The third room · Inside the lampChapter 15 of 21
A lump of light leaves an atom whole and is found whole, at one place. In between, the only rule that has ever predicted where is one that counts every way it could have gone.
Chapter 15 · Inside the lamp
Every Path
How one thing can spread like a wave and arrive like a grain. The answer is a rule for adding arrows, and you have been using it since chapter 1.
Look first
A lamp, a mirror, a detector
A small lamp shines down on a flat mirror. Off to one side sits a detector, with a screen between the two so that only reflected light can reach it. Everyone learns the law: light leaves a mirror at the angle at which it arrived. Draw the one path that obeys the law, and it touches the mirror at the middle. The rest of the mirror, it seems, does nothing.
This bench is tiny, so that what follows is large enough to see. The mirror is 48 wavelengths long, about a fortieth of a millimetre for green light. On a mirror you could hold, everything in this chapter still happens, on a much finer scale.
The bright path is the one the law of reflection allows. The faint ones are other ways the light might go, which the law rules out.
Commit to a prediction
The useless end
Cover the left quarter of the mirror with black paper. It is nowhere near the place where the light bounces.
Does the detector notice?
The detector’s reading, as a share of what the whole mirror gives, against where the paper’s edge lies. The dot marks a quarter of the mirror covered.
With the left quarter covered the detector reads 84% of what it did. Slide the paper’s edge back and forth and the reading swings between 83% and 105%. At some positions, covering part of the mirror makes the detector brighter.
No picture of light bouncing at one point can explain that. The picture that can is older than it sounds, and you met it in chapter 1.
Why the first guesses made sense
An arrow for every path
If you said the end does nothing, you were nearly right about how much it adds: by itself, under one percent. If you said a quarter less, you were right that every part of the mirror takes part. Both are caught by one rule, in three steps.
One. List every way the light could get from the lamp to the detector. Here that is by way of every point of the mirror, not only the middle.
Two. For each way, draw a small arrow. All the arrows are the same length. An arrow’s direction is set by the length of its path: it turns once round for every wavelength travelled.
Three. Lay the arrows head to tail. Draw one arrow from the start of the chain to its finish, and square its length. That is the brightness at the detector.
Below is the chain for the whole mirror. Near the middle, neighbouring paths are almost exactly the same length, because the shortest path is there and a curve is flat at its lowest point. Their arrows agree, and the chain runs straight. Toward the ends the path length changes quickly, each arrow points a little further round than the last, and the chain curls into a coil that goes nowhere.
So the middle does nearly all the work, and that is the law of reflection. But the ends are not idle. Cover part of a coil and its loose end is left wherever it happened to stop, a little nearer the finish or a little further. That is the wobble you just saw.
Above: one arrow for each path, drawn under the point of the mirror it goes by. Below: the same arrows head to tail. The long arrow from start to finish is what the detector reads.
A case you have not seen
Scratching the mirror
Now keep only the left quarter and cover all the rest. Alone it sends the detector under one percent of what the whole mirror did: its arrows go round in circles.
Take a fine point and scratch away the silver wherever an arrow points into the left half of the dial. Leave it wherever an arrow points into the right half. Half of what little mirror remained is now gone.
What does the detector read?
The left quarter, scratched. Every arrow that is left leans the same way, and the chain that was a coil runs nearly straight.
You have made a diffraction grating: a surface ruled with fine lines that sends light where no mirror would. The lines the rule asks for are not evenly spaced. They close up toward the end of the mirror, from 1.8 wavelengths apart to 1.1, and that spacing is exactly what the textbook formula for a grating requires at each place.
It is made for one colour. Light with a wavelength a fifth longer finds the stripes in the wrong places, and the detector reads 4% in place of 55%. That light goes somewhere else. A grating sorts colours by direction, which is the rainbow on the back of a compact disc.
Scratch both ends this way and throw the middle of the mirror away entirely, and the detector reads 2.2 times what the whole perfect mirror gave.
The bench
The mirror, piece by piece
The mirror is cut into twelve pieces. Press a piece to cover it, press again to scratch it, and again to restore it.
Paths, arrows and their sum
Things to try
- Cover the middle four pieces. How much light is left? Now scratch the ends.
- With the whole mirror open, cover one piece at a time from the left. Which single piece, covered, makes the detector brighter?
- Scratch the ends, then change the colour of the light. Where has the light gone?
What you can now say
Light does not take one path. To find where it will arrive, add an arrow for every path it could take, each turned once round for every wavelength of its length. Where the arrows agree there is light, and where they cancel there is none. For a single grain of light, the same sum gives the chance that it arrives.
A straight ray, the law of reflection and the focus of a lens are not rules that light obeys along the way. They are the places where the arrows agree.
