I want to make a project at differential geometry about the Hairy Ball theorem and its applications. I was thinking of including a proof of the theorem in the project. Using the Poincare-Hopf Theorem seems easy enough, but I was thinking that this proves the desired result using a stronger theorem (just like proving Liouville's Theorem in complex analysis using Picard's theorem).

Is there a simple proof of the fact that there is no continuous non-zero vector field on the even dimensional sphere? It is good enough if the proof works only for $S^2$, because that is the case I will be focusing on in the applications.

  • 14
    $\begingroup$ [insert joke about hairy balls here]. $\endgroup$ – zzzzBov Dec 12 '11 at 15:36

There is an elegant and self-contained proof by Milnor: Analytic proofs of the "hairy ball theorem", American Math. Monthly, 85 (1978), 521-524. The paper is reprinted in his collected works and can be downloaded for pay here:


  • $\begingroup$ Thank you. I'll probably use this :) $\endgroup$ – Beni Bogosel Dec 12 '11 at 12:30
  • 11
    $\begingroup$ Please consider adding the proof to your answer. Linking to an answer behind a paywall is not nearly as useful to the community. $\endgroup$ – Eric Wilson Dec 12 '11 at 13:43
  • 1
    $\begingroup$ @Eric Wilson: I don't have the paper at my fingertips, but I remembered it as a real gem, so I retrieved its coordinates. I regret the paywall as much as anybody else. Fortunately the Monthly is widely available in math libraries. $\endgroup$ – Christian Blatter Dec 12 '11 at 15:04
  • 1
    $\begingroup$ @BeniBogosel If you (or anyone else) found this useful, perhaps you can edit the answer to include a summary of the contents. $\endgroup$ – Eric Wilson Dec 12 '11 at 15:14
  • 3
    $\begingroup$ @EricWilson: The sketch of the proof can be found in the following blog post: topologicalmusings.wordpress.com/2008/07/22/… I thank Christian Blatter for posting the link. In the library of my university I could read it. The blog post summarizes the proof. $\endgroup$ – Beni Bogosel Dec 12 '11 at 18:04

The simplest I can remember off the top of my head is this:

Assume there is such a vector field. Let $v_x$ denote the vector at the point $x$. Now, define the homotopy $H: S^2\times [0, 1] \rightarrow S^2$ by the following: $H(x, t)$ is the point $t\pi$ radians away from $x$ along the great circle defined by $v_x$. This gives a homotopy between the identity and the antipodal map on $S^2$, which is impossible, since the antipodal map has degree $-1$. Hence there can be no such vector field.

  • 1
    $\begingroup$ this also works for general dimensions, with the same argument. define $H(x,t) := \cos(t) \, x + \sin(t) \, \nu_x$, as in hirsch's differential topology, which gives you a homotopy between $id_{\mathbb S^m}$ and $-id_{\mathbb S^m}$. we have $\deg (-id_{\mathbb S^m}) = (-1)^{m+1}$. $\endgroup$ – cesare borgia Jan 30 '17 at 23:49
  • $\begingroup$ Note that reflection in one coordinate has degree -1, so that the map from x to -x has degree (-1)^(n+1) where n is the dimension of the sphere. Bonus fact that any fixed point free map from the sphere to itself is homotopic to the antipodal map via parameterized normalized line segments between -x and f(x), necessarily not crossing the origin. $\endgroup$ – Morph Mar 23 '17 at 4:54

The proof can be found on this site:


  • $\begingroup$ Right, except this is not "the proof" but is "Milnor's proof". $\endgroup$ – Moishe Kohan Sep 18 at 23:41

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.