Veget Hist Archaeobot (2007) 16: 349–365 DOI 10.1007/s00334-006-0044-z

ORIGINAL ARTICLE

Conor Newman · Michael O’Connell · Mary Dillon · Karen Molloy

Interpretation of charcoal and pollen data relating to a late Iron Age ritual site in eastern Ireland: a holistic approach

Received: 9 August 2005 / Accepted: 3 January 2006 / Published online: 26 April 2006 C Springer-Verlag 2006


Abstract The study of charcoal produced by five burning episodes that occurred in a rapid succession within a ritual pit dating to the late Iron Age at Raffin Fort, Co. Meath, Ireland, reveals considerable variation in the charcoal assemblages resulting from each burning episode. Wood selection processes are considered against the background of information on woodland composition and land-use history provided by a detailed pollen diagram from nearby Emlagh Bog, the chronology of which is based on both AMS 14 C dates and tephra analysis. A human skull fragment lay on top of the charcoal layers but the radiocarbon evidence indicates that the skull predated the burnings by at least a century. This and other evidence indicate a ritual pit with the skull as a human relic. It is suggested that, in this instance, wood selection was neither random nor determined solely by availability or combustibility, but instead may have been informed by socio-religious belief systems pertaining to trees and wood. Early Irish documentary sources, which reveal a complex ethnography of wood and trees in later prehistoric and early historic Ireland, are reviewed. The results shed fresh light on aspects of late Iron Age archaeology in a part of Europe that was outside the direct influence of the Roman world. New information is provided on a distinctive feature in late Holocene Irish pollen records namely the Late Iron Age Lull (ca. a.d. 1–500). During this time, widespread regeneration of woody vegetation took place. In the subsequent early Medieval period renewed farming activity resulted in substantial decline in woodland, a pattern also seen at many other locations in Ireland.

Keywords Archaeobotany . Charcoal analysis . Pollen analysis . Iron Age . Ethnography . Ireland

Introduction Early Irish documentary sources (e.g. Lucas 1963; Kelly 1976, 1999) indicate that wood and trees were culturally charged phenomena in Ireland, as in most other countries (e.g. Frazer 1922; Yasuda 2001), during the later prehistoric and early historic periods. More than simply an exploitable, economic resource, trees were invested with ritual and totemic significance. This suggests that wood remains recovered from archaeological contexts have the potential to yield insights, not only into composition of woodlands and hence local environments, but also the cultural values obtaining in early Irish society. This paper assesses this potential through the analysis of charred wood remains from successive burning episodes in a late Iron Age ritual pit at Raffin Fort, Co. Meath. It also considers if, in this context, there is evidence for event- or context-specific wood selection. The charcoal data are evaluated against the information on regional woodland composition provided by a new pollen diagram from nearby Emlagh Bog. By combining the results of charcoal identification from the archaeological site and pollen analysis from nearby, a tangible record is obtained of the use of local environments for economic purposes and especially farming, and also the possible roles of ritual in the exploitation of local wood resources.

Communicated by F. Bittmann C. Newman Department of Archaeology, National University of Ireland, Galway, Galway, Ireland M. O’Connell (
) · M. Dillon · K. Molloy Palaeoenvironmental Research Unit, Department of Botany, National University of Ireland, Galway, Galway, Ireland e-mail: [email protected]

The archaeological background Raffin Fort, which lies 9 km north-east of Kells at 100 m a.s.l. (6◦ 45.4 W, 53◦ 47.2 N; national grid ref. N 8202 8278; Fig. 1), is a multi-period site with archaeological features dating from the Neolithic to the Early Historic period (ca. 4000 b.c. to a.d. 600; Newman 1993a,b, 1995, 1998). The monument is on a large drumlin that lies at the

350

Fig. 1 Topographical map centred on the river Blackwater/Boyne catchment. The higher land is at approximately 300 m above sea level. The locations of Raffin Fort and the coring site at Emlagh Bog are indicated (closed circles) as well as selected archaeological sites (lozenges) Fig. 2 A Photograph showing the standing stone that originally sealed the ritual pit, re-erected on site. In the background (view to the east) is the rolling, fertile countryside of north-east Co. Meath. The linear scratches on the stone are from the toothed bucket of the bulldozer; B The ritual pit during excavation (view towards north east). Remnants of the burnt soil (stained red) are visible on the sides and bottom of the pit (photographs: C. Newman, 1993)

southern edge of the Meath/Cavan drumlin belt. To the east and south, the landscape is characterised by a scatter of comparatively low hummocks of glacial deposits (Meehan and Warren 1999; Fig. 2A) and to the west lies the broad alluvial plain of the Blackwater valley. The site lies 3 km south of Moynagh Lough, the last remnant of a large lake that was drained in the 1800s (Bradley 1991). The land is fertile but there is a rich mosaic of diverse habitats and resources that includes bogs and lakes, especially when considered at the regional level. Excavations, directed by C. Newman, began in 1989 following the partial destruction of the monument. The main feature consisted of a circular enclosure ca. 65 m in diameter, defined by a bank and internal fosse that was interrupted by a causewayed entrance in the south-east quadrant (Moore 1987; Fig. 3). It was constructed between the early 4th and the late 6th century a.d., that is during the late Iron Age/early Medieval transition (14 C dates presented in Table 1). A 9 m diameter ring-ditch in the centre was surrounded by a 15 m diameter circle delineated by six free-standing posts. The concentricity of these three elements, i.e. enclosure, timber circle and ring-ditch, together with the radiocarbon dates, suggests structural contempo-

351

Fig. 4 South-facing section of the pit. Each charcoal layer represents a burning episode (BE 1–5). BE 3 (C5261) comprises a mixture of burnt soil and charcoal. The skull fragment and standing stone have been projected unto the section. The boundary between C5214 and C5215 was indistinct and hence is indicated by a dashed line

Fig. 3 Outline plan showing the Iron Age features at Raffin Fort. These comprise an internally ditched enclosure, a 9 m diameter ringditch, surrounded by a timber circle (post holes indicated) and the burial pit from which the charcoal was recovered

preserved the pit sufficiently to enable the main features and contents of the pit to be recorded and sampled satisfactorily. The late Iron Age ritual pit

raneity (Newman 1998). In the northern quadrant of the site, 5 m inside the enclosing fosse, a contemporary ritual pit was located. This was originally marked by a large stone that was probably once upright and hence is referred to as a standing stone (Fig. 2A). These features, taken as a whole, strongly suggest a ritual or ceremonial arena or sanctuary, and have been interpreted as such by the excavator (Newman 1993a,b; final excavation report in prep.). Part of the burial pit was destroyed during land reclamation in 1988. Fortunately, the large stone that lay recumbent over the pit, limited the damage caused by bulldozing and Table 1 14

The pit, which was 40 cm deep and a 90 cm diameter (maximum), had shallow concave sides and a slightly rounded base (Figs. 2B and 4). Successive layers of charcoal and soil, and the scorched red colour of the sides and base, indicate that a series of intense burning episodes took place in the pit. The fill comprised layers of burnt soil and charcoal, and towards the top of the pit fire-cracked angular stones (3–7 cm) many of which were reddened from burning. Five of the eight layers contained a high proportion of charcoal. The lowermost two layers of charcoal were

Details of radiocarbon dates from Raffin Fort and Emlagh Bog (EML I), calibrated using Oxcal 3.10

C lab. no.

Age (B.P.)

Cal age (2σ range)

Material dated

Context

110 b.c.–a.d. 140 a.d. 250–470 a.d. 240–440 a.d. 210–400

Human skull bone Charcoal (hazel) Charcoal (hazel) Charcoal (non-det.a )

Skull fragment BE-5(C5215) BE-2(C5283) BE-1(C5290)

a.d. 380–570 50 b.c.–a.d. 700

Charcoal (non-det.a ) Charcoal (non-det.a )

Enclosing fosse Ring-ditch

110 b.c.–a.d. 90b 750–150 b.c.b 920–540 b.c.b

Mainly Sphagnumc S. imbricatum S. imbricatum

180–181 cm in EML I 198–199 cm in EML I 224–225 cm in EML I

14

AMS C dates from pit charcoal AA-10281 1975 ± 50 GrA-25763 1670 ± 40 GrA-25764 1680 ± 40 UB-4004 1753 ± 32 AMS 14 C dates from other archaeological contexts OxA-6711 1595 ± 45 UB-3435 1676 ± 168 AMS 14 C date from core EML I GrA-24121 2000 ± 40 GrA-21333 2310 ± 60 GrA-21346 2630 ± 60 a

Non-determined; sent for dating prior to current wood identification Mid-points of 1σ range are as follows: a.d. 5, 340 b.c. and 785 b.c., respectively c Mainly stems and leaves of S. imbricatum and some S. sect. Acutifolium b

