Part II: The Finds. Nidum. Excavations at the Roman Auxiliary Fort 2010 to 2012: Neath. by The Glamorgan-Gwent Archaeological Trust Ltd

Nidum. Excavations at the Roman Auxiliary Fort 2010 to 2012: Neath Part II: The Finds July 2018 GGAT report no. 2018/038 Project no. P1547 National Grid Reference: SS 7465 9783

A report for Neath Port Talbot CBC By Martin Tuck MCIfA

Canteen Building (2010)

Vicus Dw

Site 716

r-y

Dwr-y-felin Comprehensive Upper School

lin

Topsoil Strip (New Building Area)

-Fe

Athletics Ground

(1993)

T1.VI (1958)

T1.V (1958) T1.II (1958)

T1.III (1958)

Structure

Area shown in Figure 2

T1.IV (1958) Site 716

Site 677 Site 716 1989

T1.I (1958)

Site 716

Site 674 Topsoil Strip (Car Park Area) Site 740

1984

Principia? 1989 Scheduled Monument (Gm215) area

1984 2nd Roman Fort and Defensive Ditch

Praetorium? 1984

(conjectured)

1984

The Glamorgan-Gwent Archaeological Trust Ltd Heathfield House Heathfield Swansea SA1 6EL

Nidum. Excavations at the Roman Auxiliary Fort 2010 to 2012: Neath, Neath Port Talbot CBC

Contents

Page

1. The Roman pottery (Figures 36-45, Plates 1-15, Tables 1-6) ....................................... 3 2. The post-Roman pottery .............................................................................................. 49 3. The clay pipes .............................................................................................................. 51 4. The glass (Figure 46, Plates 16-22) ............................................................................. 52 5. The Roman coins ......................................................................................................... 65 6. Objects of copper alloy and lead (Figure 47) .............................................................. 65 7. The Roman ironwork (Figures 48-53, Plates 23-25) ................................................... 68 8. The metalworking residues (Tables 7-9, Appendix A) ............................................... 76 9. Archaeometallurgical residues from environmental samples Site 716 (Table 10)...... 93 10. Assessment of graphitic crucible sherds...................................................................... 99 11. The Roman brick, tile and daub (Table 11) ............................................................... 101 12. The stone objects (Figures 54-56) ............................................................................. 104 13. Animal bone (Tables 12-14)...................................................................................... 107 14. The charred plant remains (Table 15) ....................................................................... 113 15. The pollen samples (Site 677) (Table 16) ................................................................. 135 16. Plant macrofossils (Site 677) (Tables 17-18) ............................................................ 137 Appendix A: Archaeometallurgical residues report (Geoarch 2013/27) from Dwr-y-Felin School, Neath (Sites 677 & 716) ............................................................................ 142 Tables Table 1. Forms and Sources, Sites 677 & 716 ........................................................................... 4 Table 2. Site 677. Vessel loss by date chart ............................................................................... 8 Table 3. Site 716. Vessel loss by date chart ............................................................................... 9 Table 4. Sites 677 and 716 combined vessel loss by date chart ................................................. 9 Table 5. Samian forms from sites 677 and 716 ........................................................................ 10 Table 6. Samian forms from sites 677, 716 and 674 ................................................................ 12 Table 7. Summary of residue classes by context (weight in g) ................................................ 83 Table 8. Metalworking residues: Details of samples selected for further analysis. .................. 91 Table 9. Weight distribution of SHC assemblage from Dwr-y-Felin, with selected comparative assemblages. Weights in g. .......................................................................... 92 Table 10. Microresidue classes by context Dwr-y-Felin School, Neath (Site 716) ................. 97 Table 11. Table of Roman tile forms ..................................................................................... 102 Table 12. Animal bone NISP and % values of Species recovered (Sites 677, 716, 740). ..... 108 Table 13. Animal bone NISP and % values of the faunal remains from Dwr-y-Felin (Sites 677, 716 and 740) by Location and Period .................................................................... 110 Table 14. No. of Measureable bones and teeth (NISP) from Dwr-y-Felin (Sites 677 and 716) ........................................................................................................................................ 112 Table 15. Analysis of the charred plant remains .................................................................... 129 Table 16. Pollen data from Site 677, Dwr-y-felin School, Neath, expressed as actual pollen counts. ............................................................................................................................ 135 Table 17. The Plant macrofossil remains from Dwr y Felin School, Neath .......................... 140 Table 18. Charcoal identification from Dwr-y-Felin, Neath - Site 677 ................................. 141

Nidum. Excavations at the Roman Auxiliary Fort 2010 to 2012: Neath, Neath Port Talbot CBC

Plates Plate 1. Context 6342. Form 27, South Gaulish samian stamp ............................................... 15 Plate 2. Context 6342. Form 27, South Gaulish samian stamp ............................................... 15 Plate 3. Context 6057. Form 37 southern Gaulish samian ....................................................... 20 Plate 4. Context 1111. South Gaulish samian form 37 ............................................................ 22 Plate 5. Form 37, South Gaulish samian .................................................................................. 26 Plate 6. Context 2053. Form 18, South Gaulish samian stamp ................................................ 27 Plate 7. Context 2053. Form 18, South Gaulish samian .......................................................... 27 Plate 8. Context 1676. Form 18, South Gaulish samian stamp ................................................ 31 Plate 9. Context 6070. Form 37, Central Gaulish samian ........................................................ 34 Plate 10. Context 6580. Form 37 South Gaulish samian ......................................................... 36 Plate 11. Context 6229. Form 37, Central Gaulish (probably Les Martres de Veyre) ............. 38 Plate 12. Context 1076. Form 37, South Gaulish samian ........................................................ 39 Plate 13. Context 5005. Form 29 South Gaulish samian ......................................................... 44 Plate 14. Unstratified. Form 37, Central Gaulish samian......................................................... 48 Plate 15. Unstratified. Form 18, South Gaulish samian stamp ................................................ 48 Plate 16. Glass No. 2 ................................................................................................................ 53 Plate 17. Glass No. 3 ................................................................................................................ 54 Plate 18. Glass No. 21 .............................................................................................................. 55 Plate 19. Glass Nos. 29, 31, 33, 34, 35, 38, 41, 44, 49, 50. Beads (left to right) a to e top row, f to j bottom row ............................................................................................................... 56 Plate 20. Gaming Counters Nos. 62 -66, 69 (left to right) a to f .............................................. 59 Plate 21. Glass No. 79 .............................................................................................................. 60 Plate 22. Glass Nos. 99, 100, 102, 106, 108. Beads (left to right) a to e ................................. 62 Plate 23. X-ray of padlock 1081............................................................................................... 71 Plate 24. X-ray plates of iron artefacts (H910, H911, H912, H913) ........................................ 73 Plate 25. X-ray plates of iron artefacts (H906, H906, H908, H909) ........................................ 75

Nidum. Excavations at the Roman Auxiliary Fort 2010 to 2012: Neath, Neath Port Talbot CBC

The Roman pottery (Figures 36-45, Plates 1-15, Tables 1-6)

By Peter Webster & Rowena Hart Introduction All pottery has been examined and recorded in an archive list according to form, fabric and the number of vessels likely to be present. It is this list which forms the basis for comments below and which, in conjunction with the stratigraphic report, was the basis for the selection of vessels which required a more detailed catalogue. As is noted or implied many times below, the soil at Neath has not been kind to the pottery. Only the most highly fired of pots retain their original surfaces completely and most have lost all or some of their surface as well as suffering degradation of the body of the fabric in the soil. A number of fabrics (eg. the North Gaulish mortarium fabric of Gillam 238 type) show a tendency to crack or even to shatter into a series of small cubes. In this situation, sherd counts, although used as a method of recording, are of little use for comparative purposes, especially in circumstances where degradation continues as the sherd dries out leading to fragmentation after the original sherds have been bagged up. Equally, surface treatments and estimates of hardness are often elusive. Sources and Dating Sources have been grouped by area and are summarised in Table 1 below. Quantification is by approximate maximum vessel numbers using only those sherds for which a vessel form can be identified. The table is arranged with the most local sources at the top and the most distant at the bottom. Amphorae have been omitted. Pottery in a local Iron Age tradition Throughout southeast Wales, early to mid Flavian sites produce small quantities of pottery in the pre-Roman tradition of bulbous jars with simple beaded or everted rims (cf. Webster, Gwilt & Horรกk 2006 for a discussion of this class of pottery; also Manning 1993, 66-8 & 23-2). In the late Iron Age such vessels will have been used for cooking, but their appearance on sites such as Usk and Caerleon seems to imply some other use in the early Roman period, perhaps as containers. They disappear within no more than a generation of the Roman conquest of Southeast Wales and one or two decades of the Flavian advance (so probably within the decade c.A.D.80-90). The number of such vessels from these Neath excavations is surprisingly small, given the date of the occupation of the first fort. However, the fabric is completely missing from the approximately contemporary early phases at Loughor, further west, and it may simply be that Neath is at the very edge of the area of production.

Nidum. Excavations at the Roman Auxiliary Fort 2010 to 2012: Neath, Neath Port Talbot CBC

677 & 716 Local IA Local Reduced Local Oxidised Other Oxidised Mica dusted Caerleon Ware White slip Severn Valley SW white slip BB1 Oxford Verulamium Terra Nigra N.Gaul

Flagon

Jug

TOTAL %

Jar 1

Bowl

1 1

77 27

20 14

7 5

10 8

677 677 716

677 716

16 10

677 716

18 8

Dish

Lid

7 2

4 5

Other

1 1

5 1

716

677 716 677 716 716 677 716

Mortar

716 677 677 716

Tazza

6 1 1

14 7

2 1

8 3 1

677 677 716 716 677 716

Lyon Samian

Beaker/ cup/ tankard

3 3

3 4 1

38 5.73

2 0.30

3 2 3 36 24 72 10.86

33 6 1 1 143 21.57

92 31 189 28.51

1 0.15

81 12.22

57 46 122 18.40

15 2.26

0 0.00

TOTAL 1

% 0.15

109 50

16.44 7.54

52 37

7.84 5.58

3 1

0.45 0.15

0.30

1 6 1

0.15 0.90 0.15

0.15

6 1 31 13 1

0.90 0.15 4.68 1.96 0.15

6 5

0.90 0.75

1 36 8 3 186 102 663 100.00

0.15 5.43 1.21 0.45 28.05 15.38 100.00

Table 1. Forms and Sources, Sites 677 & 716

Local reduced and oxidised wares Locally produced reduced and oxidised wares predominate (over 37% of all vessels), only surpassed by imported samian ware. Most belongs to the Flavian or Flavian-Trajanic period. South Wales Reduced Ware is well represented (cf. Tomber & Dore 1998, 209 & Pl.174; Manning 1993, 232-255) as one would expect. The place of production of this sandy fabric in the earlier Roman period is unknown, but we may surmise that it was further east where later production centres are known at Caldicot and Llanedeyrn, Cardiff. Alongside this, one would also expect fabrics or forms specific to the site within the Flavian and Flavian-Trajanic period and this is what we appear to have. Both oxidised and reduced versions of a local fabric have been identified and termed Neath Fabrics A and B: Neath Fabric A is the oxidised version. Although it can contain some sandy filler (as South Wales Reduced Ware) it is generally in smaller quantities, the fabric usually appears to be smoother and to be characterised by small flecks varying in colour from red to grey and apparently formed by ground up (and probably pre-fired) clay. Thicker 4

Nidum. Excavations at the Roman Auxiliary Fort 2010 to 2012: Neath, Neath Port Talbot CBC

examples of the oxidised variety of the fabric tend to have a grey core. A variety of vessel classes appear in the fabric including jars and flanged bowls. Flagons and both mortaria and mortar-like bowls appear to have been produced with a white slip, where it survives in the adverse soil conditions (cf, mortaria nos. 19, 29, 37, 102, 113, 154, 186 and 242-3, mortar-like bowl no. 120, flagons 49, 93-4 and the tazze 45 and 76). Neath Fabric B is a reduced version of Fabric A. The filler is similar to that in Fabric A, although usually in varieties of grey. It appears mainly to have been used for jars and bowls. No mortaria or mortar-like bowls in the fabric have been observed. D12 below is allied to this fabric. Most of the self-coloured jars, bowls and dishes present fit within the range of forms found generally in north-west Europe in the Flavian and Trajanic periods â&#x20AC;&#x201C; curved rim and everted rim jars, flanged dishes, flanged and carinated bowls and flanged dishes. As elsewhere, however, local preferences are visible within the general family of forms. Most noticeable is the way in which the flanged and carinated bowls are of simple type with a rim bent over in a sharp curve but without reeding or other decoration. The flanged dishes are also of the simplest. The wavy line decoration (produced with a stiff brush or a comb) common throughout South Wales is found but is perhaps less common than on many sites. Rustication on jars is distinctly scarce. Although everted rim jars are found, it would appear that simple curved rims were more popular on jars. In all, therefore, the impression is of a desire for the plain and the relatively simple. Slipped and other surface decorated fabrics As has already been indicated, the Neath soil tends to degrade surfaces, meaning that slipped finishes were probably more numerous than currently visible. The following were, however, observed in small numbers: Mica dusting was apparent on two vessels on site 716 and was probably in use, on a small scale, within the local (oxidised and reduced fabric) industry. Caerleon Ware. The red and red-brown slip of Caerleon Ware is surprisingly scarce (just one mortarium recorded from site 677, with a roughcast beaker from the other recently excavated site 674). One would not expect large quantities this far from the centre of manufacture, but the very low numbers are probably a product of limited second century occupation. White slips are certainly present on locally produced fabrics and were most noticeable on red flagons and mortaria. The local tazze, from site 677 (nos. 45 and 76) were also probably originally white slipped. South-West White Slip mortaria (cf. Hartley in Manning 1993, 415-8) Are a feature of later levels. The source is certainly east of the Severn and presumably within easy reach of its major market, Cirencester. No examples were sufficiently complete to illustrate, but examples come from contexts 1084, 1338, 1497, 1658, 1731 and 6822. Mica dusting also occurs and may similarly have been more common than is now apparent. 5

Nidum. Excavations at the Roman Auxiliary Fort 2010 to 2012: Neath, Neath Port Talbot CBC

Black-burnished ware In South Wales as a whole, Black-burnished ware (BB1) can be found at any time during the Roman occupation, but usually only forms a significant proportion of assemblages from the second century onwards (with quantities tending to increase across that century). As would be expected on sites where Flavian material predominates, Black-burnished ware was scarce on the Neath excavations. Overall the Black-burnished assemblage seems likely to belong mainly to the second century or later. On site 677 the fabric occurred in only 19 contexts (1002, 1004, 1006, 1009, 1046, 1077, 1086, 1091, 1104, 1110, 1201, 1228, 1276, 1299, 1467, 1494, 1551, 1592, 1594, 5001, 5002 and 5050), of which only context 1201 seems likely to be early. An early jar form was noted in context 1228 and seem most likely to be first century, though not dating the contexts. Examples of the flanged dishes and bowls came from 1002 (modern earthmoving), 1046 and 5001-2 (gravel extraction) and are more likely to be 2nd century. Context 1494 (extra-mural area phase 2) produced one of the few later Roman pieces from the excavations, an example of Gillam 1976, nos.11/12 which is likely to be late 3rd to 4th century. On site 716 the fabric occurred in only 12 or possibly 13 contexts (6043, 6067, 6071, 6138, 6183, 6220, 6250, 6258, 6259, 6275, 6502, 7015 with a further possible sherd from 6110). Of these, contexts 6071 and 6138 are from under roads and may be early, and contexts 6258, 6259 and 7015 included forms which are certainly 2nd century in date. None included pieces which are certainly later, although many of the contexts listed belong to the later Roman or postRoman period. Mortaria A variety of mortaria are represented and are, in the main, characteristic of the Flavian and

Flavian-Trajanic period. Particularly noticeable are examples of the north Gaulish mortarium normally referred to as Gillam 238 although a number of form variations and, indeed, fabric variations are known (cf. Hartley 1998 and Hartley & Tomber 2006, 22-4). These mortaria were current from the 60s A.D. through the Flavian period and into the very early 2nd century. An unfortunate characteristic of the fabric in Welsh soils is its tendency to fragment into small cuboid pieces and these are one of the predominant features of the Neath assemblage. Other imports are also present, including Bushe-Fox type 22/30, another North Gaulish product. Verulamium mortaria were also noted. The white slipped redware mortarium in our local Fabric A and the later South-West White Slipped mortaria have already been mentioned. Amphorae Amphorae were plentiful on the sites excavated. As one would expect, Dressel 20, the South Spanish olive oil amphora predominates but South Gaulish wine amphorae (Gauloise 4 or similar) and South Spanish fish sauce amphorae are also present. Two examples of Haltern 70 amphorae can also be noted (no.89 from context 1006 and also a fragment from context 6066 (not illustrated). In the context of Neath these must be mid first century pieces. The contents of Haltern 70 were probably defrutum, a sweet liquid by-product of the wine industry, used as a preservative especially for olives, in this case from Southern Spain (cf. Monfort 2003, especially 88-9) The similarity of the fabric to that of the common Dressel 20 amphora, not 6

