Dear Arthur, Just one question: How valid are the REML estimation of variance components when the data don't follow a normal distribution ? Hermann -- Asreml mailinglist archive: http://www.chiswick.anprod.csiro.au/lists/asreml