g90 CASSCF calc snafu

From: mxm at.at biosym.com (Max Muir )
Subject: g90 CASSCF calc snafu
Date: Sun, 7 Aug 94 16:38:44 -0700
 Hi Folks,
 I am trying to optimise the ts from reaction of a hydrogen atom
 with water.  I have a good starting geometry but I have not been
 able to set up the GAUSSIAN 90 input file correctly.
 %chk=g90.chk
 # CAS(3,6)/6-31G*
          ts from rxn of H radical with H2O
 0 2
 H
 O,1,L1
 H,2,L2,1,A1
 H,3,L3,2,A2,1,D1
 L1=0.978
 L2=1.332
 L3=0.834
 A1=99.405
 A2=166.700
 D1=0.00000
 GAUSSIAN  gets through to the MCSCF program and then . .
  ENTER MCSCF PROGRAM
  NO. OF ORBITALS = 21     NO. OF CORE-ORBITALS =  4
  NO. OF VALENCE-ORBITALS =  6      NO. OF VIRTUAL-ORBITALS = 11
  USED ACCURACY IN CHECKING CONVEGERGENCE =      .00000001
        15246 Words of memory needed for 4-index transformation.
       957327 Words available.
  NO. LOADS IN INTEGRAL SORT           3
  2ND ORD PT ENERGY     CV    -.211077   CU    -.073915   UV    -.213784
  TOTAL                     -76.584803
  **********
  CALCULATION TERMINATED
  ATTEMPT TO DO ORBITAL ROTATION GREATER THAN 45 DEGREES
  YOU HAVE CHOSEN THE WRONG STARTING ORBITALS
 **********
  Error termination in Lnk1e.
 bsh: 21078 Memory fault
 Can somebody help a computational organic chemist out?
 Regards,
 Max