More Irrationals Than Rationals — In What Sense, Exactly?

Once you hear that irrationals "outnumber" rationals, a perfectly reasonable follow-up is: what are you even counting? Both sets are infinite. Both are dense in \mathbb{R} — between any two rationals sits an irrational, and between any two irrationals sits a rational. So if you walk along the real line looking around, you see both kinds interleaved forever. How can one infinity be "more" than another when both feel equally everywhere?

The honest answer is that mathematicians use two different yardsticks to compare infinite sets, and both of them say the same thing: the irrationals are the bulk of \mathbb{R}, and the rationals are a thin splinter of it. Here are the two yardsticks, precisely.

Sense 1 — cardinality: |\mathbb{Q}| < |\mathbb{R} \setminus \mathbb{Q}|

Cardinality is a way to compare the raw size of two sets without assuming they are finite. Two sets have the same cardinality if you can set up a bijection between them — a one-to-one, onto pairing of their elements. Under this rule:

\mathbb{Q} has cardinality \aleph_0 (aleph-nought). You can list every rational as q_1, q_2, q_3, \ldots using the diagonal walk on the fraction grid, so \mathbb{Q} and \mathbb{N} are in bijection. Sets of this size are called countable. (See Rationals Are Countable, Reals Are Not.)
\mathbb{R} has cardinality 2^{\aleph_0}, also written \mathfrak{c} (the continuum). No list of reals can be complete — Cantor's diagonal argument produces, for any proposed enumeration, a real that is not on the list. Sets of this size are uncountable.
Cantor also proved \aleph_0 < 2^{\aleph_0}. They are different infinities, and the continuum is strictly larger.

Now, \mathbb{R} = \mathbb{Q} \cup (\mathbb{R} \setminus \mathbb{Q}). If the irrationals were also countable, the union of two countable sets would be countable, so \mathbb{R} would be countable — contradiction. So the irrationals are uncountable. Their cardinality equals \mathfrak{c}.

That is the first precise sense of "more": the irrationals form a strictly larger infinity. You can drop every rational into the irrationals with an injection (just name each q), but you cannot go the other way. No matter how cleverly you tag irrationals with rationals, infinitely many irrationals will be left untagged.

A short way to remember it: the rationals and the natural numbers have the same size. The irrationals have the same size as the entire real line. So asking "how many more irrationals than rationals" is like asking "how many more reals than natural numbers" — the gap is an entire jump in infinity, from countable to continuum.

Sense 2 — measure: the rationals fill zero length

Cardinality counts elements. But a more physical question is: if you pick a real number at random — say, spin a pointer over [0, 1] and read off where it lands — what is the probability that you picked a rational?

The answer is exactly zero. The rationals, despite being dense, take up no length at all inside [0, 1]. This is not sloppy language; it is a precise statement called Lebesgue measure zero, and the proof is a calculation you can do by hand.

The \varepsilon-cover argument

Here is the construction that pins it down. Since \mathbb{Q} is countable, list the rationals in [0, 1] as

q_1, q_2, q_3, q_4, \ldots

Pick any positive number \varepsilon > 0 you like — imagine \varepsilon = 0.001 for concreteness. Now cover each q_n with an open interval centred on it, where the n-th interval has length \dfrac{\varepsilon}{2^n}:

q_1 sits inside an interval of length \dfrac{\varepsilon}{2},
q_2 sits inside an interval of length \dfrac{\varepsilon}{4},
q_3 sits inside an interval of length \dfrac{\varepsilon}{8},
and in general q_n sits inside an interval of length \dfrac{\varepsilon}{2^n}.

Every rational in [0, 1] is covered (some rational is q_n for some n, and that n comes with an interval around it). The total length of all the covering intervals is

\sum_{n=1}^{\infty} \frac{\varepsilon}{2^n} \;=\; \varepsilon \cdot \sum_{n=1}^{\infty} \frac{1}{2^n} \;=\; \varepsilon \cdot 1 \;=\; \varepsilon.

