Systat 13 / How to redefine error terms?

Hi there guys,

I’m trying to do a repeated measure ANOVA with 2 grouping factors (TRAIT$ and BLOC$, with BLOC$ nested in TRAIT$) and 1 within factor (TIME).
As I have a model with fixed and random effects, I would like to redefine some error terms that are miscalculated by Systat in the GLM procedure.
I’ve tried the following command:
(‘for the effect TRAIT$, I want Systat to use the error term of BLOC$’)
But it’s not working… Do you have any idea of how I can do?

Huge thanks!