56. Quatre communes du Morbihan : Carentoir
Transcription
56. Quatre communes du Morbihan : Carentoir
EAST BRITTANY SURVEY: QUST-VILAINE UATERSHED REPORT ON FIELDWORK IN MARCH-APRIL 1985 The fourth season in a programme of fieldwork, itself part of a larger, multi-discipl inary study of the rel ati onshi p between land-use and seulement during the last two thousand years, took place between 21 March and 13 April in the communes of Ruffiac, Tréal , St-Nicol as-du-Tertre and Carentoir in the Morbihan in eastern Brittany. The aim of the study is to détermine when, how and why the exploitation of the environment changed direction within the historic period, and the effects of such changes on social groupings and labour patterns. The complète fieldwork programme consists of systematic fieldwalking of ail available ploughed fields within the four communes, together with sélective geophysical and geochemical surveys, and small excavation; in addition, environmental analysis and a survey of standing buildings is being undertaken. The larger study involves (amongst other éléments) analysis of documents, including the very detailed cadastral maps and records of the early nineteenth century (Asti 1 1 and Davies 1982a, 1982b, 1984b). The latter analysis has been completed and, as demonstrated below, is of particular significance for fieldwork. The 1985 season involved fieldwalking over large areas at wide intervais, as in 1982 - 84; fieldwalking four selected areas in 5m squares for 'total' surface collection; phosphate analysis and soil magnetic susceptibil ity survey of the four selected areas; and excavation of part of a bank and lynchet near a field from which 'total' collection had previously been made. Three days (21-23 March) were spent in préparation by three people; the main team (consisting largely of past and présent students from the Universities of London and Reading) numbered twenty-two, including the directors; it worked for twel ve days, from 24 March, and had one day off; nine people (including one director) remained for an additional week to concentrate upon the excavation. Overall , 434 mandays were spent on this year's season, including travelling time and time off; 364 of thèse were working days. Two of the main team spent most of their time assisting in a survey of standing buildings already in hand when the season started, and one spent three days taking soil samples for analysis at the Institut National Agronomique Pari s-Grignon . The weather varied between bad and appalling for much of the three weeks, with plenty of rainfall , but the crop was not high and fields were often in idéal condition for walking. The season was therefore especially productive. Fieldwalking in Runs at 50m intervais (Transect Walking) Fieldwalking over large areas was organised within transects running south/north across the communes. L (in Ruffiac), the only transect not yet walked, was completed and two adjacent transects in Carentoir (F and C), which had been inadequately covered in 1983 because of the height of the crop, were rewalked; F included the cadastral and modem village of Carentoir (see fig. 1). Except for a small area in the north of C, and for fields previously covered, ail fields under plough and with young crop within the three transects were walked at 50m 1 intervais, using collection units of 100m; field conditions, features, présence of schi ste and local pronunci ations were noted on standardised recording forms. 463 fields were thus covered, encompassing 772 hectares (4.01% of the surface area of the four communes). 30.99kg of pottery and 93.92kg of man-made building material were recovered from the transects; 45% of the pottery was médiéval, 53.6% post-medieval and 1.4% Roman. No pre-Roman pottery was found, but sixteen worked flints were recovered, three from transect L, three from C and ten from F, and also a stone axe, the stone of which has yet to be identified (F117). Two possible areas of ridge and furrow were noted, along with fourteen lynchets and eight (mostly substantial) old banks. As in previous years there were considérable variations in the concentration of recovered material, and the same conventions are hereby used to distinguish between them: fields in which more than two neighbouring units each produced five or more sherds of the same broad period (or five or more fragments of building material) have been termed 'sites 1 ; fields in which one unit produced five or more sherds or fragments of building material, and two or more neighbouring units produced one to four, or two neighbouring units each produced five or more sherds or fragments, have been termed 'probable sites'; fields in which there were irregular concentrations of material not covered by the above catégories - for example, one unit with five or more sherds of the same period - have been termed 'possible sites'. (The minimum number of finds necessary to qualify a concentration for comment has been de! iberately fixed at a low level in order to secure a wide range of possibil ities for testing; in fact, in many cases numbers were far higher than the necessary minimum. It should be stressed, yet again, that the terms are conventions for distinguishing between greater and lesser concentrations and for providing a means of référence to them; they do not necessarily dénote the position of former settlements. ) In accordance with the conventions, 187) may be classified as follows: the concentrations (total 'Sites' 'Probable' 'Possible' L 4 (2.3%) 26 (14.9%) 34 (19.5%) F 5 (3.2%) 22 (14%) 40 (25.5%) C 9 (6.8%) 21 (15.9%) 26 (19.7%) Total s 18 (3.9%) 69 (14.9%) 100 (21.6%) Ail types 64 (36.8%) 67 (42.7%) 56 187 (40.4%) (42.4%) Fields 174 157 132 463 wal ked (Percentages of fields walked per transect and in toto.) Of thèse concentrations, none produced predominantly Roman material although 4.8% produced s orne Roman material; 27.3% had predominantly médiéval, 38.5% predominantly post-medieval, and 8.6% more than the necessary minimum proportions of médiéval and post-medieval sherds. The remaining concentrations (25.7%) had a prédominance of building material, at présent treated as undatable. 