The presence in the umbral spectrum of lines from the vibration-rotation bands of the H20 molecule has been reported by several authors and the problems of identification discussed by them (Mallia and Blackwell, 1970; Mallia et al., 1970; W6hl, 1969; Wt~hl, 1971). These identifications have been based on spectra obtained in the visible region around 2 5880 and in the near infra-red at 2 9280. However, the problem of allowing for the Ha O absorption lines in the spectrum of the earth's atmosphere is a very difficult one. In these circumstances, and because more information about the H20 spectrum is now available, we have made a re-appriasal of the identification of H20 lines in the umbral spectrum using a new spectrum in the region of 2 9280 obtained at the Gornergrat station.
The apparatus for obtaining this spectrum was very similar to that used in earlier work (Mallia et aL, i970) except for an improved cooled housing for the multiplier and a more sensitive pre-amplifier. Integration was continued for about 40 minutes in order to reach a peak-to-peak noise level of about 17o of continuum. The spectrum of the adjacent photosphere was scanned immediately afterwards. The observations were all carried out within one hour of local noon.
The correction necessary to eliminate the effects of scattered light was determined from the ratio of the observed intensity in the umbra to that in the photosphere, compared with the expected ratio using the Stellmacher-Wiehr (1970) model for the umbra. A further correction to the spectrum had to be carried out in order to remove the effects of absorption in the terrestrial atmosphere, in this wavelength region arising wholly from lines of water vapour. A difficulty here arose because the published identifications of tines in this region of the photospheric spectrum (Babcock and Moore, 1947) ascribe them almost wholly to terrestrial water vapour, although line counts in nearby regions showed that some ten or more Fraunhofer lines should be present in the wavelength region we observed. In fact, as our spectra had much weaker terrestrial H20 lines than those on which the published identifications are based we were able to identify several Fraunhofer lines. These identifications, combined with a calculation of the relative intensities of the terrestrial H20 lines enabled us to derive the run of atmospheric absorption with wavelength (Figure 1, shaded portion). The residual intensity in the umbral spectrum was then corrected for this absorption, with the final result for the intrinsic umbral spectrum shown in the lower part of Figure 1.