How Many Axioms Is High-School Mathematics Built From?
Level: Advanced
Rebuilding high-school mathematics from the axioms of the real numbers
The question: what is the foundation of high-school mathematics?
The mathematics taught in high school is astonishingly varied: algebraic manipulation, functions, trigonometry, exponentials and logarithms, sequences, calculus, complex numbers, vectors, coordinate geometry, probability and statistics. From exactly how many "initial agreements" (axioms) does all of this start? This article answers that question head-on. Here is the conclusion first.
Conclusion
Most of the objects handled in high-school mathematics can be developed, with sets and logic as the substrate, on top of the ~15 axioms of the real numbers viewed as a complete ordered field. Exponentials, logarithms, trigonometry, complex numbers, calculus, sequences, and probability are all constructed as "definitions" without adding any new mathematical assumption beyond the axioms of the reals. Geometry, too, is absorbed into the reals once coordinates are introduced.
And of those 15, exactly one is essentially strong: completeness.
On "how to count" the axioms — declaring our stance first
"How many" depends on where you choose to start. To avoid confusion, let us fix our stance up front.
- Construction route: start from the Peano axioms (≈5) of the natural numbers and build $\mathbb{Z}\to\mathbb{Q}\to\mathbb{R}\to\mathbb{C}$ by definition. Along the way no new axiom is added — even completeness comes out as a theorem of the constructed $\mathbb{R}$ (the strength shows up at the $\mathbb{R}$ step).
- Direct route (our main count): axiomatize $\mathbb{R}$ directly as a complete ordered field = 15 axioms (10 field + 4 order + 1 completeness, where the 10 field axioms count nondegeneracy $0\neq1$).
The key point is that these two are not additive. The 15 of the direct route recover the naturals $\mathbb{N}$ by definition (as the smallest inductive set inside $\mathbb{R}$), so the 5 Peano axioms need not be added separately. The two routes are simply different entrances to the same mathematics (so it is not "15 + 5 = 20").
Geometry alone is an independent axis: coordinatize and you add nothing, absorbing it into the reals. If you erect synthetic geometry (points and lines as undefined primitives, Euclid-style), it is its own self-contained axiomatization of geometry from scratch — Euclid's 5 postulates plus 5 common notions, rigorously Hilbert's ~20 axioms, or Tarski's 11 axioms plus a continuity schema. These are not added on top of the 15 (like the Peano case above, they are a separate entrance). For the plane they are equivalent to the structure of the real field (more precisely a real closed field), adding no new strength. Taking the coordinate stance, this article fixes the count to the direct route above — 10 field + 4 order + 1 completeness = 15 — and proceeds with this 15 as the baseline.
This claim is a concrete instance, on the most familiar material, of the broader viewpoint of reverse mathematics — measuring how strong an axiom each theorem requires. In its language, many of the computational and constructive arguments actually used in high-school mathematics are captured at the very bottom ($\mathsf{RCA}_0$, i.e. computable mathematics) (though this does not identify high-school mathematics as a whole with a single formal system — high-school mathematics is not a formalized system). High school chiefly computes or locates its objects: it uses the intermediate value theorem informally (a sign change gives a root) and reads off maxima and minima from an increase/decrease (sign-of-derivative) table — all of which fit the weakest $\mathsf{RCA}_0$. Abstract existence theorems such as Heine–Borel and Bolzano–Weierstrass are not taught in high school but are learned in first- to second-year university analysis (in reverse-math strength they correspond to $\mathsf{WKL}_0$ and $\mathsf{ACA}_0$; however, reverse mathematics itself — which studies the systems $\mathsf{WKL}_0$/$\mathsf{ACA}_0$ — is advanced, upper-division-and-beyond material). The spread across $\mathsf{RCA}_0$–$\mathsf{ACA}_0$ describes "ordinary mathematics" (the standard analysis and algebra that reverse mathematics studies) rather than high school as such.
The starting point: the Peano axioms of the naturals
Every number begins with the naturals. What pins down the natural numbers $\mathbb{N} = \{0, 1, 2, \ldots\}$ are the Peano axioms.
The Peano axioms (essentials)
- $0$ is a natural number.
