EAST BRITTANY SURVEY

1986 EAST BRITTANY SURVEY EAST BRITTANY SURVEY

EASTER AND SEPTEMBER SEASONS 1986

Easter Season 2

Fieldwalking at 50m intervais 2

Fi g . A 3

Sampling outside the core 6

Fig. B 7

Excavations Al 16 8

Fi g. C 10

September Season 13

Fi g- 1, 2 14

H145 15

K446 18

Fig. 3 19

H80 22

Fig» 4 23

A92 ' 28

Fig. 5 9

Fig. 6 32

□t h er Work, 1985-6 33

General Comment and Acknowledgements 34

Re-f erences 35

P1 a t. e s 1 -10 f ollowing 35 The fifth and sixth seasons in a programme of fieldwork, itself part of a larger, multi-disciplinary study of the relationship between land-u.se and settlement during the last two thousand years, took place from 21 March - 5 April and 6 -- 27 September in the communes of Ruffiac, Treal, St~Nicolas-du-Tertre, , La Chapelle Gaceline, Caro, Reminiac, and Campélieac in the in eastern E

THE EASTER SEASON

The 19S6 Easter season involved fieldwalking over large areas at wide intervais, as in 1982 - 85, in the four core communes; sampling in the communes surrounding the core? and excavation of parts of a field from which 'total' collection had previously been made. Two days (21-22 March) were spent in préparation for the main season by two people. The team (consisting largely of past and présent students from the Universities of London, Reading, Durham and Sheffield) numbered twenty, including the directors, and worked for twelve days, from 23 March, with one day off. A smal1 team, of the directors and two others, returned on 28 April for three days in order to complète the excavations. Overall, 317 mandays were spent on this year's Easter seasons (303 by the main team), including travelling time and time off5 255 (244) of thèse were working days. The weather was dreadful for most of the main season, with considérable rainfail; fields were usually in idéal condition for walking and surface collection was not seriously impeded, but it made excavation al most impossible. Conditions for the return visit were much better, allowing the desired program to be completed on the excavation site»

Fieldwalking in Runs at 50m intervais (Transect Walking)

A. Within the core Fieldwalking over large areas within the four core communes was organised within transects running south/north across them» Transects A (Ruffiac), D, E, 6 (Carentoir), B and parts of J (Carentoir, Treal, St~Nicolas) were rewalked in order to increase the proportion covered of the surface area of this core (see fig» A). Except for fields previously collected, ail fields under plongh and with young crop within thèse transects were walked at 50m intervais, using collection units of 100m; field EAST BRITTANY SURVEY 1986

| TRANSECT J TRANSECT B "t— TRANSECT D

TRANSECTA -h— TRANSECT E

ci3" .y. : 75m ■ S Y •m Tréal .o-v. Lo

50m-

l—' O O ....

l'ï TRANSECT G □ «2.

; C? cm. i-. >.^'v... (Rutfiac g SOfl\

□ 1-5km apart 193 -.. > 4km apart ■ M") CP ^ A3 Carentoir

44 50m

SNicolas-du-Tertrç

•--50m-- '-. ('?• - .•0 il ■75m- «A "C3 •so*1 CI ..Q .. I

i c 7 -.- sm ..I... I \ / TRANSECT M . . 100m I fN./ < lields walked in 1966 1 commune boundories A • m site. • • . J. médiéval sites I A • 9 probable sites \ post médiéval sites

A o Q possible sites i undated sites 500m 0 conditions, features, présence of varieti£?s of schiste and local pronunciations were noted on standardised recording forms. 285 fields were thus covered, encompassi ng 372 hectares (1.947. of the surface area of the four communes). This brings the total surface area covered during five seasons to 18.54%, as mueh as can reasonably be e>:pected within the constraints of time and money.

12.49kg of pottery and 78.42kg of man-made building material were recovered frorn the transects; 6.47. of the pottery was Roman, 34.87. médiéval and 58.8?/;, post-medi evail » This is consi derabl y more brick and tile than recovered in each previous season, rather more Roman pottery and rather less médiéval. Ten worked flints were also collected, three from transect E, three from B, two from B and one from D. One possible platform and thirty lynchets were noted, of which fifteen were in Transect D. As in previous years there were considérable variations in the concentration of recovered pottery, and the saine conventions are hereby used to distinguish between thems 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'; 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, gr. 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 quai ify a concentration for comment has been deli berately fixed at a 1ow level in order to secure a wide range of possibi1ities for testing; in fact, in many cases numbers were far higher than the necessary minimum. It should be stressed, yet again, that the ternis 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 sett1ements„)

In accordance with the conventions, the concentrations (total 57) may be classified as followss

A B D E G J Total s 'Sites' 1(2.4) 2(2.4) 1(1.7) 3(4.2) 3(16.7) - 10 (3.5) 'Probable' 4(9.5) 9(11) 3(5) 2(2.8) 5(27.8) 1(8.3) 24 (8.4) 'Possible' 3(7.1) 3(3.7) 6(10) 8(11.3) 3(16.7) - 23 (8.1)

Ail types 8 14 10 13 11 1 57 (197.) (17?/.) (16.77.) (18.37;) (61.1?/.) (8.3?/.) (207.)

Fields 42 .82 60 71 18 12 285 walked Table 1 (Figures in brackets are percentages of fields walked, per transect

and in tôto.)

Of thèse concentrations, none produced predominantly Roman material although 17.57» produced some Roman material;; 12.37. had predomi nanti y médiéval, 22.87. predomi nant 1 y post-medieval, and 8.8?/. more than the necessary minimum proportions of médiéval and post-medieval sherds» 43.9% of concentrations had a prédominance of building material, at présent treated as undatable; the remainder had more than the necessary minimum of pottery and. building material» Thèse figures again stress the large quantities of brick and tile, relatively large of Roman and smal1 of médiéval pottery»

The spatial distribution of this material is as interesting as in previous years, and broadly consistent with it» There are 'blank' zones within the four communes and also zones that tend to produce more or less of a surface scatter» Surface material is usually markedly absent from areas near the commune boundaries and is also difficult to find in the fields to the north of Qui 1 vain and around Le? Bois Faux; si gni f i cant 1 y, the more intensive covera\ge achieved by rewalking many of the transects did not affect distribution patterns already noted» By contrast, concentrations of surface material are ^frequently found in zones around Le Cleu, La Touche (east of Treal) and - especially - in the southern half of Transect G, from Métairie au Joly to Le Nouai-)» The figures quoted above demonstrate the remarkably high proportion of fields in G with large concentrations of surface material. The présence of tile scatters, sometimes with smal1 quantities of Roman material associated, were notable near the northern boundary of Transect D (a Roman road) and in the neighbourhood of Marsac, La Ruaudaie and La Roche Pèlerin» Analysis of the spatial distribution of imported local schi stes » parti cularly those used for roofing materials in the area, produces some equally marked patterns» Such schi stes were présent on most fields (largely as a resuit of manuring), with the exception of some areas of 1 ande in the northern part of Transect B and of some isolated, scattered fields (D6 and 7, E267) with no other surface material. However, it was again possible to see zoning in the présence of thèse schi stes» hence, only black schi stes occurred throughout the ^Ruffiac basin (from Lodineu to Ville Robert and south to La Rivière) and also in a large zone in the north of Transects D and E (from Trelo to La Touche du Mur); in strong contrast black and pink/purple roofing fragments occurred together in zones immediately north of the Ruffiac basin and west of the Trelo zone. It was also possible, in some cases, to distinguish a range of other schi stes that were imported from nearby local sources, probably as walling material; field E230, where there was also a large tile scatter, provided a parti cularly good example of this.