352

separated from one another by a narrow lens of mineral soil that indicates relining after the initial burning. Above these, layer C5261 contained an admixture of charcoal and burnt soil and is, therefore, interpreted as representing both re-lining and burning. The uppermost deposit (C5215 and C5214) comprised a concentrated and comparatively thick deposit of charcoal on which rested a skull fragment (see below). These layers were not distinguished during excavation although the charcoal fragments were more dispersed in the lower part of the deposit (C5214). The differences in the charcoal assemblages justify making a distinction (see below) and so it is concluded that two burning episodes are represented. In summary, there were five distinct and sequential burning episodes represented in the pit, referred to as Burning Episodes (BE) 1–5 (Fig. 4). A fragment of the frontal cranium of an adult human, possibly male, was discovered in the uppermost layer of the pit (L. Buckley, unpubl. report). This lay on a bed of charcoal and angular, heat-reddened stones, and was partially crushed by the weight of the standing stone bearing directly upon it. Alongside the human remains were fragments of a possible animal rib and pelvis. The skull fragment was worn smooth around the edges which suggests frequent handling. This, and the absence of rodent or animal teeth marks, raise the possibility that the skull may have been curated. This is supported by the results of 14 C dating (see below). The idea that the recumbent stone that covered the pit was once upright is supported by the finding of an irregular arc of eleven stones – probably propping stones – on the top of the pit. The inescapable conclusion is that the stone was erected over the pit to mark, inter alia, the location of the human skull. This, together with the evidence for curation (see also below), implies that this is not a burial in the conventional sense but rather a ritual pit containing a human relic. Curation and deposition: the date of the pit and its contents Charcoal samples, representing three of the burning episodes, and a sample from the skull fragment were radiocarbon dated (Table 1; Fig. 5). The lowermost layer of charcoal (C5290; BE 1) gave a date of 1753 ± 32 b.p. (a.d. 210–400), whereas the second charcoal layer (C5283; BE 2) and the uppermost layer (C5215; BE 5) gave dates of 1680 ± 40 b.p. (a.d. 240–440) and 1670 ± 40 b.p. (a.d. 250–470), respectively (details of 14 C calibration in Table 1). The degree of overlap between these three dates supports the excavator’s original suggestion (unpubl. Interim Report 1994), based on analysis of the stratigraphy, that these burning episodes were effectively contemporary, and occurred between the 3rd and 5th centuries a.d. The skull fragment on the other hand gave a date of 1975 ± 50 b.p. (110 b.c.–a.d. 140). Thus there is no statistically significant overlap between the dates from the skull and those from the burning episodes (Fig. 5). The radiocarbon

Fig. 5 Plot of four radiocarbon dates from the pit (Oxcal 3.10 calibration program). Beneath each curve, two series of lines indicate the calibrated age ranges at 68.3% (1σ; upper series) and 95.4% (2σ; lower series) confidence levels

dates suggest that the skull was at least a century old when placed in the pit, i.e. it was a curated object. This, coupled with the placing of the skull fragment in a pit used for ritual firing and marked with a standing stone, clearly indicates ritual deposition. The skull fragment was probably a religious object or relic for a considerable period; a change in ceremonial practice and iconography is suggested by its removal from circulation and its use in a ritual context. Although commonly associated with the Celts, the curation of human skulls is an enduring and widespread phenomenon, usually explained in terms of ancestor worship, skull cults or trophy collection (Ross 1967, pp 61–127; Cunliffe 1983, pp 164–165). The possibility that human skull fragments might have been worn as amulets or masks has been suggested in cases as diverse as the late Bronze Age artificial pool at the Kings Stables, Co. Armagh (Lynn 1977) and the early Mayan site of Cuello, Belize (Hammond et al. 2002). In Ireland, the cult of the human skull appears at least as early as the Bronze Age and the tradition continued into the Iron Age. Examples from the Iron Age include the deposition of human skulls in a megalithic Passage Tomb at Carrowmore, Co. Sligo (Burenhult 1980), and human and animal remains, including human skulls, have been recorded at R´aith na R´ıg, Tara, which lies 25 km to the south of Raffin Fort (Roche 2002; see also O’Brien 1990; further contexts in Dillon et al. 2006). Combined deposits of human and animal remains, curated or otherwise, clearly expand the iconographical dimensions and potentialities of such contexts. Likewise, botanical remains may also be considered part of the ceremonial iconography. Thus, at Raffin Fort, the charcoal recovered from the pit should not only be regarded as a by-product of conflagration within the pit (which may itself have been an act of ritual cleansing), but the wood itself, and the taxa represented and their general morphology and especially the properties of the sexual reproductive parts (flowers and fruits), may also have had ritual and other significance. In short, the ethnography of the wood must also be considered, given the archaeological context from which the charcoal was recovered.

353

Material and methods Charcoal analysis During excavation each layer of the pit was bulk sampled for charcoal. The five samples that contained charcoal were sieved in the laboratory to separate the charcoal from the matrix, which consisted of small stones and burnt soil. All charcoal fragments of 2 mm or greater were identified. The small fraction (<2 mm), of which there was a considerable amount, was not analysed. Each fragment was prepared for microscopic examination by fracturing it by hand and thereby exposing a clean surface along transverse, radial and tangential planes. All three planes were examined at a range of magnifications ( × 50 to × 500) under a Nikon Optiphot incident-light microscope. For identification, a variety of reference literature was consulted (Gale and Culter 2000; Greguss 1959; Grosser 1977; Schweingruber 1990) and modern comparative material was also used. The number and weight of fragments were recorded for each charcoal type. The following details were also noted: Ring curvature During the analysis ring curvature, which can be regarded as broadly indicative of age, was noted. On the basis of the degree of ring curvature, the charcoal fragments were classified according to age categories as follows: strong curvature, ≤ 5 years; medium curvature, 6–10 years; weak curvature, 11–15 years; curvature negligible, 15 + years, i.e. mature wood. Where the pith was present, the age of the fragment was noted (equal to the number of rings present). Insect and fungi infestation Holes, round and larger than vessels, are usually regarded as caused by burrowing insects; their frequency was noted. Records were also made of charred fungal threads (cf. Asouti 2002, p 141). Ring widths In most instances, average ring width was estimated by measuring the width of the fragment in transverse view and dividing by the number of rings present.

surrounding fertile farmland supports tillage and pastoral farming. Peat depth generally ranges from 6–8 m in the intact bog. At the peat banks where sampling was carried out, the uppermost peat had shrunk by as much as 1 m, though not in a manner to render it unsuitable for palaeoecological investigations. The uppermost peat (EML I) was sampled in three sections (0–87, 87–201 and 201–360 cm). Subsequently, a core, EML II, was taken some 25 m to the east of the initial sampling site and several metres from the peat face (less shrinkage of the upper layers). This core extended from 229–593 cm (all depths are with respect to the original bog surface at the sampling site). By combining pollen spectra from EML I and II, a complete Holocene pollen profile was constructed. For this paper only that part of the profile (derived from EML I) relating to the later Holocene, and which is of direct relevance to the period to which the archaeological and charcoal data relate, is presented. Pollen sample preparation followed standard methods as implemented in the Galway laboratory. This included addition of a Lycopodium spike so as to be able to determine pollen concentration, sieving, acetolysis for 5 min, reverse sieving using a 10-µm mesh to remove fine debris and mounting in glycerol. The pollen sum – a value of approximately 1000 was achieved in most samples – is based on a total terrestrial pollen sum excluding local bog taxa. Cereal-type pollen are based on the criteria of Beug (2004) but only those pollen with size >40 µm are included in the cereal-type curve; Secale-type pollen is presented as a separate curve. Chronology The chronology of the pollen profile as a whole is based on two lines of evidence, AMS 14 C dates from the relevant part of the profile (Table 1) and tephra analysis. Six discrete tephra layers were identified in EML I by ashing 5 cm thick slices of peat and searching the ash for tephra shards (McCausland 2002). The levels at which the tephra layers were noted were subsequently re-sampled by taking 1 cm thick slices. Tephra was extracted chemically, the shards were counted by W. D¨orfler and microprobe analysis was carried out by C. van den Bogaard, Kiel.