2 As usual , the topographical position of the sites was analysed. About a third of them (29.4%) were on flat 1 and while 17.1% were on south-facing and 16.6% on south-east-f acing slopes. Upland concentrations were less notable than in 1984 (transects H, J, K) with only a quarter lying between the 50m and 75m contours; more than half 1 ay between 25m and 50m (59.9%), as in the other Carentoir transects. Only 35.3% of sites 1 ay within 250m of streams that are mapped, far less than in 1984, while a further 39% were up to 500m, and 25.6% more than 500m, away; it was largely sites in F that caused this anomaly. Comparison was systematical ly made with the early nineteenth-century pattern of land-use and settlement, as evidenced by the ancien cadastre, as usual. More than three-quarters of the concentrations (80.2%) lay more than 100m away from early nineteenth-century settlements and only 7.5% lay within 50m of them. (This is not signif icantly différent from distances from modem settlements). Surface material tends to be found within areas of nineteenth-century arable, as always - especially in bandes (literally 'bands 1 , arable divided into tenant parcels) - and the proportion was comparable with that of 1984 (70.1% of concentrations in 1985, as compared with 66.7%). The remainder occurred in nineteenth-century meadow and/or pasture (3.8%), marginal 1 ande (unculti vated land, not fallow - a high 8.6%), woodl and (0.5%), curtilage and areas of mixed land-use; ail but the latter suggest some measurable change in land-use by indicating pre-nineteenth-century arable or settlement. In some parts cadastral land-use, naming and road patterns themselves indicate former settlement sites, especially where very small fields of very mi scell aneous^ land-use are arranged in relation to tracks, like the curti 1 agesand clos that surround settlements both now and in the early 'nineteenth century (Astill and Davies 1982b: 21f, 31). Thèse cadastral suggestions coincided with fields that produced concentrations of surface material at the post-medieval 'site' C470 and, more arguably, the post-medieval 'probable site 1 C473, although no buildings were indicated there in the nineteenth century. One concentration was located in an area which had standing, inhabited, buildings in the early nineteenth century, but which is now devoid of structures or earthworks: F212, a 'possible site' (médiéval and post-medieval ) . Overall, concentrations of surface material were most frequently found in the basin to the south of Ruffiac village (the centre of L) and in the nei ghbourhood of Trignac, in the north of Carentoir commune (F); however, sites were common throughout the southern halves of F and C, particularly in the Coet Morel /Hôtel Orl and area (Carentoir). Surface material is markedly absent from fields on the northern and southern boundaries of L, on the east/west ridges (particularly on a band north west of Carentoir) and - as might be expected - on most steep slopes. There are again, then, di stincti vely 'blank' zones within the four communes and by contrast zones that tend to produce greater or smaller surface scatters. Either natural schiste or conglomerate are usually présent on the surface of fields in ail three transects, as also is imported 'roofing' si ate or schi ste; the imported material is notably absent from fields cleared of woodl and since the 1 ate nineteenth century. Fieldwalking in 5m squares ('Total' Coll ecti on ) In order to investigate the nature of sites identified in transect walking, as in 1983 and 1984 some fields were the subject of more intensive study: four fields (A107, D221, B216, B347) were gridded in 5m squares so that everything on the surface of the fields, including schiste, might be collected. A107 had been classified as a 'possible site', with comparable amounts of médiéval and post-medieval pottery; 0221 had been classified as a 'possible site', with a prépondérance of médiéval pottery in an assemblage that included Roman wares; B216 was a 'blank' field which had produced no finds; B347 was classified as a 'site 1 , with Roman pottery predominating. Soil samples for future phosphate analysis were taken from the topsoil in ail squares, at 5m intervais, and soil magnetic suscepti bi 1 i ty readings were also taken at 5m intervais, at the spot from which samples had been lifted. Fluxgate gradiometer readings were not taken since 1984 tests suggested that thèse were only useful if taken at much narrower intervais. Schiste was collected, totally from A107, D221 and B216, and in a limited sample from B347 (from one square in every nine). This was subsequently classified in three colour catégories (black/grey, green/grey and red/yellow) and three sizes (<0.5cm, 0.5-1. 