While dealing with $BO(n)$, $BSO(n)$ and $BSpin(n)$ with the universal coefficient theorem and Künneth formula, I came to have the following question:

The universal coefficient says $H^n(X;M)\cong \hom(H_{n}(X;\mathbb{Z}),M)\oplus {\rm Ext}^{1} (H_{n-1}(X;\mathbb{Z},M))$ for a $\mathbb{Z}$-module $M$.

When $X=BSpin(n)$, we know that $H^4(BSpin(n);\mathbb{Z})\cong \mathbb{Z}$ and it seems likely that once we know what $H_p(BSpin(n);\mathbb{Z})$ would be for $p=3,4$ we might be able to retrieve this isomorphism with the aid of universal coefficient theorem.

So what would be $H_p(BSpin(n);\mathbb{Z})$ at least for $p=0,1,2,3,4$?

(I have to say that the question is not about how to prove the isomorphism $H^4(BSpin(4);\mathbb{Z})\cong \mathbb{Z}$.)


  • $\begingroup$ The isomorphism you say this question is not about is not an isomorphism - $H^4(\mathrm{BSpin}(4);\mathbb{Z})\cong \mathbb{Z}^2$. See also my comment on the accepted answer. $\endgroup$ Feb 23, 2021 at 13:13

1 Answer 1


Consider the classifying space $EG$ of a given group $G$. In your case, $G={Spin}(n)$. We have a fibration $G\to EG\to BG$ where $EG$ is contractible and $G$ acts freely on it. Therefore, the long exact sequence in homotopy groups tells you that $\pi_j(BG)\cong \pi_{j-1}(G)$. But $G={Spin}(n)$ is a Lie group which is simply connected. It is also classically known that $\pi_2(G)=0$ for finite dimensional Lie groups $G$. And, we also know that $\pi_3({Spin}(n))=\mathbb{Z}$. This implies that $\pi_j(B {Spin}(n))=0$ for $j=0,1,2,3$. Moreover, $\pi_4(B {Spin}(n))\cong H_4(B{Spin}(n);\mathbb{Z})$ by Hurewicz theorem. This is also isomorphic to $\pi_3({Spin}(n))$. The universal coefficient theorem now tells you that $H^4(B{Spin}(n);\mathbb{Z})=\mathbb{Z}$.

  • $\begingroup$ I didn't thought about Hurewicz theorem and I guess that answers my question. Thanks a lot. $\endgroup$
    – HBS
    Nov 30, 2012 at 3:33
  • 3
    $\begingroup$ It may be worth commenting that a beautiful paper of Quillen ``The mod 2 cohomology rings of extra-special 2-groups and the spinor groups'' computes the mod 2 cohomology of BSpin(n). Unusually, this is done without knowing the mod 2 cohomology of Spin(n) as a Hopf algebra. That is computed by Zabrodsky and myself (math.uchicago.edu/~may/PAPERS/19.pdf). $\endgroup$
    – Peter May
    Nov 30, 2012 at 4:59
  • $\begingroup$ Do we have any result on $H_5(BSpin(n);\mathbb{Z})$ and $H_6(BSpin(n);\mathbb{Z})$ (say for n=10)? We know that $\pi_d(Spin(10))$ are $0,0,\mathbb{Z},0,0,0$ for $d=1,2,3,4,5,6$, which are simple. $\endgroup$ Feb 18, 2013 at 14:06
  • $\begingroup$ @Xiao-Gang : Write the spectral sequence (in homology) for the fibration $G\to EG\to BG$. It will follow from the $E^5$ and $E^6$ page that $H_5(BSpin(n);\mathbb{Z})=H_4(Spin(n);\mathbb{Z})$ and $H_6(BSpin(n);\mathbb{Z})=H_5(Spin(n);\mathbb{Z})$. It follows from Milnor-Moore (with your input on the homotopy groups of $Spin(n)$) that these are torsion groups. Now you just have to compute these! $\endgroup$ Feb 19, 2013 at 7:21
  • $\begingroup$ This is an old thread, but it's worth pointing out that the statement needs amendment for n=4. As $\mathrm{Spin}(4)$ is isomorphic to $\mathrm{Spin}(3) \times \mathrm{Spin}(3)$, $\pi_3(\mathrm{Spin}(4))= \mathbb{Z}^2$. Following the argument in this answer one now concludes that indeed, $H^4(\mathrm{BSpin}(4);\mathbb{Z})$ is free on two generators. (See e.g. the nLab entry on Spin(4).) $\endgroup$ Feb 23, 2021 at 13:10

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.