ANTAEUS 26 (2003) 69-84 THE ROMAN WELL IN THE PRINCIPIA AT MATRICA Péter Kovács The excavation of the Roman fort at Százhalombatta began in 19931 (fig. 1). During excavations part of the north-western corner towers, the porta praetoria and the porta principalis sinistra were unearthed. In 1995-97 we unearthed the area of the headquarters - building in trenches XVIII, XX-XXVI, XXIX-XLI (fig. 2). The well from the stone construction period of the principia was found in trench XVIII (fig. 3). This trench was our first test trench in the area of the principia. The well itself was found in 1995 but excavation of it could be finished only in 1996. In this paper I wish to make known the results of the excavation of the well. The area of the fort lies on the watershed of the river and not on the loess plateau. The distance of the well from the Danube is ca. 140 m. The level of the fort is on average 7-8 m higher (101m above sea level compared to the level of the Danube (on average 93.5 m above the sea level), therefore the high water levels in the well caused several problems during our excavations. Only a high capacity water pump (380 V, 22 mYhour) could empty the water from the well. The other great problem was that we could not cut the well in two because of its depth so that a round safety cage became necessary. This cage was intended to prevent the well from collapsing. The narrowness of the well shaft also made work more difficult. The well was found along the transversal axis of the courtyard of the stone construction period HQ-building. Both the width and the length of the courtyard measured 16.5 m. It had a portico and was surrounded by the side-wings of the building incorporating: the armamentaria, the basilica and the entrance-hall (later added). The courtyard opened in the direction of the cross-hall. The whole courtyard was covered by flagstones. The diameter of the well was 0.95 m in its upper levels (fig. 4). Its construction was contemporary with the flagstones in the courtyard, because one of these stones was laid into the upper course of the well. The flagstones of the courtyard, laid side by side, surrounded the upper structure of the well. The well was built of approximately 15-20 x 30-40 cm regular Limestone ashlars. The inner parts of the courses were of dry stone. The outer parts were most likely set in waterproof clay. The upper level of the well lies 100.54 m above sea level. The topmost course in the upper structure (especially the outer course of the well) were laid with mortar. In the section of trench XVIII looking north we could observe the backfilled pit of the well. The fill consists of clay. Fragments of a jug in yellow fabric were found in its fill. Seventy cm below the topmost course was observed a horizontally laid larger stone. It may have been used during the building of the well. The upper structure of the well was pulled down to the level of the court although it was still possible to observe all three of its levels. The diameter of the whole well was 2.7-2.8 m. The well- shaft was 90-95 cm wide. The 0.95 m wide well became widened slightly toward the bottom (its largest diameter was 1.2 m) although in the end it narrowed again (0.8 m). The bottom of the well was found at a depth of 7.1 m. The stones of the upper courses were thrown into the well. It seems that the well was completely filled with ashlars and fragments of stone carvings from the late Roman period and we could find nothing that could be dated later (not even a shard from a vessel). Stone carvings were found among the ashlars of the stone monuments in the principia (altar-, column-, basis fragments) (figs 5.1-2). The carcass of a small dog came to light among the debris. The fill of stone and bricks ended at a depth of 5.5 m. A layer of mud 1.6 m deep was unearthed under the debris layer. It contained finds and stones as well. We reached the present-day water level at a depth of ca. 3.7—4.5 m, changing according to the water level in the Danube. The concave bottom of the well narrowed and ended in large (40-50 cm wide) slabs laid horizontally side by side supporting the whole well (figs 6-7). In the middle of the bottom there1 1 Kovács 1994-95; Kovács 1995a; Kovács 1995b; Kovács 1996; Kovács 1999.