General Construction by Parabola and Conic

Summary: A systematic method for constructing any equation of any degree as the intersection of two algebraic curves (§§499–504). Choose the first curve in the form , so that is single-valued in and no complex intersections arise. Then the second curve is obtained by substituting this into the target equation and rearranging. The standard worked case takes to be the parabola ; the matched second curve is a general conic, and the discriminant selects hyperbola, parabola, or ellipse — circle when and a further condition holds.

Sources: chapter20 (§§499–504).

Last updated: 2026-05-12.


The single-valued-first-curve principle (§499)

A real root of the target equation produces a real intersection of the two construction curves and iff the recovered ordinate is real. The cleanest way to guarantee this is to choose the first curve so that is a single-valued rational function of :

For such a curve, every real produces a real . So no matter what the second curve looks like — even if it has many branches and complex ordinates of its own — the intersection at every real root of the eliminant is real. The “shadow” of complex intersection in chapter 19 is removed.

Deriving the second curve (§500)

Given the target equation of degree , the recipe is:

  1. Pick defining the first curve (a curve of order at most).
  2. Substitute into and clear denominators.
  3. The resulting polynomial in and is the equation of the second curve. Its intersections with the first curve are exactly the roots of .

The order of the second curve is fixed by and the choice of .

Worked case: quartic via parabola and conic (§500)

Target equation:

First curve — parabola (rewritten ). This is order 2, but is single-valued in .

Second curve — substitute into the quartic:

Substituting and collecting,

This is the general second-order curve in , save for the missing -term (no ). The parameters remain free, giving an infinite family of valid second curves.

Even greater variety (§501)

The first curve can be multiplied by extra terms before substitution. Replace by (multiplying by ) and add this to the quartic before substituting in:

Three free constants instead of two; the missing -term reappears.

Selecting the conic type (§502)

The conic-type discriminant of the general second-order curve in is the standard (the cross-term squared minus four times the product of the diagonal coefficients):

  • : hyperbola
  • : parabola
  • : ellipse

(The trichotomy is the §131 [[classification-of-conics|sign of ]] applied to the second curve.)

Circle subcase: a conic is a circle iff its and coefficients are equal and the cross-term vanishes. The latter requires ; the former, — i.e. . With this choice the second curve becomes a circle whose center and radius can be read off:

This recovers the circle-and-parabola construction of §§493–495 as a special case of the general method.

Higher degrees via higher parabolic lines (§503)

For an equation of degree where has only awkward factorizations, take the first curve to be a higher-order parabolic line :

This is order but still single-valued. Substituting into the target equation reduces it to a polynomial in and of degree (rounded up), defining the second curve.

Example. (degree 12). Take the quartic parabola ; then , and substitution gives the third-order curve

Order matches the target.

Multiplying by powers of (§504)

If the natural construction (parabola + matching conic) misses the target — e.g., a cubic produces a hyperbola which is less convenient than a circle/parabola — multiply the target by a power of to add easily-recognized extra roots.

Cubic example. .

  • Substitution alone gives the hyperbola . Always a hyperbola — fine but inflexible.
  • Multiplying the cubic by gives the biquadratic (with ). Apply §§493–495 to construct this as a circle-and-parabola intersection. The extra root from the multiplication is the origin and is trivially recognized; the three actual roots are the remaining intersection abscissas.

This is the technique tying the cubic Bäcker’s-rule construction back to a circle-and-parabola figure.

Caveat on real-arc coverage (§505)

Even when the construction has no complex intersections, a real arc of the second curve may not span the full range of abscissas where the eliminant has real roots. Some real roots can then go unrealized — the corresponding intersection lies on a “different branch” of the curve, or at infinity, or on the discarded conjugate of the first curve. Euler dismisses this as more curious than useful: when one chooses the first curve in form, the worry rarely materializes.