Sum of Ordinate Powers Curves

Summary: §§367–373 of chapter-16-on-finding-curves-from-properties-of-the-ordinate. Given the symmetric condition on the two ordinates of a two-valued curve , Newton’s identities pin down one polynomial relation in . The cleanest closed form is the master equation , expressing the fact that if is one ordinate then is the other. An alternate form uses the product instead. The equations extend to negative-integer and fractional exponents.

Sources: chapter16 (§§367–373). Figure 78 (in figures76-80) for the setup.

Last updated: 2026-05-11.


The problem (§367)

For each abscissa , let be the two ordinates (figure 78). We want curves with

for constant. Since and , Newton’s identities convert the condition to a polynomial relation in :

Newton identity for Required condition
1 (the diameter case, two-ordinate-sum-and-product)
2, i.e.,
3, i.e.,
4
(Newton expansion)one polynomial relation linking

Solving for in terms of gives the curve equation , with still a free function of .

The ellipse (§367)

With , choose (linear in ). The curve is

a conic. Solving the quadratic gives . The radicand is bounded, so the curve is an ellipse, centered at the origin (since the symmetry exchanges branches via reflection). This is exactly the ellipse — its two ordinates have constant sum of squares, an instance of ellipse-sum-of-squares-conjugate-diameter.

The cubic (§369)

With , the curve equation is

or after clearing the denominator and choosing , :

This is a third-order line. Euler notes it belongs to the second species of the cubic classification (see cubic-species-classification).

The irrationality (§370)

For , the relation solved for gives

The radical is generally irrational, but was required to be a non-irrational function of . Euler’s resolution: from , the value of cannot be real unless the radical is taken with the positive sign. With this sign choice, has only two values (despite being double-valued formally), as the problem requires.

The general master equation (§371)

The closed-form pattern for all is

with any single-valued function of chosen freely. The argument is immediate: if is one of the two ordinates, then is the other (since they sum to ). The condition thus reads directly.

For : ; since , the equation expands to , which is . Similarly for , and so on.

The alternate form via (§372)

Eliminating instead of : since , substitute into with as the other ordinate to get

equivalently . Solving the quadratic in ,

This is two-valued for each as required, provided is non-irrational. The advantage of the form is lower degree.

Tables for integer and fractional exponents (§373)

The chapter tabulates the cases for negative-integer reciprocal sums:

Required propertyEquation (using )Equation (using )

And for fractional exponents:

Required propertyMaster equationRationalized form
, or , or
, or

The general claim is: one equation covers all cases, whether is a positive integer, negative integer, or fraction. Equivalent forms in also exist.

Why the master equation is so clean

The clean form packages three ingredients:

  1. Vieta — the two ordinates are and , dual roles played by the two terms.
  2. No symmetric function machinery needed — the condition reads directly without Newton’s identities.
  3. Universality in — the same equation works for integer, fractional, and negative exponents.

This is the chapter’s prototype of “right-side-of-the-equation thinking”: rather than expanding to via identities, encode the condition const by substituting for the two roots.

Figures

Figures 76–80 Figures 76–80