with that capable of saturating gold and silver, we may at least speculate that the three may form a series consisting of two substances combined in different proportions. It is true that we must be extremely cautious about venturing upon hypotheses involving a compound constitution of bodies which all our efforts have hitherto proved ineffectual to decompose, but on the other hand it must be admitted that when we find so-called elements arranging themselves into a series of terms having a common difference, and when we find the terms of these series united by equality or simple relation of atomic volume, we cannot grant that their elementary nature has been absolutely established. The following substances combine relations of chemical equiv alents already pointed out, with analogies of atomic volume: Atomict Relation of

Differences of
Equivalents.

volume.

At. vol.

(Where phosphorus, arsenic, &c., are compared in the solid state, the unit of relation is of course different). It has been already remarked that in point of chemical relations generally, lead and tin are less closely united with the series than the other members composing it, but the relation between the atomic volumes of lead and antimony, the latter almost the last term at the other end of the series, is almost absolutely exact. Nitrogen is of course omitted in the second table, as we do not know what would be its atomic volume in the solid state.

In the cases of nitrogen, tin and lead, the equivalents are taken with a negative sign, as before explained.

The numbers here given for the atomic volumes are calculated from the specific gravities adopted in Gmelin's Handbook, and the latest and most reliable determinations of chemical equivalents.

It will be observed that the difference between the equivalents of copper and silver approaches very near to absolute exactness with half the difference between the equivalents of silver and gold, and as the equivalent of copper is by no means positively settled, the relation may be rigorously exact. If we take the mean between the number adopted by L. Gmelin and that adopted by the Jahresbericht (always considering cuprous oxyd as euO), we shall have for the difference between the equivalents of copper and silver the number 445, half the difference between the equivalents of silver and gold, with mathematical exactness.

In the series Hg, Cd, Mg, Zn, similar analogies are not well marked: the atomic volumes of the three first metals are not very far apart,

Hg,
Mg,*
Cd,

Atomic volume.

7.37

6.85

6.48

but the atomic volume of zinc differs considerably; it is 4.72. In each of these different series, each term differs in its equiv alent by the number 45 or a number approaching very near to 45, and yet the addition of this large amount of matter is in most cases accompanied by no change in volume, or when a change takes place, it is expressed by some simple relation to the original volume. Some of these relations of atomic volume are well known and are only here presented in view of the confirmation which they afford of the series here established, but it is believed that the connection of the atomic volume of copper with those of gold and silver, and of those of tin and lead with those of the elements of the antimony group are pointed out for the first time.

Ammermüller has noticed a fact not wholly dissimilar from this in the case of protoxyds of copper, mercury, tin and lead, which combine with a second equivalent of oxygen without change of atomic volume, the density alone increasing. But according to L. Gmelin, the specific gravities on which he based.

* Taking the sp. gr. of magnesium at 175 as determined by Deville.

his calculations are too unreliable to render the fact well established.

The numbers adopted for the equivalents in the foregoing calculations are those obtained by the latest and most reliable determinations; they are taken from the table contained in the Jahresbericht der Chemie of Kopp & Will for 1857, published in August, 1858, and the last which the author has been able to obtain at the time of concluding this paper, and have been in no case altered or modified in the slightest degree with a view to preserve or increase numerical relations, which by slight changes of this kind can be often rendered much more symmetrical. Dumas, in one of his highly interesting papers on this subject (Comptes Rendus, XLV, 709, extracted in Kopp and Will's Jahr. 1857) in his series a+xd+yd', adopts the equivalents N=14, P=31, As=75, Sb=119, Bi 207 (see Jahresbericht, p. 35, where the equivalent of Bi is erroneously printed 108, by substitution of the values given for a, x, y, d, and d', 207 is obtained): whereas the equivalent of Sb as lately found by R. Schneider, confirmed by H. Rose, and adopted by Kopp & Will is 120 3. In another place (Comptes Rendus, XLVII, 1027) Dumas has taken the equivalent of the same metal at 122, thus adopting alternately the numbers 119 and 122, neither the true one, according to the exigencies of the two series. The equivalent number of bismuth in the series just mentioned is taken at 207, whereas it should be 208. In the series a+xd we find Mg=12, Ca=20, Sr=44, Ba 68, Pb 104-the last three should be Sr 43.77, Ba 68.6, Pb 103.5. So with Li, Na, K, V, Zr, &c.

In the foregoing tables the calculated and received equivalents are placed by side of each other for comparison. The differences rarely exceed the possible errors in the determination of chemical equivalents, respecting some of which there is still much doubt. Dumas, in the paper above referred to, gives the results of many new determinations by himself, and arrives at the number 26 for both chrome and manganese, instead of the ordinarily received Cr=267, Mn=27.5. For copper his results disagreed too much to lead him to any positive conclusion.

