As Herschel discovered hundreds of double stars in the course of compiling his catalogs, he found what was poten-tially a third classification besides optical doubles and binaries: double stars in which one of the components was exceedingly dim. If it could be shown that these components had orbital motion and thus were in fact binaries, it would indicate the dimness of these compa-nions was not an artifact of distance but represented an actual intrinsic variance between the luminosity of the primary star and that of its companion. Herschel’s com-ment in 1833 urging more study of these objects indicated that such a large variation exceeded what could be expected from stellar objects. Proof of the binary nature of such pairs would imply that the dimmer object could not be a star. For something to be so intrinsically dim, it would have to be shining by reflected light. Herschel’s speculation then was not whether such planets could be observed but whether a star could be so dim. In 1833 Herschel assumed it could not. By the end of the decade this assumption was being abandoned. The determination of parallax provided astronomers with definite measurements for a handful of stars. Some of these, such as Sirius and a Centauri were shown to be intrinsically very bright, but others, such as 61 Cygni – the first star to have its parallax successfully measured – very dim.44 A small sampling of nearby stars made apparent a wide range of stellar luminosities, and Herschel’s published works reflected this growing realiza-tion. For example, in his 1833 Treatise he speculated that ‘Sirius must, in point of intrinsic splendor, be at least equal to two suns’ though he admited this proportion was probably much greater.45 By the publication of his 1851 Outlines of Astronomy, an extended and revised version of the Treatise, uncertainty regarding the scale of stellar luminosities had disappeared. Measurements of the par-allax of Sirius had by this time allowed a more definite measurement for its absolute magnitude, given by Herschel as exactly ‘63.02 times that of our sun.’46
The cosmology of John Herschel’s father had been based on the idea that celestial objects could be classified into species like plants and animals and that the characteristic properties of members of each species varied to only a limited extent.47 Though the younger Herschel was hesi-tant to explicitly embrace this assumption, his early work on double stars indicates that the classification of stellar objects – in this instance ‘subspecies’ of double stars – still played an important role in his thought. In 1833, the assumed bounds on stellar properties allowed him to speculate on the planetary nature of objects with proper-ties beyond those bounds. Yet these bounds were even then in a state of transition and within a few years would be discarded. By 1846 the astronomer Wilhelm Bessel (1784–1846) predicted an invisible stellar companion to the bright star Sirius, calculating the mass of this unseen body by measurements of Sirius’ proper motion.48 By the middle of the nineteenth century, for some luminosity itself was no longer a necessary characteristic of stars [Fig. 3]. When Herschel returned to a description of the various classes of double stars in his 1851 Outlines, he again highlighted the ‘delicate sub-class’ of doubles in which a bright star is accompanied by a very dim companion. His list of stars in this category included all those doubles offered in his 1833 paper as having possible planetary companions.49 Here, though, there is no discussion of potential planetary companions, and the stellar nature of the objects is simply assumed. The idea of what it meant to be a star had expanded. It now encompassed even the greatest disparities of luminosity in double stars that had led to Herschel’s planetary speculations twenty years ear-lier. What might have been planets had now become stars [Fig. 4].