The claim
is the positive number that multiplies by itself to make 2. It is a little more than 1.4, because , and a little less than 1.5, because .
The claim is that no fraction equals . However the whole numbers a and b are chosen, is never exactly 2. A number that cannot be written as a fraction of whole numbers is irrational, so the claim is that is irrational.
Trying fractions one by one could never prove this, because there are infinitely many of them. The proof is a proof by contradiction: suppose that is a fraction, and show that this leads to something impossible.
First, odd and even squares
The proof needs one fact: an even number squares to an even number, and an odd number squares to an odd number.
An even number is 2 times a whole number, so it can be written 2m. Its square is , which is 2 times a whole number, so it is even.
An odd number is one more than an even number, so it can be written 2m + 1. Its square is , which is one more than an even number, so it is odd.
Turn this round. If is even, then a cannot be odd, because an odd a would give an odd . So if is even, a is even.
Along the top, n runs from 1 to 8. The odd values 1, 3, 5 and 7 have odd squares, and the even values 2, 4, 6 and 8 have even squares.
Suppose is a fraction
Suppose , where a and b are whole numbers. Any fraction can be canceled down until the top and the bottom have no common factor, so choose in lowest terms. In particular, a and b are not both even.
Square both sides: . Multiply both sides by : .
Then a is even
is 2 times a whole number, so is even. By the fact above, a is even too.
Then b is even too
Since a is even, write a = 2k for some whole number k. Put 2k in place of a in : , so . Divide both sides by 2: .
Now is 2 times a whole number, so is even, and so b is even. Both a and b are even. But was chosen in lowest terms, so a and b cannot both be even. That is a contradiction.
The supposition that is a fraction has led to something impossible, so it is false. No fraction equals : is irrational.
For these values of b, misses a square by exactly 1, so the fractions , , and come closer and closer to . The proof shows that never equals a square.
Why lowest terms matters
Without lowest terms, finding that a and b are both even would be no contradiction: has both even. Choosing lowest terms at the start is what makes it one.
The same idea can be seen another way. If a and b were both even, the fraction would cancel by 2 to over , with smaller numbers. The same argument would make those even too, and they would cancel again, forever. Whole numbers cannot be halved forever and stay whole, so no such a and b exist.
Where the argument breaks for
A proof should fail where its claim is false. is a fraction, so try the same steps on it. Suppose in lowest terms. Squaring gives , so is even and a is even. Write a = 2k: then , so .
Here the argument stops. does not say that is even, so nothing forces b to be even, and there is no contradiction. For , the step that made b even was , and the 2 in it came from the 2 under the root.
The diagonal of a square
The problem below uses one fact about any square: the square drawn on its diagonal has twice its area. Call the side s and the diagonal d. Then .
The drawing shows why. Four squares of side s make one big square. Each small square is cut in half by one of its diagonals, and those four diagonals make a tilted square with side d. The tilted square holds 4 half-squares, which is 2 whole small squares. So its area is .
This is Pythagoras' theorem for the case of a square: .
Each of the four small squares has side s and area . The tilted square has the diagonal d as its side, and it is made of four half-squares, so its area is .
Worked example: No Tile Fits Both the Side and the Diagonal of a Square
Question A tiler wants a square courtyard in which one row of identical tiles fits exactly along a side and another row of the same tiles fits exactly along the diagonal. Suppose a tiles fit along a side and b tiles fit along the diagonal, where a and b are whole numbers. (a) Show that this is impossible, whatever the size of the tile. (b) The tiler makes the side exactly 12 tiles long. Between which two whole numbers of tiles does the diagonal lie?
1.Suppose a tiles fit along the side and b along the diagonal. Choose the largest tile that does this, so that a and b have no common factor. By Pythagoras' theorem, b2 = a2 + a2 = 2a2.
With the largest tile, a and b have no common factor. By Pythagoras' theorem, b2 = 2a2. 2.So b2 is even. The square of an odd number is odd, so b is even. Write b = 2c. Then (2c)2 = 2a2, which is 4c2 = 2a2, so a2 = 2c2.
b2 is even, so b is even. With b = 2c, 4c2 = 2a2, so a2 = 2c2. 3.(a) Now a2 is even, so a is even as well. Then a and b have the common factor 2, which contradicts the choice of the largest tile. So no tile fits both the side and the diagonal: ba = √2 is not a fraction.
(a) Then a is even as well, so a and b have the common factor 2. This contradiction shows that no tile fits both. 4.(b) With a side of 12 tiles, the diagonal is d tiles, where d2 = 2 × 122 = 288. Since 162 = 256 and 172 = 289, the diagonal lies between 16 and 17 tiles. It is 12√2 tiles exactly, and the last tile always has to be cut.
(b) d2 = 2 × 122 = 288, and 162 = 256 < 288 < 289 = 172, so the diagonal is between 16 and 17 tiles.
Answer: (a) It is impossible: b2 = 2a2 makes both a and b even, even after every common factor has been removed; (b) d2 = 288, which lies between 162 = 256 and 172 = 289, so the diagonal is between 16 and 17 tiles long
Common mistakes
- Saying that 17 tiles fit the diagonal because 172 = 289 is so close to 288. Close is not equal. The diagonal is slightly shorter than 17 tiles, and part (a) shows that no whole number can ever be exact.
- Leaving out the step that a and b have no common factor. Without it, the fact that both are even is not a contradiction. The common factor is removed first so that finding the factor 2 again is impossible.
The usual mistakes
Taking a calculator's answer as exact. A calculator shows only the first few decimal places of , such as 1.414. That is the fraction , but , which is not 2. Every decimal that stops is only an approximation to .
Getting the direction wrong. The proof needs "if is even, then a is even", and that is justified by odd numbers having odd squares. "If a is even, then is even" is also true, but it is not the step the proof uses.
Leaving out lowest terms. Without it, finding that a and b are both even is not a contradiction.