Home     PhD Contents     PhD Chap 1     PhD Chap 2     PhD Chap 3     PhD Chap 4     PhD Chap 5     PhD Chap 6     PhD Chap 7     PhD Chap 8     PhD Chap 9     PhD Chap 10     PhD Chap 11     PhD Bibliography     PhD Appendices     PhD Plans    [PhD PDF]

10 Case study 2: Thornhill palisaded enclosure

[Please note that part of this chapter has been published in the Proceedings of the Royal Irish Academy]

10.1 Introduction

This chapter details the analysis of the quartz assemblage from the Thornhill Neolithic palisaded enclosure. Section 10.2 introduces the Thornhill excavation and outlines the main features excavated, and then describes other excavations and finds in the vicinity. Section 10.3 details the environmental background to the area, introducing the geological and palaeoecological information. Section 10.4 presents the analysis of the quartz assemblage beginning with the condition and sources of the quartz in Section 10.4.1, followed by a brief overview of the contexts containing quartz in Section 10.4.2; Section 10.4.3 presents the analysis of the cores and Section 10.4.4 details the debitage; the diagnostic types are detailed in Section 10.4.5, Section 10.4.6 presents the main contexts and Section 10.4.7 outlines a comparison to the non-quartz assemblage. Section 10.4.8 discusses the chaîne opératoires of the quartz component, and Section 10.5 concludes the chapter with an overview and discussion.

10.2 Excavation background

10.2.1 Thornhill excavation

The Thornhill site sits on a terrace on a low ridge (38m OD) on the left bank of the River Foyle close to the present-day river’s mouth (Figure 10-1). The development-led rescue excavation was initiated in 2000 after monitoring of topsoil stripping uncovered substantial archaeological evidence. Previous investigations in the immediate vicinity had discovered a lithic scatter, as well as the SMR site of a pair of ring-ditches (Logue and Ó Baoill 2009). The excavations investigated approximately 4000m², with the finds including lithics and ceramics along with floral and faunal remains. No C14 dates or specialist reports have been produced as of yet for Thornhill, with the preliminary stratigraphic report presented to me in June 2009 being an incomplete, unedited draft, hence no page numbers are referenced (Logue 2009 pers. comm.). The retention policy for the quartz was the retention of 100% of quartz excavated, with no distinction made between natural or artefactual material during the excavation (Logue 2009 pers. comm.). Based on ceramics and the type of structures uncovered, the site has been posited as of Early Neolithic date (Logue and Ó Baoill 2009). The Thornhill excavations were divided into three arbitrary areas (Plan 3). The following excerpts and summarises the stratigraphic report (Logue and Ó Baoill 2009).

Area 1
Area 1 was given just five days to excavate and the limited excavation produced 34 pits and four stakeholes.
"Nearly all of the pits and stakeholes were subsoil-cut and there was virtually no interlinking stratigraphy between individual features...The large number of pits and the apparent deliberate placing of artefacts, and sometimes several different types of artefacts, within many of the fills has been interpreted by the site director as marking out an area that was set aside for ritual activity" (Logue and Ó Baoill 2009).  

Area 2
Area 2 covered the majority of the excavation and was initially excavated in 21 trenches; these 21 were later merged into one large trench (Logue and Ó Baoill 2009). Eight palisades and five structures were excavated. Most of the palisades – some with in situ timber – had recuts, showing signs of replacement/repair. The construction methods for the palisades were varied, with most being slot trenches within which sat posts, planks or boards, packed with stones and clay.

Palisades 1 and 1a and associated features
Palisade 1 and 1a extended for 25m with a gap interpreted as the entrance to the site, which was “subsequently blocked by a number of large ritual pits” (Logue and Ó Baoill 2009); the palisade contained possible evidence for repair or replacement of the timbers in places. The pits (Contexts 325, 6002, 542 and 784) in the entrance gap contained recuts with evidence for the placement of upright stones/timbers and also contained numerous stakeholes.
"It would appear that all four large pits probably served a ritual function.  They were big enough to hold upright stones or timber posts. Roughly 0.60m north of pit C. 784 was the southern terminal of a palisade designated Palisade 1a during the excavation (C. 3029).  It is possible that the southern most of the conjoined pits (C. 325) actually cut the terminal of Palisade 1. If that is the case, it is tempting to see the four ritual pits as having been placed in what was the entrance between a palisade. Whether they post-dated the palisades by any length of time or whether they were contemporary and actually blocked access into the settlement via this point in the palisades, with visitors having had to walk around them to get in, their siting and effect could not be more dramatic. The erection of the ritual pits and the deliberate cutting of Palisade 1 could be an act signaling a change in layout, function or population of the settlement...The nature and function of the area of the site associated with the ritual pit group clearly changed with the construction of Structure A over the pit group. Whether Structure A itself had a ritual function is unknown" (Logue and Ó Baoill 2009).

Palisades 2 and 3 and associated features
Palisade 2 was recorded for 5.7m, with evidence for possible repair. The palisade possibly cut another group of ritual pits (C. 868, C. 6001 and C. 606 within cut C. 6014) – if so, it “may be one of the later features on site and reflects a new phase of activity after the pits had gone out of use. In this regard it is similar to Structure A, whose construction over the first group of ritual pits also seemed to be a deliberate reversal of a ritual blocking of the entrance into the site” (Logue and Ó Baoill 2009). Palisade 3 (8.25m length excavated) ran parallel to Palisade 2 and had evidence of timber replacement.

Palisade 2a and associated features
Palisade 2a (1.6m long with evidence of recuts) and a series of associated postholes, stakeholes, and a pit were to the north of Palisade 2; the associated features may be evidence for a different construction technique of the palisade or else “probably less likely, is that postholes and stakeholes (Contexts 2818-1490) along with the conjoined pits C. 6014, represent a deliberate blocking of the entrance into the settlement at Thornhill” (Logue and Ó Baoill 2009).

Palisades 3a and 4 and associated features
Palisade 3a – 6.5m excavated – and Palisade 4 – 11m excavated – are seen as the continuation of the same palisade, with six phases of activity; Palisade 4 was partially burned, with a subsequent possible dismantlement at that point, with a further set of ritual pits partially cutting the palisade. At the area of burning
"[s]even flint arrowheads (almost a third of the site total) were recovered from the ashy spread and other fills of the foundation trench at this point…The high number of flint arrowheads found in the immediate area also provides evidence for an episode of conflict at the settlement…Five large pits (Contexts 423, 424, 425, 854 and 855) ran in a line and were located immediately west of Palisade 4…Running from north-south these were C. 425, C. 424, C. 423 and C. 855. Another pit (C. 854) was located immediately west of pit C. 855 and ran beyond the limit of excavation. At least three different phases of activity were uncovered here with some of the pits cutting other pits and Palisade 4 cutting one of the pits (C. 855). Some of the pits were probably for storage while some might have served a ritual purpose" (Logue and Ó Baoill 2009).

Palisades 5, 6, 7, and 8
16m of Palisade 5 and 4m of Palisade 6 were excavated and both palisades continued beyond the eastern limits of the Area 2 excavation. The relationship between Palisades 5 and 6 was not fully resolved during the excavations – it is interpreted that Palisade 5 cut Palisade 6. Palisade 8, excavated for 2m and with evidence for recuts, is interpreted as the terminal of the palisade surrounding the settlement at Thornhill – “[d]espite being immediately adjacent to Palisade 3a the relationship between the two was not clearly defined during the excavation” (Logue and Ó Baoill 2009). It is uncertain whether Palisade 7 – 7m in length, which had recuts – was in fact a palisade or the remains of a structure, but was designated as a palisade.

Structure A
Structure A was approximately 6m by 4m, defined by three foundation trenches and postholes and stakeholes, with the southern trenches cutting through the group of conjoined pits (C. 325, C. 6002 and C. 542; all within pit cut 6003); “the building was, therefore, later than them and Palisades 1 and 1a.  Whether Structure A retained a ritual function is unknown” (Logue and Ó Baoill 2009).

Structure B
Structure B was approximately 9m by 4m and divided into at least two rooms with an entrance in the northern wall.
"The entrance…had a shallow depression within the doorway into which a large flat sill stone had been placed. To the rear of the entrance a shallow curvilinear bedding trench may mark an internal screen-like porch...North of the northern wall slots…were a series of large features.  From west to east, these were a slot cut (C. 581) and four large pits (C. 6012, C. 6013, C.  671 and, most easterly located C. 361)" (Logue and Ó Baoill 2009).

Structure C
Approximately 200 stake-holes and post-holes were excavated in southern part of Area 2 and 80 may be associated with a round structure, 6m in diameter. 52 of these stakeholes created a double-walled structure with a possible porch-like entrance in the southeast with another gap in the northeast, but it was uncertain if the latter gap was deliberate or a gap created by site disturbance.
"No obvious hearth or occupation deposits relating to the structure were uncovered during the excavation and its function remains uncertain. The outline of Structure C was intermixed with that of Structure B, a rectangular structure. No clear stratigraphic relationship was determined between the two structures. Although of a different construction that would have left more ephemeral remains, it would have been expected that more of a trace of Structure C would have been encountered where the buildings overlapped, had Structure C been the later building" (Logue and Ó Baoill 2009).

Structure D
Structure D was initially noted as a linear feature by topsoil stripping.
"The main linear feature has been interpreted as the eastern foundation trench of the wall of a Neolithic building assigned…Structure D was, like Structure E, aligned north-east/ south-west. The western spur off the main linear cut has been interpreted as representing an internal partition within the structure and the large pit/ posthole (C.3099) as an internal load-bearing posthole for a roof supporting timber. Based on these features and the [other] unexcavated features, it is estimated that Structure D was roughly 8m long by 4m wide, similar in dimensions to Structure E" (Logue and Ó Baoill 2009).

Structure E and linear feature
Structure E was 7.60m long and 4.40m wide externally, defined on three sides by linear foundations with an entrance gap. 
"Although the complete outline of the structure was uncovered during the excavation, only a 3.60m long section constituting most of the northern end was fully excavated. A linear feature with a rounded terminal (C. 3305) was also partially uncovered. The function and relationship to Structure E is uncertain. The linear feature was a maximum recorded length of one metre by 0.70m wide and was 0.30m deep. The linear feature is clearly cut by the northern end of the later Neolithic building. As the feature was not further excavated, it is difficult to interpret with any authority. However, it may be that it represents the terminal of a palisade wall slot. If this is the case, it would show a very clear change in layout and function in this part of the Thornhill site" (Logue and Ó Baoill 2009). 

Area 3
While archaeological features and strata were uncovered and recorded in Area 3, none were excavated due to time constraints. The features included
"linear features and slots, post holes and stakeholes and the possible southwards continuation of defensive Palisade 4 (C. 427)…Area 3 clearly included significant archaeological remains, including from an examination of the site plan, probable structural remains (both palisades and buildings) as well as many other features and strata. It was not possible to verify the nature of these archaeological strata and features due to the time constraints" (Logue and Ó Baoill 2009).

londonderry prehistoric monuments

Figure 10-1 Thornhill environs. Elevation data from EPA (2004b)

Top of Page

10.2.2 Other Londonderry excavations

Figure 10-1 highlights that the palisaded enclosure at Thornhill sits in a landscape dotted with numerous upstanding Neolithic monuments such as megalithic tombs and standing stones, along with a number of lithic scatters identified in and around the banks of the Foyle. About one km to the northwest of Thornhill, a Middle Bronze Age lake settlement site was excavated after archaeological material was noted exposed in a drainage ditch (Ó Néill et al. 2007). A few hundred metres to the west, an excavation was undertaken in order to compliment the excavations at Thornhill and investigate crop marks identified in aerial photography ahead of development on the Thornhill site; this excavation opened up 862m² of trenches with no prehistoric structural remains noted, and a limited quantity of flint and quartz artefacts were found along with Neolithic ceramics (Ó Néill et al. n.d.). Stray finds of lithics – flint, quartz, and stone axes have been found in fields nearby, as well as along the shores of the Foyle (Figure 10-1).

