G:G94 RIC and GAMESS: constrain var to equivalence?

[Dawn Friedman <dsf at.at world.std.com>]

   This is only partly a G94 question -- my real concern is with GAMESS.
   I'm geometry-optimizing some model systems whose high symmetry is broken
 only by methyl rotors.  I want the methyl hydrogens to be free to get out of
 each other's way, but I want the C-C frame bonds to stay equivalent, at
 least until I have a partial optimization.
   This is easy to do in a Gaussian z-matrix format.  G94 can be forced to
 use the input zmatrix rather than redundant internal coordinates, though of
 course all the advantages of the RICs are lost.  (Is there a mathematical
 reason why G94 can't construct an RIC set with some variables constrained to
 be equal?)
   I haven't figured out *any* way to constrain a set of bond lengths or
 angles to be equal *but variable* in GAMESS when the entire molecule doesn't
 have symmetry.  My choices seem to be:
   (a)  Force the entire molecule, including all the methyl hydrogens, to
 exhibit at least one symmetry element.
   (b)  Add fixed constraints in $STATPT, which will hold the C-C bond
 lengths to a constant value, and hope I picked the right one :-)
   I'm working with model systems that aren't necessarily stable as single
 molecules in a full optimization, so I can't count on their optimizing to an
 energy minimum if I don't constrain them.  Should I just choose option (a)
 and let the CH3 hydrogens suffer?  Is the hybrid I'm trying to create --
 symmetric frame and free methyl hydrogens -- inherently invalid?  I'll
 summarize to the list.
   Thanks,
   Dawn Friedman
   Dawn Friedman  dsf at.at world.std.com