Snip
Tungsten has a temperature coefficient of about
0.0048/deg C - pretty similar to copper so resistance measurement with a decent 3 1/2 or 4 1/2 digit multimeter gives quite reasonable resolution with none of the nagging nuisance of cold junction compensation.Alternatively because it's a bit more sensitive than the sensors in platinum resistance thermometry (0.0039/deg C) a mod to standard instrumentation would give direct reading.
I've no information on the temperature coefficient linearity but I expect Google can produce something.
The quartz envelope totally protects the essenial resistive element and, unlike glass, is resistant to heat shock. I haven't tried it myself but I would expect it to be OK for at least partial immersion in molten metal.
The snags are the difficulty of making reliable low resistance connection to the lamp pins and the relatively low, cold, lamp resistance. High voltage, low wattage lamps are the best choice - something like 240V 50W. For connection I use mechanical connectors resued from chocolate box connector strips but some delicate work with a TIG welder would be better.
The three or four terminal resistance measurement system used in platinum resistance thermometry instrumentation is vastly less sensitive to contact resistance errors but I haven't experimented in that direction.
Jim