opinion, that this is a flue directly connected with the oven previously alluded to.

"I picked up some fragments of tile, and, on cleaning these, subsequently discovered that they were glazed1 on one surface, while some of their edges were shaped or tooled. As I was not in a position to prosecute the work of exploration; and deeming it advisable to avoid interference by irresponsible people, especially as the site adjoins a busy high-road in a populous district, I decided to refill the portion excavated.

"Shortly afterwards I communicated my discovery to Professor Anwyl, of Aberystwyth University College, and he kindly consented to co-operate with me in any further work upon the site. The Professor came over about the end of September, when we again opened the soil at the same place, and he had the opportunity of viewing the hypocaust, and agreed generally with my views concerning its Roman character and construction. Two further fragments of scored tile (marked B and c) were brought to light. We also opened the ground some 10 ft. further away, in order to see if the wall extended from west to east, and found it did so. At this point we found fragments of tiles (marked D, E, F and G) scored with semicircular lines. I also obtained from near the "oven" fragments of flanged tiles, and two pieces of slag of iron, one of which was highly glazed. Lumps of concrete, formed of rough pebbles, and of a much darker colour than the mortar, were also found, together with a large quantity of oyster and other shells.

"Professor Anwyl agreed that it was advisable to restore the ground opened, pending an appeal to the Cambrian Archæological Association for help in prosecuting the work of excavation. We therefore refilled the ground opened, and took no photographs or sketches, as we felt it would be better to delay doing this till we had started our excavations in earnest.

"There is no recognised Roman road nearer this site than that running from Tomen-y-mur, near Trawsfynydd, through Maentwrog and by the pass of Aberglaslyn, in the direction of Segontium (Carnarvon), and this is distant fully five miles away. A hundred years ago the sea covered the land in the immediate vicinity of Glasfryn, and it was not until 1811 that the land was reclaimed by the erection of an embankment or breakwater opposite the port of Portmadoc. It may well have been that this site marked the entrance to the ford or passage-way which, tradition declares, existed across the Traeth Maur (Great Sands) at low water."

1 If the glazing is not an accidental result of heat, these glazed tiles may possibly be of mediaval date.---ED.

1906

15

ROMAN REMAINS AT CARNARVON.

THE extension of the town of Carnarvon threatens to cover the site of the Roman city of Segontium, the walls of which were still visible about fifty years ago. Mr. Llewelyn Lloyd Jones, of Carnarvon, informs us that in excavating for the foundations of two new houses in a field opposite the Vicarage, he found at a depth of 18 ins. below the surface three Roman walls, about 3 ft. 6 ins. wide, running parallel with one another. The position of these walls, and of others which are known to exist, will be carefully marked on a plan, and as building operations extend, any future discoveries will be noted in the

same manner.

Segontium was the starting-point of the Eleventh Iter, the route running from that place to Deva (Chester), by Conovium, which some identify as Caerhun, in the valley of the Conway, and Varae, which has been placed at Bodfari, the pass through the range of mountains which bound the Vale of Clwyd.

A ROMAN BUILDING IN COLCHESTER CASTLE PARK.

THE exceptional dryness of the past summer has been the cause of an interesting discovery at Colchester. In the Castle Park, near the band-stand on the north side of the Castle, certain portions of the grass were particularly affected, and marks appeared which seem to show that beneath are the remains of a large Roman villa, or other important building. Six rooms of varying size are plainly discernible, and there are indications that this does not exhaust the possibilities of the site; the ground is sloping, and probably the foundations extend further to the south, east, and west. The tesselated pavement, which was discovered and preserved in situ when the Castle Park was laid out, is only a few yards away, and possibly belongs to another part of the same building.

The markings have been carefully planned, and an excavation of the whole site will be carried out; but at the present time the work is delayed by the necessity of allowing the grass to recover from the effects of the drought before the turf is removed.

A HOARD OF ROMAN COINS AT COLCHESTER.

In the early part of August a considerable find of Roman coins was made at Colchester, and about fifty of them have been recovered by Dr. Laver and deposited in the Museum, together with the fragments

of the vessel in which they were found. This vase, which has been skilfully restored by Mr. A. G. Wright, is about 6 ins. in height by about 5 ins. at its widest diameter, and its mouth is just wide enough to admit the coins. The coins recovered are all "third brass," of the reigns of Gallienus, Victorinus, and Tetricus.

THE AGE AND USE OF BRITISH STONE-CIRCLES.

AN interesting article on this subject by Sir Norman Lockyer, K.C.B., F.R.S., appeared in The Times of July 30th, 1906. The work done by Sir Norman in Egypt some fifteen years ago proved, as it is claimed, that the Egyptians carefully oriented their temples so that the rising and setting of the stars, and of the sun at certain times of the year, could be watched along the temple axis by the priest in the sanctuary; and the same methods of study have been applied here, the theory being that, from an astronomical point of view, there is the strongest resemblance between the Egyptian temples and British Stone Circles.

"In densely-populated and rich Egypt a temple was devoted to the rising or setting of one heavenly body, whether star or sun, the place of rising or setting being indicated by the long temple axis, and each sacred place contained many such temples, because there were many heavenly bodies to be watched. Now, to carry on this method of observation and worship where the population was scarce, the best and cheapest thing to do would be to build a circle to represent a sanctuary, and from its centre to imitate the various temple axes by sight-lines marked out by a stone or barrow."

