Furthermore he says "Thus the conservation law, though not an a priori mathematical truth, provides [...]".

Imho "not [...] a priori" is not the same as a simple "not".