OK, now I am thoroughly confused. After reading up on Mylar on Wikipedia (yes, I know it isn't the most reliable place for info) :
PET film (biaxially oriented - Wikipedia, the free encyclopedia)

It seems that Mylar as we know it is basically a metallized version of this PET polyester film. The mylar that we use is actually aluminum metal in PET!! Now this makes it extremely hard for me to figure out how Mylar isn't giving better/equivalent results to aluminum foil. Given the processes used to metallize Mylar, I'd assume you have very low surface roughness (which explains why it looks so much like a mirror). Also considering

The wiki on Aluminum states:

"Aluminium is one of the few metals that retain full silvery reflectance in finely powdered form, making it an important component of silver paints. Aluminium mirror finish has the highest reflectance of any metal in the 200–400 nm (UV) and the 3000–10000 nm (far IR) regions, while in the 400–700 nm visible range it is slightly outdone by tin and silver and in the 700–3000 (near IR) by silver, gold, and copper."

Hoppy, this is my theory. Let me know what you think of it. It could be, that in the case of your Spiral CFLs, your reflector geometry isn't optimal for something as highly reflective (next to zero diffuse reflections) as mylar leading to significantly higher light losses due to restrike. On the other hand, Aluminum foil, though not as reflective as Mylar, causes more diffuse reflections which basically lower this restrike as some of the light that would be reflected straight back into the bulb is now being scattered at different angles, bouncing around and in general providing better illumination. This would also explain why something like the white paint on paper behaved quite well since it is entirely diffuse in its reflection of light...again lowering restrike effects.

Something I had thought about a while back but never fleshed out in my head is the following idea:

For non optimal reflector geometries, there is a good chance that a completely diffuse reflector such as white paint or a partially diffuse reflector such as aluminum foil will provide better illumination than a perfect reflector like Mylar since the non optimal reflector design coupled with the perfectly reflecting Mylar surface will cause a significantly larger amount of restrike and thus, losses. On the other hand, diffuse reflectors in such a situation would cause some of this light that is being forced to "Restrike" due to the non-optimal reflector design in the case of Mylar to be scattered in other directions which wouldn't experience "re-strike". This effect would be even more pronounced in the case of Spiral CFLs since they have a geometry which lends itself to a lot of restrike. In addition it is next to impossible to design an "Ideal reflector" for Spiral CFLs. In this case, diffuse reflectors are probably the best materials to use.

I don't know if this is clear in words. I could draw a diagram to explain my idea.

I would postulate that for a T5 tube (restrike being less of an issue) and an optimally shaped reflector (see pic below), Mylar should still be more effective than aluminum foil or white paint (As Tony Gomez in the site linked below seems to believe).



This picture is from Aquarium Plants - Info Pages , interestingly the author claims that white paint is much better than aluminum, and hints that mylar would be the best.

I'd love to hear what people think about this idea.