G98 ECP bug

Dear CCL Masters ,
 Recently I've purchased the new G98W program and heavyly
 surprised with this buggy box ! I work on transition metal
 complexes reactivity , so I need to use ECPs. And I have a lot of
 legacy calculation results in Stevens-Bash-Krauss-Jasien-Cundary
 basis set and ECP. They are called "CEP" in G98 and "SBK" in
 GAMESS-US. Trying to make something like this
 ... memory and checkpoint specification  ....
 # opt rhf/CEP-31G pseudo=cep
 ... title, charge ,mult ...
 Pd  0. 0. 0.
 L   0. 0. -2.35
 L   0. 0. 2.35
 ... other atoms ...
 while L is the element of 4th of higher period of Periodic Table (
 Sb,As, Br, I, etc ) we'll get a error : program attached the
 palladium ECP to all of L atoms ! (I see it using GFPRINT keyword
 !) If we make this ( reorder atoms ):
 .....
 L   0. 0. -2.35
 L   0. 0. 2.35
 Pd  0. 0. 0.
 .....
 then g98 attaches L's ECP to palladium atom . As result in both cases
 we'll get wrong ECP set and wrong number of electrons in molecule !!!
 If we try to take lighter elements ECPs are selected correctly.
 Does anyone know workaround for this problem ? And I receive with
 pleasure your any help.
 Sincerely Yours,
 Grigori Shamov
 Grigori A. Shamov
 Post-graduate student
 Department of organic chemistry
 Kasan State University
 Kremlin str.18
 Kazan 420008
 Russia
 e-mail Grigori.Shamov &$at$& ksu.ru