Nidum. Excavations at the Roman Auxiliary Fort 2010 to 2012: Neath, Neath Port Talbot CBC

only betray a similar area of origin but also suggests that body sherds may be present that have not been distinguished from the more common form. One wall sherd of carrot amphora (from context 6505) hints at other imports from the Mediterranean not apparent from the assemblage from this area of the forts. Fineware With the exception of samian (discussed separately below), the sites yielded very little fineware. Although it must be acknowledged that soil action removing surface decoration and slip may have had a part to play here, nevertheless, the absences seem to be real ones. A few contexts produced examples of Lyons Ware, a pre-Flavian fineware which survives in small amounts into the Flavian period and reached a number of Welsh sites of Flavian foundation (cf. Greene 1979, 16, Fig,4, but also Webster 1992, 111). There are also a few examples of North Gaulish roughcast beakers which were probably imported from the mid Flavian period until the midsecond century but are mainly characteristic of the period c.A.D.80-110 in South Wales as they appear to have been driven out of the local market by Caerleon Ware imitations. As already noted, Caerleon Ware itself was very scarce emphasising the essentially first century character of the assemblage. There was also a piece of Terra Nigra (no.79 from pit 1712), a predominantly Flavian fabric, along with at least one local imitation (cf.no.262). Only one piece of Oxford colour coat was noted from any of the recently excavated sites (no.30 from context 6023), suggesting that the focus of later activity (probably a late reoccupation of the fort) is elsewhere. Samian

All samian was examined and is listed in the archive by form and source with approximate dates provided. All had suffered to a greater or lesser extent from soil action. Fabrics tended to be powdery and the breaks rounded. Moulded and impressed surfaces had suffered particularly badly so that detail was missing from many stamps and decorated vessels. Thus, not all stamps proved legible and, on occasion, only the general character of a decorative scheme could be discerned from a surface which had lost all or most of its slip. Given the narrow date range of the material, this probably has a minimal effect on chronological precision, but greatly reduces the amount of detail which can be obtained from decorated and stamped wares and makes recording by means of rubbing impractical as the pressure of rubbing further degrades the decorated surface. It also, of course, greatly reduces the number of pieces which merit illustration. Those selected have been photographed using a strong sidelight to enhance what decoration has survived. Although sites 677 and 716 adjoin and it was felt likely that both sites would tell a broadly similar story, the samian was first examined on a site by site basis. Site 677. The overwhelming majority of the samian was South Gaulish with a few pieces from Les Martres-de-Veyre and fragments of only about 5 vessels from Lezoux. In chronological terms this places most of the pottery before c.A.D.110. In fact, if we attach dates to all vessels for which a clear form can be determined, we find that much of the samian dates to within 20 years of the Roman conquest of central south Wales (Table 2).

Nidum. Excavations at the Roman Auxiliary Fort 2010 to 2012: Neath, Neath Port Talbot CBC

Table 2. Site 677. Vessel loss by date chart

The preponderance of vessels deposited in the period between AD 70 and 90 is clear. Given that many of the vessels which appear in the succeeding two decades can only be broadly dated to c.A.D.70-110 and thus could well have been deposited before 90, a floruit between the Flavian conquest in the mid-70s and c.90 seems likely. Site 716. The pattern which is remarkably similar to that of site 677, with a mid/late 1st century peak and a marked absence of any pottery which can be reliably dated to a period after the middle second century. The histogram for 716 is shown below (Table 3) but we have also produced another which combines 677 and 716 so as to provide a picture for the whole of the major area excavated (Table 4).

Nidum. Excavations at the Roman Auxiliary Fort 2010 to 2012: Neath, Neath Port Talbot CBC

Neath 716

Vessel loss per half decade

25.00

20.00

15.00

10.00

5.00

0.00

55 60 65 70 75 80 85 90 95 100 105 110 115 120 125 130 135 140 145 150 155 160 165 170 175 180 185 190 195 200

Years AD Table 3. Site 716. Vessel loss by date chart

Table 4. Sites 677 and 716 combined vessel loss by date chart

Nidum. Excavations at the Roman Auxiliary Fort 2010 to 2012: Neath, Neath Port Talbot CBC

S. Gaulish Site

677

Les Martres 716

677

Central Gaulish 716

677

Total

1.36

716

Form 15/17

24.83

18R

2.04

18/31

2.72

18/31R

0.34

1.70

0.34

16.67

12.93

0.00

18 or 18/31

22/23

30 33

0.68

3.74

2.38

25.17

0.68

Ritterling 12

1.70

R12 or Curle 11

0.68

Curle 11

1.36

Ritterling 1

0.34

Curle 15

0.34 100.00

Total

171

102

58.16

34.69 3.74

294

2.04

1.36

0.00

100.00

Table 5. Samian forms from sites 677 and 716

Samian forms represented on the two sites are shown above (Table 5). The preponderance of South Gaulish forms and the almost total absence of those from Central Gaul is obvious. There are a very few forms which would normally be considered pre-Flavian. Only forms 22/3 and Ritterling 1 and 12 need occasion comment. Both 22/3 and Ritterling 12 10

Nidum. Excavations at the Roman Auxiliary Fort 2010 to 2012: Neath, Neath Port Talbot CBC

are predominantly pre-Flavian but both also regularly survive long enough to appear on Flavian foundations. There are, for instance, a large number of form 22/3 from Inchtuthil (Pitts & St.Joseph 1985, 316) occupied in the mid-80s A.D., while Ritterling 12 appears on both Pennine and Scottish sites (cf. Wild 2012, 238-9) and at Caerleon and must have survived into the earlier Flavian period. The very abraded example of the plate Ritterling 1 (a more certainly pre-Flavian form) was most probably already of some age when it reached the site. Among the decorated forms, the proportion of the earlier form 29 compared with the later form 37 would suit an early to mid Flavian occupation of sites 667 and 716 (among the South Gaulish wares 38 examples of form 29 as opposed to 58 of form 37). If we consider form 37 to have been in production by about 70AD and to have supplanted the earlier form 29 by the mid-80s, then the proportions seem reasonable for a site occupied from about 75 into perhaps the early 90s (cf. the figures for Scottish sites discussed by Wild 2012, 245-6). Overall, the chart suggests a predominance of just a few forms. The two main decorated bowl forms make up about 38% of the whole assemblage and plates over 26%. Of the remaining 36%, almost half are the small cups, form 27 (16%). Chronological summary The later first century bias of the material as a whole has already been commented upon. Overall, although some forms and fabrics common in the mid first century are present, there is nothing which would be unexpected in a South Wales site of Flavian foundation. There seems no reason, therefore to question the tradition date in the mid-70s A.D. for the foundation of the fort. Equally, there can be little doubt that the majority of the pottery recovered is first century in date. Here, the samian evidence is crucial. In the table below we have added the samian from the as yet unpublished adjacent site 674.

Nidum. Excavations at the Roman Auxiliary Fort 2010 to 2012: Neath, Neath Port Talbot CBC S. Gaulish

Site

677

716

674

Central Gaulish

Les Martres

677

716

674

677

716

Total

1.26

23.66

1.89

3.47

0.32

1.58

0.32

16.09

674

Form 15/17

18R

18/31

18/31R

18 or 18/31

22/23

12.62

0.63

1.26

3.47

2.52

25.55

0.95

1.58

0.63

1.26

0.32

100.00

30 33

Ritterling 12

R12 or Curle 11

Curle 11

Ritterling 1 Ritterling 13

Curle 15

Total

171

102

53.94 32.18

317

3.79

3.47

1.89

0.63

1.26

0.00

2.84

100.00

Table 6. Samian forms from sites 677, 716 and 674

Even if we add in the samian from site 674 (Table 6) which certainly cuts through part of the later fort, the amount of samian from Les Martres-de-Veyre (exporting mainly c.A.D.100-120) and Lezoux (exporting mainly from c.A.D.120) is very low (about 10% of the whole). This may 12

Nidum. Excavations at the Roman Auxiliary Fort 2010 to 2012: Neath, Neath Port Talbot CBC

be distorted by the small number of vessels overall from site 674 (a calculation based just on the 23 samian vessels from 674 gives c.52% South Gaulish, 9%, Les Martres and 39% Lezoux) but the predominance of South Gaulish ware from the sites as a whole is overwhelming. Within the South Gaulish assemblage, the preponderance of form 18 over 18/31 and the relatively high ratio of the decorated form 29 against the later 37 probably argues for an occupation mainly before c.A.D.90 on the sites examined. Further than that, it is probably not safe to go on a site which was not abandoned but simply rebuilt on a marginally different site. The samian evidence is in no way contradicted by the other pottery, although, of course, this does include some few vessels which belong to later activity on or near the sites, extending beyond the samian importing period. Economic and Social Implications The Neath assemblage is reasonably typical for a Flavian/Trajanic site and shows the degree to which the ceramic trade of the later first and early second century relied upon non-British sources for specialist wares. Finewares are almost exclusively samian and, where they are not, they are mainly Gallic imports. Gaul also forms the main source of mortaria although, here, some British sources (including a local one) are also evident. Amphora-born commodities consist largely of South Spanish olive oil, although there is also some evidence for other South Spanish products including fish products and for South Gaulish wine. A number of very large flagons probably also indicate the importation of liquid commodities. Here we can chart this later first and early second century reliance upon non-British sources for specialist wares, as well as the importance of local production of kitchen wares by rendering part of the chart, table 5, in the form of a pie chart, placing sources clockwise from local to distant: Sites 677 & 716: Vessel Sources Local IA Local Reduced Local Oxidised Other Oxidised Mica Dusted Caerleon Ware White Slip

Severn Valley SW White Slip BB1 Oxford Verulamium Terra Nigra N Gaul Lyon Samian

This shows fairly clearly how at least 32% of vessels were local, while around 50% were from Gaul with the remainder made up of pottery from Southern Britain from the Severn Valley 13

Nidum. Excavations at the Roman Auxiliary Fort 2010 to 2012: Neath, Neath Port Talbot CBC

eastwards. This reliance upon imports would appear even greater if we were to include amphora borne commodities, mainly South Spanish olive oil in this case. The range of vessel classes in use on the Neath sites is fairly typical for a military site in South Wales. Again part of Table 1 can be represented as a pie chart: Sites 677 & 716: Classes of vessel

Flagon Jug Beaker Jar Bowl Tazza Mortarium

Dish Lid

As one would expect, jars and bowls predominate with a reasonable spread of other forms. Overall the assemblage seems very utilitarian in character, with an emphasis on cooking rather than on food preparation or consumption. However, this is offset by the quantity of samian present (43% of all vessels), all of which can probably be regarded as tableware. If we break down the table of samian forms present on sites 677 and 716 into broad types we arrive at: Plain cups Plain bowls Plain dishes Decorated bowls Decorated beakers

21.09% 3.74% 36.39% 38.1% 0.68%

We should perhaps also consider that the areas excavated were mainly working areas rather than those occupied by accommodation and this might well have altered the balance of vessels deposited there. The Catalogue The catalogue has been divided by area: Defences, Intervallum, Interior, Extra-mural. Where feasible each section is sub-divided in a way which mirrors the structural report above. Catalogue numbers 50-63 and 194 and 195, which were originally attributed to the oven areas are now known to be from the extra-mural areas; in order to maintain the catalogue numerical sequence these have not been rearranged. Sub-divisions are arranged, where possible, in 14

Nidum. Excavations at the Roman Auxiliary Fort 2010 to 2012: Neath, Neath Port Talbot CBC

numerical order of context. See Part III Figures 36-45 for illustrations referenced by integer and D number i.e. 2. (D271). The Defences Drain under the Rampart Context 6338: With South Gaulish samian form Ritterling 12 (c.A.D.50-80 in a Welsh context): 1. (D270). (Not illustrated) Neck of a large, probably two-handled flagon in Fabric A. Context 6342: 2. (D271).Form 27, South Gaulish (Plates 1, 2). The stamp reads CIINâ&#x20AC;˘So retrograde, die 15a of Censor i of La Graufesenque, NOTS2, 335-341 c.A.D.70-90.

Plate 1. Context 6342. Form 27, South Gaulish samian stamp

Plate 2. Context 6342. Form 27, South Gaulish samian stamp

The Rampart Construction Context 1108 included a sherd of South Gaulish samian form 29 (c.A.D.65-85). Also from the Rampart: 15

Nidum. Excavations at the Roman Auxiliary Fort 2010 to 2012: Neath, Neath Port Talbot CBC

Context 1009, a post-hole, with a Les Martres samian bowl form 37 (c.A.D.100-120): 3. (D37). Curved rim jar in light grey fabric with a brown core. Curved rims appear to have been popular at Neath in the Flavian-Trajanic period, cf. Heywood & Marvell 1992, Fig.7, 6-12 and ibid. 10 for this vessel. Rampart Breach Context 1091 with South Gaulish samian forms 18R or 18/31R, 27 and 29: 4. (D61). (Not illustrated). soil action.

Mortar-like bowl in light orange fabric A somewhat eroded by

5. (D62). (Not illustrated). Flanged bowl, or more probably dish in Black-burnished ware; cf. Gillam 1976, no.57 (early to mid 2nd century). Context 1467: 6. (D129). (Not illustrated). Flanged dish in Black-burnished ware; Gillam 1976, no.56 (early to mid 2nd century). Context 1555 (a clay deposit) included a fragmentary Dressel 20 amphora rim of MartinKilcher type 18 (Peacock & Williams 1986, Fig.65). Mid/late to late 1st century. Also: 7. (D143). Jar in light grey with a mid grey exterior surface shading to grey/pink internally. There is a shoulder groove suggesting a decorated vessel of the same general type as Manning 1993, Fig.108, 13.2. Mid 1st to early 2nd century. Comment The material from the rampart is mainly consistent with an early Flavian date (c.A.D.70-80) but with pieces which suggest that the rampart recess and the post-hole 1009 are later and probably early 2nd century. Fort Ditches Fill of outer ditch, context 6268: 8. (D255). Dressel 20, South Spanish olive oil amphora; Martin-Kilcher type 20 (Peacock & Williams 1986, Fig.65). Late 1st to early 2nd century. The Intervallum Underlying the internal road (via sagularis) Context 6588: With South Gaulish samian form 18 (c.A.D.70-90): 9. (D293). Mortarium in a fine granular cream fabric, probably a Verulamium region product. A stamp has been placed diagonally and reads ]LVGVD, the counter-stamp of the potter Albinus, cf. Frere 1972, Fig.145, 5-10 & 1984, 282. c.A.D.60-90. 10. (D294). (Not illustrated). Rim fragment, broken at the edge of the spout in a coarser (and more common) Verulamium fabric. Mid-late 1st century. 16

Nidum. Excavations at the Roman Auxiliary Fort 2010 to 2012: Neath, Neath Port Talbot CBC

Context 6658: The context included South Gaulish samian form 27. Clay deposits under the Via Sagularis: Context 6117: With South Gaulish samian form 18 or 18/31: 11. (D212). Mortarium of Gillam 238 type in light buff; Hartley 1977, Fig.2.1, 3F. Mid 1st to early 2nd century. Context 6127: 12. (D213).

Everted rim jar in brown fabric with traces of a grey surface.

Context 6138: The context included South Gaulish samian form 37: 13. (D214).

Curved rim jar in grey Fabric B.

14. (D215).

Flanged bowl or dish in Black-burnished ware. 2nd century.

