From: Jim Propp <propp@math.mit.edu>
Date: Fri, 2 Dec 94 17:55:33 EST
To: rusin@math.niu.edu
Subject: Re: curvature

I think your answer is a good one.  But I've realized that my question makes
no sense unless one first addresses the question of how one *defines* curvature
to a random layperson.  Eigenvalues clearly aren't the way.  Some people 
who replied to my posting used the Gauss map to define curvature, and I 
think that this must be the right approach.  Combining this with the notion 
of signed area and the index of a map, it's easy to get the result I asked
for.

Jim Propp