One km to the southeast of Thornhill, on the right bank of the Foyle, a sub-rectangular Neolithic structure was partially excavated during a rescue excavation at Enagh (Figure 10-1), along with a Neolithic possible hut a few hundred metres away; finds included lithics and Neolithic ceramic; a Bronze Age site was also excavated nearby during the same excavations (Mc Sparron 2003). Both of these Neolithic sites overlook the Foyle and are inter-visible with the Thornhill complex (Mc Sparron 1998, 2003). Three km south of Thornhill, another Neolithic structure was excavated at Caw (Figure 10-1), again in a development-led excavation (Bowan 2003). The Caw structure produced a date of 3905–3707 cal BC from charcoal in a wall slot (Bowan 2003), while the Enagh structure produced two dates – 4230-3790 cal BC from charcoal in a construction slot, and 3770-3530 cal BC from hazelnut shells (Mc Sparron 2003, 11). The possible hut at Enagh produced a date of 3950-3530 cal BC from charcoal (Mc Sparron 2003, 12).

Top of Page

10.3 Environmental background

10.3.1 Geology

The following description of the central Northern Irish geology is abbreviated from Long and Mc Connell (1997). The predominant bedrock around Thornhill is, similar to Belderrig in the previous chapter, the Dalradian Supergroup (Figure 10-2); here the Dalradian sedimentary series were metamorphosed, forming quartzites, psammites, psammitic schists, and pelitic schists.  Later tectonic activity formed granites east of Lough Foyle with a consequent metamorphosing of the local rock; minor igneous intrusions – metadolerite sills – occurred before this granite formation, which initially formed as quartz-bearing basaltic, doleritic, and gabbroic tholeiites. On the southern and eastern shores of Lough Foyle are a sequence of carboniferous rocks – limestones, sandstones, siltstones, mudstones, and conglomerates (with quartz pebbles noted). The eastern boundary of the carboniferous sequence is formed by flint-bearing chalks whose eastern boundary is in turn formed by the Antrim Tertiary basalt plateau; at this time, further dolerite intrusions occurred on present day Lough Foyle’s northeastern shore (Figure 10-2).

This simplified geological description highlights that Thornhill is positioned at a source of quartz pebbles with vein quartz further afield – Long and Mc Connell (1997) note a very pure quartz vein which is up to 15m wide some 40km to the west-northwest of Thornhill. In situ flint is available c. 25 km to the east, with quartzites available the same distance to the west, and sandstones, siltstones, and mudstones available to the east.

lough foyle geology

Figure 10-2 Central Northern Ireland bedrock geology. Data from GSI (2005b)

I conducted a rapid survey over one day to identify possible sources of quartz, focusing on the left bank of the Foyle close to the Thornhill site, covering around 3km of the shore (Figure 10-3). The right bank was not surveyed due to the expanse of mud flats and apparent lack of rocky shore; little could be surveyed away from the shoreline due to a lack of outcrops in the vicinity. On the shore, quartz veins (approximately 20cm max. width) were noted in the steep-sided, schist cliff face on the shoreline approximately 200m southeast from the site; much of the cliff is covered in dense vegetation so the extent of the availability is uncertain.

At the point of the quartz veins the shoreline is dominated by schist shingle. Following the river to the mouth to the northeast, the shingle gives way to pebbles/cobbles, with quartz (flint pebbles are also common) becoming more abundant and sizeable approximately at a point due east of Thornhill, with this continuing for a few hundred metres. The rocky shore is then replaced by mud flats for a few hundred metres, with the sandy/rocky shore once again appearing. At this point, quartz pebbles and cobbles are abundant, with numerous large cobbles (up to approximately 20cm max. length) apparent; this stretch continues on for a further few hundred metres to the north. 40 quartz cobbles were tested, with 90% proving to be smooth-grained, high quality quartz lacking in detrimental major fracturing. Therefore, at least at present, quartz is abundant 200m from Thornhill, and especially abundant in sizeable beach cobbles approximately 2km northeast.

foyle estuary quartz sources

Figure 10-3 Quartz sources on Foyle Estuary. Geological data from GSI (2005b)

Top of Page

10.3.2 Palaeoecology

The River Foyle and Lough Foyle area’s quaternary deposits were principally laid down during the drumlin readvance and the subsequent glacial retreat, leaving a series of drumlins, eskers, moraines, and flat-topped terraces (McCabe 2002; NIEA 2009). Cooper and Gault (2002) note that post-glacial sea levels were about -30m OD c. 9500 cal BC with a subsequent rise to about +3m OD c. 5500 cal BC after which the sea level fell to the present-day level. Palaeoecological investigations were carried out on a Middle Bronze Age lake settlement site a few hundred metres northwest and down slope from Thornhill  providing a localised record around the lake (Ó Néill et al. 2007) (Figure 10-1). At c. 3000 cal BC the environs was a closed woodland represented by 90% arboreal taxa, with pine, oak, alder, and hazel as the predominant trees (Ó Néill et al. 2007, 42). For the Bronze Age, Ó Néill et al. (2007, 46) suggest the record
"reveals a gradual change from a closed, wooded landscape around the lake basin to a more open one in which farming activity is apparent towards the end of the third millennium BC. There is little evidence for specific tree clearance, with only gradual reductions noted in hazel and oak. Charcoal values do not rise until after grass and herb values start to increase, and burning does not seem to have been a factor in the general opening of the woodland".

Unfortunately the Neolithic section of the record proved to be problematic (Logue and Ó Baoill 2009). However, Plunkett (2009 pers. comm.) has reported that Neolithic section becomes problematical at c. 3000 cal BC; for c. 4000-3500 cal BC the “pollen record suggests a heavily wooded landscape (oak-hazel-elm-pine, with a good deal of alder carr probably around the bog) in the bog's catchment at this time”. Therefore, with the Thornhill site posited as Early Neolithic, during this time the area a few hundred metres away remained heavily forested, with the woodland opening up in the Bronze Age. This identification of the lack of widespread deforestation in the Neolithic suggests that while the inter-visibility of Neolithic structures noted above is theoretically possible, the woodland would negate this.

Top of Page

10.4 Analysis

10.4.1 Quartz condition and sources

The following analysis is concerned with vein quartz, therefore terms such as artefact imply vein quartz artefacts unless other materials are specifically mentioned. The Thornhill vein quartz assemblage was 4047 pieces of which 813 were artefacts, nine were possible artefacts, 74 were indeterminate, with the rest (78%) deemed to be natural. There were also six rock crystal debitage. The vast majority of the quartz was in a fresh condition with sharp edges, with a minor amount of artefacts having damaged to significantly damaged edges. Two cores, 17 debitage, and two indeterminate pieces were either burnt or possibly burnt; the two cores were burnt, while the remainder were possibly burnt. The natural component was dominated by sub-angular to rounded pieces, along with some angular pieces – these are found in the natural soils in the area, which is glacial till. Many of the natural pieces in the glacial till can take the form of knapped artefacts, making judgements on the more angular pieces difficult. Consequently, half of the indeterminate pieces were angular fragments and 13 were possibly bipolar cores. Seven indeterminate pieces had rounded or sub-rounded edges; two of the possible bipolar cores, three other possible cores and two possible retouched pieces. These seven pieces were difficult to determine because the degree of edge roundedness ruled them out as definitive artefacts, but it was speculated that some may have been either collected/deposited as ‘antique’ artefacts or as items that looked like artefacts – six of these seven came from the structures or pits.

21% of the vein quartz artefacts (53% of the cores and 13% of the debitage) were identified as quartz-beach, 4% as quartz from schist bedrock, and the rest were of indeterminable source. 23% of the vein quartz artefacts (58% of the cores and 14% of the debitage) had at least 1% cortex. Compared to the Belderrig assemblage, therefore, there is a much greater degree of cobble usage at Thornhill. This is not necessarily related to the greater proportion of bipolar cores at the site (see below). While 53% of the bipolar cores had cortex and 45% were identified as quartz-beach, 61% of the platform cores had cortex and 52% were identified as quartz-beach: this means that 85% of the cortical cores from both core groups were quartz-beach. Allowing for some quartz-beach cores to have been completely decortified thus reducing their perceived total, a rough estimate could be around a 60-80% quartz-beach assemblage for the initial cores. As noted in Section 10.3.1, good quality cobbles are available in abundance in the environs of Thornhill. Quartz veins are also available in the schist bedrock, but it is unclear to what extent quartz was quarried. The largest piece of quartz from the assemblage was a natural, sub-rounded block weighing 335g, found in a posthole; this large block was presumably derived from the glacial till, highlighting that sizeable blocks were available from there. Whether such quartz blocks from the till were chosen for knapping is uncertain, and it is possible that veins were quarried instead. It is also possible that some of the cobbles used were also collected from the glacial till, but no clear evidence such as striations on the cobbles were noted. An analysis of striations on glacial cobbles has shown that they can appear on 33% of quartz cobbles from glacial deposits (Wentworth 1936, 86-7).

Comparing the cortex proportions of the three datasets shows that in terms of the cores, the Belderrig and Thornhill assemblages have substantially different proportions of cortical cores; while the Experiment dataset has more cortical cores, the Thornhill assemblage has substantially more cores with >49% cortex (Table 10-1). However, looking at the debitage gives a completely different picture with Thornhill having 14% cortical debitage and Belderrig having 8%, both far fewer the Experiment assemblage at 40%. Unlike the other datasets, Thornhill had, as with the cores, proportionally more >49% cortex flakes than 1-49% cortex flakes.

Artefact Dataset Cortex
0% 1-49% >49% Total 0% 1-49% >49%
Count Count Count Count % % %
Core Belderrig 36 17 2 55 65.5 30.9 3.6
Experiment 58 98 11 167 34.7 58.7 6.6
Thornhill 57 26 51 134 42.5 19.4 38.1
Debitage Belderrig 3328 147 131 3606 92.3 4.1 3.6
Experiment 865 320 261 1446 59.8 22.1 18.0
Thornhill 530 37 52 619 85.6 6.0 8.4
Total Belderrig 3364 164 133 3661 91.9 4.5 3.6
Experiment 923 418 272 1613 57.2 25.9 16.8
Thornhill 587 63 103 753 78.0 8.4 13.7

Table 10-1 Cortex proportions. Thornhill, Belderrig, and Experiment cores and debitage

The relatively high proportion of >49% cortex cores at Thornhill suggests that the cores at Thornhill were not decortified to a great extent before being brought to the site for use or deposition (as was surmised from some of the Belderrig cores). The low proportion of cortical flakes suggests that the debitage assemblage does not represent a knapping floor assemblage to a great degree, and that the ‘missing’ cortical flakes are either elsewhere in the unexcavated areas of the site, or at a further remove from the site altogether. There is no apparent distinction between contexts in terms of cortex proportions of the artefacts, with all having the same low proportion, but with the topsoil/surface having a slightly lesser proportion of cortical artefacts than the rest (the differences between core types is discussed below). Therefore, there is a clear mismatch in the cortex proportions between cores and debitage indicating a complex pattern of core and debitage use and/or deposition throughout the site’s contexts.

Almost all the quartz had a smooth-grained appearance with just three platform cores, one possible radially split core, and one bipolar flake having a sugar-grained appearance and three bipolar cores, one platform core, four platform flakes, two bipolar flakes and one <20mm debris having a sugar/smooth-grained appearance. Table 10-2 provides the most common visual characteristics, which breaks down into eight groups, and covers 95% of the >10mm artefacts; the remaining 5% are slight variations of the main groupings. All of these have a smooth-grained appearance, and 78% are semi-opaque; the majority of the semi-translucent artefacts have a cloudy hue and cloudy glass appearance. A third were vitreous with a metallic grey hue.

