This is a little experiment I decided to try last night out of boredom, and much to my surprise it worked. I had one red wire go to the lamp as well as the white wire and I disconnected the blue wire and the other red wire and it actually worked. See it here. When I took the starter out, nothing happened, much like on a preheat ballast with no starter. Manual preheating also worked, although I left that part out of the video because it's not as fun to watch as a nice glow starter. I don't know if anyone has done this before but I figured it would be cool to share. I don't know whether or not it's a good idea to install this setup in a fixture, but it's still fun to try out and pretty cool as well. With that said, enjoy!

It depend on the ballast, how it respond to the short circuity imposed by the starter, mainly at lamp EOL. For some designs this is no problem (even less heat stress then open-circuity; mainly HPF using capacitor as ballasting impedance), other might overheat (LPF).

You know, I actually want to try this on a HPF ballast next. I don't have any 1 lamp HPF ballasts right now but its actually interesting that it might work better. What is also strange is that the ballast didn't get very hot during this experiment. I believe the ballast has thermal protection so if it got hot at EOL, it would shut down until it cools off.

When the lamp run, there is no difference towards normal configuration and the stressful starting has short duration to heat up anything.
You might use twin lamp ballasts in the same way, connect only one end-connection to the tube and the starter connect to other end's (like you did it for 1lamp ballast). For ballast with serial lamp connection you do not have to wire the middle point to the ballast at all...

I just tried it using a 2 lamp HPF magnetic rapid start ballast meant for F32T8 lamps and I used an F32T8 that is a bit mercury starved. This lamp worked on a F30/40 preheat ballast for a little bit, then just started blinking continuously on that ballast, and it worked on the F32T8 rapid start ballast with no problem at all. It didn't try to restrike or anything. I used one red wire and one blue wire and then I wired a starter in in between on the lamp. The starter I used was an FS4.

The RS ballast more easily start the lamp then preheat one (assume preheat configuration in both cases), because it has higher OCV. But be careful about HPF ballasts: The leading current phase of ballasting capacitor is compensated by the magnetizing current on the secondary, what yield only small (resistive phase) current in the primary. When the lamp does not strike, the primary has to handle all core magnetisation, so it ask for high current (the input VA would be ballast OCV times lamp current), the input VA's would be about 3..5x nominal lamp wattage, so the current would be higher then normal in the same ratio, what would overheat it.

The magnetic RS F32T8 ballast has the high OCV required by the F32T8 lamp. That is why it can keep the lamp going while the other one can't. On a side note, I also discovered that on the 1 lamp ballast, I can use one red wire on one end of the lamp and the blue wire on the other end rather than white and line, and get the same results, which eliminates the need of getting another wire from both neutral and the lamp end that would normally go to the white wire. Also, be sure you know what you are doing. I cooked one of my 1 lamp F30/40 ballasts once by accidentally plugging the blue wire into power instead of the black one. You definitely don't want to do that.

I tried this on a cheap ballast that could not start lamps on its own and now it works fine as a preheat ballast (still need to do more testing).