I wonder, on most today's fluorescents run on DC, the mercury migrates on the positive side, but how about these old AC/DC fluorescent? (I have one, a Westinghouse F15T8/D) How do they run on direct current with no problem?

If they run on low enough current density, the diffusion process of mercury neutral atoms is fast enough to redistribute the mercury back along whole tube.
Old tubes operate at low density in large diameter tubes mainly to keep them cool, so mercury pressure low (for lamp efficacy), side effect was they had no problems running on DC.
But newer (amalgam, cold-spot tips,...) technology allow to keep the mercury pressure at the same, low level, but at higher lamp operating temperature, allowing narrower tubes for the same current (to reduce selfabsorption), so the diffusion has narrower clearance to spread the mercury back, so is slower and not sufficient for DC operation anymore. But present day when DC supply is needed, electronic HF ballasts are used, so there is no issue of mercury migration anymore, as traditional DC ballasts are only resistive, so quite lossy...

I will borrow this chain littlebit....

I have some electronic ballasts that uses 12vdc as mains. Im not home rihgt now so I cant check the brand of those, but can somebody make any guess is the form(ans other vales) of output electricity how close to of those electronic ballast that uses 230(or 120)V to mains. I was just wondering how the use of that kind of ballast will effect expected lifetime of lamp and whether to use T8 or T12 lamps with that kind of ballast. wattage is 18W. Im planning to convert some FL fixtures to work with 12vdc and install those in my summerhouse. I could check this myself if I had an oscilloscope but I dont have one.

@Roi:
12V ballasts do have three forms:
1. Single transistor flyback inverters
  - Simplest ballast circuits
  - These tend to underpower the lamp, when aranged as "instant start" (without filament heating supply) the lamp life would be short, if operated on weak battery
  - these tend to supply the lamp with DC current, but DC blocking capacitor might be added
2. Two transistor Royer oscillator - Generate symmetrical HF Ac voltage, then uses capacitor as ballasting reactive element to control the arc current (used in monitors to power CCFL backlight)
  - No danger of rectification
  - Output power strongly depend on voltage, so if designed as "instant start" (without filament heating supply) the lamp life would be short, if operated on weak battery
3. Step-up transforming inverter with inductor as ballasting impedance - i have seen these mainly in 24V fixtures for buses - i think the best quality i've seen.
  - These use decent programmed start, so no limitation of switching cycles
  - Usually stabilzed lamp current - so no dependency on battery votage

Ok, so even newer F15s can actually run on DC? That's cool, I should try it out.

Ok, so even newer F15s can actually run on DC? That's cool, I should try it out.

But beware of correct working current - to have correct electrode temperature, otherwise they would sputter off quickly.

Of course. That's the problem with those battery-powered F8T5 striplights, too little current.