Euclid’s Elements, Pasch’s Axiom

For centuries, the gold standard for mathematical reasoning wasn’t just inspired by Euclid’s Elements – it was Euclid’s Elements. Compiled around 300 BCE, this monumental 13-book collection systematically derived a vast body of geometry and number theory from a small set of explicit starting points. It begins with fundamental plane geometry (Book I covers basic constructions, congruence theorems like SAS, AAS, SSS, and culminates in the Pythagorean theorem). It progresses through circles, polygons, the groundbreaking theory of proportions developed by Eudoxus capable of handling irrational magnitudes (Book V), elementary number theory including primes and divisibility (Books VII-IX), classification of irrationals (Book X), and finally solid geometry, culminating in the construction and classification of the five Platonic solids (Books XI-XIII).

The foundation upon which this edifice rests consists of three types of initial statements:

Definitions: Descriptions of basic concepts like “a point is that which has no part” or “a line is breadthless length.” While intuitive, these are not operational definitions in the modern sense.
Postulates (P1-P5): These assert fundamental geometric capabilities: P1: Drawing a straight line between any two points. P2: Extending a finite straight line continuously. P3: Describing a circle with any center and radius. P4: All right angles are equal. P5: The famous (and controversial) parallel postulate.
Common Notions (CN1-CN5): General logical or arithmetic principles considered self-evident: CN1: Things equal to the same thing are equal to each other. CN2: If equals be added to equals, the wholes are equal. CN3: If equals be subtracted from equals, the remainders are equal. CN4: Things which coincide with one another are equal¹ (the basis for superposition proofs). CN5: The whole is greater than the part. Euclid’s genius was demonstrating how much could be logically deduced from just these principles.

This axiomatic-deductive method, demonstrated so powerfully in the Elements, became the very idea of how rigorous mathematical proof should proceed. The work stood as the paradigm of certainty for over two millennia. Yet, precisely because it felt so solid and complete, its subtle imperfections were slow to emerge. When mathematicians in the 19th century, driven by a quest for absolute certainty (think of the work on calculus foundations by Weierstrass, Cantor, Dedekind), began stress-testing its foundations, they discovered that Euclid’s arguments sometimes relied implicitly on visual intuition drawn from diagrams—assumptions about continuity, order, and intersections—rather than solely on his stated Definitions, Postulates, and Common Notions.

One of the first to really nail down a specific missing piece was Moritz Pasch in 1882. He pointed out something fundamental about how lines interact with triangles that Euclid just… assumed.

This discovery was part of a broader push for a truly modern axiomatic system. Enter David Hilbert. Around 1899, he published his Grundlagen der Geometrie (Foundations of Geometry), aiming to rebuild geometry from the ground up. Hilbert’s approach was radical:

Undefined Terms: Forget descriptive definitions. Start with undefined primitives: “point,” “line,” “plane,” “lies on,” “between,” “congruent to.” Their meaning comes only from the axioms they satisfy.
Axiom Groups: He organized axioms into logical clusters: Incidence (I), Order (II), Congruence (III), Parallels (IV), Continuity (V). This allowed studying their roles and independence.
Pure Logic: Proofs must be purely formal deductions, no diagrams needed.

Hilbert’s system became the new standard. But the story didn’t end there. Later, Alfred Tarski (starting in the 1920s) developed another axiomatization with different goals.

Different Primitives: Tarski used just “point,” a three-term relation for “betweenness,” and a four-term relation for “congruence” (segment $ab$ is congruent to segment $cd$ ).
First-Order Logic: Crucially, Tarski formulated his axioms entirely within first-order logic – the basic logic of “for all” ( $\forall$ ) and “there exists” ( $\exists$ ) quantifiers, logical connectives (and, or, not), and relations between variables.
Metamathematical Power: Why does first-order logic matter? It opens the door to powerful tools from mathematical logic. Tarski achieved stunning results: he proved elementary geometry (based on his axioms) is complete (every statement expressible in the language is either provable or refutable) and decidable (there’s an algorithm to determine the truth of any statement!). This came via a deep technical property called quantifier elimination.

So, the journey to understand geometry’s foundations took us from Euclid’s powerful method, through Pasch and Hilbert’s focus on missing axioms and rigorous proof, to Tarski’s logical analysis using the machinery of modern logic.

Let’s zoom in on that specific gap Pasch found, see why Hilbert cared about proving its independence using weird algebraic tricks, and how it fits into this bigger picture.

