From the cracks to a cure
In the previous guide we caught Euclid red-handed. His very first proof builds an equilateral triangle by crossing two circles — and then quietly assumes the circles meet, though no postulate guarantees it. Elsewhere he slides a point 'between' two others as if betweenness were already defined, and he picks up a triangle and lays it on another to compare them, a motion no axiom licenses. None of these gaps make his conclusions wrong; they make his reasons incomplete. The fix is not to throw Euclid away but to rebuild the foundation so that every such step is backed by an explicit rule.
That rebuild is David Hilbert's, from his 1899 book Grundlagen der Geometrie (Foundations of Geometry). His goal was ruthless honesty: write down every assumption, hide nothing, and let no picture do work that a sentence has not earned. The result is the modern model of what the axiomatic method should look like. The system is now called the Hilbert axioms, and it organises everything into five families: incidence, order, congruence, continuity, and parallels. The rest of this guide walks through those five, one at a time.
Incidence and order: who lies on what, and what comes between
The first family, the incidence axioms, pins down the bare relationship of points lying on lines and planes. Two of them you already know in spirit: any two distinct points lie on exactly one line, and every line has at least two points on it. A third quietly forbids a degenerate world by demanding that at least three points do not all sit on one line — without it, 'geometry' could collapse to a single line and nothing interesting would ever happen. These axioms say nothing about distance or angle; they are pure 'which things touch which.'
The second family fixes order — the relation of one point being between two others, which Euclid used freely but never defined. Hilbert's order axioms make betweenness behave: if B is between A and C then B is between C and A; of any three points on a line exactly one is between the other two; and a line can always be extended past its endpoints. The crown jewel here is Pasch's axiom: if a line enters a triangle through one side, it must exit through one of the other two. That sounds laughably obvious, which is exactly why Euclid never wrote it down — and exactly why so many of his proofs secretly depended on it.
Why does Pasch matter so much? Because it is what lets you talk rigorously about the inside and outside of a figure. Once a line through a triangle must leave through a second side, the plane splits cleanly into a half-plane on each side of any line, and 'this point is inside the triangle' stops being a guess from the diagram and becomes a provable fact. Order axioms are how a flat sheet of plane earns its sense of in, out, and between.
Congruence without picking the triangle up
Euclid's most charming cheat was 'superposition' — he proves two triangles congruent by imagining one lifted and laid on the other. But sliding a figure through space is a motion, and no postulate of his guarantees that rigid motions exist or preserve shape. Hilbert refuses the trick entirely. Instead, his congruence axioms treat 'congruent' as a primitive relation between segments and between angles, governed by rules — you never move anything; you just compare.
The rules are the ones you would hope for. On any ray from a point, you can lay off a segment congruent to a given one in exactly one way (this is how a compass behaves, made axiomatic). Segment congruence adds: if AB is congruent to A'B' and BC to B'C' with B between A and C, then AC is congruent to A'C'. Angles get a matching pair of axioms. And then comes the load-bearing one: if two sides and the included angle of one triangle are congruent to those of another, the triangles are congruent. You will recognise that — it is side-angle-side. In Euclid it was a (flawed) theorem; in Hilbert it is taken as an axiom, because superposition can no longer be allowed to prove it.
Continuity: filling in the line with no gaps
Now we return to that very first crack: why do the two circles in Euclid's triangle construction actually meet? The answer is continuity — the line and plane must have no holes for the intersection point to live in. Hilbert handles this with two axioms working together. The first is the Archimedean axiom: given any two segments, you can lay copies of the smaller end to end until you pass the longer one. No segment is infinitely long compared to another; there are no 'infinitesimal' lengths hiding off to the side.
The Archimedean axiom alone is not enough — it rules out lengths that are too big or too small, but it still permits a line riddled with pinprick gaps, like the rationals, where a crossing point can slip through unmade. The second axiom seals every gap: Hilbert's completeness axiom, the geometric twin of Dedekind completeness. Informally, if you split a line into a 'left' part and a 'right' part with every left point before every right point, there is exactly one point at the seam. With that, the points on a line correspond perfectly to the real numbers, and two circles that ought to cross genuinely do.
no continuity: o---o---o---o---?---o---o (the crossing point
^ could fall in a GAP)
Archimedean: no length is infinitely larger or smaller
completeness: every "cut" of the line has exactly one seam point
together ==> points on a line <--> the real numbers, gap-free
so two circles that should meet, DO meet.Parallels — and why this group stands apart
The fifth and final family has just one member, and Hilbert states it in the clean modern form due to John Playfair: through a point not on a given line, there is exactly one line parallel to it. This Playfair axiom is logically equivalent to Euclid's tangled fifth postulate but far easier to reason with. Hilbert deliberately quarantines it as its own group, set apart from the other four, and that placement is the whole story of the next three guides.
Here is the deep point, stated honestly. The first four groups — incidence, order, congruence, continuity — make no commitment about parallels at all. Everything you can prove from those four alone belongs to what is called neutral geometry, the common ground shared by Euclidean and non-Euclidean worlds. The parallel axiom is the single dial you may set one way or another. Set it to 'exactly one parallel' and you recover ordinary Euclidean geometry; set it to 'more than one parallel' and a different but equally consistent geometry opens up. That this is a free choice, not a forced truth, is the liberation we hinted at earlier — and we will follow it properly two guides from now.
One caution, to keep us honest. Demanding more axioms is not free: the more you assume, the more you risk assuming something false — a hidden contradiction that would make the whole system worthless. So a natural fear arises. How do we know these five groups never quietly disagree with one another? And how do we know the parallel axiom truly is a separate dial, not secretly forced by the other four? Those are exactly the questions of consistency and independence, and answering them is the job of the very next guide.
What Hilbert actually bought us
Step back and look at the shape of the achievement. Hilbert did not discover new triangles or prove a surprising new theorem about circles. He did something quieter and more powerful: he made the rules of geometry complete and explicit, so that for the first time every line of every proof could be checked against a written axiom rather than against a persuasive picture. Euclid's results survive intact — what changed is that they now rest on ground you can audit.
Keep the five groups in your pocket as a checklist: incidence (what touches what), order (Pasch and betweenness), congruence (comparison without motion), continuity (Archimedean plus completeness, the line without gaps), and parallels (the one free dial). Together they are a precise, gap-free reading of the axiomatic method you met in the foundations rung. With them in hand, the next questions almost ask themselves — are the axioms consistent, are they independent, and what happens when you turn that last dial — and the rest of this rung answers each in turn.