- Each natural number $n$ has a successor $S(n)$ (the next number).
- $0$ is not the successor of any number.
- The successor operation $S$ is injective ($S(m)=S(n) \Rightarrow m=n$).
- Mathematical induction: if $0$ has a property $P$ and $P(n)\Rightarrow P(S(n))$, then every natural number has $P$.
Essentially these 5 items. Written in second-order logic, the structure of the naturals is fixed uniquely (up to isomorphism) by just 2–3 axioms.
Addition, multiplication, and order are all defined from these ($n+0=n,\ n+S(m)=S(n+m)$, etc.). So "$1+1=2$" and "commutativity of addition" are not axioms but theorems. The starting point is nothing more than the 5 items above.
Building the numbers: ℕ→ℤ→ℚ→ℝ→ℂ are all "definitions"
From the naturals, every number system used in high school is built by stacking definitions alone, and in this construction no new axiom is needed at all. Even completeness — the star of the next section — is here derived as a theorem of the constructed $\mathbb{R}$ (one can prove that the $\mathbb{R}$ built from Dedekind cuts is complete). There is, however, one step where the strength concentrates: the step from $\mathbb{Q}$ to $\mathbb{R}$. If instead you axiomatize $\mathbb{R}$ directly (next section), this same completeness becomes "the single new axiom." A theorem in the construction, an axiom in the axiomatization — either way the strength gathers at the $\mathbb{R}$ step.
Caveat: "definition" presupposes basic set operations
"Definition" here is understood to allow basic set operations such as forming pairs and quotienting by an equivalence relation (quotient sets). Strictly, these rest on the axioms of set theory (pairing, separation, etc.). Within high-school mathematics, such operations are tacitly assumed as the common sense of "logic and sets." So, precisely, "definition only" means "adding no new mathematical assumption beyond the axioms of the reals."
- $\mathbb{Z}$ (integers): regard a pair of naturals $(a,b)$ as "$a-b$" and identify $(a,b)\sim(c,d) \Leftrightarrow a+d=b+c$. Negative numbers are defined this way.
- $\mathbb{Q}$ (rationals): regard a pair of integers $(a,b)$ ($b\neq0$) as "$a/b$" and identify $(a,b)\sim(c,d)\Leftrightarrow ad=bc$. This yields the field of rationals.
- $\mathbb{R}$ (reals): defined via Dedekind cuts of rationals, or Cauchy sequences of rationals (What are the real numbers?). Here, for the first time, the strong property of completeness arises.
- $\mathbb{C}$ (complex): regard a pair of reals $(a,b)$ as $a+bi$ and define multiplication so that $i^2=-1$. This is added structure on $\mathbb{R}^2$ — no new axiom required.
The functions, too, are all "definitions"
The exponential, logarithmic, and trigonometric functions central to high school can all be defined as functions on the real field. For example, $$ \exp(x)=\sum_{n=0}^{\infty}\frac{x^n}{n!},\qquad \sin(x)=\sum_{n=0}^{\infty}\frac{(-1)^n x^{2n+1}}{(2n+1)!} $$ given as power series (whose convergence is guaranteed by completeness) or as solutions of differential equations. No new axiom is added. The addition formulas covered in Review of Trigonometric Functions are likewise all theorems from here.
The single strong addition: completeness
Axiomatizing the reals directly as a "complete ordered field" lets you see the whole foundation of high-school mathematics at a glance. The axioms split into three groups.
The rational field $\mathbb{Q}$ satisfies both the field axioms and the order axioms. What separates $\mathbb{R}$ from $\mathbb{Q}$ is one thing only: completeness. That limits converge, that suprema exist, that the number line has no holes — all of this is thanks to this single axiom (axioms of the real numbers). (The existence of an algebraic number such as $\sqrt{2}$ actually needs only real-closedness rather than full completeness, as the section on circle intersections below makes precise; completeness, being stronger, supplies it too.)
Why "4" order axioms (settling the count)
Taking $\lt$ as primitive, the order is exhausted by exactly 4 axioms: ① Trichotomy (for any $a,b$, exactly one of $a \lt b,\ a=b,\ b \lt a$ = total order), ② Transitivity, ③ Compatibility with addition ($a \lt b \Rightarrow a+c \lt b+c$), ④ Compatibility with multiplication ($0 \lt a,\ 0 \lt b \Rightarrow 0 \lt ab$).