Comparison of the distribution of surface material was made with the early n i neteenth-cent ur y p atterri of 1 and-use and settlement, as e v i d e n c e cl b y the ancien cadastre , as u s u a 1 . More than three-quarters of concentrations of pottery (76.7V.) 1 ay more than 100m away from early nineteenth-century settlements and only 12»5% 1ay within 50m of them» (The former is only slightly higher than distances from modem settlements and the latter proportion the same.) Surface material tends to be found within areas of ni neteenth-century arable, as always - especially in bandes (1 itéraily 'bande', arable divided into tenant parcels) - and the proportion was comparable?, though slightly smal 1er, with that of earlier years (58,9% of concentrations in 1986, as compared with 70.1% in 1985). The remainder occurred in nineteenth-century meadow or pasture (an unusually high 12.5%), marginal 1 ande (unc u11 i vat ed land - 10.77.), curtilage (5.47.) and areas of mixed 1 and—use. Concentrations of material in early ni neteenth-century 1 andes (like B2S and D52) and meadow (ilke BS7 and B90) demand some further investigation, since they imply either arable use or settlement previous to the nineteenth century; sites 1ike G21S, 220 and 221, lying in the distinetively managed landscapes associated with pet i ts château;-: , also warrant further investigation because such areas tend to préserve more traces of earlier landscapes. Add i t i onal 1 in some parts cadastral land—use, naming and road p a11 er n s i n t h emse1ves suggest. former settlement sites (Asti 11 and Davi.es 1982b. 21f 31); such cadastral suggestions coincided with a concentration o-î >ur ace material at the 'probable site B85.

As usual, the 1 opographic position of the sites was also analysed. Again, a third of them loo. OA) were on fiât .Land whiie 14/. were on south-faci ng and 19.37. on east-facing slopes. Upland concentrât i ons were not especially notable, with only a quarter lying between the 50m and 75m contours (28.IX)5 almost a half 1 ay between 25m and 50m (49.17.). Somewhat less than a half of the? sites 1 ay within 250m of mapped streams (427.) , while more than a quarter (31.57.) were more than 500m away from them.

£<. Samplinq ou.tside the core

Sampling outside the core was organisée) in 2km transects radiating from it and thèse were walked in the s ame way as transects within the core. Ail available fields in one, Transect N we>re covered; tl 11s ran for 10km slightly west of north, from the north-west boundary of Ruffiac, and passed through the communes of Carc , Reminiac and Augan and into the south oi Zampefrieac, running just to the w est of Augan commune centre, Here there was a marked change in the topography - as also in the bedroc and in modem land management — as we moved north 5 i t was rnuch higher with steeper hi 11 s and some north/south valleys; but farms (and often fiel ds and houses) were biqger and their property less dispersed, with more cattle and altogether a more wealthy air; and, despite récent remembrement, fewer remembrement mounds than in the core. Transect M, running slightly south 01 east from the south~-e?ast corner of Carentoir, was bequn, but most of it remains to be covered (see fig. B)» 209 fields were walked (of which 6 were in M), encompassing 257 hectares (18% of the surface area of Transect N).

7.86kg of pottery and 13.81kg of man-made building material were recovered from thèse transects; 1.37. of the pottery was pre-Roman„ 1% Roman, 29.7% médiéval and 68% post-medieval5 six worked flints were? collected. One possible platform, a mound and ten lynchets were also noted. In accordance with the usual conventions, concentrations of surface material (total 51) may b e c 1 a s s i f i e d as f o 11 o w s r,

N M Total s Si tes' 4 ( 27. ) 4 (1.9%) ' Probable' .8 (8.97.) 3%) 20 (9.67.) Possib1e' Î7 (13.3%) 27 (12.97.)

Fields 203 6 209 walked Tab 1 e

Of thèse concentrations 9.87» produced some Roman and pré—Roman sherds; 9.87» had predomi nanti y médiéval, 54.97. prédominant!1/ post-medi eval , and 11.87. more than the necessary minimum proportions of médiéval and post-medieval sherds. 13.77 had a. prédominance of building material and the rest had more than the necessary minimum of building material and pottery. Despite the changes in topography and land management the proportion of sites located and quantities of material collected is entirely comparable with those in the four core communes. However, there is less médiéval material than has usual1 y been found there and it was a considérable contrast to collect pre-Roman sherds from the surface. However, the distribution pattern of material does not significant1 y deviate from that found in the core. There are again areas with little or no material 'blank ' zones - and areas with more or less of a surface scatter. So, material tends to concentrate in the La Bonneraye/Le Printemps/Saint-Nicolas (Reminiac) area and in the area around ^the château of Touraille, stretching north from it over the? Campeneac commune boundary to La Tieulais; and there are a distinetive number of 1ow concentrations of post-medieval material in the Les PIacieux to Ville Glâird area. There are large 'blank' zones from La Ravraie to Le Boulay and also north west of Augan, although such zones are not so characteristic of the commune boundaries as they are in the core communes.

Se h i stes in gênerai re-f lect the pinker colour of local naturals and there are consequently far more fields with a mixture of black and pink/purple roofing schi stes, often with a prédominance of the pink. However, there were still zones that could be distinguisheds black roofing schi stes overwhelmingly predominated in the Boussac/Bramboc/La Buzardiere area and there was very little surface schi ste at ail, of either type, in the Patis de Boussac . In the northern part of the transect grey schi stes were used for roofing, and thèse are évident on the fields; smal1 quarries, like that near Les Toulans, are still in use?»