Pollen analysis As part of an integrated study into the history of land use in the Blackwater Valley, Co. Meath, a core was taken in the Wilkinstown bog complex. Individual parts of this complex are referred to locally by names such as Fletcherstown, Emlagh and Oristown Bogs. Sampling was carried out in Emlagh Bog near the centre of the original bog complex and approximately 30 m to the north of a minor road where the bog was being actively cut. The sampling site lies between Kells and Raffin Fort, at 7 km to the north-east of the former and 4.4 km to the south-west of the latter (6◦ 46.7 W, 53◦ 45 N; national grid ref. N807 786; Fig. 1). Today, the peripheral parts of the bog complex are largely cut over and carry conifer plantations. The

Results In all, 383 charcoal fragments were identified from the five samples. Sample numbering reflects the context numbers used in the archaeological excavation (Fig. 4). The results are presented as follows: In Fig. 6 percentage frequencies of the various charcoal types based on fragment count and dry weight respectively, are shown. Bar charts illustrate the percentage frequencies and percentage dry weights of the various charcoals in the five samples as a whole (Fig. 7). The most frequent charcoal type overall is hazel at 35.8, or 47.5% by weight (Fig. 7). This is followed in descending

354 Fig. 6 Pie charts showing the percentage fragment frequency and percentage dry weight of wood types from BE 2–5. BE 1, from which only a few fragments of charcoal were identified, is not shown

frequency by oak (18.5%), ash (17%), Sorbus/Crataegus (8.9%; includes rowan/whitebeam, hawthorn and crab apple (Malus sylvestris), i.e. the charcoal-type Pomoideae), alder (7.6%), willow/aspen (6.5%), elm (3.9%) and Prunus spp. (2.1%; includes wild cherry, bird cherry and black-

thorn). When considered on a weight basis, the overall picture remains broadly similar apart from hazel and ash, the former being much more important when weight is considered (47.5% based on weight) while the latter is less important (9.2% based on weight).

355

Fig. 7 Pie charts show mean percentage representation of composite pollen groups (those included in the pollen sum) in zones B, C and D. Zone C, which represents the Late Iron Age Lull, has increased representation of woody taxa and low NAP values. The two central bar charts show the mean percentage representation of the main AP taxa in zone C. Maximum and minimum percentage values (triangles) are also indicated. Note: in the lower diagram the vertical scale is exaggerated by 5. The bar charts on the right hand side show charcoal fragment representation. The bars in the upper and lower diagrams

show fragment counts as a percentage of total fragments and weight as a percentage of total weight. Maximum percentage counts and the maximum percentage weights are also indicated (triangles). All of the taxa were not recorded in at least one sample so zero is the minimum value in the case of each taxon. Abbreviations of pollen taxon names: Be, Betula; Co, Corylus; Fr, Fraxinus; Qu, Quercus; Sa, Salix; Ta, Taxus and Ul, Ulmus. Abbreviations of charcoal taxon names are as above with following exceptions: Sa, Salix/Populus and So, Sorbus/Crataegus/Malus

As regards the individual layers, there is considerable inter-layer variation (Fig. 6). The lowermost layer (C5291/5290; BE 1) yielded only Salix/Populus (3 fragments; not shown in Fig. 6). In 1995, prior to the current charcoal investigations by M.D., charcoal from C5290 had been removed and submitted for radiocarbon dating. Unfortunately, this wood was not identified. However, the charcoal fragments retrieved from C5291 probably included remnants from C5290 and are regarded as indicative of the composition of the charcoal in C5290. In the next layer (C5283; BE 2), Salix/Populus dominates (frequency 43.5%; weight 55%), Corylus is important and Ulmus, Quercus and Alnus are also represented. Layer C5261 (BE 3) is distinctive in that Corylus dominates (especially if weight is considered), Alnus is im-

portant and Sorbus/Crataegus and Salix/Populus are minor components. The uppermost layers (C5214 and C5215, BE 4 and 5), are distinctive in that, in addition to Corylus, tall canopy trees are well represented (mainly Quercus and Fraxinus; Ulmus is a minor component) and also Sorbus/Crataegus. Differences between these layers include the presence, albeit in very small quantities, of Salix/Populus and Alnus in C5214 while Prunus is better represented in C5215. Ages and forms of wood present. While ring curvature does not give a strict qualitative indicator of age, the results nevertheless are useful in that they give insights into the age of the wood that was burned. Mature wood (c. >15 years) is presumed to be derived mainly from large branches and trunks, while young wood ( ≤ 15 years) is presumed to derive from young branches and twigs.

356

Fig. 8 Percentage pollen diagram, EMLI, from Emlagh Bog. Main curves only are given. Colour coding is used to indicate the main ecological/land-use indicator groups. Individual pollen curves that

are not shaded have a magnified horizontal scale which is indicated at the base of each curve. A closed circle is used to highlight those parts of the curves with low values

In excess of half the fragments analysed derive from mature wood, i.e. the rings have low curvature and furthermore they are, by and large, narrow (<2 mm) which is typical of mature wood. The main exceptions are as follows: (i) all Alnus fragments were estimated to be ≤ 10 years and in C5261 the 17 fragments noted were ≤ 5 years, and (ii) Sorbus/Crataegus fragments were estimated to ≤ 10 years.

Quercus (oak). Quercus is well represented in the upper charcoal layers only (C5214 and C5215). The fragments are from mature wood. Oak is slow burning and gives out substantial heat as it burns which would have made it a natural choice for a fire. Both Q. petraea and Q. robur are usually regarded as native in Ireland, though it appears that only Q. petraea is represented in the macrofossil record (Jessen 1949, p 247); it is also the main tree in the Atlantic oak woodlands on poor, acidic soils in western Ireland (e.g. the Atlantic oak woodlands of Killarney National Park; Kelly 1981). On the other hand, Q. robur is characteristic of the more fertile parts of Ireland such as the study area. Unfortunately, it is difficult to distinguish these species on the basis of wood anatomy (Grosser 1977) so a distinction was not attempted here. On the basis of present-day distribution pattern and ecological preferences, it is probable that the charcoal derives mainly, if not exclusively, from Q. robur, the pedunculate oak. Fraxinus (ash; F. excelsior). Fraxinus is recorded in the two uppermost layers only, where it is well represented. Ash makes great fuel, burned green or dead, and this may have influenced its selection (see also pollen vs. charcoal representation, Fig. 7). As in the case of oak, ash charcoal from mature wood only was recorded.

Comments on individual charcoal taxa Corylus (hazel; C. avellana). The charcoal data as a whole show that hazel was the main tree/shrub used in the burning episodes (Figs. 6 and 7). It accounts for 35.8% of all charcoal fragments identified. Hazel was widely exploited in both prehistory and historical times for its nutritious nuts and supple rods which were widely used for building. Its coppice-like growth form makes it relatively easy to cut and there are normally substantial quantities of dead wood available near ground level. Pollen analytical studies indicate that hazel was of greater importance in Ireland for most of the Holocene than elsewhere in western Europe. Keeping these considerations in mind, and also its importance in the study area (see pollen diagram, Fig. 8), its abundance in the charcoal assemblages is not surprising.

357

Sorbus/Crataegus (rowan/whitebeam/hawthorn/crab apple). This charcoal type is well represented especially in the uppermost layers. Woodlands and woodland-related environments are the normal habitats for the various woody plants that may be represented in this charcoal type. An important habitat, especially for hawthorn (Crataegus), is the so-called Mantel or edge communities of woodlands (cf. Wilmanns and Brun-Hool 1982). The spines of hawthorn (also blackthorn, Prunus spinosa) afford it a degree of protection from grazing and enable it to establish and survive in the face of light grazing. It is noteworthy that only young wood is represented in the charcoal, suggesting that the charcoal derives from twigs and small branches, and possibly from immature specimens growing outside the woodland proper. Alnus (alder; A. glutinosa). Alnus is moderately represented in the charcoal assemblages and is not recorded in the top and bottom layers. Interestingly, only young wood is represented (particularly in C5261 where the fragments were ≤ 5 years). Alder was probably largely confined to damp/wet areas. Alder wood does not burn well. Salix/Populus (willow/aspen). Willow/aspen charcoal was recorded in all but the uppermost layer. It was the main charcoal in the basal layers (in the lowermost layer only three fragments – Salix/Populus – were identified; because of the very limited number of specimens, the results are not plotted in Fig. 6). Populus is seldom recorded in Irish pollen diagrams and then mainly in the early Holocene. It is assumed that willow (one or more of several possible willow species) is mainly, or probably exclusively, represented in the charcoal records. Ulmus (elm). Elm was recorded in three samples where it has moderate to low representation. In the early Holocene woodlands of eastern central Ireland, elm made a major contribution. After the Elm Decline, its contribution to woodland cover in the Irish midlands oscillated considerably, due to phases of clearance in the context of farming activity (disease may also have been involved) followed by phases with strong regeneration, until final woodland clearances in relatively recent times (Moore 1967; O’Connell 1980; Holocene pollen diagram from Emlagh Bog (K. Molloy, unpublished); partial diagram presented in Fig. 8). Only U. glabra is regarded as native to Ireland (Webb et al. 1996). Prunus spp. [includes wild cherry (P. avium), bird cherry (P. padus) and blackthorn (P. spinosa)]. Prunus charcoal has low representation in the uppermost two samples. Blackthorn may have been common in Mantel vegetation while wild cherry would be expected to occur in the woodlands. Bird cherry may also have been represented. Today, it is largely confined to the northern part of Ireland (Preston et al. 2002). Webb et al. (1996) regard it as introduced but this view is not universally accepted. Regional vegetation and land-use history based on pollen analytical data A partial pollen diagram showing the main percentage curves for the relevant period is presented in Fig. 8. For

