The problem of exercising the freedoms of reparameterization of polyno
mial or rational curve segments to achieve a ''parametric flow'' close
st to the unit-speed or are-length representation is addressed. A quan
titative measure of ''closeness'' to are-length parameterization is fo
rmulated and, according to this measure, the problem of identifying th
e optimum rational reparameterization of a degree n polynomial curve i
s shown to be analytically reducible to the determination of the uniqu
e real root on (0, 1) of a quadratic equation. Examples indicate that,
in practice, the algorithm can produce significantly more uniform par
ameter variation across the extent of typical Bezier curves. The gener
alization of the method to reparameterization of rational curves is mo
re difficult, however, and does not admit generic reduction to a polyn
omial equation in even the simplest context (the conics).