I am guessing that they do not mean the full flame, as you have experienced, Roger. Rather I believe this is a form of pre-ignition/final cumbustion. As you have noticed, there are several colors related to a premixed gas (a typical canister stove.) A lot depends on which elements and radicals are being broken down and recombined in the heat of the flame. Of interest to this discussion, is three basic regions, usually corresponding roughly to the heat of the flame. A lighter blue/whitish “flame front” around a “center core” that is colorless, since it is too cool to support combustion. The flame front in turn cools off and continues to combust with other less reactive molecules producing a dimmer/darker blue fading to non-visable flame.
The central core is not actively combusting. It is heating up to ignition temperature and is far cooler than the around 2000C degrees of the actual flame front. Pressure/Volume drive this through the little holes, essentially keeping the flame spreader or grate far cooler than the flame front. By turning it down to a very low level, you can actually have the flame front in very close proximity to the spreader, heating it to much higher temps than simply turning up the stove and allowing the gas pressure to cool it as it passes by. I don’t think of this as an “inside” flame, rather a stalled or slow motion ignition sequence of the gas. But if you consider the heating of the gas to kindling temperature, I suppose that would be correct. I kind’a doubt the rep had the english to explain what was happening in any detail. I had some old WG stoves that acted similar, I got rid of them…a grey PEAK 1 from way back and an old Coleman Exponent 1. Hmmm, I think I still have the Exponent now that I am thinking about it. Turning it down very low actually heated the stove itself more than turning it up.

