bulgie 

Sorry about the slow response, I missed your question.

I haven't done this exact procedure, closest I have done is softening the stay tip to allow shaping for the freewheel/chain clearance on the inside of the right dropout. That's a safer place to be messing with the heat treatment.

The tire clearance point on the chainstays is harder to get evenly hot, but a oxy/Mapp torch would be a good choice. If you only have an air torch (no oxygen), I wouldn't expect that to be hot enough, but I haven't tried it.

You want a very smooth general heat without hot/cold spots, so a large soft flame is best, and keep it moving. Heating one side of the tube will cause it to curve ("witch wanding"), so go for even heat all around. I expect better results with a dummy axle installed but maybe it's better to let the parts move however they want? Opinions please.

I'm not an expert on this but I think it only needs to come to red heat more or less for an instant, for thin wall tubing. You can assume that the temperature you see on the outside is the same all the way through, unlike with a thick part. Maybe there is some slight delay while the metal ions rearrange themselves into the high temperature stable crystal lattice, essentially erasing the prior heat treatment. Someone here will probably know. But my understanding is that once you get it red, you're done. Cooling needs to be slow but not too slow. Normalized is a stronger state than annealed, which is what you get when you make it cool very slowly in an oven. For thin wall tube, cooling naturally in still, room-temperature air is solidly in the range of Normalizing -- neither annealed nor quenched.

There is some self-quench when you only heat part of a tube, where heat is sucked away from the hot part by conduction to the cold part of the tube. This is unavoidable to some extent, but probably you can mitigate it by heating a larger area to a temperature below the tempering range. My gut feeling is that it's not a huge concern. I'd be more worried about the over-tempered part of the HAZ, just past the part that got red hot. That'll be the weakest point, so try to keep it away from lug edges or the chainstay bridge. In fact if there's a bridge, I might be inclined to remove it. It's more likely to cause a fatigue crack than it is to prevent one by any "reinforcing" effect. The larger the HAZ, the larger and softer the over-tempered area will be, so that would indicate working fast and not heating more of the tube than necessary.

On a short-chainstay racing bike (very likely, on a 753 frame), you won't have much room between the tire clearance point and the BB shell lug points, so you are in danger of having the weakest part of the HAZ right there at the stress riser of the lug point. Thinning the points to feather into the tube reduces the thickness discontinuity, probably worth it for improving the fatigue endurance.

Mark B