The reader should be thoroughly familiar with the K-Ar method, as explained in the previous article, before reading any further.
J is a factor which depends on the nature of the neutron bombardment.
Now the bad news is that there is no way we can somehow manipulate this data to give us a correct date for the sample.
But the good news is that we do know that there's a problem; whereas if we'd analyzed the same rock using the K-Ar method, then it would have supplied us with a date and there'd have been no sign in the K-Ar data of anything wrong with it.
The ideal scenario according to Bowen's reaction series would see a granite melt begin crystallizing a cumulate assemblage of plagioclase and hornblende (ie; tonalite or diorite), which is low in K (and hence Rb) but high in Sr (as this substitutes for Ca), which proportionally enriches the melt in K and Rb.
This then causes orthoclase and biotite, both K rich minerals into which Rb can substitute, to precipitate.