ease of description the diagram is divided into local pollen assemblage zones (PAZs) A to E. Given that the core was taken in the middle of a large bog, the pollen curves reflect changes in vegetation and land use at a regional level, i.e. the pollen spectra integrate pollen arising from a wide area around the bog. The pollen curves therefore reflect general trends in vegetation change and land use rather than localised developments. Tephra layers centred on 73–74 and 96–97 cm were identified and geochemically assigned to Hekla 1 (a.d. 1104) and the a.d. 860 eruption (Sluggan B, Pilcher et al. 1995), respectively (C. van den Bogaard, pers. comm.). Three further layers were geochemically analysed but it was not possible to assign them with confidence to a known eruption. Yet another layer, detected in the initial sampling (135– 140 cm; dates to c. a.d. 430) and relating to the part of the profile presented here, was not re-found in the 1 cm thick subsamples; however, it is assumed that the layer exists but further re-sampling is required to pinpoint it and obtain shards for microprobe analysis. The age/depth relationship for the part of the profile presented here is constrained by the tephra layer at 96 cm (all samples 1 cm thick; uppermost depth quoted) that is assigned to the Hekla a.d. 860 eruption and the calibrated age of AMS 14 C dates from 180, 198 and 224 cm. To facilitate plotting an age/depth curve, the mid-points of the 1σ probability range for the above dates were used (Table 1). Zone A (end of the 3rd century B.C.) Only a single spectrum of this zone is presented here. This spectrum is characterised by high Corylus and moderate Poaceae and P. lanceolata representation (but the trend is towards increasing values). The high Corylus values mark the end point of a period of more than 300 years (early Iron Age) during which there was reduced farming which facilitated the regeneration of woody vegetation and especially hazel (complete profile will be presented elsewhere). During this period there is an interruption in the cerealtype curve that had been more or less continuous since the mid Bronze Age (c. 1700 b.c.). This serves to emphasise the overall decline in farming and especially arable activity during the early Iron Age, the end of which is recorded in the single spectrum presented in zone A. Zone B (300 B.C.–A.D. 20) Key features of this zone are the decline in Corylus and a strong increase in P. lanceolata and especially Poaceae. AP (cf. curves marked ‘Trees’ in Fig. 8) decline from levels in excess of 80% in the immediately preceding spectra (uppermost spectrum presented in zone A) to 44% at the end of zone B. On the other hand, NAP taxa, indicative of grassland, increase from low values (less than 16% in the three spectra that immediately precede zone B) to 54% at the top of zone B.

358

The features referred to above, and also the unbroken curve for cereal-type pollen, indicate sustained and substantial farming activity that resulted in expansion of open farmland at the expense of woodland. Tall canopy trees, especially elm and ash, were cleared (these are well represented in the immediately preceding part of the profile which is not reproduced here) but most striking is the clearance of hazel scrub which serves to emphasise the intensity of farming during this period. The composition of the NAP component (cf. ‘Herbs-pasture’ curves in Fig. 8) and the exceptionally high values for Poaceae (maximum 46%) and Plantago lanceolata (peaks at 14%) indicate that the farming economy was mainly pastoral based. Cereal growing, however, was probably also relatively important at least within the context of prehistoric farming in Ireland which was mainly pastoral based (cf. O’Connell and Molloy 2001; Kelly 1998). Zone C (A.D. 20–500) In this zone, the trends seen in the previous zone are reversed, i.e. the AP curves expand strongly (Ulmus, Fraxinus, Taxus (minor but continuous curve) and especially Corylus) and NAP curves decline and reach particularly low values in the middle of the zone (e.g. P. lanceolata falls to ≤ 0.6% in two spectra). Three subzones are recognised which serve to designate an initial phase (C-i) when AP expands strongly, which is followed by a phase when NAP has particularly low representation and Ulmus and Quercus are most strongly represented (C-ii), and finally a phase of short duration during which NAP starts to recover and AP declines (especially Ulmus; C-iii). The changes described above are interpreted in terms of regeneration of woody vegetation in response to a substantial reduction in farming (subzones C-i and C-ii) that is followed by initiation of woodland clearance as farming activity increases substantially (subzone C-iii). Interestingly, cereal-type pollen is represented throughout and in subzone C-i achieves the highest values up to this point in the profile, i.e. since the inception of farming in the Neolithic. In other words, farming which included cereal growing, continued in the region although at low levels, particularly in the late third and most of the fourth century a.d. (subzone C-ii). Noteworthy also, particularly in the context of the charcoal records, are the wide diversity of woody plants in the pollen records. The unbroken curve for Taxus suggests that yew was present but was not frequent (Taxus is generally regarded as over-represented in pollen records). Other pollen taxa of note include Sorbus (both rowan and broadleaved Sorbus species may be represented here; see below), Sambucus (almost certainly S. nigra, elderberry), the only species of the genus regarded as native in Ireland and Britain; Preston et al. 2002), Prunus (may include pollen of one or more Prunus species; see below), Crataegus (hawthorn) and Ilex (holly). All of these pollen taxa are severely under-represented in pollen records due to low pollen production and dispersal (mostly insect pollinated). The values, although low, can therefore be regarded

as highly significant in terms of indicating presence of the plants in question at a regional level. Zone C is regarded as the Late Iron Age Lull (LIAL), a feature first described by Mitchell (1965) from the classic Littleton Bog profile and subsequently recorded in several pollen diagrams from western Ireland (e.g. Jeliˇci´c and O’Connell 1992; Molloy 2005; Molloy and O’Connell 1993, 2004; O’Connell and Molloy 2005; O’Connell et al. 2001) and also in the adjoining Co. Louth (Weir 1995). The impression given by the profiles in the publications cited is that there was a substantial reduction rather than complete abandonment of farming activity during the LIAL. The continuous cereal-type curve in the present profile indicates that this is certainly the case within the pollen source area of EML I. Likewise, the 14 C dates from the burning episodes at Raffin Fort (Table 1; Fig. 5) support the hypothesis of human activity in the study region at/or before (cf. 14 C date for the human skull), and also during, the Late Iron Age Lull. If the dating is accurate (and there is no reason to believe otherwise), then the burning episodes can be placed at the beginning of subzone C-iii when the first signs of increased human activity within the LIAL are detected in the pollen record. Zone D (A.D. 500–730) This zone is characterised by a substantial decline in Corylus and Fraxinus, and expansion of NAP curves. The latter include all the classic indicators of farming activity such as Poaceae, Plantago lanceolata and Rumex, which are primarily indicators of pastoral farming. Cereal-type pollen is also well represented and Secale-type (presumed to include only pollen of rye) is recorded for the first time. Evidence such as that available here for renewal of farming activity, increased importance of cereal cultivation including the introduction of rye and impact on woodlands can be seen in many Irish pollen profiles at about this time (e.g. Fuller 2002; Molloy 2005; O’Connell et al. 2001; O’Connell and Molloy 2005; Weir 1995), although the pattern elucidated in these publications may not pertain in all parts of Ireland (Hall 2005). The upsurge in farming is usually regarded as associated with the introduction of Christianity, which in turn marks the beginning of the early Medieval period in Ireland. Zone E (mid eight century A.D.) The single spectrum that is presented here marks the beginning of a phase of further intensification of farming activity dating to the mid Medieval period. Apart from increases in Poaceae and P. lanceolata, cereal-type pollen is more strongly represented and Secale is consistently recorded. Increased importance of arable farming is also emphasised by curves for Artemisia and Apiaceae, i.e. pollen of weeds associated with arable farming (Brown et al. 2005; O’Connell and Molloy 2005; O’Connell and N´ı Ghr´ainne 1994). This increase in farming activity is not

359

surprising given the many important monastic settlements in the wider region (e.g. the monastery at Kells founded in a.d. 804; Herbert 1996). Pollen and charcoal analyses – complementarity in palaeoenvironmental reconstruction Pollen and charcoal analyses provide complementary information on the environmental and other contexts of past human activity. Pollen data from regional pollen profiles enable forest composition and overall levels of human activity to be reconstructed based on careful interpretation of the pollen curves of woody taxa and NAP taxa respectively. Furthermore, important insights can be gained into farming economies from the curves of taxa indicative of pastoral (especially Poaceae and P. lanceolata) and arable (cf. cereals and weeds associated with arable activity) farming (Behre 1981). An important strength of pollen data is their secure statistical basis, particularly if sufficient pollen are counted and identification is reliable. Charcoal data, on the other hand, provide much specific information on wood that has been subjected to firing in circumstances where combustion is incomplete. Since it can be assumed that such wood derives from, by and large, local contexts, charcoal records are normally regarded as indicative of local presence of the trees/shrubs in question. As in the case of pollen data, the interpretation of charcoal data in terms of abundance of particular species is not straightforward and particularly so since selection of wood for burning in the past was often strongly influenced by cultural, religious and other ethnographic considerations rather than simply local abundance or ready availability of particular species. For these and other reasons, it is therefore interesting and useful to compare pollen and charcoal data with a view to gaining a more comprehensive insight into local and regional environments, and the nature of human interaction with the local environment. To facilitate such a comparison, a diagram with the relevant pollen and charcoal data in summary form is presented in Fig. 7. The pie-charts show mean percentage representation of composite pollen groups in pollen zones B, C and D. Additional information relating to the intervals represented by the PAZs is presented in Table 2. The representation of the main trees in zone C, which corresponds with the period to which the charcoal in the pit relates, is shown in bar charts (see the two central charts) and charcoal representation in the five layers in the pit (taken as a whole) is shown on the right hand side.

