In October 1859, German physicist Gustav Kirchhoff announced the results of his investigations with chemist Robert Bunsen on the dark lines that interrupt the otherwise continuous solar spectrum (1). These lines had puzzled practitioners and theorists alike since they were first observed in 1814 by German optician Josef von Fraunhofer (2).Now it seemed that Bunsen and Kirchhoff had finally confirmed what others had long suspected, namely, that an individual metal produces its own characteristic pattern of bright spectral lines when it is burned. Furthermore, Kirchhoff asserted that Fraunhofer's lines "exist in consequence of the presence, in the incandescent atmosphere of the sun, of those substances which in the spectrum of a flame produce bright lines at the same place."News of his claim spread quickly throughout the scientific world. In England, Bunsen's former student, Henry Enfield Roscoe, wrote to the secretary of the Royal Society, George Stokes (3): "Have you seen in the last no. of the Annales ... a short note about Kirchoff's [sic] discovery...?"Soon, Roscoe was offering public lectures on the subject to interested scientists and laymen alike. After one such presentation to the Chemical Society, moderator Warren De La Rue remarked (4)[I]f we were to go to the sun, and to bring away some portions of it and analyze them in our laboratories, we could not examine them more accurately than we can by this new mode of spectrum analysis....What really excited De La Rue, a stationer known for his photographs of the sun and moon, was the potential this method of analysis portended for astronomy. After all, French philosopher Auguste Comte in 1835 had clearly defined the domain of questions considered legitimate for Earth-bound observers to ask about the denizens of the celestial realm. "We can imagine the possibility of determining the shapes of stars, their distances, their sizes, and their movements," he declared, "but there is no means by which we will ever be able to examine their chemical composition" (5).Despite this obvious hindrance, professional astronomers worked productively and creatively throughout the 19th century to make many important discoveries: the successful determinations of solar and stellar parallax, the discovery of Neptune, and confirmation of the existence of an unseen companion to the star Sirius--to name a few. Meanwhile, their amateur colleagues, many of whom were no less serious about, or adept at, studying the sky, sifted patiently and tirelessly through the heavenly haystack hoping to be the first to find one of the proverbial needles that lay hidden there.But De La Rue was right. Coupling the spectroscope to the astronomical telescope did revolutionize the way astronomy was performed, realigning the very boundaries of what astronomers considered to be acceptable research.A quarter of a century after Kirchhoff's announcement, historian of astronomy Agnes Clerke marveled at the youthful audacity of a new science she called "astronomical or cosmical physics" (6). "It promises everything," she wrote, "it has already performed much; it will doubtless perform much more." And, she identified the enterprising English amateur astronomer William Huggins (1824-1910)--not Kirchhoff, Roscoe, or De La Rue--as one of stellar spectroscopy's principal founders.A London silk merchant and self-taught amateur astronomer, Huggins joined the Royal Astronomical Society (RAS) in 1854. Soon after, he retired from commercial life and moved to the suburb of Tulse Hill, where he had a substantial observatory built to house his instruments. In 1856, as he recorded his first observations in a bound notebook, Huggins began his metamorphosis from curious dilettante to confident, self-directed observer. Along with his personal correspondence and publications, these notebooks help us trace the incremental career choices by which he, a recognized novice, shaped the inner dynamics of a new research agenda in a tradition-bound exact science.Toward the end of his long career, Huggins waxed nostalgic as he witnessed his efforts being eclipsed by those of individuals like Hermann Carl Vogel (1841-1907), Edward Charles Pickering (1846-1919), and George Ellery Hale (1868-1938). With editorial assistance from his wife and collaborator, Margaret Lindsay Murray, he set to work putting his many achievements before the public, beginning in 1897 by publishing a personal retrospective entitled "The New Astronomy" in Nineteenth Century, a popular magazine of the day (7). In this stirring narrative--long a favorite source for Huggins's biographers--he recalls that soon after establishing his Tulse Hill observatory he became "a little dissatisfied with the routine character of ordinary astronomical work, and in a vague way sought about in my mind for the possibility of research upon the heavens in a new direction or by new methods." Luckily, he tells us, he had attended a lecture in 1862 given by his "friend and neighbour, Dr. W[illiam] Allen Miller."Miller, a professor of chemistry at King's College, was repeating a well-received lecture on "The New Method of Spectrum Analysis" he had delivered some months earlier. In it, he drew particular attention to Fraunhofer's pioneering work on spectroscopy, which, Miller noted, had included an examination of the spectra of Venus and several prominent stars. But celestial spectroscopy was merely a prologue to the real purpose of Miller's talk, namely, to update the audience on what was currently known about spectrum analysis and how it was being used in the chemical laboratory.Thirty-five years later, Huggins held the evening as his personal epiphany (7). The news of "Kirchhoff's great discovery," he proclaims, "was to me like the coming upon a spring of water in a dry and thirsty land..."A sudden impulse seized me, to suggest to [Miller] that we should return home together. On our way home I told him of what was in my mind, and asked him to join me in the attempt I was about to make, to apply Kirchhoff's methods to the stars.According to Huggins, Miller obliged by coming to work in his observatory "on the first fine evening."Unfortunately, Huggins's observatory notebooks contain no record of collaborative spectroscopic work with Miller in 1862. We know from correspondence and published sources that Miller was preoccupied at the