Most of the fossils alluded to in the sequel of this paper were found in the variegated shales of the Onondaga group-an horizon below those which have yielded the oldest previously known fish remains in America and at least as old as the lowest fish-bearing rocks elsewhere. The exact stratum is a bed which I have named the Bloomfield Sandstone from its exposure near the town of New Bloomfield in Perry County and which lies at the top of the variegated shale of the Onondaga about 200 feet below the water lime. The fossils themselves are of two kinds. First, or least abundant, are small spines apparently of Selachian fish much resembling those described from the Ludlow rocks in England. These by their form and appearance admit of no doubt concerning their nature. Second, are a number of flattish, elongated, striated shields whose appearance instantly suggests a relationship to the wellknown Pteraspis of the same rocks. Nor is this appearance delusive. Closer examination confirms the first impression in every point and leaves no room for doubt that these Onondaga shales, so barren of animal life, contain older vertebrate fossils than any previously described from this county.

To confirm these conclusions, microscopic evidence can be cited, if necessary. Thin sections cut and mounted in Canada balsam show, not indeed the ordinary structure of bone, which is totally wanting, but a structure so closely resembling that of the Ludlow Pteraspis that the near affinity of the two is placed beyond question.

Sufficient differences, however, exist both in the outer shape and the minute structure, as shown by my own specimens of both the English and American fossils and by a comparison of the latter with Prof. Huxley's figures as given in the Quarterly Journal of the Geological Society (August, 1858), to enable me to distinguish between them, both as genera and species, and I propose for the fossils now brought forward the generic name Palæaspis and in this new genus I place the two following species: Palæaspis elliptica, Palwaspis bitruncata.

For the species previously mentioned I propose the name Onchus Pennsylvanicus.

It is not possible in an abstract like this to enter into minute details of description and structure. These must be reserved for another place.

But in regard to the age of these fossils compared with that of

the Onchus and Pteraspis of the English beds, I may say that in the Table given above the Ludlow beds are palæontologically considered the equivalent of the American Lower Helderberg and Waterlime and that in consequence the fossils they contain must be stratigraphically younger than those which I here bring forward from strata at least two hundred feet lower down. Only a single specimen has yet been found from the Lower Ludlow in England. All the rest came from the Upper Ludlow about 1000 feet higher up.

Shortly after the discovery of these fossils, when working in the Red Shale of the same county and about a thousand feet below the Bloomfield sandstone, I met with a thin bed consisting in great part of what appeared to be the comminuted scales of fish. Their broken and fragmentary condition has thus far prevented my making a successful study of these, but I cannot from their general appearance and structure entertain much doubt that they are also fish remains and I strongly incline to the view that they are coprolitic in origin. This carries the existence of fish down nearly to the base of the Onondaga group.

Further, on the same day I discovered another layer crowded with fragments presenting an appearance very similar to that just described, but at least 500-600 feet lower still. Though in broken condition and from the refractory nature of the matrix exceedingly difficult to extract or examine, yet their appearance is enough to satisfy me that these also are comminuted fish plates or scales bearing in many cases the characteristic sculpture on their outer faces. There are also with them a few imperfectly preserved spines similar to those found in the Bloomfield Sandstone. Coming as they do from the Iron Sandstone they carry, as will be seen by reference to the Table, the existence of fish (and therefore of Vertebrates?) back nearly to the middle of the Clinton Group and consequently to an horizon considerably lower and older than any that have hitherto yielded fish remains in America or elsewhere.1

Into the zoological affinity of Palæaspis this is not the place or time to enter. Its resemblance to the English Pteraspis is so close that the two must stand or fall as fish together. Regarding

1 It is perhaps right to add that in thus speaking I exclude the much disputed "Conodonts" of Prof. Pander from the Lower Silurian rocks of St. Petersburg. Should these bodies prove the remains of fish they are of yet older date.

the latter, Huxley says (Q. J. G.. S, Aug., 1858): "No one can, I think, hesitate in placing Pteraspis among fishes." In this preliminary notice I claim no more than the admission of Palæaspis to the standing which has been accorded to Pteraspis. I shall hope on some future occasion here or else where to state some of the details at greater length.2

GEOGRAPHIC CLASSIFICATION, ILLUSTRATED BY A STUDY OF PLAINS, PLATEAUS AND THEIR DERIVATIVES. By W. M. DAVIS, Harvard University, Cambridge, Mass.

[ABSTRACT.1

THE relations of a number of geographic forms may be exhibited by means of a classification based first on the peculiarities of structure and second on the amount of erosion that they have suffered. Conditions of accumulation, elevation and deformation give subheadings under the first class, while the stage of advance in the destructive processes of erosion that cut all land forms alike down to a base-level plain of denudation provides further opportunity for subdivision. In no group of forms does this method of classification find better illustration than in those characterized by a structure of horizontal layers.

