It definitely was the bad rotor arm! What was actually happening was that attempts with the main HT lead connected properly were resulting in sufficient voltage, for sufficient time, at the rotor arm for it to break down, before the plugs could fire. Arcing to the centre contact changed the voltage and the time, less voltage and a very much faster rise time, such that the rotor arm did not break down before the plugs fired. The relationship between breakdown voltage and time for a spark gap is a complex matter, even more so as one of the three gaps was in an air/petrol mixture at a much higher pressure than the arcing gap and the rotor arm to contact gap.
In the 1950s and 60s there were all sorts of daft devices for allegedly obtaining more power or better fuel consumption. One was basically a spark gap in a perspex casing that fitted inline with the main HT lead. It "may" actually have made a difference, as it would sharpen the pulse at the plugs, making a contaminated plug more likely to spark properly. Not legal now, as it would cause a massive breach of EMC regulations.
In this case, fortuitously, maybe even freakishly, circumstances produced a similar effect.
The ballast etc did of course need sorting out but was not the main cause of the problem.
It is only after a well-described event that such a diagnosis is possible. I would not have guessed correctly if I had been there, until the rotor arm was changed.
The modern way is to improve the timing accuracy by causing a very fast breakdown of the plug gap, despite having no other air gaps in the system, by using highly optimised coil packs and computerised solid state control. Fit and forget... Seemingly, plugs are good for about 50k miles. There is advantage in fitting such a system to older vehicles, albeit you would have the expense of getting it mapped properly on a rolling road.