angles to the bar, when, as should be the case, it was supported. These results show, as we had expected, that parallel currents increase each other, while opposite currents diminish each other.

The experiment was repeated, approaching the armature with its axis at right angles to the axis of the bar, when results similar to those of Jamin were obtained.

The delicacy of these experiments required a great number of repetitions. This was done, and care was also taken to carefully clean the armature each time lest any moisture from the hands, or other foreign matter, should make it adhere to the bar. Chemically pure iron was used to prevent the armatures acquiring a permanent magnetism.

I have thus attempted to show that the results of M. Jamin's experiments, although undoubtedly correct, do not warrant the conclusions respecting Ampère's theory which he has drawn, but, on the contrary, are a direct consequence of that theory: first, by investigating the condition of the currents in the armature; and, secondly, by showing that contrary results are obtained by making the armature very long in proportion to its diameter, and approaching it always with its longer axis parallel to that of the bar.

XVIII.

MELANOSIDERITE:

A NEW MINERAL SPECIES, FROM MINERAL HILL, DELAWARE COUNTY, PENNSYLVANIA.

BY JOSIAH P. Cooke, JR.

Read, May 11, 1875.

FOR the specimens of this mineral I am indebted to Mr. W. W. Jefferis, of West Chester, Pa., who informs me that the locality is not over 500 yards from the Hallite locality (see these Proceedings, Vol. IX., page 59), lying a little off from the serpentine range in a bed of clay.

It is very brittle, and its Sp. Gr., in two determinawater), was 3.390 and 3.392. Color, black, with a tinge of Subtranslucent; and, in thin Heated before the blowpipe

The mineral is compact and amorphous. fracture is conchoidal. Hardness about 4. tions (made with alcohol, but referred to Lustre, vitreous inclining to resinous. red. Streak, brownish red to brick red. plates, cherry red by transmitted light. in a closed tube, it decrepitates and gives off water. In the forceps it fuses at about 4 to an iron-gray mass, which is strongly attracted by the magnet: On platinum wire with soda, it dissolves with effervesWith borax glass it gives the reactions for iron. The powder dissolves very readily in muriatic acid; but, as soon as the amount added exceeds a certain limit, the solution gelatinizes.

cence.

In the following analysis, the water was determined by ignition; the iron and alumina were weighed together, and the amount of iron in the ignited precipitate determined by titration. The silica was separated and weighed in the usual way.

The last column gives the percentage corresponding to the symbol Fe, Si H; and it is obvious that the composition of the mineral is remarkably definite, and the agreement of the analyses with theory very close.

Melanosiderite is closely related to the sesqui-hydrates of iron. If the silica could be regarded as an impurity, the ratio between the sesquioxide and the water would be precisely that of Limonite; and, on the other hand, if we regard the basic radical as neutralized both by the silicon and the hydrogen, then the ratio would be that of Göthite. Melanosiderite differs, however, most markedly from both of these minerals in its specific gravity.

Moreover its vitreous lustre, its fusibility, its definite composition, and, above all, its reaction with acids (gelatinizing), indicate that it is a true silicate. The mineral which it most closely resembles, both in external appearance and specific characters, is Hisingerite, but this contains thirty-six per cent of silica. Melanosiderite, however, is evidently a highly basic compound of the same class; and the new species has a special interest, arising from the circumstance that it is a definite natural example of a class of compounds which are so characteristic of the ferric radical. Its amorphous colloidal structure is wholly in accordance with this view of its constitution. The name Melanosiderite is derived from Greek μhas and oídŋoos, and refers to the very striking black color of the mineral.

The analytical work in this investigation has been done by Mr. W. H. Melville, of the Senior Class in Harvard College.

XIX.

ON TWO NEW VARIETIES OF VERMICULITES,

WITH A REVISION OF THE OTHER MEMBERS OF THIS GROUP.

BY JOSIAH P. COOKE, JR.,

Erving Professor of Chemistry and Mineralogy at Harvard College,

AND

F. A. GooсH,

Assistant in the Chemical Laboratory.

Presented, May 11, 1875.

SINCE the publication of the writer's first monograph on the vermiculites,* two new varieties of this group of minerals have been brought to his notice by Mr. W. W. Jefferis, of West Chester, Pa., who has most kindly furnished the materials for the following investigation. The first of these varieties which occurs at Lerni, Delaware Co., Pa.—has the following characters: The unaltered mineral is of a dull sea-green color, has a highly developed micaceous structure, is an aggregate of rough hexagonal plates, and of very imperfect external form. It is transparent in moderately thin laminæ, and is free from enclosed foreign matter. The optical characters of the mineral closely resemble those of the Culsagee variety of vermiculite, the angle between the optical axes varying in different parts of the same laminæ from 18° to 0°. Its hardness is about 1.5, and three determinations of its specific gravity (taken in alcohol at 23° C.) gave 2.409, 2.368, and 2.373. Heated in a closed tube, it gives off water acid in reaction, changes color, and doubles its volume. Heated before the blowpipe, it fuses to a dirty enamel.

The mineral was prepared for analysis by drying at 100° until its weight was constant, and in this condition was easily decomposed by hydrochloric acid. The bases, after the separation of silica, were con

*The Vermiculites, their Crystallographic and Chemical Relations to the Micas, by Josiah P. Cooke, Jr. These Proceedings, vol. ix. p. 35. The analyt ical work in this second paper has been done by Mr. Gooch.

verted into nitrates, and separated by Deville's process. The iron and alumina were weighed together as oxides, and the magnesia as the pyrophosphate. The amount and condition of the iron were determined by decomposing the mineral by the process described by the writer in Am. Jour., vol. xliv. page 347, estimating by a standard solution of potassic permanganate the ferrous iron at once, and the total amount of iron after reduction to the ferrous state. Small amounts of lithium and potassium were found by the spectroscope in the residue obtained in Smith's process for the determination of alkalies, but the total amount of alkali probably did not exceed onetenth of one per cent of the mineral. The mineral, dried at 100°, lost upon ignition,

(1.) 11.65

(2.) 11.67

(3.) 11.71

(Mean.)

11.68

per cent of its weight. The details of these determinations will be referred to hereafter. Taking, then, 11.68 as the percentage of water in the mineral, the results of analysis are as follows: —

The second of these two varieties of vermiculite occurs at Pelham, Mass. It has a greenish-yellow tint, resembling closely in color the Culsagee variety. In the specimen examined, however, the scales. were very much smaller than those of the Culsagee variety, and exhibited no regular outline. Examined with the microscope, they appeared equally free from interpenetrating material. The plates do not show the striation observed both in Culsageeite and Jefferisite. No evidence of macling could be found. The plates were optically biaxial,