For the determined reader: the sum, the spiral, and the limits
The sum
A path by way of the point x of the mirror has length L(x). Its arrow is the complex number e2πiL(x)/λ, and the detector reads the squared size of their sum over the uncovered mirror:
brightness = | ∫ e2πiL(x)/λ dx |²
Near the shortest path L(x) grows as the square of the distance from it, and the chain of arrows is the curve called Cornu’s spiral. For a long mirror the sum settles to a value fixed by the middle alone: the wavelength divided by how sharply L(x) curves there. On this bench that value is 26.0, in units of the mirror’s length in wavelengths. The bench’s short mirror gives 29.5, because its ends still count.
The grating
Where the path length changes by one wavelength over a distance d along the mirror, the arrows make one full turn. That distance satisfies d (sin i − sin o) = λ, with i and o the angles of the incoming and outgoing paths from the upright. This is the grating equation, and the scratches found by the rule obey it to within a quarter of one percent.
Where this model stops
The mirror is a line, not a surface. The second direction multiplies every sum by the same factor and changes none of the ratios. Every arrow has the same length: the model ignores the way light weakens with distance and with the angle at which it leaves a surface. Light is treated as having no direction of vibration, which chapter 4 showed it has; with polarisation each path carries two arrows, not one. And the lamp is a point of one colour. A wide or white lamp blurs the coils, as chapter 3 would lead you to expect, and leaves the middle.
One more case
The pinhole again
Go back to the camera of chapter 1: a round hole, a screen 50 millimetres behind it, green light from a distant lamp. You found there that a hole exactly two zones wide leaves darkness at the centre of the lamp’s image. The reason can now be said in a line. The paths through the outer ring are half a wavelength longer than the paths through the middle, so the two sets of arrows point opposite ways, and cancel.
Make the hole eight zones wide, 0.94 millimetres. The centre is dark again: four rings cancel the other four. Now paint out every second ring with ink. Half of the hole is covered.
What happens at the centre of the image?
The open hole, eight zones wide. Eight arrows, four up and four down: nothing at the centre.
Every second zone blocked. Four arrows, all up. Both pictures are 0.6 mm across and share one brightness scale.
Each open ring adds an arrow twice as long as the light would have with no card at all. Four that agree make an arrow eight times as long, and the brightness is its square: 64 times the unobstructed lamp. The bright spot is 0.032 millimetres wide, less than half the width of the spot from a one-zone pinhole, which gives 0.089. You have made a lens out of nothing but ink. It is called a zone plate, and it is how X-rays are focused, since glass hardly bends them.
Now turn the lamp down until one grain of light passes at a time. Each lands at one point of the screen, and nothing in the sum says where. What the sum gives is the chance for each place.
From least time to every time
In 1662 Pierre de Fermat proposed that light going from one point to another takes the path that needs the least time. It gave the right law for light entering glass or water, and it troubled people, because it seemed to say that light knows in advance where it is going and compares the routes.
The wave picture of Huygens and Fresnel removed the mystery for bright light. A wave goes every way at once, and the path of least time is simply where its parts arrive in step. That was the sum of chapter 1.
The grain of light brought the mystery back. In 1933 Paul Dirac wrote a paper suggesting that the quantity classical physics minimises might enter quantum mechanics as the turning of an arrow. Richard Feynman heard of the paper at a beer party in the Nassau Tavern in Princeton, from Herbert Jehle, newly arrived from Europe, and the two read it in the library the next day. Dirac had written that two quantities were analogous. Feynman, by his own account, asked what use that was, tried making them equal, and found that the established equation of quantum mechanics came out. Beginning in that doctoral work and finishing in 1948, he made it into a complete method: add an arrow for every history a thing could have. It became one of the foundations of the theory of light and electrons, for which he shared the Nobel Prize of 1965. In 1983 he explained it to a general audience in four lectures in Los Angeles, with no mathematics beyond the adding of arrows. They were published in 1985 as QED.
Where this explanation stops
Between the lamp and the click
The rule tells you what to compute, and it has never been caught out. It does not tell you what the light is doing between the lamp and the detector, and it rules out the two obvious pictures. A grain that went by way of the middle of the mirror could not care whether the far end was covered. A wave spread over the whole mirror could not arrive, whole, at one point of a screen.
Physicists have used this rule for most of a century without agreeing on what it describes. What is light, between the lamp and the click?
End of chapter 15
Added to your notebook
To find where light will arrive, add an arrow for every path it could take, each turned once round for every wavelength of its length. Where the arrows agree there is light. For one grain, the sum gives the chance that it arrives.
An arrow for every path. But arrows of what? What is it that light is a ripple in?
Chapter 16 · The Field