SanDiego_Art - 11:01 pm on Mar 3, 2010 (gmt 0)
I don't think you'll necessarily be testing the performance between adgroups, b/c google will be choosing (based on their success metrics - not yours) which is "better" (makes them more money...) You will not get a true split of traffic between adgroups.
As already suggested, you should have the ads with the different styles in the SAME ad group. Testing mulitple ads that are very similar will not give you any real actionable results. You want to test the different sytles directly against one another. To do this, put them in the same ad group.