What Pasch Pointed Out (The “Obvious” Insight)

Imagine a triangle formed by three non-collinear points $a, b, c$ . A line segment, say $ac$ , is intuitively the set of points between $a$ and $c$ . Now, draw a line $L$ that crosses this segment $ac$ but doesn’t hit any of the vertices $a, b,$ or $c$ . Where does the line go next? It must, it seems, emerge from the triangle by crossing one of the other two segments, $ab$ or $bc$ . It can’t just vanish inside, or somehow loop back out through $ac$ .

Pasch made this formal. Let $B(x,y,z)$ mean “point $y$ is strictly between points $x$ and $z$ “.

Pasch’s Axiom (PA): If a line $L$ (not hitting vertices $a,b,c$ ) intersects segment $ac$ (i.e., $\exists d \in L$ with $B(a,d,c)$ ), then it must intersect either segment $ab$ (i.e., $\exists e \in L$ with $B(a,e,b)$ ) or segment $bc$ (i.e., $\exists f \in L$ with $B(b,f,c)$ ).

This axiom is the bedrock of plane separation – the idea that a line divides the plane into two distinct regions or “sides”. This property is fundamental; without it, we can’t rigorously talk about polygons having an interior and exterior, or prove many basic topological facts about the plane (it’s a precursor to things like the Jordan Curve Theorem). Euclid used these ideas constantly, but Pasch insisted they needed an explicit axiomatic basis.

Is It Free? The Independence Puzzle

Could Hilbert derive PA from his other axioms (Incidence I, Congruence III, Continuity V)? If so, it would be logically implied, not fundamental. To prove independence, Hilbert needed two “models”:

Model $M_1$ : A geometry where I, III, V and PA hold. The standard Euclidean plane $\mathbb{R}^2$ works. Points are pairs $(x,y)$ of real numbers. Lines are solutions to linear equations. Congruence uses the Pythagorean distance $\sqrt{(x_a-x_b)^2 + (y_a-y_b)^2}$ . Crucially, PA holds here because the real numbers $\mathbb{R}$ are complete (they have the least upper bound property, or equivalently, every Dedekind cut corresponds to a real number). This completeness ensures lines are continuous and truly separate the plane into two connected half-planes (the Plane Separation Property, PSP), from which PA follows directly.
Model $M_2$ : A geometry where I, III, V hold, but PA fails. This requires more ingenuity.

The existence of both models demonstrates PA is independent – it’s an essential assumption, not a consequence.

Sketching the Weird Universe ( $M_2$ )

Hilbert’s stroke of genius for $M_2$ was realizing that coordinate geometry could be built over fields other than the real numbers. He needed a field $F$ with specific properties:

It must be an ordered field: It needs a total order relation $<$ that respects the field operations (if $a < b$ , then $a+c < b+c$ ; if $a < b$ and $c>0$ , then $ac < bc$ ). This allows comparison of coordinates and defining $B(a,d,c)$ by checking if $d$ ‘s coordinate lies between $a$ ‘s and $c$ ‘s according to $<$ .
It must support congruence: The axioms of congruence (especially SAS, Side-Angle-Side) need to hold. Using the standard “squared distance” $d^2(a,b)=(x_a-x_b)^2+(y_a-y_b)^2$ requires properties guaranteed by ordered fields (like sums of squares being positive). Making congruence fully work often involves ensuring certain square roots exist in $F$ .
The Crucial Twist: Non-Archimedean Order! Hilbert chose an ordered field $F$ that violates the Archimedean Property. This means $F$ contains infinitesimals ( $\epsilon > 0$ but $\epsilon < 1/n$ for all positive integers $n$ ) and their reciprocals, infinite elements ( $M > n$ for all $n$ ). Think of a number system with elements “infinitely smaller” or “infinitely larger” than any standard rational or real number.

How Pasch Fails in $F^2$

Now, imagine the coordinate plane $F^2$ built over such a non-Archimedean field $F$ . Define “betweenness” $B(a,d,c)$ purely algebraically using the non-Archimedean order $<$ . Here’s the conceptual failure: A line $L$ can intersect segment $ac$ of triangle $abc$ . Algebraically, this just means there’s a point $d$ on $L$ whose coordinate(s) lie between those of $a$ and $c$ according to the order $<$ . But, because the field $F$ has infinitesimals, the line $L$ might pass infinitesimally close to the vertex $b$ . Because $F$ contains infinitesimals, the difference in coordinates between points on $L$ near $b$ , and $b$ itself, might be infinitesimal. The line L might “slip through” an infinitesimal gap. Even though visually it seems to cross segment $ab$ or $bc$ , there might be no point $e$ on $L$ whose coordinates are strictly between $a$ and $b$ according to $<$ , and similarly no $f$ strictly between $b$ and $c$ . The algebraic definition of betweenness, rooted in the non-Archimedean order, fails to capture the topological notion of crossing in this weird space. The line enters ( $B(a,d,c)$ is true) but doesn’t register an exit ( $B(a,e,b)$ and $B(b,f,c)$ are false for all $e,f \in L$ ). Pasch’s axiom fails.

