MathOverflow is a question and answer site for professional mathematicians. Join them; it only takes a minute:

Sign up
Here's how it works:
  1. Anybody can ask a question
  2. Anybody can answer
  3. The best answers are voted up and rise to the top

This must have been known to the ancients, but I am having some trouble finding the references: what can be said (especially geometrically) about the normal closure of an element in a surface group? Especially an elliptic element (in an orbifold group)...

share|cite|improve this question
@Igor: what do you want to know about these subgroups? They are typically free products of cyclic groups generated by the conjugates of the element. – Mark Sapir Apr 9 '11 at 21:49
@Mark: what you say is already (a) interesting and (b) not obvious to me. Why is that true? For a torsion-free surface group, since the abelianization is 1-dimensional, the normal closure is of infinite index, hence infinitely generated free, but it is not clear what the canonical generators are (are all the $a g a^{-1}$ the free generators?) and with torsion that is less clear. Plus, if you look at the quotient by the normal closure, do you get something interesting? – Igor Rivin Apr 10 '11 at 0:35
@Igor: By a theorem of Delzant for every element $g$ of a torsion-free hyperbolic group $G$ the normal closure of some power $g^n$ is freely generated by conjugates $hg^nh^{-1}$. I think that in the case of surface groups $n$ should be 1. Every torsion-free hyperbolic group is SQ-universal by Olshanskii. The proof implies that every 1-related group embeds into the quotient you described. In fact if you take "any random" free subgroup $F$ of the surface group $S$, and every element $r$ in $F$, then $F/\ll r\gg$ naturally embeds into $S/\ll r\gg$. – Mark Sapir Apr 10 '11 at 1:22
@Igor: Delzant, Thomas(F-STRAS-I) Sous-groupes distingués et quotients des groupes hyperboliques. (French) [Distinguished subgroups and quotients of hyperbolic groups] Duke Math. J. 83 (1996), no. 3, 661–682. - the paper by Delzant. – Mark Sapir Apr 10 '11 at 2:09
@Mark: Thanks for the reference! I looked briefly at the paper, and if I understand correctly, the family of elements $h g^n h^{-1}$ always satisfies the small cancelation condition for $n$ large enough. That's not quite satisfying (since one can always get this kind of result by ping-pong in groups which are themselves very far from free (eg, lattices in higher rank lie groups). If you can actually get $n=1,$ that's a different matter entirely.... – Igor Rivin Apr 10 '11 at 2:42

This doesn't say much about the elliptic case, but there is Hempel's theorem that the normal closure of an element $\alpha$ contains a (power of a) simple curve $\beta$ if and only $\alpha$ is a (power of a) simple curve. Furthermore, any such $\beta$ is either a power of $\alpha$ or a commutator of $\alpha$ with a simple loop $\gamma$ intersecting $\alpha$ in one point.

You should be able to glean some structural information about the normal closure from his paper.

John Hempel, One-relator surface groups, Mathematical Proceedings of the Cambridge Philosophical Society (1990), 108: 467-474

share|cite|improve this answer
@Richard: thanks, I saw that paper (and there is a sequel by Jim Howie), but I am not quite sure yet what to make of it... – Igor Rivin Apr 10 '11 at 16:38
Sure thing. If you're interested in this kind of topological statement, there is also a theorem of Maskit that the covering corresponding to $\langle \langle \alpha \rangle \rangle$ is planar. – Richard Kent Apr 10 '11 at 18:22

Your Answer


By posting your answer, you agree to the privacy policy and terms of service.

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