Reading passage
Of all the sciences, mathematics is the only one that claims certainty, and pure mathematics is the part of the discipline that pursues certainty for its own sake. Where applied mathematics borrows its problems from physics, engineering or finance, pure mathematics generates its questions internally, from the consequences of its own definitions. Its objects, such as prime numbers, infinite dimensional spaces and abstract symmetries, need not correspond to anything physical at all. The field is often described as an art as much as a science, and its practitioners have been notably unapologetic about its lack of obvious use.
The engine of the discipline is proof. A theorem is not accepted because it is plausible, supported by examples or valuable in practice, but because it follows by strict deduction from explicitly stated axioms. Euclid's Elements, composed around 300 BCE, established the template: begin with a handful of assumptions and derive, step by step, conclusions that no measurement could ever guarantee. A single counterexample destroys a conjecture, whereas a valid proof stands for eternity. This standard of rigour is unmatched elsewhere, which is why mathematical results, once secured, are never withdrawn but only generalised or superseded.
During the nineteenth and twentieth centuries the subject underwent a profound abstraction. Algebra ceased to be the manipulation of numbers and became the study of structures such as groups, rings and fields, each defined by operations satisfying a few axioms. Geometry escaped physical space altogether, giving rise to topology, which asks which properties of shapes survive stretching and bending. This generality is the source of the field's power: a theorem about groups applies at once to crystal symmetries, number systems and shuffles of cards, because it concerns the shared pattern rather than any particular instance.
Nowhere is the culture of purity expressed more memorably than in number theory, the study of the primes. The mathematician G. H. Hardy, writing in 1940, boasted that nothing he had done would ever make the least difference to the comfort or discomfort of the world, and he regarded this innocence as his subject's noblest feature. The primes seemed the perfect emblem of that ideal: beautiful, mysterious and utterly remote from application. Hardy would have been astonished, and perhaps dismayed, to learn that within a few decades the multiplication of large primes would underpin modern cryptography.
The pattern repeats with almost comic regularity. Non-Euclidean geometry, developed in the nineteenth century as a purely logical exploration of what happens when Euclid's parallel postulate is denied, turned out to be exactly the language Einstein required for general relativity. Group theory, formalised without any physical purpose, became indispensable to quantum mechanics and the classification of elementary particles. The physicist Eugene Wigner famously called this correspondence the unreasonable effectiveness of mathematics. Whether mathematics is discovered, which would explain its fit with nature, or invented, which makes that fit a standing mystery, remains a genuinely open question.
The pursuit of proof can also demand extraordinary patience. Fermat's Last Theorem, a claim scribbled in a margin around 1637, asserts that no three positive integers satisfy a certain equation when the exponent is greater than two. For more than three centuries the statement resisted every assault and became the most famous unsolved problem in mathematics. Andrew Wiles finally proved it in 1995, using machinery from algebraic geometry that Fermat could not have imagined, and correcting a gap in his own first attempt along the way. The episode is often cited as evidence that persistence, not luck, settles the great problems.
Modern proofs have grown so large that they strain the classical ideal of individual verification. The classification of finite simple groups, completed in stages across the twentieth century, spans tens of thousands of journal pages written by hundreds of authors, and no single person has read it all. The four colour theorem, which states that any map can be coloured with four colours without neighbours sharing one, was proved in 1976 only by reducing the problem to many cases checked by computer. Some mathematicians objected that a proof no human can survey forfeits the very understanding that proof is supposed to provide.
In response, a quieter revolution is under way. Proof assistants, software that checks every inference against formal axioms, can now verify arguments that humans cannot reliably audit, and several major theorems have been translated into machine-checkable form. Advocates foresee a mathematics that is simultaneously more certain and more collaborative, its results accumulating in verified libraries. Detractors worry that something essential is lost when conviction is outsourced to software. The dispute is ultimately about what mathematics is for, guaranteed truth or human understanding, and the pure tradition insists that the two should coincide.