摘要： A favourite quotation by astronomers is a passage by the French philosopher, Auguste Comte (1798–1857). The nineteenth lesson of his Cours de Philosophie Positive appeared in 1835 and was one of several lessons dealing with the theory of knowledge in astronomy. With reference to the stars, he wrote: We understand the possibility of determining their shapes, their distances, their sizes and their movements; whereas we would never know how to study by any means their chemical composition, or their mineralogical structure, and, even more so, the nature of any organized beings that might live on their surface. In a word, our positive knowledge with respect to the stars is necessarily limited solely to geometric and mechanical phenomena, without being able to encompass at all those other lines of physical, chemical, physiological and even sociological research which comprise the study of the accessible [i.e. terrestrial] beings by all our diverse methods of observation. [1] A little later he continued: ‘I persist in the opinion that every notion of the true mean temperatures of the stars will necessarily always be concealed from us’ [1]. These passages may be amusing in the light of present knowledge, and it seems probable that Comte was ignorant of Fraunhofer’s investigations from about 1814 to 1823 in which he described the absorption lines in solar and stellar spectra (see Sections 2.4 and 2.5). In any case, the implications of Fraunhofer’s spectroscopic work were far from apparent at that time, even in scienti?c circles. However, Comte has been much maligned, mainly by astronomers ignorant of his overall positivist philosophy. Immediately preceding the much quoted passage on chemical composition of the stars is the statement: ‘Every research that is not ?nally reducible to simple visual observations is therefore necessarily disallowed in our study of the stars’ [1]. Comte therefore preached that true science was impossible if not based on direct observation or experiment (in the case of astronomy, on observation), a philosophy that scientists today should be happy to espouse. In spite of my defence of Comte’s views on the composition of the stars, which were admittedly erroneous yet pardonable, this book is largely a study of the subsequent investigations by many astronomers who reached the opposing view to Comte’s by the analysis of starlight using the spectroscope or spectrograph. This is the story of how ?rst the different chemical elements were identi?ed in stars from their spectra, of how the temperatures and other physical properties of the outer layers of stars were measured, and ?nally of how the chemical composition of the stars themselves has been quantitatively determined. The story described here spans more than three centuries if the prismatic dispersion and analysis of sunlight by Isaac Newton in 1666 is taken as the starting point. Alternatively, it encompasses nearly two centuries from the early investigations of Joseph Fraunhofer mentioned earlier, or nearly one and a half centuries from the rebirth of stellar spectroscopy with work by Huggins, Secchi and others in 1863. The studies of Newton and Fraunhofer were relatively isolated events in the history of astronomical spectroscopy, and the continuous development of the subject took place from about 1860 onwards. The scope of this book covers the main events in this development until about the end of the twentieth century, though the last decade or so of the century is not covered in the same detail as the earlier epochs. The remainder of this introductory chapter concerns some of the basic concepts in spectroscopy, which may be skipped by those already familiar with them. In addition, some historical comments on the development of astronomical spectroscopes and spectrographs are included.