Concentrations of material in N tend to be on higher land (ail over 50m and 58,27» over 75m), reflecting the generally higher topography, and tend to be cl oser to streams (517. within 250m of a mapped stream) . Fewer sites are on flat land (14.37.) than is gênerai in the core, and more are on south- (32.7), north- (18.47.) and west--facing (12.27.) slopes - again reflecting the h i g h e r t o p o g r a p h y „

Excavations. Al16

Excavation was designed to sample the medieva\l settlement. of Al 16, located by fieldwalking and classified as a "site', work having begun in 1985 with investigation of the surrounding earthworks. The main aims of this enquiry were, firstly, to assess the state of archaeological préservation and, secondly,

9 to cl i se over the rel at i onshi p between material found on the surface and subsoil features, Work on the excavation was eventually abandoned in the main season because of the rising water table; however, the smalI team that returned in late April was able to complète the trial. Altogether excavation- took place on 8 days, with an average of 6 people per day, thereby absorbing 42 working days.

Two 6m squares were excavated (see fig. G), One, T3, was located on the highest part of the field, where surface prospection had identified both large amounts of pottery and concentrations of phosphates (Asti 11 and Davies 1983s 16, 18). The second, T4, was sited over an area of phosphate concentration but 1ow density of pottery, on an earthwork platform in a lower part of the field. A third cutting, located in an area of high magnetic susceptibiIity enhancement but 1ow pottery and phosphates, was planned but had to be abandoned because of the weather. Quadrats with 6m si des were chosen since a smal1er cutting could fall within a structure, failing to reveal its existence. However, in the later stages of the excavation a machine (JCB) was used to eut three 1m wide trenches to extend the investigation. The first (T7) ran north west from T3 across a slight break in si ope, and the other two (T5 and T6) were eut in order to section the platform near T4 (see fig. C) . The-? sections were cleaned and drawn and features in the bottom of the trenches recorded.

The two 6m squares were mai ni y excavated by hand. The plough soi 1 was removed in 10-20cm spits by shovels the spoi1 was searched for finds (including schi ste) on the si de of the excavation. As a control a randomly chosen 60cm square (57. of the main cutting), was removed separately and wet sieved through an 8mm mesh. The density of finds in this plough soil was low, as the following table showss

no. of finds no. of finds depth recovered by hand in control T3 surface (fieldwalking) 0—10 cm 10-20 cm 20—30 cm

T4 surface (fieldwalking) 1 N/A 0-20 cm 6 2 20-40 cm 9 1 40-60 cm 4 - 60-70 cm 2 Table 3

T3 did not produce évidence of structures. The plough soil was only 30cm deep, with plough-scarred natural schi ste immediately below it. It is therefore likely that any structural évidence in this high part of the field has been destroyed by ploughing, a factor which may explain the relatively large amount. of pottery found on the surface by comparison with that recovered by excavation. Quartz and conglomerate blocks

°\ EBS 85-86

0 20 40 60m were -found embedded in the sur-face of the naturel schi ste; since several groups of blocks of a similar character were noted in the machine trenches, widely dispersed over the field, they could be natural features»

T7, a 13m~long, lm--wide, north-west extension to T3, located a shallow ditch eut into the natural. It had a sloping bottom (50 cm wide) and was fi lied with plough soil. A small négative feature, 20cm wide, was also noted in section. Neither feature produced finds. The pottery from the plough soil of both T3 and T7 was entirely médiéval, and of a similar character to that recovered from the surface, that is 907. coarse wares of fabric 1, 67. table wares of fabrics 5 and 6, and 47. storage vessels of fabric S.

Most of T4 was dug by hand in a similar way to T3, but in the later stages a machine was used to remove some of the colluvium. Two im-wide trenches were also eut by machine. One (T5), 22m long, extended from the north-west corner of T4 and eut the edge of the earthwork platform at right angles; the other (T6) was 23m long and ran from the south-east corner of T4 to the southern end of the platform. A section of the whole earthwork was thus obtai ned.

The earliest évidence came from T4, where three inter-cutting ditches were eut into the natural» It was only possible to section two of them in the time avail cible and the task of establishing the relationship between the three features was postponed until 1987» The sectioned ditches were both U-shaped and were 14cm and 20cm (contexte 4 and 5 respectively) deeps eamples for pollen and soil analysis were taken from context 5» No finds were recovered from the features, but four sherds of pottery came from the surface of the ditch fillings, whose ■fabrics were différent from those previously recorded from the site. Three of thèse sherds were of a sandy fabric with large rounded quartz inclusions (1mm) and a notable absence of mica. One, a ri m sherçj, has parallèle with later Iron Age forms from Brittany (La Tene Finale) (Biot, Briard and Papes 331—9)» The other fa\bric was coarse, with lairge angular lumps of quartz (2—4mm) and no mica» This pottery maiy be prehistoric and could indicate the date of the features excavated in T4»

Overlying thèse features was about 70cm of a 1oamy soil, interpreted as colluvium. The long exposure of this deposit in T5 and T6 made it clear that the platform earthwork was a lynchet, similar in character to the? lynchet sectioned in 1985 on the other si de of the valley. Around the crest of the lynchet a slight change in the colour and texture of the 1owest colluvium (T5, context 4) may represent the; original soil which formed the lynchet» A column of saimples was taken through the de;posits for soil analysis. Pottery from the colluvium was médiéval, similar to that recovered from the surface and from the other trenches; the majority were coarse wares (547. of fabric 1, 387. of fabric 3), with a small amount of fabric 5, a fine ware (87).

The résulte of thèse excavations are important» It is clear from T3 and T7 that where soils are shallow the only archaeological features that are likely to survive are those

il which cleeply penetrate the natural 5 thèse can only be understood by extensive excavation,, Such large—scale work was impossible and it is therefore difficult to assess whether the area with dense pottery and phosphates marks the site of the settlement» If i t does, then much of the airchaeol ogi cal assemblage was found on the surfaice (see Table 3),, Identification of the lynchet by T4, T5 and T6 confirma the 1985 resuits suggested by T2? the lower si opes of the valley had been in arable use in the médiéval period but had been turned over to meadow by the nineteenth century, perhaps by the early post médiéval period (Asti 11 and Davies 1985s 92-5, 97)»

If the occupation in T4 is prehistoric then it is the first of its kind for the? région and has important implications for survey work» Its low-lying position, in a wet valley bottom and sealed by 70cm of colluvium, means that this kind of feature can only be detected by excavation, because even the deepest ploughing is unlikely to bring material to the surface, even assuming the laxnd is cultivable» This may be one reason for the dearth of prehistoric material recovered in the core communes from fieldwalking» The T4 features were excavated in an area where there was a small amount of médiéval pottery and high phosphate concentrations» While the small quantity of pottery is commensurate with the agricultural nature of the médiéval land-use, the high phosphates around T4 may be an indication of the extent of the much earlier site sealed by colluvium. The phosphate distribution of Al 16 may therefore reflect occupation of many periods, and consequently may not relate directly to the distribution patterns of the pottery. THE SEPTEMBER SEASON

The 1986 September season involved sample excavation of parts of four fields (H145, K446, H80, A92) from which 'total' collection had previously been made (fig. i). One day (5-6 Sept) was spent. in préparation for the main season by two people. The team (consisting of volunteers from a wide range of places) numbered twenty-one, including the directors and Alan Lane, and worked for eighteen days, from 7 September, with two days off? Anne Gebhardt joined it for soil sampling for the second half of the season, and was assisted by a fri end for the 1ast week. Overal1, 481 mandays were spent on the September season, including travelling time and time off§ 437 of thèse were working days. The weather was poor during the second week, with considérable rainfall, but only two complète days were lost in t 11 e field.

