This is going to be complete conjecture on my part - especially as this may be a new, fandangled ballast with DMX control or the like - but if it's old, then the following may just be the trick (you did not say whether the dimming was achieved with a 0-10VDC, or the 10V was modulated in some other way).
The principle applied is called a "magenetic amplifier". It was used extensively in 3-phase voltage regulation systems (for professional VCRs etc.) where the natural tendency of 3-phase is to cancel out.
But the same can be very successfully done with single phase. You need two identical transformers at least about twice the rating of the load. The primaries are wired in series, and this then in series with the load (see pic below - hope the link works!).
The secondaries, however, are wired 'back-to-back' so as to CANCEL each other out. If a DC current is injected (with the aid of a voltage, of course - note, you're only working against the DC resistance of the two secondaries in series), the primaries become a lower impedance. The more the current, the lower the impedance.
You can try this at home if you wish, it is perfectly safe (well, should be as all of you play with at least mains on a daily basis! so simply carry on respecting the mains as you always do and it won't bite!). I like to go down in voltage on the secondary (say about 50V), but not too far (else the controlling current goes up).
Now, putting this in to principle with the dimmable tubes.
It may just be possible the fittings are adjusting the arc current in the fluoro tube with the same principle? There is likely to be a "max" you can reach where the core becomes saturated thus controlling the maximum arc current (else you destroy the tube), but it might just be possible to get a reasonable brightness range out of the tube. Hmm, you've got me thinking! I must dig through my scrap and see what I can find and see how much I can change the brilliance.
[This message has been edited by marcspages (edited 11-11-2004).]