Oh dear! OK Arne, here I pop. This could be a long winded answer.
Early western junk rigs were reported to heel less, even when they had more sail area than their Bermudan cousins. This is not surprising as the flat airfoil produces very little lift at the low angle of attack before knife edge separation produced a separation bubble behind the luff . That is, assuming the luff was tight and straight, which it often wasn't. From what I observed I believe that flat sails were routinely oversheeted to get high pressure on the windward side and therefore stalling and not generating the much greater low pressure force on the lee side. The resultant lift/ drag ratio would be very low so downwind performance would be good, but upwind poor, as we all know.
Arne demonstrated that putting camber in the sail would increase the L/D ratio and give greater drive on all point of sail. The SJR was my attempt to analyse and improve the windward performance, and to simplify the ever increasing spaghetti controls apparently required to set the rig and to shift it fore and aft on various points of sail. I was looking for a low stress KISS rig.
When I owned Poppy I did not meet a junk rigged boat that I could not beat, except when Bertrand's small cat was on the reach as by being unballasted was not restricted to waveform speeds so could reach at over 10 knots. I always matched similar cruising rigged Bermudan boats to windward and even when they flew chutes off the wind, but of course could not beat the expensively prepared racing brigade who have had years of development under their belts. I'm not sure what is happening in the performance tests, but probably due to my early training in experimental work I hold little reliance on tests in uncontrolled conditions and which have the results 'factored'. That is why I elected to use the Island Race to compare rigs, and am happy with the performance when sailing alongside comparable Bermudan boats. Overall position in the Island races requires tidal knowledge beyond my experience. (I approached one potential winning helmsman in an effort to get good junk publicity, but he wanted a £1,000 fee for his services!)
Going back to heeling, the SJR does heel less than Bermudan, but only in certain situations. To windward there should be no difference, and that is borne out in practice. As the sheets are eased the L/D of the Bermudan rig starts to fall off, and the ratio of forward drive to leeward heeling force will impair the performance. As the SJR sheeting is fixed as the rig rotates the L/D does not increase so the boat accelerates when the total force moves forward as it bares away and does not heel as much as the Bermudan boat. This is good up to squared off and the wind about 135° from the bow, at which point it becomes clear that the total thrust force it forward of 90° to the rig and to windward, and which can only be explained by the inbuilt jib sheeting angle which proves the efficiency and importance of the cambered jibs. Arne has correctly identified this situation, and is right in saying that I would not set the rig greater than squared off. Even then in gusty conditions it was advisable to sheet in a little to stop the windward total forces. The best performance is obviously with the total thrust is straight ahead.
It is a major regret that Arne did not get to sail Poppy with me as I believe we would have gained from the resultant analysis (and possibly arguments). I have been lucky enough to sail Johanna on two occasions with him, and both were great educational experiences for me.
I agree that the 'lesser heeling' on the reach was probably what brought out the hooligan in me. Lets face it, if the boat doesn't heel too much, why reef? Just let the forward drive make the boat go faster. The rig allows for smaller area for equivalent performance, so why reef, unless you want to slow down.
'nuff said.
Cheers, Slieve.