L’étude d’une zone de production sidérurgique antique située dans la partie orientale de la forêt domaniale de Sillé-le-Guillaume (Sarthe) a amené à reprendre les recherches sur l’établissement rural d’époque romaine le plus proche. Celui-ci, découvert en 1844, n’était connu que par de courts comptes-rendus des dégagements effectués dans la seconde moitié du xixe s. Ces travaux avaient mis au jour une statuette de divinité masculine et un bâtiment maçonné comptant quatre pièces dont une pourvue d’une mosaïque. Une cinquantaine d’inhumations découvertes dans les pièces de l’édifice indiquait une réoccupation funéraire des lieux durant le haut Moyen Âge. Le programme de recherche pluriannuel entamé en 2008 a permis de reconnaître une villa de près de 2,4 ha fondée dans le deuxième tiers du ier s. apr. J.-C. Les sondages menés sur environ les deux tiers des constructions repérées rendent compte du développement et des mutations qu’a connu le site de son origine jusqu’à son abandon définitif dans le courant du xe s.
The study of an ancient iron and steel production area located in the eastern part of the Sillé-le-Guillaume state forest (Sarthe) led to a renewal of research into the nearest rural settlement dated to the Roman period and situated between the localities of Roullée and La Selle in the commune of Mont-Saint-Jean (Sarthe), about 35 km northwest of Le Mans. This settlement was excavated during the summer of 1844 and again in 1873. Knowledge of it was limited to brief accounts of these efforts carried out during the second half of the 19th century. This work led to the discovery of a male divinity statuette and a masonry building with four rooms, within one of which a mosaic was revealed. Approximately fifty burials installed within the levels of the building destroyed by fire indicate a funerary reoccupation of the site, attributed to the Later Roman Empire.Beginning in 2008 and spanning until 2020, a multi-year research programme on the Roullée/La Selle site has resulted in 10 excavation campaigns and 3 geophysical surveys. In total, 58 test trenches were excavated, spread over 3.7 ha, and corresponding to a cumulative area of just over 5,000 m². In addition, 7.4 ha were surveyed using a variety of methods (electrical resistance, magnetic and ground-penetrating radar). This allowed for the identification and understanding of evidence related to ancient occupations spanning an area of 4.5 ha. The remains or traces corresponding to the villa, however, are concentrated within approximately 2.4 ha. The excavations were therefore primarily carried out within the agricultural portion of the site and only represent a fifth of the total settlement area. The project is ongoing and this report presents a mere summary of the main results, some of which remain provisional at this time.During the Early Roman Empire, the Roullée/La Selle site belonged to the city of Aulerci Cenomani, but was located a short distance from the border of the neighbouring civitas, belonging to the Aulerci Diablites. This area thus represents a boundary zone, located more than 15 km from any known settlement and in an area devoid of evidence attesting to other archaeological occupations. The nearest contemporary rural settlements are located more than 6 km away. The villa would nevertheless have been located near the intersection of two roads, one linking Entrammes to Oisseau-le-Petit and the other, Neuvy-en-Champagne to Jublains. The ancient buildings are located at the mouth of a valley that cuts into the northern flank of the Coëvrons syncline, the last eastern foothill of the Armorican Massif, currently covered by the Sillé-le-Guillaume national forest. The thalweg provides access to this massif, the highest points of which loom above the surrounding plateau by more than 120 m. Thus, this zone would have allowed for easy access to the Roman iron ore workshops located in the forest, the closest of which, the Saut du Cerf, is situated less than 1.2 km south of the villa.The ancient rural settlement was founded around AD 30-40, on a plot of land ostensibly untouched by any prior activity. It was composed of three parts: to the south-east a residential building was installed on the highest point, and to the north-east-south-west, two wings of annexes face the valley below. To date, no enclosure device has been identified, but the existence of an installation leaving little or no trace in the ground (such as embankments or hedges) cannot be precluded. Though still very rarely attested to for the western tip of Gallia Lugdunensis, the longitudinal axial plan is characteristic of the villae, but presents in this case the particularity of being installed at the bottom of a valley and not within the plain.The buildings attested to for the first phase have stone foundations (flashings or low walls), but their elevations were most likely made of earth and wood. The initial layout was modified before the end of the 1st century, probably between AD 70 and 90. The two buildings in the western wing were enlarged, a forge was built between the eastern annexes, and the residential area was walled in. The most significant changes, however, occurred between AD 120 and 160. Based on observations of the pars urbana’s western