A proposition is given which provides an easily justified reason as to why attention should be confined to estimable parametric vectors when formulating linear hypotheses.The possibility of justifying one's linear estimation effort on the estimable parametric functions via an identifiability condition is also mentioned.