We now come, in Plate 28, to a cloud of singular beauty. It forms rapidly in a clear sky, its first traces bearing a striking resemblance to cirro-macula, but the floccules, instead of remaining semi-transparent or dropping cirrus threads, rapidly become opaque balls of cloud which lengthen upwards. This upward tendency causes the formed cloudlets to have their longer axes vertical, which is very characteristic. It might be named alto-cumulus castellatus, or high-turreted cloud. Mr. Ley named it stratus castellatus, or turret-cloud, but it certainly belongs to the cumulus section of the alto group. Thunder weather is the invariable condition for its production. If it is seen, at least in England, thunderstorms are certain to be recorded not very far away. When this particular photograph was being taken in South Devon, very destructive storms were recorded in Brittany and in the English Midlands, and the anvil-shaped tops of unmistakable thunder-clouds were visible above the horizon while the exposure was being made. [Illustration: Plate 28. HIGH TURRETED CLOUD. (_Alto-cumulus Castellatus._)] Another form of almost equal beauty is shown in Plate 29. The rounded balls make their appearance as semi-transparent spots upon the sky, and in their general characters might easily be mistaken for cirro-macula. But a few minutes will be enough to decide the question. The little spots rapidly grow denser, frequently becoming ragged at the edges; they never drop down the slender filaments which usually descend from cirro-macula, and their edges are never denser than their central parts, which, it will be remembered, was a frequent feature of the true speckle cloud. The cloudlets are obviously rounded balls arranged in patches, which may turn gradually into alto-stratus by their fusion, or, after an existence of minutes or hours, the whole may disappear by a disintegration of each ball, by its breaking up into a ragged mass and melting away. The altitude at which this cloud forms is between 5000 and 9000 metres, according to measurements made by the writer, the actual specimen figured being about 7000. It is almost as characteristic of thunder weather as the last, but whereas Plate 28 shows a variety which is most often seen before 3 p.m., since it only occurs while the cloud planes are rapidly rising, the one before us may be formed at almost any time of day, but most frequently occurs in the afternoon. An imperfect form of it is frequently met with about sunset, in which the rounded balls are not usually so well defined as when the sun is high above the horizon. Alto-cumulus glomeratus would be a suitably descriptive name. [Illustration: Plate 29. HIGH BALL CUMULUS. (_Alto-cumulus Glomeratus._)]