We have now only to explain how it happens that on very rare occasions the colors, instead of lying in irregular patches, form definite fringes round the borders of the cloudlets. The circumstances that give rise to this special form of the phenomenon appear to be the following: While the cloud is in the process of growth (that is, so long as the precipitation of vapor into the crystalline state continues to take place), so long will the crystals keep augmenting. If, then, a cloudlet is in the process of formation, not only by the springing up of fresh crystals around, but also by the continued growth of the crystals within it, then will that patch of cloud consist of crystals which are largest in its central part, and gradually smaller as their situation approaches the outside. Here, then, are conditions which will produce one color round the margin of the cloud, and that color mixed with others, and so giving rise to other tints, farther in. In this way there comes into existence that iris-like border which is now and then seen. The occasional upsetting of the crystals, which is required to keep them fluttering, may be produced in any of three ways. The cloudlets may have been formed from the blending together of two layers of air saturated at different temperatures, and moving with different velocities or in different directions. Where these currents intermix, a certain amount of disturbance will prevail, which, if sufficiently slight, would not much interfere with the regularity of the crystals, and might yet be sufficient to occasion little draughts, which would blow them about when formed. Or, if the cold layer is above, and if it is in a sufficient degree colder, there need not be any previous relative motion of the two layers; the inevitable convection currents will suffice. Another, and probably the most frequent, cause for little breezes in the neighborhood of the cloudlets is that when the cloudlets are formed they immediately absorb the heat of the sun in a way that the previously clear air had not done. If they absorb enough, they will rise like feeble balloons, and slight return currents will travel downward round their margins, throwing all crystals in that situation into disorder. I do not include among the causes which may agitate the crystals another cause which must produce excessively slight currents of air, namely, that arising from the subsidence of the cloudlets owing to their weight. The crystals will fall faster wherein cloud masses than in the intervening portions where the cloud is thinner. But the subsidence itself is so slow that any relative motions to which differences in the rate of subsidence can give rise are probably too feeble to produce an appreciable effect. Of course, in general, more than one of the above causes will concur; and it is the resultant of the effects which they would have separately produced that will be felt by the crystals.