<a href=showthread.php?s=&postid=9244887#post9244887 target=_blank>Originally posted</a> by BeanAnimal
Joshua it all depends on the bulb placement within the structure, the desired focus, the diameter of the bulb (restrike and shadowing), the size and shape of the reflector mouth... and a host of other things.
A "smooth arc" is a rather poor shape for a reflector. The target shape is a derivative of a parabola. The shape is hard to bend so facets are used at the narrower parts of the shape. The facets also help to accommodate the bulb position and eliminate restrike (light reflected back into the bulb). The problem with doing DIY bending on a break is very simple. Even if you could determine the proper angles and facet widths, you would have trouble bending those angles accurate enough to make it worth the while.
You best bet would be to get your graph paper and math/science book out and draw a parabola to the scale you think will work. You will then need to build a form of some kind to to use as you work the shape to match. You will idealy need to bend a "bat wing" type of shape instead of a smooth curve. The crease in the top of the parabola is to partly counteract the restrike. Best guess is that a DIY "arc" is going to be 50% (or worse) less efficient than a decent engineered design. There are just way to many variables to deal with, that just can not be guessed at.
Remember these are not cylindrical light sources like a fluorescent tube. They are pretty much point sources. So all of those SE batwing reflectors waste a TON of light out of the ends. That is what makes the lumenarcs great. They take the light from all directions and get a good portion of it back down to the tank with as few secondary reflections as possible.
I think one of sanjays tests included the regent hammertone reflector. That may give some of you an idea how it compares to some of the better designs that were out at the time of the test. 
