Comparing these two options with the Australian ledgers shows WM20 50280 being in Australia with the AMF serial 86318, WM20 58201 isn't listed but fits in a gap in a sequence of machines that are.
C4557966 was in Polish service in the UK. It has the larger fuel tank. If the Australian group photos are a guide then machines by serial C4561693 were being fitted with the later-pattern tank.
I'm still trying to work out how they could have arrived at a system of using the last four digits and being certain that they were never going to duplicate across vehicle types (OK, there is the theoretical 'C' prefix which isn't present !)or subsequent contracts. They only had a range of 5000 numbers...there couldn't have been a system behind this, could there?
Could it be that the whole is from the Polish sequence and that at some point a decision was made to overpaint the first three letters in a different colour ?
I could almost be tempted to see a darker colour behind the last four digits so maybe the entire number is a Polish replacement.
I doubt that an ex Polish machine would turn up in Australia in 1942, and if it did the original number had already been lost so they couldn't have noted it in the Australian ledger.
Its also possible that the machine has been retro fitted with a side damper knob, so maybe "if" the last four digits are original then it could be earlier, 4199789 or 4349789 perhaps?
In 1940/41 the Dutch free forces also started with C128xxxx numbers.
Later pictures show that these numbers were replaced by the more "usual" ranges.
Neither the Nortons nor the BSA's had Census numbers with any relation to the mating frame/engine numbers.
Totally random.