Giuseppe Peano

Giuseppe Peano

Mathematician and logician · 1858–1932 · Italy

Arithmetic with its assumptions exposed

Peano helped make arithmetic and mathematical notation objects of explicit formal study. His work encourages a useful habit: separate primitive symbols, axioms, definitions and proofs. A definition tells us how to use a new expression; an axiom states an assumed property; a proof establishes a consequence. Those roles are easily blurred when familiar arithmetic feels self-evident.

Peano did not invent induction, numbers or axiomatic reasoning. Dedekind and other predecessors are essential to the history. Nor do five short slogans prove every truth of arithmetic. Understanding why requires distinguishing versions of the axioms and what counts as a formal proof.

Career and publications

Born near Cuneo on 27 August 1858, Peano studied at the University of Turin, graduating in 1880. His career combined analysis, geometry, logic, education and the development of mathematical language. His 1889 Arithmetices principia, nova methodo exposita presented arithmetic in a symbolic framework. In 1890 he published a continuous space-filling curve, and his collaborative Formulario project sought a systematic mathematical reference in symbolic notation. He also promoted Latino sine flexione as an auxiliary language. He died in Turin on 20 April 1932. Scholarly biography.

These were related interests in mathematical expression, but they should not be collapsed into one inevitable route to software. The connection to today's inductive datatypes is a modern conceptual comparison. Specific historical influence, such as Russell's encounter with Peano's school, requires documentary evidence rather than a diagram connecting famous names.

A modern successor presentation

In the following teaching model, natural numbers begin with 0. Let S be the successor operation. We require:

1. 0 belongs to the domain of natural numbers.
2. S maps natural numbers to natural numbers.
3. No natural number has successor 0.
4. If S(a) = S(b), then a = b.
5. Induction: a property holding at 0 and preserved by S holds throughout the natural numbers.

The original historical convention used 1 as the initial number; the 0-based version here is a modern choice, not a literal transcription of the 1889 text. Counts of “five axioms” also suppress background assumptions about equality and logic. Historical axiom summary.

The fourth condition is injectivity. It forbids two different numbers from merging under successor. The third prevents returning to the starting point. But those two restrictions alone do not ensure that every object in a candidate domain is reached from 0. That is one job of induction.

Worked mechanism: definitions are directional

Define addition by recursion on its second argument:

a + 0 = a

a + S(b) = S(a + b).

Use the abbreviations 1 = S(0), 2 = S(1), and so on. Then:

2 + 3
= 2 + S(2)
= S(2 + 2)
= S(S(2 + 1))
= S(S(S(2 + 0)))
= S(S(S(2)))
= 5

Each line uses a defining equation or the numeral abbreviation. This does not yet prove addition is commutative. In particular, a + 0 = a is a defining clause, but 0 + a = a needs an argument for arbitrary a.

Prove it by induction. At a = 0, 0 + 0 = 0 by the base definition. Suppose 0 + a = a. Then 0 + S(a) = S(0 + a) = S(a), first by the recursive definition and then by the hypothesis. The base and step establish the claim for every natural a.

This is an original modern derivation. It illustrates how a proof follows a recursive definition and why reversing arguments is not free. Defining multiplication by a × 0 = 0 and a × S(b) = a + (a × b) similarly produces calculations, while algebraic laws require further proofs.

What induction excludes

Consider a candidate structure made from two disjoint pieces: the ordinary naturals and a separate copy of all integers. On the natural-number piece, successor behaves normally. On the integer piece, it adds one and remains in that piece. The distinguished 0 belongs only to the natural-number piece.

Successor is injective, no element maps to the distinguished 0, and every element has a successor. Yet the integer component contains elements never reached from the distinguished 0 by finitely many successor steps. Let X be precisely the natural-number component. X contains 0 and is closed under successor, but it is not the whole domain. Induction over all subsets rules out this structure.

This countermodel is a teaching construction, not an assertion that it is a model of full first-order Peano arithmetic. It shows specifically why the non-induction successor conditions are insufficient.

First-order and second-order arithmetic

If induction quantifies over every subset of the domain under full second-order semantics, the resulting successor characterization is categorical: its models have the intended natural-number structure up to isomorphism. If induction is expressed in first-order logic, it is normally a schema with an instance for each formula. It does not quantify over arbitrary subsets, and first-order Peano arithmetic has nonstandard models. Semantics and categoricity.

Categoricity is not the same as having an effective proof system that derives every truth. Full second-order validity does not have a sound, complete effective proof calculus. Conversely, first-order logic has a completeness theorem for logical consequence, while a particular first-order arithmetic theory can be incomplete: some sentences are not decided by its axioms. The two uses of “complete” concern different objects.

Gödel's results apply to effectively axiomatized theories with sufficient arithmetic strength under the relevant consistency assumptions. They do not show that arithmetic proofs are worthless. They show why “all arithmetic truths follow by an algorithm from these few axioms” is an unjustified claim. Gödel's original paper concerned formal systems associated with Principia Mathematica and related systems; describing it as only a paper about a five-axiom version of Peano's 1889 text is historically inaccurate. Incompleteness and its hypotheses.

A curve is not a dimension-preserving bijection

Peano's space-filling curve supplies another lesson in precise definitions. A continuous map from an interval onto a square can be surjective without being injective. It does not show that an interval and square are homeomorphic. A drawn approximation also is not the entire mathematical limiting curve. When comparing such constructions with modern spatial indexing, distinguish the continuous limit, a discrete traversal, and locality guarantees; each is a different claim. Peano's curve in historical context.

Exercises

1. Expand 3 × 2 using the multiplication equations above and the addition equations. Mark which steps are definitions. Then explain why two examples cannot prove multiplication commutative.
2. Prove that no finite nonempty domain can satisfy injective successor together with a distinguished element 0 outside the range of successor. State which elementary fact about finite sets you use.
3. Explain why the two-component structure above passes the non-induction successor tests but fails full second-order induction. Identify the subset witnessing failure, and explain why that observation alone is not a classification of all nonstandard models of first-order arithmetic.

Compare your reasoning with the worked solutions

1. Evaluate the recursive definition

3 × 2 = 3 + (3 × 1) = 3 + (3 + (3 × 0)) = 3 + (3 + 0) = 3 + 3 = 6. The first three rewrites use multiplication's clauses; the next uses addition's base clause. Expanding 3 + 3 applies the successor clause three times. The trace establishes this input's value. Commutativity is a universal statement over two arguments, so checking 3 × 2 and 2 × 3 supplies examples rather than the required general derivation.

2. Finite injectivity forces surjectivity

An injective function from a finite set of k elements to itself has k distinct outputs. There are only k possible outputs, so every element must be in its range. That contradicts the requirement that the distinguished 0 is absent from the range. Thus a structure meeting these successor conditions must be infinite. This proves infinitude, not uniqueness of its infinite structure.

3. Closure is not enough without induction

Within each component successor is injective; images never cross components, so the combined function is injective. No element of the natural component maps to its initial 0, and no integer-component element maps into that component at all. However, X consisting only of the natural component contains 0 and is successor-closed while excluding every extra integer. Full second-order induction demands that such an X contain the entire domain, yielding failure. First-order induction ranges over definable properties instead of all subsets; analyzing its models needs those additional logical details and the arithmetic operations.