The filament used in the Nernst lamp was a zirconia ceramic (also containing some yttrium oxide) which becomes electrically conductive after being heated. As the filament was a ceramic, it could operate in air. Zirconia melts a 2715 ?C.http://www.youtube.com/watch?v=uGxSwumcFtU
"The red, green, and blue brightness of the Welsbach gas mantle was also measured. The red-blue ratio thus found corresponded to a color temperature of about 2800 K. The measurements showed that for the mantle the green brightness was about 30 per cent too high for an exact color match. That is, if the curve that represents the spectral distribution of a Welsbach mantle be made equal in the red (665nm) and in the blue (467nm), the two curves would differ in the green where the intensities for the Welsbach mantle are about 30 per cent higher than those for the black body."Abstract Bulletin of the Physical Laboratory of the National Electric Lamp Association (1913), p548
The interesting thing to note here is that both the Welsbach gas mantle lamp and Nernst lamps emit spectral curves that more closely correlate to a higher
color temperature than the temperature they are actually heated to, an effect sometimes referred to as "candoluminescence". This is believed to be because the thoria does not efficiently radiate infrared., while the 0.5-1% cerium dopant efficiently emits in the visible spectrum, and the thoria is transparent to visible light. In both cases the spectral curves are very smooth, no characteristic spikes from rare earth elements, it is just not quite the same shape as an ideal black body.
The thoria used in the old mantles is slightly radioactive, so the modern gas mantles often use ytteria instead. I think the older style thoria mantles give off a slightly whiter color of light though.