Appearance
11.12 — Why the Same Few Numbers Run the Universe
Give the same exam to a thousand students and draw a chart of their marks. A few students do very badly and a few do very well. Most land near the middle. The chart makes a smooth hump, high in the middle and falling away on both sides. It is called the bell curve, because of its shape. People's heights make the same shape. So do the small errors in any careful measurement.
Chapter 7.6 gives the formula for the height of that hump. In its simplest form, where x measures how far a mark is from the middle, it is:
\text{height at } x = \frac{1}{\sqrt{2\pi}}\,e^{-x^2/2}
Read it aloud: "one over the square root of two π, times e to the power minus x squared over two". You do not need to use it. Just look at what is inside it.
- There is π, the number of circles.
- There is e, the number of growth.
- There is \sqrt2, hidden inside \sqrt{2\pi}. The square root of two things multiplied is the square roots multiplied, so \sqrt{2\pi} = \sqrt2 \times \sqrt\pi.
Yet the question was about marks in a classroom. There is no circle in a classroom, and nothing is growing.
The last six chapters met these numbers one at a time. They came from different centuries and different countries, and from questions that had nothing to do with each other. A Babylonian scribe measured a square. Archimedes measured a circle. Bernoulli asked about a bank. Cardano played with an impossible equation. Euclid cut a line to draw a pentagon. And still, again and again, they turn up inside each other's formulas, and inside the laws of physics.
This page explains why. The explanation has four parts. First, these numbers are forced on us, not chosen. Second, each one answers the simplest possible question of its kind. Third, they are linked by exact formulas that we have now derived. Fourth, nature uses them because nature keeps doing the same few simple things.
1. These numbers are forced, not chosen
Imagine a civilisation on a planet a thousand light years away. Its people count in base 8, because they have eight fingers. They have never heard of Greece or India or Babylon.
They would still find π. Their digits would look different, because they write numbers in base 8. But the number would be the same, because any circle, anywhere, has the same ratio of distance round to distance across. In Chapter 11.7 we saw that this follows from the fact that all circles have the same shape. Nobody decided it. Nobody could change it. The Indiana legislature tried in 1897, as Chapter 3.3 tells, and failed.
The same is true of every number in Chapters 11.6 to 11.11.
- Every square, anywhere, has a diagonal \sqrt2 times its side.
- Every quantity whose growth equals its size grows like e^x.
- Every quarter turn, done twice, is a half turn, so i exists in any system of numbers that can turn.
- Every regular pentagon has a diagonal \varphi times its side.
These are called mathematical constants. Their values are fixed by logic alone, once you have said what you are talking about. They would be the same in any universe, with any laws of physics, or with no physics at all.
2. Each one answers the simplest question of its kind
Why are there so few important constants? Mathematics has endless possible numbers. Why do the same five keep coming back?
Because each one is the answer to the simplest possible question in one area of mathematics. Here they are side by side.
| Area | The simplest question you can ask | The answer |
|---|---|---|
| Lengths | How long is the diagonal of the simplest shape, a square? | \sqrt2 |
| Curves | How far round is the simplest curved shape, a circle? | π |
| Change | What grows at a rate exactly equal to its own size? | e |
| Direction | What, done twice, gives a half turn? | i |
| Proportion | What cut makes the whole relate to the part as the part relates to the rest? | \varphi |
| Adding up | How far does the simplest endless sum, 1 + \frac12 + \frac13 + \cdots, run ahead of the area under \frac1x? | \gamma |
Simple questions get asked over and over, by everyone, in every subject. Any calculation that is complicated enough will, somewhere inside it, ask one of these simple questions. When it does, the matching constant appears.
Each one is something an action leaves unchanged
There is a second way to see the same thing. Each of these numbers is a fixed point. That means it is something that a simple action leaves exactly as it was.
- Take a guess x and replace it by the average of x and \frac2x. The one number this action does not move is \sqrt2 (Chapter 11.6).
- Take a number x and replace it by 1 + \frac1x. The one positive number this does not move is \varphi (Chapter 11.10).