Context 6156: The context included South Gaulish samian form 27. Comment As with the rampart material, most of the material found under the road point to a Flavian date. One fragment, D215, from context 6138 seems anomalous (cf. comments on no.100 below) and more likely to be early 2nd century. The Via Sagularis Road Area A: Context 6107 (initial resurfacing): The context included South Gaulish samian form 18 (c.A.D.70-90). Context 6095 (2nd resurfacing): The context included South Gaulish samian form 36 (c.A.D.70-110) Context 6508, under the 3rd resurfacing (6059): 15. (D282). Handle of an amphora or large flagon in pink very micaceous fabric. In the absence of an obvious candidate among amphora forms a large flagon is likely, although the likely size suggests use as a liquid commodity container. Road Area B: The section produced no diagnostic pottery Road Area C: 17

Nidum. Excavations at the Roman Auxiliary Fort 2010 to 2012: Neath, Neath Port Talbot CBC

Deposits overlying the road foundation were undifferentiated (Context 1004) and included a South Gaulish samian bowl, form 37 with zonal decoration (probably c.A.D.70-90). Also: 16. (D25). Flagon neck (12 fragments) in creamy white fabric. The even rings surmounted by a narrow top ring ally this with mid 1st century forms such as Usk Fortress form 4 (Manning 1993, Fig.3 & p. 19). A mid to mid/late 1st century date may be suggested. 17. (D26). Rim, probably of a small flask in eroded light orange, probably fabric A with a flaring neck and double rim. 18. (D27)). (Not illustrated). Part of the spout and flange of a Verulamium region mortarium in sandy buff fabric. A worn and partial stamp may read CIF probably retrograde and a stamp of Bruccius; cf. Frere 1972, Fig 145, no.13 & Frere 1984, 283. C.A.D.80-120. 19. (D28). Mortarium in pink Fabric A with slight traces of a white slip. The context also included the flange of a mortarium or mortar-like bowl in light orange Fabric A. 20. (D29). Mortarium in orange Fabric A with white quartz trituration grits (8 fragments). The form seems to imitate the imported mortaria of Gillam 238 type (Hartley 1977, Fig.2.1, 3) 21. (D30). Lid in Black-burnished ware (at least 6 pieces); cf. Wallace & Webster 1989, 90-1. The Black-burnished lid was made from the mid 1st century until at least the late 2nd. So, although there is no reason why this Neath example should not date from the early occupation of the site, it should be noted that Black-burnished ware is rare upon the site and appears mainly in 2nd century or later contexts. Road surface, context 6108 included South Gaulish samian form 27 (c.A.D.70-110). Comment Material from the road is mainly Flavian with a few probable Flavian-Trajanic or Trajanic pieces. Area between the Via Sagularis and the Ramparts Charcoal and clay from the ovens had accumulated against the northern edge of the road, context 1201: The context included South Gaulish samian forms 18, 27 and 37(2). Also: Context 1201: 22. (D89). Two handled flagon in orange. The type appears at Usk (Manning 1993, 20-21 & Fig.4, type 8) and also Caerleon (Nash-Williams 1929, Fig.30, 42) suggesting at least a mid 1st to early 2nd century context. 23. (D90). Jar in light grey-buff with a filler of fired clay. The form resembles the Usk fortress type 11.1 and this is probably a burnt example of a vessel similar to no.43 from context 1731 below. Mid-mid-late 1st century. 24. (D91).

Curved rim jar in light brown with a grey surface around the rim.

25. (D92). Gillam 238 type mortarium; cf. Hartley 1977, Fig.2.1, 3C and no.114 from context 2035 below. c.A. D.60-110. 18

Nidum. Excavations at the Roman Auxiliary Fort 2010 to 2012: Neath, Neath Port Talbot CBC

26. (D93). Verulamium mortarium in granular buff, cf. Frere 1972, Fig.106, no.224, Frere 1984, 268. Mid-late 1st century. 27. (D94).

Flanged bowl in pink fabric A.

28. (D95). Small flanged bowl in Black-burnished ware. Most of these bowls are 2nd century but an earlier date in our area cannot be ruled out. Context 6516: pre-fort surface under 6023 29. (D283). slip.

Mortarium in light orange Fabric A with a grey core and faint traces of white

Context 6023: With South Gaulish samian forms 18, 29 and Ritterling 12: 30. (D196).

Jar in grey South Wales Reduced fabric, burnt and sooted externally.

31. (D197). Dish in pink to grey fabric heavily burnt but probably originally Oxford red colour coated ware; cf. Young 1977, C47-8. Late 3rd to 4th century. 32. (D198).

Lid in granular grey-brown fabric with a dark grey surface. Wheel thrown.

South of the Via Sagularis Context 6057: 33. (D201). Form 37, Southern Gaulish (Plate 3). The design is heavily abraded and the ovolo frieze has largely been removed in finishing. Below is panel decoration with a saltire, the upper quadrant of which includes the quadruple leaf, Knorr 1919, Tb.12, top row, 4th from left, Webster 1987, H4. To the left is a much eroded arcade formed from an inverted festoon resting on columns; cf. Mees 1995, Taf.77, 4 (Germanus III) for a broadly similar design.and ibid. Taf.193, for a very similar saltire alongside an arcade (Sulpicius). c.A.D.80-100.

Nidum. Excavations at the Roman Auxiliary Fort 2010 to 2012: Neath, Neath Port Talbot CBC

Plate 3. Context 6057. Form 37 southern Gaulish samian

Ovens Oven 1023 and area: Context 1272 (between 1217 and 1269) included South Gaulish samian form 37. Oven 1039 and area: Context 1786: 34. (D164). Flanged rim jar in light orange with a grey core, probably local in manufacture. cf Heywood & Marvell 1992, Fig.7, 2 from primary levels in the 1958 excavations Context 1861: 35. (D169). below.

Flanged bowl in light orange fabric with a grey core; cf. no.82 from context 1822

36. (D170). Mortarium in orange Fabric A with a paler orange core and surface. The trituration grit was probably white quartz. A surface slip is possible but has largely disappeared apart from a few burnt remnants. Context 1919: 37. (D171). the rim.

Mortarium in Fabric A with traces of white slip, burnt brown and sooted over

Context 1140 (by oven 1039) included a fragment of South Gaulish samian bowl form 37, probably c.A.D.90-110. Context 1827 (South of oven 1039):

Nidum. Excavations at the Roman Auxiliary Fort 2010 to 2012: Neath, Neath Port Talbot CBC

38. (D167). Mortarium of Gillam 238 type. Soil action has caused most of the sherds to fragment but a portion of spout remains with an eroded stamp visible (perhaps a single line Veranius stamp), cf. Hartley 1977, Fig.2.1, 3A. Mid 1st – early 2nd century. Oven 1513/1579 and area Context 1710 (associated with 1579): 39. (D153). Mortarium in light fawn with pinkish tinges. Although the fabric itself is smooth, both on the surface and in the break it has numerous broken fragments of ?quartz. The fragment from 1710 has been badly eroded by soil action but the form seems to follow the more rounded North Gaulish forms and a continental origin seems likely. Oven 1645 and associated deposits Context 1731: The context included South Gaulish samian forms 18, 22/3, 27, 29, 35 and Ritterling 12. Of these, 22/3 and Ritterling 12 are predominantly pre-Flavian although they do occur on Flavian foundations. Overall, the samian would suit a date early in the occupation of the Neath fort. In addition, there was what may be the broken and burnt base of a South Gaulish wine amphora of Gauloise 4 type, a neck fragment of a Verulamium flagon, a worn fragment of Verulamium mortarium as Frere 1972, Fig.106, 225 c.A.D.60-110 (see no.214 from context 1046 below) and the base of a rough cast beaker from North Gaul (probably mid–late Flavian). Along with a flanged bowl in orange with a grey core, possibly imitating the samian form Ritterling 12 (cf. no 58 from context 1111 below) there was: 40. (D155). (Not illustrated). Form 37, South Gaulish. An abraded sherd of panel decoration shows a small portion of a large saltire to the left, with, to the right, the legs of what is probably a bestiarius or, more probably, the gladiator with trident, O.1013E, over a small dog, O.1968, chasing a hare. Below is a basal wreath of S-shaped gadroons. c.A.D.70-100. 41. (D156). Dressel 20 South Gaulish olive oil amphora, Martin-Kilcher type 12 (Peacock & Williams 1986, Fig.65). Mid to mid/late 1st century. 42. (D157). Ring-necked flagon in light orange-red fabric. The even rings suggest a 1st century date. 43. (D158). Jar in light grey fabric. The form resembles Usk fortress type 11.1 (Manning 1993, 22-6 & Fig.4). Probably Flavian. 44. (D159).

Curved rim jar in reddish brown with a grey core and grey surface.

45. (D160). Tazza in pink to orange fabric with traces of a white slip. The origins of the form are discussed by Greene in relation to the Usk fortress tazze (Manning 1993, 40-1 and Fig. 6, type 26. The tazza is well represented at Caerleon (eg. Nash-Williams 1932, Figs.61-2, nos.408427, all apparently from 2nd century or later contexts) but is much rarer on auxiliary sites. 46. (D161). (Not illustrated). Mortarium of Gillam 238 type, cf. Hartley 1977, Fig.2.1, 3B. Mid 1st – early 2nd century.

Nidum. Excavations at the Roman Auxiliary Fort 2010 to 2012: Neath, Neath Port Talbot CBC

Unstratified from the ovens area: Context: internal oven, unstratified: 47. (D190). Dressel 20 amphora, Martin-Kilcher type 12 (Peacock & Williams 1986, Fig.65). Mid to mid/late 1st century. 48. (D191). Amphora rim in light buff with a light orange core; the form (Camulodunum 186C, PĂŠlichet 46) indicates use as a container for South Spanish fish sauce; cf. Peacock & Williams 1986, 122-3, Martin-Kilcher 2003, (especially Fig.5, F). Mid 1st to early 2nd century. 49. (D192). Ring neck flagon in pink Fabric A with a grey core and traces of white slip traces of white slip. Comment The scanty yield of pottery from the ovens all belong to the mid first to early second century and a Flavian to Flavian-Trajanic period of use seems likely. Context 1111 (Note: Catalogue numbers 50 â&#x20AC;&#x201C; 63, and 194 and 195, from the extramural area) With South Gaulish samian forms 18, 29, 36 and 37: 50. (D68). Form 37, South Gaulish (Plate 4). Five fragments from the same vessel show a basal wreath of triple leaves below panel decoration. Only the very bottom of the panels survive but the Pan, O.714, a running hare and a figure which is probably D.115 (the right half of the erotic O.A but without its accompanying flute player) can be recognized. The style suggests a later South Gaulish piece, perhaps c.A.D.90-110.

Plate 4. Context 1111. South Gaulish samian form 37

51. (D69). (Not illustrated). Form 37, Les Martres-de-Veyre. Three fragments may be from the same vessel. The surface of the two decorated pieces has been almost totally eroded by soil action. A basal wreath of triple leaves is visible below panel decoration. A small fragment of 22

Nidum. Excavations at the Roman Auxiliary Fort 2010 to 2012: Neath, Neath Port Talbot CBC

the panels shows a running ?panther (probably O.1570 used by X-9 and X-13) above, with an acanthus spray, probably Rogers 1974, K2, used by several Les Martres potters (including X9 and X-13), below. c.A.D.100-120. 52. (D70). Flagon neck in orange fabric. The prominent upper ring suggests a late 1st to 2nd century date. 53. (D71).

Everted rim jar in brown fabric with a grey core and grey surface.

54. (D72). Necked jar in light grey South Wales Reduced fabric, cf. Manning 1993, Fig.110, 25.1. Late 1st to early/mid 2nd century. One of at least five necked jars. 55. (D73). Curved rim jar in light grey fabric with a darker surface and heavy sooting on the neck externally; cf. Manning 1993, Fig,110, 25.1 (late 1st – early 2nd century). . One of at least five curved rim jars. 56. (D74).

Jar in light grey South Wales Reduced fabric, sooted on the neck.

57. (D75).

Jar in light grey Fabric B.

58. (D76). Flanged bowl in light grey fabric with traces of a darker surface and sooted externally. One of the mid 1st to early 2nd century flanged and carinated bowls. The context also yielded the flange of a light grey bowl possibly imitating the samian form Ritterling 12 and, therefore, mid to mid/late 1st century. 59. (D77). Mortarium of Gillam 238 type, cf. Hartley 1977, Fig.2.1,3B and no. 46 above from context 1731. c. A.D.60-110. 60. (D78). Mortarium in light orange with a grey core. There is a groove on the rim. The trituration grits are quartz-like. 61. (D79). (Not Illustrated). Dish in brown fabric with a dark grey surface, burnished on the exterior of the rim. The form appears to be derived from Gallo-Belgic forms (cf. no.259 from context 6240 below). The lid no. 62 below is in a similar fabric and would fit the dish. 62. (D80). (Not Illustrated). Lid in brown fabric with a grey surface. Cf.no.61 above. Context 1770 (fill of 1769): 63. (D162). Jar in orange slightly gritty fabric. The high shoulder suggests a mid 1st to early 2nd century vessel. Fired clay deposits and waste associated with the ovens Context 1256: Clays associated with the context included South Gaulish samian form 18 (c.A.D .70-90). Context 1293: 64. (D109). (Not illustrated). Handle from a Black-burnished ware jug. The attachment to the wall was through the ‘mortice and tenon’ arrangement seen in Wallace & Webster 1989, Fig.2, 9.

Nidum. Excavations at the Roman Auxiliary Fort 2010 to 2012: Neath, Neath Port Talbot CBC

65. (D110). Bowl or, more probably, a dish in light grey South Wales Reduced fabric. The rim has been formed by flattening the edge, thus widening it to a slight wedge. Context 1252 (underlying 1130-1) with South Gaulish samian forms 18, 27, 37 and Curle 11: 66. (D106).

Curved rim jar in light grey Fabric B.

67. (D107). Mortarium in granular light buff fabric from the Verulamium region; cf, Frere 1972, Fig. 106, 225 & no. 214 from context 1046 belwo. Mid-late 1st century. The context also contained fragments of a mortarium of Gillam 238 type. Context 1052 included a South Gaulish samian cup, form 35. Context 1553 included South Gaulish samian form 37 (probably c.A.D.70-90). Context 2009 included South Gaulish samian form 18R (c.A.D.70-90). Context 6440 included South Gaulish samian form 18 (c.A.D.70-90). Context 6474: 68. (D278). 16.2.

Everted rim jar in hard fairly smooth grey fabric; cf. Manning 1993, Fig.109,

Context 6983 with South Gaulish samian form 18 (c.A.D.70-90): 69. (D309). Base and two sherds of a beaker in greenish buff with traces of a brown colour coat. A Lyon product, probably a roughcast beaker as Greene 1979, Fig.14. Mid 1st century, Context 6988 included South Gaulish samian form 18 (c.A.D.70-90). Context 6998 with South Gaulish samian form 36: 70. (D310). (Not illustrated). Form 29, South Gaulish. A small fragment of upper zone shows panels with palmate leaves arranged on their sides in rows to the left and what may be part of a figure to the right. For a similar arrangement see Mees 1995, Taf. 152, 1 (Murranus), c.A.D.60-80. 71. (D311). Mortarium in light granular cream fabric from the Verulamium region, cf. Frere 1984, Fig.111, 2652 9. Late 1st-mid 2nd century. Context 7018 included South Gaulish samian form 18 (c.A.D.70-90). Other features in the intervallum Structures Context 1625 (Associated with 1370): 72. (D146). Neck of a small flask or jar in granular white fabric. Probably a product of the Verulamium region which has come in as a make-weight within a consignment of mortaria. 73. (D147). Jar in light grey South Wales Reduced fabric. The form closely resembles Usk fortress type 11.1-2 and a mid-late 1st century date can be suggested. Structure 1795, 1841: 24

Nidum. Excavations at the Roman Auxiliary Fort 2010 to 2012: Neath, Neath Port Talbot CBC

Context 1841 included South Gaulish samian forms 18 and 29 (c.A.D.65-90). Context 1844 (over 1841 with South Gaulish samian forms 18R and 27: 74. (D168). Mortarium in granular buff fabric from the Verulamium region. About half the rim survives including a stamp reading SECVNDV[S], cf. Frere 1972, Fig.146, 37 dated there c.A.D.65-95. Pits Pits are listed in numerical order of context Context 1497: 75. (D139). Eight fragments of Gillam 238 type mortarium; cf.Hartley 1977, Fig.2.1, 3B and no.46 above for further references. c.A.D.60-110. Pit 1549: 1152, 1497-8, 1872, 1875-6 Context 1152 included South Gaulish samian forms 15/17 and 35 and: 76. (D86). Tazza in eroded light red fabric with traces of white slip. Probably Fabric A; cf. no.45 from context 1731 for a more complete example and a discussion. 77. (D87). Lid in granular brown fabric shading to light grey near the surface and with a dark grey surface Pit 1593: 1594=2153, 1800, 1804 Context 1594: 78. (D145). Form 37, South Gaulish (Plate 5). Five fragments, probably from the same bowl. Three fragments, two of which may join (edges are too rounded for certainty) show an abraded zonal decoration divided by fine bead rows. The terminal of the ovolo tongue appears as a small blob (but may be a trident). It lies beneath a wavy line. Below the ovolo frieze is a zone of triple leaves as Knorr 1919, Tb.12, top row, 3rd from right (Webster 1987, H8) from bowls stamped by Iucundus, Mercator, Niger, Secundus and Vitalis. Below this is a winding scroll with stirrup buds in both upper and lower lobes, except for one lower lobe which is divided with leaf tips below the division. The upper lobes contain a tendril on which perches a small bird, similar to O.2271A The lowest zone is largely missing but appears to have been a wreath, perhaps of chevrons. The general zonal style is one familiar from the Pompeii hoard (Atkinson 1914) although without close parallels there. A date c.A.D.70-90 seems likely.

