Your explanation is too good! But while you said about about P values all three( sample, columns and interaction ) is less than 0.05 then as you said we rejected null hypothesis. But when I did with my data, I have got like (0.025128 for sample, 0.582991 for columns and 0.971607 for interaction) so this means that I can only reject the first one which is sample while H02 and H03 columns and interaction will remain same as there is no significant difference ?