\3> EBS 86

Fi g. 1

H145

10m 0 10 20 30 UQ 50 m

F i g. 2

\4- H145 (Tréai ZB154a)

This field was classified as a 'possible site' (médiéval) after •f i el dwal ki ng at 50m intervais in 1984» Intensive collection •from the whole of its surface in March 3.986 indicated smal 1 quantities and fairly even distribution of médiéval and post-medieval pottery. Although there were areas of slightly higher concentration, the scatter looked like a manuring scatter..

H145 is a flat field that is located near the crest of an exposed south-facing ridge on the south--west edge of the settlement of Quoiqueneuc in Treal, in the area of a smal1 ninth-century monastery. The field is bounded to the north by the main east-west road leading to Quoiqueneuc, and to the west by the road to Le Passair. The ancien cadastre shows H145 to have been within an area of largely arable land-use and smal1 enclosures (çl_os) , with two buildings and ? walled yards inside its western edge - since démolished. The site of thèse buildings is presumably now marked by a remembrement mound near the junction of the two roads.

Excavation strategy was the same as used on Ai 16. A 6m square was sited over one of the slight pottery concentrations located by intensive surface collection. The plough soi 1 was removed by pick and shovel but 57. of the square was reserved for sieving, to act as a control over recovery rates of material. Further areas

Trench 9 The plough soil was excavated in 10cm spits. The amount of pottery decreased as the depth increased, as did fragments of rock. The marked différence between the 1atter and the bedrock, and their range (siltstones, grès Csandstone3, conglomerate, quartzite) suggest that they are plough scatter derivihg from nearby structures rather than from the local bedrock. The depth of the plough soil varied between 30 and 50cm and in some places overlay the natural clay subsoil and in others the natural bedrock. No archaeological features were encountered.,

Trench 11 The plough soil was excavated by hand and found to be between 35 and 45cm deep. It 1ay above the natural bedrock. No a\rchaeol ogi ca 1 f eatures were f ound ,.

Trench 13 This machine—eut trench was 35m long and no features were o b s e r v e d i n s e c t. i on o r p 1 a n «

Trench 14 T h i s mac h i n e - c u t t. r e n c h w as 28 m long and did not reveal any f eat ures.,

Trench 12 The trench was excavated by machine and -for most of its 28m length merely provided a record of the depth of plough soil» However in its extrême north end a section of stone walling was excavated- The wall, 30cm high, consisted of coursed blocks of schi s te and conglomérats which were bonded with a thin yellow clay. it was 80—90cm thick and was laid directly on to a compactée! plough soil (8). Butted on to the north face of the wall was a dense layer of smalI (15—20cm in diameter) pièces of shale and sandstone - presurnabl y some kind of hardstanding or floor (9)« The northern extent of the standing was established by augering and was found to peter out some three mètres north of the wall. The eastern extent. of the wall was similarly established and found to end 1 » 2m from T12, with no sign of another wall or hardstanding»

Trench 10 Excavation produced the following séquence» The natural clay sufosoi1 contained a shallow dépression (21) and a smal1 hollow (23) , but there was nothing to suggest thèse were not natural1 y formed» A shallow pit (10), about 25cm deep, was dug into an apparently natural subsoil (5). This was filled with blocks of schi ste (including slate) and conglomerate, most of which showed no sign of having been worked (11). A silty soil had accumulated in the interstices of thèse stones, and this produced 11 sherds of pottery, ail médiéval (9 of fabric group 1, and i of fabric group 3), and 2 pièces of tile (one a tequla).

The upper part of the pit continued to be filled in during the médiéval period (3); 32 médiéval pottery sherds were found, of which 25 were of fabric group 1, 2 of group 3 and 5 of a new fabric. The ri m forms were similar to those recovered from the médiéval settlements of and , and are probably of t. we 1 f t h - t o t h i r t een t h -c en t ur y date ( An d r e , 19 74 , 32-3 5 An d r e , 1982, 168-9); see fig. 6, H145-1, 2, 3.

A thick (40cm) layer of 'Pplough soil (2) accumulated over the filled-in pit and the rest of the trench. The pottery suggests that this Dccurred during the post-medieval period for 13 sherds of post-medieval pottery were found; 39 médiéval sherds (28 of group 1, 3 of group 3 and 8 of a new fabric) were also recovered, their number increasing with depth - which may suggest that they came from the underlying levels» This layer also contained numerous rock fragments, including 18% roofing slate and 3.154 of the harder stones used Ic^cally for building (14% congl omerate, 11?/» quartzite and 6% qres) . The area appears ta have been devoted to rubbish disposai during the médiéval period; it is possible that the pit, or perhaps tree hole, was filled with materi al from a building.

Comment The results of thèse excavations are extremely useful. The absence of archaeological subsoil features in H145 supports the view that the surface pottery scatter is the product of manuring» It is difficult to interpret the remains of the wall and hardstanding in T12. They were in ail probability connected with the buildings recorded in the ancien cadastre, but the lack of a return to the wall, and the petering ont of the hardstanding, argues against them being part of the actual structures a boundary wall and yard are more likely. It is unfortunate that neither produced dating évidence.

The pit in T10 is important because it shows that that part of Quoiqueneuc was occupied by the thirteenth century; hence, the buildings recorded in the ancien cadastre on the edge of H145 did not represent a post-medieval extension of an earlier settlement., but an aspect of the médiéval settlement pattern i tself. K446 (Ruffiac ZA106)

K446 is si tuâted on the eastern outskirts of Beculeu 1 n Ruff i ac

Although road changes make corrélation difficult, it seems that a t. t h e ti me o f the ancien cadastre K446 was part o f an arable field, close to two small curtilages, which may lave been abandoned earlier house plots. (The cadastral plan of Beculeu as a whole shows a great nurnber of thèse empty curtilages and suggests that the settlement of Beculeu had once been considerably larger.) The possible house sites, together with the surface prehistoric material, made the field a good candidate for sample excavation.

