HE village of Box, which is about five and a half miles east of Bath, just within the Wiltshire border, was in Roman times about two miles distant from either of the two great roads leading from Bath in an easterly and northerly direction to Silchester and Cirencester respectively.

A writer in the Gentleman's Magazine of 1831 (part I., 595-6), in a letter upon the topography of Box, says :

"the tradition of the place was, that there had been formerly found baths which had been considered to have been Roman,"

and :

"that several beautiful tesselated pavements had formerly been found in the churchyard and gardens adjoining, but no spot could be pointed out where the same might with certainty be found,"

but the writer continues:

"a year or two ago, in a garden belonging to Mr. Mullins, adjoining the churchyard, in making some additions to a very old building,3 the workmen sinking for a foundation struck upon the mutilated remains of a tesselated pavement about two or three feet below the surface of the ground. It appeared to have been part of a large square, and the part now discovered was evidently

For a large part of the expense of the illustrations accompanying this paper the Society is indebted to the kindness of Mr. W. Heward Bell. For one of the photos reproduced we are indebted to Miss Coote, of Bath; for another to Mr. Hardy, of Box; and for the remainder to Mr. Sidney Brakspear. The drawing of the capital was kindly made by Mr. E. Cook, of Devizes, and all the plans and drawings of pavements, &c., by Mr. H. Brakspear. The British Archeological Association has kindly allowed us to make use of the description and plan of the portions excavated in 1881, which were given in their Journal.--[ED.]

2 He occupied the house now called " The Wilderness," and rented the land whereon both the excavations of 1881 and 1902 took place.

3 This was certainly the "Old Parsonage House," that stood until 1853 upon the north-west angle of plot D (plan of site), and of which were found certain foundations, in part made of hypocaust pile laid on their sides.

one of its corners.

It had a wide ornamental border of no remarkable beauty, but what I particularly recollect (the few moments I had opportunity of seeing it) was that there were evident effects of repeated fires having been made apparently about the middle of the square; for the tessera toward the centre were burnt from their original colours to a brick red, and the redness diminished in intensity as it approached the border, near which the colours were again all perfect. This pavement must have been discovered when the old building was erected, for it appeared to run under its foundation, and if so the remainder must have been then destroyed. The portion of it lately found, however, was considered worthy of preservation, and has been, it is hoped, safely secured from injury by means of large flag-stones carefully placed over it."

This letter was followed in 1833 by a further notice in the same magazine (p. 357) from the hand of the Rev. George Mullins, Rector of Ditteridge, who lived at Box, in the house now called "The Wilderness"

'In the supplement to the first part of your vol. ci., p. 596, a correspondent in speaking of the antiquities of the village of Box, mentions the existence of some Roman remains in my garden. The discovery which he alludes to was that of a tesselated pavement found there a few years ago. I believe that until that discovery no site of Roman remains could be pointed out in the village, although tradition spoke of their existence. I have now to announce a further and more important discovery, which has taken place within the present year. In excavating some earth at the distance of forty-three yards north from the pavement before alluded to, evident traces of another, but in a state of destruction, presented themselves; and in a line leading from this immediately to the former I found seven stone pillars, of rough workmanship, and near them an altar-like erection, consisting of several stones, and a piece of stone of a semi-circular shape, about a foot across and eight inches thick, partially excavated on each side as if for the purpose of holding something. This stone bears evident traces of fire. Distant twenty-eight yards to the west are the mutilated remains of a tesselated pavement of blue stones, ornamented with two red borders, the tesseræ nearly an inch square, and the blue stones entirely decomposed. This pavement in its original state must have been at least 10 or 12 feet square. The remains of a wall on the south side were covered with several flues, made of whole bricks, supported by iron cramps; and underneath the bed on which the pavement was laid, made of coarse gravel and mortar, were large flags, supported by pillars of stone forming a regular hypocaust. The discovery of this pavement induced me to make an opening at a point where I might conjecture from the direction of the hypocaust where remains would be found ; at a depth of 4 feet below the surface of the earth I discovered a third tesselated pavement, very nearly perfect, apparently forming a passage from some other part of the building. It is 9 feet wide and 28 feet long, and turns at a right angle 6 feet, where it is broken up; but it evidently extended much further in both directions. In one corner is a curious stone cut in

form as though intended for a seat, but now only 8 inches high. The pattern of this passage is particularly elegant. The ground colour is white; the exterior tessera coarse; the blue lines of smaller cubes form the borders, 6 feet about, and the intermediate space is filled with semicircles, forming waving lines, blue, crossing each other at right angles; these are again intersected by others of half the diameter, with their extremities united in the centre, and terminating in small crosses shaded with red and yellow, white and blue, and producing the most beautiful effect. I have preserved several specimens of the plaster from the walls, the colours of which are very bright when first exposed to the air. The patterns were principally imitations of Egyptian marble, with elegant coloured borderings, but I have not been able to discover a decided figure on any of the pieces. One small vase holding about a pint, apparently of British workmanship, was taken up near the pillars, but unfortunately damaged by the spade. Numberless fragments of earthen vessels, small pieces of pavement, tessera, and Roman bricks, are now in my possession. One room was evidently paved with square red bricks, quite plain. One small coin only was found, and this was so far corroded as to be wholly illegible. There is another piece of pavement (which, however, I have never seen), in a distant part of my garden; and the whole of the original buildings, if square, must have covered a considerable portion of ground, the most distant of the pavements being at least fifty yards apart."

