Just discovered this entry by accident, while I was looking for something else...
Anyway, regarding CK's query on why there are several plots for the 7'3" + 8'3" wheelbase, they are there primarily to show there are many possible solutions, particularly when playing around with the amount of reduction of the springrate on the middle axle. To be honest, apart from the intuitive feel everyone shares about the middle axle needing to be a little weaker than the outer two, no one really knows what this degree should be. In context though, the variations in middle axle strength shown in the plots are exceptionally small compared to previous eras of springing, where 6-coupleds were often not more than 0-4-0s with a middle axle lightly sprung or even floating around in the air somewhere.
The 7'3" + 8'3" plots are also there because it became my sort of default yardstick experiment in the early days - the first plot was done before the modern spreadsheets came into existence, and was a bit of an exercise in convincing myself the mathematics would scale properly. (In those days, much beam software tended to blow a fuse if you plugged in a beam diameter less than 0.1m.) The first plot was also done a long time before the HL pannier chassis was a glimmer in Chris Gibbon's eye. 10 years ago, I think the only commonly available chassis around for the wheelbase was the old Comet one, and I had no idea where any future manufacturers would put their frame spacers, so it was a case of trusting to luck to a large extent. Later on Chris introduced his Collett Goods chassis, and of course his CSB jig, and these days I try to spot published plots to a 0.5mm increment, but it's not always possible, not that anyone should get too hung up about the first place of decimals - experience seems to indicate that to the nearest 0.5mm is good enough given the other errors in the system (arising chiefly from friction), but generally it is a good objective to reasonably maximise a beam length - "design safe", in modern vacuous soundbite terms.