-----Original Message-----
From: bsouthey@iastate.edu <bsouthey@iastate.edu>
To: Janet Roden <jjr997@aber.ac.uk>
Date: 05 January 2000 15:17
Subject: Re: Fortran error
>Hi,
>Can you send the .pin file to the ASREML list? Or at least the line that
>creates the error and the associated solutions?
Please see attachment. As you will see, the .PIN file is a very simple one.
I don't believe this is the problem in itself.
>The problem appears to be numerical overflow and suggests your .pin is not
>correct.
>
>I tend to start simple and buildup .pin especially contrasts and ls means
with
>interactions - ASREML does not check for estimability! You MUST tell
ASREML
>the correct functions to use. So use SAS (glm) or Genstat (?) under a
fixed
>effects model (drop the random terms from the model) to find the
appropriate
>functions to use.
I have analysed the data using a fixed model in Genstat prior to analysis
with ASREML in order to identify the correct fixed effect model to use.
>Regards,
>Bruce
>
>