5cm and >1.5cm) in an attempt to find criteria for distinguishing between local and imported material. A107 lies just below a flat, exposed hilltop on a south-facing slope 75-80m high. The area was arable when the cadastral survey was recorded, the field being the same size and shape as at présent but divided into bandes; an area of 1 ande lay to the north and the nearest settlement (La Boulardaie) lay 130m away. Pottery, building material and schi ste were collected from an area of 1 hectare, distributed ail over the field, with no obvious clustering; magnetic susceptibil ity readings did not vary very much. Quantities of pottery, and of brick and tile especially, were low: 0.41 sherds of médiéval pottery per square (2.72g), 0.55 sherds of post-medieval pottery per square (4.4g) and 0.29 fragments of brick and tile (6.58g). The assemblage included five worked flints, one pièce of haematite and two pièces of tegul a. Although black/grey and grey/green schiste were présent ail over the field, and looked like natural , red/yellow schi ste was distributed in a pattern similar to médiéval and modem pottery and could have been introduced on to the surface. The small quantities of archaeol ogical material recovered, and its gênerai distribution, suggest that it was brought on to the field in the course of manuring; quantities, pattern of distribution and topographie position are similar to those of B319, walked in 1983 and also interpreted as a manuring scatter (Astill and Davies 1984a: 20). Six fields in the near vicinity were also walked at 50m intervais; thèse produced some material but no notable concentrations. D221 lies on the 45m contour in a flat area. In the early nineteenth century it 1 ay on a track and was part of a block of arable in the ' château landscape' associated with La Meule, 125m to the west (a landscape where seigneurial 1 and management introduced distinctive rectangular field shapes, greater expanses of meadow and woodl and and diversions of streams and roads for essentially aesthetic purposes [Astill and Davies 1982b: 22]). An area of 0.62 hectares was walked, yielding 0.31 Roman sherds per square (2.77g), 0.52 médiéval sherds (2.72g), 1.15 post-medieval sherds (4.1g) and 6.36 fragments of brick and tile (115.57g). The assemblage included second-century central Gaulish Samian and rims of third/fourth-century types; thirty-eight pièces of tegul a and twenty-nine of imbrex; three pièces of haematite (310g) and two worked flints. Magnetic susceptibil ity readings did not vary much, but although médiéval and post-medieval pottery were gênerai ly distributed over the field, both Roman wares and brick and tile cl ustered in the north-west quarter. Although quantities of Roman pottery were small, those of brick and tile were unusually large, and their distribution, together with the nature of the assemblage, suggests that the field once contained a Roman-period structure. Médiéval and post-medieval pottery, by contrast, probably arrived as a resuit of manuring. B216 lies on a slight north-west si ope at 65m in an area that was extensive 1 ande in the early nineteenth century. Previously, there was little to suggest that it was cultivated before the twentieth century and it is clearly in a zone that was marginal for most of the historic period: the nearest settlement (La Bridelaie) is 500m away. 0.41 hectare was walked. No pre-medieval or médiéval pottery was recovered and no brick or tile; but 0.05 sherds per square of récent post-medieval pottery were collected and eleven pièces of haematite (1561g). There was little variation in magnetic susceptibil ity readings. The results confirm the impression that this 1 and was not brought into cultivation until the twentieth century, and confirm the classification made on the basis of transect walking - the field real ly is 1 bl ank 1 . B347 lies on a slight east-facing slope at 35m, beside a stream, and its size is 0.44 hectare. In the early nineteenth century it was part of an area of water meadow, 250m from the nearest settlement of Le Cleu. 0.3 Roman sherds per square were collected (2.11g), with 0.4 médiéval sherds (1.63g), 0.31 post-medieval sherds (1.71g) and 2.38 fragments of brick and tile (50.82g). The assemblage included second-century Roman wares, fifteen fragments of tegul a and one of imbrex. Médiéval and post-medieval pottery were gênerai 1 y distributed over the field but both Roman pottery and brick and tile cl ustered in the north-eastern third. As with field 0221, this cl ustering was sufficiently pronounced (and sufficiently distinct from the distribution of médiéval and post-medieval pottery) to suggest that a Roman-period structure once stood on the field. Magnetic susceptibil ity readings produced unusually high values, and rather more variation than did the other fields, with some tendency for lower readings in the area of the brick/tile/Roman cl ustering. A pl atf orm at the north-western edge of the field produced little material of any type; it is likely that this area has been ploughed down to the bedrock. Quantities of natural schi ste from this field were enormous, some squares producing 45kg; the sampling strategy proved to be of very limited value since it was impossible to gain a sensé of overall distribution and difficult to correl ate the collected squares with the cl usters of other classes of material. 