Picture Perfect Proof. How square root of 2 is irrational

We can view proofs from different perspectives as well as people

“Every being cries out silently to be read differently.”

Simone Weil, quoted in Francis Su, Mathematics for Human Flourishing

The goal for this proof is that in one picture, you will be convinced beyond a shadow of a doubt that $\sqrt{2}$ cannot be expressed with a finite number of digits.

Two quick lenses before the proof

1) A small paradox. Our goal is to prove $\sqrt{2}$ is irrational. But if a number is irrational, we can’t express it as a ratio of whole numbers. So how do we prove something we can’t “write down” as a neat fraction? Rather than chasing digits, we’ll change the picture.

2) Change dimensions. We do know a defining truth about $\sqrt{2}$: it’s the number whose square is 2. That’s a 2-dimensional clue. So rather than thinking about one-dimensional lengths, we’ll think in areas—with squares. Let $B$ be the side of a big square and $s$ the side of each small square. If $\sqrt{2}$ were a ratio of whole numbers $\left(\frac{B}{s}\right)$, then $\left(\frac{B}{s}\right)^2 = 2$, i.e., $B^2 = 2\cdot s^2$. That translates cleanly into a question about areas of squares.

A quick warm-up (why “double-covered = not-covered” is natural)

Take two identical $5\times 5$ squares and place them side-by-side to make a $5\times 10$ rectangle. Same pieces, same total area.

Now slide the left $5\times 5$ square right, say by 1 unit inside that rectangle:

• On the left edge you create a not-covered strip that’s $1\times 5$.
• In the middle you create a covered twice strip that’s also $1\times 5$.

You can also think of it as not counted area and double counted area. Either way, because the total area hasn’t changed, the covered twice area equals the not-covered area. This simple “slide principle” is the only engine we’ll need when we move to squares.

When a Square Speaks

We’ll show—by picture alone—that no matter what whole numbers you choose, two identical small squares can never have exactly the same total area as one larger square. That kind of equality is easy if the larger shape is a rectangle (as in the warm-up). It becomes truly interesting—and decisive—when we require the larger shape to be a square. From there, it will be almost immediate that $\sqrt{2}$ can’t be a ratio of whole numbers.

We’ll do this in two parts:

1. Geometric descent: assume such an equality exists; the picture manufactures a smaller equality of the same kind—then smaller again, and so on. That’s impossible in the positive whole numbers.
2. Equivalence to $\sqrt{2}$: why “two equal smalls = one big” would force the side-length ratio $\frac{B}{s}$ to be $\sqrt{2}$—so the impossibility above is exactly “$\sqrt{2}$ is irrational.”

Part 1 — The geometric descent (the jigsaw that shrinks)

Imagine—just for contradiction—that you’ve found whole numbers where the areas match:

• two identical small squares of side 5 (area 25 each, total 50), and
• one big square whose side is some whole number $W$ with area $W^2 = 50$.

(We know there’s no such $W$, but the point is to assume there is and let the picture do the work.)

Place the two $5 \times 5$ squares inside the big $W \times W$ square—one tucked into the top-left corner, the other into the bottom-right—so they overlap in the center.

Look at what the placement creates:

• Two uncovered corner squares (top-right and bottom-left).
• One central overlap square (the region covered twice by the smalls).

By the slide principle, double-counted area = not-counted area. In this square setting, those regions are themselves squares (by symmetry and alignment), so we get:

(central overlap area, counted once) = (sum of the two uncovered corner areas).

Even better, their side lengths are integer differences you can read off the picture:

• Each uncovered corner square has side $W − 5$ (big side minus small side).
• The central overlap square has side $(5 + 5) − W = 10 − W$ (two small sides minus the big).

So from our assumed example $[(5, 5) → W]$ we’ve produced a smaller example of the same kind:

two identical small squares of side $(W − 5)$ equal in area to one big square of side $(10 − W)$.

Why “smaller”? Since $W$ must lie strictly between $5$ and $10$ (otherwise the two $5\times 5$’s wouldn’t fit and overlap as arranged), both $W − 5$ and $10 − W$ are positive whole numbers smaller than $5$ and $W$, respectively.

Thus the same “two smalls = one big” pattern appears at a smaller scale—again with whole-number sides. Repeat the placement with these new squares and you get an even smaller trio; repeat, repeat…

But the positive whole numbers cannot descend forever. You can’t keep producing strictly smaller positive whole numbers without end. That contradiction means our very first assumption was wrong:

No two identical whole-number squares can have the same total area as one larger whole-number square.

(And nothing was special about choosing side $5$; the same descent traps any supposed whole-number example.)

Repeat once more

Part 2 — Why that is exactly “$\sqrt{2}$ is irrational”

If such an equality of areas ever existed with whole-number side lengths—two smalls (side $s$) and one big (side $B$)—it would say:

$B^2 = 2\cdot s^2$

Divide by $s^2$ and take square roots:

$\dfrac{B}{s} =\sqrt{2}$

But $\frac{B}{s}$ would then be a ratio of whole numbers—i.e., rational—equal to $\sqrt{2}$. That contradicts the geometric descent we just proved. Therefore:

$\sqrt{2}$ cannot be written as a ratio of whole numbers. $\sqrt{2}$ is irrational.

Pause and notice (let the picture do its work)

Before we move on, let's sit with the picture so the entire story is visible: two identical $5\times 5$ squares tucked into one larger $W\times W$ square, overlapping in the middle and leaving two uncovered corners. You now know that with whole-number side lengths, this arrangement can never make “two small areas = one big area.” The jigsaw forces a smaller copy of the same setup, and whole numbers can’t descend forever.

That single visual fact already implies the headline: if equality of areas ever held, we’d have $B^2 = 2\cdot s^2$, so $\frac{B}{s} =\sqrt{2}$ would be a ratio of integers—rational. Our picture shows that can’t happen. So $\sqrt{2}$ is irrational. Can you truly see why $\sqrt{2}$ MUST be irrational from this picture alone?

A glimpse of the near-hits (full story in the Epilogue)

Consider this chain of fractions:$\frac{3}{2}, \frac{7}{5}, \frac{17}{12}, \frac{41}{29},\dots$ Each fraction approaches the true value of $\sqrt{2}$, alternating above/below. Square them and you get near-misses for 2: high, low, high… ever tighter. For one snapshot: $(17/12)^2 \approx 2.0069$. That “almost-equal” rhythm is exactly what our square picture predicts: close forever, never equal.

If you’d like the generator for these best approximations (and an animation), the Epilogue walks it step-by-step.