Grain Opacity Hue Appearance Core Debitage Core/
debitage
Retouched
pebble
Total
Smooth Semi-
opaque
Cloudy Cloudy glass 22 62 - - 84
Metallic grey Vitreous 57 196 1 - 254
Milky white Vitreous 27 113 - - 140
White Dull white 2 9 - - 11
White Vitreous 11 66 1 1 79
Semi-
translucent
Cloudy Cloudy glass 3 124 - - 127
Metallic grey Vitreous 2 18 - - 20
Milky white Vitreous - 12 - - 12

Table 10-2 Thornhill vein quartz. Grain, opacity, hue, appearance for 95% of ≥10mm artefacts which fall in main eight groups

Overall, 20% of the artefacts had a milky white hue; this quartz was vitreous, and had a plastic-like feel and appearance which was for the most part quite distinctive. The distinction between this milky white quartz and the white and metallic grey quartz was not, however, absolute, with a certain degree of gradation and artefacts feeling and appearing less or more ‘plastic’ than others. A number of milky white, vitreous natural quartz pieces were noted in the assemblage, suggesting that this type of quartz is present in the glacial till. While no thin sections were taken of the Thornhill artefacts, Julian Menuge (2009a pers. comm.) examined two artefacts macroscopically for grain size; one sample had grain sizes of 2-7 mm, while the other sample was 1-3 mm. Therefore, the vein quartz is broadly similar in grain size to the Belderrig quartz.

10.4.2 Contexts overview

Table 10-3 provides a summary account of the proportions of artefacts, indeterminate pieces and natural pieces by context, with Appendix A- 86 providing a full account. In the following analysis a number of contexts with small quantities of artefacts have been variously grouped together as ‘other’, and the contexts are described as ‘context summary’, with the full list of contexts in Appendix A- 86. The artefacts came from over 50 individual contexts and over 100 fills, as well as 28 strata. 60% of the artefacts came from the topsoil and 18% came from the pits, with the remaining contexts having 8% or less each; excluding the topsoil, the average artefact per context was five.

Context summary Artefact Indeterminate Natural Total
Count Count Count Count
Other 45 15 116 176
Palisade 66 - 261 327
Pit - possible pit 147 7 623 777
Structure 27 5 68 100
Stratum 49 2 106 157
Topsoil/surface 494 45 1977 2516
Total 828 74 3151 4053
% % % %
Other 25.6 8.5 65.9 100.0
Palisade 20.2 - 79.8 100.0
Pit - possible pit 18.9 0.9 80.2 100.0
Structure 27.0 5.0 68.0 100.0
Stratum 31.2 1.3 67.5 100.0
Topsoil/surface 19.6 1.8 78.6 100.0
Total 20.4 1.8 77.7 100.0

Table 10-3 Thornhill quartz. Context summary: artefact, indeterminate, natural

Of the main quartz-bearing features, the structures had relatively few artefacts, but had the greatest ratio of artefacts to indeterminate/natural pieces. The pits and topsoil had similar ratios of artefacts to each other, but the Area 1 pits had a greater proportion of artefacts than the Area 2 pits. The 19 burnt or possibly burnt artefacts came from the topsoil (n=6), pits (n=7), palisades (n=2), structures (n=2), a linear slot cot (n=1), and an uncertain context (n=1). The pits, therefore, are over-represented with the burnt or possibly burnt artefacts, with three of the seven from the pits coming from one pit.

Top of Page

10.4.3 Cores

16% of the Thornhill vein quartz artefacts were cores, with a further two artefacts categorised as core/debitage and another six as possible cores (Table 10-4); in addition to these were the 13 indeterminate pieces that were possibly bipolar cores mentioned above. Just over half of the cores were bipolar, a third were platform cores, with the remainder being combined bipolar/platform cores and radially split cores (Appendix A- 87, Appendix B- 24 and Appendix B- 25). Given that a bipolar knapping event will produce a multitude of cores compared to a direct percussion event, this ratio of 5:3 does not imply a dominance of a bipolar technology at the site and instead may imply the converse, as does the dominance of platform flakes. A minor proportion of cores were a combination of techniques, with both bipolar-on-platform and platform-on-bipolar cores present, especially in Area 1 which also had, proportionally, substantially more platform cores than Area 2. A quarter of the cores were core fragments, with the majority of these from the topsoil and none from the structures (Table 10-5). This may suggest post-depositional fracturing of the cores from the topsoil as opposed to core fragmentation during knapping. Many of the cores exhibited step fractures on the cores, with three of the platform cores being heavily step fractured.

Artefact Count %
Core 134 16.4
Core/debitage 2 0.2
Flake 454 55.5
Debris 218 26.7
Retouched pebble 1 0.1
Possible core 6 0.7
Possible debitage 3 0.4
Total 818 100.0

Table 10-4 Thornhill vein quartz artefacts and possible artefacts, excluding hammerstones

Context summary Core Count % context
Palisade Complete 10 83.3
Fragment 2 16.7
Pit Complete 21 87.5
Fragment 3 12.5
Topsoil Complete 42 62.7
Fragment 25 37.3
Structure Complete 5 100.0
Other Complete 24 92.3
Fragment 2 7.7
Total Complete 102 76.1
Fragment 32 23.9

Table 10-5 Thornhill core fragmentation by context

Over half of the 72 bipolar and three bipolar-on-platform cores came from the topsoil with the pits containing 11% and remaining contexts containing less than 10% each; almost all of the fragments came from the topsoil, with one from the palisade and one from ‘other’ (Table 10-6). The palisade contained six bipolar cores, three of which formed a refit set knapped on a block of vein quartz.

Context summary Complete Fragment Total Complete Fragment Total
Count Count Count % % %
Topsoil/surface 22 20 42 41.5 90.9 56
Pit - possible pit 8 - 8 15.1 - 10.7
Stratum 7 - 7 13.2 - 9.3
Palisade 5 1 6 9.4 4.5 8.0
Linear/slot cut 4 - 4 7.5 - 5.3
Structure 4 - 4 7.5 - 5.3
Other 3 1 4 5.7 4.5 5.3
Total 53 22 75 100.0 100.0 100.0

Table 10-6 Thornhill bipolar and bipolar-on-platform cores: context

44% of the bipolar and bipolar-on-platform cores were identifiable cobbles, with remainder either unidentifiable cobbles (lack of cortex) or from blocks; almost half were non-cortical, with the pits and structures having the greatest proportions of cortical cores (Appendix A- 88). Many of the cores are relatively large, and thus do not appear to have been worked to exhaustion, with a general spread in sizes and a number of outliers in the upper range; Figure 10-4 and Figure 10-5 provide the length by width and width by thickness of the cores and also include the radially split cores. Figure 10-6, Figure 10-7, Figure 10-8, and Figure 10-9 provide boxplots for both complete and fragment bipolar cores. Three of the bipolar cores weighed over 80g and one weighed over 180g, while the overall median weight of complete bipolar cores was 20g (Figure 10-9).

=quartz bipolar platform cores quartz bipolar platform cores

Figure 10-4 [top] Thornhill. Complete bipolar, bipolar-on-platform and radially split cores: length/width ratio

Figure 10-5 [bottom] Thornhill. Complete bipolar, bipolar-on-platform and radially split cores: width/thickness ratio

=quartz bipolar cores boxplot quartz bipolar cores boxplotquartz bipolar cores boxplotquartz bipolar cores boxplot

Figure 10-6 Thornhill. Boxplot. Bipolar core length

Figure 10-7 Thornhill. Boxplot.Bipolar core width

Figure 10-8 Thornhill. Boxplot.Bipolar core thickness

Figure 10-9 Thornhill. Boxplot. Bipolar core weight

Appendix A- 89 provides the means for complete and fragment bipolar cores. Using the log transformations, the 50 complete bipolar cores were analysed for any discernable differences for the various metrics between the Areas, contexts, and cortex presence. The differences in means of the five metrics for the three variables were not significant (Appendix A- 90 and Appendix A- 91). The three bipolar-on-platform cores all fell in the upper range for length and two for width and thickness as well.

As noted, the complete bipolar cores were generally large. The Thornhill and Experiment bipolar cores were compared using the log transformations, with the Belderrig dataset excluded due to its low numbers. Between the Thornhill and Experiment datasets, the difference in means for all metrics was significant (Table 10-7), with the Thornhill cores heavier and larger in all dimensions, with a lower length/weight ratio. Figure 10-10, which also includes the two from the Belderrig assemblage, highlights that the Thornhill bipolar cores are in the upper size range, especially in terms of width and thickness thus accounting for the low length/weight ratio.

Source Dependent Variable Type III
Sum of Squares
df Mean
Square
F p
Dataset Weight (LOG) 8.003 1 8.003 60.856 0.000
Max Length (LOG) 0.542 1 0.542 25.134 0.000
Max Width (LOG) 0.697 1 0.697 30.696 0.000
Max Thickness (LOG) 1.134 1 1.134 48.663 0.000
Length/weight ratio (LOG) 4.380 1 4.380 65.686 0.000

Table 10-7 GLM. Thornhill and Experiment complete bipolar cores

The platform and platform-on-bipolar cores came primarily from the topsoil and pits, with the majority of the core fragments from the pits (Table 10-8). Half (n = 26) were single platform cores, which included five single platform-on-bipolar cores; one of the platform-on-bipolar cores from Area 1 was retouched as a scraper; this medium-sized core also contained a direct percussion flake removal, creating a surmised finger rest on the scraper Appendix B- 24.[25]

foyle estuary quartz sources

Figure 10-10 Scatterplot: length, width, thickness. Complete bipolar cores for three datasets

A greater than average proportion of the single and multiplatform cores came from the pits while a greater than average proportion of the dual, opposed and dual, right angled cores came from the topsoil (Appendix A- 92). Over half were cobble-based, with the remainder either unidentifiable cobbles or from blocks, with three non-cobbles cores having cortex; a greater proportion of the single platform cores were cobble-based and cortical cores (Appendix A- 93). Four single platform and one dual, right angled core were conically-shaped cores.

Context summary Complete Fragment Total Complete Fragment Total
Count Count Count % % %
Topsoil/surface 19 1 20 41.3 20.0 39.2
Pit - possible pit 13 2 15 28.3 40.0 29.4
Palisade 5 1 6 10.9 20.0 11.8
Stratum 3 - 3 6.5 - 5.9
Posthole 1 - 1 2.2 - 2.0
Structure 1 - 1 2.2 - 2.0
Linear/slot cut 1 - 1 2.2 - 2.0
Other 3 1 4 6.5 20.0 7.8
Total 46 5 51 100.0 100.0 100.0

Table 10-8 Thornhill. Platform and platform-on-bipolar cores: context

Two single platform cores and two dual, opposed cores were core-on-flakes and were both small and large cores; one of these single platform cores also had distal impact marks. Overall, five single platform and five multiplatform cores (20% of total platform cores) had distal impact marks suggesting platform-on-anvil cores, and these included both smaller and larger cores; two of these single platform also had possible evidence of bipolar strikes. Another five platform-on-bipolar cores also exhibit distal impact marks, but in these cases from the bipolar striking.

As well as the bipolar core scraper mentioned previously, two of the platform cores were retouched – a mid-sized, retouched complete multiplatform core came from the same fill as the scraper, and a small, retouched single platform core fragment came from Pit 607. It is unclear if the latter is a fragment as a result of a core break during knapping, or through breakage during use. One refit set was noted – in Area 3, a possible post/stakehole (C.2349) produced a single platform core, with two refitted flakes; along with these three artefacts were 10 angular fragments categorised as indeterminate, but possibly natural, as well as 6 natural quartz, one of which weighed 335g making it the largest piece of quartz in the assemblage.

Appendix A- 94 provides the metrics’ means by platform core type for the complete cores; all the core types’ means have large standard deviations, highlighting the wide range in sizes for all types. As with the bipolar cores, the complete platform cores were analysed for any discernable differences between the core types, contexts, Areas, cortex presence, and quartz type for the various metrics, also using the log transformations. For core type, context, and Area, the differences in means for the various metrics were not significant (Appendix A- 95). For cortex presence, however, the difference in means was significant for all metrics (Appendix A- 95), with the cortical cores being on average heavier and larger in all dimensions, and with a lower height/weight ratio than the non-cortical cores (Table 10-9). For the quartz sources (beach and generic quartz), the differences in means were significant for all metrics except for length/weight ratio, with the quartz beach larger (Appendix A- 95); the differences between the cores with cortex, and the cobble-based ones can be discerned by the increase in p values for all metrics for the quartz types, highlighting the slight difference between the cobble-based and block-based cortical cores. Figure 10-11 and Figure 10-12 show scatterplots of length/width and width/thickness, highlighting the lack of grouping by core type and the dominance of cortical cores in the middle and upper size ranges.