Nidum. Excavations at the Roman Auxiliary Fort 2010 to 2012: Neath, Neath Port Talbot CBC

Plate 5. Form 37, South Gaulish samian

Context 1800 included South Gaulish samian form 18 and 29 (c.A.D.65-90) Pit 1712: Context 1713 (fill of 1712) included South Gaulish samian form 37, the flange of a mortar-like bowl in pink-buff fabric A and with slight traces of painted white decoration on the flange (perhaps a frieze of festoons). Also: 79. (D154). (Not illustrated). A small rim fragment of Terra Nigra, probably from a bowl such as Greene 1979, Fig.46. Mid to late 1st century. From the fill of pit 1712 (context 2148) with South Gaulish samian form 37: 80. (D175). (Not illustrated). Lid or shallow dish in granular dark grey with a brown core, cf. Heywood & Marvell 1992, Fig. 36, 52 & 55. Pit 1840 Context 1822 (fill of Pit 1840) with South Gaulish samian forms 18R and 29 (c.A.D.65-90): 81. (D165). Flagon neck in bright orange fabric; the rim is of the so-called ‘Hofheim’ type. The bell-shaped collar resembles that found on Central Gaulish moulded and glazed vessel (cf. Déchelette 1904, pl.II, 60) and a mid 1st century date seems likely for our piece. 82. (D166). Flanged bowl in orange with a grey core. Probably intended to be reminiscent of the samian form Ritterling 12. Mid-late 1st century. Pit 2013 (cut by pit 2016), 2014 Context 2014: 83. (D172).

(Not illustrated). Curved rim jar in grey Fabric B.

Pit 2052-3 26

Nidum. Excavations at the Roman Auxiliary Fort 2010 to 2012: Neath, Neath Port Talbot CBC

Context 2053 with South Gaulish samian forms 18, 27 and 37: 84. (D174). Form 18, South Gaulish (Plates 6, 7). A near complete small plate stamped SEVER+, die 7e” of Severus iii of La Graufesenque, NOTS8, 252-267. Hartley and Dickinson date Severus iii’s period of production to c.A.D.65-95. The low wall of our example suggests manufacture earlier, rather than later within this period. Perhaps c.A.D.65-80.

Plate 6. Context 2053. Form 18, South Gaulish samian stamp

Plate 7. Context 2053. Form 18, South Gaulish samian

Context 2148: see context 1712 Pit 6035 and related contexts:

Nidum. Excavations at the Roman Auxiliary Fort 2010 to 2012: Neath, Neath Port Talbot CBC

Context 6037 included a base in pink with a grey core, possibly very abraded, perhaps originally mica dusted. Context 6479 with South Gaulish samian forms 18, 27, 27g, 29, 67 and Ritterling 12, which, as a group, probably dates c.70-85): 85. (D280). (Not illustrated) Mortarium in a fabric similar to that of Gillam 238 type, much eroded by soil action; The rim more closely resembles those of Hartley Group I (1977, Fig.2.1, 1) suggesting that this is Hartleys sub-group iii (1998, 206) from North Gaul. Mid 1st – mid 2nd century. Context 6516 included South Gaulish samian form 18 (c.A.D.70-90). Other intervallum contexts (in numerical order) Context 1005 included a South Gaulish samian form Ritterling 12, the flange and spout of a mortarium in pink Fabric A, much eroded by soil action. There is also a very eroded samian plate, lacking all slip but probably a Les Martres form 18/31 dating c.A.D.100-130 and: 86. (D31). (Not illustrated). Form 29, South Gaulish. The stamp was poorly impressed and has suffered from erosion in the soil. It reads OF…EC…. A stamp of Secundus ii of La Graufesenque is possible but not certain; cf. NOTS8, 170-182. The form confirms a date c.A.D.60-85. Context 1006 with South Gaulish samian forms 15/17, 18 and 37 and ten burnt fragments of a North Gaulish roughcast beaker, fragments of a curved rim jar, a very eroded flanged bowl and an everted rim jar in grey, also 87. (D32). Ring-neck flagon in granular orange fabric. The even rings resemble those found on mid 1st century examples (eg. Manning 1993, Fig.3, 4, Fig.148, 2-4) and a mid or mid to late 1st century date seems probable. 88. (D33). Dressel 20, South Spanish olive oil amphora, Martin-Kilcher, type 17, mid/late 1st to early/mid 2nd century. A wall sherd bears the graffito, IVII, perhaps part of an indicator of volume. 89. (D34). Handle in granular buff fabric very similar to that of the common Dressel 20 amphorae but probably from a Haltern type 70 (Peacock & Williams 1986, class 15, pp.115-6). For a review of forms and dating, see Monfort 2003. Probably mid-late 1st century in this context. 90. (D35).

Everted rim jar in grey. One of the mid 1st to early 2nd century series.

91. (D36). Flanged bowl in Black-burnished ware burnt light red near the surfaces. The very upright wall suggests a vessel early in the series; cf. Gillam 1976, 34 (early to mid 2nd century). See also context 1227 below. Context 1053 included a South Gaulish samian cup form 35. See also context 1170 below. Context 1117 (charcoal rich clay) with South Gaulish samian forms 18, 27, 29, 35, 37(3) and a flanged bowl, either Ritterling 12 or Curle 11, also 28

Nidum. Excavations at the Roman Auxiliary Fort 2010 to 2012: Neath, Neath Port Talbot CBC

92. (D82). (Not illustrated). Form 18, South Gaulish. The stamp appears to read OIIMM. Not identified. c.A.D.70-90. 93. (D83). Large ring-necked flagon in light orange fabric. The surface has been severely eroded by soil action but was probably originally white slipped. The form resembles that found at Usk (Manning 1993, 19 & Fig.3, 4 where the type is discussed). A mid 1st century date for our piece seems possible. 94. (D84). Small ring-necked flagon in similar fabric to the above and with trace of white slip. The rim is more everted than the Usk fortress fabric type but the even rings are still compatible with forms found in the Usk fortress (cf. Manning 1993, Fig.18, 9-10). A mid or mid-late 1st century date may be suggested. 95. (D85).

Jar in orange fabric with traces of a darker surface burnt grey externally.

Context 1170 (under 1053): 96. (D88).

Flanged rim, probably from a jar in brown granular fabric with a grey surface.

Context 1187: The context included South Gaulish samian forms 15/17 and 29. Context 1227 (under 1006): 97. (D97). Narrow necked jar in light grey Fabric B with a darker surface much eroded by soil action. The filler appears to be fragments of fired clay. Jars with double rims such as these are commonest in the later Roman period in this area (cf. Manning 1993, 237, type 10) but the occasional earlier example has been noted and this piece may be one such. Context 1228 (between 1101 and 1052) included South Gaulish samian forms 27g, 29, 35 and 37 (including one probably dating 90-110), a necked jar, at least one curved rim jar and a jar similar to Usk fortress type 11.1 (see no.43 from context 1731 above), also 98. (D98). Dressel 20 South Spanish olive oil amphora neck, Martin-Kilcher type 17 (Peacock & Williams 1986, Fig.63). Mid/late 1st to early/mid 2nd century. 99. (D99). Jar in Black-burnished ware or a related fabric (7 fragments). The dark grey fabric is edged in light orange with a dark grey burnished surface externally and a lighter grey one internally. The upright rim has a zig-zag decoration. The characteristics of Black-burnished Ware jars are discussed by Holbrook and Bidwell (1991, 95) and this jar would fit well into their early period (mid 1st to early 2nd century). 100. (D100). Flanged bowl in Black-burnished ware, sooted externally and burnt orange to grey especially externally. Flanged bowls in Black-burnished ware are mainly 2nd century in date but earlier vessels closely resembling contemporary flanged and carinated bowls do occur (e.g. Manning 1993, Fig.27, 7). It seems more likely, however, that the present piece is early 2nd century in date; cf. Manning 1993, Fig.127, 23.1 but also Holbrook & Bidwell 1991, 97 on the two Exeter vessels mentioned under type 23. 101. (D101). Mortarium in granular light buff fabric from the Verulamium region, Frere 1972, Fig.106, 226, Frere 1984, 269. c.A.D.70-110. 29

Nidum. Excavations at the Roman Auxiliary Fort 2010 to 2012: Neath, Neath Port Talbot CBC

102. (D102). Mortarium in orange-pink Fabric A with traces of white slip and with quartz trituration grits. Context 1255 included a South Gaulish samian form 18. Context 1414 (clay deposit): 103. (D126). Everted rim jar in light grey fabric with a pronounced lid seating. Context 1442 (clay deposit) included two joining fragments of mortarium flange in pink fabric A with a grey core and traces of a white slip. Also: 104. (D128). Curved rim jar in grey South Wales Reduced fabric. Three fragments, two of which show sooting externally. Context 1551 (clay deposit) included South Gaulish samian form 18 (c.A.D.70-90), the abraded rim of a Verulamium mortarium and: 105. (D140). Narrow-necked jar in light grey Fabric B with a darker surface. With at least one other narrow necked jar and a fragment from the girth of a jar with wavy line decoration. 106. (D141). Dressel 20 South Spanish olive oil amphora, Martin Kilcher type 17 (Peacock & Williams 1986, Fig.65. Mid/late 1st to early 2nd century. 107. (D142). Flanged bowl in granular red-brown fabric with a grey surface sooted externally. Probably a local variation on the common mid 1st to early 2nd century flanged and carinated bowl. Context 1592 with South Gaulish samian form 18: 108. (D144). (Not illustrated). Curved rim jar in light grey South Wales reduced fabric shading to red-brown internally and sooted on the rim. Context 1595 included a South Gaulish samian form 18, an orange flagon neck and the flange of an orange mortarium probably of Welsh manufacture. Context 1676: 109. (D148). Form 18, South Gaulish (Plate 8). The stamp is partly damaged but reads MACâ&#x20AC;˘RIâ&#x20AC;˘MA, with the MA ligatured in both cases, die 2a of Macer i, NOTS5, 177-180. c.A.D.55-70.

Nidum. Excavations at the Roman Auxiliary Fort 2010 to 2012: Neath, Neath Port Talbot CBC

Plate 8. Context 1676. Form 18, South Gaulish samian stamp

110. (D149). (Not illustrated). Necked jar with an upright beaded rim in hard granular mid grey fabric. For the form see Heywood & Marvell 1992, Fig.7, 11-12. Probably Flavian. 111. (D150). About half the rim of a Gillam 238 mortarium in light buff. The surface has been eroded by soil action but traces of flint grits survive. This is a North Gaulish product, Hartleyâ&#x20AC;&#x2122;s Group II, 1977. Fig.2.1, 3B. Cf. No. 46 above for further discussion. C.A.D.65-10. 112. (D151). (Not illustrated). Mortarium in light red fabric A. The surface is heavily eroded by soil action but what little remains suggests a slightly darker red surface. 113. (D152). Mortarium in pink-red fabric A with traces of a white slip. The trituration grits barely survive but appear to have been grey-white quartz scattered both internally and over the rim. A deliberate imitation of the north Gaulish Gillam 238 mortaria seems likely. Context 2035: 114. (D173). Mortarium of Gillam 238 type (4 fragments); cf. Hartley 1977, Fig.2.1, 3C. Mid 1st to early 2nd century. Context 6068 (fill of gully 6083), with a South Gaulish samian form 27: 115. (D205). Flanged bowl in light orange (9 fragments). The slight curve to the flange suggests an attempt at imitation of the earlier Curle 11 samian bowls. Probably mid/late 1st century. Context 6069 included South Gaulish samian form 27. Context 6078: 116. (D208). (Not illustrated). Form 37, South Gaulish. Six highly abraded fragments, probably all from the same vessel. The basal wreath consisted of chevrons. The main zone is almost totally abraded away but appears to have been a continuous design rather than panels. A separate sherd, probably from the same vessel, shows a fragment of ovolo with a bifid terminal to the tongue over a fine wavy line. Below is a running dog and what may be a pendant motif. Probably c.A.D.80-100.

Nidum. Excavations at the Roman Auxiliary Fort 2010 to 2012: Neath, Neath Port Talbot CBC

117. (D209). Flanged bowl in red-brown to dark grey granular wheel-thrown fabric with a dark grey surface. One of the mid 1st to early 2nd century flanged and carinated bowl series. 118. (D210). (Not illustrated) Mortarium in buff fabric of Gillam 238 type. Mid 1st-early 2nd century. Gully 6083, see 6068 above. Context 6281with a South Gaulish samian form 18/31 (c.A.D.90-110): 119. (D257). (Not illustrated). Two adjoining fragments of jar base in light red fabric with a grey mica dusted surface. Context 6447 included South Gaulish samian form 18 (c.A.D.70-90). Context 6450: 120. (D276). Mortarium or mortar-like bowl in orange-pink Fabric A, possibly originally white slipped. Cf. no.243 from the post-Roman context 6445. Context 6464: 121. (D277). Mortarium of Gillam 238 type; Hartley 1977, Fig.2.1E. Mid 1st â&#x20AC;&#x201C; early 2nd century. Context 6527 with South Gaulish samian forms 27 and 36 was a fragment of Ritterling form 1, a mid 1st century form. Also: 122. (D285). (Not illustrated). Mortarium in a fabric similar to that of Gillam 238. Mid 1st to mid 2nd century. Context 6590: 123. (D295). Flanged rim jar in micaceous grey with a brown core; the rim appears to be a derivative of jars such as Usk Fortress type 11 (Manning 1993, Fig.4) and a mid to late 1st century date seems possible. Context 7015: With a curved rim jar in brown burnt and sooted on the rim: 124. (D312). Flanged dish in Black-burnished ware. 2nd century. Context 7026: 125. (D313). Wall sided mortarium in a light yellow fabric which shares many of the characteristics of some Gillam 238 mortarium fabrics (including a tendency to shatter in the adverse Neath soil conditions). Cf. Holbrook & Bidwell 1991, Fig.85, C/B.2. A North Gaul product? Mid 1st century. Context 7027: 126. (D314). (Not illustrated). Mortarium spout in light orange with quartz trituration grits

Nidum. Excavations at the Roman Auxiliary Fort 2010 to 2012: Neath, Neath Port Talbot CBC

The fort south of the intervallum Context 6623 included South Gaulish samian forms 18, 29 and 35. Phase 1. Initial clearance Context 6954 (fill of channel 6928) with a South Gaulish samian form 27: 127. (D306). Mortarium in bright orange Fabric A with a grey core. The trituration grits included white quartz. Context 6612 with South Gaulish samian forms 18 and 27: Phase 1. 128. (D296). (Not illustrated) Form 29, South Gaulish. An abraded fragment of lower zone shows a zonal arrangement with a festoon containing a stirrup bud below and straight gadroons above.c.A.D.70-85. Context 6659 included South Gaulish samian forms 18 and 27g. Phase 1. Context 6827 (Fill of 6826): 129. (D303). Curved rim jar in light grey fabric with a darker sooted surface externally. 130. (D304). Everted rim jar in light grey Fabric B. Context 6688 included South Gaulish samian form 27. Phase 1a Context 6689 included a South Gaulish samian form 27, burnt grey. Phase 1. Context 6690 with a South Gaulish samian form 27g: Phase 1. 131. (D301). Four joining fragments making up the complete neck of a ring-necked flagon in dark grey fabric with an orange surface. A fragment of handle also survived. The small upper ring suggests a mid-late 1st century date. Context 6545: Phase 1: 132. (D287). (Not illustrated). Four fragments of a very abraded rim of Lyons beaker; cf. no.69 from context 6983 above. Mid 1st century. 133. (D288). (Not illustrated). Sixteen fragments of a handled vessel in granular light cream fabric, probably from the Verulamium region. The handle profile suggests a flagon Context 6585 with South Gaulish samian form 18 or 18/31. Phase 2: 134. (D292). Curved rim jar in grey fabric sooted on the rim externally. Phase 2. Context 6672 with a South Gaulish samian form 18: 135. (D300). (Not illustrated). Mortarium flange in granular light cream fabric from the Verulamium region; cf. Frere 1984, Fig.110, 2647. Mid-late 1st century. Context 6577, Phase 1. 136. (D289). Dish base in fairly sandy orange fabric with mica, quartz and fired clay inclusions and a grey core. There is a slight footring and the lower part of a curving wall, suggesting a rather thick version of a vessel reminiscent of the samian form 18. Perhaps midlate 1st century. 33

Nidum. Excavations at the Roman Auxiliary Fort 2010 to 2012: Neath, Neath Port Talbot CBC

Context 6907 included South Gaulish samian forms 18 and 27. Phase 1. Context 6908 included a South Gaulish samian form 18 (c.A.D.70-90). Context 6070 with South Gaulish samian forms 18, 27 and 37. Phase 2: 137. (D206). Form 37, Central Gaulish (Plate 9). A large double bordered medallion contains the scarf dancer O.361A, within a plain double-bordered medallion; cf. Rogers 1999, Pl.12.3 for use of this motif within a different medallion. Probably c.A.D.120-150.