Table 2 Summary information on pollen data that are presented as pie charts in Fig. 7. Ages based on interpolation between points dated by an AMS 14 C date and micro-tephra layers

The distinctiveness of the LIAL (zone C), obvious already in the pollen diagram, is particularly clear in the pie-chart presentation of the pollen data. Corylus representation has increased by 44% with respect to zone B, clearly demonstrating the size of the increase in hazel pollen production. This may not fully reflect the expansion that took place in the hazel population in that pollen production per individual hazel specimen was probably depressed as tall canopy trees such as ash and elm (also oak and yew but to a lesser extent) expanded. The expansion of woody vegetation was a response to a reduction in farming activity (cf. substantial decline in P. lanceolata). However, some farming activity, including arable farming, persisted as is shown by the unbroken cereal-type curve (see also above). Comparison of the representation of woody taxa in the pollen and charcoal data shows several interesting features (bar charts, Fig. 7). Corylus is well represented in both which suggests that hazel was both readily available and used for burning. Quercus and Fraxinus are more strongly represented in the charcoal record. Fraxinus pollen representation is higher in zone C than in earlier zones suggesting the increased availability of ash as a source of wood. It should also be noted that Fraxinus is probably underrepresented in the pollen record compared to Quercus (cf. Andersen 1973); in other words, oak and ash may have been equally important in the regional vegetation during much of zone C. Ulmus is a rather minor element (also variable) in the charcoal record. Its pollen curve suggests slow regeneration during zone C and differential clearance with respect to other trees near the end of the zone (subzone C-iii). The charcoal evidence suggests that elm was not preferred for burning but local scarcity and/or purposeful avoidance may also have had a role. Conifers are poorly represented in the pollen record and are absent from the charcoal record. Mainly single pollen grains of Pinus were recorded (maximum 4 per pollen spectrum) which are probably attributable to long-distance transport for which Pinus pollen is well known. The failure to record pine as charcoal also suggests scarcity or absence. It is quite probable that pine was locally and even regionally extinct by this time (but see Kelly 1999; Feehan 2000; O’Connell and Molloy 2005). Taxus, on the other hand, represented in zone C by a slender, although well-defined, pollen curve is not recorded in the charcoal record. Yew is regarded as a relatively high pollen producer (Bradshaw 1981) so it was probably a relatively minor component in the regional woodlands. Its absence from the charcoal record may be due to local rarity or even absence or it may have been avoided for cultural

PAZ (EML I)

Chronology

D C B

a.d. 500–730 a.d. 20–500 300 b.c.–a.d. 20

a

No. of pollen spectra No. of terrestrial pollen counteda 6 12 5

5434 12444 4346

Taxa included in the pollen sum; pollen of purely local origin, i.e. bog taxa such as Calluna and Cyperaceae, are excluded

360

and other reasons, e.g. on account of its highly poisonous berries. Trees and shrubs of wet habitat such as alder and willow (poplar may be represented in the Salix/Populus charcoal type, although this is unlikely; see above) are relatively well represented in the charcoal record. The correspondence with the pollen record is good if allowance is made for the strong over-representation of Alnus (high pollen production and dispersal) and the under-representation of Salix. Given the preference of alder and willow for wet soils, local availability (e.g. on the drumlin or its sides) is not expected. It may, however, have been available in wet habitats in the general vicinity of the drumlin. Selective gathering is also a possibility. The absence of Betula in the charcoal record is noteworthy. On ecological grounds and present-day distribution patterns, absence within the general vicinity of Raffin Fort is unlikely. In other words, deliberate avoidance for use in fires is quite possible (see below). However, it may also be argued that birch may have been relatively scarce in the local environment at Raffin. Its average pollen representation in zone C is 6.7% but it should be borne in mind that it is over-represented in pollen diagrams due to high pollen production and excellent dispersal. Furthermore, over-representation in the pollen records from bogs is also expected on account of its preference for wet and nutrient-poor habitats such as bog margins (cf. the importance of birch timbers in the Iron Age trackway at Corlea, Co. Longford; Raftery 1994, p 99). When comparing the pollen and charcoal records for Rosaceae several considerations should be borne in mind. In the pollen record, it can be assumed that woody Rosaceae are greatly under-represented. Another factor to be considered is the difficulty of distinguishing Rosaceae pollen to genus level, e.g. Sorbus which includes trees with quite different ecological preferences such as rowan (S. aucuparia) and whitebeams (e.g. S. hibernica, S. aria and S. rupicola). In charcoal analysis it is also often difficult/impossible to identify charcoal of Rosaceae to species or even genus level; for example, as already pointed out, Prunus and Sorbus/Crataegus potentially aggregate several genera and/or species. Nevertheless, comparison of the records yields additional interesting information. The pollen records indicate that woody Rosaceae (e.g. Crataegus) and also other minor trees/tall shrubs such as elder (Sambucus) and holly (Ilex) were present although probably not plentiful in the region. Elder and holly are not recorded in the charcoal record. On the other hand, Sorbus/Crataegus and Prunus are well represented in layers C5214 and C5215 but are only weakly represented in the pollen record. This is probably attributable to poor pollen production and dispersal rather than scarcity in the landscape. This serves to highlight the complementarity of charcoal and pollen analyses in reconstructing past woodland composition and use.

Trees, wood and woodlands: evaluation in the context of ritual and religion While the radiocarbon dates indicate that the skull fragment was placed in the pit at Raffin Fort sometime between the 3rd and early 5th century a.d., the skull itself is dated to about the Birth of Christ (Fig. 5). If the later end of the age range (early 1st/mid 2nd century a.d.; Fig. 5) is considered, then there is broad contemporaneity with a small, bronze La T`ene fibula from the site. This indicates not just a protracted mid to later Iron Age horizon at Raffin but also points to its broader cultural milieu, which, in ethnographical terms, can be characterised as insular La T`ene or Celtic. Though evidently insular in their expression, religious practices in Iron Age Ireland would have compared favourably with those known from Britain and western Europe generally (Cunliffe 1997). Insights into the social and religious significance of wood and trees among Iron Age peoples on the Continent, obtained through Classical writers and from archaeological sources, may therefore be of relevance in evaluating the evidence derived from the charcoal preserved in the ritual pit at Raffin Fort. It is widely believed that the Celts worshipped at natural places; in groves, woodland and forests and at springs, bogs and lakes (e.g. Cunliffe 1997; Ross 1967, 1986; Green 1995, 1997). Theirs was a religion of nature, where gods and goddesses inhabited and embodied natural features. Trees, woodlands and wood, in particular, played an important role in the belief systems of the Celts. Classical sources attest to a multiplicity of sacred groves in continental Europe. Lucan, writing in the 1st century a.d., describes a sacred grove in the town of Massilia (Marseille), which was discovered by Caesar’s legions as they invaded Gaul: A grove there was, untouched by men’s hands from ancient times, whose interlacing boughs enclosed a space of darkness and cold shade, and banished the sunlight from far above . . . gods were worshipped there with savage rites, the altars were heaped with hideous offerings . . . yew trees fell down and rose again, the glare of conflagration came from the trees that were not on fire, people never resorted to worship there at close quarters but left the place to the gods. The Civil War III, 399–425 (translated by Duff 1928). Similar accounts are given by Strabo (12.5.1–2; translated by Jones 1960) who writes of the reunification of three Galatian tribes (Celtic) at a drunemeton (oak-sanctuary or sacred oak-grove) in Galatia in Asia Minor, and by Tacitus who speaks of forest clearings in Anglesey, Wales as the last stronghold of the druids against Rome (The Annals XIV, 30, 1; translated by Jackson 1937). Dio Cassius (Dio’s Roman History LXII, 7, 1–3; translated by Cary 1968) refers to sacred woods in Britain where human sacrifices were carried out. The Celtic priests or druids referred to in classical literature are described as worshipping the oak tree and rarely carrying out a ceremony without utilising some part of it (Aldhouse-Green 2000; Linnard 1982).

