Now it seems to me that the reduction in thickness cannot be due to compression alone, but that a portion of the substance of the lower layers must have been removed. It is not easy to see why the temperature of the earth's crust, under a widely extended and practically permanent ice-sheet of great thickness, should ever fall below the freezing-point; and it is a matter of observation that at all seasons of the year vast rivers of muddy water flow into the frozen sea from beneath the great glaciers which are the issues of the ice-sheet of Greenland. Ice is a very bad conductor, so that the cold of winter cannot penetrate to any great depth into the mass. The normal temperature of the surface of the earth's crust, at any point where it is uninfluenced by cyclical changes, is at all events above the freezingpoint, so that the temperature of the floor of the ice-sheet would certainly have no tendency and fall below that of the stream passing over it. The pressure upon the deeper beds of the ice must be enormous at the bottom of an ice-sheet 1,400 feet in thickness-not much less than a quarter of a ton on the square inch. It seems, therefore, probable that under the pressure to which the body of ice is subjected a constant system of melting and regelation is taking place, the water passing down by gravitation from layer to layer until it reaches the floor of the ice-sheet, and finally working out channels for itself between the ice and the land, whether the latter be subærial or submerged.

I should think it probable that this process, or some modification of it, may be the provision by which the indefinite accumulation of ice over the antarctic continent is prevented and a certain uniformity in the thickness of the ice-sheet maintained-that in fact ice at the temperature at which it is in contact with the surface of the earth's crust within the antarctic regions cannot support a column of itself more than 1,400 feet high without melting. It is suggested to me by Professor Tait that the thickness of the ice-sheet very probably depends upon its area, as the amount of melting through squeezing and the earth's internal heat, will depend upon the facility of the escape of the water. The problem is, however, an exceedingly complex one, and we have perhaps scarcely sufficient data for working it out.

The Fauna of the Deep Sea.-I can scarcely regret that it is utterly impossible for me on this occasion to enter into any details with regard to the relations of the abyssal fauna, the department of the subject which has naturally had for me the greatest interest. Recent investigations have shown that there is no depth limit to the distribution of any group of gill-bearing marine animals. Fishes, which, from their structure and from what we know of the habits of their congeners, must certainly

live on the bottom, have come up from all depths, and at all depths the whole of the marine invertebrate classes are more or less fully represented. The abyssal fauna is of a somewhat special character, differing from the fauna of shallower water in the relative proportions in which the different invertebrate types are represented. It is very uniform over an enormously extended area, and in this respect it fully confirms the antici pations of the great Scandinavian naturalist, Lovén, communicated to this Association in the year 1844. It is a rich fauna, including many special genera and an enormous number of special species, of which we, of course, know as yet only a fraction; but I do not think I am going too far in saying that from the results of the Challenger expedition alone the number of known species in certain classes will be doubled. The relations of the abyssal fauna to the fauna of the older Tertiary and the newer Mesozoic periods are much closer than are those of the fauna of shallow water; I must admit, however, that these relations are not so close as I expected them to be-that hitherto we have found living only a very few representatives of groups which had been supposed to be extinct. I feel, however, that until the zoological results of these later voyages, and especially those of the Challenger, shall have been fully worked out, it would be premature to commit myself to any generalizations.

I have thus attempted to give a brief outline of certain defensible general conclusions, based upon the results of recent research. Some years ago, certain commercial enterprises, involving the laying of telegraph cables over the bed of the sea, proved that the extreme depths of the ocean were not inaccessible. This somewhat unexpected experience soon resulted in many attempts, on the part of those interested in the extension of the boundaries of knowledge, to use what machinery they then possessed to determine the condition of the hitherto unknown region. This first step was naturally followed by a development of all appliances and methods bearing upon the special line of research; and within the last decade the advance of knowledge of all matters bearing upon the physical geography of the sea has been confusingly rapid-so much so, that at this moment the accumulation of new material has far outstripped the power of combining and digesting and methodizing it. This difficulty is greatly increased by the extreme complexity of the questions, both physical and geological, which have arisen. Steady progress is, however, being made in both directions, and I trust that in a few years our ideas as to the condition of the depth of the sea may be as definite as they are with regard to regions to which we have long had ready access.

ART. XLIII.-Notes on Antimony Tannate. No. II; by ELLEN SWALLOW RICHARDS and ALICE W. PALMER.

THE next point of interest was to determine whether the method of titration as given in the preceding paper (this Journal, p. 196) was applicable to tannin-holding substances other than nut-galls and sumac. The following tests were made for this purpose:

Leaves of sweet-fern (Comptonia asplenifolia) from near
Boston, gathered the middle of May

The same, gathered on the Kennebec River, Maine, the last
of July.

Sample of catechu

Sample of ground hemlock-bark from Vermont

Crushed quercitron bark

Sample of kino..

Congo tea..