5 Excavations (EBS85 Tl and T2) Small -scale excavation of earthworks was undertaken in order to begin investigation of the soils and archaeology within field boundaries near sites identified in fieldwalking. As indicated previously, one of the areas of greatest environmental potential lies in attention to the pedology and sedimentol ogy of the soils themselves, especially where this can be precisely rel ated to the archaeol ogi cal évidence (Astill and Davies 1984c: 58). The features chosen were close to A116, a concentration of médiéval pottery ('site') discovered in the course of transect walking in 1982. In 1983 a 'total' collection was organised over the same area, with geophysical and geochemical prospection. The results suggested that there was domestic occupation on some part of the field, at least during the médiéval period (Astill and Davies 1984a: 20). A116 lies to the north of Ruffiac, between Coetion and Le Vivier. It is located in a small damp valley bottom, on a north-facing si ope, at 70m. At least one platform can be seen. To the north of the field, in the lowest part of the valley, is an area of permanent pasture in which there is a prominent bank and a possible platform (see fig. 2). The bank runs diagonally across the valley bottom. 130m to the west of this bank, at the bottom of the south-facing slope of the valley, is a pronounced lynchet some 1.8m high. This area is near the northern periphery of the Ruffiac commune, and the ancien cadastre indicates that it was a zone of extensive 1 ande in the early nineteenth century. Set within the 1 ande was the petit château of Coetion, with its metai ries (associated farms) of La Touche Gourelle, Bas Coetion, Ruis, Gayon, and Le Vivier. The surroundings of the château seem to have been deliberately landscaped with long, straight, tree-lined approach roads, copses and a fish pond, and the farms are surrounded by the large rectangul ar blocks of arable characteristic of the ' château landscape". Since the château appears to have been built by the seventeenth century, and the associated landscaping at least considerably pre-dates the nineteenth century, the settlement at Allô was presumably abandoned before seigneurial interests put their mark on this landscape. The bank, which lies in an area of nineteenth-century meadow and is not shown on the ancien cadastre, may therefore relate to earlier land-use; the lynchet lies at the edge of the château arable, about which - at least - it should furnish some useful information. After three days préparation, excavation took place over eight days with a team of nine people, making a total of 52 working days. The weather during this period was appalling, with torrential rain and high winds: one complète day was lost and extremely difficult conditions prevailed on another three. Excavation therefore had to be more limited than was intended. The excavation strategy was similar to that used by Martin Bell to sample lynchets and valley bottoms in the south of England (Bell 1977; 1983). A 2m-wide trench, 14m long, was eut across the large lynchet (Site Tl), and a lm-wide trench, 18m long, across the bank (Site T2). This was done using a JCB, a method that would allow a preview of the stratigraphy . lm-wide trenches were excavated to the side of the machine cuts, and thèse were divided into mètre squares to facilitate excavation and recording. Soil was excavated using trowels and ail the finds, with the exception of schiste, were three-dimensional ly recorded. Because of the quantities recovered and the lack of time, the schi ste was collected in 5cm spits. The bad weather prevented total excavation of the trenches. In Trench 1 work was concentrated in areas where features had been observed in the machine-cut trench: 64% (9 of the 14 mètre squares) was dug. As no features were seen in the machine trench, alternate squares were excavated in Trench 2 and hence 50% (9 of the 18 mètre squares) was dug. Soil samples for micromorphol ogical and pollen analysis were taken in columns from the sections, using purpose-made meta! containers. Small test holes were also eut by machine at every 15m to the north of Trench 1 (up the side of the south-facing valley) in order to record the depth of the subsoil (see fig. 2). Trench 1. A shallow ditch was located, eut into the subsoil and natural schi ste. It had a slightly sloping bottom (33cm wide) and was situated 5m north of the crest of the présent lynchet; the ditch was on approximatel y the same alignaient as the lynchet. The ditch was fi 1 1 ed with a homogeneous silty loam which contained two sherds of médiéval pottery and two pièces of (? médiéval) tile. The fill of the ditch was indistinguishable from the overlying thick 1 ayer of loam, which constituted the main body of the lynchet and lay immediately below the modem plough soil. The loam and the plough soil produced 112 pièces of brick/tile and 92 sherds of pottery. Most of the pottery was of fabric 1 (64%), the most common médiéval type found in surface collections throughout the study area, with 11% of fabric 5, a soft cream fabric used for médiéval table wares. There was a small proportion of both Roman and early médiéval types (3% [fabrics 13 and 16] and 2% [fabric 10] respecti vel y; see below, 10). The absence of the highly fired quartz-tempered wares characteristic of the région in the sixteenth to early eighteenth centuries was notable: only 3% of post-medieval pottery was recovered, and that mostly modem. There was no apparent zoning of particular fabrics, which would suggest a constant accumulation of soil produced by near