Cortex Weight Max Length Max Width Max
Thickness
Length/Weight
ratio
Non-cortical N 18 18 18 18 18
Mean 51.96 41.07 33.38 25.06 1.44
Std. Deviation 55.75 16.65 13.63 8.33 0.90
Median 33.82 36.25 29.95 23.75 1.30
Cortical N 28 28 28 28 28
Mean 107.91 55.25 45.67 34.86 0.94
Std. Deviation 97.21 22.64 19.06 15.78 0.77
Median 56.44 47.45 41.25 30.10 0.83
Total N 46 46 46 46 46
Mean 86.02 49.70 40.86 31.03 1.13
Std. Deviation 87.21 21.48 18.02 14.11 0.85
Median 48.13 45.25 37.65 27.5 0.89

Table 10-9 Thornhill. Complete platform cores: mean and median by cortex presence

The Thornhill cores contained both a much greater size range and substantially smaller cores compared to the Belderrig assemblage, with the Belderrig cores producing a tighter cluster of cores in the middle and upper range of length/width/thickness (Figure 10-13). Using the log transformations, five metrics of the Thornhill and Belderrig cores were compared. The differences in means between the two assemblages were significant for all metrics (Table 10-10). Overall, the Belderrig cores’ means for all metrics were greater, with smaller standard deviations, and a much lower length/weight ratio which is the result of much thicker cores (Appendix A- 96). Appendix A- 96 includes the medians for the metrics, highlighting the greater size of Belderrig cores.

Source Dependent Variable Type III
Sum of Squares
df Mean
Square
F p
Dataset Max Length (LOG) 0.186 1 0.186 8.911 0.004
Max Width (LOG) 0.151 1 0.151 6.919 0.010
Max Thickness (LOG) 0.178 1 0.178 8.731 0.004
Weight (LOG) 1.922 1 1.922 13.618 0.000
Length/weight ratio (LOG) 0.911 1 0.911 14.490 0.000

Table 10-10 GLM. Belderrig and Thornhill complete platform cores

quartz platform cores ratio

Figure 10-11 Thornhill. Complete platform cores: length/width scatterplot by cortex presence

quartz platform cores ratio

Figure 10-12 Thornhill. Complete platform cores: width/thickness scatterplot by cortex presence

quartz platform cores ratio

Figure 10-13 Complete platform cores. Belderrig and Thornhill

Top of Page

10.4.4 Debitage

Debitage accounted for 82% (n= 672) of the vein quartz artefacts (Appendix A- 97), with a further six pieces of rock crystal debitage (Appendix A- 98). The following analysis focuses on the vein quartz. Flakes were 68% of the debitage, a higher proportion than the Belderrig or Experiment datasets; the topsoil contained 59% flakes, with most of the other contexts having 80% or over. There was only a small proportion of <10mm debris, highlighting the difference in contexts between the knapping floors of the experimental and Belderrig assemblages and the varied contexts of the Thornhill, none of which appear to be knapping floors or the deposition of small knapping floor artefacts (Figure 10-14). This stands in contrast to the high proportion of cores in the Thornhill assemblage compared to the other datasets. As with Belderrig, there is an apparent diminutive proportion of <20≥10mm slivers compared to the Experiment dataset; however, the sliver/debris ratio for <10mm debris is greater for Thornhill (60:40) compared to Belderrig (52:48), as seen in the sharper rise to the right of the chart for Thornhill (Figure 10‑14). A similar proportion of <10mm debris came from the topsoil, the structures, and the pits, with less from the palisade, and none from any other contexts.

quartz debris proportions

Figure 10-14 Debris proportions: Experiment, Belderrig, and Thornhill

There were 454 flakes, with 54% from the topsoil; 14% were identified as bipolar flakes (Table 10-11 and Appendix B- 26). The palisades and structures had above average proportions of bipolar flakes, and, as with the bipolar cores, the pits had below average proportions of bipolar flakes. Overall, 5% of flakes were complete with 3% (n=2) of the bipolar flakes complete (Appendix A- 99) – for the Experiment dataset, a much higher proportion of bipolar flakes were complete (37%) compared to the platform flakes. While the pits contained 21% of the flakes, they contained 48% of the complete flakes, and the structures contained 4% of the flakes but contained 13% of the complete flakes; many contexts had no complete flakes (Appendix A- 99). This discrepancy in the distribution of complete flakes suggests three possibilities:
1. A part of the higher rate of fragmentation of the flakes in the palisades, strata, and topsoil does not originate from fracturing during knapping, but rather from use/post-depositional factors.
2. The greater proportion of complete flakes in the pits and structures originates from the purposeful deposition of complete flakes in these contexts.
3. A combination of the above two.

Using correspondence analysis (see Shennan 1997), Figure 10-15 plots the relationship between the flake fragmentation and the five main contexts. While the complete flakes are far from all contexts, they are closest to the pits and structures and furthest from the topsoil; the mesial fragments are closest to the topsoil, while the closest relationships are stratum and distal missing, and structures and proximal missing.

Context summary Bipolar Platform Total Bipolar Platform Total
Count Count Count % context % context % context
Other 2 17 19 3.1 4.4 4.2
Pit/posthole - 1 1 - 0.3 0.2
Posthole - 3 3 - 0.8 0.7
Linear/slot cut 1 1 2 1.6 0.3 0.4
Palisade 14 30 44 21.9 7.7 9.7
Pit - possible pit 9 84 93 14.1 21.5 20.5
Structure 7 11 18 10.9 2.8 4.0
Stratum 1 27 28 1.6 6.9 6.2
Topsoil/surface 30 216 246 46.9 55.4 54.2
Total 64 390 454 100.0 100.0 100.0

Table 10-11 Thornhill vein quartz flakes. Technique and context

quartz context correspondence

Figure 10-15 Thornhill. Correspondence analysis. Flake fragments and context summary

Appendix A- 100 provides a full list of fragment types for bipolar and platform flakes, with Table 10-12 providing a summary account. As noted in the previous chapter, for the categorising of flakes the definitive bipolar flakes were labelled ‘bipolar’ while platform flakes and flakes in general were labelled ‘platform’, and no third category of flake type was used. While this will possibly inflate the perceived amount of ‘platform’ flakes, in the fragment groups of mesial, distal, and proximal missing (Table 10-12), these three groups account for just 30% of the overall flake fragments.

Fragment grouped Bipolar Platform Total Bipolar Platform Total
Count Count Count % % %
Complete 2 21 23 3.1 5.4 5.1
Mesial L/R missing - 5 5 - 1.3 1.1
Distal missing 17 90 107 26.6 23.1 23.6
Proximal missing 2 35 37 3.1 9.0 8.1
Lateral 2 19 21 3.1 4.9 4.6
Mesial 1 68 69 1.6 17.4 15.2
Distal 2 32 34 3.1 8.2 7.5
Proximal 38 120 158 59.4 30.8 34.8
Total 64 390 454 100.0 100.0 100.0

Table 10-12 Thornhill. Bipolar and platform flake fragment groups

54% of the assemblage’s ≥10mm debitage had complete or partial platforms, which is a much greater proportion than the Belderrig or Experiment datasets (which had 33-35%), but with a similar proportion of complete flakes to Belderrig. Looking at the platform flakes, almost all the complete platforms were flat, with just one faceted platform (Appendix A- 101). There was a high degree of platform collapse, with may relate to the relative thinness of the flakes (see below). Unlike the Belderrig assemblage, that had no evidence for platform preparation, 11% of the platform flakes had platform preparation – 8% prepared and 3% abraded. Two debitage exhibited evidence for the use of a punch for indirect percussion – one proximal flake fragment with a prepared platform that had collapsed and one rock crystal <10mm sliver.

The greater presence of bulbs and compression rings on the Thornhill platform flakes highlights the different type of quartz compared to the Belderrig quartz – 6% of the flakes had bulbs compared to 2% in the Belderrig assemblage and 5% of the experimental flakes; 3% of the Thornhill flakes had compression rings, while the Belderrig assemblage had none visible and the experimental assemblage had 0.3%. Just over half of the flake terminations were feather, with very few step terminations, which stands in contrast to the degree of step terminations visible on the cores; a much higher proportion of plunging and retouched terminations were in the Thornhill assemblage compared to the Belderrig. 45% of the flakes were straight. For the bipolar flakes, the majority of the platforms were collapsed; for those with complete platforms, the majority were rounded (Appendix A- 102). No bulbs or compression rings were present. 58% of the flakes were straight and 60% of the terminations were feather.

The breaks were dominated by longitudinal only breaks, with few transverse and transverse only breaks, and with very few longitudinal or transverse clean breaks (Appendix A- 103). The platform flakes had a slightly lesser proportion of siret breaks than the Belderrig assemblage and the bipolar component had just one. Little patterning could be discerned in terms of the distribution of break types by context, except that the palisades had a greater proportion of both siret and clean breaks than other contexts. While the siret proportions of the Belderrig assemblage were compared to the experimental assemblage to ascertain whether soft or hard hammer was used, this analysis was not conducted on the Thornhill assemblage because the debitage appears to have been a selection of certain flakes brought from a knapping floor onto the excavated area. However, as 100% of the flakes with complete platforms exhibited impact marks, this suggests the use of hard hammer – for the experimental dataset, 79% of the complete soft hammer flakes had impact marks compared with 96% of the hard hammer direct and bipolar flakes.

The complete flakes from the assemblage were generally small with only a handful greater than 25mm in length; the mean length was 20.5mm. Comparing the means of the techniques, the (two) complete bipolar flakes were slightly longer than the complete platform flakes, but the bipolar were narrower and half the thickness of the platform flakes (Appendix A- 104). Figure 10-16, Figure 10-17, and Figure 10-18 provides the length, width, and thickness for complete and fragment flakes, comparing stratum, topsoil, pit/possible pit, palisade, and structure.

Figure 10-16 Boxplot. Length. Thornhill flakes: stratum, topsoil, pit/possible pit/palisade, structure

Figure 10-17 Boxplot. Width. Thornhill flakes: stratum, topsoil, pit/possible pit/palisade, structure

Figure 10-18 Boxplot. Thickness. Thornhill flakes: stratum, topsoil, pit/possible pit/palisade, structure

Figure 10-19 Thornhill. Scatterplot. Bipolar and platform flakes for selected fragment groups

The complete flakes from the pits and structures were smaller, and especially thinner, than those for the topsoil, but the same pattern does not hold for the flake fragments from the pits, where the median width and thickness was greater and median length just slightly lower than the topsoil component. The stratum contained only one, small, complete flake, but their flake fragments’ median length and width was the greatest overall. Figure 10-19 shows the length/width scatterplot for the fragment groups with the larger fragments – complete, lateral, proximal missing, distal missing, and mesial left/right missing. Few flakes are >40mm in length and the larger flakes are primarily the proximal missing and distal missing fragments. Over half of the complete platform flakes have a length/width ratio of 1:1 or less with none over 2:1; both bipolar complete flakes have a length/width ratio of close to 2:1.

Compared to the Belderrig assemblage, the length/width and width/thickness scatterplots show that the complete platform flakes from Thornhill are much smaller, and relative to size, much thinner with almost all having a width/thickness ratio of 2:1 or greater (Figure 10-20 and Figure 10-21). The Belderrig mean length/weight ratio was 7:1 while Thornhill’s was 10:1; comparing the median, the difference was even greater, with Belderrig’s at 4:1 and Thornhill’s at 9:1. These ratios highlight the considerable comparative thinness of the Thornhill flakes. Consequently, the mean platform thickness was also much less for Thornhill (Figure 10-22); however, while Thornhill’s mean platform thickness was 40% less than Belderrig, their platform width was only 25% less (median platform thickness – 45% less; median platform width – 23% less). Using the log transformations, the two datasets’ complete flakes were compared for seven metrics. The differences in means between Thornhill and Belderrig were significant for all metrics tested, apart from max. width (Table 10-13).

