TEETER . 0 -— SCIENCE. 8 December 1967, Volume 158, Number 3806 to produce absorption or emission spectra. We know from laboratory studies the chief spectral lines produced by each species in the range of wavelengths from 911.6 to 6800 A. Our predictions of the ultraviolet spectra of stars are based on the idea that any theory and model, representing well the part of the spectrum of a particular atom or ion observed in the spectral region 3000 to 6800 A, should be also valid for prediction of the spectrum in the ultraviolet spectral region. Ultraviolet Spectra of Stars The ultraviolet spectra of stars are discussed from both theoretical and observational viewpoints. Anne B. Underhill and Donald C. Morton From the shape, strength, and wavelength of stellar absorption lines, one can deduce information about the com- position, density, temperature, and state of motion of the gas forming stellar atmosphere. If emission lines are observed, further particular conclusions can be drawn about the physical characteristics of the outer parts of the stellar atmosphere. Ground-based observations of stars, with spectrographs illuminated by light collected with large telescopes, permit one to study spectra from 3000 to 6800 A and into the infrared in the parts of the spectrum that are not obscured by absorption bands originating in the earth’s atmosphere. Extension of the observations into the infrared is done with recording devices sensitive to low levels of light for stars that are rather cool and so produce much radiation at long wavelengths. In practice, most stellar spectroscopy is done between the lower limit ‘of wavelength, 3000 A, imposed by the ozone absorption bands of the earth’s atmosphere and about 6800 A —the long limit of wavelength of most panchromatic photographic emulsions. The stellar spectra are usually recorded photographically; thus one may use most efficiently the limited amount of observing time that is available for high-resolution spectroscopy with large telescopes. Dr. Underhill is professor of astrophysics at the University of Utrecht, The Netherlands. Professor Morton is a research astronomer at Princeton University Observatory. 8 DECEMBER 1967 No atom, ion, or molecule in a gas Rockets and satellites provide means of transporting telescopes and spectrographs above the ozone of Earth’s atmosphere. Once an instrument is above the ozone layers, one may expect to record stellar spectra from 3000 A to the Lyman limit of hydrogen at 911.6 A. Since interstellar space contains neutral hydrogen atoms, radiation emitted from stars more distant thas the sun is absorbed strongly at wave- lengths shorter than 911.6 A. At extremely short wavelengths (soft x- rays), interstellar space becomes again fairly transparent, but normal stars are not expected to radiate a measurable amount of energy at such wavelengths. It is practical and convenient to define the ultraviolet spectrum of stars as lying between 911.6 and 3000 A. Two questions arise: (i) Which stars are expected to radiate in the ultraviolet spectral range at a detectable level? (ii) Is it reasonable to expect the spectral lines observed in the ultraviolet to contribute information not already available from the normally observed spectral region? To answer these questions we must consider the theoretical models used to represent stellar atmospheres, and the validity of our theory of the formation of stellar spectra. One should remember that stellar atmospheres consist of gas (atoms, ions, and molecules) through which a stream of radiation is flowing. The atoms, ions, and molecules interact with the radiation ever radiates only part of its spectrum, although we frequently observe only part of the spectrum because of the restrictions imposed by our observation techniques (untransparent windows, in- cluding the earth’s atmosphere; insensitive recording devices at some wavelengths; and suchlike). An important point is evaluation of the reliability of our theories of the formation of stellar spectra. Some weaknesses are well known, and devia- tions between predictions made with the imperfect theory, and observations can be interpreted qualitatively. One can gain new information about the stellar atmosphere by study of spectral lines when the deviations between observed and predicted line profiles and line strengths are largest. From our studies of stellar spectra with groundbased spectrographs we think we know what sort of spectral lines yield the most information. Naturally, surprises greet us in the ultraviolet spectral re- gion, but they are a bonus that comes with a successful program of observation—they are not the primary reason for making the observations. Before looking at the answers to the questions just posed and at the available observational material, we should remark that the light reaching the earth from the stars has passed through vast distances in interstellar space. The gas and dust lying between the earth and the stars absorb stellar light and form interstellar absorption lines in a stellar spectrum. These interstellar spectral features are expected to, giva pitch ia- formation about ‘the physical conditions 1273