It is sometimes written as "3," but that just bundles ① into the single phrase "is a total order." Conversely, if you take $\le$ as primitive and split it into reflexivity, antisymmetry, transitivity, and totality, the pure order axioms become 4, and together with the 2 compatibility axioms some count 6. This article settles on the most standard $\lt$-based count of 4. Counting the field axioms as 10 (including nondegeneracy $0\neq1$) gives exactly "Field 10 + Order 4 + Completeness 1 = 15" (omit $0\neq1$ and it is 14 — either way, a dozen-and-a-half or so, and we take 15 as the baseline).
Completeness (the least-upper-bound property)
Every nonempty set $S$ of reals that is bounded above has a least upper bound (supremum) $\sup S$.
This least-upper-bound property quantifies over all subsets, so it is a second-order property. The reals being determined uniquely up to isomorphism (a complete ordered field is essentially unique) rests on this second-order completeness; weakened to a purely first-order axiom system, uniqueness is lost (by the Löwenheim–Skolem theorem).
Equivalent restatements: every bounded monotone sequence converges / every Cauchy sequence converges / the Archimedean property + nested intervals.
Where completeness matters: existence
Completeness does not come into play when you merely compute the value of finitely many arithmetic operations or elementary functions. You do not need completeness to evaluate $3+5$ or $\sin(\pi/6)$. What completeness newly produces is "existence" — the guarantee that points such as a supremum, a limit, or a convergent subsequence really are there.
Caution: "computations" involving limits do need completeness
"Not needed for computation" applies only to finitely many operations. For computations involving limits — the convergence of a Taylor series, the value of an infinite sum, the justification of any quantity containing $\lim$ — completeness guarantees that the limit exists as a real number. "Computation" and "existence" are continuous with one another, and the moment you step into infinite operations, completeness shows its face.
More precisely, the essential boundary is not "finite vs infinite" itself but whether you have an effective error bound (a modulus of convergence). The value of an elementary function such as $\sqrt2$ or $\sin 1$ carries an explicit error bound (the Taylor remainder, or the error of Newton's method), so any desired precision is produced by a finite computation on rationals. This is why a calculator returns $1.41421\ldots$ for $\sqrt2$ and need not return the completed real $\sqrt2$ (the symbol $\sqrt2$ itself). By contrast, for the convergence of a bounded monotone sequence or the existence of a supremum, you know it converges but cannot in general compute how fast, so completeness is needed to guarantee non-constructively that the limit exists. The dividing line is whether the value can be generated as an approximation.
The representative cases that genuinely require completeness itself are the following three "limit-flavored existence" facts. They break in fields that are not complete, such as the rationals or the real algebraic numbers.
- A set of reals bounded above has a supremum $\sup$.
- A bounded monotone sequence converges.
- A bounded sequence has a convergent subsequence (Bolzano–Weierstrass).
On the other hand, there are many examples of "vanishes over $\mathbb{Q}$ but exists over $\mathbb{R}$" that do not require completeness itself (being closed under square roots suffices; see below).
- A real number $\sqrt{2}$ with $x^2=2$ exists.
- Two circles have an intersection point (next figure).
- The intermediate value theorem for polynomials.
The "intersection of two circles" is a fine example where the difference in existence between $\mathbb{Q}$ and $\mathbb{R}$ is visible. The figure below is Euclid's very first proposition, the construction of an equilateral triangle.
Why it "slips through"
The intersection of the two circles $x^2+y^2=1$ and $(x-1)^2+y^2=1$ is $\left(\tfrac12,\ \pm\tfrac{\sqrt3}{2}\right)$. The $y$-coordinate contains $\sqrt3$, and $\sqrt3$ is not rational. So within $\mathbb{Q}^2$ the intersection becomes a "point that does not exist," while it does exist in $\mathbb{R}^2$.