continuous ploughing of the field to the north, at least during the médiéval period (see fig. 3). The trial holes up the slope to the north of Trench 1 showed that the subsoil occurred at a shallow depth (about 30cm), in contrast to that recorded at the north end of Trench 1 (lm), and at the crest of the lynchet (1.8m), indicating notable col 1 uvi ation . The schi ste recovered from the lynchet was not derived from the yellowish-red type which constituted the natural bedrock in the trench. Most was of a grey/green colour. The greatest amounts were recovered at depths of between 5 and 40cm, as was also true of the other material. The pottery, brick, tile and also this schi ste were probably brought on to the field in the course of manuring. Trench 2. The remains of a bank, only 20cm high, were found lying on the natural, midway along the trench. The bank was made of redeposited natural clay; no associated structural features were observed nor dating évidence recovered. The bank appears to have been sufficiently big for soil to accumulate behind and eventually over it, thus producing a shallow lynchet. This soil was a homogeneous loam. It produced only seven pièces of brick and tile and 47 sherds of pottery. Most of the pottery was of fabric 1 (44.7%); there was also a notable proportion of the médiéval tableware, fabrics 5 (12.5%) and 6 (21.3%). The post-medieval wares were mainly nineteenth-century types (8.5%). As in Trench 1, there was no apparent zoning of particular fabrics. The quantities of the (non-local) schi ste progressi vely decreased with depth, and the small amount of brick and tile is in striking contrast to that in Trench 1. The excavations are important for demonstrating the archaeol ogical potential of small -scale work on field boundaries in this area. The ditch in Trench 1 is likely to be an earlier field boundary, perhaps marking the division between cultivable 1 and and the wet areas of the valley bottom. In Trench 2 the bank may have marked the boundary between plough land and meadow and could also have acted as a kind of flood barrier; the shallowness of the loam deposit would argue for a shorter period of ploughing in the lower northern slope of the valley (the area immediately behind this bank). The absence of brick and tile coul d indicate that this happened before brick and tile were in common use. Both trenches have shown that lynchets could have a complicated history, with their origins in other features. From the point of view of land-use history, the pottery suggests that this area of Ruffiac commune was intensively cultivated from the 1 ater twelfth century (although small amounts of earlier pottery could reflect earlier activity), while the absence of early post-medieval wares suggests a lapse in arable cultivation during the early modem period or changes in manuring practice. Since the two trenches showed différent amounts of soil buildup and material in the two areas, changes in manuring practice rather than lapse in cultivation may be more relevant in explaining the data from Trench 1. The land-use around Trench 2 clearly changed when it was exploited as meadow, apparently in the early post-medieval period. The range of pottery found in both field boundaries reflects closely that recovered from 'total' collection of Allô. The désertion of the médiéval settlement may have been connected with the remodelling of this area when or shortly after the château was built. It is therefore hoped that further investigation of Allô will clarify the relationship between settlement and surrounding earthworks. To date, nothing suggests that this area was intensively cultivated before the Roman period, but the weight of évidence suggests intensive cultivation in the late middle âges that was limited in the early modem period with changes in the social status of the 1 andowner and in his method of 1 and management. 8 Envi ronmental , Pottery, Architectural and Language Work, 1984-85 In the course of the last year pollen analysis has been carried out on samples taken during 1984, in particular from buried soils beneath banks in woods near Le Vivier (not far from the excavation site) and Le Rond Point (Carentoir), areas of extensive 1 ande in the early nineteenth century. Although the sample was small, in both cases there were indications of cereal cultivation in phases previous to those dominated by grass and calluna heathlands. Further samples have also been tested for diatom préservation: diatom floras do survive, though often in a fragmented state, and it does not look profitable to pursue this work in the near future. The preliminary stages of analysis of the pedology and sedimentol ogy of the soils have been initiated. Dr Marie-Agnès Courty and M. N. Fedoroff, with their assistant Anne Gebhardt, visited the study area during the season and proposed - as part of an investigation into the application of microscopic techniques to the identification of traces of agricultural and pastoral activities in the soils and sédiments a sustained programme of soil analysis in close association with other field and archive work in the area. Samples were taken from Trench 1 and Trench 2; more will be taken during the summer; and micromorphological analysis will begin in September. The samples taken from 'total' collection sites in 1984 for phosphate analysis have ail been processed and a study was made of phosphate concentrations at the deserted médiéval settlement at Kerlano. This