So I've made a bit of a discovery. I've said before that the Philips CPO-TW lamps for 45 and 60W look identical, same size arc tube. Well I've gotten hold of an energy monitor and these are the results for the identical looking 45 and 60W lamps:
CPO 60W on 60W ballast = 66W (60W + 6 ballast loss)
CPO 45W on 45W ballast = 66W (60W + 6 ballast loss) Exactly the same!
CPO 45W on 45W ballast with permanent dim to 55% = 51W (45W + 6 ballast loss)
I used three of each ballast and took an average but they were all within 1 or 2 watts anyway. Two of the three runs had a brand new lamp, the third had a used lamp (~3 years nightly service), the used lamp didn't affect the power.
Also visually, the 60 and 45 at full power seemed to be pretty much identical in terms of brightness. The 45W on the permanent dim ballast was noticeably dimmer.
So it seems like the 45 and 60W CPO lamps are literally identical, and the 45 is just a 60 with a different etch. To get 45W out of a 45 you need to run it with a permanent dim ballast.
Interesting; we only have a 60W ballast available, although we do have different wattage lampses, if we can find them that is..!
This mystery has now been solved. The 45W ballasts drive the lamps at 60W for the first 20 minutes or so before settling down to the correct power draw. Early CPO lamps had no difference between them, 45 and 60 lamps could run on either ballast no problem, but apparently more modern ones are different. You can run a 60W lamp on 45W gear but not the other way around.