pavilion and using data from the geophysical surveys, it is possible to suppose that the residence buildings were in turn destroyed and then rebuilt in solid form and provided with robust masonry walls, cement floors, painted plasterwork, stucco and at least one mosaic. The stream running through the villa, whose original course is unknown, was then diverted to create a pleasure canal running along the south-western portion of the residential enclosure. Following these endeavors, the eastern wing of the courtyard was demolished and the annexes were moved to the northeast, perhaps to clear the view of the new pars urbana’s northern façade. Two of these buildings were equipped with hydraulic installations: a forge and a possible mill. These facilities presuppose the existence of at least one other diversion from the river or one of its springs. The western buildings were also further extended during this period.Several radiocarbon dates performed on oak charcoal from domestic and artisanal fireplaces belonging to the second phase of the eastern wing of the commons’ courtyard suggest the use of wood felled during the Middle and Late La Tène periods for fuel. These ancient timbers may have come from the dismantling of the annexes belonging to the original manifestation of the villa. This hypothesis implies the reuse, at the time of the settlement foundation, of timbers that were already 100 to 300 years old. This reuse may be related to the opportunistic recovery of a nearby protohistoric occupation or to the transfer of a pre-existing farm. These dates therefore raise questions about the origin of the estate’s establishment and development.Although it is not always possible to reconstruct the exact chronology of the transformations, the development of a monumental dimension to the residence likely continued until the end of the 2nd century. It was mainly expressed through the extension of the pars urbana’s northern pavilions, which were also linked by a gallery measuring over 70 m long.In the agricultural courtyard, the annexes identified are related to cereal farming, with two barns and two possible mills (originally animal-drawn, and subsequently hydraulic). However, the most significant production is iron metallurgy. Two successive workshops attest to forging activity on the site between the year 70 and the second half of the 2nd century. More than 400 kg of waste indicate a sustained production, echoing the ironworks of several thousand cubic metres identified in the nearby forest.This phase of development was followed by a more slow-moving period. Signs of abandonment, or at least a lack of maintenance, of the residence are perceptible from the end of the 3rd century, as implied by the filling in of the pleasure canal. Perhaps because of the lack of a guide fossil, it is difficult to establish the precise nature of the occupation during the 4th century. Coins from the 280s, including some official mints, as well as a nummus of Constantine II, indicate money circulation on the site until the early 4th century. In addition, seeds found in a pit within one of the annexes have been radiocarbon dated to the 3rd-4th century. Thus, it appears likely that there was a continuation of cereal production during this period. The absence of ceramics characteristic of the Late Roman period, derived from Early Christian Atlantic group or Argonne terra sigillata, is noteworthy, though the absence might simply be related to the very partial excavation of the residence.Even if it is difficult to situate the event with chronological precision, it is likely that the destruction of the pars urbana occurred before the 5th century. Indeed, at this time the first burials were installed within the ruins of the northern pavilions. Primarily observed during the 19th century, this necropolis accommodated anywhere between several dozen and several hundred individuals between the 5th and 8th centuries. A round fibula and a back plate from a belt buckle, both discovered out of context, point to the presence of burials within which the deceased would have been dressed.Concurrent with this funerary reoccupation was a change of function for the ancient agricultural annexes. Several hundred shards of culinary ceramics testify to the presence of a population living within the buildings belonging to the former agricultural courtyard. The exact dating of these items is still uncertain, but a bulk of the remains may indicate an attribution around the 6th century.It is not possible to firmly establish the continuity of occupation between the Roman and early medieval periods. However, the maintenance of earthen and wooden buildings for nearly eight centuries suggests that, if there was a hiatus, it was short-lived.The medieval settlement appears to have declined before the year 1000. The final elements attesting to site use are a few 10th-11th century shards recovered within the pavilion during the 19th century excavations, as well as 11th century mule shoes found on a path leading to the early medieval necropolis. The later remains can be linked to the spreading of manure on land plots destined for cultivation from the 14th century onwards. The farms of Roullée and La Selle were not established until the late 18th century.