- Take a function and replace it by its rate of change. The one function this does not change is e^x (Chapter 11.8).
- Take any number and multiply it by i four times. It is left exactly as it was, because four quarter turns make a full turn. i is the simplest action, other than doing nothing, that comes back to the start when repeated (Chapter 11.9).
- Take a point and turn it round a circle by a full turn. It comes back exactly where it started. The size of that full turn, measured along a circle of radius 1, is 2\pi (Chapter 11.7).
Simple actions are repeated everywhere: averaging, dividing, taking a rate of change, turning. Whatever a repeated action leaves unchanged is what you are left with at the end. That is why these numbers are where so many calculations end up.
3. The web: every number is tied to the others
The second surprise is that these numbers are not separate. They are joined by exact formulas. Every link below has been derived in this book, most of them in the last six chapters.
Here is how to read the map. Each circle is a number, or in one case the prime numbers. Each line is an exact formula containing both ends. The numbers on the lines match the list below.
- \gamma and e. 1 + \frac12 + \cdots + \frac1n \approx \ln n + \gamma. The natural logarithm is the logarithm to base e. Derived in Chapter 11.11.
- e and π. \int_{-\infty}^{\infty} e^{-x^2}\,dx = \sqrt\pi. The area under the bell-shaped curve built from e is the square root of π. Derived in Chapter 5.7, and explained in Chapter 11.7 as a hidden circle.
- e, i and π. e^{i\pi} = -1. Growth in a sideways direction is turning, and π of it is a half turn. Derived twice, in Chapter 5.8 and in Chapter 11.9.
- i and \sqrt2. \sqrt i = \frac{1+i}{\sqrt2}. Half of a quarter turn is 45°, the angle of a square's diagonal. Derived in Chapter 11.9.
- \sqrt2 and π. Viète's product, derived just below.
- \varphi and π. \varphi = 2\cos\frac{\pi}{5}. Derived just below.
- The primes and π. The chance that two random whole numbers share no common factor is \frac{6}{\pi^2}. Explained in Chapter 11.7, from Euler's prime product in Chapter 11.1.
- The primes and e. The number of primes below n is about \frac{n}{\ln n}. This is the prime number theorem, which Chapter 11.4 describes and Chapter 11.P checks with real counts.
Two of these links have not been derived yet. Here they are.
Link 5: square roots of 2 that build π
In 1593 the French lawyer and mathematician François Viète published the first infinite product in history. It builds π out of nothing but 2s and square roots:
\frac{2}{\pi} = \frac{\sqrt2}{2} \cdot \frac{\sqrt{2 + \sqrt2}}{2} \cdot \frac{\sqrt{2 + \sqrt{2 + \sqrt2}}}{2} \cdots
Read aloud: "two over pi equals root two over two, times root of two plus root two, over two, times..." and so on forever. Each new factor puts one more "two plus root" inside the last.
Before the steps, recall what sine and cosine are (Chapter 11.9). Walk round a circle of radius 1 through some angle. The cosine is how far across you end up, and the sine is how far up.
The proof uses one rule from Chapter 3.5, called the double-angle rule:
\sin 2\theta = 2\sin\theta\cos\theta
Read it as: "the sine of twice an angle equals two times the sine of the angle times its cosine". Check it with numbers. Take \theta = 30°. Then \sin 60° = 0.866. And 2 \times \sin 30° \times \cos 30° = 2 \times 0.5 \times 0.866 = 0.866. It matches.
Now here is Viète's product, step by step.
