Weighted wedgies

I'm tearing my hair out trying to get weighted wedgies generated from mappings ("vals" if you like) to agree with those generated by unison vectors ("monzos" if you like). In Prime Based Errors and Complexities I said that you should multiply by the log of the relevant prime instead of dividing to weight the monzos. Well, that seems to work. Then I implied that you should divide the weighted monzo wedgie by the product of the primes to get the complement of the weighted val wedgie. That unfortunately doesn't work :(

Where there's a single unison vector the old rule does work, so that must be how I tested it. For a 7-limit planar temperament, the monzo wedgie has to be divided by 5. Why 5? I've no idea...

Any ideas?

Graham