What I now propose to do is to inquire whether we may not find the true figures and proportions of the ancient constellations in another way--viz., not by looking for them among the constellations as at present bounded and figured in our star-maps, but by searching the heavens themselves for them. This general method of search occurred to me very long ago while I was preparing various star-atlases, but the special mode of illustration here adopted occurred to me lately, while preparing for young astronomers in the United States a series of monthly maps showing the skies towards the north, south, east, and west, at different times of the night all the year round, and in various latitudes within the limits of the States. When I was in America I noticed, as I travelled about over a tolerably wide range of latitude, that the varying attitudes assumed by several of the constellations suggested features of resemblance to different objects. In constructing maps, simple in appearance, but based in reality on careful calculations, this characteristic came out more clearly. Adopting a particular way of presenting the connection between the various stars of a constellation, I often found the figure suggested which had actually been associated with the group of stars thus connected. Lastly, the idea of extending this method to other cases naturally occurred to me, and some of the results are presented in the present essay. The method of delineation referred to is simply that of connecting the stars of a group by lines, _ad libitum_, that is, not merely introducing so many lines as will connect all the stars into a single set, but where necessary to complete the delineation of the imagined figure, adding other lines connecting pairs of stars belonging to the group, yet not so many that _every_ pair of stars is connected by a line. The lines, again, need not be straight. On the contrary, where a group of stars forms a stream, the natural way of joining them is by lines so curved as to follow the serpentine course thus suggested. And in other cases a slight curvature of the lines joining pairs of stars will seem permissible, because corresponding to a configuration suggested by the stars themselves. It is easily seen that in some of the simplest cases, the figure associated with a constellation is at once suggested by this method of delineation. For instance, take the case of the Northern Crown. [Illustration: Fig. 32.--The Northern Crown.] [Illustration: Fig. 33.--The Dolphin.]