hexadecapole moment (bug?)

[Omololu]

Dear MOLPRO users,
I am trying to calculate a hexadecapole (z^4) moment
of carbon dioxide (CO2) using finite field but i
encounter the following problems:
1) the finite-field Hartree-Fock (HF) result is
radically different than the result obtained from
calculating the expectation value of z^4, i.e. <z^4>
2) i get two very different results depending on what
machine i use. Using a 32 bit (Linux) machine o get   
-924.8382626 while i get 371.577683 from a 64 bit
(Sun) machine. 
The RHF expectation value i get is <zzzz>=
-277.37575790 on both machines, so at least this
calculation of the expectation value works, i think.
My input is below. (i hope this is not a bug. i am
using MOLPRO 2002.6 with the readop_multipole_np patch
applied). Thanks in advance for any help.

o.

Input:

***,CO2 hexadecapole
memory,20,m

GTHRESH, oneint=1.d-14, energy=1.d-9, step=1.d-5,
twoint=1.d-14, prefac=1.d-14
GTHRESH, coeff=1.0d-5, zero=1.0d-14

r1=1.162453 ang
r2=1.162453 ang
theta=180
geometry={nosym;C;
          O1,C,r1;
          O2,C,r2,O1,theta}
basis=avdz

expec,mltp4
hf;
f=[0.000001,-0.000001]
method=[hf,mp2]
k=0
do i=1,#f
field,zzzz,f(i)
  do m=1,#method
  k=k+1
  $method(m)
  core;
  e(k)=energy
  end do
end do

k=0
n=#method
do m=1,#method
  k=k+1
  hexz(m)=(e(k)-e(k+n))/(f(1)-f(2))
end do

table,method,hexz
title,hexadecapole (zzzz) result for CO2
---

=====
***************** SEEK GOD! *******************

__________________________________
Do you Yahoo!?
Exclusive Video Premiere - Britney Spears
http://launch.yahoo.com/promos/britneyspears/