Proof by infinite descent
A proof by infinite descent, also called Fermat's method of descent, is a special kind of proof by contradiction. It shows that a statement cannot be true for any number. The method assumes that if the statement were true for one number, it would also be true for a smaller number, and so on forever. But that cannot go on forever, so the first assumption must be wrong. The method uses the well-ordering principle and is often used to show that an equation, such as a Diophantine equation, has no solutions. It should not be confused with infinite regress.
Usually, a person shows that if a solution to a problem existed, it would lead to a second solution, connected to one or more smaller natural numbers. That second solution would lead to a third, then a fourth, and so on. There cannot be an endless supply of ever-smaller natural numbers. So, by mathematical induction, the first idea that a solution exists must be wrong.
There is another way to describe this method. One assumes a solution exists and then finds the smallest one, called a minimal counterexample. Then one tries to prove that this smallest solution would lead to a still smaller one. This again shows that any solution would lead to a contradiction.
The earliest uses of this method appear in Euclid's Elements. One example is Proposition 31 of Book 7, where Euclid proves that every composite integer is divided, or "measured," by some prime number. The method was developed much later by Fermat, who gave it its name and often used it for Diophantine equations. Two examples are showing that the Diophantine equation r2+s4=t4 has no solutions, and proving Fermat's theorem on sums of two squares. That theorem says an odd prime p can be written as a sum of two squares when p≡1(mod4) (see Modular arithmetic and proof by infinite descent). Using this method, Fermat showed that many classic Diophantine equations have no solutions, such as the problem of four perfect squares in arithmetic progression.
A modern mathematician might see Fermat's "method of infinite descent" as using the inversion of the doubling function for rational points on an elliptic curve E. The idea starts with a hypothetical non-trivial rational point on E. Doubling a point on E roughly doubles the length of the numbers needed to write it, measured in digits. So "halving" a point gives a rational number with smaller terms. Since these terms are positive, they cannot keep getting smaller forever.
Number theory
In the number theory of the twentieth century, the infinite descent method was used again. It was developed further and linked to the main ideas of algebraic number theory and the study of L-functions. Mordell showed that the rational points on an elliptic curve E form a finitely-generated abelian group. His proof used an infinite descent argument of the kind Fermat had used.
To extend this to an abelian variety A, André Weil needed a clearer way to measure the size of a solution. He used a height function, an idea that became very important. Later, this work was linked to Galois cohomology, where some abstract groups matched the "descents" of Fermat's time. The Mordell–Weil theorem was the start of a very large area of study.
Application examples
See also: Fermat's right triangle theorem § Fermat's proof
Infinite descent can show that the square root of 2 (√2) is irrational. This means it cannot be written as a fraction of two whole numbers. This proof may be the earliest known example of infinite descent. The Pythagoreans discovered that the square root of two is irrational, but little is known for certain about when or how this happened. People often mention the name of Hippasus of Metapontum. According to legend, Hippasus was killed after sharing the discovery. The ancient Greeks did not have algebra, so they worked out a geometric proof using this idea. An algebraic version shows that if √2 could be written as a fraction, it could always be written with smaller parts, ad infinitum. This is impossible for whole numbers, so √2 must be irrational.
The same idea is used to study other equations. One proof shows that a Pythagorean triangle cannot have two sides that are each a square or twice a square. There is no smallest such triangle. This means the equations r2 + s4 = t4, r4 + s2 = t4, and r4 + s4 = t2 have no solutions using whole numbers, except trivial ones. The first of these equations helps show that q4 + s4 = t4 has no solutions. This is a special case of Fermat's Last Theorem.

Quiz time!
4 questions about Proof by infinite descent
What kind of proof is a proof by infinite descent a special form of?
Keep exploring
This article is a child-friendly adaptation of the Wikipedia article on Proof by infinite descent, available under CC BY-SA 4.0.