Why the sum equals \varepsilon: it is a geometric series with first term \tfrac{1}{2} and common ratio \tfrac{1}{2}, so it sums to \tfrac{1/2}{1 - 1/2} = 1. Multiply through by \varepsilon.

So every rational in [0, 1] fits inside a union of intervals whose total length is \varepsilon. But \varepsilon was arbitrary — take \varepsilon = 10^{-6}, or \varepsilon = 10^{-100}, or smaller. The rationals are contained in covers of length arbitrarily close to zero. The only non-negative number less than every positive \varepsilon is zero itself. So the "length" of \mathbb{Q} \cap [0, 1] is zero.

Each rational $q_n$ gets wrapped in an interval of length $\varepsilon/2^n$. The geometric series sums to $\varepsilon$. Send $\varepsilon \to 0$: the cover vanishes, but every rational is still inside it. So the rationals have measure zero — no length, despite being dense.

What this means for [0, 1] \setminus \mathbb{Q}

The whole interval [0, 1] has length 1. The rationals inside it have length 0. Subtracting, the irrationals inside [0, 1] have length 1. If you pick a real number uniformly at random from [0, 1], the probability it is rational is 0; the probability it is irrational is 1.

So the second precise sense of "more" is: by length, the real line is essentially all irrational. The rationals occupy zero length. Every bit of visible length on the number line is irrational length.

Why this is surprising — density vs measure

Here is the strangeness to sit with. Density and measure are independent properties.

Density says: every open interval (a, b) \subset [0, 1], no matter how tiny, contains a rational. You cannot zoom anywhere and escape them. They are truly everywhere.
Measure zero says: the total length occupied by the rationals is zero. You cannot accumulate any visible length from them, no matter how many you gather.

Both are true simultaneously. The rationals are like fine dust sprinkled on the real line — present at every location, yet contributing no weight. The \varepsilon-cover is how that is possible: the enumeration lets you pack cleanup intervals that shrink geometrically, so the covered total stays bounded even as every point is captured.

Compare with a set that has positive measure — say [0, 1] itself. You cannot cover [0, 1] with intervals of total length less than 1, because any cover has to span a length of 1 somehow. The difference is that [0, 1] is uncountable; you cannot list its points, so the geometric-series trick is unavailable.

A dart-throwing check

Try this as a sanity test. Imagine throwing a dart uniformly at [0, 1]. The probability that the dart lands in any sub-interval (a, b) is b - a — proportional to length. Now the probability that the dart lands exactly on q_1 is zero (single points have no length). The probability it lands on q_2 is zero. By countable additivity of probability, the probability it lands on some rational — the union of all these zero-probability events indexed by n — is

P(\text{rational}) \;\le\; \sum_{n=1}^{\infty} P(\text{hit } q_n) \;=\; \sum_{n=1}^{\infty} 0 \;=\; 0.

Why countable additivity matters: this argument only works because you can list the rationals. If you tried the same thing for the irrationals in [0, 1], you would not even be able to write down the sum — there is no enumeration to sum over. The axioms of probability let you add countably many zeros and still get zero, but they do not let you add uncountably many and conclude anything.

So the dart lands on an irrational with probability 1. The "typical" real number is irrational, transcendental, and uncomputable; the rationals you can actually write down are a measure-zero decoration on a line that is essentially all irrational.

The two senses agree — in spirit

These two answers to "more in what sense" come from different branches of mathematics (set theory vs. measure theory), and they use different definitions of "small". But they point at the same structural fact: the rationals are a scaffold that is very thin and very orderly, and the reals have vastly more going on than the rationals can describe. Cardinality says: too many reals to list. Measure says: too many reals to fit in zero length. Either way, the irrationals carry the weight.

If you want the cardinality side told in full, see Rationals Are Countable, Reals Are Not and Cantor's Diagonal Argument. If you want the broader companion, Irrationals Outnumber the Rationals is the "big picture" version of this same idea. This satellite is the precise answer to the meta-question you should always ask when a mathematician says one infinite set is larger than another: in what sense.