Source Dependent Variable Type III Sum
of Squares
df Mean
Square
F p
Dataset Max Length (LOG) 0.322 1 0.322 8.195 0.005
Max Width (LOG) 0.116 1 0.116 2.470 0.119
Max Thickness (LOG) 0.736 1 0.736 12.025 0.001
Platform Width (LOG) 0.233 1 0.233 4.153 0.044
Platform Thickness (LOG) 0.737 1 0.737 11.927 0.001
Weight (LOG) 2.781 1 2.781 6.693 0.011
Length/weight ratio (LOG) 1.209 1 1.209 5.380 0.022

Table 10-13 GLM. Belderrig and Thornhill complete quartz flakes

quartz platform flakes ratio quartz platform flakes ratio

Figure 10-20 Length/width scatterplot. Complete platform flakes: Thornhill and Belderrig

Figure 10 21 Width/thickness scatterplot. Complete platform flakes: Thornhill and Belderrig

quartz context correspondence

Figure 10-22 Means. Complete platform flakes: Belderrig and Thornhill

Top of Page

10.4.5 Diagnostic types

The Thornhill assemblage contained 40 diagnostic types and three possible retouched artefacts; two of the 40 were retouched rock crystal flakes; nearly half came from the topsoil and almost a third came from the pits (Appendix A- 105). The biggest collection of diagnostic types from the same fill came from Pit 140, which had three retouched flakes and a dual, opposed core scraper along with 15 flakes, five cores and a core/debitage; apart from the topsoil this context/fill was the second largest collection of artefacts together, after Pit 2966 which contained a rock crystal retouched flake along with 2 cores, 18 flakes, and 12 <20mm debris in a single fill (Appendix A- 105). Overall, there were seven scrapers, two borers, one retouched pebble, two retouched cores, one retouched bipolar flake, 24 retouched platform flakes, and three retouched debris. One scraper was on a bipolar core while another was on a core/debitage with the remainder on platform flakes. Four of the five artefacts on flakes were convex-retouched on the distal end, with two on plunging terminations; the other was a lateral-retouched scraper (Appendix B- 26). Scrapers on plunging terminations were also noted in the flint assemblage (Nelis 2004). Table 10-14 provides the angle, delineation, and localisation for the 31 retouched artefacts (excluding scrapers and borers). The majority had abrupt retouch, with just one low angled. Half had rectilinear retouch, and 10% had concave. The largest grouping was abrupt, rectilinear retouch on lateral edges.

Angle Delineation Localisation Count Angle Delineation Localisation Count
Abrupt Concave Proximal left 2 Semi-abrupt Concave N/A 1
Convex N/A[26] 2 Convex Distal 1
Distal 2 Distal left 1
Distal left 1 Lateral left 2
Lateral left 1 Rectilinear Distal 1
Lateral right 1 Distal right 1
Rectilinear N/A 2 Lateral right 1
Distal 3 Low Convex Distal left 1
Distal left 1
Lateral left 3
Lateral right 4

Table 10-14 Thornhill. Retouched artefacts (excluding scrapers and borers). Angle, delineation, localisation

Only two of the diagnostic flakes were complete, with a number being small fragments (Figure 10-23 and Figure 10-24). These scatterplots highlight that the scrapers are generally the largest, but that two are quite small and these are small fragments rather than small artefacts per se. The diagnostic flakes’ means were compared to the non-retouched flakes of the assemblage to ascertain any differences in sizes. The analysis excluded the retouched cores, core/debitage, and debris in order to compare flakes with flakes and in order to avoid bias created by small fragments, only artefacts ≥1g were analysed; the resultant flakes for analysis were two borers, five scrapers, and 17 retouched flakes, and 285 non-retouched artefacts for analysis (Appendix A- 106). While the scrapers’ mean and median metrics were larger for weight and the three dimensions, the differences in means between the four artefact types for the four metrics were not significant (Appendix A- 107).

quartz diagnostic types ratioquartz diagnostic types ratio

Figure 10-23 Thornhill complete and fragment diagnostic types. Length/width

Figure 10-24 Thornhill complete and fragment diagnostic types. Width/thickness

Top of Page

10.4.6 Palisades, structures, and pits

This section focuses on the main contexts of the palisades, structures, and pits, with the artefacts from other contexts presented in the appendices – Appendix A- 108 for stake/posthole and linear/slot cuts; Appendix A- 109 for the strata; Appendix A- 110 for the topsoil; and Appendix A- 111 for miscellaneous finds. The majority of the quartz artefacts from the palisades came from Palisade 4 (see Plan 4 for all Area 2 contexts mentioned), and derived from 13 fills from the palisade; half of the artefacts came from the upper fill and fills that were capping fills of two pits interpreted as originating from the burning episode of Palisade 4 (Appendix A- 112). No difference was discerned between the upper and lower fills in terms of artefact types, except that the three retouched artefacts all came from the upper fill or capping fills. The palisade had a high proportion of bipolar flakes; C.320, the capping fill of Pit C.607 had six bipolar flakes along with one platform flake and two single platform cores.

Out of the five structures, quartz artefacts came from three – Structures A, B, and D. While Structure E contained the greatest quantity of non-quartz artefacts – 30 from Structure E, 14 from Structure B/C, and eight from Structure A (Nelis 2004, 332) – it produced no quartz artefacts. The upper fill of Structure D’s eastern foundation trench contained a platform flake fragment (Appendix A- 113 and Plan 5).

Structure A’s quartz came from C.311, the southern discontinuous, foundation trench (Appendix A- 113 and Plan 6). C.318, interpreted as packing fill above the basal context, contained a small complete bipolar core with a large cavity in the centre of the core, and a platform flake fragment. C.319 was one of the uppermost fills and overlay C.318; it contained a bipolar core, a small borer, two bipolar flake fragments, one platform flake fragment, and two debris. The latter platform flake was a thick fragment on different quartz than the rest of the assemblage, and one of the debris was >20mm and was possibly burnt.

Structure B contained quartz in nine contexts (Appendix A- 113 and Plan 7). C.717, interpreted as an internal porch, contained one complete bipolar flake, and an indeterminate piece with flake removals but with rounded edges. C.430 was sealed by a large flat stone interpreted as a threshold stone, and contained a complete bipolar core. C.571, a fill of a southern side foundation trench, had two platform flakes, one fragment and one complete. The other southern side foundation trench terminated in a posthole C.583 which contained one complete bipolar core formed on one of the few sugar/smooth materials from the assemblage, and four indeterminate pieces; three of these were possible cores, but with sub-rounded edges - a possible bipolar core, a possible platform core, and a possible core/flaked pebble. C.527, fill of the east wall foundation trench, contained two <10mm debris. C.545, an upper fill of a northern foundation trench contained a thick, heavy, complete multiplatform core that was burnt, along with a complete platform flake. The final stretch of the northern foundation trench contained two fills, with the lower fill, C.715, containing four bipolar flake fragments and two platform flake fragments; the two platform flakes were proximal fragments that conjoined. The fill (C.548) of a large posthole in the northeastern corner of the eastern room contained a platform flake fragment and an interior stakehole (C.379), possibly forming part of an internal division of the structure, also contained a platform flake fragment.

Structure A cut through three conjoined pits (for Area 2 pits see Appendix A- 114), which were interpreted as blocking the palisade entrance. C.325, the southerly of the three, contained several dozen fills with artefacts from four – C.2791 contained three possibly burnt platform flake fragments, C.2727 contained one bipolar flake fragment, C.2810 contained a complete bipolar core, and C.2813 contained a complete bipolar core and two possible bipolar cores. The central pit of the three had a posthole (fill C.2846), with a small platform flake fragment. Pit C.784, near to the conjoined pit group and interpreted as also blocking the palisade entrance, was cut by C.783, which was possibly related to Structure A; pit C.784 contained five fills, with artefacts in one; this had one conically-shaped complete single platform core, and one indeterminate piece that is possibly a retouched flake fragment, but had rounded edges.

A second group of ritual pits (C.606, C.6001, and C.868), also surmised to have been constructed to deliberately block the settlement entrance (Logue and Ó Baoill 2009), were located 1.5 m southwest of pits conjoined ritual pits C.325 and C.784. Of these three pits, two contained artefacts – Pit C.606 and C.868. Pit C.868, which contained a possible recut to hold a post and cut Palisade 3a, contained about 30 fills with artefacts in one – it held a dual, right angled core and two platform flakes. Pit C606, which cut Palisade 4, contained 11 fills with a bipolar flake and a platform flake in one fill.

Five large pits ran in a line immediately west of Palisade 4, of which four contained artefacts. Pit C.425 had seven fills with artefacts in three – fill C.477 had a platform flake fragment and an indeterminate piece (a possible bipolar split cobble and also possibly burnt); fill C.478 contained a retouched platform flake fragment and a <20≥10mm sliver; fill C.479 contained two platform flake fragments. Pit C.855 had 10 fills with two artefacts from unlisted fill(s) – a complete platform flake and a <20≥10mm debris which was possibly burnt. Pit C.424 cut Pit C.425 and had five fills plus an upper fill (C.1461) which may represent the burning episode of Palisade 4 (Ó Baoill 2009 pers. comm.) and contained a quartz hammerstone; three other fills contained quartz – C.492 contained a platform flake fragment, a core/debitage, and a quartz hammerstone; C.493 contained a platform flake fragment and two <20≥10mm debris; C.494 contained a complete multiplatform core. Pit C.423 cut Pit C.855 and had three main fills with artefacts in all – fill C.511 contained two platform flake fragments; C.512 contained a single platform core fragment and a complete multiplatform core; C.513 contained a possible flake. Pit C.423 had three deposits above, possibly related to the burning of Palisade 4, one of which (C.509) contained a retouched pebble, a complete single platform core, and a platform flake fragment.

Between the two sets of conjoined pits, Pit C.607, which was cut by the construction of Palisade 4, contained artefacts in four of the eight contexts. There were 1 bipolar flake and 15 platform flakes, one of which was possibly burnt and another was a scraper.

Pit C.2966, which is not in the stratigraphic report or site plan but is recorded by Logue (2003), contained the most artefacts of the pits; it contained 33 artefacts, including a bipolar and multiplatform core, a bipolar flake, 17 platform flakes, a retouched rock crystal platform flake, with the remainder being <20 mm debris. Adjacent to Structure B was a pair of pits, one of which (C.360) contained a single fill with a small platform flake fragment with a cavity on its ventral face. South of the structure was another pit (C.529) which contained a complete platform flake scraper, two small retouched flake fragments, a platform flake fragment, and two <10mm debris – no fill was recorded for these artefacts.  A possible slot cut adjacent to Structure B, which was recorded on plan as 4.4m in length, contained in the upper fill (C.582) three complete bipolar cores, a complete single platform core, and platform flake fragment and a >20mm debris.

The pits in Area 1 (Appendix A- 115) are interpreted as situated in an area set aside for ritual activity (Logue and Ó Baoill 2009). Of the 35 pits in Area 1, artefacts were uncovered from 10; eight were single fill pits while two contained three fills each with artefacts from one fill each. The pits had a greater proportion of platform cores to bipolar cores than any contexts from Area 2. While artefacts came from 10 pits, half of the artefacts came from one of the pits – Pit C.140 contained six cores, one of which was retouched as a scraper, and 18 flakes, three of which were retouched; one of the flake fragments was of a different quartz that was possibly burnt. As well as the 10 pits with artefacts, there were six pits with only natural quartz present and nine of the 10 pits with artefacts also contained natural quartz; while these 10 artefact-bearing pits had a greater ratio of quartz artefacts to natural quartz pieces compared to the pits in Area 2, Area 2 had more pits with only artefacts and no natural quartz and Area 1 had more pits with only natural quartz (Figure 10-25).

The Area 2 pit group C.423, 424, 425, and 855 lay outside of the palisades, and apart from Pit C.855, these pits contained a large quantity of natural quartz, especially small, complete beach pebbles (pebbles defined as GD 4-64mm). Figure 10-26 presents the proportions of natural quartz in three pits, sub-dividing the quartz into beach pebbles and lumps derived from the glacial till; however, it is possible that some of the pebbles also came from the glacial till. While the excavators noted that these pits may represent storage pits (Logue and Ó Baoill 2009), these pebbles and lumps were all small, suggesting that the pits do not represent a caching of lithic material.

