Introduction

One of the first surprising facts encountered in measure theory is:

$$m\big((\mathbb R\setminus\mathbb Q)\cap[0,1]\big)=1$$

In words:

The irrational numbers occupy the entire measure of the interval [0,1].

Since the rational numbers have measure zero,

$$m(\mathbb Q\cap[0,1])=0,$$

removing them leaves the full measure intact.

This naturally raises a deeper question:

Do all subclasses of irrational numbers also have measure one?

At first glance, the answer might seem to be yes. After all, irrational numbers themselves occupy the entire interval from the perspective of measure.

Surprisingly, the answer is no.

Some subclasses of irrational numbers have measure one, while others have measure zero. In fact, some uncountable sets of irrational numbers have measure zero.

This distinction reveals the richness hidden within the irrational numbers.

---

Irrational Numbers Are Not One Homogeneous Group

The set of irrational numbers contains many different families:

• Algebraic irrationals
• Transcendental numbers
• Computable numbers
• Normal numbers
• Liouville numbers
• Badly approximable numbers
• Fractal subsets

Each family behaves differently from a measure-theoretic perspective.

Measure theory allows us to ask:

How much of the continuum does this family occupy?

The answers are often surprising.

---

Algebraic Irrationals

An algebraic irrational number is a solution of a polynomial equation with integer coefficients.

Examples include:

$$\sqrt2$$

$$\sqrt3$$

and

$$\frac{1+\sqrt5}{2}.$$

These satisfy equations such as:

$$x^2-2=0$$

and

$$x^2-x-1=0.$$

Measure of Algebraic Numbers

There are only countably many integer polynomials.

Each polynomial has finitely many roots.

Therefore:

$$\text{Algebraic Numbers}=\bigcup_{n=1}^{\infty}\text{Roots of Polynomial}_n$$

is a countable union of finite sets.

Hence:

$$m(\text{Algebraic Numbers})=0.$$

Although algebraic irrationals are among the most familiar irrational numbers, they occupy zero measure.

---

Transcendental Numbers

A transcendental number is not the root of any polynomial with integer coefficients.

Examples include:

$$\pi$$

and

$$e.$$

Since:

$$\mathbb R=\text{Algebraic}\cup\text{Transcendental},$$

and algebraic numbers have measure zero,

it follows that:

$$m(\text{Transcendentals}\cap[0,1])=1.$$

Thus:

Almost every real number is transcendental.

This is one of the great surprises of modern mathematics.

---

Computable Numbers

A number is computable if an algorithm exists that can generate its digits to arbitrary precision.

Examples include:

$$\pi,\quad e,\quad \sqrt2.$$

Since there are only countably many computer programs, there are only countably many computable numbers.

Therefore:

$$m(\text{Computable Numbers})=0.$$

An astonishing consequence follows:

Almost every real number is non-computable.

Most real numbers cannot be completely described by any finite algorithm.

---

Normal Numbers

A number is normal if its digits are distributed exactly as random digits would be.

For example, in base 10:

• Each digit appears with frequency 1/10.
• Each pair appears with frequency 1/100.
• Every finite block appears with the expected frequency.

In 1909, Émile Borel proved a remarkable theorem:

Almost every real number is normal.

Thus:

$$m(\text{Normal Numbers}\cap[0,1])=1.$$

Despite this, proving that specific numbers such as:

$$\pi$$

or

$$e$$

are normal remains an open problem.

---

Liouville Numbers

Liouville numbers can be approximated exceptionally well by rational numbers.

One example is:

$$L=\sum_{n=1}^{\infty}10^{-n!}.$$

These numbers played an important role in the first proofs of transcendence.

However:

$$m(\text{Liouville Numbers})=0.$$

Although there are infinitely many Liouville numbers, they occupy no measure.

---

Badly Approximable Numbers

These numbers resist approximation by rationals.

The golden ratio:

$$\phi=\frac{1+\sqrt5}{2}$$

is the most famous example.

The set of badly approximable numbers is uncountable.

Yet:

$$m(\text{Badly Approximable Numbers})=0.$$

This is particularly striking because it demonstrates that:

Uncountable does not imply positive measure.

---

The Cantor Set

Perhaps the most famous example is the Cantor set.

It consists entirely of numbers whose ternary expansions contain only the digits:

$$0\quad\text{and}\quad2.$$

The Cantor set is:

• Uncountable
• Perfect
• Closed

yet:

$$m(\text{Cantor Set})=0.$$

This shows once again that enormous infinite sets can have zero measure.

---

Measure One Versus Measure Zero

The irrational numbers themselves satisfy:

$$m((\mathbb R\setminus\mathbb Q)\cap[0,1])=1.$$

However, many important subclasses satisfy:

$$m=0.$$

A rough summary is:

Class

Measure

Irrationals

1

Transcendentals

1

Normal Numbers

1

Algebraic Numbers

0

Computable Numbers

0

Liouville Numbers

0

Badly Approximable Numbers

0

Cantor Set

0

Thus:

Being irrational does not determine the measure of a subclass.

---

Why Researchers Care

Modern number theory often studies questions of the form:

What properties hold for almost every real number?

Researchers investigate:

• Rational approximation
• Digit distributions
• Dynamical systems
• Ergodic theory
• Fractal geometry

The goal is often to determine whether a particular property defines a set of:

$$m=0$$

or

$$m=1.$$

This is the central theme of metric number theory.

---

The Deep Insight

When students first learn that:

$$m(\mathbb R\setminus\mathbb Q)=1,$$

it is tempting to think that irrational numbers form one giant undifferentiated mass.

The reality is far richer.

Within the irrational numbers exist countless subclasses.

Some occupy the entire continuum from the viewpoint of measure.

Others are so thin that they occupy none at all.

Measure theory teaches us that infinity alone is not enough to understand size.

To understand the continuum, we must ask not merely:

How many numbers are there?

but rather:

How much of the continuum do they occupy?

That question continues to drive active research in number theory, ergodic theory, probability, and fractal geometry today.

---

References

1. Borel, É. (1909). Les probabilités dénombrables et leurs applications arithmétiques. Rendiconti del Circolo Matematico di Palermo.
2. Billingsley, P. Probability and Measure. Wiley.
3. Falconer, K. Fractal Geometry: Mathematical Foundations and Applications. Wiley.
4. Hardy, G. H., & Wright, E. M. An Introduction to the Theory of Numbers. Oxford University Press.
5. Katznelson, Y. An Introduction to Harmonic Analysis. Cambridge University Press.
6. Royden, H. L., & Fitzpatrick, P. Real Analysis. Pearson.
7. Rudin, W. Real and Complex Analysis. McGraw-Hill.
8. Tao, T. An Introduction to Measure Theory. American Mathematical Society.
9. Walters, P. An Introduction to Ergodic Theory. Springer.
10. Khinchin, A. Y. Continued Fractions. Dover Publications.