Rationals Aren't Enough — Even for Practical Work

You have probably thought: every measurement I make rounds off to a decimal. A carpenter's tape reads 1.41 m, not \sqrt{2} m. An engineer's CAD file stores 3.14159265, not \pi. A computer's float is always rational. So in "practical" life, rationals seem to do the whole job. Why does the real number line need irrationals at all?

The honest answer is that the moment practical work crosses from counting into geometry, calculus, music, or signal processing, rationals stop being closed under the operations you need. You are not free to round at the end — the rounding changes the theorem. Here are the places where this actually bites.

1. Geometry demands \sqrt{2} the moment you draw a square

Put a unit square on graph paper. The side is 1 m. The diagonal, by Pythagoras, satisfies d^2 = 1^2 + 1^2 = 2, so d = \sqrt{2}. Now try to express d as a ratio p/q of integers.

Suppose p/q = \sqrt{2} with \gcd(p, q) = 1. Then p^2 = 2q^2, so p^2 is even, so p is even. Write p = 2k: then 4k^2 = 2q^2, i.e. q^2 = 2k^2, so q is even too. But p and q were supposed to share no factors. Contradiction. Why: the argument forces a factor of 2 into both p and q, violating the lowest-terms assumption. This proof is Euclid's, from around 300 BCE.

So no fraction p/q equals the diagonal of a unit square. Yet the diagonal physically exists — you can draw it with ruler and compass.

Rational-only geometry would have to declare: "the diagonal has no length." That is absurd. The number system has to contain the point, so we extend to include \sqrt{2}.

2. Rationals are not closed under the square root

A number system is only "enough" if the operations you need keep you inside it. The rationals are closed under addition, subtraction, multiplication, and non-zero division — those stay rational. But:

• Square root — \sqrt{2}, \sqrt{3}, \sqrt{5}, \sqrt{6}, \sqrt{7}, \sqrt{8}, \sqrt{10}, \ldots are all irrational. The square roots that are rational are rare: \sqrt{n} is rational only when n is a perfect square. In a list of the first 100 integers, only 10 have rational square roots.
• Logarithm — \log_{10}(2) is irrational. So is \log_{10}(3), \log_{10}(5), \log_2(3), and almost every logarithm you will ever compute. The exceptions are the trivial ones like \log_{10}(100) = 2.
• Exponential — e^1 = e = 2.71828\ldots is irrational. So is 2^{1/12}, the frequency ratio of a semitone in equal temperament.
• Trigonometric functions — \sin(30°) = 1/2 happens to be rational, but \sin(1°), \cos(1°), \tan(1°) are all irrational. Pick any random angle in degrees and its sine is almost surely irrational.

So every time a calculator or scientific formula involves \sqrt{\,\,}, \log, \exp, or a non-trivial trig value, the exact answer leaves the rationals. You can round to 16 decimals for display, but the equation is about an irrational number.

3. Music theory: the twelfth root of 2

Take a real example: why do piano keys sound "in tune" across keys? A semitone (say C to C#) is a frequency ratio of r, and 12 semitones make an octave (ratio 2). So r^{12} = 2, which gives

This number is irrational — in fact, it is transcendental (not a root of any polynomial with rational coefficients, a harder property than just "not a fraction"). Why: if r = p/q were rational, then r^{12} = p^{12}/q^{12} = 2 would say p^{12} = 2 q^{12}. The number of 2s in p^{12} is a multiple of 12; in 2q^{12} it is one more than a multiple of 12. These can never match.

Piano tuners use 2^{1/12} every day. They cannot round it off and keep the octaves pure. The equal-tempered scale lives in the irrationals, not the rationals. A "rationals-only" music theory would force you to pick a key and detune every other key — the compromise that Baroque composers actually made before equal temperament was adopted.

4. Calculus needs completeness, which rationals lack

Calculus is the mathematics of limits. A limit is what a sequence is "trying to approach." Consider the rational sequence

Every term is a terminating decimal, so every term is rational. The terms get closer and closer — their differences shrink to zero. In a complete number system, they converge to a limit. That limit is \sqrt{2}. In the rationals alone, this sequence has no limit — the point it is approaching is not in the system.

This is not a nitpick. The intermediate value theorem says a continuous function that is negative at x = a and positive at x = b must be zero somewhere in between. Take f(x) = x^2 - 2 on \mathbb{Q}: it is -1 at x = 1 and +2 at x = 2, continuous, and never zero for any rational input. The theorem fails on the rationals. Every serious calculus theorem — IVT, Bolzano–Weierstrass, extreme value theorem, convergence of Cauchy sequences — needs the number line to have no holes. The rationals have holes; the reals don't. Why this matters practically: numerical methods like Newton's method and bisection rely on IVT to guarantee that the root they are hunting for actually exists. Without completeness, those methods would converge to nothing.

5. GPS, signal processing, and physics are irrational by nature

Modern engineering sits on transcendental constants. GPS clock corrections use \sqrt{1 - r_s/r} — an irrational square root. The Fourier transform decomposes a signal into sines and cosines at frequency \omega = 2\pi f: that 2\pi is irrational, and the discrete Fourier transform uses e^{2\pi i / N}, whose components are \cos(2\pi/N) and \sin(2\pi/N) — irrational for almost all N. A pendulum's period T = 2\pi\sqrt{L/g} and radioactive decay N(t) = N_0 e^{-\lambda t} both involve irrational constants.

When an ISRO engineer computes a launch trajectory, every constant in the equations is \pi, e, a square root, or a logarithm. The final number sent to the thruster is rounded to a rational (hardware accepts finite bits), but the theory that produced it lives in the reals.

Rationals vs reals: the trade-off

The rationals give you computational exactness — you can write \tfrac{22}{7} and that is what it is. The reals give you algebraic and geometric completeness — every sensible equation has a solution, every limit exists, every continuous curve crosses where it should. No single number system gives you both perfectly. But when you ask "which is more useful for mathematics as a whole?", the answer is unambiguous: the reals, because without completeness, geometry and calculus simply do not work.

So when can you live in the rationals?

Some domains genuinely do. Number theory, combinatorics, elementary arithmetic, and pure algebra over \mathbb{Q} all stay inside the rationals — because their questions are about divisibility and ratios, not about lengths, limits, or continuity. The rationals are a perfectly good home for them.

But the moment you bring in a Pythagorean triangle that isn't 3–4–5, or take a square root, or compute a compound interest with a continuous rate, or ask for the period of a pendulum, or tune a piano — you leave \mathbb{Q} whether you like it or not. The irrationals aren't a luxury tacked on for pure-maths aesthetes. They are the points at which the answers to practical questions actually sit.

Related: Why We Need Irrational Numbers · Irrationals Between the Rationals · Real Numbers — Properties · The Rational Line Has Holes, the Real Line Does Not