Since there were no obvions concentrations of surface material, it was decided to eut by machine three i-metre~wide trenches, 30m aipart, down the si ope of the field (from east to west s T23, T24, T25), such that the areas producing Iron Age pottery could be tested. The si des of the trenches were cleaned by hand and recorded; the lowest part of the plough soil was excavated by hand, as were the features located thereby. The trenches were extended whèrever it was necessary to establish the character of features, and in two cases (T26, T27) this was done by machine. Features proved difficult to identify because they were often slight and because the natural subsoil and bedrock varied so much. In ail approximately 52 working days were spent on site and the average size of the team was six.

Trench 25 This was the most westerly trench and was 39m long. The depth of plough soil varied from 36cm at the south, uphill, end to 1.10m lower down the valley side. The plough soil (1, 36) overlay the natural clay subsoil (37), into which features had been eut. A small pit (35) was excavated in the south end of the trench, and a séries of uneven holes, interpreted as root holes (40), were also found. The remaining features consisted of two irregular spreads of charcoal (39, 43), which produced no finds. However seven abraded body sherds of Iron Age pottery were recovered from the surface of the .natural.

Trench 24 T24 was 81m long, and the depth of the plough soil ranged from 15cm at. the top to 1.0m at the bottom of the si ope. The clay Subsoil and schiste Iay i mmed i ai t e 1 y b e ow the plough soil, and s e v e r a 1 f e a t u r e s had been eut into this natural (45). At the southern end a v shaped diteh, 60cm wide and 35cm deep (44), produced two médiéval pottery sherds (fabric group 1). The most K446 prominent feature was a ragged v-shaped ditch, 1.3m wide and 40cm deep (46), which followed the slope down the valley side -for at 1 east 90m; subsidiary machine euts (T26, T27) located the ditch on the saine alignment and f ai 1 ed to produce évidence of a tujrn. Both ditches contai ned small quanti ti es of non-local qres and quartzite fragments. The fill was homogeneous and produced 3 médiéval and 7 Iron Age sherds. At the north end of T24, where the greatest depth of plough soil had sealed some shallow features eut into the natural, were located two small gulleys (49, 52/53), whose fill contained only Iron Age pottery (3 and 4 sherds respect!vely), and the end of a small ditch or pit ( 50 ) .

Trench 23 This trench was 42m long, and for the most part the plough soil (varying in depth between 25cm and 45cm) rested on the undisturbed natural subsoil and schi ste. In five places an apparent 1 y linear arrangement of quartz blocks set in natural clay was found (12/13, 20, 23, 28, 29); excavation showed that they were probably formed natural1 y.

The archaeological features in the trench concentrated in the southern (highest) end, nearest the settlement of Beculeu. A shallow (15cm) pit rnay be dated by the single médiéval sherd in its filling (5), and may be associated with other features which apparent1 y constitute a roadway. A deep (2m) ditch or holloway (66) had either been eut or worn into the natural schi ste. Dver the bottom had been laid a layer of compactée! schi ste in which there was at 1 east one rut. (64). On this surface accumulated a layer of silt (63), which was in turn covered by soil which had been tipped in from the south edge of the holloway (61, 62). A new surface of seh iste fragments had then been laid (60) , which was in turn masked by a thick layer of silt (58). Lenses of small seh iste fragments (57, 59, 56) show that attempts were made to consolidate small areas of the road surface. Further silting occurred (22), which filled completely the dépression of the holloway. 6 médiéval sherds came from this layer. A macadamised surface of quartz and bitumen was then laid over the filling of the holloway, and for a considérable distance to the south so that the road was virtually doubled in width (18, 21, 55). The surface had been fragmented through plough destruction.

Five mètres south of the holloway a ditch 1.5m wide and 45cm deep was excavated (.1.0); its filling contained one médiéval sherd of pottery (7, 8, 9). It was apparently on the same alignment as the holloway and had silted up by the time the macadam road surface had been laid over it. The ditch may have acted as a drain or boundary for the holloway.

T26 was eut to the north west of T23 to check that thèse features constituted a road: the same séquence of levels was recorded.

Comment The résulte of thèse limited excavations are difficult to interpret. The lack of clear médiéval features would suggest that the 1ow and even scatter of médiéval pottery recovered in fieldwalking was not produced by occupation on this field. The density of négative features that are associated with Iron Age

<2o pottery are, however, likely to be an indicator of occupation (see also HSO) . In fieldwalking the ' greatest amount o-f this pottery came -from the areas eut by T24, the trench that produced the most convincing prehistoric -features. The long ditch in this trench could be one si de o-f an enclosure, but the -failure to -find a turn would argue against this, and may suggest that it was eut merely to drain the land. The mixed date range o-f the pottery ■from this ditch makes it impossible to suggest a firm date for i t.

The évidence for a holloway, replaced by a métalled road, is particularly interesting. It demonstrates that the routes into Beculeu have changed. More importantly, the pottery from the holloway and associated ditch gives an indication that the earlier route was in use in the médiéval period» Thin section analysis of the siIts in the holloway wi1i provide an interesting comparison with those samples taken -from the lower plough soi 1 s in T24 and T25, which may be expected to be colluvial in charaicter.

^1 H80 (Treal ZN4b)

1-180 is located right on the western edge of Treal commune, some 150m east of La Hattaie (fig. 4). The field is in a low-lying position and si opes very slightly down to the north. The ancien cadastre shows that H80 was an area of mixed land-use, parti y cultivated in strips but also with 1 andes (uncultivated) and pasture, in small enclosures; field names (landes and fri ches) over the whole area indicate that it had not been characteristical 1 y cultivated before the 1820s.

The field was first walked in 1984 and was classified as a 'possible site' (médiéval). In April 1986 it was intensively walked and this produced a 1ow and even scatter of médiéval and post-medi eval pottery; small quanti ties of Iron Age pottery were also recovered (not more than two sherds from a five-metre square), most coming from the western third of the field.

Médiéval and post-medieval pottery distribution suggested a manuring scatter; that of the Iron Age was possibly derived from occupation (cf. A116). The aim of the excavation was therefore to test. thèse suggestions. The even nature of the médiéval pottery distribution meant that there was no obvions area for excavation, and so the first trench was placed over one of the 'concentrations' of Iron Age pottery. Initial ly ai 6m square (T15) was excavated as at Al 16, A92 and H145, and then five trenches (1 mètre wide) were eut north-south across the field at 30m intervais using a machine (from west to east. T16/17, T18, T19, T20, T21). The sections were cleaned by hand and recorded; the remains of the plough soil were removed, and the features thus revealed were excavated, by hand» Extensions were made in some trenches in order to clarify the character of some features, for example to discover if a négative feature was a ditch or pit. A machine was used to remove the depth of plough soil in thèse cases, as it ailso was to check the f armer's cl ai m that a building existed in a part of the field that otherwise would not have been sampled (T22). In ail 87 working days were spent on site, with an average team of 6.