Precisely: the intersection's existence is "closure under square roots," not "completeness itself"
Note that what the intersection's existence really requires is that "positive numbers have square roots" — that is, being a Euclidean field, or more generally a real closed field — and not completeness itself. Indeed, the field of real algebraic numbers $\mathbb{R}\cap\overline{\mathbb{Q}}$ is not complete yet is real closed, so this intersection exists there too. Completeness is a stronger property that implies real-closedness, so "it exists over $\mathbb{R}$" is true; but crediting the intersection's existence to completeness alone is, strictly, an overstatement.
What genuinely requires completeness is the limit-flavored existence above — supremum, monotone convergence, Bolzano–Weierstrass (which break in fields that are real closed but not complete). Constructions and the existence of algebraic points are fine in a real closed field. Measuring precisely this "different theorems require different strengths" is exactly what reverse mathematics does. This is also why high-school straightedge-and-compass geometry can be justified with axioms weaker than full completeness.
Geometry is the same: coordinatize and add zero axioms
High-school geometry problems look, at first, as though they need their own "axioms of geometry" separate from the reals. But once you introduce coordinates, geometry is translated wholesale into the algebra of the reals.
Geometry → reals dictionary
| Geometric term | Definition in ℝ |
|---|---|
| point | a pair of reals $(x,y)$ |
| line | solutions of a linear equation $ax+by+c=0$ |
| distance | $\sqrt{(x_1-x_2)^2+(y_1-y_2)^2}$ |
| angle $\theta$ | $\cos\theta = \dfrac{\vec a\cdot\vec b}{|\vec a|\,|\vec b|}$ (orthogonality: dot product $=0$) |
| circle | $(x-a)^2+(y-b)^2=r^2$ |
Through this dictionary, the Pythagorean theorem is the distance formula itself, the inscribed-angle theorem is a dot-product computation, and the existence of the centroid or circumcenter is the solvability of a linear system. Geometric theorems turn into algebraic identities over the reals, and not a single new axiom is required. If you set up synthetic geometry separately (Euclid-style, with points and lines as undefined primitives), it becomes its own self-contained axiomatization of geometry (Euclid's 5 postulates plus 5 common notions, Hilbert's ~20 axioms, Tarski's 11 axioms plus a schema, and so on). Hilbert's system builds its coordinate field internally without using coordinates, so it is not added to the 15; for the plane it is an equivalent alternative foundation, the same strength as the structure of the real field (a real closed field). On the coordinate stance of this article it remains zero additions = 15.
(Advanced) High-school geometry is "mechanically decidable"
From here on is an advanced aside. Tarski proved that first-order Euclidean geometry is decidable — the truth of any geometric statement can be decided by an algorithm. The decision method is quantifier elimination, implemented in computer algebra as CAD (cylindrical algebraic decomposition). So high-school geometric theorems are not only "derivable from the reals" but can be proved by computation after coordinatization. If interested, see Quantifier Elimination and CAD.
Logic is a premise distinct from the axioms
When we say "high-school mathematics is derivable from 15 axioms," we are in fact tacitly assuming logic as an inference engine. Logic is not derived from the axioms. Nor is it "another axiom layer beneath the axioms." Logic is another premise standing alongside the axioms; only when both are present do theorems get built. By analogy, the axioms are the "materials" and logic is the "blueprint."
The logic taught in high school — propositions, necessary and sufficient conditions, converse/inverse/contrapositive, De Morgan's laws, proof by contradiction, counterexamples — can be checked mechanically with the truth tables of propositional logic (Sets and Logic). So the laws of logic themselves "can be proved," but not by deriving them from the axioms; they are checked within logic.
A hidden choice: high-school mathematics adopts "classical logic"
Of the proof-by-contradiction used routinely in high school, the form "assume $\neg P$, derive a contradiction, and therefore conclude the positive proposition $P$" is in fact a principle specific to classical logic. It does not hold in constructive (intuitionistic) mathematics.
This is worth making precise. Not all proof by contradiction is classical-specific. The form proving a negative proposition — "assume $P$, derive a contradiction, therefore $\neg P$" — and $P \Rightarrow \neg\neg P$ do hold in intuitionistic logic. What is specific to classical logic is removing the double negation, $\neg\neg P \Rightarrow P$ (equivalent to the law of excluded middle); the "derive a positive proposition from the contradiction of its negation" form above is exactly this. Proving a negative proposition such as "$\sqrt2$ is irrational" is constructively legitimate; only when one obtains a positive proposition by contradiction is classical logic being chosen. That is the precise picture.
