AMBER 4.0 Energies. What are they?

 AMBER 4.0 purports to output energies from MD runs in Kcals (pg 79
 of the manual).  This energy does not seem to be kcals/mol, as a box
 of 200 water molecules gives twice the energy as a box of 100 molecules.
 My best guess is that since the AMBER code accepts a 'built-in' solvent
 such as TIP3p water as a single 'molecule' built up of many residues, the
 energies are reported in (kcal)/(mol of boxes of whatever).  Is this true?
 Also, has anyone had any difficulty using TIP4p water in AMBER?  The
 pure solvent that I have run drops dramatically (~400 kcal) in energy
 after about 30-40ps, and from there on fluctuates only about 1 kcal up or
 down in total energy.  It seems to have formed some type of rigid
 'polywater'structure.  It has been suggested to me that this is caused by
 the charge center being given a mass of ~12amu, and that it might be corrected
 by setting this mass to zero, but AMBER will not accept a zero mass.  I
 would be very interested to hear about other experiences with TIP4p in AMBER.
 Please feel free to e-mail me at Kgrafton' at \`  I will
 post a short summary of responses if the infomation gained warrants it.
 K. Grafton
 Department of Chemistry
 University of Oklahoma at Norman, USA.