The majority of features excavated were sections of ditches. An account. of al 1 thèse would be repetitious; instead the détails are tabulated after a discussion, by trench, of the other contexts. Al I the ditches had been dug into the natural clay subsoil and se h i ste (3) , and were filled by soil which had ai chairacter similar to the plough soil, but slightly more compact. The location of the ditches is shown on the gênerai site plan.

Trench 15 The plough soil in this 6m square was excavated in 10cm spits. The détail* of the pottery excavated, compared with that recovered by field walkinq, is tabula\ted below» The increase of prehi stor i c material with depth would argue that. it was derived from subsoil features of that. daite. Rock fragments, of which 307. were roofing slate and 507. soft shale (Bri overi an ) , were more prolific in the upper levels and are likely to dérive from plough scatter »

aa

Context 1 (plough soil) No. Wt. on sur-face 2 prehistoric .005 kg 1 médiéval „025kg 1 post médiéval .002kg 0-10cm 2 prehistoric .016kg 7 médiéval .078kg 23 post médiéval .072kg 10--20cm 1 prehistoric „ 003kg 1 médiéval .021kg 8 post médiéval .030kg 20-30cm 10 prehistoric .074kg 1 médiéval .009kg 5 post médiéval .088kg

Cut into the subsoil was an irregular curving ditch, 50-70cm wide, 50cm deep (4). One possible interprétation of this feature is that it represents the eaves trench, 16.8m in diameter, of a circular structure. No internai features were located in T15 however, nor was the other si de of the 'eaves trench' located in T17, although T17 could have gone through a causewayed entrance. Alternatively, context 4 might have been a sinuous drainage or boundary ditch. The filling of this ditch produced 17 sherds of Iron Age pottery, one a rim (see fig. 6, H80-8), and 4 pièces of tile.

Trench 17 This trench was the only one to produce évidence for superimposed ditches, ail three of which contained Iron Age pottery.

Trench 16 4.5m south of the ? eaves trench in T15 a collection of at least 4 intercutting pits was excavated,, Thèse may have been dug to extract clay. Ail the pits seem to have been filled at the same time. 4 separate layers were identified; thèse produced the greatest amount of prehistoric material from the whole site. The 1owest fill (45) produced 11 pièces of tile (0.365kg) and 8 sherds of Iron Age pottery (0„120g). Layer 57 overlay 45 and produced no finds. 22 overlay this and contained 60 pièces of daub (0.168kg), 30 fragments of tile (0.632kg) and 38 sherds of Iron Age pottery (0.396kg), some of which are rims (fig. 6, 1-180-3, 5 and 10). This layer also contained several large, f1at slabs of Cambrian siItstone like that obtainatale from a narrow band 1.5km to the north of the field. The final layer of thèse pits (7) contained the most material: 72 pièces of daub (0.153kg), 39 Iron Age sherds (0.293kg), including rims (fig.6, H80-1, 4, 6, 7, 9) and 89 pièces of tile (1.718kg). The latter are parti cularly interesting. they are of a low-fired clay which has large inclusions of quart:-: (3~7mm) , and at. least one edge has a chamfer (see fig. 6, H80-ii). Ail the tile recovered from the site is of this type. Two pièces of slag also came from this 1ayer.

This feature? produced the best évidence of settlement from the site; the pits may have been originally dug to obtain clay for daub, and then backfilled with occupation débris. The pits were cut through by a ditch which contained modem glass (56). Rock fragments from the fill of one ditch (8) were of parti cular interest by comparison with those recovered from T15. they included several large blocks of hard stone (quartz, quartzite, altered schiste, greywacke) , smal 1er fragments of hard stone (including grès) but no black roofing slate and negligible amounts of soft shaïe.

Trench 18 This trench produced few features, but in the southern part a small ditch (10) had been dug through a pit, approximately 34cm in diameter and 60cm deep- It had filled u.p with clay, probably as a resuit of silting (53), but at the bottom there was layer of ash and charcoal mixed with burnt clay (54). It produced no finds, but a comparable feature was recovered in Ti9 (19). Another pit (12) was also located to the north (75cm diameter, 45cm deep), and although the fill did contain some charcoal there was no évidence of burning in si tu as with 53.

Trench 19 Three ditches were excavated in this trench (15, 16, 20)p another passible ditch was also located, but it had been substantial 1 y modified. The ditch seems to have been extended to croate a larger and deeper pit (19). The pit base contained much charcoal and ash, and the natural schi ste at the bottom had also been burnt (61). The pit was then allowed to silt up, a layer of clay with some charcoal accumulated (60), followed by perhaps a more deliberate backfilling with more clay (58) and a mass of burnt stone which had been thrown in from the south (59). Thèse were nearly ail hard blocks of conglomerate (387.), quartzite (277.), greywacke (147.) and quartz (107.). The stone may have been used when the pit was used for firing, piled on to the south si de, and then pushed back. No finds were recovered from this feature. Half a mètre to the north a pit containing burnt clay fragments was excavated (18) and may have been associated with the pit 19.

Trench 20 Five ditches were excavated in this trench (33, 34, 35/50, 37/51, 40), and the détails are tabulated below. Ditch 40 included some blocks of hard stone similar to those from T19 (59). In two cases the ditches appear to have been reçut (35/50, 37/51).

Trench 21 Four ditches were excavated (41, 42, 47, 49), détails of which are given below. The only other feature wais a modem tree hole (46) ". Trench 22 An irregular area Of approx i matel y S by 4m was cleared of plough soil and a deposit of hard stone blocks waswas foundfound, mixed with clay which contained modem glass and a sherd of post médiéval pottery (62)■ Half of thèse blocks were quartz, a tenth were quartzites of a type obtainabie 3km to the north and another tenth were siltstone of a type obtainable 5km to the south. The stone seems to have been piled into a large irregular pit which had been cut into the natural subsoil (48). It muet represent. collapse from some structure in the vicinity, but the cadastral map shows no trace of any structure or track in the area in the early nineteenth century and nothing suggests that the structure was of greater antiquity than that.

Ditch détails IA = Iron Age pottery No. Width Depth Shape Width of Finds & Comments (m) (m) bottom illustrations T15/4 .50-70 .50 ^ J .20 17 IA(H80,8),4 tile

T17/26 . 85 . 78 T17/25 . 80 .50 2 IA, 2 tile T17/52 2-2 . 48 T17/27 . 70 . 70 T17/29 . 78 . 90 1 flint flake

T16/8 1 50 1.0 ., 10 7 IA,1 tile, 8 daub 3 f i 11 s T16/9 . 58 . 40 T16/56 . 70 . 68 modem qlass 2 fi1ls T16/23 1. 70 . 75 T16/24 . 30

T18/10 47 . 60 T18/11 37 .54

60 . 46 ? terminal T19/16 58 . 51 T19/20 30 .51

T20/33 45 „ 60 T20/34 76 .71 T20/35 04 .51 reçut of T20/50 44 T20/51 36 . 45 . 16 T20/40 .1. 30 .70 i 1 Is T21/41 1.13 T21/42 1.75 T21/47 "21/49 1.0

Comment The lack of certain médiéval features would encourage us to regard scatters of the same character as H80, and indeed H145, as a p r o d u c t. o f m a n u r i n g.

