Rigor, Infinity, and the Axiomatic Method

The development of modern logic cannot be explained by a desire for better notation alone. Nineteenth-century mathematics raised questions about the objects and principles that mathematical arguments presupposed. Analysis required precise accounts of limits and continuity. The study of infinite collections challenged familiar comparisons of size. Alternative geometries changed the status of assumptions long treated as descriptions of a uniquely determined space.

These developments did not form one coordinated campaign. Their interaction nevertheless made it increasingly useful to distinguish a mathematical structure, an axiomatic description of it, and the reasoning used to study that description.

The previous note explained how quantified language displays dependencies within an argument. This note considers the mathematical developments that made those dependencies—and the assumptions surrounding them—so consequential.

Analysis and the control of approximation

Calculus had produced extensive and reliable mathematics before its nineteenth-century reconstructions. Its earlier arguments used several approaches, including geometric intuition, limits, and infinitesimal quantities. The task was not to replace an entirely unsuccessful discipline, but to identify conditions under which its successful methods could be justified.

Bernard Bolzano's 1817 work on continuity and Augustin-Louis Cauchy's Cours d'analyse of 1821 were important stages. Cauchy organized analysis around limits while continuing to use infinitesimal quantities. Later presentations associated with Karl Weierstrass made the dependence between tolerances explicit through inequalities.

Consider the claim that $f(x)=x^2$ is continuous at a real number $a$. We must show that for every $\varepsilon>0$, a positive $\delta$ can be chosen so that

The factorization

suggests the proof, but the second factor must also be controlled. Require $|x-a|<1$. Then

It therefore suffices to choose

The conclusion follows by substitution into the preceding inequality. The proof does not merely state that the graph has no visible break. It supplies a choice whose dependence on $a$ and $\varepsilon$ is explicit.

This kind of argument also reveals the need to distinguish continuity at each point from uniform continuity on a domain. A $\delta$ allowed to depend on $a$ answers a different question from a single $\delta$ that must work at every point.

The rigorization of analysis thus gave logical form a concrete mathematical role. It also brought a further question into view: what are the real numbers over which these quantifiers range?

Dedekind and the construction of number systems

Richard Dedekind's 1872 account of continuity constructed real numbers through cuts in the rational numbers. One common modern presentation describes a cut by a nonempty proper subset $L\subset\mathbb{Q}$ that is downward closed and has no greatest member.

The rational number $r$ determines the cut

But there are cuts not determined by a rational endpoint. For instance,

specifies the location associated with $\sqrt{2}$. The construction does not begin by assuming that this location already contains a real number. It uses a precisely characterized collection of rational numbers to supply the new object.

Several constructions of the reals were developed, including approaches using equivalence classes of suitable sequences. Their coexistence is instructive: the significant question is often whether the constructions yield isomorphic ordered fields with the required completeness property, rather than whether they use literally identical underlying objects.

Dedekind's 1888 Was sind und was sollen die Zahlen? pursued a structural account of natural numbers. A distinguished initial element, an injective successor operation, and an appropriate minimality condition characterize a simply infinite system. Recursion then requires justification: defining addition by repeatedly applying successor is a mathematical construction whose existence and uniqueness must be established.

Peano's 1889 presentation gave an influential symbolic axiomatization of arithmetic. The relation between these projects should not be flattened into a claim that one author invented the natural numbers. They clarified different aspects of what it means to specify a number system and reason within it.

Cantor and the comparison of infinite sizes

A finite collection cannot be placed in bijection with a proper subset of itself. Infinite collections can behave differently. The map $n\mapsto 2n$ pairs the natural numbers with the even natural numbers.

Georg Cantor made one-to-one correspondence the basis of a systematic comparison of sizes. This permits a precise distinction between a collection being infinite and one infinite collection being strictly larger than another.

Cantor's 1874 work established an uncountability result; his later diagonal method gives an especially clear general argument. Let $X$ be any set. Suppose that a function

were onto the power set of $X$. Define

If $D=f(d)$ for some $d\in X$, then

a contradiction. Thus no such surjection exists.

This is a positive theorem about size. The construction of $D$ is confined to the existing set $X$; it does not require a set containing absolutely everything. The distinction will be crucial when similar diagonal reasoning appears in the paradoxes of unrestricted set formation.

