glass,inch thick, from the rim of a patera; (16) several fragments of the sides and moulded bases of square and round bodied ampulla, of bluish and greenish glass, so often met with and employed as cinerary urns in Roman cemeteries; (16a) a small bit of opaque cobalt blue enamel or frit.

Along with these glass fragments of a number of vessels, which would obviously have formed a notable collection if found complete, there were the following specimens in lead :--(17) a leaden weight, with corroded remains of the staple of an iron ring, marked x by slightly indented lines on the shoulder, and weighing 7lbs. 14 ounces; (18) a leaden disc, weighing 8 ounces; (19) several strips of sheet lead, 3 to 4 inches in length by about inch in width and inch in thickness. (20) In addition to the above there were several fragments of querns, of vesicular lava from the Eifel; (21) a much-corroded second bronze coin of Trajan, (legend undecipherable); (22) a lump of glassy slag or "scum," apparently waste from a broken "pot, composed of whitish opaque glass mixed with clay. (Fig. 2.)

In recording these glass specimens it should be recalled to mind that a melon-shaped ribbed bead, inch in diameter, of cobalt blue translucent glass, along with the unique specimen of embossed Samian, which was figured and described in my preceding report (plate IV, fig. 15, and p. 27, Trans., vol. xiv, n.s.), was found in the tumbled foundations of a house on the east side of the clay floor I.

Fragments of Roman cut glass, of an equally interesting character, were found in the clay floor of the annexe, situated outside the rampart on the west side of the civitas. Those represented on plate X, No. 1, figs. 1, 3, and 7, are portions of crystal glass drinking cups or goblets with upright sides, ornamented by angular cordons and encircling bands of incuse oval facets, such as have been fully illustrated

PLATE XI.

and described by Apsley Pellat in his Curiosities of Glass Making, p. 136, plate III, fig. 3; and by Dr. Anderson and Mr. James Curle, jun., in the Proc. of the Scot. Soc. of Antiquaries, 1896, p. 189, et seq. The pattern on that represented by No. 3, is formed by deeply-cut intersecting grooves, with intervening cones and ridges. One of the facets only is polished, showing that the vessel was rejected owing to a breakage during the operation, and that the process was being performed near the same spot.

OVAL FURNACES OR LEARS.

All three specimens were met with in the soft clay surrounding two large oval furnaces or ovens, which were each enclosed in a separate platform of heaped-up boulder-clay, such as has already been described. The shape of each was likewise preserved and traced by means of their calcined interior lining. They were peculiar in having two openings, a stoke-hole at one end and a horizontal flue or channel at the other, or upon one side. Closely adjoining the stoke-hole of No. I (plate XI, Nos. 1 and 3) there was the base of a small circular furnace for heating a caldron or meltingpot, and in front of both openings an ornamental hearth stamped with twenty-one indented rings 2 inches in diameter. The hearth in front of No. II was carefully laid with stone flags and a Roman brick, cracked and broken by heat, measuring 15 inches by 11 inches by 24 inches. Crossing diagonally underneath the floor and platform surrounding the latter (No. II) was a circular flue, 6 to 7 inches in diameter, resembling a rabbitburrow, which commenced at a small clay furnace or fire-hole beneath the N.E. angle, and ended in two exits on either side of the opposite S.W. angle, at a total distance of at least 32 feet, the flue being still open and blackened with soot internally throughout its entire length. Close to the side of the stoke

hole of the oven II, the flue expanded into a chamber I foot wide, which was covered over with fragments of amphora, carefully set in stiff clay.

The interior dimensions of the two ovens, stokeholes, flues, and furnace were :

OVEN I.-Length 4 feet 6 inches, breadth 2 feet, height of upright sides 10 inches. Stoke-hole-Length 2 feet,

breadth 1 foot 3 inches. Channel or flue-Length 2 feet, breadth 9 inches.

OVEN II.-Length 5 feet, breadth 2 feet 3 inches, height of upright sides 4 inches. Stoke-hole and flue-Length 2 feet, breadth 10 inches.

FURNACE-Length 2 feet 6 inches, breadth 1 foot 2 inches to I foot 4 inches.

They are both represented in plan and section on plate XI, No. I by a photograph. The special purpose of these ovens and the small furnace adjoining No. I is very difficult to define. The position of the hearths inside the area of the clay floors, and the provision of a separate underground flue for heating purposes beneath the floor of No. II, negative the hypothesis that they correspond to the central pit or chamber of a channelled. or composite hypocaust. The most feasible explanation of them appears to be that the small furnace was employed for heating a glass-melting pot, and the adjoining ovens as "lears," into which the newly-fashioned glass vessels, while still hot, were immediately placed, in order to be annealed, i.e., gradually cooled, by being slowly withdrawn from the region of the fire by means of a flat wooden shovel or "peel."

Also, judging from the abundance of glass fragments found upon the site of the workshop on the north side of the fortified area, and the presence of glass-making materials without the usual proportion of potsherds and other ordinary Romano-British remains, it would appear evident that the three pairs of furnaces there situated were similarly employed