I has answered my own question (post no6, about 3 1/2 years ago!) and I've just updated the link as that had moved.
The resistance of heating thermistors rises very rapidly at the desired temperature. If they didn't do that, there would have to be a separate control mechanism.
What I found, and is confirmed by the EPCOS data sheet, is that the resistance falls slightly as the temperature increases from room temperature to just below the desired temperature. The glue gun cooled in the time it took me to unplug it and measure it, so it was well below the desired temperature, so the resistance had fallen to near the minimum. As it cooled further, the resistance increased slightly, but far less than the increase that must have occurred at the desired temperature.
It is possible that the heating thermistors are made with a combination of PTC and NTC parts, but they are sold and used as a single component.