The évidence that. has corne from K446 and more certainly H80 would suggest that small surface scatters of prehistoric pottery

<2

While T15 and T16 have produced settlement data, it is more difficult to interpret the features in the other trenches» Some (e.g. 56, 48, 64) resuit from récent activity in the field, but. most of the ditches cannot. be dated. The façt that ail the profiles of the ditches in each trench (with two possible exceptions - T19/20, T20/33) are différent implies that they were not continuons boundaries or drainage ditches» The anci en cadastre shows part of the field to have been divided into narrow bands, but the excavated ditches neither run parai 1el nor the fui1 length of the field» They are therefore unlikely to be of a médiéval or post-medieval date and may possibly be contemporary with the 'prehistoric settlement» Some of those close to the pit group 7 do have small quantities of prehistoric material (8, 25) but others in the same trenches do not» This might imply that the disposai of rubbish was confined to the settlement area, and therefore it would be difficult to show that outlying parts were contemporary» The absence of prehistoric pottery from the areas around thèse ditches may imply that they do not enclose other parts of the settlement»

The occurrence of pits beyond the area of settlement that have been used for burning might argue for spécialised activities taking place within small enclosures beyond the residential area» One possible interprétation of H80, then, is that it was the site of a late prehistoric settlement, the focus of which lay in the west of HSO and beyond it, with associated enclosures to the east » A92 (Ruffiac ZK67a)

A92 is very close to Les Landes de la Ruée on the crest o-f a ridge some 700m from La Hattaie (fig. 5). At the time of the aneien cadastre the field was un d er arable cultivation and was on the periphery of the ' c h la t eau landscape' of the Château de la Ruée to the south»

This field was classified as a 'site' after wide—spaced fieldwalking, on the basis of a tile scatter, and was thought to be part of a settlement. that spread over the neighbouring fields which were also walked (A93. , A93, A94) » The sporadic finds of Roman pottery, to the exclusion of material of other periods, suggested the site might be Roman» In February 1986 intensive surface collection was carried out, and in September excavation was started on two 6m squares. The method of excavation was the same as that used on Al 16, the position of the excavated areas being determined by the resuits of intensive collection» One 6m square (T28) was in the area which produced the most tile, the other, some 15m away, was placed over the highest concentration of Roman pottery (T29)»

It was not possible to complète the excavation of the trenches in the time avallable, so work was stopped at a suitable level» In ail some 28 working days were spent here with an average team of ten people» We hope to return to complète excavation in 1988.

Trench 28 The plough soil, 20cm deep, was excavated in 10cm spits. A high quantity of tile was recovered, which was sorted according to fabric and shape (i.e. tegulae, i mbri ces and flue tile). Three f abri es were i dent, i f i ed s group 1 was highly micaceous and contained much quartz, varying in size up to 2mmp group 2 had no mica but fréquent iron oxide inclusions; group 3 was similar to group 1, but. had larger quartz inclusions and was more highly fired» The gross détail of tile recovered is as followss

Context. i (plough soil) No. Wt. on surface 155 8.96kg 0-10cm 486 19.745kg 10-20cm 475 32.34kg

385kg of hard stone blocks were recovered, of which half were quartzite, 147. congl omerate, 13.7. qres and 87. hard Brioverian shale; there were negligible quantities of roofing slate. 25 sherds of Roman pottery were also recovered and 2 sherds of médiéval (8 Roman and 1 médiéval in 0-10cm; 17 Roman and 1 médiéval in 10—20cm. No pottery had been found in intensive fieldwalk i ng)„

There appeared to be a lower plough soil at. 20cm (context 2); its surface was plough scarred in the southern third of the trench, and it is likely that ploughing has destroyed much évidence. In the northern two thirds there was a dense layer of broken tile and building stone which looked like the destruction layer of a building. The total number of tile fragments recovered from this context was 775, weighing 95.65kg. A complète tegula was found, and this weighed 5»5kg» Counting al 1 the tegula fragments and f eaturel ess pièces of tile the destruction deposit rep resent s approx i matel y 18 complète tegulae. 335kg of stone were also recovered, in comparable proportions to that -front content 1, except that there was negligible hard shale, 107. quarts and a higher proportion of grès. There was no roofing slat'e.

Conte;;t 2 also appeared to -fill a ditch—like feature which was defined but not excavated (10). It was of the same size and alignaient as the ditch found in T29. The ditch in T2B seemed to have been dug through a 1 ayer ot compactée! schi ste fragments which was deliberately laid as a -floor or hardstanding (9). This underlay the destruction 1 ayer o-f tiles and stone; it has yet to be excavated.

The pot ter y recoverec front bot h contexts appears to have a similar range of types and forms. Most are coarse grey wares with riiïts which suggest a first- to second-century date (see f g ■ & i A92-1, 2. 4), but there are sherds of central Gaulish samian which have a mid second-century date ; i g. 6, A92-6).

Trench 29 The plough soil (3) was removed in two 10cm spits» As the fieldwalk i ng data suggested, it contained 1ess tile than that of T2S:

Dontext (plough soi 1) No Wt. on surface 113 2.7kg 0-10cm 197 9. 6kg 10—20cm 216 5.4kg

The unfeatured pièces and tequla fragments r ecovered were the approx i mate équivalent of 2 6 complète tequlae. The d i str i but i on of tile through the p1ough soil was also différent, for neither the amount nor the size of fragments increased with depth. Quantities of stone were also far smaller than in context 1 of T28 (2.6kg), though the proportions were precisely the same as found there. Indeed there was no sign of a destruction 1ayer in T29, which < Suggest s that this area was further front the buildings, and that the tile and stone had been scattered there.

The pot ter y however was consistent with that front T28. The plough soil produced 2 médiéval (fabric 1), 3 post-medieval and 34 Roman sherds. The intensive surface collecton had produced one sherd each of médiéval and post-medieval pottery and 13 Roman sherds. The majority of the Roman pottery was from coarse wares in a grey fabric with rims which are characteristic of the first and second centuries (see fi g. 6, A92-3, 5, 8), but there was soute central Gaulish samian of the first and second centuries, and sente terra—niara-1 yp e vesse1 s.

