A low-light-level television camera and a television spectrograph were used to study the variation of color with elevation angle in auroras having type-B red lower borders. It was found that much of what is commonly described as type-B aurora results from a simple separation between the prompt red and blue molecular nitrogen emissions on the one hand and the persistent green oxygen line on the other hand. The red and blue emissions combine to produce magenta which appears on the leading edge of rapidly moving auroral features. True type-B aurora, not dependent on motion, was observed to result from a suppression of the green line at low altitudes, presumably due to quenching. While the quenching of the green is the predominant effect, there are, in addition, changes in the shade of magenta resulting from an enhanced red/blue ratio associated with the phenomenon of auroral hems (bands of enhanced luminosity along the lower borders of auroral curtains). In one case there was also evidence for a redistribution of the intensities of the nitrogen first positive bands, possibly by intercollisional transfer of energy from the W states to the B states as proposed by Benesch [1981>. In sum, evidence was found to support each of the prominent theories for type-B auroras and that the various mechanisms can coexist. ¿ 1998 American Geophysical Union