site, partially excavated by M. Patrick André nearly twenty years ago, is located in the granité upland of the Landes de Lanvaux, 30km west of the study area, and is the nearest known surviving médiéval settlement (André 1974). At least one building in the settlement had a central hearth, and middens (represented by pottery dumps) were located around the buildings immediately outside their walls. In an attempt to understand phosphate concentrations and magnetic susceptibil i ty readings from the fields of the study area, and to pursue the problem of surface scatter arising from middens, it was decided to investigate phosphate and magnetic susceptibil ity readings from this known settlement with its known pattern of rubbish disposai. In December 1984 soil samples and magnetic suscepti bil i ty readings were taken at mètre intervais along transects intersecting in the middle of an unexcavated building. In gênerai both magnetic susceptibil ity readings and phosphate quantities were very low, at less than 10 Si/kg and lOOppm, but over the site of the building, especially inside and immediately outside its walls, there were higher levels of 26-37 Si/kg and 200-300ppm. The experiment suggests that we might expect structures and middens in the study area to produce relatively high levels. Pottery from the 1984 season has been sorted and classified by Astill, Cook and Wright, and compared with the existing fabric séries. No changes have been suggested for this séries, which now comprises 16 fabric groups for prehistoric, Roman, médiéval and early post-medieval pottery. In December 1984 the fabric séries was compared with that generated by survey undertaken around St Malo by the Centre Régional Archéologique d'Alet. Similarities in some of the médiéval fabrics were noted, and there was a striking visual similarity between fabric 10 and pottery from a kiln found at Guipel ( 1 1 1 e et Vilaine). The kiln produced an archeomagnetic date of 895-945 AD. A survey of ail standing buildings has been in progress in the study area since October 1984, undertaken by Pete Addison, working as a fui 1 -time research assistant. To date, 4000 buildings have been recorded on standardised recording forms, coding their attributes to enable electronic sorting. Thèse forms are accompanied by notes and drawings, together with 1000 photographs. Although the main work has been of recording, provision of a dating framework for the undated buildings is at an advanced stage, assisted by the récognition of characteristic seventeenthand 1 ate nineteenth-century types. It is clear that some settlements have a prépondérance of one or other of the two common types of building - the single cell , with ground floor hall and storage loft above, and the long-house, combining byre and hal 1 . Local pronunci ation material collected while fieldwalking has again suggested that the influence of the vernacular language is évident in scattered parts of the study area; initial stress on Trignac and Kerhal , for example, are notable. Pi scussi on The 1985 season produced the highest number of sites in transect walking and by far the greatest quantity of material, including very large amounts of brick and tile. The season also allows further observations on the problems identified in 1984 and some reinf orcement of the suggestions made there, with some useful focussing on early modem, pre-cadastral use: the relevant problems are those of distinguishing manuring from settlement scatters and of determining the extent of scatter generated from inhabited buildings (Astill and Davies 1984c: 55-8). It is even clearer that crude quantities of surface material are no sure guide to the ci rcumstances of its déposition: distribution over the field is as significant as quantity and fields - and even gardens - beside long-establ ished inhabited settlements do not necessarily produce the greatest quantities of material; 27 fields hard by buildings tended to have as much post-medieval as médiéval pottery but only seven produced sufficient concentrations to warrant classification; fields penetrating the buildings at Rangera, Bussonaie, Pied, Gaincru and early médiéval Lodineu ail produced no, or next to no, material. As in 1984, even fields with inhabited structures in the early nineteenth century produced no more material than that sufficient to qualify as a 'possible site' (F212); and, as noted above, most of the high concentrations occurred more than 100m from présent and cadastral settlements. In the particular case of Roman wares, only small quantities have been recovered and the 'total' collections made from D221 and B347 both suggest that Roman sites generate little pottery on the surface, though they may produce large quantities of brick and tile. 0f course, sometimes fields by inhabited settlements produce plenty of material, such as other parts of Gaincru, the field beside La Charmille on the outskirts of Ruffiac, and those beside the château of Peccaduc; most notable are the quantities in and lO around the small settlement of Trignac and on the western outskirts of Carentoir village, a striking contrast with that collected around Treal centre in 1984 (Astill and Davies 1984c: 56). In fact, the combination of material again suggests that modem settlements generate little scatter and that - in this study area - large quantities indicate 1 ate médiéval and/or early modem (twelfthto seventeenth- century) activity. Indeed, further, surface finds suggest that