Step 1. Use the rule backwards. Write any angle x as twice its half:
\sin x = 2\sin\frac{x}{2}\cos\frac{x}{2}
Step 2. Do it again to the new sine. \sin\frac{x}{2} = 2\sin\frac{x}{4}\cos\frac{x}{4}. Put that in:
\sin x = 4\sin\frac{x}{4}\cos\frac{x}{4}\cos\frac{x}{2}
Step 3. Keep going n times. Each time, the front number doubles, the angle inside the sine halves, and one more cosine joins the line:
\sin x = 2^n \sin\frac{x}{2^n}\;\cos\frac{x}{2}\cos\frac{x}{4}\cdots\cos\frac{x}{2^n}
Step 4. Let n grow without end. Here we need one fact. For a tiny angle, measured in radians, the sine is almost exactly the angle itself. Check it. \sin 0.1 = 0.09983, and \sin 0.01 = 0.0099998. The picture explains why. For a tiny angle, you have walked only a tiny way round the circle. That short walk is almost a straight line going straight up. So the distance walked (the angle) and the height reached (the sine) are almost the same.
As n grows, the angle \frac{x}{2^n} becomes tiny. So \sin\frac{x}{2^n} is almost exactly \frac{x}{2^n}. Then 2^n \times \sin\frac{x}{2^n} is almost exactly 2^n \times \frac{x}{2^n} = x. In the limit, the front of step 3 becomes just x. Divide both sides by x:
\frac{\sin x}{x} = \cos\frac{x}{2}\cdot\cos\frac{x}{4}\cdot\cos\frac{x}{8}\cdots
Step 5. Choose x = \frac{\pi}{2}, a right angle. Then \sin x = 1, and the left side is \frac{1}{\pi/2} = \frac{2}{\pi}:
\frac{2}{\pi} = \cos\frac{\pi}{4}\cdot\cos\frac{\pi}{8}\cdot\cos\frac{\pi}{16}\cdots
Step 6. Find each cosine with square roots. We need a rule that gives the cosine of half an angle. It takes four small moves.
- Chapter 2.5 gives a double-angle rule for the cosine: \cos 2\theta = \cos^2\theta - \sin^2\theta. (Here \cos^2\theta means \cos\theta \times \cos\theta.)
- The point you reach on the circle is 1 away from the centre. By Pythagoras, across squared plus up squared is 1: \cos^2\theta + \sin^2\theta = 1. So \sin^2\theta = 1 - \cos^2\theta. Put that into move 1: \cos 2\theta = \cos^2\theta - (1 - \cos^2\theta) = 2\cos^2\theta - 1.
- That rule works for any angle, so use it for half an angle. Twice half an angle is the whole angle: \cos\theta = 2\cos^2\frac{\theta}{2} - 1.
- Rearrange to get the half-angle on its own. Add 1 to both sides: 1 + \cos\theta = 2\cos^2\frac{\theta}{2}. Multiply both sides by 2: 2 + 2\cos\theta = 4\cos^2\frac{\theta}{2}. The right side is \left(2\cos\frac{\theta}{2}\right)^2. Take square roots: 2\cos\frac{\theta}{2} = \sqrt{2 + 2\cos\theta}.
Now start from a right angle. Its cosine is 0, because after a quarter turn you are straight up, zero across. Then halve again and again:
- 2\cos\frac{\pi}{4} = \sqrt{2 + 0} = \sqrt2
- 2\cos\frac{\pi}{8} = \sqrt{2 + 2\cos\frac{\pi}{4}} = \sqrt{2 + \sqrt2}
- 2\cos\frac{\pi}{16} = \sqrt{2 + \sqrt{2 + \sqrt2}}
Divide each by 2 and put them into step 5. That is Viète's product.
Watch it close in. After one factor, \frac{2}{\pi} gives π as 2.828. After two, 3.061. After three, 3.121. After seven, 3.14151.
Look at those numbers again. 2.828 is the half-perimeter of a square inside a circle of radius 1. 3.061 is the same for an eight-sided polygon. Viète's product is Archimedes' doubling of polygons from Chapter 11.7, written as one formula. The square roots of 2 appear because the first polygon is a square, and a square's diagonal is \sqrt2.
Link 6: the golden ratio is a cosine of π
Chapter 11.10 found the golden triangle inside a pentagon. It has angles 36°, 72°, 72°, and long sides \varphi when its short side is 1. Inside it was a smaller triangle, ACF, with two sides of length 1 and a long side AC = \varphi. Its two base angles are both 36°.