Plate 9. Context 6070. Form 37, Central Gaulish samian

Context 6071 with South Gaulish samian forms 27, 29 and Curle 11. Phase 2: 138. (D207). Mortarium in light orange. The filler contains mixed grit with much quartz. Larger pieces appear on the surface probably intentionally. With a worn mortarium in smooth Fabric A. Context 6478=6432. Phase 2: 139. (D279). Large necked jar in sandy dull light red with a grey surface. Cf Manning 1993, Fig.107, 6.1 although the Neath example is more likely to be 1st century than later. Context 6557 (fill of channel 6746) included a South Gaulish samian form18 (c.A.D.70-90). Context 6045 included South Gaulish samian forms 18 or 18/31, 29 and 37. Phase 2: Context 6053, with a flange from a Verulamium mortarium. Phase 2: 140. (D199). Curved rim jar in grey South Wales reduced ware. 141. (D200) (Not illustrated). Curving flange of a mortarium in granular cream from the Verulamium region. Mid/late 1st to early 2nd century. Context 6437. Phase 2: 34

Nidum. Excavations at the Roman Auxiliary Fort 2010 to 2012: Neath, Neath Port Talbot CBC

With South Gaulish samian form 18 (c.A.D.70-90): 142. (D272). Everted rim jar in hark grey fabric with a red-brown core. Context 6521 with a jar in South Wales Reduced Fabric similar to Usk fortress type 11: 143. (D284). Curved rim jar in light grey Fabric B with a darker surface; cf. Manning 1993, Fig.110, 27.2 but not necessarily as late as the 2nd century date assigned there. Context 6537, with a Les Martres samian bowl form 37 (c.A.D.100-120), and a small fragment of North Gaulish rough cast beaker, probably late 1st century in a South Wales context. Phase 2: 144. (D286). (Not illustrated). Form 37, Central Gaulish (possibly Les Martres-de-Veyre), two joining fragments. The moulded surface has been almost totally removed by soil action but the ‘shadow’ of a winding scroll design remains, the lower lobe of which contain a large ‘ivy’ leaf, perhaps Rogers 1974, J1, and a smaller leaf. A design similar to Terrisse 1968, Pl.41, 261 seems possible. Probably c.A.D.100-120. Context 6049. Phase 2: The context included South Gaulish samian form 29, and a basal fragment of North Gaulish roughcast beaker; probably later 1st century in a South Wales context. Context 6621. Phase 2: 145. (D297). Necked jar in bright orange fabric with mixed quartz inclusions and a grey surface externally. Context 6790. Phase 2: 146. (D302). (Not illustrated). Form 27, South Gaulish, six fragments. The stamp reads IMV..M Comment The Les Martres samian from context 6537 in particular places this phase in the early 2nd century. Context 6525. Phase 2: The context included South Gaulish samian form 37. Other contexts from the fort interior in numerical order of context Context 6092: 147. (D211). Flanged bowl in brown with a dark grey surface, probably burnt Blackburnished ware. There are traces of lattice decoration on the wall. Possibly 2nd century. Context 6161: 148. (D216). Curved rim jar in light grey Fabric B; cf. Manning 1993, Fig. 110, 26.2 where a 2nd century date is suggested although, at Neath, this seems more likely to be a variant on a common 1st century rim. 35

Nidum. Excavations at the Roman Auxiliary Fort 2010 to 2012: Neath, Neath Port Talbot CBC

Context 6444 with a South Gaulish samian form 27: 149. (D274). Curved rim jar in Fabric B. Context 6567: The context included South Gaulish samian forms 18 and 37. Context 6580: 150. (D290). Form 37, South Gaulish (Plate 10). Panel decoration shows, to the left, a triplebordered medallion containing a cupid, probably O.436 and with looped tendrils with lanceolate leaf terminals in the lower spandrels. To the right is a small panel with running dog. Below is a basal frieze of S-shaped gadroons. The cupid is used by Vitalis (cf. Knorr 1919, Taf.83, 2) who also uses the triple medallion, but his output of decorated bowls were mainly form 29. A broadly similar design is illustrated by Hermet (Pl.82, 1) and included the gadroons. c.A.D.80100.

Plate 10. Context 6580. Form 37 South Gaulish samian

Context 6582: 151. (D291). (Not illustrated). Flanged bowl or dish in Black-burnished ware burnt light grey to orange. Context 6652 with two examples of South Gaulish samian form 18 (c.A.D.70-90): 152. (D298). (Not illustrated). Four abraded fragments of Lyons beaker; cf. nos. 69 & 132 from contexts 6545 and 6983, also Greene 1979, Fig.13, 5. Mid 1st century. 153. (D299). Curved rim jar in light grey Fabric B. Probably from a jar such as Nash-Williams 1929, Fig.29, 16. (Mid-late 1st century). 36

Nidum. Excavations at the Roman Auxiliary Fort 2010 to 2012: Neath, Neath Port Talbot CBC

Context 6678 included South Gaulish samian form 18 or 18/31. Context 6686 included South Gaulish samian form 27. Context 6904: 154. (D305). Mortarium in orange-pink Fabric A with a grey core and traces of a white slip. Context 6960: 155. (D307). Everted rim jar in light grey with signs of a darker surface. Context 6971: 156. (D308). Small ring-necked flagon in light orange fabric. The small upper ring suggests a mid-late 1st century date. Comment Although the material from Phases 1 and 2 is insufficient to provide clear dates for the clearance and construction of the fort, it is noticeable that vessels of the mid to mid/late first century predominate. There is certainly nothing which would not support a foundation date in the mid 70s and nothing which need be later than the mid-Flavian period. The extra-mural areas The Western Vicus Context 6201: 157. (D229). Six fragments of mortar-like bowl in buff fabric with a grey core Context 6183 (cleaning of feature 6225) with South Gaulish samian forms 27 and 37: 158. (D225). Curved rim jar in light grey Fabric B. One of at least 3 curved rim jars. This example is unusually thin walled. 159. (D226). Flanged dish in light grey South Wales Reduced Fabric. Feature, context 6227 with South Gaulish samian form Curle 15, a South Gaulish sherd perhaps adapted as a counter and an everted rim jar in grey: 160. (D230). Curved rim jar in red-brown with a dark grey surface. 161. (D231). Flanged bowl in light grey Fabric B. One of the mid 1st to early 2nd century flanged and carinated series. 162. (D232). Mortarium in granular light cream from the Verulamium region. There is faint evidence of a two line stamp, with the first line starting M and the second E, a stamp of Matugenus, cf. Suggett 1954, Fig.6, 2-3; Frere 1984, Fig.118, 85. Late 1st-early 2nd century. Feature, context 6229 with South Gaulish samian forms 18, 18R and 37 (probably c.90-110), a Les Martres 27, a probable beaker fragment in bright orange with traces of mica dusting: 163. (D233). Form 37, Central Gaulish (probably Les Martres de Veyre), Plate 11. Three fragments, two decorated and joining. The ovolo frieze has been completely removed in 37

Nidum. Excavations at the Roman Auxiliary Fort 2010 to 2012: Neath, Neath Port Talbot CBC

finishing. The remaining zonal decoration shows a frieze of vine scrolls, Rogers 1974, M18 used by X-12 and others over one of festoons (perhaps Rogers 1974, F74, also used by X-12, cf. S&S, Pl.41, 482), much abraded, containing what could be a beaded circle as in S&S, Pl.41, 477 and separated by a pendant. c.A.D.100-125.

Plate 11. Context 6229. Form 37, Central Gaulish (probably Les Martres de Veyre)

164. (D234). Ring neck flagon in light grey fabric A. A local version of a common mid-late 1st century type. 165. (D235). Necked jar in light grey fabric with traces of a darker surface; cf. Manning 1993, Fig.107, 7.3 although an earlier date seems likely. A grey handle from this context may imply a handled jar. 166. (D236). One of at least 3 curved rim jars in light grey. There are faint traces of rouletting below the shoulder. 167. (D237). Everted rim jar in light grey. 168. (D238). Flanged bowl in light grey with a darker surface. 169. (D239). Dish with wedge-shaped rim in light grey with a darker surface. Feature, context 6230: 170. (D240). Seven fragments of flanged bowl or dish in mid grey fabric. Feature, context 6257: 171. (D243). Curved rim jar in a granular grey fabric. Comment The pottery from the western vicus exactly mirrors that from the early forts and there is no reason to suppose that it was occupied outside the period of fort occupation. 38

Nidum. Excavations at the Roman Auxiliary Fort 2010 to 2012: Neath, Neath Port Talbot CBC

The northern extra-mural area Ditches: Context 1470 with a South Gaulish samian form 18 (c.A.D.70-90). Phase 1: 172. (D133). (Not illustrated). Lid (or less probably a shallow bowl) in light grey South Wales Reduced ware; similar to, or the same vessel as, D121 from context 1338. Context 1470 (Fill of 1480, Ditch/drain) Phase1. Included South Gaulish samian forms 36 and 37, and several fragments of a mortarium in grey fabric with an orange surface showing the scar of the spout. Context 1076 with South Gaulish samian forms 18R and 27and a Les Martres form 37. Phase 3:

Plate 12. Context 1076. Form 37, South Gaulish samian

173. (D53). Form 37, South Gaulish (Plate 12). The fragment shows the very bottom of a very abraded ovolo frieze, perhaps including a trifid terminal over the erotic group O.A, used by a variety of South Gaulish potters. Probably c.A.D.80-110. 174. (D54). (Not illustrated). Form 37, Central Gaulish. A small fragment shows an eroded ovolo with rosette-ended tongue over a wavy line. Below is part of the Apollo. O.92 used by a variety of Lezoux potters. Perhaps c.A.D.120-150. 175. (D56). Curved rim jar in reddish brown shading to grey and with a dark grey surface. The flaring profile could denote a later Roman vessel. 176. (D57). (Not illustrated). Jar in Black-burnished ware; probably Gillam 1976, nos.1-2 (early to mid 2nd century). Context 1469 with South Gaulish samian form 29. Phase 3:

Nidum. Excavations at the Roman Auxiliary Fort 2010 to 2012: Neath, Neath Port Talbot CBC

177. (D130). Rim in red-brown fabric with a dark grey surface. Probably from a lid-seated jar, although a flange is not impossible. Context 1077 from the Phase 2 horizon included South Gaulish samian form 18 and 37 and a samian flanged bowl, either Ritterling 12 or Curle 11, along with a Les Martres form 37. Context 1320: Phase 2. The context included South Gaulish samian form 37 and a Les Martres bowl form 37 (c.A.D.100-120), a mortarium rim in Fabric A with traces of a white slip and a spout fragment from a further example. Also: 178. (D112). (Not illustrated). Necked jar in light orange with a grey core. The filler shows a tendency to leach and may have been calcitic. The fabric has deteriorated due to soil action, but this is probably Severn Valley Ware; cf. Manning 1993, Fig.132, 3.3. Mid 1st to 2nd century. Context 1333 (over 1496). Phase 2: The context included South Gaulish samian forms 18 and 37. Also: 179. (D113). Jar in dark grey granular fabric. The vessel appears to be wheel-thrown but is otherwise similar in form to early Black-burnished jar types (e.g. Manning 1993, Fig.27, 3) and a mid to late 1st century date may be suggested. 180. (D114). (Not illustrated) Rectangular foot c.2 x 1.5 cms where it rested on the ground and c.1.5 to 2 cms in height in a light grey granular fabric with a clinker-like feel. It appears to have supported a bowl-like vessel and may have served as a crucible. Context 1338 with South Gaulish samian forms 18, 18R, 27, 29 and 37. Phase 2: 181. (D115). (Not illustrated). Form 29, South Gaulish. The stamp has been almost totally removed in finishing but a fragment of what is probably O remains. The form largely went out of production by the mid 80s so a date c.65-85 seems assured. 182. (D116). Small fragment of Dressel 20, South Spanish olive oil amphora rim, MartinKilcher type 15 (Peacock & Williams 1986, Fig.65). Mid to mid-late 1st century. 183. (D117). Everted rim jar in light grey with a darker heavily sooted rim externally. 184. (D118). Everted rim jar in light grey South Wales Reduced fabric with a groove, perhaps above decoration near the girth; cf. Manning 1993, Fig.109, 16.5 for the general type. Mid 1st to early 2nd century/ 185. (D119). Curved rim jar in light grey fabric. One of two curved rim jars. With fragments of Gillam 238 mortarium and the spout of a cream mortarium probably also of Gillam 238 type: 186. (D120). Mortarium in orange Fabric A, probably originally with a white slip. 187. (D121). Lid in light grey South Wales Reduced fabric. Context 1366 included South Gaulish samian form 37. Phase 2. 40

Nidum. Excavations at the Roman Auxiliary Fort 2010 to 2012: Neath, Neath Port Talbot CBC

Context 1494 with South Gaulish samian forms 18R or 18/31R, 29 and 37: 188. (D134). Lid-seated jar in light grey South Wales Reduced fabric. The entire rim seems to be present. 189. (D135). Jar rim in Black-burnished ware; the degree of flare suggests a late 3rd or 4th century piece, cf. Gillam 1976, no.12. The context also has part of the sunken handle of a Black burnished ware jar as Gillam 1976, no.15 (probably mid 1st to mid 2nd century in a South Wales context). 190. (D136). Jar in mid grey fabric shading to reddish brown on the interior surface. The form resembles that of late 3rd and 4th century jars in Black-burnished ware (cf. no.189 above) and this could be of similar date. 191. (D137). Mortar-like bowl in dull pink Fabric A with a grey core. Road context 1061. Phase 3: With South Gaulish samian form 18 (2): 192. (D43). (Not illustrated). Mortarium in cream of Gillam 238 type. This is Hartleyâ&#x20AC;&#x2122;s Group II (1977, 9-11 & Fig.2.1, no.3; 1998, 201-6; Hartley & Tomber 2006, 22-4), c.A.D.65110. Context 1495 clay deposit cut by pit 1486 (which itself produced no diagnostic pottery). Phase 3. 193. (D138). Curved rim jar in light grey South Wales Reduced fabric; cf. Manning 1993, Fig.110, 26.2 (2nd century). Related to building discovered 1993: Context 1111 194. (D66). (Not illustrated). Plate floor, probably form 18, South Gaulish. The surface has been eroded. Part of a badly worn stamp remains, perhaps ]IRâ&#x20AC;Ś.I. Probably c.A.D.70-90. 195. (D67). Flanged bowl in light grey Fabric B. The few traces of the original surface suggest that it may have been in mid grey. Cf. Heywood & Marvell 1992, Fig.34, 20 for the form. Context 1388. Phase 3: 196. (D125). Mortarium of Gillam 238 type. The rectangular panel for a stamp is visible but the stamp itself is much abraded. It may have been of two lines; cf. Holbrook & Bidwell 1991, Fig.81, C26 for a similar piece stamped by Q.Valerius Veranius. Context 1073 (a possible post-hole): The context included three fragments of a Les Martres samian bowl form 37 (c.A.D.100-120) with lead rivet. Context 1081. The posthole included South Gaulish samian forms 18 and 27. Context 1270: 41

Nidum. Excavations at the Roman Auxiliary Fort 2010 to 2012: Neath, Neath Port Talbot CBC

197. (D108). Straight sided dish in grey fabric with a darker surface; probably South Wales Reduced fabric. Context 1339: The context included South Gaulish samian forms 27 and 37. Also: 198. (D122). Jar in light grey fabric. 199. (D123). Mortar-like bowl in pink with a thick black core Other Extra-Mural Features Contexts are listed in numerical order Context 1065 (clay deposit): With South Gaulish samian forms 27 and 29, and a burnt fragment of North Gaulish roughcast beaker: 200. (D45). (Not illustrated). Form 37, probably Les Martres-de-Veyre. The decorated surface has almost completely abraded although an ovolo frieze over a (?)bead row is visible, but the main body of the design has almost completely vanished. A freestyle or scroll design seems likely as there is no evidence of panel divisions. c.A.D.100-120. 201. (D46). Everted rim jar in light grey South Wales Reduced fabric. One of the mid 1st to early 2nd century series. 202. (D47). Curved rim jar in grey South Wales Reduced fabric; with at least one other curved rim jar. 203. (D48). Jar in grey South Wales Reduced fabric; perhaps as Heywood & Marvell 1992, Fig.35, 26. Late 1st to early 2nd century. 204. Not used Context 1075: With South Gaulish samian form 36 and 37, a Les Martres bowl, perhaps form 29 (c.A.D.100120), an abraded mid grey flanged bowl fragment and a burnt curved rim jar: 205. (D50). Bead rim bowl in light orange. Probably intended to be reminiscent of the samian form 37. Probably Flavian. 206. (D51). Mortarium in light orange Fabric A. The rim form suggests a mid 1st to early 2nd century date. 207. (D52).

