I am reading up on Stiefel-Whitney classes $w_k(T_M)$ on $m$-manifolds; $k=1,2,\dots,m$, which are described as obstacles to extending a trivialization of a bundle, from the $k$-skeleton (assume manifolds are nice-enough to admit a cell-decomposition); for $k=1$, it is just the obstruction to orientability (we just define $w_1$ over embedded circles $C$, and define $w_1(C)=0$ if $C$ preserves orientation, and $w_1(C)=1$ otherwise. A nowhere-zero vector field is equivalent to orientability)

Question: anyone know of an example of a manifold $M$, say for $\dim \leq4$ that admits a trivialization over its $1$-skeleton (i.e., $M$ is orientable), but so that the trivialization does not extend to the $2$-skeleton? (and, is this trivialization over the $1$-skeleton a line bundle as the restriction to the $1$-skeleton of the $\mathbb R^4$-bundle?)

Even better if there are examples of trivialization over $1$- and $2$-skeleton, but not $3$-skeleton, etc.

Thanks in Advance,


  • 7
    $\begingroup$ Most compact orientable surfaces work, no? $\endgroup$ May 21, 2011 at 0:24
  • 3
    $\begingroup$ In dimensions larger then $2$, the answer is any orientable manifold which does not admit a spin structure . Any trivialization over the $2$-skeleton automatically extends to the $3$-skeleton, since $\pi_2 SO_n$ is trivial. $\endgroup$ Dec 18, 2011 at 8:23

1 Answer 1


Well this has been open for a while, but you want a manifold with non-trivial second homology class right? What about $\mathbb{C}P^2$?

  • 1
    $\begingroup$ Dear cduston, Actually the question is answered by Mariano's comment. Regards, $\endgroup$
    – Matt E
    Nov 18, 2011 at 3:25
  • $\begingroup$ Sorry, but I think he wanted an example of a non-spinable manifold, of which $\mathbb{C}P^2$ is. Did I miss the point? $\endgroup$
    – levitopher
    Nov 23, 2011 at 22:46
  • $\begingroup$ Dear cduston, Actually, maybe I missed the point. I thought that he wanted a manifold whose tangent bundle could be trivialized over the $1$-skeleton, but not over the $2$-skeleton, hence any compact orientable surface of genus different from $1$ seems to do the job. But your interpretation, that he wants a manifold with $w_2 \neq 0$, also makes sense. (I presume that when you say "non-trivial second homology class", you mean non-trivial second Stiefel--Whitney class.) Regards, $\endgroup$
    – Matt E
    Nov 24, 2011 at 3:15

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.