settlements throughout the modem period may have generated less than 50m of scatter: in L90 finds occurred within 20m of the buildings at the side of the field and not beyond; at the petit château of Gre'e Orlain there were fields that were 'blank' a short distance between inhabited buildings and 'probable sites'. Of course, it is not merely past settlement that is of interest but past land-use too. As demonstrated by the 'total' collection sites of D221 and B347, even one field can produce distribution patterns that differ by period; the cl ustering of one class of material can suggest structures, the lack of cl ustering of another can suggest the effects of manuring. Comparison with cadastral land-use can also be helpful here. Sites like Trench 2 - which fall in areas of early nineteenth-century meadow or 1 ande are likely to represent pre-nineteenth-century activity, and as the excavation clearly demonstrates sometimes cultivation, pace our earlier suggestion (Astill and Davies 1984c: 58). This year concentrations on meadow/1 ande most frequently produced a prédominance of post-medieval pottery. This suggests that many of the concentrations were in areas of early modem arable; in the case of those in the 1 ande they may indicate the taking in of new arable in marginal areas in the context of population pressure. Here, both this year' s excavations and the récent pollen analysis provide compl ementary material, with their suggestions of cultivation in areas that were certainly - at différent periods peripheral 1 ande. Fields walked on recently cleared woodl and also produced some compl ementary material, as in 1984. Again, much of the established woodland of the early twentieth century lies in 'blank' areas that produce no surface material when the wood is cleared and fields walked; again, this suggests no médiéval or early modem cultivation in those parts. However, a small proportion of recently cleared woodland did produce surface material. This was sometimes Roman and sometimes médiéval but it was also sometimes post-medieval. In the 1 atter cases the woodland (deciduous as well as coniferous) can hardly be more than two hundred years old. In such ways surface scatters, in association with other évidence, are beginning to suggest precisely localisable changes in land-use, especially during the 1 ast millenium. This year also saw a major attack on the problem of the local schi stes. The problem is a simple one: because of outcrops and deep ploughing, fields more often than not have a surface scatter of freshly broken and/or abraded schi ste; schiste is also the most common building material in use in the study area between the seventeenth and early twentieth centuries. Given that vacated buildings are often left to di sintegrate , collapsed buildings are likely to leave a surface scatter of schi ste; where this is distinguishable from natural, the distinction ought I \ to be noted since the material may be just as significant as surface brick and tile. We have made some progress in characterising the local schi stes - which are of very mixed character and considérable local variation - by identifying small quarries within the study area and comparing samples with material from buildings and from surface collections (principally during May 1985). This is sufficient to make it clear that a proportion of the commonly occurring surface material on our fields is imported - from good quai ity modem si ate to harder schi stes, both used largely for roofing but also within the mud si urry cores of walls. The harder schi stes are of three distinctive colours - black, pink-purple and grey - and can only very rarely be found on inhabited buildings now, although piles of both black and pink-purple sometimes survive in the fields beside collapsed structures. The présence of this material on the fields therefore indicates imports into the study area for building purposes. It is so common on the surface of présent arable that it is reasonable to explain most of it as the product of manuring activities; it only does not occur notably on recently cleared woodland (F231, F594, C405, C436, for example). It is therefore at least a useful indicator of pre-twentieth-century arable. However, there are also différences in colour combinations for there are zones with ail three colours, others without pink-purple and others with black only; for example, the dense scatters on the long-used fields on the western outskirts of Carentoir have ail three ^colours while the area of the ' château landscape 1 around Gree Orlain and Herblinaie has only black and grey. (In fact, black zones do not seem to be distinct from black and grey.) This needs more observation and more testing but at présent it suggests that the présence of pink-purple imported roofing schi ste is an indicator of areas of 1 ate médiéval and early modem arable while its absence is an indicator of land more recently taken into cultivation. Potentially; then, this is an excepti onal ly important source of information about land-use in the pre-cadastral period. As for the local schi stes that form the bulk of walling materials, although it i_s_ possible to distinguish thèse by eye in the field and even localise them within the study area, it is probably impractical to attempt to instruct a teamamateurs to do so. However, it would be possible, in some areas, to be précise about the source of local building material and deduce patterns of its movement within