How do high school and university differ?
Everything above sharpens the boundary between high-school and university mathematics. In a word:
The difference (in a word)
High-school mathematics "computes inside the structure ℝ." University mathematics "builds that ℝ from axioms, proves its properties, and then swaps the axioms to build other structures too."
The shift comes down to four points.
- Computation → proof: the deliverable changes from "a number (answer)" to "a theorem."
- Implicit axioms → explicit axioms: the completeness high school used silently is brought into the open by university via the ε-δ method, reproving limits, the intermediate value theorem, and the extreme value theorem from completeness (Why be rigorous?).
- Concrete → abstract: remove completeness and order from the field axioms to study "any structure satisfying the axioms" — groups, rings, fields, metric spaces, topological spaces. Abstraction is the operation of deleting axioms and seeing what survives.
- Descriptive definitions → creative definitions: from "a circle is the set of equidistant points" (description) to "a group is a set satisfying …" (creation). The definition becomes the starting point.
In university, non-constructive existence (objects that exist but cannot be written down) and pathologies like functions continuous everywhere but differentiable nowhere, not met in high school, come to the fore. These are precisely matters of completeness and of the constructive/non-constructive distinction — the true identity of the part high school skipped as "obvious."
The ceiling: Gödel and "derivable < true"
This article's claim was that "high-school mathematics is derivable from 15 axioms." A natural question then arises — is "derivable" the same as "everything that is true"? The answer is "no," and here lies the precise upper bound. Combine logic with sufficiently strong arithmetic and, by Gödel's incompleteness theorem, there necessarily appear propositions that are true yet unprovable from the axioms. That is,
The limit of derivability
In a consistent axiom system containing enough arithmetic, $\{\text{propositions derivable from the axioms}\} \subsetneq \{\text{true propositions}\}$.
What "true" means here
To avoid misunderstanding: this "true" means true in the standard model $\mathbb{N}$ of the natural numbers. The proposition Gödel constructs is true in $\mathbb{N}$ but unprovable from the axiom system in question. "True" is nothing absolute or mystical; it refers to the truth value in the standard world of the naturals that we intend.
Fortunately, high-school mathematics never hits this wall, because the propositions it handles do not deliberately use arithmetic strong enough for incompleteness to surface. But when one says "all of it is derivable from the axioms," it is honest to keep this ceiling in mind — derivability and truth are strictly distinct.
Summary
Key points
- Almost all of high-school mathematics is derivable from ℝ = the 15 axioms of a complete ordered field (10 field + 4 order + 1 completeness, the 10 field axioms counting nondegeneracy 0≠1), with logic as the inference engine.
- This is not added to "Peano ≈5." The construction route (build $\mathbb{Z}\to\mathbb{Q}\to\mathbb{R}\to\mathbb{C}$ by definition from the Peano axioms ≈5) and the direct route (axiomatize ℝ directly, ~15) are different entrances to the same mathematics. The strength concentrates at a single point — completeness of ℝ (a theorem in the construction, the one new axiom in the axiomatization).
- What completeness purely supplies is limit-flavored "existence" (supremum, monotone convergence, Bolzano–Weierstrass). It is not needed for finitely many arithmetic/elementary-function evaluations, but is needed to justify computations involving limits. The existence of √2 or of two circles' intersection actually does not require completeness — a real closed field (closed under square roots) suffices.
- Geometry is absorbed into the reals with zero axioms once coordinatized, and is moreover decidable (CAD).
- Logic is a premise distinct from the axioms (materials = axioms, blueprint = logic). High school adopts classical logic, whose specific principle is proof by contradiction yielding a positive proposition (double-negation elimination).
- High school computes inside ℝ; university proves and abstracts ℝ from axioms.
- The upper bound is Gödel incompleteness — derivable ⊊ true.
The broader framework of "measuring how strong an axiom each theorem requires" is called reverse mathematics. This article walked its bottom layer using the most familiar material of all: high-school mathematics.
