@xmaslightguy Yes you're right, they only work with switchstart circuits and electronic ballasts though, they wont work with rapid start even on 240v mains.

I've considered ordering an American ballast, I have a variac that I could use if I bought just a standard 110V one. I also want a slimline ballast as such things are almost non existent here.

@Eliot_240 From my understanding, UK T8's are F36T8 (rather than the US F32T8)?. There was a time when i used to see the F36T8's sold here as well, but not any more. Advertisements claimed they could be used in existing F40T12 fixtures. I shoulda bought some back when they were available LOL.

If shipping wasn't so expensive i'd say order a multi-volt (120v-277v 50hz/60hz) F32T8 ballast from a US seller (the ballasts can be had for cheap on eBay, but overseas shipping would make it expensive). It
would work as far as lighting the lamps.

@xmaslightguy

The T8 tubes here in the UK are different to yours and are designed as a retrofit for T12s, every electronic T8 ballast I've come across uses EOL protection. EU and British standards most probably require all ballasts to have a protection feature. Instant start ballasts with one wire going to each cathode are something else that doesn't seem to exist here.

Most of the ballast here in the UK are low power factor copper-iron chokes and it's up to the manufacturer of the light fitting whether to install an additional power factor correction capacitor. Power factor correction is never bothered with for domestic installations as this just bumps the cost of the light fitting up, you may struggle to find a light fitting of high power factor aimed at domestic use.

You can simply wire a capacitor across the live and neutral to convert any low power factor magnetic ballast to a high power factor one, it seems to be about 1uf for every 10 watts of lamp load and polyester film capacitors are most common. This is only a rough guide though and I recommend you check first which capacitor you need.

@TiCoune66 some slimline ballasts can run shorter lamps...if it lists F48T12, then it'll run F40T12's as
well. If not you can run 2 F40T12's in series to equal 1 F96T12
Slimline ballasts will sometimes overheat from running a bad lamp (but at least with mine it takes a few
hours before they get hot enough for the Thermal Protection to shut down the ballast)

@Eliot_240 Check eBay .. there are tuns of listings for F32T8 IS ballasts that will work (at least on
the US-eBay site)

@nodgen:
Here (Czech Republic) electric meters are required to have tolerance within (+0/-2%) of apparent on the measured real power and i think 0.1% of wattage range at low loads. Mechanical meters were way better, but only when the cause of the low power factor was only the phase shift.
Actually electronic meters had quite long time difficulty to reach this accuracy generally (mainly the one at low power level), but unlike on mechanical types, the harmonic distortion does not make the error worse. These days the technology improved, so when new meters installed, they are electronic. But still majority of replacements are restored and recalibrated mechanical types (the law here require valid calibration certificate for any measurement instrument used for billing, what expires after few years, but even those 30 and more years old meters are able to pass these criteria after normal maintenance, so are still repeatedly used)
But the reason is not he accuracy or so, but available features (remote read-out, multi-tariff,...) and mainly it's own electric consumption: Mechanical meters consume about 5W (those newer ones, older types even 15..20W), while the electronic ones consume below 1W.

I think last CFL ballast I used must have had the filament fusing protection feature but it was a 23W CFl ballast that I hooked up to an F40T12 with the two wires that go to each filament joined together so protection was bypassed.

Most CFL's i've seen used at least the "filament fusing" protection concept, but the ballast is not designed to reliably survive this, so very frequently the ballast die just before the filament break. The protection is here meant only to avoid excessive heat buildup, but not to keep the ballast working.
An exception are "unlimitted ON-OFF cycles" types (e.g. Osram Facility), what usually use quite sophisticated ballast (per CFL "standard"; it usually contain the same controller IC as is used in "mature" high quality fluorescent ballasts, so include preheat timer, that always correctly preheat cathodes even after very short OFF-ON cycle and reliable EOL protection).
But even these suffer from the fact, then ballast components are designed for similar life expectancy as the lamp, what mean there is not much lifetime remaining on the ballast (mainly the electrolytic tank and lamp resonance capacitor and even coil's bobbin material get quite crispy due to the long heat exposure)

Might it have anything to do with how power usage is metered? Only in the last few years here in the US has electric meters been able to accurately measure power usage from low power factor devices. I could be wrong about that, but the standard mechanical meters that we have always used will under-register when power factor is below 1. Again, I could be wrong, but there is now a big push to replace mechanical meters with electronic meters for a number of reasons, power factor being one of them. Therefore, power factor might not be as important here as it once was.

-Nelson

I have a slimline ballast that is "half-dead", but seems to run lamps in parallel. Also I think the good side for vacuum side is the working one. I could try but is it going to overheat the ballast, even with small lamps?