high-intensity headlamps have made the brightest sources brighter, and there is no settled answer to how the dimming should track the glare, nor to whether the dim image should be uniform or dim only the bright points — which a camera-based display mirror, now appearing in some cars, could in principle do.
A day–night rear-view mirror is a glass wedge silvered on its back, so it sends out two reflections about 2nα apart — 12.7° for a 4° wedge — one from the silver at 82.5% and one from the bare front at Fresnel’s 4.26%, fixed by the index of the glass; tilting the wedge by 6.3° hands the driver the faint one and aims the bright one at the dark roof lining, 19 times dimmer with nothing tinted. By day the faint reflections show as ghosts beside a bright headlight, and the silvered image is dispersed by the wedge, 4.4 arcminutes from blue to red, so the dim night image, a pure reflection from the surface, is the sharper of the two.
Part 13 of 13
This essay is one argument about Refraction. The others:
The objects named here
The third axis, after the field and the reading path: the things themselves, and every essay that touches each one.
DispersionFresnel reflectionGhost imageRefractive indexSnell's law
- The angle the rainbow has to be, and why nobody chose it dispersion, refractive index, snell's law
- The colour at which glass disappears dispersion, fresnel reflection, refractive index
- The window a diamond is cut inside dispersion, refractive index, snell's law
- The angle past which light cannot leave refractive index, snell's law
- The angle that is two angles dispersion, refractive index
- The answer that cannot come first dispersion, refractive index