Thanks for the replies (and all the good work!), chaps.
It appears that diskspare.device uses 12 blocks per track (rather than 11?) and, optionally, tracks beyond 79. I'm sure I had better documentation for it, back in the day, in which the trick was more fully explained - from an earlier version probably "backed up" on a diskspare floppy!
I'm trying to test things myself, but I'm currently having trouble getting ADFRead to work with regular FFS floppies.
I suspect one of my floppy drives might be at fault - I'll do a swap later.
I can say that disk2FDI failed in all of my attempts with diskspare. Hmmm... maybe I should just acquire a working Amiga!