One curve wearing three masks
Back on the conics rung you learned the three curves as distinct citizens, each with its own equation and its own eccentricity: the ellipse closed and tidy, the parabola open with one arm, the hyperbola flung into two branches. The single number eccentricity told them apart — below one an ellipse, exactly one a parabola, above one a hyperbola. That classification is perfectly true in the flat Euclidean plane. But it is a classification by shape, and projective geometry has spent this whole rung teaching you that shape is not what it cares about.
Recall the central move of this rung. In guide 1 you adjoined a point at infinity for every direction, and all of them together formed the line at infinity, completing the ordinary plane into the projective plane. In guide 2 you gave every point three homogeneous coordinates (x : y : z), so that the finite point (a, b) became (a : b : 1) and a point at infinity got z = 0. A projective transformation can then drag the line at infinity to any line it pleases, and that single freedom is exactly what melts the three conics into one.
It is all in how you meet the horizon
Here is the picture that makes the merger feel inevitable. Take any conic and ask one question: how many times does it cross the line at infinity? A projective conic is just the zero set of a homogeneous second-degree equation in (x : y : z), and you find its meeting with the horizon by setting z = 0 and solving what is left — a plain quadratic in x and y. A quadratic over the reals has two roots, one double root, or no real roots, and those three cases are precisely ellipse, parabola, and hyperbola.
Conic in homogeneous coords, then set z = 0 to meet the horizon:
x^2 + y^2 - z^2 = 0 --> z=0 gives x^2 + y^2 = 0
no real solution => ELLIPSE
(misses the line at infinity)
y z - x^2 = 0 --> z=0 gives x^2 = 0
one double root => PARABOLA
(tangent to the line at infinity)
x^2 - y^2 - z^2 = 0 --> z=0 gives x^2 - y^2 = 0
two roots x=y, x=-y => HYPERBOLA
(crosses the line at infinity twice)Read that off slowly, because it reframes everything you knew. An ellipse never reaches infinity — it misses the horizon entirely, no real points with z = 0. A parabola is tangent to the line at infinity: its two ends both run off toward one single direction, meeting the horizon in one doubled point. A hyperbola crosses the horizon in two distinct points, and those points are exactly the directions of its two asymptotes — the asymptotes are the lines through the conic's two points at infinity. The eccentricity test of the old rung and this crossing-count are the very same distinction, told once in metric language and once in projective language.
Now the punchline. A projective transformation can carry any line to any other line, so it can send the line at infinity to a line that misses your conic, or touches it, or cuts it twice — your choice. Apply the right transformation and you turn a hyperbola into an ellipse, or an ellipse into a parabola, simply by relocating where 'infinity' sits relative to the curve. The curve in the projective plane never changed; only the horizon moved. That is the precise, constructive meaning of 'all conics are one' — not a slogan, but a transformation you can actually write down.
Five points pin a conic down
If conics are this flexible, what nails one in place? The clean answer is a counting argument you can almost do in your head. The general homogeneous second-degree equation A x^2 + B x y + C y^2 + D x z + E y z + F z^2 = 0 has six coefficients, but multiplying the whole equation by a non-zero constant gives the same curve — so only their ratios matter, leaving five genuine degrees of freedom. Each point you require the conic to pass through imposes one linear condition on those coefficients. Five points, five conditions, and the conic is determined.
This is the theorem of the conic through five points: through any five points, no three of them collinear, there passes exactly one conic. It is the conic world's answer to 'two points determine a line', scaled up by the extra degrees of freedom. Notice the honest fine print in 'no three collinear': if three of your five points lined up, a conic could only catch all three by containing that whole line, which forces it to be a degenerate conic — a pair of lines rather than a smooth ellipse or hyperbola. The general-position clause is exactly what keeps the answer a genuine curve.
Duality hands you the pole and polar
Now collect the dividend that guide 3 promised. The principle of duality swaps points and lines throughout projective geometry, and a conic is exactly the kind of object that turns this swap from a slogan into a tool. Fix a conic. To any point P the conic assigns a line, and to any line the conic assigns a point, in a way that is mutually consistent — the line attached to P is its polar, and P is in turn the pole of that line. This pairing is the conic's own personal version of duality, a point-line correspondence woven from the single curve.
The construction is concrete, not mystical. If P sits outside the conic, draw the two tangent lines from P to the curve; they touch at two points, and the line joining those two touch-points is the polar of P. If P sits on the conic, its polar is simply the tangent line there. And if P sits inside, you recover the polar by drawing any two chords through P and joining the diagonal points of the resulting complete quadrangle — the same quadrangle gadget from guide 4 doing quiet structural work again. Crucially, the whole construction uses only straightedge incidence; no lengths, no angles, nothing metric. That is why pole and polar are genuinely projective.
The pole-polar pairing also obeys a perfectly reciprocal law that is itself a tiny instance of duality: if point P lies on the polar of point Q, then Q lies on the polar of P. Such a P and Q are called conjugate with respect to the conic, and this reciprocity is what lets you dualise any theorem about points on a conic into a matching theorem about tangent lines. The conic, in other words, is not just a curve sitting in the projective plane — it is a machine that realises the abstract duality of guide 3 as an honest, drawable correspondence.
Why the projective view is worth the climb
Step back and feel what this rung bought you. On the conics rung the three curves were stubbornly separate, each needing its own theorems. Here a single homogeneous equation covers all three, a single counting argument (five points, one conic) governs them uniformly, and a single duality (pole and polar) runs through every one. Statements that took three cases on the flat page collapse to one case in the projective plane — and that economy is not cosmetic, it is the whole reason mathematicians reach for the projective viewpoint when they study these curves seriously.
Two honest boundary markers before you move on. First, projective unification deliberately discards the metric data — foci, directrices, eccentricity, the very things the conics rung built — so the two viewpoints are partners, not rivals: when you need to aim a satellite dish you go back to the focus, and when you want the cleanest classification you come up here. Second, even within projective geometry the degenerate conics (line-pairs, a double line, a single point) are a separate stratum the smooth classification sets aside, and serious work has to handle them explicitly. Naming where a beautiful idea stops is part of holding it honestly.
And the view from here keeps opening upward. The pole-polar machine, the five-point theorem, the merging of the three conics — these are the first real fruits of treating geometry as the study of what survives a projective transformation. The rungs above frame that idea grandly: a geometry simply is a group of transformations and the invariants it preserves, with projective geometry near the foundation and Euclidean geometry sitting inside it as the special case that also remembers distance and angle. The conic, which looked like three curves down below, is one of the clearest places to watch that whole hierarchy snap into focus.