Drop a line from F straight down to the middle of AC. It cuts ACF into two right triangles. Each has a long side of 1, an angle of 36°, and a base of half of AC, which is \frac{\varphi}{2}. In a right triangle, the base next to an angle equals the long side times the cosine of that angle (Chapter 3.5). So:
\frac{\varphi}{2} = 1 \times \cos 36°
And 36° is a tenth of a full turn, which is \frac{2\pi}{10} = \frac{\pi}{5} radians. So:
\varphi = 2\cos\frac{\pi}{5}
Check: \cos 36° = 0.80902, and twice that is 1.61803. The golden ratio is a measurement of a circle, because the pentagon is the shape you get by marking five equal steps around a circle.
4. Why nature uses these numbers
We have seen why mathematics keeps meeting these numbers. The harder question is why nature does. Why should a pendulum, a bacterium, a radio wave and a sunflower care about questions asked by Greek geometers and an Italian gambler?
Chapter 11.5 told the story of Eugene Wigner's 1960 essay on "the unreasonable effectiveness of mathematics". Nobody has fully answered him. But for these particular numbers, a large part of the answer is plain. Nature keeps doing a few simple things over and over, and each simple thing has its own constant.
Nature changes according to how things are now
Almost every law of physics has the same form. It says how fast something changes, based on how it is right now. Newton's law of motion is like that. So are the laws of heat, of electricity, and of chemical reactions. Chapter 6.1 calls such laws differential equations.
The simplest such law is "change in proportion to the amount". Its answer is e^{kt} (Chapter 11.8). That is why e appears in radioactive decay, in cooling, in drug clearance and in population growth.
The next simplest law is about something that gets pulled back. Pull a weight on a spring down, and the spring pulls it back up. The further you pull, the harder it pulls back. Newton's law says the pull changes the weight's speed, and the speed changes its position. So here the rule is: "the rate of change of the rate of change is a fixed number times the position, with a minus sign", because the pull is always back towards the middle.
Which motion obeys that? Try e^{ikt}, the turning arrow of Chapter 11.9. Each time you take its rate of change, you multiply it by ik. In Chapter 11.8 we saw that e^{kt} changes at k times itself, and the same works with ik. Do it twice and you have multiplied by ik \times ik = i^2 k^2 = -k^2. The result is a fixed number with a minus sign, exactly what the spring needs. So the spring's motion is a turning arrow, and its up-and-down shadow is a sine wave. Every vibration, every sound and every light wave follows this same law. And a sine wave carries π, because it is the shadow of a turning circle. So e, i and π appear together in physics because the two simplest laws of change produce them.
Space looks the same in every direction
Empty space has no special direction. Nothing in physics prefers up over sideways. So anything that spreads out from a point, such as light, sound, or gravity, spreads evenly over a sphere. Spheres bring 4\pi (Chapter 11.7). That is why the laws of electricity and gravity carry a 4\pi when written in their natural form.
Many small random pushes add up to a bell
Your height is the result of thousands of genes and many years of food and sleep, each nudging it a little up or down. In Chapter 7.6 we prove that whenever many small independent pushes add up, the result follows the bell curve, whatever the pushes are. The bell curve's formula contains e and π, as this page began by showing. So any quantity built from many small random causes carries those two numbers.
Growth fills the space that is free
A growing plant tip puts each new bud where there is most room. In Chapter 11.10 we saw that this rule, and nothing else, produces the golden angle. So \varphi appears wherever things grow by adding new pieces into the largest gap.
In short: these numbers are not secret messages and not coincidences. They are the fingerprints of the few simple things the universe does again and again. It changes according to its present state. It turns and vibrates. It spreads out evenly. It piles up random effects. It grows into free space. Each of those actions leaves its own constant behind.
5. Constants of mathematics and constants of physics
Physics has constants too, and they are a completely different kind of number. It is important not to mix them up.
Physical constants with units
The speed of light is 299,792,458 metres per second. The constant in Newton's law of gravity is 6.674 \times 10^{-11} in metre, kilogram and second units. Planck's constant sets the size of quantum effects, the odd behaviour of very small things such as atoms. It is 6.62607015 \times 10^{-34} in the same units.