The analogies here presented, all depending upon the same or approximately the same number, extend therefore

To the series Pb, Sn, N, P, As, Sb, Bi.

To the series Hg, Cd, Mg, Zn.

To the series Au, Ag, Eu.

To the magnesia group, including Mn, Fe, Co, Ni, U, Co, and some of the metals also classed in the three preceding series.

The interesting paper of Prof. Cooke (Memoirs Amer. Ac., 2d ser., vol. v) to which the author's attention has been called since concluding this paper, will be more particularly referred to in the Second Part.

To the metals belonging to the group Ti, Ta, W, V, Mo, Te and No; Sn belongs also to this series as well as to the first.

To the platinum group, Rh, Ru, Pd, Pt, Ir, Os.

To Є, B, Si.

To G, Al, Zr.

The differences between Cl and Br, Br and I, approximate to the same number, as likewise do the relations between Li, Na and K, and between Ca, Sr and Ba.

This relation, therefore, extends to no less than forty-eight of the elementary bodies: to all except those as yet imperfectly understood, most of which may yet range themselves under the same law, and except the oxygen group, oxygen, sulphur, selenium and tellurium, substances which stand alone and unmistakably apart from the other elements.

Philadelphia, Nov. 10, 1859.

ART. XIV.-Remarks on the Dissolution of Field Ice; by CHAS. WHITTLESEY, of Cleveland, Ohio.

THE interesting paper of Col. Totten, U. S. A., in the November number of this Journal for 1859, upon the rapid disappearance of ice in the northern lakes, recalls some observations that I had an opportunity to make on Lake Superior a few years since.

On the 8th of March, 1855, the inhabitants of Eagle River, a village in Haughton County, situated upon the most northerly part of Point Kewenaw, were engaged in procuring ice for their summer use. The severity of winter in that latitude (47° 22' north) had so far relaxed, that the surface of the field was slightly porous from the direct action of the sun. There had been no rain; the atmosphere was clear and cool, but on the sunny side of houses and other objects the snow melted rapidly in the day time.

a

Below the soft and moist surface, at a few inches in depth, the ice was solid and pure to the bottom, its thickness being thirty inches. The blocks which the people were cutting out, were taken about 1000 feet from the shore. One of them nearly in the form of a cube, of thirty inches on each face, was suffered to lie upon the unbroken ice, its natural surface uppermost, as represented in the figure here inserted.

Block of ice 30 inches thick.
a a, upper surface.

I was thus enabled to take a direct view of the progress of its decay, as no doubt others have done many times, upon these lakes. As the force of the sun increased, the porous part on the

surface increased rapidly in depth, lines or planes of separation extending downward from it into the hitherto transparent and homogeneous mass. There were not at any time horizontal planes visible, indicating layers or lamination, in the original structure. A thin film of matter followed each newly formed crevice downwards, and bubbles of air rose continually through the same to the surface. These planes of division converged below, giving the block the appearance above represented, of inverted spikes or rudely formed pyramids, with their bases upward. By ten o'clock A. M. the upper half of the block was divided in this manner. The figures were somewhat regular and were principally triangular and rectangular, reminding me of the imperfectly columnar red trap of the north shore of Lake Superior. By noon the block was so far disintegrated that it fell to pieces under a single blow, and remained a pile of roughly formed spikes, pyramids and prisms of various lengths. After this as so much new surface was exposed to the sun it melted very fast. The newly cut ice was still solid and clear except a few inches at the surface.

There seemed to be in the block that had so suddenly lost its form and solidity, a process of contraction, arising from an increase of temperature. I presume that this appearance can be thus accounted for. No doubt the planes of division existed in the solid ice, as results of the crystallization in freezing. The general law of structure in all masses slowly crystallizing from a state of fusion is the production of a prismatic structure perpendicular to the cooling surfaces. Basalt assumes its polygonal figures in obedience to the same law, and the structure of ice is quite in accordance with it. Its effects are not wanting even in some pastes, like starch and domestic cake.

This structure exists often where it is concealed. An ingot of block tin shows no crystalline structure, but by slow fusion the amorphous parts melt and run out leaving a skeleton of crystalline prisms. Ice is in the same predicament, and since in freezing water expands one-seventh of its volume, the first result of the fusion of a part of it is to dissect out the prismatic masses, leaving them standing isolated by reason of their being on a larger scale than the fluid volume from which they were formed. In this process the air bubbles no doubt materially assist by opening channels of escape for the ice-water.

What I have stated may assist in explaining why immense fields of fresh water ice disappear in a single gale of a few hours duration. When the temperature rises above 32° the ice soon loses its cohesion, and the first agitation breaks it up. In popular phrase it sinks, and is thus lost sight of suddenly; but in truth it is dissolved by the warmer water acting upon the fragments in the shape of little columns and pyramids such as Col. Totten saw strewed along the shore of Lake Champlain.