Pilling-Bedworth rule

I would like to comment on the application of the Pilling-Bedworth rule. In most cases of oxidation, it is the metal ion that migrates in the oxide film. If this is true, then the metal ion will go through the oxide layer and form new oxide on the outer surface. Since this new oxide is not constrained, there seems to me to be no reason why the difference in the volume of the metal oxide to the volume of the metal should be used to predict the continuity of the oxide film. For example, for sodium, one would predict, by this rule, a porous oxide when actually a dense, protective oxide is found. For tungsten, one would predict a compact oxide while the opposite is found, ft seems to me that the continuity of the oxide layer is determined more by the possible transformations or reactions of the film which forms initially. 

---