The amount of Sr that was already in the rock when it formed, for example, should be proportional to the amount of Sr that is currently there. Any isochron whatever with a positive age and a constant concentration of N can be constructed by such a mixing.
In the century since then the techniques have been greatly improved and expanded. It is not even necessary for the fractionation to treat parent and daughter equally, as long as it has the same preference for one over the other in all minerals examined; this will also preserve the isochron. Yet you state it as fact. Kennedy in Geoscience Reports, Spring , No. Suppose that A has a for the sake of argument, uniform concentration of P1 of parent, D1 of daughter, and N1 of non-radiogenic isotope of the daughter. Obviously, directly sampling mantle magma will tell us a lot, but I am not sure if it will tell us anything about radioactive dating.