Lid in grey fabric.

Context 1096 (an irregular stone and pebble deposit): With a South Gaulish samian cup form 27, and a curved rim jar in grey: 208. (D63). Eighteen fragments of flanged bowl in red-brown sandy fabric (perhaps an oxidised version of South Wales reduced fabric). The form closely resembles the samian form Ritterling 12 and the earliest form Curle 11 (though without the characteristic trailed decoration 42

Nidum. Excavations at the Roman Auxiliary Fort 2010 to 2012: Neath, Neath Port Talbot CBC

of the latter); cf. Heywood & Marvell 1992, Fig.7, 3 and no.263 from context 1003 below. A date c.A.D.60-80 seems likely. Context 1361 (clay deposit) included South Gaulish samian form 29, a mortarium in pink fabric A (soil action has entirely eroded the surface). Context 1441 (loam and stone) with a South Gaulish samian form 27g: 209. (D127). (Not illustrated) Eight fragments along with smaller sherds of a Gillam 238 type mortarium, Hartley 1977, Fig.2.1, 3D. c. A.D.60-110. Context 1550 (clay deposit) included South Gaulish samian form 18. Context 6255 (charcoal rich clay) included South Gaulish samian form33 Context 6262 (clay deposit): 210. (D253). Flanged bowl in light grey fabric with traces of a darker surface. One of the few flanged bowls from the excavation with evidence of a reeded rim. Context 6264 (clay deposit): 211. (D254). Curved rim jar in light grey fabric. Context 6265 (clay deposit) included South Gaulish samian form 37. Gravel extraction pits Contexts are listed in numerical order Context 1046 included a South Gaulish samian bowl form 37 and South Gaulish cup sherd, a mortar-like bowl in light orange (probably late 1st to early 2nd century) and an orange bowl with wedge-shaped flange. Also: 212. (D39). Jar in Black-burnished ware with zig-zag decoration on the neck; Gillam 1976, no.3 ( mid-late 2nd century) 213. (D40). Jar in Black-burnished ware. The angle of the neck suggests a vessel broadly similar to Gillam 1976, nos.11 (late 3rd-early 4th century) although one would prefer a larger piece and some of the lattice decoration for certainty. 214. (D41). (Not illustrated) Mortarium in granular off white with a grey core (9 fragments); from the Verulamium region. The incomplete rim suggests a strongly curving flange as Frere 1972, Fig.106, 225 or Fig.110, 372 (the date for which is suggested as c.A.D.60-100 in Frere 1984, 269-270). 215. (D42). Flanged dish or, more probably bowl, in Black-burnished ware; cf.Gillam 1976, no.35. 2nd century. One of 3 rim sherds probably from the same vessel. Context 3001 included South Gaulish samian form 18 (c.A.D.70-90). Context 5001 with South Gaulish samian forms 18, 29, 35 and 37, three curved rim jars in grey and an eroded mortarium in orange and an abraded Verulamium mortarium: 216. (D176). Curved rim jar in grey fabric heavily sooted on the rim. 43

Nidum. Excavations at the Roman Auxiliary Fort 2010 to 2012: Neath, Neath Port Talbot CBC

217. (D177). Curved rim jar in grey with dark grey surface. One of two similar curved rim jars. 218. (D178). (Not illustrated) Bead rim jar in light grey fabric. 219. (D179). Jar in Black-burnished ware. 220. (D180). Flange rim jar in dark grey with an orange surface. 221. (D181). Large flange-rim jar in light orange Fabric A. 222. (D182). Flanged bowl in orange fabric. 223. (D183). Flanged bowl in grey with a sooted surface 224. (D184). Flanged dish in Black-burnished ware; Gillam 1976, no.56. Early-mid 2nd century. 225. (D185). Small mortar-like bowl in pink with a dark grey core. Context 5002: 226. (D186). Sixteen fragments, many joining, of a flanged bowl in Black-burnished ware burnt buff on the surface in places; 2nd century. Context 5005 with South Gaulish samian form 18: 227. (D187). Form 29, South Gaulish (Plate 13). Two small joining fragments of lower zone, show a winding scroll. The extant lower lobe contains looping tendrils around a central inverted arrow-like motif, Knorr 1919, Tb.7, centre, Webster 1987, C25 recorded with stamps reading ‘Of.Cotoi’ and ‘Iust’. c.A.D.65-85.

Plate 13. Context 5005. Form 29 South Gaulish samian

228. (D188). Flanged bowl in grey South Wales Grey fabric. 44

Nidum. Excavations at the Roman Auxiliary Fort 2010 to 2012: Neath, Neath Port Talbot CBC

Context 5017 with South Gaulish samian forms 18 and 29 (c.A.D.65-90): 229. (D189). Mortarium in pale cream fabric. Context 5086 included a large basal fragment of a mortarium in micaceous off white with a pink core. The few trituration grits which survive appear to be quartz. The fabric has red-brown inclusions (? fired clay). The source is unknown. Later Roman features Contexts are listed in numerical order Context 1062 With a small eroded fragment of dish in light grey with a wedge-shaped rim (cf. D110 from context 1293). 230. (D44).

(Not illustrated) Lid in grey fabric with a darker surface.

Context 1066: 231. (D49). Mortarium in buff fabric with traces of flint-like grits on the rim and internally; Gillam 238 type, cf. no.D43 from context 1061 above. C.A.D.65-110. Context 1316 with a South Gaulish samian form 27: 232. (D111). Everted rim jar with possible lid seating in light grey South Wales Reduced fabric sooted on the rim externally; cf. Manning 1993, Fig.109, type 14. Mid 1st to early 2nd century. Context 1473 with South Gaulish samian forms 18, 35 and 37: 233. (D131). Everted rim jar in light grey. 234. (D132). Fourteen fragments (and many other scraps) from a flanged (and probably carinated bowl) in light grey South Wales Reduced fabric with a darker surface. This simple form of flange seems to be a local preference within the common mid 1st to early 2nd century flanged bowl series. Context 6258 with South Gaulish samian forms 18 and 37: 235. (D244). Ring-neck flagon in pink Fabric A. 236. (D245). With four other curved rim jars: Jar in light grey with darker surface; cf. Manning 1993, Fig.110, 26.23 (2nd century). 237. (D246). Everted rim jar in light grey Fabric A. Soil action has eroded all the surface. 238. (D247). Bowl base in light grey with a smooth surface. Probably from a hemispherical bowl reminiscent of the samian form 37. Probably mid-late 1st century. 239. (D248). Flanged bowl in smooth orange fabric shading to light grey in places. 240. (D249). Lid or shallow bowl in grey fabric. 241. (D250). Lid in Black-burnished ware; cf. Wallace & Webster 1989, Fig.2, 10. 1st and 2nd century. 45

Nidum. Excavations at the Roman Auxiliary Fort 2010 to 2012: Neath, Neath Port Talbot CBC

Context 6443: 242.(D273). Mortarium or mortar-like bowl in pink Fabric A with traces of white slip (20 fragments). The form is probably intended to be reminiscent of the common Gillam 238 mortaria and a mid-late 1st century date seems likely. Context 6445: 243. (D275). Mortarium or mortar-like bowl in orange-pink Fabric A, possibly originally white slipped. Context 6275: 244. (D256). Necked jar in light grey Fabric B. Context 6286 with a South Gaulish samian form 18: 245. (D258). Necked jar in light grey. 246. (D259). Curved rim jar in light grey South Wales Reduced fabric. One of at least 5 curved rim jars. 247. (D260). Curved rim jar in light grey fabric (about half the rim). 248. (D261). Everted rim jar in light grey with a darker surface; cf. Manning 1993, Fig.109, 16.4. Late 1st to early 2nd century. With at least 3 other everted rim jars. 249. (D262). Everted rim jar in light grey fabric with lid seating. 250. (D263). Jar in fawn grey with grey surface. The groove on the rim is presumably for a lid. 251. (D264). Flanged bowl in light grey fabric sooted externally. One of at least five flanged bowls. 252. (D265). Flanged bowl or dish with an internal bead perhaps to reduce spillage. 253. (D266). Small flanged bowl in light grey fabric. 254. (D267). (Not illustrated) Lid in granular dark pink with a dark grey surface; Heywood & Marvell 1992, Fig.7, 13. 255. (D268). Lid or shallow dish in grey-buff to light grey. 256. (D269). Lid in grey fabric. Context 6259 with a Les Martres samian form 27 (c.A.D.100-150): 257. (D251). Flanged bowl or dish in Black-burnished ware. Probably 2nd century. Post Roman contexts Roman vessels from post-Roman contexts all reflect the range of forms and sources shown from stratified contexts. A small number of vessels have been selected for illustration because of their intrinsic interest. A fuller list is to be found in the archive.

Nidum. Excavations at the Roman Auxiliary Fort 2010 to 2012: Neath, Neath Port Talbot CBC

Context 1086 with a South Gaulish samian form 27 and 29: 258. (D58). Flanged bowl in light grey fabric [Fabric B?] (3 fragments). The curved junction between wall and flange appears to be a popular local variant on the mid 1st to early 2nd century flanged and carinated bowl. Context 6240 with Les Martres samian form 37 (c.A.D.100-120), a necked jar, an everted and a curved rim jar in grey, and a burnt mortarium or mortar-like bowl in Fabric A and a mortarium of Bushe-Foxe type 22-30: 259. (D242). Dish in light grey fabric with traces of a darker surface. The form, is derived from Gallo-Belgic dishes such as Camulodunum (Hawkes & Hull 1947, Fig.47) type 8. Probably mid to late 1st century in this context. Context 1002 included Central Gaulish samian forms 18/31 and 37, a Les Martres bowl, South Gaulish samian form 18 (3), 27, 29 (2), 37 (2), a small fragment of North Gaulish roughcast beaker, two flanged bowls in grey, a lid in brown with a dark grey surface and: 260. (D22). Bead rim bowl in Black-burnished ware. The form is related to the mainly 2nd century forms, Gillam 1976, nos.50-52, but our example probably lies both typologically and chronologically between these and the so-called â&#x20AC;&#x2DC;War Cemetery Bowlsâ&#x20AC;&#x2122; (Wheeler 1943, Fig.72, nos.171-181). Probably mid-late 1st century. 261. (D23). Mortarium in buff fabric with the vestiges of flint-like grits internally and over the rim. Form Bushe-Fox 22-30 from N.Gaul; cf. Manning 1993, 400-1, type imp.18, Hartley & Tomber 2006, 24. c.A.D.70-150. Context 1032 (a modern pit): 262. (D38). Bowl base in light grey Fabric B with traces of a darker surface. The interior shows a pronounced omphalos, suggesting a local imitation of a Terra Nigra form, Greene 1979, Fig.47. Flavian. Context 1003 (topsoil over Via Sagularis) with a South Gaulish samian cup form 35: 263. (D24). Flanged bowl in light orange with a grey core; probably related to the bowls which imitate the samian form Ritterling 12 as no.208 from Context 1096 above. Probably Flavian. Unstratified 264. (D193). (Unstratified) Form 37, Central Gaulish (Plate 14). The ovolo is Rogers 1974, B231 (used mainly by Cinnamus), over a bead row and a zone of vine leaves, Rogers 1974, H24. Also in the field is an unidentified vine tendril partly overlaid by the leaves. The bottom of the design is both eroded and has been scuffed in finishing. c.A.D.140-170.

Nidum. Excavations at the Roman Auxiliary Fort 2010 to 2012: Neath, Neath Port Talbot CBC

Plate 14. Unstratified. Form 37, Central Gaulish samian

Plate 15. Unstratified. Form 18, South Gaulish samian stamp

265. (D194). Form 18, South Gaulish (Plate 15). The floor of the plate with the stamp OFRONTI, die 16a of Frontinus of La Graufesenque, NOTS4, 101-113. c.A.D.70-95.

Nidum. Excavations at the Roman Auxiliary Fort 2010 to 2012: Neath, Neath Port Talbot CBC

2. The post-Roman pottery By Joyce Compton All of the pottery from Sites 677, 716 and 740 has been recorded by sherd count and weight, in grams, by context, and the data entered onto an Excel spreadsheet for ease of manipulation. The spreadsheet forms part of the archive. The pottery recorded in ninety-two contexts across the four excavated areas amounts to a total of 1,268 sherds, weighing 15.5kg, and includes a small quantity of unstratified material. In general, the assemblage is fragmentary (average sherd weight 10.4g) and in a poor and abraded condition. Almost all of the pottery is of later post-medieval and modern date, although earlier types are also present. By far the largest proportion of the pottery by weight was recorded in Site 677 contexts; the pottery from Site 740 (representing watching-brief material) amounted to only a small amount. A small piece of medieval glazed ridge tile, now fragmented, was recorded in context 1002 (Site 677). Most of the fragments are abraded and some joins are uncertain. Two sherds were examined by Alice Forward who suggests that the tile is in a transitional Vale Fabric, perhaps indicating a later medieval date. The assemblage falls into three main pottery types, modern ceramics, local red earthenware and North Devon gravel-tempered ware, in slightly different proportions at each site. These pottery types date largely to the 19th century, perhaps into the early 20th, with earlier pieces noted within the two coarseware assemblages. A range of earlier ceramics, West Country slipwares, Staffordshire slipwares, salt-glazed stonewares and tin-glazed earthenware, was also recorded in small amounts. Together, these four earlier post-medieval pottery types account for almost 10% by weight of the assemblage. Ten sherds (70g) of medieval pottery were also recorded (from Site 677). A number of abraded and formless sherds (10 sherds, weighing 109g, from Sites 677 and 740) are undated; these are as likely to be Roman as post-medieval, since most have no surfaces or closely datable features. Modern ceramics These form the largest component, mainly comprising white earthenwares, nearly all of which are blue transfer-printed and include large numbers of plate sherds. Makersâ&#x20AC;&#x2122; marks consist of partial stamps (mostly individual letters) or symbols, and are not identifiable. There is surprisingly little Victorian stoneware, noted in just eleven contexts, although bottles, inkwells and large flagons or cisterns are all represented. Other vessel types present in small quantities are teapots, kitchen-ware bowls and Mocha ware mugs and jugs. The modern ceramics are in a poor condition, with an average sherd weight of 5.8g. Local red earthenware Much of this pottery type consists of large pans and bowls, usually thick-walled with internal glazes. These thick walls and bases account for the relatively high average sherd weight (18.1g) for this assemblage component. A slip-decorated sherd from Site 716 is pierced to form a strainer-plate. Other slip-decorated sherds derive from platters and dishes; these may indicate an earlier post-medieval date. A pipkin rim sherd from Site 677 may also be from an earlier post-medieval vessel. 49

Nidum. Excavations at the Roman Auxiliary Fort 2010 to 2012: Neath, Neath Port Talbot CBC