Exercises
Problem 1
List the axioms of the reals that the rational field $\mathbb{Q}$ satisfies and those it does not. Give one fact of high-school mathematics that "fails" in $\mathbb{Q}$.
Problem 2
Describe a strategy to show that $\sqrt{2}$ "exists" using completeness (the least-upper-bound property). Use the set $S=\{x\in\mathbb{Q}: x>0,\ x^2 < 2\}$.
Problem 3
Find the intersection of the two circles $x^2+y^2=1$ and $(x-1)^2+y^2=1$, and verify that its coordinates do not lie in $\mathbb{Q}^2$. This intersection is a standard example of the gap between $\mathbb{Q}$ and $\mathbb{R}$ — but is completeness itself actually required for its existence? Argue using the article's "real closed field" discussion.
Show solutions
Solution to Problem 1
$\mathbb{Q}$ satisfies the field and order axioms (arithmetic and order are in place). What it does not satisfy is completeness. For instance $S=\{x\in\mathbb{Q}:x^2 < 2\}$ is bounded above but has no supremum in $\mathbb{Q}$ (the supremum is $\sqrt2\notin\mathbb{Q}$). As a result, in $\mathbb{Q}$ facts of high-school mathematics break: "$x^2=2$ has no solution," "a bounded monotone sequence may fail to converge," and so on.
Solution to Problem 2
$S=\{x\in\mathbb{Q}:x>0,\ x^2 < 2\}$ is nonempty ($1\in S$) and bounded above ($2$ is an upper bound). By completeness, $\alpha=\sup S$ exists as a real number. If $\alpha^2 < 2$, slightly increasing $\alpha$ keeps it in $S$, contradicting that it is an upper bound; if $\alpha^2 > 2$, a slightly smaller number is an upper bound, contradicting minimality. Hence $\alpha^2=2$, i.e. $\alpha=\sqrt2$ exists. Completeness produces existence.
Solution to Problem 3
Subtracting the two equations gives $-2x+1=0$, i.e. $x=\tfrac12$. Substituting, $y^2=1-\tfrac14=\tfrac34$, so $y=\pm\tfrac{\sqrt3}{2}$. The intersections are $\left(\tfrac12,\pm\tfrac{\sqrt3}{2}\right)$. Since $\sqrt3\notin\mathbb{Q}$, the points are not in $\mathbb{Q}^2$ ($\mathbb{Q}$ is not closed under square roots, so the intersection slips through). However, the coordinates are the algebraic number $\sqrt3$, and the intersection exists in any field where positive numbers have square roots (a Euclidean / real closed field). Indeed the real algebraic numbers $\mathbb{R}\cap\overline{\mathbb{Q}}$ are not complete yet contain the intersection. So what is truly required is closure under square roots, not completeness itself. Completeness is genuinely required only for limit-flavored existence such as suprema, monotone convergence, and Bolzano–Weierstrass. $\square$
Frequently Asked Questions
How many axioms is high-school mathematics built from?
Almost all of high-school mathematics can be developed from 15 axioms of "ℝ = a complete ordered field" (10 field + 4 order + 1 completeness). Exponentials, logarithms, trigonometry, complex numbers, calculus, sequences, and probability are all constructed as definitions on top of this real field without adding new axioms. If you treat synthetic geometry separately, it becomes its own self-contained axiom system (about 20 for Hilbert); this is not added to the 15 but is an equivalent alternative foundation for the plane, of the same strength as the real field.
What is the completeness axiom for?
Completeness (the least-upper-bound property) is needed to guarantee existence — of limits, of the real number √2, of the value in the intermediate value theorem. The rationals lack completeness, so √2 and limits drop out and two circles that ought to meet slip past each other. High-school mathematics can use limits and irrationals freely thanks to this single completeness axiom.
How does high-school mathematics differ from university mathematics?
High-school mathematics computes inside a given real field. University mathematics rebuilds that field from axioms (proving theorems from completeness via the ε-δ method) and then swaps axioms to study other structures such as groups, rings, and topological spaces. The shift from computation to proof and abstraction, and from implicit to explicit axioms, is the essential difference.