An examination of a number of circles in the West of England has led to the conclusion that Arcturus was used as a clock-star, to watch the flow of time during the night, at these circles between B.C. 2330 and B.C. 1420.1

Another purpose was the indication of the rising or setting places of the warning or morning stars, i.e., stars rising or setting "heliacally," or an hour before sunrise, such observations being necessary to enable the priests to know when to prepare for the morning sacrifice at the chief festivals. The Pleiades were observed rising and Antares setting, and the dates given by an examination in connection with these stars are about the same as those found in connection with the "clockstars." Stones, or barrows, are also found, indicating the direction in which sunrise or sunset was to be looked for at the critical times of the year the beginning of May, August, November, and February.

1 The circles referred to are Tregaseal, The Hurlers, Merrivale, Fernworthy, Stanton Drew, and Merry Maidens.

The May sunrise is thus provided for in all the circles surveyed except The Hurlers. Sir Norman summarises his conclusions as follows:

"If we accept the dates thus astronomically revealed, several interesting consequences follow. The British circles were in full work more than a thousand years before the Aryans or Celts came upon the scene, if the time of their arrival favoured by archæologists is anything like correct. Stonehenge began as a May temple a British Memphis-and ended as a Solstitial one, like that of Amen-Ra at Thebes. Another conclusion is that, whatever else went on some four thousand years ago in the British circles, there must have been much astronomical observation and a great deal of preparation for it. Some of the outstanding stones must have been illuminated at night; so that we have not only to consider that the priests and deacons must have had a place to live in, but that a sacred fire must have been kept going perpetually, or that there must have been much dry wood available. The question, then, is raised whether dolmens, chambered barrows, and the like, were not places for the living and not for the dead, and therefore whether the burials found in some do not belong to a later time."

DISCOVERIES AT OLD KILPATRICK, DUMBARTONSHIRE.

THE examination by Mr. Ludovic MacLellan Mann, F.S.A. Scot., of the ancient structure of wood and stone at Old Kilpatrick, on the Clyde, has been carried on from June to September. The area covered by the structure (the precise nature of which is yet by no means clear) has turned out to be much more extensive than was at first conjectured, and indeed its limits on the side farthest from the present river margin have still to be determined. The position of the kitchen midden (and it is possible that one existed in connection with the structure) has unfortunately not being ascertained, Perhaps the most important object yet recovered is the fragment of the rim and upper portion of a vessel of dark-coloured earthenware, possessing features which may be sufficiently distinctive to enable experts to determine its age, and thus to furnish some hint as to the period when the site was occupied. Massive oak logs, some mortised, have been extracted from the foundations. The numerous worked objects in wood and stone, the stones used for polishing, sharpening, and pounding, the remains of animal bones and of cereals and fruit, the plaques of shale worked into shape, and in some cases perforated, and other relics recovered from the site, constitute a mass of material of immense archæological value.

A BERKSHIRE "DENE-HOLE."

ON the brow of a hill in a field on North Heath Farm, about five miles north of Newbury, at an elevation of about 430 ft., one of the farm hands noticed that the top soil of sandy loam had caved in,

disclosing a circular opening of considerable depth. This aperture, about 1 ft. 6 ins. in diameter, was found to lead by a sloping entrance passage, carefully hollowed out of the chalk, to a pear-shaped pit, gradually enlarging from the surface downwards for a distance of some 14 ft., the bulbous end being about 7 ft. in diameter, and the depth about 7 or 8 ft., the roof being carefully rounded off. Unfortunately, owing to the passage and dome-like chamber being partially choked with soil, a careful inspection was not possible, and it is difficult to give a decided opinion as to the purpose of this singular excavation. But the assimilation of its shape and structural affinities to what are known as 66 dene-holes," which, either in isolated pits scattered singly, or in groups, abound in certain parts of the chalky districts of Kent and Essex, generally on high gronnd, and have been found at Perborough and other places in the Newbury neighbourhood, afford some clue to its identification. The excavation of hollows in the chalk for the storing of grain is a method of the greatest antiquity, and is a common practice over the whole world. Beside the distinct assertion of Diodorus that corn was preserved in pits in Britain, there can be little doubt that this method, which Tacitus says was employed by the Germans (the ancestors of the Belgae and other tribes of Northern France), was also used by some of those tribes who migrated to Britain. If the Britons-if such they were-who quarried here did so for the purpose of obtaining chalk, they knowingly and wilfully concentrated their efforts of every kind so as to ensure the least and worst possible return for their labour. The pits probably extend along the whole brow of the hill, as in many other places where they have been discovered. In the course of a cursory examination, several flint-flakes, cores, and a well-worked scraper were picked up, scattered over the surface of the brow of the hill, bearing testimony to the occupation of the site by the flint folk who here pastured their flocks and grew their corn in the Neolithic Age; and within view is an ancient British barrow on Rowbury farm-a survival of the Domesday Hundred of Roebery. Unmistakeable marks were found in the walls and dome of the pit, as were also observed in the famous Hangman's Wood Dene-holes, at Grays, in Essex, showing that these chambers owe their origin to a bronze- or iron-using people.

We are indebted for the above information to Mr. Walter Money, F.S.A. It must, however, be remembered that mined chalk has always been preferred for agricultural purposes to surface chalk, and may have been worth the extra labour necessary to obtain it. It is possible that the truth lies between the two theories, and that the