Fix some $g \geq 2$, let $\Gamma_g$ be the mapping class group of a genus $g$ surface, and let $\pi : \Gamma_g \rightarrow Sp(2g,\mathbb{Z})$ be the projection. In

Meyer, Werner Die Signatur von Flächenbündeln. Math. Ann. 201 (1973), 239–264.

a $2$-cocycle $\tau_g$ on $Sp(2g,\mathbb{Z})$ is constructed with the following property. Let $X$ be an arbitrary closed surface and let $f : X \rightarrow \Gamma_g$ be a homomorphism. There is an associated fiber bundle $E_f \rightarrow X$ whose fiber is $\Sigma_g$. Letting $\tau(E_f)$ be the signature of the $4$-manifold $E_f$, we then have

$$-\tau(X) = f^{\ast}(\pi^{\ast}([\tau_g]))([X]) \in \mathbb{Z}.$$

Meyer also proves that $\tau(X)$ lies in $4 \mathbb{Z}$ and that any such integer can be achieved.

I have seen the following two claims in several places with references to Meyer's paper:

  1. The homology class $[\tau_g] \in H_2(Sp(2g,\mathbb{Z}))$ is divisible by $4$.
  2. The element $\frac{1}{4} [\tau_g] \in H_2(Sp(2g,\mathbb{Z}))$ generates $H_2(Sp(2g,\mathbb{Z}))$.

But these don't appear to actually be proven in Meyer's paper! Can anyone either tell me how to prove these or point me to a paper which contains the proofs?


At least the first statement is proved in Meyer's thesis (Die signatur von lokal...), and a newer proof can be found in:

MR1823605 (2002a:57038) Reviewed Hoster, Matthias(D-MNCH) A new proof of the signature formula for surface bundles. (English summary) Topology Appl. 112 (2001), no. 2, 205–213. 57R20 (55R10 55R40 57T10)

I am guessing that since both publications begin with "Die Signatur", and have the same author, people accidentally cite the wrong one.

| cite | improve this answer | |
  • $\begingroup$ Thanks, this is very helpful! I'll hold off on accepting it for the moment in case someone manages to answer my second question. $\endgroup$ – Patrick Apr 13 '14 at 23:44

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.