The method of exhaustion is a technique of mathematical proof used to establish areas, volumes, and relationships between geometric magnitudes. Developed in ancient Greece, it replaces a curved figure with successively refined figures whose measurements are already understood. The remaining difference can be made smaller than any prescribed positive magnitude. A proposed result is then established by showing that any departure from it would contradict a sufficiently accurate finite approximation. The method anticipates aspects of integration, but its classical arguments do not require modern notation for infinite processes. (mathshistory.st-andrews.ac.uk)
Historical development
The method is traditionally associated with Eudoxus of Cnidus, a fourth-century BCE mathematician whose work also developed a theory of proportions. His own writings have not survived; the method is known through later mathematical texts. Euclid presents its underlying exhaustion principle in Book X of the Elements and applies it extensively to plane and solid geometry in Book XII. (mathshistory.st-andrews.ac.uk)
Archimedes, working in the third century BCE, extended exhaustion arguments to important problems involving circles and curved regions. In the introduction to his Quadrature of the Parabola, he describes the assumption that repeated addition of the difference between unequal areas can exceed any given finite area. He also identifies earlier results about circles, spheres, pyramids, and cones that depended on comparable assumptions. These statements place his proofs within an established tradition rather than presenting exhaustion as his own invention. (mathshistory.st-andrews.ac.uk)
The exhaustion principle
Book X, Proposition 1 of the Elements states that, given two unequal magnitudes, repeatedly removing more than half of the larger magnitude, and then more than half of each remainder, eventually leaves a remainder smaller than the originally specified smaller magnitude. This guarantees a finite construction sufficiently accurate for any particular comparison. (euclids-elements.org)
The associated Archimedean property says, in the setting of positive magnitudes of the same kind, that a sufficiently large multiple of one can exceed the other. It rules out a positive magnitude that remains smaller than every repeated subdivision relevant to the argument. Archimedes explicitly invokes this assumption for areas. (mathshistory.st-andrews.ac.uk)
In a modern numerical reformulation, suppose successive remainders satisfy
[ 0\leq R_n\leq q^nR_0,\qquad 0<q<1. ]
For every positive tolerance (\varepsilon), some finite (n) gives (R_n<\varepsilon). This expresses the exhaustion principle in the language of a limit. It does not mean that a finite approximation must eventually coincide exactly with the curved figure; only that its discrepancy becomes smaller than the particular magnitude needed in the proof. (euclids-elements.org)
Structure of an exhaustion proof
A typical argument begins with a proposed area or volume and constructs approximations with known measurements. It then uses proof by contradiction to exclude an excess or deficiency. If the proposed result differs from the true magnitude by a positive amount, the construction is refined until its remaining error is smaller than that amount. Geometric comparisons then produce an impossibility. Where both alternatives must be excluded, the reasoning is often called a double reductio. (aleph0.clarku.edu)
Some proofs use both inscribed and circumscribed figures, producing lower and upper bounds. Others use only inscribed figures together with a separately established identity for their measurements. Exhaustion is therefore more than an approximation procedure: the decisive requirement is a demonstrated bound on the remainder, sufficient to support an exact conclusion. The circle and parabola illustrate these different arrangements. (web.math.utk.edu)
Circles and solid figures
Book XII, Proposition 2 of the Elements proves that the areas of circles are proportional to the squares of their diameters. Euclid inscribes polygons and repeatedly bisects the remaining arcs. Adding the corresponding triangles removes more than half of each remaining circular segment, so the uncovered area can be reduced below any specified positive area. Similar inscribed polygons in two circles supply the ratios used in the contradiction argument. (aleph0.clarku.edu)
In Measurement of a Circle, Archimedes establishes that a circle has the same area as a right triangle whose perpendicular sides equal its radius and circumference. Written in modern notation, this is
[ A=\frac12 rC. ]
Inscribed and circumscribed polygons support the exclusion of both unequal alternatives. The theorem relates area to circumference without requiring the circle to be treated as a polygon with infinitely many sides. (ms.uky.edu)
Book XII also applies exhaustion to solid figures. Among its results are that a cone occupies one-third of the cylinder with the same base and height, and that sphere volumes are proportional to the cubes of their diameters. The constructions use appropriate inscribed pyramids, prisms, or polyhedra. (aleph0.clarku.edu)
Quadrature of the parabola
Archimedes’ Quadrature of the Parabola determines the area enclosed by a parabola and a chord. An initial inscribed triangle has the chord as its base and the same height as the segment. Further triangles fill the remaining regions, with each generation contributing one-quarter of the preceding generation’s total area. (mathshistory.st-andrews.ac.uk)
If the initial triangle has area (T), the successive contributions form a geometric series:
[ T,\quad \frac{T}{4},\quad \frac{T}{16},\quad \ldots. ]
Archimedes establishes a finite identity equivalent to
[ T+\frac{T}{4}+\cdots+\frac{T}{4^n} +\frac{T}{3\cdot4^n}=\frac43T. ]
The final correction becomes arbitrarily small, while the inscribed triangles exhaust the segment. Contradiction arguments consequently establish that its area is exactly (4T/3), rather than merely approximately that value. (web.calstatela.edu)
Relationship to calculus
Exhaustion is an important predecessor of calculus. Its use of controlled approximations resembles the limiting processes underlying Riemann sums and the Riemann integral. Nevertheless, the classical method operates with geometric magnitudes and finite constructions, rather than a general theory of functions and integration. It proves a candidate relationship by excluding every positive discrepancy; it does not by itself provide a universal procedure for discovering the candidate result. (mathshistory.st-andrews.ac.uk)