Furthermore, sometimes even the setting sun produces a rainbow; surely at that time, being near[30] the earth, he strikes the clouds on their lower side. And yet then, too, the bow is only a semicircle, though the clouds receive the sunlight on their lower and darker portions. The Stoics, who hold that the light is reflected in the cloud as 4 in a mirror, make the cloud hollow like the section of a ball. Such a mirror, being but part of a circle, cannot, they think, reproduce a whole circle. I give my adherence to the proposition, but I cannot agree to the argument in its support. For, if the whole figure of a circle placed opposite a concave mirror is reproduced in it, then there can surely be nothing to prevent the whole of a ball being seen in a semicircular circular mirror. Besides, we have already shown that 5 complete rings resembling a rainbow surround the sun and the moon at times. Why should the circle be complete in the halo, but never in the rainbow? And then again, why should the clouds that receive the sunlight be always hollow ones, and not sometimes flat or bulging? [30] The common reading makes this adjective refer to clouds--the clouds which are near the earth. Aristotle says that rainbows are formed, after the autumnal equinox, at any hour of the day, but in summer only either in the early part of the day, or when the sun has begun to sink. The cause of this is obvious. In the first place, about midday the great heat of the sun 6 dispels the clouds: he cannot be reflected in the clouds which he breaks up. But in the early morning and as he sinks toward the west, his rays have less power, and can thus be resisted and reflected by the clouds. In the second place, the sun is not wont to form a bow except when he faces the clouds in which it is formed. When the days are 7 shortening in autumn, his rays are always slanting. Therefore, he has some clouds facing him that he can strike, at any part of the day, even at the hour at which he attains his meridian height. But in the summer season he sails right overhead. Therefore, in the great altitude of his midday course, he looks down on the earth too directly to encounter any clouds. He has them at that period all beneath him. IX I must now go on to speak of _Streaks_ (watergalls, sun-dogs), which 1 are as bright and varied as the rainbow, and commonly received by us as equally indicative of rain. No great labour need be spent in explaining them, for they are just incomplete rainbows. They have the variegated 2 appearance of the bow, but none of its curve. They lie in a straight line. They are formed near the sun, as a rule, in a moist cloud that has begun to break up. Thus, they have the same colour as is found in the rainbow, but there is a difference in the shape, due to the corresponding difference in the clouds over which they stretch. X