Cinchona flavor

Ground cloves

Chestnut-oak from Careyville, Tenn..

Tannin. Per cent.

7.56

8.00

7.07

29.70

41.50

7.00

4.60

9.60

7.03

3.00

We also prepared a quantity of antimony-tannate from each of these substances in the same manner as we had prepared it from commercial tannin and sumac. The composition is given as follows:

Sweet-fern (May)..

H.

Per cent.

Sweet-fern (July).

These analyses showed that the composition of the precipitate was influenced by one of two causes: either the formula of the so-called tannin which united with the antimony contained more C and H than di gallic acid,-that is, it must be something like SbOC3H8O17 or Sb2(C25 H21O12)3—or the antimonytanuate, which was formed in the solution, acted as a mordant, and carried down with it coloring matters which might or might not affect the titration, but which did affect the combustion.

To determine how far this latter cause could be held responsible, we prepared antimony tannate from the sample of tannin

which we used for all our experiments, and having washed by decantation so as to keep the gelatinous precipitate in the best condition for absorbing color, we treated solutions of several of these substances with a quantity of antimony tannate corresponding to the estimated quantity of tannin contained in the solution, so as to have the conditions the same as in the previous precipitations; in one case we increased the amount of antimony tannate. The composition of the antimony tannate thus treated in the different solutions, together with the average composition of antimony tannate as we have already obtained it from tannin, sumac and nut-galls is given as follows:

Sb.

C.

H.

Per cent.

Per cent. Per cent.

In the case of sweet-fern and quercitron, the results are very nearly those obtained by direct precipitation with tartar emetic. The possible reason for this will be considered later.

We were greatly surprised by the behavior of the solution. of hemlock-bark. In all cases after treating the solution with the previously prepared antimony-tannate, we precipitated the remaining tannin by tartar emetic as usual, and noted the quantity required as compared with that required for the precipitation of tannin in the titration. The sweet-fern and quercitron gave a precipitate about one-third less than that from the original solution. In the case of hemlock there was scarcely a trace of a precipitate, showing that the antimony tannate had dragged down or united with all the substance which had been supposed to be tannin.

In order further to test the character of the supposed coloring matter in these substances, we made a series of trials with mordanted yarn. A brown-red color was obtained from hemlock on wool mordanted with tin chloride, and on cotton mordanted with alumina. A brilliant yellow color nearly equal to that from quercitron was obtained from sweet-fern on both the wool and the cotton.

We then tested solutions of all the substances upon which we had been working with cloth mordanted in the usual way for calico-printing (with iron and alumina in alternate stripes), in order to show the presence of tannin and coloring matter at the same time. This test divided the substances into two classes, the one showing the deep black of tannin on the iron stripe, and a yellow more or less intense on the alumina; the other giving on the iron stripe a faint brownish-black, corre

sponding in dullness to that produced by gallic acid, and on the alumina a dull reddish-brown. To the first class belong nut-galls, sumac, sweet-fern leaves, bark of quercitron, black oak, white oak and chestnut oak and bearberry leaves; to the other, hemlock, catechu, kino, fever bark, cinchona bark and congo tea. For our further investigation we took sweet-fern as the type of the former, and hemlock as that of the latter class. The yellow in sweet-fern seems closely allied to, if not identical with the quercetin derived from oak-bark. A solution of sweet-fern guarancined (i. e., boiled with very dilute sulphuric acid), behaves like a solution of quercitron-bark,—a black gummy mass being formed, and the solution depositing yellow flakes which dye intensively.

Two pieces of cloth of equal size, the one dyed with one gram of sweet-fern leaves, the other with one gram of quercitron-bark, showed rather more tannin and less yellow for the sweet-fern, and more yellow and less tannin for the quercitron.

A single trial of the amount of yellow in sweet-fern, by weighing the antimony-tannate which had carried down the yellow with it, and which had been added in known quantity gave 2.5 per cent, and the amount of tannin in the filtrate had decreased about three of the eight per cent. This indicates that the antimony combines with a portion of the coloring matter, as well as with the tannin. This is further shown by the fact that the quercetin-like color obtained by guarancining was precipitated by antimony. The formula of this portion of the color must be very near to that of di-gallic acid, since the per cent of C and H in the precipitate from sweet-fern after the original solution had been treated with antimony tannate and the tannin then precipitated by tartar emetic was C 44:32 and H 3:39, and the composition of the precipitate when tartar emetic had been added directly to the solution without previous treatment with the antimony tannate was C 44.9 per cent and H 3.9 per cent.

Heppe (Die chemischen Reactionen) gives the formula of quercetin as CH18O12 and that of quercetin acid as C15H1007, which, corresponding to our formula of antimony tannate, would give respectively:

Sb.

C.

H.

.Ó

The result of this is that the process of titration with tartar emetic, when applied to the class of substances holding this yellow coloring principle, would give too high results. We