§ 12. Still greater confusion in aspect is induced by the apparent change caused by perspective in the direction of the wind. If Fig. 3 be supposed to include a quarter of the horizon, the spaces, into which its straight lines divide it, represent squares of sky. The curved lines, which cross these spaces from corner to corner, are precisely parallel throughout; and, therefore, two clouds moving, one on the curved line from _a_ to _b_, and the other on the other side, from _c_ to _d_, would, in reality, be moving with the same wind, in parallel lines. In Plate 66, which is a sketch of an actual sunset behind Beauvais cathedral (the point of the roof of the apse, a little to the left of the centre, shows it to be a summer sunset), the white cirri in the high light are all moving eastward, away from the sun, in perfectly parallel lines, curving a little round to the south. Underneath, are two straight ranks of rainy cirri, crossing each other; one directed south-east; the other, north-west. The meeting perspective of these, in extreme distance, determines the shape of the angular light which opens above the cathedral. Underneath all, fragments of true rain-cloud are floating between us and the sun, governed by curves of their own. They are, nevertheless, connected with the straight cirri, by the dark semi-cumulus in the middle of the shade above the cathedral. [Illustration: FIG. 83.] § 13. Sky perspective, however, remains perfectly simple, so long as it can be reduced to any rectilinear arrangement; but when nearly the whole system is curved, which nine times out of ten is the case, it becomes embarrassing. The central figure in Plate 65 represents the simplest possible combination of perspective of straight lines with that of curves, a group of concentric circles of small clouds being supposed to cast shadows from the sun near the horizon. Such shadows are often cast in misty air; the aspect of rays about the sun being, in fact, only caused by spaces between them. They are carried out formally and far in the Plate, to show how curiously they may modify the arrangement of light in a sky. The woodcut, Fig. 83, gives roughly the arrangement of the clouds in Turner's Pools of Solomon, in which he has employed a concentric system of circles of this kind, and thus lighted. In the perspective figure the clouds are represented as small square masses, for the sake of greater simplicity, and are so beaded or strung as it were on the curves in which they move, as to keep their distances precisely equal, and their sides parallel. This is the usual condition of cloud: for though arranged in curved ranks, each cloud has its face to the front, or, at all events, acts in some parallel line--generally another curve--with those next to it: being rarely, except in the form of fine radiating striæ, arranged on the curves as at _a_, Fig. 84; but as at _b_, or _c_. It would make the diagram too complex if I gave one