Update on this subject: I have learnt some interesting stuff the last two days with the help of a good friend who knows a lot about lighting and with the help of Excel. Those of you not interested in the how and why of this may stop reading after this: you can´t use series-coupled light bulbs to make a dimmer for a 1000W+ unit, unless you are both of these things:
1. A total maniac 2. Rich.
For the rest of you, and for the TMC record if anyone else would have this brilliant idea again, here goes (white lab coat on, please):
If you want to use a series coupled component to make a low-tech dimmer for a heater of 1000W+, it will be bulky. VERY bulky. It might actually be easiest for you to use your electric stove, your fridge and your freezer as the dimmer, because the sort of light bulb you would need is a little hard to find. As far as I know, the only bulb you could get that would cut it is the 2000-5000W+ UV light arc bulbs they used in the light boxes for making printing plates for offset printers. Available second hand through your local offset printer supplier
. Buy the entire light box - while you are at it, buy another one since you will need two if you want to be able to get 80% of max power for the heater. They are cheap and can probably also be used for killing horses.
The conclusion of the earlier messages in this thread is correct, with one little thing that we all forgot about.
To get 1/10th of the effect on a 1000W element we said: get a 100W bulb, right? Lets say we take a 1000W light bulb? That would give a 50-50 power division, right? Yes. Right, if you only look at the percentage of available power that each component gets. There is just one little thing. How much *actual* effect has each component? With these values, the heater has *250W* and the bulb also *250W*
Since this is series coupling
1. Total effect depends on total resistance, which in this case is doubled.
2. Double the resistance=>half the total effect, 500W. Each component now gets 50% of that. Voilà : 500W/2 = 250W.
I made an Excel sheet with the necessary formulas and attached it to this message. Play with it if you like. This was fun! With other relations between bulb and heater effects things get even stranger (I tried putting in my current temporary values, 500W bulb/1350W heater, and got 266W for the lamp, 98W for the heater!), but remember that the *total resistance* is what makes the difference here. My electrically wise friend was also very surprised by these results, by the way. But we checked and doublechecked them. I also tried it with my stuff, and it works in real life too. Light bulbs in series works well for "dimming" domestic audio amplifiers. But then, they are usually at or below 100W. A roaster should be at least ten times that.
I learnt a lot from this... the exponential nature of effect and other such really interesting stuff. Ehhrrmmm, by the way, anyone needs a 500W halogen lamp?
As you can see in simonp:s excellent article a common light dimmer must be modified to withstand around 1200W. I would need it to handle about 2000W... This is the playground of pro league dimmers, the stuff they use for concerts, theaters and so on. For now, that is way out of my league since I am only experimenting and don´t know if I will need all those watts, but I must have headroom and I don´t want to blow too many mains fuses. Tomorrow it´s off to the nearest warehouse and we´ll see what I come up with. I promise to follow safety regulations and be legal 