E-felow the plough soil was a more compac soil (4) which was -free of tile and had only 0.9kg of stone, and appeared to be similar to context 2 in T28. A third of the st.one5 were siltstone, and they included little quartzite and no roofing slate. The surface of this 1ayer also produced 11 first-century sherds, including a flagon base and a pièce of central Gaulish samian. There was sorte évidence that this 1 ayer had been dug into in modem times, for a pit contai ning charcoal and glass was found. More important!y a ditch was defined which had been dug into context 4. It wa\s only possible to excavate a narrow, half~metre~wide, section, but this showed that the ditch was 1.5m wide and 0.5m deep. It had steeply sloping si des wi th a narrow flat bottom, and produced 17 first- to second-century sherds, mostly of grey coarse wares (see fig» 6, A92-9), but there was a rim of a terra-niqra-type vessel (fig. 6, A92-7). The same ditch may also have been located in T28. It is important to return to complète the excavation of this trench.

Comment The small amount of work done on A92 demonstrates the potential of the site,, While it h as not been possible to show conclusively that the tile scatter represents the remains of buildings, the distribution and volume of the tile alone would make it difficu.lt to escape this conclusion. Even this limited excavation encourages us to interpret as Roman settlements the tile scatters with small amounts of Roman pottery which have been located in the course of fieldwalking„

The narrow date range of the pottery is also interesting. Re-ex ami nat i on of the pottery front both the wide-spaced and intensive fieldwalking has failed to show any Roman pottery that is défi ni tel y later than the second century. This is also true of the material from the neighbouring fields, which alrnost certainly enclose the rest of the settlement (A91, A93, A94). It is therefore unlikely that this settlement continuée! to be occupied in the later Roman period.

3\ H80

3& w~ork ontsi de the main season, 1985-86

During the past year work has continuée) on a number of other aspects of the project, both field and archivai. Mi cromorphcDl ogi cal analysis of some soil s from the core communes has been^ begun by Anne Gebhardt under the supervision of Marie-Agnes Court.y in Paris. Soil samples taken from 1985 'total' collection sites for phosphate analysis have been processed in Reading and resuits have yet to be correlated with pottery distributions from those fields. Samples taken from the ditch of Tl and from low-lying land by the river Aff for pollen analysis in 1985 have also been processed in Reading. Analysis of pottery f abri es has continuée), the sorti ng and clasification of pottery from the 1985 season being completed; further fabri es have been identified, the fabric séries now comprising 20 fabric groupe. Work on local quarries has allowed the provenancing of local schi ste types; those imported from outside the study area can therefore be distinguished so that some sensé of the movement of 1 ocal 1 y-occurr i ng schi stes; for building and roofing purposes is being established. Détailed reports on ail this material wi11 be ready in 1987-8.

During the months of February, March and April 1986 a team of four worked full-time on intensive surface collection within the core area, adding a further thirteen fields to the existing thirteen from which 'total' collection has- been made. The survey of standing buildings in the four communes, undertaken throughout 1984-5, was completed in October-Decembers 4589 buildings have now been recorded in détail, on standardised recording forms; this archive, together with 795 photographs, was deposited in Uni versi ty Collège London 3. ibrary in January 1986. Between January and June 1986 détailed démographie mater i al relaiting to the core communes was collected; this involved the extraction of relevant data from ail parish records from the 1540s until the 1790s5 it will allow a simple baptism/marriage/burial count for ail four communes and detailed fami 1y reconstitution for Treal. (Post 1790s démographie data is already available.) Since June 1986 transcription of land-use information from miscel1aneous thirteenth- to eighteenth—century archives in , Rennes and Nantes has been in hand, and should be completed by early September. The information generated by ail thèse major areas of the project will be collated with data from the annual fieldwork seasons during the year 1987-8. General Comment Three of the five sites excavated this year have now produced évidence of later prehistoric occupation (Al 16, HSO, K446) despite minimal surface indication. While it would be prématuré to généralise, it is noticeable that ail three sites are located in low-lying positions, in two cases near the bottom of small valleys (A116, K446). None of thèse sites has produced any notable quantity of Roman material, though the date range of the fabrics from H80 is likely to span the first centuries BC and AD and cover the early Roman period. By contrast, the site at A92 is clearly Roman, though apparent!y limited to the first and second centuries AD. Fieldwalking over the last five years has located other pot en ti al Roman sites on exposed ridges, in positions similar to that of A92. The proximity of this field to HSO (800m) must suggest that the early Roman period was a time of pronounced settlement relocation, in which occupation of low-lying sites gave way to that of the higher ridges.

The Roman site at A92 is also notable for the collection of stone in its building destruction Iayer. Thèse were hard stones, from the région but. mostly not from the immédiate vicinity, including a large proportion of quarts!te and none of the roofing material that occurs abundantly on the surface of fields in the study area. The late Iron Age ditch fills in T16 on HSO (as also the Liridated fills from T19 and T20) are comparable, though containing 1ess quartsite and more sandstone. Ail this is qui te distinct from the building tradition of the sixteenth ta nineteenth centuries, as evidenced by standing buildings, which predominant1 y used irregular blocks of the immediately local soft Brioverian shale and siltstones from the north and south periphery of the study area, with some conglomerate and quartz blocks interspersed. The mixed fragments of H145 and of T15 on HSO, as also the blocks from T22 on HSO, would not be at ail inconsistent with destruction material from buildings of this latter date.

Both s.easons' work were undertaken with the authori sati on of the Ministère de la Culture, Direction des Antiquités de Bretagne, and many thanks are due to M. Le Roux, director of the ci rconscri pti on, and to the conservateur M. Clément and his assistant Michael E

18 December 1986

REFERENCES

André 1974 site médiéval de Kerlano en Plumelec Arc h eoV. 1 oq i e en Br et a g n e . i i , 27/-34. Andr^ P. 1982 MJn village médiéval breton du onzième siècle Archeologi e Medieyale, xii, 155—74. Astill, G. and Davies, W. 1982a 'Un nouveau programme de recherche sur 1 e terrain dans l'Est de la Bretagne', Arch. en Bretagne , ;■; x ;•; v , 24-42. Astill, G. and Davies, W. 1982b 'Fieldwalking in East Brittany, 1962 ' , Camb r i d g e Med i e va 1 Ce 1 tic St. ud i es - i v , 19-31. Astill, G. and Davies, W. 1983 'Recherches sur le terrain dans 1'Est de la Bretagne (EBS) - 1983 Arch . en Bretagne

Astill, G. and Davies, W. 1984 'Prospection dans l'Est de la Efretagne. Résultats de? la campagne de Mars-Avril 1984', Dossiers du CRAA, xii, 49-59. Astill, G. and Davies, W. 1985 'Prospections archeol oçj i ques dans I 'est de la E