North Devon gravel-tempered ware Large pans and bowls also occur in this fabric, their thick walls and bases accounting for an average sherd weight of 19.3g. Although most of this fabric is heavily tempered with quartz, giving rise to its name, some sherds appear to have no temper (North Devon gravel free). Itâ&#x20AC;&#x2122;s possible that this fabric is reserved for finer vessels such as jugs. A pipkin-type handle was observed; this may derive from an earlier post-medieval vessel. Cream-coloured earthenware A small quantity of creamware was recorded. This pottery type is in the poorest condition, with average sherd weight of 3.5g, and showing much spalling throughout. Some sherds are the true warm cream-coloured earthenware which could indicate a late 18th century date for some of this assemblage. Sherds of interest include a plate base sherd with an asterisk-shaped impressed stamp beneath and the base from a coffee can, both from Site 677. Salt-glazed stoneware Three-quarters of this small assemblage comprises 18th-century fine white Staffordshire stonewares. Forms include mugs and tankards and plates with â&#x20AC;&#x2DC;barley-and-basketâ&#x20AC;&#x2122; moulded designs on the rim. Earlier stoneware forms include a tankard rim sherd with manganesecoloured reeding around the rim and two sherds of grey Westerwald stoneware, with cobalt decoration, and of possible 17th-century date. Staffordshire slipwares These slipwares were first produced in the 17th century but their main production period was during the 18th, with press-moulded dishes continuing into the 19th century. The source is usually given as Staffordshire but these slipwares were also made in Bristol. Vessels have a pale buff fabric, covered with either brown or yellow glaze; the latter commonly has applied brown slip which is trailed and/or combed producing feathered patterns. Tankards, posset pots and press-moulded dishes were all noted in the assemblage. West Country slipwares This pottery type has a distinctive inclusion-free fabric, pale red-orange with a pale grey core, and nearly always white-slipped beneath the glaze. Much of the pottery has incised designs and patterns (Sgraffito), which very often exposes the darker body through the design. Common vessel types are platters and jugs, both of which are in evidence here, although only small amounts of slipware were recorded. The Nether Stowey area in Somerset is a known production site for this ware during the 17th to mid-18th centuries, although a Glamorgan source cannot be ruled out. Other early post-medieval pottery types amounted to eight sherds (63g) of tin-glazed earthenware and three (10g) of blackware (Cistercian ware). Both types were common during the 17th and early 18th centuries. Medieval pottery Ten sherds (70g) of medieval pottery were recorded, all from Site 677. Both cooking pots and table wares are represented, although most sherds are small and abraded. Most will be locally made, although Redcliffe in Bristol is a probable source for at least one glazed sherd, and Vale 50

Nidum. Excavations at the Roman Auxiliary Fort 2010 to 2012: Neath, Neath Port Talbot CBC

Fabric was identified by Alice Forward. The pottery has not been closely dated, due to lack of forms and the small sherd size, but a broad 13th/14th century date is likely. Conclusions Throughout the post-medieval period, pottery was sourced in Staffordshire, Bristol and the West Country, as well as locally, although no makersâ&#x20AC;&#x2122; marks were noted. A full range of pottery types was recorded, although assemblages were largely composed of disparate body sherds. Early post-medieval pottery from Somerset and the Barnstaple/Bideford area demonstrates the importance of the Bristol Channel in the movement of goods. The sherd of Redcliffe ware shows that this was true in the medieval period also and the presence of early German stoneware indicates trading further afield. The condition of the pottery, its small sherdage and the low incidence of rim sherds perhaps indicates a lack of primary deposits. In a discussion of the assemblage from excavations on the gas pipeline as it crossed the fort, it was suggested that the bulk of the pottery arrived via nightsoil manuring from elsewhere. If so, this occurred over an extended period of time since the post-medieval pottery spans the 17th to 19th centuries. The lack of certainly-identifiable 20thcentury pottery might confirm this theory; as improved sanitary conditions would have halted the practice. Whether the medieval pottery appeared via the same mechanism is debatable but entirely possible. The presence of the medieval pottery is interesting, as the castle and medieval town lie on the other side of the river Neath, and Neath Abbey, to the west, lies across the river Clydach.

3. The clay pipes By Charlotte James A total of 311 fragments of clay pipe were recovered from the three sites at Dwr y Felin School, Neath. The pipes are from 36 contexts and three unstratified groups, the largest collection (210 pieces) being from context 1002: the greater part of this assemblage consisted of undiagnostic pipe stem fragments. Applying the rough guide to dating for pipe stem assemblages from a single contexts using the bore diameter of the stem, here most frequently 2.3mm, suggests a late 17th early 18th century date for the collection in this context, which also included three pipe bowls with large flat heels, of mid-late 17th century date. The remainder of the assemblage comprised twenty undiagnostic bowl fragments, three 17th century bowls, one late 17th-early 18th century, two 18th century and nine 19th century bowls. Three marked pieces were recorded, two of which had the markings of a known maker William Pardoe and another badly damaged stamp possibly that of Richard Nooney.

Nidum. Excavations at the Roman Auxiliary Fort 2010 to 2012: Neath, Neath Port Talbot CBC

The glass (Figure 46, Plates 16-22)

By Rowena Hart Summary A total of 1152 glass fragments were recovered from three sites at Dwr y Felin, Neath (site numbers 674, 677, 716). This assemblage included 288 fragments of post-medieval or modern date leaving 864 fragments of Roman glass. The post-medieval and modern glass is not considered in this report. The total of Roman fragments included 23 counters, 39 beads and three probable fragments of window glass with the remainder comprising vessel fragments. The assemblage from site 740 comprised only post-medieval and modern fragments and is not further considered here. The Roman glass recovered from the three sites covers the period from the second half of the first century AD to the end of the second century. Some fragments of polychrome glass and blue green pillar moulded bowls highlight first century occupation while blue green bottles, jugs, melon beads and gaming counters represent both the first and second centuries. The entire assemblage from the three sites, when considered together, forms the largest recovered from excavations in Neath. Several of the vessel or object types recovered are represented by more than one fragment. There is a relatively high proportion of diagnostic pieces revealing a narrow range of vessels and objects in use. Introduction A total of 864 fragments of Roman glass had been recovered from three sites and 244 excavated contexts with 109 unstratified fragments. Aims and Objectives The aim of the examination of the glass was to determine the nature of the glass in order to provide dating and interpretive information about the site and its contexts through the glass recovered. The report is intended to conform to the Standard and guidance for the collection, documentation, conservation and research of archaeological materials (CIfA, Published December 2014). Methodology The glass was visually inspected, with each fragment assessed for colour, form/typology, other diagnostic information and date. This information was then analysed in order to provide an interpretation of the assemblage to provide context and site information. Results The Roman glass assemblage is described below. Site 674. A total of 20 fragments of Roman glass vessels were recovered from ten contexts and a single fragment recovered as an unstratified find. Most were small undiagnostic fragments of vessel glass. 52

Nidum. Excavations at the Roman Auxiliary Fort 2010 to 2012: Neath, Neath Port Talbot CBC

Site 677 A total of 422 fragments of Roman glass were recovered from 90 contexts including 49 unstratified fragments. Most were small undiagnostic fragments of vessel glass. The diagnostic pieces are catalogued below. Vessels 1. Context 1000. Figure 46, 2.a A small rim fragment from a flask with a rim folded outward and back over. The internal diameter of the rim is 1.4cm and the external rim diameter is 4.6cm. 1st-2nd Centuries. 2. Context 1002. Plate 16. A very small fragment of glass with a colour between purple and brown. The glass is patterned with white rods forming a polychrome mosaic cup or bowl. 1st â&#x20AC;&#x201C; early2nd Centuries.

Plate 16. Glass No. 2

3. SF92. Context 1002. Plate 17. A small rim fragment from a bowl. Polychrome with dark green ground with yellow spirals which may be part of a floral motif. The fragment is too small to be certain. Mid 1st Century.

Nidum. Excavations at the Roman Auxiliary Fort 2010 to 2012: Neath, Neath Port Talbot CBC

Plate 17. Glass No. 3

4. Context 1005. Figure 46, 6.a. A single fragment of a ribbon handle in blue green with a central rib. Most likely from a convex jug or conical jug with long neck. Both forms are common from late 1st to late 2nd centuries. 7cm in length, 3.3cm in width and 5mm in thickness. 5. Context 1052. Figure 46, 2.b. A flask rim in blue green. The rim is formed by pulling the glass out, slightly downward, folded back inward and slightly downward. The internal rim diameter is 2cm and the external rim diameter is 6.4cm. 1st -2nd centuries. 6. Context 1065. Not Illustrated Fragment of wide blue green ribbon handle. 1st - 2nd Centuries. 7. Context 1076. Figure 46, 2.c. Rim, neck and handle fragment in pale blue green. Part of a square bottle. 1st to end of 2nd Centuries. 8. Context 1111. Not illustrated. Two small adjoining fragments from the body of a jug with a single vertical rib. The glass is dark yellow/brown. Late 1st to late 2nd century Neronian/early Flavian. 9. Context 1117 Not Illustrated. Blue green ribbon handle with reeding. 1st - 2nd Centuries. 10. Context 1141. Not illustrated. Blue green glass fragment of cylindrical bottle with a very small fragment of ribbed handle. 1st - early 2nd Centuries. 11. Context 1153. Not illustrated. Complete rim fragment from an unguent bottle. Rim diameter is 20mm. 2nd- early 3rd Centuries. 12. Context 1187. Figure 46, 3.a. Blue green glass fragment of square bottle base with concentric decoration. A total of seven square bottle bases were recovered from site 677 (three) and 716 (four). These comprise a series of concentric rings, some with a dot in the centre (677.12/13/17 and 716. 8/10/15/16). This is a very common motif (Cool and Price 1995 and dot motif. 1st - 2nd Centuries. 54

Nidum. Excavations at the Roman Auxiliary Fort 2010 to 2012: Neath, Neath Port Talbot CBC

13. Context 1201. Not Illustrated. Blue green glass fragment of square bottle base with concentric circle and dot motif. 1st - 2nd Centuries. 14. Context 1224. Figure 46, 4.a. A small fragment of dark amber/brown glass with a single rib. Likely part of a bowl of jug. The rib has a width of c.4mm. 1st – 2nd Centuries. 15. Context 1228. Not Illustrated. Pale blue green body fragment from a long necked jug. The fragment has a single rim running vertically. 1st – 2nd Centuries. 16. Context 1229. Not Illustrated. Rim fragment from a pillar moulded bowl in blue green. 1st century – early 2nd Centuries. 17. Context 1255. Figure 46, 3.b. Blue green glass fragment of square bottle base with concentric circles and dot motif. 1st – 2nd centuries. 18. Context 1333. Not Illustrated. Small body fragment from a pillar moulded bowl. 1st –early 2nd Centuries. 19. Context 1338. Not Illustrated. Amber glass fragment from a mould blown cup with square pattern. 1st Century. 20. Context 1338. Figure 46, 6.b. Reeded ribbon handle in blue green including body fragment where the handle meets the shoulder of the bottle. 4.5cm in width. 1st – early 2nd Centuries. 21. SF87. Context 1470. Not Illustrated. Plate 18. A small fragment of a bowl or cup in pale blue with marvered blobs in yellow and white. 1st - end of 2nd Centuries.

Plate 18. Glass No. 21

22. Context 1551. Figure 46, 2.d. Complete wide rim from a bottle. Internal rim diameter 3.6cm and external diameter 6.6cm. 1-2nd Century. 23. SF93. Context 1594. Not Illustrated. A very small fragment of cobalt blue glass with white applied decoration. 55

Nidum. Excavations at the Roman Auxiliary Fort 2010 to 2012: Neath, Neath Port Talbot CBC

24. Context 1594. Not Illustrated. Small fragment in cobalt blue from the body of a convex or conical jug. The fragments have part of two vertical ribs. 1-2nd Century. 25. Context 1713. Not Illustrated. Blue green rim fragment from a pillar moulded bowl. 1st â&#x20AC;&#x201C; early 2nd Century. 26. Context 1869. Figure 1.a. Blue green rim fragment from a pillar moulded bowl. The diameter would have been approximately 19cm and the widest point of the pillars 1.5cm. 1st â&#x20AC;&#x201C; early 2nd Century. 27. Context 5001. Not Illustrated. Fragment of blue green ribbon handle with a central, vertical rib. 1st-2nd Century. Beads A total of 34 beads were recovered from site 677. Of those 30 were faience melon beads and four were glass beads. 28. SF 35. Context US. Not Illustrated. A single fragment of turquoise faience melon bead. No glaze survives. 16mm in height. 1-2nd Centuries. 29. SF81. Context US. Plate 19.a. A single fragment of turquoise faience melon bead. No glaze survives. 16mm in height, 21mm external diameter, 10mm internal diameter. 1-2nd Centuries.

Plate 19. Glass Nos. 29, 31, 33, 34, 35, 38, 41, 44, 49, 50. Beads (left to right) a to e top row, f to j bottom row

30. Context 1001. Not Illustrated. A very small fragment of turquoise faience melon bead. 1st2nd Centuries. 31. SF 3. Context 1065. Plate 19.f. A single, small fragment of turquoise faience melon bead. No glaze survives. 12mm in height, 18mm external diameter, 8mm internal diameter. 1-2nd Centuries.

Nidum. Excavations at the Roman Auxiliary Fort 2010 to 2012: Neath, Neath Port Talbot CBC

32. SF 4. Context 1067. Not Illustrated. A single, large fragment of cobalt blue glass bead with gadroons around the circumference only. 18mm in height, 30mm external diameter, 14mm internal diameter. 1-2nd Centuries. 33. SF 6. Context 1075. Plate 19.g. A single fragment of turquoise, faience melon bead. 14mm in height, 18mm external diameter, 8mm internal diameter. 1-2nd Centuries. 34. SF 84. Context 1077. Plate 19.c. A single fragment of turquoise faience melon bead. Some glaze survives. 18mm in height. 1-2nd Centuries. 35. SF 8. Context 1117. Plate 19.h. A complete cobalt blue glass melon bead. 15mm in height, 24mm external diameter, 9.5mm internal diameter. 1-2nd Centuries. 36. Context 1139. Not Illustrated. A single, small fragment of turquoise faience melon bead. No glaze survives. 17mm in height. 1-2nd Centuries. 37. Context 1152. Not Illustrated. A single, small fragment of turquoise faience melon bead. No glaze survives. 1-2nd Centuries. 38. SF 11. Context 1187. Plate 19.i. A single fragment of cobalt blue glass melon bead. 12mm in height. 1-2nd Centuries. 39. SF 38. Context 1228. Not Illustrated. A complete turquoise faience melon bead with some glaze remaining. 13.5mm in height, 17mm external diameter, 8mm internal diameter. 1-2nd Centuries. 40. SF45. Context 1228. Not Illustrated. A single, small fragment of turquoise faience melon bead. Glaze survives. 1-2nd Centuries. 41. SF 20. Context 1316. Plate 19.j. A single, large complete cobalt blue glass bead. 18mm in height, 30mm external diameter, 14mm internal diameter. Plain with no gadroons. 1-2nd Centuries. 42. SF22. Context 1338. Not Illustrated. A single fragment of pale turquoise faience melon bead. No glaze survives. 18mm height. 1-2nd Centuries. 43. SF23. Context 1338. Not Illustrated. A single fragment of pale turquoise faience melon bead. No glaze survives. 14mm height. 1-2nd Centuries. 44. SF88. Context 1368. Plate 19.e. A single fragment of pale turquoise faience melon bead. No glaze survives. 1-2nd Centuries. 45. Context 1465. Not Illustrated. A single fragment of turquoise faience melon bead. Some glaze survives. 11mm in height. 1-2nd Centuries. 46. Context 1494. Not Illustrated. A single fragment of turquoise faience melon bead. Some glaze survives. 14mm in height. 1-2nd Centuries. 47. Context 1553. Not Illustrated. A single, small fragment of turquoise faience melon bead. Some glaze survives. 11mm in height. 1-2nd Centuries.