Figure 10-25 Thornhill pits: artefact/natural ratios. Rows 1 and 2: Area 1 pits. Pit C.325, Posthole C.2839, and Pit C.784 = conjoined pits.
Pit C.423, Pit C.424, Pit C.425, and Pit C.855 = pit group. Pit C.606 and Pit C.868 = conjoined pit

Figure 10-26 Thornhill Pits C.423, C.424, C.425. Pebbles = derived from beach. Lump and fragment = pieces presumably derived from
glacial till with lumps deemed as 'complete' and fragments being fragmented lumps

Pit C.423’s (no section diagram available) base was interpreted as having been timber-lined and the basal layer (C.513) contained one possible flake fragment on a poor quality quartz along with eight complete pebbles, one split pebble, five lumps, and two fragments; the next layer (C.512) contained two cores along with six pebbles, one split pebble, nine lumps, and two fragments; the next layer (C.511) contained two flakes fragments along with 40 complete pebbles, six split pebbles, eight lumps, and five fragments. Pit C.424 also contained numerous pebbles in the basal layer and a core in the next layer (Figure 10-27). C.425 also had no artefacts in the basal layer, with the artefacts appearing in the upper layers, appearing with numerous pebbles (Figure 10-28).

While many of the quartz pieces in these and other pits could be interpreted as being deposited through the backfilling of soil which contained quartz, the high proportion in many of the deposits, especially of small beach pebbles, suggests that these represent purposive deposition of ‘natural’ quartz and not a random deposit of pebbles and lumps or caching of lithic raw material; out of the 21 pits which contained 343 pebbles, there were just two beach cobbles (defined as GD 64-256mm), and almost all the pebbles and lumps were GD <20mm, with pits from Area 1 containing larger natural pieces than the Area 2 pits. While small flint pebbles were noted on the Foyle shoreline during the present survey (see above, Section 10.3.1), this material was not deposited in pits in the same manner as quartz suggesting a difference in the treatment of these materials.

While many of the quartz pieces in these and other pits could be interpreted as being deposited through the backfilling of soil which contained quartz, the high proportion in many of the deposits, especially of small beach pebbles, suggests that these represent purposive deposition of ‘natural’ quartz and not a random deposit of pebbles and lumps or caching of lithic raw material; out of the 21 pits which contained 343 pebbles, there were just two beach cobbles (defined as GD 64-256mm), and almost all the pebbles and lumps were GD <20mm, with pits from Area 1 containing larger natural pieces than the Area 2 pits. While small flint pebbles were noted on the Foyle shoreline during the present survey (see above, Section 10.3.1), this material was not deposited in pits in the same manner as quartz suggesting a difference in the treatment of these materials.

quartz pit deposits

Figure 10-27 Thornhill Pit C.424 contexts. Adapted from Logue and Ó Baoill (2009)

quartz pit deposits

Figure 10-28 Thornhill Pit C.425 contexts. Adapted from Logue and Ó Baoill (2009)

Top of Page

10.4.7 Comparison to non-quartz assemblage

Unlike the Belderrig assemblage, the Thornhill assemblage contained a substantial non-quartz component. Nelis (2004) analysed the Thornhill non-quartz flaked stone assemblage as part of a Ph.D. on the Northern Irish Neolithic, and the following comparison is based on her thesis. Unfortunately, the non-quartz analysis was conducted before the stratigraphic report was completed; therefore, it is not possible to compare the spatial distribution of the assemblage, as only a limited amount of context descriptions were available to her. Nelis (2004, 79-89) provides a glossary of terms used in her thesis, but some terms in the glossary remain undefined. Unusually, Nelis (2004, 82) describes ‘debitage’ as including cores which are categorised as separate from debitage in this thesis, and by most lithic analysts; she comments that the term debitage “describes all material resulting from primary reduction which is not subsequently modified or utilised as a tool”. For what is termed ‘debris’ in this thesis, Nelis (2004, 88) appears to use the term ‘angular shatter’, saying that ‘angular shatter’ is “chips and spalls” but does not define what chips and spalls are; shatter also includes ‘flake shatter’ (following Andrefsky (1998)) which are defined as flake fragments in the present analysis. While Nelis (2004, 85) comments that ‘microdebitage’ is defined by size, she does not state the size definition she uses. Not so unusually, but incorrectly (see Section 4.5), Nelis (2004, 89) describes a ‘stone tool’ as “an artefact that has been secondarily modified by retouch or use…debitage is not considered to be tools, but would be considered artefacts”. In a number of instances, Nelis (2004) provides contradictory quantities of artefacts; in order for clarity here, the differing quantities have been cited. Therefore, with these comments in mind, the non-quartz component will be compared the quartz.

One chert/basalt and 616 flint flaked stone artefacts were analysed. Cores accounted for 1.8% (n=11) of the assemblage and core tools for 0.6% (n=4), giving a total of 15 cores (Nelis 2004, 296); however, Nelis (2004, 301) then states that there were 14 cores in total. Complete flakes and blades accounted for 21% (n=135) of the assemblage, “flake/blade shatter” for 24% (n=148), “angular shatter” for 10.4% (n=64), and “modified” for 35% (n=216) (Nelis 2004, 296). With limited spatial information, Nelis (2004, 331) noted that 40% of the non-quartz assemblage was from the “topsoil/unlocated”; while Structure E contained 30 non-quartz artefacts – the largest quantity from the structures – no quartz artefacts were noted; Structure A had six non-quartz modified pieces and two flake/blades. Overall, 19% of the assemblage was burnt/heated, with 15% described as heavily burnt; 13% of the “modified assemblage” was also burnt (Nelis 2004, 298-9). This proportion is at odds with the quartz component, which had 2.3% burnt or possibly burnt artefacts – this significant difference cannot be put down entirely to the difficulties in identifying burnt quartz, and may result from a different treatment of materials, or different contexts of burnt material.

The modified proportion of the flint assemblage is very high, with few cores; just over half of the modified assemblage consisted of scrapers, knives, and projectiles. As with the quantities of cores, Nelis (2004, 311-30) provides different quantities and/or proportions for the modified artefacts, with Table 10-15 providing the variations noted.

Type Count % Alternative
count
Alternative
%
Scraper 46 20.9 45 Not stated
Projectiles 41 18.6 42 19.1
Edge retouched pieces 63 28.6 N/A 28.3
Knives 26 11.8 N/A Not stated
Miscellaneous tools 23 10.45 N/A Not stated
Unretouched but utilised[27] pieces
Flakes and blades 14 6.4 N/A N/A
Cores 4 1.8 N/A Not stated
Flaked flint axe fragments 3 1.4 N/A Not stated

Table 10-15 Thornhill modified non-quartz artefacts. Count and alternative count from Nelis (2004, 311-30)

Nelis (2004, 297) noted “a clear dichotomy between the knapping debitage and the modified tools…in terms of quality and apparent source of material;…the core assemblage infers…a primary industry largely reliant on very small…beach pebbles…[while] many of the tools, and much of the small scale  debitage, are of a very translucent, highly vitreous quality”.

While Nelis did not remark on a possible Later Mesolithic flint component at Thornhill in the main section covering the assemblage, in a later discussion she suggested that
"Thornhill includes more utilised flakes and blades which bear resemblance, at least superficially, to the butt trimmed forms of the previous [Mesolithic] industry, and although they may not compare with the heaviest of types found at that time, they are larger than many of the other tools within the assemblage. This may indicate contact with those familiar with these previous industries; the fact that they seem to have served as expedient cutting [sic] may hint at a way in which the two industries may have come together and complimented one another" (Nelis 2004, 340, emphasis added).

However, as noted by Woodman (Woodman 1992), large flakes should not automatically be designated as Mesolithic because Neolithic flakes can be the same size if not bigger that Mesolithic flakes. Nelis then comments on the Enagh Neolithic structure (across the river, see Section 10.2.2) where a posthole
"yielded a single butt trimmed blade…Again, this piece echoes an industry more commonly associated with the Mesolithic, and its appearance at Enagh Lough, with similar tools being found at Thornhill, may indicate a continuance of this tool type into the Early Neolithic, either as a result of contact with those more familiar with such tools, or as a residual element of the earlier industry” (Nelis 2004, 341).

It is not apparent, however, that the artefacts from Thornhill are specifically butt trimmed forms, as no mention was made of butt trimmed forms in the main analysis and these artefacts were categorised as ‘utilised’ by Nelis – with ‘utilised’ defined by her as non-retouched artefacts (Table 10-15) – and initially described as superficially resembling butt trimmed forms, therefore suggesting that they lacked retouch; butt trimmed forms are classified as such if they are retouched on the proximal end. This inconsistent, if not misleading, account of the artefacts can lead to misunderstandings of the material culture – for example, Smyth (2006a, 169) cited Nelis’ analysis by arguing that “networks may have been open to the occupants of Thornhill and Enagh, where the recovery of bulky, butt trimmed tools…suggests that there was direct and indirect access to larger sized raw material.” Smyth’s account has therefore taken the artefacts to be diagnostic butt trimmed tools, rather than superficially resembling butt trimmed forms.

Figure 10-29 compares the proportions of artefact types for quartz and non-quartz – while a similar proportion of both components were unmodified flakes, the quartz component had substantially more debris and cores, and the non-quartz had substantially more modified artefacts. Figure 10-30 re-categorises the non-quartz modified component as either flakes or cores (whichever is appropriate) and compares the proportions of the total flaked stone assemblage.

Overall, 69% of the assemblage were flakes, with slightly more from the non-quartz component but the quartz component had significantly more flake fragments (5% complete quartz flakes and 48% complete non-modified, non-quartz flakes); the quartz cores account for 10% of the overall assemblage, with non-quartz cores accounting for just 1%. The significantly different proportion of complete flakes in the quartz and non-quartz component makes a graph such as in Figure 10-30 difficult to interpret. The comparison of the fragmentation rate of the chert and quartz materials produced during the experimental knapping highlighted that quartz produced 4.5 times more fragments than the chert thus skewing direct comparisons.

As noted, there were either 14 or 15 cores – one was a single platform core, with the remainder being bipolar or scalar, “all of which have been exhaustively worked” (Nelis 2004, 302). Nelis (2004, 88) describes scalar cores as “non-platform based percussion techniques, similar to bipolar reduction, but with no visible evidence for the use of the anvil technique”. There were 135 complete flakes and blades and 148 fragments; five were bipolar, 25 were pressure flakes, with the remainder platform flakes. While Nelis (2004, 302) states that the “single platform core indicates that platform technology was known on site, but…bipolar reduction was preferred”, the dominance of platform flakes at the site, and the dominance of a direct percussion technique in the Neolithic in general, makes this statement questionable. 75% of the complete flakes and blades were less than 32mm in length, with an average length of 23mm, which is greater than the average complete quartz flake length of 20.5mm.

quartz non quartz flaked tools thornhill quartz non quartz flaked tools thornhill

Figure 10-29 Thornhill quartz and non-quartz flaked artefacts – % per material. Non-quartz data from Nelis (2004), excluding the unworked material cited as artefacts and considers her ‘angular shatter – chips and spalls’ as debris. Quartz component includes rock crystal and excludes nine possible artefacts, one core/debitage and one retouched core/debitage

Figure 10-30 Thornhill quartz and non-quartz flaked artefacts – % per total assemblage. Non-quartz data from Nelis (2004), excluding the unworked material cited as artefacts and considers her ‘angular shatter – chips and spalls’ as debris; modified artefacts included as either flake or core. Quartz component includes rock crystal and excludes nine possible artefacts, and two core/debitage

Top of Page

10.4.8 Chaîne opératoires

The chaîne opératoires of the quartz component at Thornhill consisted of the collection by the communities of practice of beach cobbles with half the cores identified as based on cobbles; it is unclear to what extent veins were quarried, and it is possible that blocks were also available in the glacial till, along with some of the cobbles; the cobbles, however, did not show evidence of striations that would indicate glacially-derived cobbles. The communities of practice had a clear preference for smooth-grained quartz, with very few sugar or sugar/smooth-grained quartz present, with the pits over-represented with these quartz types. This is similar to the Belderrig assemblage’s lack of sugar-textured quartz (Section 9.5); the experimental knapping showed that the sugar-textured quartz generally produced less fragments per strike than the quartz with a smooth-grained appearance (Section 6.4.1.1). A fifth of the quartz was a distinctive milky-white quartz which had a plastic-like appearance and feel.