Why Independence Matters

Hilbert’s $M_2$ construction definitively showed that the intuitive properties of order and plane separation captured by PA are not guaranteed by simply having points, lines, and a notion of congruence. Order is an independent pillar of Euclidean geometry’s structure.

Pasch’s Ghost in Euclid’s Machine? Proposition I.7

So, did Euclid stumble over this specific gap? Many point to Proposition I.7. It essentially states that you cannot construct two distinct triangles $ACB$ and $ADB$ on the same base segment $AB$ , on the same side of $AB$ , such that the corresponding legs are equal ( $AC=AD$ and $BC=BD$ ). This proposition is crucial; Euclid uses it to prove the fundamental Side-Side-Side (SSS) congruence theorem (I.8).

The problem lies in “on the same side“. Euclid provides no axiom or definition for this. Rigorously defining the “sides” of a line requires the Plane Separation Property, which, as we saw, is deeply connected to Pasch’s Axiom and the other axioms of order. Without these, the concept of “side” has no formal meaning, and the proof of I.7 (and subsequently I.8) rests on shaky ground.

A Word of Caution: While I.7 is a compelling example, attributing the first need for order axioms solely to it is complex. Even Proposition I.1 (constructing an equilateral triangle) implicitly assumes that the two circles drawn must intersect at some point. This intersection property isn’t guaranteed by Euclid’s postulates alone and relies on unstated continuity assumptions, which are related to, but distinct from, order. The foundations are deeply interwoven, but I.7’s direct appeal to “sides” makes its reliance on order particularly clear.

Tarski’s Logic and Geometry’s “Tameness”

Returning to Tarski, his first-order axiomatization (which included an equivalent of Pasch’s axiom) allowed for powerful logical analysis. His proof that this theory admits quantifier elimination was key. Conceptually, it means that any geometric statement, no matter how complex with nested “for all” ( $\forall$ ) and “there exists” ( $\exists$ ) quantifiers, can be systematically translated into an equivalent statement involving only basic algebraic operations (polynomial equations and inequalities) on coordinates, without any quantifiers.

This technical result has profound consequences:

Decidability: Quantifier elimination provides a recipe (an algorithm) for determining the truth or falsehood of any statement. Although computationally very expensive, its existence proves that elementary geometry is decidable. Unlike Peano arithmetic, where Gödel showed undecidable statements must exist, there are no such fundamentally unanswerable questions in basic Euclidean geometry. It’s logically “tame”.
Completeness: Decidability also implies completeness for Tarski’s system: every statement expressible in its language is either provable or refutable from the axioms. There’s no room for independent statements like the Continuum Hypothesis in set theory.

Tarski thus showed that Euclidean geometry, when carefully axiomatized within first-order logic, is a remarkably stable and predictable mathematical structure from a logical standpoint.

Algebra <-> Geometry: Still True!

The deep connection shines through: Hilbert used algebraic field properties to probe geometric axiom independence. Tarski linked the decidability of geometric statements to the algebra of real closed fields via quantifier elimination. Different layers of geometric structure (Incidence, Order, Congruence, Parallels, Continuity) correspond intricately to different layers of algebraic structure (field operations, ordering, metric properties, specific field characteristics).

Conclusion: From Gaps to Foundations

The journey from Euclid’s intuitively grounded Elements to the modern axiomatic understanding reveals the layers beneath familiar geometry. Pasch exposed a crucial gap in handling order. Hilbert, using non-standard algebraic worlds, proved this order was a logically independent requirement. Tarski, employing the lens of first-order logic, mapped the boundaries of geometric decidability, linking it profoundly back to algebra. Exploring these foundations is more than just historical footnote-checking; it’s about sharpening our understanding of proof, illuminates the interplay between intuition and rigor, and reveals the deep, elegant connections between the study of space, number, and logic.

Share this:

Related

Leave a comment Cancel reply