The site of the furnace is close to a large yew-tree; it is semi-circular, and the fuel for it was wood. The hot gases and smoke were drawn under the floors of the rooms, which were supported on hypocaustal pillars of burnt clay tiles mortared together, or of stones such as are used in the walls. Brick tiles were used nearest the furnace, because the heat would have cracked the stone. Fragments of circular columns are also used, but these are diverted from their original purpose. The floors of the rooms were of double slabs of stones, cemented together, so as to prevent the smoke from coming up through the floors. The tiles were roughened with a tool usually, to give a grip to the mortar. Accidental marks, made on them when wet, are often seen: the footprints of dogs, dents from the nails of a sandal, thumb-marks, showing the lines of the skin, and the mark of a man's bare foot, showing the great toe. The baths had been cemented all over with pink cement, probably made with brick-dust. The rooms {94} were plastered inside, and the plaster decorated with deep red, terre-verte and yellow ochre. Evidently they had been replastered sometimes over the paint, and coloured again on the top, just as we put on successive wall-papers. The level of the floors had been raised nearly 3 feet since the building was first made. Over its ruined walls there grows the pretty little purple "Erinus Alpinus," which is said by some to have made a mysterious appearance only since the excavations were begun. Did it spring from seed which had long lain dormant, having been originally brought from Spain by one of the Asturians who garrisoned this fort? It is an attractive legend, and I would like to believe it true; but the hard cold fact is that somebody remembers its having been deliberately planted on the ruins after the excavations were made! A very interesting point which must not be missed is the way in which the stone thresholds of the gateways have been worn by the chariot-wheels passing over them. It must have been a dreadful jolt for the occupants to cross these high thresholds when they were new, but Roman soldiers were of course above minding little things like that! The ruts are just over 4 feet 6½ inches apart, exactly the distance of the wheel-marks we see in the streets of Pompeii. {95} The Vallum runs into the circumscribing ditch at Cilurnum; that is to say, the two ditches coalesce at the south of the fort. What has been called the Roman "Villa" at Cilurnum is now definitely recognized as the bath-house provided for the comfort of the troops, or perhaps for the officers only. It is the best-preserved building on the Wall-line, one of its chambers still standing twenty-three courses of stones, or 9½ feet high. The great storehouse at Corstopitum, which might claim to rival it, is 2½ miles south of the Wall.