Prove that a regular curve parametrized by arc length such that all its normal lines pass through a single point is contained in a circumference.

Suppose $\alpha:I\rightarrow \mathbb{R}^3$ is a curve such that all its normal lines pass through $P$. Then we can write that for every $s\in I$ there exist a $r_a$ such that:

$$\alpha(s)+r_s n(s)=P$$


$$r_s n(s)=P-\alpha(s)$$


$$\alpha'(s) \cdot (P-\alpha(s))=\alpha'(s)\cdot r_s n(s)=0$$

since $(P-\alpha(s))'=-\alpha'(s)$ which implies that the distance $|P-\alpha |$ is constant.

My question is, how do I prove that $\alpha$ is in a circumference not in a sphere.



To show that the curve lies on a plane, it is enough to show that it is torsion-free (I'll let you figure out why) or that its binormal vector is constant (same). In your situation, the quickest way to show this is probably to use the Frenet-Serret formulas.

  • $\begingroup$ I know that I should show that the curve is torsion-free and I think I can derive this from the equation $ \alpha(s)+r_s n(s)=P$ but I don't know why $r_s$ is a derivable function to apply the rule for multiplication. $\endgroup$ – YTS Sep 22 '14 at 21:29
  • $\begingroup$ Just differentiate again. Assume $\alpha$ is arclength-parametrized, for starters. $\endgroup$ – Ted Shifrin Sep 23 '14 at 0:11
  • 1
    $\begingroup$ If I'm not mistaken, $r_s = (P - \alpha_s)\cdot n(s)$,right? So $r_s$ is a differentiable function of $s$. I think you know how to proceed. $\endgroup$ – Seub Sep 23 '14 at 1:21

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.