I realized another possibility is that there are too many records and little
variation. This can lead to very small standard errors of the variance
components. The .vvp file will show that this. One solution that I have used
is to multiply bwt by say 10. Then you just have to remember to untransform
What I would do first run is a simple animal model and see if it works.
Also, the .asr file is the most valuable so please send it in the future.
Asreml mailinglist archive: http://www.chiswick.anprod.csiro.au/lists/asreml