time with his own research problems. Huggins, meanwhile, was taking advantage of an edge-on orientation of Saturn's rings to search for as-yet undiscovered moons of that planet. He did not find any--in fact, no one did--and the time and effort he expended on this, like many of his other short-lived or unsuccessful projects, eventually disappeared from the record.Huggins and Miller soon communicated the results of their first efforts in celestial spectroscopy to the Royal Society, leaving no doubt they accomplished the work Huggins described in "The New Astronomy." But the vividness of his later recollections belies the fact that his early research program underwent not so much a radical transformation in 1862, as a gradual and fitful shift.Operating on the periphery of the RAS, Huggins's independent research agenda was characterized by flexibility, alacrity, and openness rather than the traditional virtues of diligence, tenacity, and servitude which, in his view, restricted contemporary institution-bound observers. In the newly emerging field of celestial spectroscopy, no one, including Huggins, knew which line of research would prove the most fruitful. So, rather than, say, systematically catalog the spectra of all northern hemisphere stars, he chose to explore a number of different subjects in innovative and often technically challenging ways.In August 1864, Huggins gamely tested the spectroscope's capacity to resolve questions about the nature of nebulae. It was a bold move that ultimately propelled him to a position of prestige and authority among fellow astronomers, capturing their imaginations and increasing their awareness of the potential of spectrum analysis to generate new knowledge about the heavens.In May 1866, he became the first to analyze the spectrum of a nova (T Coronae). When the nova was lost from view, Huggins moved on to other projects: devising a method to observe solar prominences without an eclipse, spectroscopically determining the chemical composition of meteors, visually corroborating reported changes in lunar surface features, and even attaching a thermopile to his telescope to measure the heat of celestial bodies. In light of the mixed success of these projects, that he attempted them at all reveals much about his willingness, like a shrewd businessman, to take risks to establish himself securely in the forefront of research in the new astronomy.One risky project that Huggins undertook in 1867 resulted in what is arguably one of the greatest, and certainly the most influential, of his many contributions to the new astronomy: his development of a spectroscopic method to determine the motion of a celestial body along its line of sight, a method integral to modern astronomical research. It is hard to conceive that in pursuing this project, Huggins relied only on direct visual comparisons of terrestrial and celestial spectra.His notebook entries reveal this project as an audacious effort fraught with overwhelming mensurational and interpretative difficulties. But in announcing his findings to the Royal Society, Huggins expressed satisfaction that he had resolved the problems he had faced, stressed the care he had taken to insure the reliability of his measurements, and invoked the name of famed physicist James Clerk Maxwell to underscore the soundness of the theoretical foundation on which he based his conclusions. His tone was one of confidence and spirited adventure. Indeed, it could be said that Huggins's success in introducing this method to astronomical research lay in his ability to persuade his contemporaries that he had, in fact, accomplished what he claimed despite the overwhelming difficulties the method entailed.Although few understood the physical theory on which his line-of-sight measurements were based, and implementing his method was largely beyond the resources and ability of many of his fellow amateurs; celestial mechanicians, including the Astronomer Royal, George Biddell Airy, recognized it as an aid to their mission of charting the positions and motions of celestial bodies.It has been tempting for all who have read "The New Astronomy" to place it in a different category from Huggins's other published work: to see it as truer for its candor, more accurate for its detail, closer to the way things actually happened than any of his formal scientific papers. But, it is a synthetic account composed of selected events recalled some 35 years after the fact; like a well-laid string guiding us through a labyrinth after the puzzle has been solved, it excludes ambiguous choices, wrong turns, and dead ends.Huggins was not a disinterested observer of the events that led to the growth of interest in analyzing the light of celestial bodies. He was an active participant in them. He could not tame his eclectic and opportunistic research style to fit the image of the methodical and systematic scientific investigator, but he could--and did--construct a more conforming public account of himself and his work, thus artfully situating his own benchmark contributions to celestial spectroscopy in the evolving lore of 19th-century spectrum analysis.Huggins's unpublished notebooks and correspondence restore our contact with the visceral forces that drove him to venture outside the boundaries of acceptable research. They bring to light his instrumental and methodological ingenuity, his endless concern for priority, his power of persuasion, and his perseverance. Finally, they reveal him as a man who is more interesting and certainly more complex than the cautious, focused, and methodical individual portrayed in the public record.References and NotesG. Kirchhoff, Philos. Mag., 4th ser., 21, 196 (1860).Published in three segments. J. Fraunhofer, Edinburgh J. Sci. 7, 101-113 and 251-252 (1827); 8, 7-10 (1828).H. E. Roscoe to G. G. Stokes, 24 February 1860, from Stokes papers, Add. MS 7656.R788, Cambridge University Library.W. De la Rue, Chem. News 4, 130 (1861).A. Comte, Cours de Philosophie Positive, vol. 2 (Baillière, Paris, ed. 2, 1864) p. 6, translation by the author.A. Clerke, A Popular History of Astronomy During the Nineteenth Century (Macmillan, New York, ed. 2, 1887), p. 180.W. Huggins, Nineteenth Century 41, 907 (1897).H. Schellen, Spectrum Analysis in Its Application to Terrestrial Substances (Longmans Green, London, 1872), p. 466.