The bridge we still have to build
Look back at where this rung began and where it has arrived. Guide 1 cut a cone with a plane and called the slices conics — a purely spatial picture. Guides 2 through 4 threw the cone away entirely and defined the same curves flat on the page, by foci, directrices, and eccentricity: the parabola as points equidistant from a focus and a line, the ellipse and hyperbola by a sum or difference of two focal distances. Two completely different stories. We asserted they describe the same curves, but we never showed it. That missing proof is the gap this guide closes first.
Here is the honest difficulty. On the cone, the curve lives in tilted three-dimensional space and there is no focus in sight — just a slanted plane crossing a cone. On the page, the focus is the star of the show but the cone has vanished. To bridge them we must find the focus inside the cone — produce, from the pure geometry of cone and plane, the special point and the special line that the flat definition demands. In 1822 a Belgian mathematician, Germinal Pierre Dandelin, did exactly this with an idea of startling beauty: drop spheres into the cone.
Dropping spheres into the cone
Take a cone and a cutting plane that meets it in an ellipse. Now imagine inflating a sphere inside the cone, below the plane, until it is wedged snugly: it touches the cone all the way around in a horizontal circle, and it just kisses the cutting plane at a single point. Call that point F_1. Inflate a second sphere above the plane the same way, touching the cone in another circle and the plane at a point F_2. These two are the Dandelin spheres. The claim — almost unbelievable the first time you hear it — is that F_1 and F_2 are precisely the two foci of the ellipse.
The proof rests on one humble fact you met rungs ago: the two tangent segments drawn from an external point to a sphere have equal length — the spatial cousin of tangents from a point to a circle. Pick any point P on the ellipse. The straight line from the cone's apex through P runs along the surface of the cone and pierces both tangent circles, at points T_1 (on the lower sphere's circle) and T_2 (on the upper one's). Now watch the magic: the segment PF_1 and the segment PT_1 are both tangents from P to the lower sphere, so they are equal. Likewise PF_2 = PT_2.
P on the ellipse. F_1, F_2 = the two contact points (the foci).
T_1, T_2 = where line (apex -> P) meets the two tangent circles.
P F_1 = P T_1 (two tangents from P to the LOWER sphere)
P F_2 = P T_2 (two tangents from P to the UPPER sphere)
P F_1 + P F_2 = P T_1 + P T_2 = T_1 T_2
T_1 T_2 = distance along the cone between the two fixed
circles = the SAME constant for every P.
=> P F_1 + P F_2 = constant (the ellipse's defining property)Add them: PF_1 + PF_2 = PT_1 + PT_2 = T_1T_2, the distance measured along the cone from the lower tangent circle to the upper one. But T_1 and T_2 always sit on those two fixed circles, and the slant distance between two parallel circles on a cone is the same no matter which generating line you travel down. So PF_1 + PF_2 equals a constant that does not depend on P — which is exactly the two-foci definition of the ellipse from guide 3. The cone slice is the focal ellipse, proven, not asserted. The same construction, with one sphere in each nappe of the double cone, turns the constant sum into a constant difference and gives the hyperbola.
And where the directrix was hiding
The spheres give up the directrix too, which is the deeper payoff. Each Dandelin sphere touches the cone along a horizontal circle, and that whole circle lies in a flat plane. Intersect that plane with the cutting plane and you get a line — and that line is the directrix. With a little trigonometry comparing the tilt of the cutting plane against the half-angle of the cone, one shows that for any point P on the curve the ratio (distance from P to the focus) / (distance from P to that line) is a fixed number, the same for every P. That ratio is the eccentricity, and this is the focus-directrix property of guide 4 emerging straight out of the solid figure.
Every quadratic is a conic: the general equation
Now the second loose end. Across this rung the conics arrived in tidy 'standard' equations — y = x^2 for a parabola, x^2/a^2 + y^2/b^2 = 1 for an ellipse — but those assume the curve is politely centred and axis-aligned. The honest general object is the general second-degree equation: A x^2 + B x y + C y^2 + D x + E y + F = 0, the most general polynomial equation of degree two in x and y. The remarkable theorem is that every such equation, whatever its six coefficients, graphs to a conic — possibly tilted, possibly shifted, possibly degenerate, but always a member of the family.
The new character is the B x y cross term. The two linear terms D x and E y only slide the curve around — you neutralise them by completing the square, the same move you used to find a circle's centre. But B x y does something the others cannot: it rotates the curve off the axes. A hyperbola you already know intimately, x y = 1, is nothing but the standard hyperbola spun 45 degrees; the cross term is the spin written algebraically. To read such an equation you must first undo that rotation.
Undoing it is rotation of axes: turn the coordinate frame by a cleverly chosen angle theta so that, in the new tilted coordinates, the cross term disappears. Choosing theta to satisfy the condition that kills B (it works out to cot(2 theta) = (A - C) / B) leaves you a clean standard equation in the rotated frame, which you then recognise on sight. So the full recipe is: rotate to remove B x y, then translate by completing the square to remove the linear terms, and the bare standard conic is revealed.
The discriminant sorts them in one stroke
Often you do not actually need to rotate — you only need to know which kind of conic you have. There is a quantity built from the top three coefficients that answers this instantly and never changes when you rotate the axes: the discriminant B^2 - 4AC. Because rotation leaves it untouched (it is an invariant), you can read off the species of the curve directly from the messy tilted equation, without straightening anything. One sign, one verdict.
- If B^2 - 4AC < 0, the curve is an ellipse (and a circle is the special case A = C, B = 0). Think of x^2 + y^2 = 1: here B^2 - 4AC = -4 < 0.
- If B^2 - 4AC = 0, the curve is a parabola. Think of y = x^2, i.e. x^2 - y = 0: here A = 1, B = 0, C = 0, so B^2 - 4AC = 0.
- If B^2 - 4AC > 0, the curve is a hyperbola. Think of x y = 1, i.e. x y - 1 = 0: here A = 0, B = 1, C = 0, so B^2 - 4AC = 1 > 0 — a hyperbola in disguise, exactly as we said.
One honest caveat, and it matters. The discriminant names the type, but a few equations collapse into a degenerate conic — the leftovers you get when the cutting plane passes through the apex of the cone instead of cleanly across it. Then x^2 - y^2 = 0 is not a hyperbola but the pair of lines y = x and y = -x; x^2 + y^2 = 0 is a single point; x^2 = 1 is two parallel lines; and x^2 + 1 = 0 is nothing at all (no real points). The discriminant cannot see these by itself, so always glance at whether your equation actually factors or empties out before declaring a genuine curve.
What the conics rung leaves you holding
Three definitions, one family. The conics arrived as slices of a cone, were reborn as focus-and-directrix loci on the flat page, and were classified by a single number, eccentricity. The Dandelin spheres welded the spatial story to the flat one, and the general second-degree equation with its discriminant welded both to algebra: a conic is a plane section of a cone, and a focal locus, and a degree-two equation, and these are three views of the same objects. That triple identity is the quiet triumph of this rung.
And this is not where the conics' story ends — it is where it opens out. The reflective property you have met sends headlight beams and satellite dishes to a focus and lets planets trace orbits governed by a single focal equation. Looked at through the lens of the rungs still above you, a conic becomes a projective curve where ellipse, parabola and hyperbola merge into one shape distinguished only by how they meet the line at infinity — and, further up, the simplest interesting algebraic curve of all. You have learned the family completely at this level; the ladder above shows the same curves wearing grander names.