The placement will suggest something like 400W or so, so most likely probe start on a CWA.
These ballasts feed the lamps by pretty constant current, so the real power delivered to the lamps when they are still cold is rather low and it increases only as they warm up.
That means they start to warmup veeryyy slooowlyyy, only when "tipping over" certain temperature/pressure state, the warm up speeds up.
Here what differs is the time, when the lamps reach the "tipover" point, so the difference could be really tiny. It could just be as tiny as a bit different mounting position (slightly different angle of the starting probe,...).
Or just random production tolerance variation, batch-to-batch difference, different level of wear, till even different arctube construction of different maker, really all is possible...
Thank you for explaining everything clearly. Just wondering, how did you know it was 400W probe start on a CWA?