These numbers depend on our choice of units. Measure the speed of light in miles per hour and you get about 671 million. The light is the same; only the ruler and the clock changed. So the digits of these constants tell you about human history, about the length of the metre and the second, not about nature.
Since 1983 the speed of light has been exact by definition. The metre is now defined as the distance light travels in \frac{1}{299{,}792{,}458} of a second. Planck's constant became exact in the same way in 2019, when it was used to define the kilogram.
Physical constants without units
Some physical constants have no units at all, because they are a ratio of two things measured in the same units. These do not depend on any human choice. A civilisation counting in base 8 would measure the same number.
The proton is about 1,836.15 times heavier than the electron. These are two of the tiny pieces that make up every atom. Nobody knows why the number is this size.
The fine-structure constant is the most famous of them:
\alpha \approx \frac{1}{137.036}
The Greek letter α is "alpha". This number measures how strongly light and electric charge pull on each other. Volume IV, Parts 7 and 8 show how it controls the colour of every atom's light and the strength of every chemical bond. Calculations suggest that if it were a few percent different, stars would struggle to make carbon, and chemistry as we know it might not exist.
Here is the important contrast. π can be derived. α cannot. Nobody has ever found a reason why α is near \frac{1}{137} rather than some other value. It has to be measured. The physicist Richard Feynman called it "one of the greatest damn mysteries of physics: a magic number that comes to us with no understanding by man".
Interestingly, when you write α out in terms of other constants, its formula contains a 4\pi. That 4\pi is the mathematical constant from the surface of a sphere, sitting inside a physical one, for exactly the reason given in section 4.
A warning about numerology
Because α looks so mysterious, many people have tried to "explain" it by playing with π, e and small whole numbers until something matches. The English astronomer Arthur Eddington, one of the finest scientists of his time, argued in 1929 that α was exactly \frac{1}{136}. When better experiments showed it was nearer 137, he produced a new argument for exactly \frac{1}{137}. Today it is measured as 137.035999, which is neither.
With enough freedom to combine numbers, you can match any value to a few digits. That is not an explanation. The constants of mathematics in this Part each came with a derivation: a clear chain of reasoning from a simple question to the number. A formula that fits but has no such chain is a coincidence until proven otherwise.
The three kinds side by side
| Kind | Example | Where its value comes from | Same in another universe? |
|---|---|---|---|
| Mathematical | π, e, \sqrt2 | logic alone | yes |
| Physical, no units | α, proton-to-electron mass ratio | measurement | nobody knows |
| Physical, with units | speed of light, gravity's constant | measurement and our choice of units | the digits are not even fixed here |
Whether the unit-free physical constants could have been different is one of the deepest open questions in science. Some physicists hope a future theory will derive them the way we derive π. Others think they are simply accidents of our universe. Volume VII, Part 15 looks at what that question means for the bigger questions people ask about why the universe is the way it is.
6. Where this shows up in your life
The phone in your pocket uses all of them at once. When it finds your position by GPS, it is doing every chapter of this Part together.
- The satellites move in orbits whose timing carries π (Chapter 11.7).
- Their radio signals are turning arrows, written with e^{i\theta} (Chapter 11.9).
- The distance to each satellite comes from the speed of light, a physical constant, times the travel time of the signal.
- The phone turns those distances into a position using Pythagoras, square roots of sums of squares, the same arithmetic as \sqrt2 (Chapter 11.6).
- The satellite clocks are corrected for Einstein's relativity by about 38 microseconds a day. Without that correction, your position would drift by about 10 kilometres a day. Volume IV, Chapter 6.9 computes it.
- The phone's estimate of how sure it is about your position follows the bell curve, with its e and π (Chapter 7.6).
A Babylonian square, a Greek circle, a Swiss question about interest, an Italian puzzle and a Danish surveyor's map all meet inside one small blue dot on your screen.
Next: 11.P — Worked Problems puts the great results of this Part to work with real numbers.