numerous. No detailed record was kept, but it seems certain that they included stone foundations, broken masses of building material from hypocausts, a small bronze statuette of Minerva, embossed Samian and other potsherds, and many fragments of inscribed box-tiles, all of one

Fig 29 Plan of Villa at Allen's Farm, Plaxtol

pattern. In 1858 Major Luard carried out a small excavation slightly to the south of the remains just mentioned, in an adjoining ash plantation near the stream, and part of a building, about 60 ft. long, was uncovered. The excavators do not always seem to have understood what they had before them. But their published descriptions suffice to show that the building was a bathhouse. It possessed hypocausts and apses, as bathhouses do; further, it contained a small bath presumably for cold water, 5½ ft. by 9 ft. in extent, paved and faced with tiles laid in concrete, sunk 2 ft. below the level of a neighbouring drain, and perhaps provided with steps; a room containing wood ashes and charcoal, suggestive of heating apparatus, and some masonry at the east end, suggestive of latrines. Its walls were of stone, or, as in the circular part of the southern apse, alternate courses of stone and tile. No minor remains are recorded as found in this building. It lay close to the other foundations, noticed in 1857; the walls of the two were thought to agree in direction, and the whole probably formed one extensive villa. But whether the baths formed a detached bathhouse or a part of the residence we cannot now decide. A well constructed with timberwork, like that of Beakesbourne (p. 145), has been since explored, but yielded nothing beyond black earth. Cemeteries and burials have also been noticed in the vicinity.⁴⁹
   Peculiar interest attaches to the inscribed box-tiles, which illustrate the Romanization in a curious detail.⁵⁰  They were covered all over their four faces with a pattern of irregular lettering in relief, impressed on the clay before firing, and produced with a wooden stamp cut out clumsily by a knife; one can still detect the marks of the wood fibres and the hacking of the knife on the surfaces of the better preserved fragments. The exact shape of the stamp is uncertain. It contained three lines of letters, which are constant in their relative positions, and must therefore have been cut on the same wooden block. But the character of the lines is puzzling. Each line contains one word. The top line (as shown in P1. XXVI, no. 1) has parietalem, the l (third letter from the end) being oddly formed. The second line has Cabriabantu—or perhaps rather nu—in large letters placed reversely to the first line and followed by the semblance of a stop. The third line, also containing large letters, is imperfectly preserved, but seems to end in ricavit; it cannot be restored with certainty, but seems to be the perfect of some verb meaning to make. Fabricavit occurs to the mind, but the first three letters do not resemble fab; possibly the reading is favricavit, a form intermediate between fabricavit and the French forgeat; the rare rubicavit does not suit all the conditions well. The third line is separated from the second by a bar, which does not denote (as one might have expected) the end of the inscription.⁵¹ Strangely enough, these words do not exactly fill each one line. They
   ⁴⁹  Luard, Arch. Cant. ii, i—6; hence C. R. Smith, Coll. Ant. iv, 217; Payne, Coll. Cant. 179 for the well; remains in Maidstone Museum. See also Topographical Index.
   ⁵⁰  I (F. H.) have taken my readings (P1. XVI, no. 1) from seven fragments in Maidstone Museum, very kindly lent me for examination, compared with plate 6 in Arch. Cant. ii. Previous writers read only one line Cabriabanti (Luard, Arch. Cant. ii, 4.; hence C. R. Smith, Roman London, 115, misquoting by memory; and Corpus Inscr. Lat. vii, 1238). Liverpool museum has a fragment (Mayer Coll. 6134.) acquired by Mayer from the representatives of Bryan Faussett, the Kentish antiquary, which bears a label in Faussett’s handwriting stating that it was found about 1773 in London, near Bishopsgate, ‘on the site of the new Excise Office, once Gresham College,’ in Broad Street. But this London provenance may be an error. Any Kentish antiquity might easily have found its way into Faussett’s hands, and it is unlikely that so peculiar a tile should occur once in London, often at Plaxtol, and nowhere else. We may suspect therefore that the specimen was found at Plaxtol in the 18th century and its provenance misunderstood.
   ⁵¹  The line or bar between CABRIABANV. and the illegible word next to it does not denote that the inscription ended here. On P1. XVI it is clear the top of one stamping has been erased by the bottom of another. The stamp placed second on the tile had at top parietalem : this has obscured, and in turn been obscured, by an existing  - continued