The high proportion of >49% cortex bipolar and platform cores suggests that the cores were not decortified by the communities of practice before being brought to the site for use or deposition, while the low proportion of cortical flakes suggests that the debitage assemblage does not represent a knapping floor assemblage to a great degree, and that the ‘missing’ cortical flakes are either elsewhere in the unexcavated areas of the site, or at a further remove from the site altogether. Therefore, there is a clear mismatch in the cortex proportions between cores and debitage indicating a complex pattern in the chaîne opératoires of core and debitage use and deposition throughout the site’s contexts. No clear knapping floors were identified in the excavation – as well as the lack of cortical flakes, just a small proportion of the assemblage consisted of <10mm debris – with the quartz artefacts distributed over 50 individual contexts, 100 fills, and 28 strata.

The chaîne opératoires of the communities of practice involved a multitude of techniques and combination of techniques including bipolar, direct percussion freehand platform and platform-on-anvil, indirect percussion, bipolar-on-platform, and platform-on-bipolar. Half of the cores were bipolar and a third were platform cores; this suggests that while the use of a bipolar technique was frequent in the chaîne opératoire it was not necessarily prevalent, and the dominance of platform flakes backs up this interpretation – 14% of the flakes were bipolar flakes. For the direct percussion component, single platform cores dominated, and the use of dual, right angled cores is a distinctive difference to the Belderrig chaîne opératoires. A number of the cores showed evidence of the reuse of large flakes as cores. As with the Belderrig assemblage, it is unclear whether this use of core-on-flakes by the communities of practice at Thornhill was fortuitous or a purposive strategy, as further evidence may be masked by the reduction of the cores. While many of the platform cores were knapped to a small size, the bipolar cores were rarely knapped to exhaustion by the communities, with the majority left quite large. This lack of extensive reduction of the bipolar cores by the communities of practice may partly explain the apparent low amount of bipolar flakes noted in the assemblage.

Unlike the Belderrig communities, the communities of practice at Thornhill used a number of platform preparation procedures, with 8% of the flake platforms prepared, 3% abraded, and one platform was facetted. The knapping appears to have been geared towards small, thin flakes, with almost no blades noted in the assemblage; comparisons to the experimental dataset in terms of visible impact marks on the complete platforms suggests the use of hard hammer impactors. Few of the flakes were complete, especially for the bipolar flakes, and the pits and structures contained above average proportions of complete flakes. The implications for the chaîne opératoires are uncertain with three possibilities raised – the higher rate of fragmentation in other contexts did not originate from knapping fractures but from use/post-depositional factors; the higher proportion of complete flakes in the pits and structures represents the purposeful deposition of complete flakes; or a combination of the these two factors.

While the knapping appears to have been geared towards the production of small, thin flakes, compared to the Belderrig assemblage a sizable proportion of thicker flakes with plunging terminations were noted, and a number of these were retouched on the thick, plunging termination. This suggests that the chaîne opératoire included the purposive knapping of plunging flakes for subsequent use as scrapers, which was also noted for the non-quartz assemblage (Nelis 2004). Along with these retouched plunging flakes, were a series of retouched flakes with the most common type having abrupt, rectilinear retouch on lateral edges. Unlike the Belderrig assemblage, however, the diagnostic types were not significantly larger than the non-retouched flakes. A number of the cores were also retouched, but without use wear analysis it is unclear to what extent other cores, as well as non-retouched debitage, were subsequently used for various tasks.

The deposition of the quartz artefacts by the communities of practice was spread amongst a variety of contexts, particularly pits, palisades, and structures, beyond the artefacts from the topsoil and the smaller number from strata. As noted, the pits and structures had a relatively high proportion of complete flakes deposited in them and the complete flakes from these contexts were smaller, and especially thinner, that the complete flakes from the topsoil. While the partially excavated Structure E contained the largest quantity of non-quartz lithic artefacts, it contained no quartz artefacts, with no quartz artefacts from Structure C either. The cores and debitage from Structure B were deposited, often singly, in the foundation trenches and a posthole terminating a foundation trench, underneath the threshold stone, and in interior postholes, with a particular emphasis on bipolar cores and flakes; the one platform core from Structure B was a thick, heavy, burnt multiplatform core. The 10 quartz artefacts from Structure A came from two fills of one foundation trench, with four of the 10 being bipolar cores or debitage; one of the platform flakes was a thick flake fragment of a different quartz than the rest of the assemblage. While the partially excavated Structure D contained no non-quartz artefacts, it contained one quartz platform flake fragment in its foundation trench. Smyth (2006a), in reviewing evidence for Irish Neolithic structures, has argued that such deposition of objects in foundation trenches need not all be interpreted as accidental occurrences, but rather can be seen as purposeful acts of digging and filling of the structures’ foundations. The deposition of lithics in the Thornhill structures point to differences in the chaîne opératoires between structures and materials – Structure E contained many flint but no quartz, while Structure A’s quartz deposition was limited to the one foundation trench only; Structure B’s deposition of quartz was more varied that Structure A, and no quartz was noted from Structure C whose plan intermingled with Structure B (Plan 3). As well, compared to the other contexts, while the structures had had relatively few artefacts they had the greatest ratio of artefacts to indeterminate/natural pieces of quartz.

18% of the quartz artefacts were uncovered from pits. Pits are strongly associated with Neolithic activity in a variety of settings, including structures (for discussions of pits and structures in the Irish Neolithic, see Smyth 2006a, 175-79, passim). As with the deposition of material in structures, Smyth has commented that these are not necessarily all refuse dumps and so forth, and moreover, even pits which may appear to be “artefact-poor” should not be dismissed as meaningless: “viewing these pits as meaningless, or at the very least peripheral to the activities being carried out in and around the house, goes against what we know about practices of digging and filling in the Neolithic” (Smyth 2006a, 67). At the Thornhill site, the cluster of 35 pits in Area 1 was interpreted as an area “set aside for ritual activity” because of the “large number of pits and the apparent deliberate placing of artefacts, and sometimes several different types of artefacts, within many of the fills” (Logue and Ó Baoill 2009). 10 of these contained quartz artefacts, with most being single fill pits. Half of the Area 1 artefacts came from Pit C.140 that contained six cores, one of which was retouched as a scraper, and 18 flakes, three of which were retouched. These 10 pits had a greater ratio of quartz artefacts to natural quartz pieces compared to the pits in Area 2, and a greater proportion of platform cores to bipolar cores than any contexts from Area 2. However, Area 2 had more pits with only quartz artefacts and Area 1 had more pits with only natural quartz.

Top of Page

10.5 Overview and Discussion

The excavations at the Thornhill site uncovered a series of palisades enclosing five structures along with numerous pits, stakeholes, and linear features, with the site, as yet undated, interpreted as Early Neolithic on the basis of ceramics. The 100% quartz retention policy adopted during the excavation resulted in the collection of 4047 pieces of quartz, of which 813 were artefacts, nine were possible artefacts, and 74 were indeterminate.

The Thornhill quartz chaîne opératoires were more beach cobble-based than the Belderrig assemblage, with 53% of the cores identified as beach cobbles; allowing for some beach cobbles to have been decortified thus reducing their recognition, it is possible that the cores consisted of 60-80% beach cobbles. Just 13% of the debitage were cortical artefacts, which is much lower than the Experiment dataset, suggesting that the cortical flakes are either elsewhere in the unexcavated areas of the site, or at a further remove from the site altogether. The very high proportion of >49% cortical cores compared to the Belderrig and Experiment assemblages, and the large size of many of the cores (especially the bipolar cores), highlights that the communities of practice did not systematically knap the material to exhaustion, which was noted by Nelis for the flint bipolar cores. The high degree of >49% cortical cores and the lack of cortical debitage also points to a clear mismatch between cores and debitage, indicating a complex pattern of core and debitage use and/or deposition throughout the site’s contexts.

While much of the quartz assemblage is beach cobble-based, it is unclear if this is comprised completely of collecting locally available quartz or whether other non-local beach cobbles were brought to the site by the communities of practice. However, only a few artefacts appeared to be of distinctly different quartz. Nevertheless, the fact that much of the beach cobble component of the assemblage appeared somewhat similar does not necessarily imply that it was all sourced locally. Though it is unclear what the other sources of quartz besides the beach cobbles were. A large, natural sub-rounded block was found in a posthole, highlighting that large blocks were available in the glacial till, but it is uncertain whether this material was chosen or whether quartz veins, available today 200m to the east on the cliff  forming the bank to the Foyle Estuary, were quarried.

The greater presence of bulbs and compression rings on the Thornhill platform flakes highlights the different type of quartz compared to the Belderrig quartz – 6% of the flakes had bulbs compared to 2% in the Belderrig assemblage and 5% of the experimental flakes; 3% of the Thornhill flakes had compression rings, while the Belderrig assemblage had none visible and the experimental assemblage had 0.3%. However, through macroscopic examination, the quartz is identified as of a similar grain size to the Belderrig quartz examined, and none appears to be fine-grained quartz.

The chaîne opératoires of the communities of practice involved a multitude of techniques and combination of techniques including bipolar, direct percussion freehand platform, direct percussion platform-on-anvil, indirect percussion, bipolar-on-platform, and platform-on-bipolar. The range of techniques used was, therefore, greater than used by the Belderrig communities. A number of the cores showed evidence of the reuse of large flakes as cores. As with the Belderrig assemblage, it is unclear whether this use of core-on-flakes by the communities of practice at Thornhill was fortuitous or a purposive strategy, as further evidence may be masked by the reduction of the cores.

Half of the cores were bipolar cores. A notable absence for the Thornhill assemblage were any conical pieces, which were produced during the experimental bipolar knapping, and noted by Knutsson as deriving from both platform and bipolar knapping (see Section 6.4.2). It is unclear if the lack of conical pieces in the Thornhill assemblage relates to differing materials, techniques, or depositional practices. While the use of a bipolar technique was frequent in the chaîne opératoire it was not necessarily prevalent, and the dominance of platform flakes backs up this interpretation – 14% of the flakes were bipolar flakes (for possible under-recognition of bipolar flakes, see below). The higher degree of beach cobble usage at Thornhill compared with Belderrig is not necessarily related to the frequent use of a bipolar technique: 85% of both bipolar and platform cortical cores from Thornhill were identified as beach cobbles. Furthermore, the communities of practice at Thornhill do not appear to have used the bipolar technique to reduce small platform cores further, as is noted in other lithic traditions (e.g. Callahan 1987; Knutsson 1988a; Ballin 2008). Indeed, at Thornhill many of the complete bipolar cores are larger than the complete platform cores, indicating that the latter were frequently knapped to a smaller size. Neither is this difference related to the quality of the quartz, as similar quartz was knapped with both techniques.

A third of the cores were platform cores, with single platform cores dominating, and the dual, right angled cores a distinctive difference to the Belderrig chaîne opératoires. 20% of the platform cores had evidence of a platform-on-anvil technique. The Thornhill platform cores contained a much greater size range and substantially smaller cores compared to the Belderrig assemblage, with the Belderrig cores producing a tighter cluster of cores in the middle and upper range of length/width/thickness. Overall, the Belderrig platform cores’ means for all metrics were significantly greater, with smaller standard deviations, and a much lower length/weight ratio which is the result of much thicker cores.

Debitage accounted for 82% (n= 672) of the vein quartz artefacts with a further six rock crystal debitage. Flakes were 68% of the debitage, a higher proportion than the Belderrig or Experiment datasets; the topsoil contained 59% flakes, with most of the other contexts having 80% or over. There was only a small proportion of <10mm debris, highlighting the difference in contexts between the knapping floors of the experimental and Belderrig assemblages and the varied contexts of the Thornhill, none of which appear to be knapping floors or the deposition of small knapping floor artefacts. This stands in contrast to the high proportion of cores in the Thornhill assemblage compared to the other datasets. A similar proportion of <10mm debris came from the topsoil, the structures, and the pits, with less from the palisades, and none from any other contexts. The 100% retention policy suggests that the lack of <10mm pieces is not related to the collection strategy of the excavation.