Cantor's theory also distinguished cardinality from order type. Two sets can have the same size while carrying different well-orders. Ordinals encode the latter structure. The resulting arithmetic of infinity was not merely an extension of finite counting rules; it required new definitions and theorems.

These developments gave foundational work both a subject and a difficulty. Infinite sets were mathematically productive, but the principles governing their formation needed careful examination.

The following table makes the diagonal construction explicit. Each edit changes the proposed list, and the missing set is constructed anew from that list.

Geometry and the changing role of an axiom

Euclid's parallel postulate had long been treated as a candidate for derivation from the other assumptions of geometry. Nineteenth-century work associated with Gauss, Lobachevsky, and Bolyai revealed a different possibility: coherent geometries could be investigated in which a corresponding Euclidean parallel principle failed.

The development of models, notably through Beltrami and Klein, provided ways of relating non-Euclidean geometry to other mathematics. Such an interpretation establishes a relative result. If a contradiction could be derived in the interpreted geometry, it could be translated into a contradiction in the background mathematics used to construct the model.

This does not provide an assumption-free guarantee. It explains how one consistency question depends on another.

Riemann's broader investigation of geometric structures further separated geometry from the assumption that only one mathematical form of space was possible. The axiomatic method could now be used to explore families of structures rather than merely to organize truths about a single intuitively given object.

Hilbert's Grundlagen der Geometrie of 1899 supplied a systematic account of incidence, order, congruence, parallels, and continuity. He investigated the relations among axioms as well as the geometric conclusions obtained from them. The work sharpened a method with a long history; it did not invent axiomatic reasoning.

A small example of independence

The strategy of using models to separate assumptions can be illustrated without reconstructing an entire geometry.

Take the group axioms and consider the additional statement

The integers under addition satisfy the group axioms and this commutativity statement. The permutations of three objects, under composition, form a group in which commutativity fails.

For a concrete calculation, let $\sigma$ interchange $1$ and $2$, and let $\tau$ interchange $2$ and $3$. With functions composed from right to left,

Consequently, the two compositions are different.

Assuming a sound deductive calculus, commutativity cannot be a consequence of the group axioms, because a group exists in which it is false. Its negation cannot be a consequence either, because a commutative group exists.

The example separates several questions. Both structures satisfy the same basic axioms. They need not be isomorphic. An additional assertion may hold in one and fail in another. Axioms can therefore be investigated through the range of their models.

Four questions that should remain distinct

These mathematical developments encouraged several forms of foundational inquiry:

• Construction: How can the relevant objects be defined within an accepted background?
• Axiomatization: Which principles characterize the structures being studied?
• Categoricity: Do those principles determine a structure up to isomorphism, under the chosen semantics?
• Consistency: Can a contradiction be derived from the principles?

The answers are not interchangeable. A theory can be consistent without determining a unique structure. A construction can establish relative consistency without justifying the background in which the construction was carried out. A categorical description may require stronger semantic assumptions than an effective first-order calculus can capture.

The pressure for foundations consequently came from several directions at once. Analysis demanded control of approximation; arithmetic demanded an account of recursion and number; set theory expanded the mathematics of infinity; geometry made the independence of assumptions a practical mathematical question.

The next difficulty was more direct. Some apparently natural formation principles produced contradictions. The next note examines those failures and the different repairs proposed in response.

Sources and further reading

• Bernard Bolzano, Rein analytischer Beweis (1817); Augustin-Louis Cauchy, Cours d'analyse (1821).
• Richard Dedekind, Stetigkeit und irrationale Zahlen (1872) and Was sind und was sollen die Zahlen? (1888), translated in Essays on the Theory of Numbers.
• Georg Cantor, “Über eine Eigenschaft des Inbegriffes aller reellen algebraischen Zahlen” (1874) and “Über eine elementare Frage der Mannigfaltigkeitslehre” (1891). The earlier uncountability proof and later diagonal method should be distinguished.
• David Hilbert, The Foundations of Geometry, an English translation of the work originating in 1899.
• Dedekind's Contributions to the Foundations of Mathematics, for the relationship among construction, structural characterization, and recursion.