361

There is good correlation between the archaeological evidence and written sources in this regard, including depictions of trees on coins, altars and in metalwork from the later Iron Age (cf. Aldhouse-Green 2000; Webster 1995). There is also evidence for the characterisation of trees, leading to the selection of certain wood types for particular social or religious activities. Boughs of oak, for example, accompanied a burial in an oak coffin at Gristhorpe, near Scarborough (Ross 1967, p 33). Likewise, the frequent finding of hazel in the context of late Iron Age bog bodies has also been noted (Aldhouse-Green 2000). The medieval Dutch tradition associating the loss of a cow’s milk with willow has been suggested as a possible explanation for the use of willow wattle in domestic contexts, but not where animals were stalled, within an Iron Age long-house excavated in the Assendelver Polders (Therkorn et al. 1984). The 190 wooden images and idols recovered from the shrine of Sequanna, which marked the source of the river Seine, were all of oak. These dated to the 1st century a.d. The primacy of oak as a venerated tree also finds expression in Ireland where, for example at Navan Fort, Co. Armagh, a huge oak post, felled in 95/94 b.c., was erected at the centre of the socalled ‘Forty Metre Structure’. It has been interpreted as an axis mundi and a symbol of the druidic caste (Lynn 1992). Druidism itself probably represents a forest cult and the ˆ (oak) (Linnard etymology of the word druid is from δρυς 1982, p 10). Early Irish literature preserves an impressive corpus of traditions associated with sacred trees (Ir. B´ıle) and groves, as well as a detailed ‘wood lore’ recording the symbolism and powers of different tree species (Lucas 1963; MacCoitir 2003). Indeed, aspects of such customs survive into the present (and not just in Ireland, see Rival 1998) in the guise of saint’s trees and so-called ‘Rag Trees/Bushes’, which are frequently associated with holy wells. Phenomena such as these, and the fact that early church sites were sometimes located on the sites of, or near, trees or groves sacred to pagans (e.g. Saint Patrick founded his first church at Druim Sailech, the Hill of the Sallows or Willows; Lucas 1963), attest to the resolution of pagan and Christian traditions in Ireland and elsewhere. Inauguration sites of the late prehistoric and early Medieval periods often had b´ıle associated with them, and these commonly consisted of an ash, yew or oak tree (Lucas 1963; Manning 1988; Watson 1981). In such contexts, the growing tree symbolised the fecundity and life-line of the tribe; the felling of such a tree, such as the Ash of Uisneach, signalled decline or loss of power. The characteristics of different trees, such as strength, flexibility, longevity, etc., figure totemically in Irish tribal and personal names. IVA-, meaning ‘yew tree’, appears in compound names like IVACATTOS (‘yew’ and ‘battle’) and is the root of the ´ tribal name ‘Eoganachta’ (McManus 1997, §6.6). The tribal or dynastic name ‘U´ı Fidgeinti’ means the ‘People of the Wood’ (E. Bhreathnach, pers. comm.). The U´ı Fidgeinti are indirectly associated with the magical well at Segias which is traditionally identified as the source of the river Boyne [though Bhreathnach (1999, pp 85–86) argues that it could equally refer to the source of the Shannon]. Segias, it is

recorded, obtained its supernatural powers from hazel nuts that dropped into it from a nearby hazel bush. According to legend, the ‘Salmon of Knowledge’ acquired its special wisdom by swallowing one of these hazel nuts and later transferred it to the legendary Fionn mac Cumhaill (whose name means ‘Seer Son of Hazel’) when he ate of its flesh. Wood lore permeated many aspects of Irish life during the early centuries a.d. Nowhere is this better illustrated than in the case of the Ogham alphabet (referred to in Irish as In Beithe-luis-nin, which means ‘the birch-rowanash’), a stroke script created between the 4th and the 5th centuries a.d. Ogham provides evidence for the complex significance of wood and trees in pre-Christian Ireland and the perpetuation of tree lore into the early Christian period (McManus 1997). The letters (Ir. Feda which means trees) of the Ogham alphabet are arranged into four groups of five characters, and many of them are named after trees [McManus 1997, §3.14–5; Nelson and Walsh (1993) suggest that such interpretation should be treated with due caution]. The first group, for instance, the Aicme Beithe, or birch group, comprises the letters B, L, F, S and N which correspond to the initial of the letter-name, vis Beithe/birch; Luis/rowan; Fern/alder; Sail/willow and Nin/ash (McManus 1997, §1.3; MacCoitir 2003). If only for the purposes of kenning, this arrangement suggests, in the case of individual letters, reference to specific characteristics of each species, and in the case of the groups, agreement on perceived common attributes across each quintuplet of trees. In addition, Ogham inscriptions are read in the same manner as one would climb a tree, namely ascending from the lower right side (McManus 1997, §1.5). There is a significant concentration of Ogham stones in the Blackwater valley and particularly in the area around Raffin Fort (e.g. Moore 1990; Newman 2005). The discovery of a small Ogham stone in the enclosing fosse at Raffin Fort, a context broadly contemporary with the ritual pit, is particularly significant in that it locates at Raffin a person with a knowledge of Ogham and therefore well versed in wood lore. In Table 3, this idea is progressed a step further by presenting the concordance between the wood types employed in each burning episode and their Ogham equivalents (including letter group or aicme; the aicmi – plural of aicme – are ordered from 1–4). The intriguing possibility arises that the charcoal/wood types represent a meaningful letter sequence in the Ogham system. Though a definitive statement is not possible, the information presented in Table 3 raises the possibility that a particular rubric influenced/determined wood selection. Moreover, ignoring, as it does, frequency or weight, this table reminds us that in a ritualistic context such as this, wood types of low incidence are potentially as important as, or maybe even more important than, the better represented types. For example, placing twigs of alder or Pomoideae or Prunus into the fire may, under the circumstances, have been a highly symbolic act. Wood from the first two letter groups dominates, i.e. ´ the Aicme Beithe (Group 1) and Aicme hUath (Group 2). However, given the problems of distinguishing at species and/or genus level charcoal of trees such as Salix and

362 Table 3 Concordance of Latin, English and Old Irish nomenclature, Ogham letter equivalents and Ogham letter group (letter, name of group or aicme and aicme number; aicmi are numbered 1–4) Latin name Burning Episode 5 Corylus avellana Quercus Fraxinus excelsior Ulmus glabra Pomoideae Crataegus Sorbus aucuparia Sorbus (broad-lvd. spp.) Malus sylvestris Prunus P. avium P. spinosa Burning Episode 4 Corylus avellana Quercus Fraxinus excelsior Ulmus glabra Alnus glutinosa Salix Populus tremula Pomoideae Crataegus Sorbus aucuparia Sorbus (broad-lvd. spp.) Malus sylvestris Prunus P. avium P. spinosa Burning Episode 3 Corylus avellana Alnus glutinosa Salix Populus tremula Pomoideae Crataegus Sorbus aucuparia Sorbus (broad-lvd. spp.) Malus sylvestris Burning Episode 2 Corylus avellana Quercus Ulmus glabra Alnus glutinosa Salix Populus tremula Burning Episode 1 Salix Populus tremula

Evaluation

English name

Old Irish name Letter

Aicme

Aicme no.

c c c c

Hazel Oak Ash Wych elm

Coll Dair Nin Ailm

C D N A

Beithe ´ hUath Beithe Ailm

1 2 1 4

p p p p

Hawthorn Rowan Whitebeam Crab apple

´ hUath Luis Fionncholl Cert

U L C? Q

´ hUath Beithe ´ hUath ´ hUath

2 1 2 2

p p

Wild cherry Blackthorn

Idad Straif

I Z

Ailm Muin

4 3

c c c c c p p

Hazel Oak Ash Wych elm Alder Willow Aspen

Coll Dair Nin Ailm Fern Sail Edad

C D N A F S E

Beithe ´ hUath Beithe Ailm ´ hUath Beithe Ailm

1 2 1 4 2 1 4

p p p p

Hawthorn Rowan Whitebeam Crab apple

´ hUath Luis Fionncholl Cert

U L C? Q

´ hUath Beithe ´ hUath ´ hUath

2 1 2 2

p p

Wild cherry Blackthorn

Idad Straif

I Z

Ailm Muin

4 3

c c p p

Hazel Alder Willow Aspen

Coll Fern Sail Edad

C F S E

Beithe ´ hUath Beithe Ailm

1 2 1 4

p p p p

Hawthorn Rowan Whitebeam Crab apple

´ hUath Luis Fionncholl Cert

U L C? Q

´ hUath Beithe ´ hUath ´ hUath

2 1 2 2

c c c c p p

Hazel Oak Wych elm Alder Willow Aspen

Coll Dair Ailm Fern Sail Edad

C D A F S E

Beithe ´ hUath Ailm ´ hUath Beithe Ailm

1 2 4 2 1 4

p p

Willow Aspen

Sail Edad

S E

Beithe Ailm

1 4

Evaluation of the presence/absence of the particular taxon based on the charcoal evidence is indicated as follows: c, certain; p, possible