the study area over time. More quarry samples and some further limited testing of 'total' collection sites would therefore be useful, particularly from a site with a known, though collapsed, cadastral settlement; the nature of the red/yellow schi stes unevenly distributed on A107 and the grey/green schi ste introduced into the field bounded by the lynchets and eut by Trench 1, need some further investigation. The next season Next year will see two main seasons of fieldwork: fieldwalking and small excavation from 22 March - 5 April 1986 and three weeks of excavation during September. Meanwhile Pete Addison will complète the survey of standing buildings; a small team will dévote February, March and April to 'total' collection; analysis of pollens from the Mauffrais podsols and the River Aff \1 backswamps will continue; and work on the local schistes will be pursued intermittently by Wendy Davies. In the Easter season transects with uneven coverage will be rewalked, especially the northern part of E and parts of A and B, to achieve a relatively even survey of ail parts of the study area. The process of sampling the surrounding communes will be begun: transects will be taken radially from the area of intensive study, and fields within them walked at 50m intervais. Excavation of Allô will begin, in order to ascertain the relationship of this year's excavated features with the settlement and at the same time make some assessment of the quai ity of préservation of sites; two 25m squares within the field have already been prospected using a f 1 uxgate gradiometer in order to assist choice of areas for excavation. Thereafter further sites from which 'total' collection has been made will be investigated in small areas to continue exploration of the meaning of surface scatters and to extend the range of environmental study. Investigation of the pedology and sedimentol ogy of lynchet deposits will continue to be integrated with archaeol ogical investigation of the sites. This report, in French, will appear in the next Dossier of the Centre Régional Archéologique d'Alet. The season's work was undertaken with the authorisation of the Ministère de la Culture, Direction des Antiquités de Bretagne, and many thanks are due to M. Le Roux, director of the circonscription, and to the conservateur M. Clément and Michael Batt for facilitating the smooth runing of the programme. Excavation was undertaken with the permission of the owners of Allô, M. and Mme. d ' Argentre", and the tenant, M. Poyou, and the Kerl ano site was tested with the permission of M. Le Labourier; without their help much of this year's season would have been impossible and we are therefore immensely grateful to them. Fieldwalking and excavation were financed by the British Academy, the Society of Antiquaries of London, the University of Reading, the University of London Central Research Fund and University Collège London; the buildings survey is financed by the Leverhulme Trust; we are in ail sensés indebted to thèse bodies for their support. We are also especially grateful to Professor P.-R. Giot, Dr L. Langouet, Professor G. Mei ri on-Jones and Dr S. Wright for devoting their time and expertise to particular aspects of the project; to Dr R. Battarbee for testing samples containing diatoms, Dr T. Stevenson and Judi Darley for work on pollens, Robin Iles and Martin Cook for testing phosphate samples and Anne Gebhardt for micromorphological work; to Bill Campbell for considérable assistance with computer mapping and very generous use of his software 'Mapics'; to Steve Ford, Lorraine Mepham, Liz Musgrave, Eric Norton, Frances Raymond, Kate Sergeant, Mog Tingle and Cathy Wilkey for their invaluable help in supervising work in the field; and, as ever, to our team, who worked stoically in awful conditions and ensured such a productive season. We should like to express our thanks, as ever, to ail who have given time, labour and energy, as well as to those who have provided financial support. G. G. Astill Department of Archaeol ogy University of Reading Wendy Davies Department of History University Collège London Whi tekni ghts Reading RG6 2AA Gower Street London WC1E 6BT 30 July 1985 REFERENCES André", P. 1974 'Le site médiéval de Kerl ano-en-Pl umel ec (Morbihan)', Archéologie en Bretagne, ii, 27-34. Astill, G. and Davies, W. 1982a 'Un recherche sur le terrain dans l'Est de Bretagne, xxxv, 24-42. nouveau programme de la Bretagne', Arch. en Astill, G. and Davies, W. 1982b 'Fieldwalking in East Brittany, 1982', Cambridge Médiéval Celtic Studies, iv, 19-31. Astill, G. and Davies, W. 1984a 'Recherches sur le terrain dans l'Est de la Bretagne (EBS) - 1983', Arch. en Bretagne, xxxix, 13-23. Astill, G. and Davies, W. 1984b 'Prospections archéologiques dans l'Est de la Bretagne', La Prospection Archéologique en Haute-Bretagne , Dossiers du """Centre Régional Archéologique d'Aï et, no. G, 251-60. Astill, G. and Davies, W. 1984c Bretagne. Résultats de la Dossiers du CRAA, xi i , 49-59. 'Prospection campagne de Bell, M. 1977 Excavations Archaeol ogi cal Collections, cxv. at dand l'Est de la Mars-Avril 1984', Bishopstone, Sussex Bell, M. 1983 'Valley sédiments as évidence of prehistoric land-use on the South Downs', Proceedings of the Prehistoric Society, xlix, 119-150. 1+ EAST BRITTANY SURVEY 1985 £J3 fields walked m 1965 ■— — commune boundaries A • ■ sites A c* U probable • médiéval si tes sites ■ post médiéval sites A o □ possible sites A undated sites Fig. 1