A Strange New Symmetry
By now you have built the projective plane by gluing a point at infinity onto each direction and a single line at infinity holding them all. The payoff was clean: in this plane any two distinct lines meet in exactly one point, with no annoying parallel exception. And of course any two distinct points still lie on exactly one line. Read those two sentences back to back and something almost suspicious jumps out — they are the same sentence with the words point and line swapped.
This is no accident, and it is the whole story of this guide. In the projective plane points and lines stand on an astonishingly equal footing. The only relationship we ever truly need between them is the one incidence relation: a point 'lies on' a line, or equally a line 'passes through' a point. Notice that 'lies on' and 'passes through' are just two phrasings of the very same fact — A is on line m exactly when m goes through A. The relation does not care which of the two objects you call the subject.
Why the Swap Is Honest, Not Magic
It would be easy to leave duality as a charming slogan — 'just swap the words.' But a slogan is not a proof, and the reason the swap actually preserves truth deserves to be spelled out. The honest version comes straight from the homogeneous coordinates you met in the last guide. A point there is a triple (x, y, z), not all zero, with scaling ignored. A line is the equation a x + b y + c z = 0, and that line is pinned down by its own triple of coefficients (a, b, c), again with overall scaling irrelevant.
Stare at the incidence condition a x + b y + c z = 0. The point lies on the line exactly when this sum is zero. But the expression is perfectly symmetric in the two triples: it pairs (x, y, z) with (a, b, c) and asks whether they are 'orthogonal' in this projective sense. Swap which triple you call the point and which you call the line, and the very same equation reads true. So a point and a line are each just a triple-up-to-scale, and incidence is one symmetric equation linking them. Swapping the names cannot break a theorem because the underlying algebra never noticed which was which.
point P = (x, y, z) line m = [a, b, c] incidence: a x + b y + c z = 0 read one way: point P lies on line m read other: line m passes through point P the expression is symmetric in (x,y,z) <-> (a,b,c) so swapping 'point' and 'line' preserves every incidence
The Dictionary You Translate By
To use duality in practice you keep a small dictionary and translate a true statement word by word into its dual. 'Point' becomes 'line' and 'line' becomes 'point.' 'The point lies on the line' becomes 'the line passes through the point.' Two phrases dualize especially nicely: 'the line joining two points' turns into 'the point of intersection of two lines,' because joining points is the exact mirror of intersecting lines. Three points being collinear (sharing one line) dualizes to three lines being concurrent (sharing one point).
- Write the theorem using only incidence language: points, lines, 'lies on', 'passes through', 'collinear', 'concurrent', 'the line joining', 'the point meeting'. Strip away any mention of distance, angle, or betweenness — those are not projective and have no dual.
- Swap every 'point' with 'line' and every 'line' with 'point' throughout the whole statement, leaving the logical skeleton untouched.
- Swap the paired phrases too: 'collinear' with 'concurrent', and 'the line joining two points' with 'the point common to two lines'.
- Read the new sentence aloud. By the duality principle it is automatically a theorem — already proved, with no separate work needed, as long as the original lived purely in the projective plane.
Watching Twins Pop Out
Let us run the machine on the two facts we opened with. Start from 'two distinct points lie on exactly one line.' Translate: point becomes line, line becomes point, and 'lie on' becomes 'pass through.' Out comes 'two distinct lines pass through exactly one point' — that is, any two lines meet in a single point. The hard-won meeting theorem of the projective plane is simply the dual of the obvious fact that two points determine a line. One sentence, dualized, hands you the other for nothing.
A juicier example is coming in the next guide, but it is worth previewing. Desargues' theorem says that if two triangles are in perspective from a point — three lines through corresponding vertices all meet at one centre — then they are in perspective from a line, meaning the three intersection points of corresponding sides are collinear. Dualize it and you get a statement that turns out to be its own converse: the dual of Desargues is essentially the converse of Desargues. Some configurations are so balanced that duality folds them back onto themselves.
When a figure dualizes into a figure of the same shape, we call it a self-dual configuration. The humblest example is a triangle: three points and the three lines joining them. Swap points and lines and you get three lines and the three points where they cross — the very same triangle, just described from the line side. A more striking one is the complete quadrangle versus the complete quadrilateral, dual partners that the next guide will use to pin down the harmonic relation among four points.
Pole and Polar: Duality You Can Draw
Duality so far has been a logical operation on sentences. But there is also a concrete, drawable version that attaches a definite line to each point and a definite point to each line, once you fix a conic. Given a circle (or any conic) and a point P, the construction of pole and polar produces a specific line p called the polar of P; run it backwards and the line p hands back the point P as its pole. This is duality made visible: a genuine geometric correspondence trading points for lines and back.
The beautiful coherence is that pole-polar respects incidence in a dual way: if a point Q lies on the polar of P, then P lies on the polar of Q. Incidence on one side becomes incidence on the other, perfectly. So this construction is not just an analogy to the word-swap duality — it is a faithful model of it, drawn in ink with a conic as the dictionary. Abstract duality and this concrete pole-polar duality are two faces of the same coin.
What Duality Does Not Do
It is tempting to over-trust duality, so let us draw the boundary honestly. Duality is exact only for statements built purely from incidence — points, lines, and which lies on which. The moment a claim mentions a distance, an angle, congruence, betweenness, or 'the midpoint', it has left projective territory, and there is no dual to it. There is no length-preserving partner waiting on the other side, because lengths are not projective quantities; they are not even preserved by the projective transformations of the previous rung.
A second honest caveat: duality guarantees that the dual statement is a theorem, but it does not hand you a clever or illuminating proof of it. The proof you get is exactly the dual of the original proof, line for line. That is logically airtight, yet it may feel mechanical, and it will never tell you anything the original did not already encode. Duality doubles your stock of theorems; it does not double your supply of fresh ideas. Both halves of that sentence are worth holding onto.