of intersecting curves; but the lowest figure in Plate 65 represents, in perspective, two groups of ellipses arranged in equidistant straight and parallel lines, and following each other on two circular curves. Their exact relative position is shown in Fig. 2, Plate 56. While the uppermost figure in Plate 65 represents, in parallel perspective, a series of ellipses arranged in radiation on a circle, their exact relative size and position are shown in Fig. 3, Plate 56, and the lines of such a sky as would be produced by them, roughly, in Fig. 90, facing page 128.[4] [Illustration: FIG. 84.] § 14. And in these figures, which, if we look up the subject rightly, would be but the first and simplest of the series necessary to illustrate the action of the upper cirri, the reader may see, at once, how necessarily painters, untrained in observance of proportion, and ignorant of perspective, must lose in every touch the expression of buoyancy and space in sky. The absolute forms of each cloud are, indeed, not alike, as the ellipses in the engraving; but assuredly, when moving in groups of this kind, there are among them the same proportioned inequalities of relative distance, the same gradated changes from ponderous to elongated form, the same exquisite suggestions of including curve; and a common painter, dotting his clouds down at random, or in more or less equal masses, can no more paint a sky, than he could, by random dashes for its ruined arches, paint the Coliseum. § 15. Whatever approximation to the character of upper clouds may have been reached by some of our modern students, it will be found, on careful analysis, that Turner stands more absolutely alone in this gift of cloud-drawing, than in any other of his great powers. Observe, I say, cloud-_drawing_; other great men colored clouds beautifully; none but he ever drew them truly: this power coming from his constant habit of drawing skies, like everything else, with the pencil point. It is quite impossible to engrave any of his large finished skies on a small scale; but the woodcut, Fig. 85, will give some idea of the forms of cloud involved in one of his small drawings. It is only half of the sky in question, that of Rouen from St. Catherine's Hill, in the Rivers of France. Its clouds are arranged on two systems of intersecting circles, crossed beneath by long bars very slightly bent. The form of every separate cloud is completely studied; the manner of drawing them will be understood better by help of the Plate opposite, which is a piece of the sky above the "Campo Santo,"[5] at Venice, exhibited in 1842. It is exquisite in rounding of the separate fragments and buoyancy of the rising central group, as well as in its expression of the wayward influence of curved lines of breeze on a generally rectilinear system of cloud. [Illustration: FIG. 85. _To face page 118._] [Illustration: 67. Clouds.]