54% of the assemblage’s ≥10mm debitage had complete or partial platforms, which is a much greater proportion than the Belderrig or Experiment datasets (which had 33-35%), but with a similar proportion of complete flakes to Belderrig. This high proportion of flakes with complete or partial platforms suggests that while the proportion of bipolar flakes (14% of the total flakes) may be less than it should be, the low proportion of bipolar flakes is an actual pattern. The lack of an extensive reduction of the bipolar cores by the communities of practice noted above, and the lack of cortical flakes in the assemblage, may partly explain the apparent low amount of bipolar flakes. Overall, 5% of the flakes were complete with 3% (n=2) of the bipolar flakes complete – for the Experiment dataset, a much higher proportion of bipolar flakes were complete (37%) compared to the platform flakes (8%). While the pits contained 21% of the flakes, they contained 48% of the complete flakes, and the structures contained 4% of the flakes but contained 13% of the complete flakes; many contexts had no complete flakes. This discrepancy in the distribution of complete flakes suggests that either a part of the higher rate of fragmentation of the flakes in the palisades, strata, and topsoil does not originate from fracturing during knapping, but rather from use/post-depositional factors; or that the greater proportion of complete flakes in the pits and structures originates from the purposeful deposition of complete flakes in these contexts (see below); or a combination of the previous two possibilities.

Looking at the platform flakes, almost all the complete platforms were flat, with just one faceted platform. There was a high degree of platform collapse, with may relate to the relative thinness of the flakes. Unlike the Belderrig assemblage, the chaîne opératoire of the Thornhill communities of practice included platform preparation – 8% of the platforms were prepared and 3% were abraded. Nelis (2004, 865) has suggested that Early Neolithic assemblages analysed by her generally “lack extensive edge preparation or platform faceting”. Therefore, having 11% of the platform flakes exhibiting platform preparation is not necessarily a low proportion compared to non-quartz assemblages, and highlights a significant difference with the Belderrig chaîne opératoires.

Compared to the Belderrig assemblage, the complete platform flakes from Thornhill are significantly smaller, and relative to size, much thinner. The Belderrig’s complete flake median length/weight ratio was 4:1 while Thornhill’s was 9:1. These ratios highlight the considerable comparative thinness of the Thornhill flakes. Consequently, the mean platform thickness was also much less for Thornhill; however, while Thornhill’s median platform thickness was 45% less than Belderrig, their median platform width was only 23% less.

While the Thornhill non-quartz assemblage contained a high proportion of diagnostic types, the quartz component contained a much smaller proportion – 40 diagnostic types and three possible retouched artefacts, with two of the 40 being retouched rock crystal flakes. Unlike the Belderrig diagnostic types, the Thornhill diagnostic types’ means for the various metrics were not significantly different than the non-retouched artefacts. Overall, there were seven scrapers, two borers, one retouched pebble, two retouched cores, one retouched bipolar flake, 24 retouched platform flakes, and three retouched debris. One scraper was on a bipolar core while another was on a core/debitage with the remainder on platform flakes. Four of the five on flakes were convex-retouched on the distal end, with two on plunging terminations; the other was a lateral-retouched scraper. Scrapers on plunging terminations were noted in the flint assemblage as well, and the higher proportion of flakes with plunging terminations compared to the Belderrig assemblage suggests that the chaîne opératoire included the purposive knapping of plunging flakes for subsequent use as scrapers.

The quartz artefacts came from over 50 individual contexts and over 100 fills, as well as 28 strata. 60% of the artefacts came from the topsoil and 18% came from the pits, with the remaining contexts, primarily the palisades and structures, having 8% or less each. Area 1 contained 35 pits and has been interpreted by the excavators as an area set aside for ritual activity, while many numerous pits in Area 2 have also been interpreted as signifying ritual activity, including the blocking of the entrance to the site. The contexts (apart from the topsoil) with the largest collection of artefacts were two pits – Pit C.2966 and Pit C.140. Pit C.2966 in Area 2 contained the largest collection, with a rock crystal retouched flake along with 2 cores, 18 flakes, and 12 <20mm debris in a single fill. Pit C.140 in Area 1 contained the second largest collection of artefacts and largest collection of diagnostic types; it held three retouched flakes and a dual, opposed core scraper along with 15 flakes, five cores and a core/debitage in a single fill – half of the artefacts from Area 1’s 10 pits came from this one pit. The pits were over-represented in the proportion of diagnostic types (over a third of the diagnostic types and less than a fifth of the overall count) and also in the proportion of sugar/smooth- or sugar-grained artefacts; the vast majority of artefacts were smooth-grained, with only 16 sugar/smooth- or sugar-grained artefacts. The pits contained half of them but less than a fifth of the overall assemblage; the topsoil contained 25% of them, but had 60% of the overall assemblage. While numerous pits have been interpreted by the excavators as being ritual in function, as noted previously, Smyth (2006a, 67) has argued that even seemingly artefact-poor pits should not be dismissed as meaningless or peripheral to the main activity at a given site. Therefore, while some pits can be seen as explicitly ritual in nature, others can be seen as indicative of ritualised activity.

Moreover, the deposits of quartz in other contexts such as the structures can also be seen as ritualised activity in the communities of practice’s chaîne opératoires. Overall, the structures contained few artefacts but held a greater ratio of artefact to natural quartz than other contexts, suggesting that the appearance of quartz artefacts, and possibly the natural quartz, in them was not random. The quartz deposition was not, however, the same for all the structures – while Structure E contained the largest quantity of non-quartz flaked artefacts, it contained no quartz, and no natural quartz pieces; Structure D contained no non-quartz and contained one quartz flake fragment along with one natural quartz piece. While Structure B and Structure C’s plans intermixed, with Structure B possibly the later structure, no quartz was retrieved from Structure C and Structure B had quartz deposition, often of single artefacts, in the foundation trenches as well as under the threshold stone and internal features.

Structure B’s (and Structure A’s) artefacts had a greater emphasis of bipolar cores and flakes compared to the pits. However, the deposition of the cores in the structures does not appear to represent knapping taking place, but rather the deposition of cores in the various features such as the foundation trenches. The one platform core from the Structure B was a thick, heavy, complete multiplatform core that was burnt, found along with a complete platform flake in the northern foundation trench. One of the few sugar/smooth-grained materials from the assemblage was a complete bipolar core from a posthole which terminated the southern side foundation trench of Structure B, found with four indeterminate pieces. These four indeterminate pieces had rounded or sub-rounded edges but, as was speculated above, three of them – which all resembled cores – may have been either collected/deposited as ‘antique’ artefacts or as items that looked like artefacts. A fourth possible ‘antique’ artefact was also possibly a core, and came from an internal porch, deposited with a bipolar flake. A further two indeterminate pieces were interpreted as possible ‘antique’ artefacts – Pit C.138 in Area 1 and Pit C.784 (part of the conjoined ritual pit group blocking a palisade entrance and later cut by Structure A) both contained an indeterminate piece each, that were possible retouched artefacts.

As noted above, while the proportion of complete flakes was low and similar to the Belderrig assemblage, the pits and structures held an above average proportion of them. In discussing the deliberate fragmentation of artefacts in the southeastern European Neolithic Chapman (2000) suggested that the both complete and fragmented objects can be interpreted as an enchainment, linking both people together and linking people to objects, viewing this in terms of dividual and individual personhood (see also Fowler 2004; Jones 2005). For example, commenting on whole ceramic pots, Chapman (2000, 43) suggested that these may have represented “integration, wholeness, group solidarity and its successful constitution”. Smyth (2006a, 228-9) has argued that “[p]erhaps the most striking feature of Neolithic pits…is the frequency with which these contexts are associated with broken or fragmented material, in particular the pits from the early and later Neolithic”. At Thornhill, there appears to be the purposive deposition by the communities of practice of complete quartz flakes in the pits and structures, of a material that has a much higher fragmentation rate than other knapped stone.

As well as this deposition of complete flakes, many of the pits had depositions of complete pebbles and lumps of natural quartz along with lesser quantities of split pebbles and fragments of lumps. These do not appear to represent a caching of lithic raw material, and some pits contain only natural pieces in basal layers, with upper layers containing a mixture of natural and artefactual material. This suggests a structuring of deposits in some pits. The presence of an abundance of ‘natural’ quartz in various pits either with ‘artefacts’ or alone raises the question of what is deemed as natural. For the present analysis, which is examining the use of quartz in the lithic technologies, it is easy to define natural as material that does not show signs of use or modification and artefacts as those that do. Clearly, the pits point that such a designation breaks down in the face of prehistoric practices. Are these artefacts, manuports, or ecofacts (for definitions see Kipfer 2000)? While these quartz pieces are not artefacts in terms of a lithic technology, there is a clear manipulation of materials that goes beyond the use of quartz for ‘stone tools’, a manipulation not observed with ‘natural’ flint. These ‘natural’ quartz pieces are therefore part of the material culture, and cultural repertoire, of the communities of practice at Thornhill (see also Cooney 2008 for Neolithic engagements with stone and deposition of 'non-artefact' stones in pits).

While this deposition of ‘natural’ quartz points to a different treatment of the material compared to flint, the use of flakes with plunging terminations for convex scrapers parallels the use of flint and highlights that some of the chaîne opératoires of the materials was similar. While scrapers accounted for about 21% of the diagnostic flint artefacts and about 7% of the overall flint assemblage, they accounted for 18% of the diagnostic vein quartz types but just 0.8% of the overall vein quartz assemblage, highlighting the significant difference with the amount of diagnostic flint types. Around 19% of the flint diagnostic types were projectiles and 12% were knives, but none were noted in the quartz assemblage. This highlights the limitations of the vein quartz – which was not suited to pressure flaking – for the tool repertoires of the communities of practice at Thornhill. As noted in Chapter 3, just one quartz projectile has been identified in the Irish archaeological record – a rock crystal arrowhead from the Ballyglass no. 13 Neolithic house (Warren Forthcoming). The minor proportion of diagnostic types in the Thornhill quartz assemblage is comparable to all the other Neolithic assemblages noted in Chapter 3. With a lack ‘types’ in quartz assemblages, these are not suitable for dating assemblages from a typological perspective; quartz scrapers, however, are sometimes noted, and the Thornhill convex end scrapers are typologically and technologically similar to the non-quartz examples. Otherwise, the retouched quartz component consists primarily of retouched flakes, which are difficult to fit into convenient typological boxes. Technologically, the quartz assemblage appears broadly similar to other Neolithic house assemblages outlined in Chapter 3. However, in light of the analytical difficulties that quartz assemblages present to lithic analysts discussed in Chapter 7, it is difficult to draw clear comparisons with previous research that has not taken the fracture mechanics of the material into consideration. Compared to the Mesolithic quartz scatter at Belderrig, however, the Neolithic Thornhill assemblage is distinctly different and appears geared towards the production and use of smaller, thinner flakes knapped using a wider range of techniques, combinations of techniques, and technical procedures than the communities of practice at Belderrig.

[25]Unfortunately, the context fill number it came from, while noted on the bags, is not registered in the site archive as a context fill (Ó Baoill 2009 pers. comm.); this same fill also produced a retouched multiplatform core along with three other cores and eight flakes, one of which was a scraper. Therefore almost a quarter of this fill’s artefacts were retouched.[return]
[26]N/A indicates retouched debris or pebble where localisation could not be discerned due to lack of proximal/distal end to orientate.[return]
[27]The 'utilised pieces' were identified macroscopically. [return]

Home     PhD Contents     PhD Chap 1     PhD Chap 2     PhD Chap 3     PhD Chap 4     PhD Chap 5     PhD Chap 6     PhD Chap 7     PhD Chap 8     PhD Chap 9     PhD Chap 10     PhD Chap 11     PhD Bibliography     PhD Appendices     PhD Plans    [PhD PDF]

Top of Page