363

Populus, Crataegus and Sorbus and species of the genus Prunus, various letter options arise. Populus pollen was not recorded in the EML I profile while Salix is well represented. It is reasonable therefore to assume that the charcoal samples probably derive from willow. It follows that Group 4 (Aicme Ailm, which includes Populus and Prunus avium) may be represented only by Ulmus (BE 2, 4 and 5). It is also possible, of course, that the wood types were chosen not because of the Ogham letters with which they were associated, but rather because of other attributes, such as the briatharogaim, i.e. poetic descriptions of the letter names (MacCoitir 2003, p 160). Luis (meaning rowan) is variously described as ‘luster of eye’, ‘friend of cattle’ and ‘sustenance of cattle’. Fern (alder), a species traditionally associated with death in Ireland because of the markedly red sap it produces when freshly cut, is characterised as ‘vanguard of warriors’, ‘milk container’ and ‘protection of the heart’. The combination of the elements, wood, stone and bone, in the burial pit at Raffin Fort suggests a complex iconography. Bradley (2002) has argued that as symbols of passing and lasting, wood and stone may have been connected with the transience of the living and the permanent or everlasting realm of the ancestral dead. Bloch (1995, pp 213–215) observed the same phenomenon at work in the traditional Malagasy society where wood is used to build the houses of the living and stones the houses of the dead. The combustion of wood may, in such contexts, represent the transition from life to death, from mortality to immortality. Applied to Raffin Fort, the combination of the burnt wood and stone may engender a quite complex iconography of transition from life to death, transience to permanence, which is, in turn, imparted to the human skull fragment. Conclusions It is clear from the foregoing discussion that trees, and the wood derived from them, were socially and culturally charged materials in later pre-historic and early Medieval Ireland: they had totemic significance and the properties of different types of trees and wood, including their aesthetic attributes informed how they were characterised and employed by society. The social significance of wood suggests, therefore, that its selection for different purposes and events may have been determined by quite complex and deeply rooted social customs: combustibility or easy availability, for instance, may not have been the only or the main criteria applied in the case of wood collected for burning during religious ceremonies. The evaluation of the results of charcoal analysis from the ritual pit at Raffin Fort seeks to accommodate such possibilities. The carbonised wood from the burial pit represents eight distinct charcoal types, potentially derived from a much larger number of woody species indicative of a relatively wide spectrum of habitats. Whether the wood was stockpiled ahead of time or collected specifically for the occasion, it is unlikely that the wood came from a single location. There is, however, sufficient variety in topogra-

phy and edaphic conditions in the vicinity of Raffin Fort that all of the species could potentially be found within easy reach of the site (for example, within a kilometre) and certainly within the wider region, a hypothesis supported by the pollen data (zone C in pollen diagram; Figs. 7 and 8). Assembling the various wood types recorded in the charcoal assemblages was not necessarily an onerous task. The comparison of the charcoal and pollen data provides interesting insights into selection processes that may have been involved in wood collection at this late Iron Age site. Birch and holly, for instance, which were probably fairly readily available – based on the pollen evidence and also the rather wide ecological preferences of the species in question – are not represented in the charcoal assemblage. In both instances, it is quite likely that the omission was deliberate. On the other hand, alder, which is not readily combustible, may have been specifically chosen for symbolic reasons. The persistence of the seed-bearing cones and the prolific production of catkins in early spring, for instance, would not have gone unnoticed and may have resulted in attribution of particular qualities to alder. The fact that the Pomoideae and alder charcoal consisted of young, twiggy wood raises the possibility that flower- or fruit-bearing branches were selected, possibly for aesthetic and iconographical reasons. In contrast, the charcoal representing willow and tall-canopy trees, i.e. oak, ash and elm, is exclusively mature wood from large branches or trunks. It is quite possible that the wood in question consists of recycled construction timbers. Despite the significant insights into past societies, particularly with respect to ethnographical considerations that can accrue from integrated investigations – here charcoal and pollen analyses – into wood and its availability and use, investigations in which ethnographical considerations have been given due prominence have seldom been applied to palaeobotanical data from archaeological sites in Ireland. Clearly, a highly interesting dimension in the study of ancient wood remains has thus been largely ignored and, furthermore, a fundamental component of socio-ritual practices in the past has been neglected. Finally, the paper illustrates the ‘added value’ that accrues from an integrated application of palaeoecological and archaeological techniques. This is particularly the case given that the study relates to an intriguing period in Irish prehistory, i.e. the late Iron Age, and a distinctive feature in the post-glacial pollen analytical record from Ireland, namely the Late Iron Age Lull. Acknowledgements Facilities and help with charcoal analysis were accorded to M. Dillon by W. D¨orfler and H. Kroll during a visit to the Institut f¨ur Ur- und Fr¨uhgeschichte, University of Kiel. M. Hennessy, C. McCausland, G. Mills and M. Moriarty assisted with coring. W. D¨orfler prepared tephra samples for microprobe analysis and C. van den Bogaard carried out the analysis and provided commentary on the results. Assistance and advice with various aspects of the excavation were provided by E. Bhreathnach, J. Bradley, L. Buckley, E. Herring, R. Legg and S. Meece. R. Legg made available the topographical map that is reproduced in Fig. 1 and A. Gallagher assisted with the preparation of Figs. 3 and 4. Two referees, V. Hall and an anonymous referee, provided helpful comments. The excavations were funded by

364 the Heritage Service, Department of the Environment, Heritage and Local Government. Financial support towards the pollen analytical investigations and radiocarbon dates was provided under the HEA Programme for Research in Third Level Institutions (PRTLI-2).

References Aldhouse-Green M (2000) Seeing the woods for the trees: the symbolism of trees and wood in ancient Gaul and Britain. University of Wales, Aberystwyth Andersen ST (1973) The differential pollen productivity of trees and its significance for the interpretation of a pollen diagram from a forested region. In: Birks HJB, West RG (eds) Quaternary plant ecology. Blackwell Scientific Publications, Oxford, pp 109– 115 Asouti E (2002) Charcoal analysis from C¸atalh¨oy¨uk and Pinarba
i: two Neolithic sites in the Konya Plain, south-central Anatolia, Turkey. PhD thesis (unpublished), Institute of Archaeology, University College London, London Behre K-E (1981) The interpretation of anthropogenic indicators in pollen diagrams. Pollen Spores 23:225–245 Beug H-J (2004) Leitfaden der Pollenbestimmung f¨ur Mitteleuropa und angrenzende Gebiete. Friedrich Pfeil, Munich Bloch M (1995) Questions not to ask of Malagasy carvings. In: Hodder I (ed) Theory and practice in archaeology. Routledge, London, pp 213–215 Bradshaw RHW (1981) Modern pollen-representation factors for woods in south-east England. J Ecol 69:45–70 Bradley J (1991) Excavations at Moynagh Lough, County Meath. J R Soc Antiq Ireland 121:5–26 Bradley R (2002) The land, the sky and the Scottish stone circle. In: Scarre C (ed) Monuments and landscape in Atlantic Europe. Routledge, London, pp 122–138 Bhreathnach E (1999) The construction of the stone fort at Cahercommaun: a historical perspective. Discovery Prog Rep 5:83–91 Brown AG, Hatton J, O’Brien CE, Selby KA, Langdon PG, Stuijts I, Caseldine CJ (2005) Vegetation, landscape and human activity in Midland Ireland: mire and lake records from the Lough KinaleDerragh Lough area, Central Ireland. Veget Hist Archaeobot 14:81–98 Burenhult G (1980) The archaeological excavation at Carrowmore, Co. Sligo, Ireland: excavation seasons 1977–1979. Theses and Papers in North-European Archaeology 9, Institute of Archaeology, University of Stockholm, Stockholm Cary E (1968) Dio’s Roman history, vol. VIII (English translation). Heinemann, London Cunliffe B (1983) Danebury, anatomy of an Iron Age hillfort. Batsford, London Cunliffe B (1997) The ancient Celts. Oxford University Press, Oxford Dillon M, Newman C, Molloy K, O’Connell M (2006) Environment and ritual in a late Iron Age context: an example from Raffin Fort, Co. Meath, Ireland. BAR, Int Ser (in press) Duff JD (1928) Lucan. The Civil War. Books 1-X (English translation). Harvard University Press, Cambridge (Mass.) Feehan J (2000) Fraoch M´or Mh´aigh Rechet: the Great Heath of Maryborough. D´uchas, Dublin Frazer JG (1922) The golden bough: a history of myth and religion. Macmillan, London Fuller JL (2002) Past vegetation and land-use dynamics in Mayo Abbey, central Mayo. Archaeol Ireland 16:20–23 Gale R, Culter D (2000) Plants in archaeology. Westbury, Otley Greguss P (1959) Holzanatomie der europ¨aischen Laubh¨olzer und Str¨aucher. Akad´emiai Kiad´o, Budapest Green MJ (ed) (1995) The Celtic world. Routledge, London Green MJ (1997) Exploring the world of the druids. Thames and Hudson, London Grosser D (1977) Die H¨olzer Mitteleuropas. Springer, Berlin Hall VA (2005) The vegetation history of monastic and secular sites in the midlands of Ireland over the last two millennia. Veget Hist Archaeobot 15:1–12

