of approximately 30000: the 725 - 616 line at 29.84 microns, the 441 - 312 line at 31.77 microns, and the 432 - 303 line at 40.69 microns. One additional emission line, the 441 - 414 line at 37.98 microns, and one absorption feature at 38.08 microns that we attribute to a blend of the 1313,0 - 1312,1 and the 1313,1 - 1312,2 water lines were observed in grating mode ata resolving power of 2000. The observed emission line fluxes were 3.2 x 10-19, 6.3 x 10-19, 2.3 x 10-19 and 2.8 x 10-19 Wcm-2 respectively, and the equivalent width of the absorption feature was ~ 10 km s-1. To within the possible errors in the flux calibration, the observed emission line fluxes can be accounted for simultaneously by a model similar to that of Chen & Neufeld (1995), given a mass-loss rate in the range (0.5 - 3) x 10-5 Msolar yr-1 . This range lies at least a factor ~ 2 above an independent estimate of the mass-loss rate that may be derived from dynamical considerations, and at least a factor ~ 30 above previous estimates based upon the interpretation of CO observations.
[