green, or gray, with which the gypsum is cut up, are darkened with dendrites of various dimensions and sometimes very elegant. These dendrites are likewise found in limestones, chalk, building-stones, lithographic stones, and compact marbles; in sandstones, granite, and various other crystalline rocks. They are not always black; some are the color of rust; some are metallic, and consist of common pyrites between sheets of slate, or copper, or native silver, or gold. Finally, besides superficial dendrites, deep ones are known, which are developed across the mass of the stones. The best-known specimens of this kind are those which make appropriate the special designation of arborized agate (Fig. 3).
This name, like that of dendrites, shows that a vegetable origin was at first attributed to these accidents. Sometimes fancy went further; and Fig. 4 represents, from Mylius, whom we have already quoted, the figure of a dendrite in which the author saw a landscape—a plain traversed by a river and bordered by a chain of wooded hills, and pierced with caves. It is easy to discover that dendrites have none of the characteristics of the vegetable ramifications with which we are at first inclined to compare them, and, when we study them under a sufficient magnifying power, the crystalline structure of most of them appears distinctly. This is especially the case with the black dendrites, which are most, abundant, and is shown in the originals of Figs. 2 and 4, which I have particularly studied, and have been able to produce artificially. It is evident that these dendrites, which consist of a hydrated oxide of manganese—the acerdesis of mineralogists—are the result of a precipitating action exercised by calcareous rocks on water containing traces of metallic salts. Hence we might expect to obtain an imitation