This great group may be first divided and subdivided according to special conditions of formation and composition; thus we have (a) Strata of fragmental or calcareous materials of marine, lacustrine, fluviatile or æolean origin, all of common occurrence.

(b) Sheets of overflowing lava, greatly expanded in Idaho and India, less in France and Germany.

(c) Ice and snow plains, still existing in Greenland, and of special interest there as representatives of the vast snowy desert that spread in relatively recent times over the lands about the North Atlantic.

(d) Salt plains, deposited by evaporation of inland lakes.

(e) Peat and other vegetable deposits, accumulated by organic growth. These last two divisions include deposits generally of relatively small area and short duration, which gain a long life only when preserved by burial.

For further details on this subject see the Q. G. J. S., for February, 1885.

The means of revelation or birth of such of these deposits as are formed beneath water affords further ground for their classification. The greatest number have been (a) directly lifted out of water by an upheaving force of unknown origin that acts vertically without any superficial appearance of lateral crushing, as in the Colorado and other plateaus. Corresponding closely with this, and perhaps in some regions of moderate elevation confounded with it, is (b) the laying dry of shallow sea-bottoms by the deepening of the ocean elsewhere. The sheet of water has been withdrawn from lake deposits (c) by evaporation, (d) by cutting down outlet, probably in some cases (e) by tilting the basin so as to allow better drainage, and (f) by melting away of barrier, as suggested in the case of Lake Agassiz and very many small glacial lakes. The water has been (g) displaced by delta and littoral deposits.

But of greater geographical importance than these early and constructional characteristics, are the later, destructional ones, determined by erosion, inasmuch as they comprehend the topographical form that we actually observe. It is this division of the subject that is worthy of most attention from the comparative geographer, and in assigning a region of horizontal structure to its proper place in this classification, the following sketch of the lifehistory of such a land mass may serve as a guide. Just as the surface of the deposit rises above its base-level of erosion-—or, in the case of lake deposits, as the governing base-level sinks below the surface, a smooth, unbroken plain is revealed. Its drainage is imperfect, for the newly formed streams have not yet had time to establish their channels, and much of the rainfall stands on the ground until it evaporates. Faint depressions on the surface, the effect of slight irregularity in the original distribution of the deposit, hold the water back to form shallow marshy lakes. The smoothness of the surface and the shallow lakes are indeed truly infantile features, retained only during the earliest life of the plain, and soon lost in its further development. This development finds its opportunity in the gaining of an effective elevation above the base-level of the region and in the persistent down-cutting of the streams. Rivers establish their courses, the smooth plain is trenched across by their meandering channels, and all the lakes disappear. This is adolescence. The channels will be narrow and steep walled in regions of relatively rapid elevation, but broadly open in regions that have risen slowly, and I believe that

rate of elevation is thus of greater importance than climatic conditions in giving the cañon form to a valley. Of course the channels must be shallow in plains of small altitude, while they may become deep in regions that have been raised to a great height; but great cañons can occur only in young plateaus, for cañons are marks of a precocious adolescence, and in this feature, the Colorado plateaus are truly American. Adolescence as thus defined includes that part of a plateau's early life in which the stream channels are narrow and well marked; but as the valleys increase in number and open widely so as to consume a good share (one-sixth to onefifth) of the plateau mass measured above the existing base-level of discharge, then adolescence merges into maturity. Maturity may be said to last through the period of greatest diversity of form, or maximum topographic differentiation, until about threequarters of the original mass are carried away; and through this period when the geographic names for this group of forms should be most applicable, there are none whatever that can be properly applied, nor is there any that includes this group of forms as a whole.

During maturity no vestige of plain surface remains (except in slow-rising, low plains that never acquire any marked relief), and the region is too broken to be named a plateau. To call it a hilly or mountainous country resting on a plateau would be an unappreciative, indeed a vicious, style of nomenclature. Indeed there is no suitable technical name yet proposed for such a surface, although it possesses a most peculiar and well-defined topography. Long lines of cliffs are the most pronounced features; their upper edge is of constant altitude, but their direction is exceedingly irregular, continually varying as they pass from reëntrant to projecting angles. When standing on the bench or platform of such a cliff, its extension can be traced for miles, contouring around the spurs that separate ravines and valleys. Every cliff marks the outcropping of a more resistant stratum or group of strata than those that over and underlie it, and the natural variety in the sequence of harder and softer layers gives origin to the many variations on a single type that are observed in the scenery of broken plateaus. The cliffs are separated by sloping surfaces, that in the younger plateaus are steep and largely covered with debris from the cliff above, forming a talus almost down to the cliff below; but with the recession of the cliffs, the talus proper is restricted