Hammond N, Mather J, Saul FP (2002) Ancestral faces: a Preclassic Maya skull-mask from Cuello, Belize. Antiquity 76:951– 952 Herbert M (1996) Iona, Kells and Derry: the history and hagiography of the monastic familia of Columba. Four Courts Press, Dublin Jackson J (1937) Tacitus. The Annals, Books XIII–XVI (English translation). Harvard University Press, Cambridge (Mass.) Jessen K (1949) Studies in late Quaternary deposits and flora-history of Ireland. Proc R Ir Acad, Ser B 52:85–290 Jeliˇci´c L, O’Connell M (1992) History of vegetation and land use from 3200 b.p. to the present in the north-west Burren, a karstic region of western Ireland. Veg Hist Archaeobot 1:119–140 Jones H (1960) The geography of Strabo, vol. V, books X–XII (English translation). Heinemann, London Kelly DL (1981) The native forest vegetation of Killarney, south-west Ireland: an ecological account. J Ecol 69:437–472 Kelly F (1976). The old Irish tree-list. Celtica 11:107–124 Kelly F (1998) Early Irish farming. Dublin Institute of Advanced Studies, Dublin Kelly F (1999) Trees in early Ireland. Ir For 56:39–57 Linnard W (1982) Welsh woods and forests: history and utilization. National Museum of Wales, Cardiff Lucas AT (1963) The sacred trees of Ireland. J Cork Hist Archaeol Soc 68:16–54 Lynn CJ (1977) Trial excavations at the Kings Stables, Tray Townland, County Armagh. Ulster Journal Archaeol 40:42–57 Lynn CJ (1992) The Iron Age mound in Navan Fort: a physical realization of Celtic religious beliefs? Emania 10:33–57 MacCoitir N (2003) Irish trees. Myths, legends and folklore. Collins Press, Cork McCausland C (2002) Tephra as time markers: tracing volcanic eruptions of Holocene age in a peat profile from Emlagh Bog, Co. Meath. B.Sc. thesis (unpublished), Department of Botany, NUI, Galway, Galway McManus D (1997) A guide to Ogam. An Sagart, Maynooth Manning C (1988) A note on sacred trees. Emania 5:34–35 Meehan RT, Warren WP (1999) The Boyne valley in the Ice Age. A field guide to some of the valley’s most important glacial geological features. Meath County Council and Geological Survey of Ireland, Navan and Dublin Mitchell GF (1965) Littleton Bog, Tipperary: an Irish agricultural record. J R Soc Antiq Ireland 95:121–132 Molloy K (2005) Holocene vegetation and land-use history at Mooghaun, south-east Clare, with particular reference to the Bronze Age. In: Grogan E (ed) The North Munster project, Vol. 1. The later prehistoric landscape of south-east Clare. Wordwell, Bray, pp 255–301 Molloy K, O’Connell M (1993) Early land use and vegetation history at Derryinver Hill, Renvyle Peninsula, Co. Galway, Ireland. In: Chambers FM (ed) Climate change and human impact on the landscape. Chapman and Hall, London, pp 185–199 Molloy K, O’Connell M (2004) Holocene vegetation and land-use dynamics in the karstic environment of Inis O´ırr, Aran Islands, western Ireland: pollen analytical evidence evaluated in the light of the archaeological record. Quat Int 113:41–64 Moore F (1990) Appendix 4: Ogam stones. In: Eogan G Ballynee souterrains, County Meath. J R Soc Antiq Ireland 120:60–62 Moore JJ (1967) Zur pflanzensoziologischen Bewertung irischer nacheiszeitlicher Pollendiagramme. In: T¨uxen R (ed) Pflanzensoziologie und Palynologie. Junk, The Hague, pp 96–105 Moore MJ (1987) Archaeological inventory of County Meath. Government Stationary Office, Dublin Nelson EC, Walsh WF (1993) Trees of Ireland. Native and naturalized. Lilliput Press, Dublin Newman C (1993a) ‘Sleeping in Elysium’. Archaeol Ireland 7(3):20– 23 Newman C (1993b) The show’s not over until the Fat Lady sings. Archaeol Ireland 7(4):8–9 Newman C (1995) Raffin Fort, Co. Meath: Neolithic and Bronze Age activity. In: Grogan E, Mount C (eds) Annus archaeologiae: proceedings of the OIA winter conference 1993. Organisation of Irish Archaeologists, Dublin, pp 55–65

365 Newman C (1998) Reflections on the making of a royal site in early Ireland. World Archaeol 30:127–141 Newman C (2005) Re-composing the archaeological landscape of Tara. In: Bhreathnach E (ed) The Kingship and Landscape of Tara. Four Courts Press, Dublin, pp 361–409 O’Brien E (1990) Iron Age burial practices in Leinster: continuity and change. Emania 7:37–42 O’Connell M (1980) The developmental history of Scragh Bog, Co. Westmeath and the vegetation history of its hinterland. New Phyto 85:301–319 O’Connell M, Molloy K (2001) Farming and woodland dynamics in Ireland during the Neolithic. Biol Environ (Proc R Ir Acad, Ser B) 101:99–128 O’Connell M, Molloy K (2005) Native woodland composition and dynamics: a long-term perspective based on a Holocene pollen profile from Inis O´ırr, Aran Islands, western Ireland. In: Doyle C, Little D (eds) Ireland’s native woodlands (on CD-ROM). Woodlands of Ireland, Dublin, pp 20–47 O’Connell M, N´ı Ghr´ainne E (1994) Palaeoecology [of Inishbofin]. In: Coxon P, O’Connell M (eds) Clare Island and Inishbofin. Field Guide No. 17. Irish Association for Quaternary Studies, Dublin, pp 60–103 O’Connell M, Molloy K, McMahon H (2001) Reconstructing prehistoric farming activity and human impact at a fine spatial resolution: palaeoecological investigations at Mooghaun, Co. Clare, western Ireland. In: Schauer P (ed) Beitr¨age zur Siedlungsarch¨aologie und zum Landschaftswandel. Regensburger Beitr¨age zur Pr¨ahistorischen Arch¨aologie 7, pp 161–186 Pilcher JR, Hall VA, McCormac FG (1995) Dates of Holocene Icelandic volcanic eruptions from tephra layers in Irish peats. Holocene 5:103–110

Preston CD, Pearman DA, Dines TD (eds) (2002) New Atlas of the British and Irish Flora. Oxford University Press, Oxford Raftery B (1994) Pagan Celtic Ireland. The enigma of the Irish Iron Age. Thames and Hudson, London Rival L (ed) (1998) The social life of trees. Anthropological perspectives on tree symbolism. Berg, Oxford Roche H (2002) Excavations at R´aith na R´ıg, Tara, Co. Meath, 1997. Discovery Prog Rep 6:19–165 Ross A (1967) Pagan Celtic Britain. Studies in iconography and tradition. Routledge, London Ross A (1986) The pagan Celts. Batsford, London Schweingruber FH (1990) Anatomy of European woods. Haupt, Bern Therkorn LL, Brandt RW, Pals JP, Taylor M (1984) An early Iron Age farmstead: site Q of the Assendelver Polders project. Proc Prehist Soc 50:351–373 ´ Watson A (1981) The king, the poet and the sacred tree. Etudes Celtiques 18:165–180 Webb DA, Parnell J, Doogue D (1996) An Irish Flora, 7th edn. Dundalgan Press, Dundalk Webster J (1995) Sanctuaries and sacred places. In: Green MJ (ed) The Celtic world. Routledge, London Weir DA (1995) A palynological study of landscape and agricultural development in County Louth from the second millennium BC to the first millennium AD. Discovery Prog Rep 2:77–126 Wilmanns O, Brun-Hool J (1982) Irish Mantel and Saum vegetation. In: White J (ed) Studies on Irish vegetation. Royal Dublin Society, Dublin, pp 167–174 Yasuda Y (ed) (2001) Forest and civilizations. Roli Books and Lustre Press, New Delhi