When William Herschel sent to the Royal Society his 1784 paper on the construction of the heavens, he was still of the opinion that there were true nebulae to be found alongside those that were simply distant "star clusters in disguise" (as he later put it 1 ), and he had persuaded himself that observers could tell the difference: true nebulae shone with a milky appearance, while star clusters disguised by distance appeared mottled, 'resolvable'. But, as we saw in Part 1, 2 he soon after came across M17, which we know as the Omega Nebula, and then M27, the Dumbbell Nebula; and to his surprise each contained both nebulosities, milky and 'resolvable'. Already convinced that the Milky Way is the optical effect of our immersion in a layer or 'stratum' of stars (the Galaxy), and that he had encountered the remnants of other strata that had already fragmented into nebulae and clusters as a result of gravitational attraction, 3 he interpreted these two objects as entire strata of stars, their stars in the middle distance appearing to us as resolvable nebulosity and those in the far distance appearing as milky. He concluded that, contrary to what he had hitherto believed, all nebulae, whether milky or resolvable, were formed of stars.With the puzzle concerning nebulosity apparently solved, Herschel prepared a second paper on the construction of the heavens, dated New Year's Day 1785 and one of the greatest in the history of astronomy. 4 He begins in dramatic fashion with an imagined model of a stellar universe in its infancy: "Let us then suppose numberless stars of various sizes, scattered over an indefinite [but not infinite?] portion of space in such a manner as to be almost equally distributed throughout the whole." Gravity is the agent that will bring about change in this distribution: "The laws of attraction, which no doubt extend to the remotest regions of the fixed stars, will operate in such a manner as most probably to produce the following remarkable effects." And he goes on to describe the "Formation of Nebulae" -that is, star systems -likely to result from irregularities in the hypothetical distribution, places where the gravitational pull is greater than normal.He instances the presence in a particular region of either one large star, or unusually many ordinary ones. The large star will attract those stars immediately around it, and this will result in a regular (spherical, or near-spherical) cluster of stars -Form I. The unusually numerous ordinary stars will also draw in those around, and this will result in an irregular cluster, its shape depending on the layout of these ordinary stars -Form II. Combinations of these situations will result in more complex configurations, Forms III and IV, as well as in Form V, "great cavities or vacancies by the retreat of the stars towards the various centers which attract them".Does this mean, he asks himself, that clusters will one day end in what we today JHA, xlii (2011)