Tronrud's parameter(b-factor) on xplor(3.851 on line)

[SATOSHI FUJII]

Hello,
I am refining  structures with the modified individual B-factor refinement
at 1.8A resolution on xplor3.851. 

lines:
REMARKS file bref_tnt.restr
REMARKS contains Dale Tronrud's parameters for atomic temperature factor
REMARKS refinement (published in J. Appl. Cryst. 29, 100-104 (1996) )
REMARKS for main chain atoms, 898 examples

 bsigma = ( name N  or name CA ) = 4.6
 bsigma = ( name CA or name C  ) = 7.4
 bsigma = ( name C  or name N  ) = 6.6
 bsigma = ( name C  or name O  or name OT* ) = 6.3

REMARKS alanine side chains, 80 examples

 bsigma = ( resnam ALA and ( name CA or name CB ) ) = 3.5

REMARKS arginine side chains, 36 examples

 bsigma = ( resnam ARG and ( name CA or name CB ) ) = 8.5
 bsigma = ( resnam ARG and ( name CB or name CG ) ) = 16.5
 ..
REMARKS no weights on angles
 asigma = ( not name H* ) = 1000000

I have checked the log file and  there are 98 bond_B_restraints groups and
the number and rmsd in statistics for each group are incorrect. It shows
that the above  script doesn't work.  I guess that the group number for
atom is considered only by the final evaluation and a few bond restraints
for the pair atoms with same group number are selected under the assigned 
sigma_bonds.  There is also a same question that in the case of the
original brefinement.inp script,  what is the  sigma_bonds for CA-CB bonds
which must be assigned  with the different group number. (it must be
default)
In xplor 3.1,  It looks  go through but the sigma_bonds comes from the last
assigned selection and this script also did not work .
Is there some tricks that I should try?

I would appreciate any suggestions. With many thanks in advance

Satoshi Fujii