Although no plan accompanied this letter the late excavations have made it quite easy to identify the various pavements described, which will be referred to later when dealing with the respective chambers to which they belonged.

The next notice of the villa, so far as is known, occurs in the British Archæological Journal for 1860 (xvi., 340), in a short article by H. Syer Cuming, quoting a letter received from a friend of his who occupied a house on the site (probably that formerly occupied by Mr. Mullins,) who says:—

"My garden is full of Roman remains―tiles somewhat ornamented, but broken, bricks, tesselated pavement, fused iron, &c. I send all away to mend the roads; they are a perfect nuisance. We cannot put a spade into the ground without bringing up these impediments to vegetable growth. There is a bath quite perfect, in the centre of the garden. It has been opened, but is covered up; and a beautiful pavement runs all about. The bits I dug up were white and black, very coarse work."

Mr. Cuming goes on to say :

"I begged my friend to send me up some of the remains from his garden, which he has kindly done, and a selection is now before you, consisting of tesselæ of rather rough-hewn dice of white limestone, and portions of four

square flue pipes of red terra-cotta from the hypocaust, which are useful for comparison with examples found in London and elsewhere. The majority of the pipes met with at Box are scored with the common diamond pattern, a few have only bands of parallel lines, whilst one fragment is decorated with a succession of waved or undulated furrows of more novel design. From more careful examination of these air conduits I am inclined to regard them as the products of different kilns, for they vary both in hardness and colour and above all in the character of the material, which must have been obtained from different localities. The only example of stucco which has reached me is a fragment of the fine variety called albarium; but much of the walling has been painted in fresco, in imitation of African marbles, with elegant coloured borderings." 1

In his Aqua Solis, published 1864, Preb. Scarth, in his map marks the sites of seventeen or eighteen Roman villas existing within a radius of seven miles round Bath, of which this at Box is one, another at Ditteridge is scarcely more than a mile away, and a third at Colerne not more than two miles. Of the Box villa he says:-"The tesselated floors of three rooms were found here quite perfect, but the patterns are plain and the work coarsely executed. The most interesting portion is the remains of the bath; the sides and circular end of which were covered with tessera of white lias. Careful drawings have, I understand, been made of these pavements, which were situated in gardens in the middle of the village. The remains of a hypocaust have also been found, with several pillars entire, and a Roman bath is also stated to have been found on the south side of the churchyard."

use.

On page 127 he remarks:-"There are certain particulars of these villas (round Bath) which are worthy of notice. The regularity of their form. They were either built round a court, and formed three sides of a square; or else ran in a straight line, often with a projecting portion at right angles to the main body of the building. They were all provided with a hypocaust and baths, and had tesselated pavements of elegant workmanship. They were accompanied with outbuildings, and situated in an area of some extent enclosed by a boundary wall, within which were interments of two kinds, viz., cremation and inhumation. The villas were supplied with earthenware utensils of every description, and with glass, both for windows and domestic Coins are found in the greatest abundance and to the latest period of the Roman occupation. The situations are well-chosen, and the villas are for the most part represented at the present day by elegant modern country houses in the same localities and near the same sites. They were always well supplied with water, and the wells were of excellent construction. The villas round Bath do not seem to have equalled in dimensions those laid open in other parts of England, as at Woodchester or Bignor, nor the elegant remains which exist at Lydney, in Gloucestershire . . . The superstructure of these villas is a subject which has caused much perplexity; and antiquarians are not decided as to whether the upper portions were constructed of stone or wood. I am inclined to think that wood must have furnished the materials of the upper portions, and that the stone walls were only carried to a certain

In the British Archæological Journal for 1887 (xliii., 47–55) is a very interesting article by Mr. R. Mann, in which these notices are all quoted, dealing with a portion of the villa excavated in 1881 and accompanied by a plan.

This portion was situated to the south of that recently uncovered, and has been since partly built over. Mr. Mann's description of the remains will be referred to later on.

Some Roman remains are said to have been found under the porch of "The Wilderness" about 1888, but no record was kept of their character.

In 1895 the small piece of pavement under the lawn of the same house was discovered.

In 1897 Mr. Hardy, provision merchant, of Box, purchased the walled garden (plot A, plan of site), with the narrow strip of land that gives access to it from the road on the south. This land was part of that occupied by Mr. Mullins and referred to in his letter quoted above. The tradition of the Roman finds on the site caused Mr. Hardy to begin excavating, with the result that the greater part of the buildings in his garden were uncovered.

For various reasons it was impossible to make a thorough examination and survey of the site at that time, much as Mr. Hardy wished it, so that the whole was filled in again and the matter remained in abeyance till the summer of last year (1902). Mr. W. Heward Bell, of Seend, then most generously offered to bear the expense incurred in having the whole excavated and planned. The work was commenced in August, 1902, and continued until the end of January, 1903, and it is only right to record that it was to the untiring energy of Mr. Hardy, in constantly superintending

height above the ground-floors. The remains seem to indicate that they were hastily plundered and then set fire to, and that the roofs and timbers fell in upon the floors, which are often found indented and covered with burnt matter and roofing tiles. After remaining in this condition, it may be, for centuries, the portions of the walls still standing were afterwards used as quarries, when stone was needed for other buildings or to make enclosures. The Saxon population left them in ruins; the Norman and mediæval inhabitants used them as materials, and thus little is left to our time except the foundations, and that which lies buried under their debris.”