Hello, for some reason TukeyHSD() seems not to want to play with aov-results with an Error() function:> Res1<-aov(H2O~location+topo+horizon+pF+Error(location:topo:horizon)) > TukeyHSD(Res1, "topo")Error in TukeyHSD(Res1, "topo") : no applicable method for "TukeyHSD" However you could do it by hand by doing all pairwise comparisons and adapt the confidence interval with Tukey or Scheffe. Or do I make a statistical thinking error? Thanks, Wouter