Nidum. Excavations at the Roman Auxiliary Fort 2010 to 2012: Neath, Neath Port Talbot CBC

48. SF 66. Context 1592. Not Illustrated. A single fragment of pale turquoise faience melon bead. Some glaze survives. 20mm in height, 26mm external diameter, 10.5mm internal diameter. 1-2nd Centuries. 49. SF 82. Context 1592. Plate 19.b. A single fragment of pale turquoise faience melon bead. No glaze survives. 1-2nd Centuries. 50. SF 86. Context 1713. Plate 19.d. A complete turquoise blue faience melon bead. Some glaze survives. 11mm in height, 14mm external diameter, 6mm internal diameter. 1-2nd Centuries. 51. SF 71. Context 1731. Not Illustrated. A single, small fragment of turquoise faience melon bead. Some glaze survives. 15mm in height. 1-2nd Centuries. 52. SF 79. Context 1731. Not Illustrated. A complete, pale turquoise faience melon bead. Some glaze survives. 17mm in height, 19.5mm external diameter, 8mm internal diameter. 1-2nd Centuries. 53. Context 1822. Not Illustrated. A single, small fragment of pale turquoise faience melon bead. No glaze survives. 15mm in height. 1-2nd Centuries. 54. Context 1841. Not Illustrated. Two small fragments of pale turquoise faience melon bead. Some glaze survives. 15mm in height. 1-2nd Centuries. 55. Context 1841. Not Illustrated. Two small fragment of turquoise faience melon bead. Some glaze survives. 15mm in height. 1-2nd Centuries. 56. Context 1869. Not Illustrated. A single, small fragment of turquoise faience melon bead. Some glaze survives. 12.5mm in height. 1-2nd Centuries. 57. SF 17. Context 3002. Not Illustrated. A single fragment of turquoise faience melon bead. 12mm in height, 14mm external diameter, 5mm internal diameter. 58. Context 5001. Not Illustrated. A single, small fragment of pale turquoise faience melon bead. No glaze survives. 16mm in height. 1-2nd Centuries. Counters A total of twelve complete or part glass counters were recovered from site 677. Of these seven were white and five were black. They came from eight different contexts while three were unstratified.

Nidum. Excavations at the Roman Auxiliary Fort 2010 to 2012: Neath, Neath Port Talbot CBC

Plate 20. Gaming Counters Nos. 62 -66, 69 (left to right) a to f

59. SF48. Context US. Not Illustrated. A chipped white glass gaming counter. 19mm diameter 60. SF83. Context US. Not Illustrated. A very small fragment of black glass gaming counter. 61. SF90. Context US. Plate 20.f. A complete white glass gaming counter. 16mm diameter. 62. SF7. Context 77. Plate 20.d. A complete black glass gaming counter. Diameter 15mm. 63. SF40. Context 1006. Plate 20.c. A complete black glass gaming counter. 17mm diameter. 64. SF26. Context 1076. Plate 20.b. A complete black glass gaming counter. More oval in shape with the shorter axis measuring 15mm and the longer axis 22mm. 65. SF89. Context 1091. Plate 20.e. A fragment of a white glass gaming counter. 17mm diameter. 66. SF33. Context 1201. Not Illustrated. Two adjoining fragments of a black glass gaming counter. 14mm diameter. 67. SF34. Context 1201. Not Illustrated. A complete white glass gaming counter. 16mm diameter. 68. SF12. Context 1228. Plate 20.a. A fragment of a white glass gaming counter. 14mm diameter. 69. SF52. Context 1252. Not Illustrated. A complete white glass gaming counter. 15mm diameter. 70. SF73. Context 1844. Not Illustrated. A complete black glass gaming counter. 18mm diameter. Other objects

Nidum. Excavations at the Roman Auxiliary Fort 2010 to 2012: Neath, Neath Port Talbot CBC

71. SF 91. Context 1032. Not Illustrated. A fragment of twisted stirrer in blue green glass. Surviving length 31mm and diameter 10mm. 72. Context 1519. A very small fragment of probable Roman window glass Site 716 A total of 176 fragments of Roman glass were recovered from 83 contexts including three unstratified fragments. Most were small undiagnostic fragments of vessel glass. The diagnostic pieces are catalogued below. Vessels 73. Context 6000. Not Illustrated. Body fragment from a blue green square bottle. 1st â&#x20AC;&#x201C; 2nd Centuries. 74. Context 6000. Not Illustrated. A single fragment of bottle rim in blue green. 1-2nd Centuries. 75. Context 6023. Not Illustrated. A very small fragment of cobalt blue glass with white marvered blobs similar to fragment 716.21. Dates etc. 76. Context 6023. Not Illustrated. Two small adjoining fragments in pale green from the base of a candlestick unguent bottle. 1st-early 3rd. 77. Context 6037. Not Illustrated. Lower body fragment, cup/bowl. Pale blue green. Two vertical mould blown lines. 78. Context 6037. Not Illustrated. Lower body fragment, cup/bowl. Colourless.

Plate 21. Glass No. 79

79. SF 1. Context 6057. Not Illustrated. Plate 21. A single fragment from the body of a pillar moulded bowl in dark translucent green with yellow and red rods. 1st â&#x20AC;&#x201C; early 2nd centuries. 60

Nidum. Excavations at the Roman Auxiliary Fort 2010 to 2012: Neath, Neath Port Talbot CBC

80. Context 6067. Not Illustrated. A fragment of bottle base in blue green with part of a circle and dot motif. 1-2nd Centuries. 81. Context 6067. Not Illustrated. Two small fragments of pale blue green glass each with a small mould blown rib. 1-2nd Centuries. 82. Context 6078. Figure 46, 3.c. A fragment of bottle base in blue green with part of a circle and dot motif. 1-2nd Centuries. 83. Context 6098. Not Illustrated. A fragment of cobalt blue handle with marvered white decoration. 1-2nd Centuries. 84. Context 6257. Figure 46, 2.e. A single fragment of bottle rim in blue green. The rim is folded out, up and back in. Internal diameter is 7.4cm and external rim diameter is 12.2cm. 12nd Centuries. 85. Context 6431. Not Illustrated. A narrow neck fragment of an unguent bottle. Probable 12nd Centuries. Internal – fired clay deposit. (not Illustrated) 86. SF 25. Context 6440. Not Illustrated. A single rim fragment from a pillar moulded bowl in pale blue green. mid 1st – early 2nd Centuries. 87. Context 6448. Not Illustrated. Two fragments from a blue green bottle. One is a body fragment and the other is a base fragment with circle and dot motif. 1st – 2nd Centuries. 88. Context 6448. Not Illustrated. Two fragments from a blue green bottle. One is a body fragment and the other is a base fragment with circle and dot motif. 1st – 2nd Centuries. 89. Context 6479. Not Illustrated. A fragment of cylindrical blue green bottle. 1st – early 2nd Centuries. 90. Context 6567. Figure 46, 1.b. A fragment of blue green pillar moulded bowl. 16.2cm diameter. 1st – early 2nd Centuries. 91. Context 6622. Not Illustrated. A fragment of blue green ribbon handle from a bottle/jug. 1st – 2nd Centuries. 92. Context 6652. Figure 46, 5.a. A fragment of a flask with rim and ¾ of neck. The rim is folded out, up and in then flattened. The rim internal diameter is 0.8cm and the external rim diameter is 2.3cm. This fragment is part of No. 96. The trails of bubbles can be seen to run in the same lines. Pale blue green. 2nd Century. 93. Context 6983. Not Illustrated. A small fragment of cobalt blue bowl with white marvered blobs. Similar to 716.3. 1st – 2nd Centuries. 94. Context 6983. Not Illustrated. A small neck fragment from a pale blue green unguent bottle. 1st – 2nd Centuries. 95. Context 6986. Not Illustrated. A fragment of pale blue green cylindrical bottle with very small part of two ribs from a handle adhering to the shoulder 1st – early 2nd Centuries. 96. Context 6988. Figure 46, 5.a. A small fragment of pale blue green unguent bottle same vessel as No. 92. Surviving length is 11.4cm. 1st-3rd centuries. 61

Nidum. Excavations at the Roman Auxiliary Fort 2010 to 2012: Neath, Neath Port Talbot CBC

Beads A total of twelve beads were recorded from ten contexts with two of them being unstratified. All except one were faience melon beads. The remainder was a glass melon bead.

Plate 22. Glass Nos. 99, 100, 102, 106, 108. Beads (left to right) a to e

97. SF 14. Context US. Not Illustrated. A single fragment of pale turquoise faience melon bead. Some glaze survives, particularly on the interior. 21mm in height. 1-2nd Centuries. 98. SF 21. Context US. Not Illustrated. A single fragment of pale turquoise faience melon bead. Some glaze survives. 14mm in height. 1-2nd Centuries. 99. SF 7. Context 6023. Plate 22.b. A single fragment of pale turquoise faience melon bead. Some glaze survives. 18mm in height. 1-2nd Centuries. 100. SF 8. Context 6023. Plate 22.c. A single fragment of pale turquoise faience melon bead. Some glaze survives. 14mm in height. 1-2nd Centuries. 101. SF 10. Context 6023. Not Illustrated. A single fragment of pale turquoise faience melon bead. Some glaze survives. 14mm in height. 1-2nd Centuries. 102. SF 3. Context 6079. Plate 22.a. A single fragment of turquoise faience melon bead. Some glaze survives. 21mm in height. 1-2nd Centuries. 103. Context 6117. Not Illustrated. A tiny fragment of turquoise faience melon bead. 1-2nd Centuries. 104. SF 13. Context 6479. Not Illustrated. A single fragment of cobalt blue glass melon bead. 1-2nd Centuries. 105. SF 24. Context 6652. Not Illustrated. A single fragment of turquoise faience melon bead. Some glaze survives on interior surface. 11mm in height. 1-2nd Centuries. 62

Nidum. Excavations at the Roman Auxiliary Fort 2010 to 2012: Neath, Neath Port Talbot CBC

106. SF 27. Context 6673. Plate 22.d. A single fragment of pale turquoise faience melon bead. No glaze survives. 9.5mm in height. 1-2nd Centuries. 107. SF 31. Context 6432. Not Illustrated. A complete pale turquoise faience melon bead. No glaze survives. 12mm in height, 14mm external diameter, 5mm internal diameter. 1-2nd Centuries. 108. SF 29. Context 6808. Plate 22.e. A complete turquoise faience melon bead. Some glaze survives. 12mm in height, 16mm external diameter, 5mm internal diameter. 1-2nd Centuries. Counters A total of five counters or parts of counters were recovered from five different contexts at site 716. All counters were white and roughly circular in plan with the usual convex upper surface and rough base. One counter was nearer oval in shape. 109. SF5. Context 6030. Not Illustrated. A slightly chipped, white glass gaming counter. More of an oval shape measuring 16mm diameter across shorter axis and 20mm along longer axis. 110. SF4. Context 6138. Not Illustrated. A slightly chipped, white glass gaming counter. 16mm diameter. 111. SF23. Context 6447. Not Illustrated. A very small fragment of a white glass gaming counter. 112. SF19. Context 6585. Not Illustrated. A complete white glass gaming counter. 14mm diameter. 113. SF18. Context 6590. Not Illustrated. A complete white glass gaming counter. 12mm diameter. Discussion Coloured glass The fragments of coloured glass were all recovered from sites 677 and 716. The polychrome vessel fragments were all small and determining vessel type and profile was not always possible. The polychrome colours included dark purple/brown with white rods (Plate 16), dark green with a yellow spiral or floral motif (Plate 17) and dark green with yellow and red rod decoration (Plate 21). Brightly coloured vessels are known in British sites from the conquest to early Flavian occupation. Similar polychrome vessels are found in military contexts such as within the barracks at Caerleon within a Flavian-Trajanic context (Price 1995b, 87 Fig 9, 1). Coloured monochrome vessel fragments were also small. This made determining vessel type and profiles difficult in most cases. The catalogues fragments include an amber bowl/jug in amber with single vertical rib (14, Figure 46 4.a), an amber mould blown cup (19) and a cobalt blue jug with two vertical ribs (24). These are likely to correlate to a late Neronian/early Flavian date (Price and Cottam 1998, 152-4). Fragments of monochrome glass with applied/marvered decoration were also very small with similar poor outcomes for form and profile details. This type of applied and marvered decoration is found on numerous vessel types. This decoration ranges in date from the end of 63

Nidum. Excavations at the Roman Auxiliary Fort 2010 to 2012: Neath, Neath Port Talbot CBC

the first quarter of the first century to the end of the second century. This decoration is rare after the early Flavian period (Price and Cottam 1998, 30; Williams 2007, 20). A tiny fragment of cobalt blue glass with white applied decoration was recovered from site 677 (23), as was a fragment of probable bowl/cup in pale blue with marvered blobs in white and yellow (21, Plate 18). A fragment of cobalt blue handle with marvered white decoration (83) and two body fragments of cobalt blue glass also with white marvered decoration (75 and 93) were also found. Blue Green Glass This colour is the most common colour of glass through the Roman period. The colour was achieved naturally through the presence of iron oxide in sand and this lead to significant variation in the colour. Bottles and household vessels were predominantly blue green in the first and second centuries. Blue green pillar moulded bowls (16, 18, 25, 26, and 86, 90) are common in first century Flavian contexts although they are still used into the early second century (Price and Cottam 1998, 44). Bottles in blue green were common from this assemblage of bottle glass being larger than any other vessel type. This is almost always the case from Roman sites of this type. Cylindrical bottles (89 and 95) date from the first to early second centuries whilst the square bottles (7, 12, 13, 17 and 73) continue to the end of the second century. A total of seven fragments of square bottle bases with concentric circles were recovered from the excavation (12, 13, 17, 80, 82, 87, 88). This is a common motif (Cool and Price 1995, Figures 11.8-12). Hexagonal bottles (3) go out of use in the mid second century although some occasional fragments are found through to the middle of the third century. Blue green convex or conical jugs in blue green are not in use beyond the end of the second century. Some forms are only in use during the first century however the fragments in this assemblage do not allow for such precise dating (Price and Cottam 1998, 147-155). The jug fragments in this assemblage are attested by ribbed fragments and ribbon handles (4, 8, 15, 24, 27 and 91). A single fragment of glass stirring rod was recovered from site 677 (71) from the fill of a modern trench. Glass rods are thought to date from 1st and 2nd Centuries although they are found throughout the Roman Period (Charlesworth 1972, 215). The rods were used to stir cosmetic preparations including perfumes (Isings 1957, form 79). When they were stored in an unguentariua the oil based preparations would often separate and require stirring prior to use. Beads A total of 48 beads were recovered from the excavations. Site 674 produced five faience melon beads and a single glass melon bead. Site 677 produced 30 faience melon beads and four glass beads. Finally, site 716 produced eleven faience melon beads and four glass beads. These beads are found in 1st- and 2nd-century contexts (Guido 1978, 100). The glass beads are similarly gadrooned as the faience examples with the exception of a single, large biconical bead in cobalt blue (41) and a large cobalt blue glass bead with gadroons on the circumference of the bead (32). No comparison could be found for this bead however it can be assumed it has the same date range as the melon beads.

Nidum. Excavations at the Roman Auxiliary Fort 2010 to 2012: Neath, Neath Port Talbot CBC

Gaming counters All of the glass gaming counters were circular in plan, flat-bottomed, convex 'blobs' of glass. Black and white examples were recovered from each site at Neath. Glass gaming counters are found throughout the Roman period. Roman towns and forts produce more glass gaming counters than rural sites. Conclusion The Roman glass recovered from the three sites covers the period from the second half of the first century AD to the end of the second century AD. Some fragments of polychrome glass and blue green pillar moulded bowls highlight first century occupation while blue green bottles, jugs, melon beads and gaming counters represent both the first and second centuries. The fragments were all very small which might point to the larger pieces being systematically collected for cullet. When systematic collection of glass for cullet takes place on a site, only the smallest fragments might have escaped collection. The entire assemblage from the three sites, when considered together, forms the largest recovered from excavations in Neath. Several of the vessel or object types recovered are represented by more than one fragment. There is a relatively high proportion of diagnostic pieces revealing a fairly narrow range of vessels and objects in use.

5. The Roman coins By Edward Besley 1. 716/6257, Probably a denarius 1st to 2nd century AD. Weight 3.08g 2. 677/1002 (SF01), Copper alloy, remains of an as? 3. 677/1228 (SF37), Copper alloy: fragmentary as, unworn, probably Flavian (reign of Vespasian, 69-79 AD), with reverse Spes standing left, [S] â&#x20AC;&#x201C; C 4. 677/1592 (SF55), Copper alloy; probably an as or dupondius of 1st -2nd century date 5. 677/5048 (SF32), Denarius, uncertain (encrusted). Present weight 3.64g

6. Objects of copper alloy and lead (Figure 47) By Evan Chapman Most of the nonferrous metal retrieved was fairly nondescript. All was in a poor state of preservation, suggesting very adverse soil conditions, and most was fragmentary. Amongst those pieces that could be identified there was a high preponderance of horse harness related items: seven out of the thirteen copper alloy objects catalogued were definitely or possibly such. Most interesting amongst these are two phallic mounts, neither of which can be readily closely paralleled.

Nidum. Excavations at the Roman Auxiliary Fort 2010 to 2012: Neath, Neath Port Talbot CBC

Copper alloy Site 677 1. 1081 (SF002). Small globular terminal. Diameter 7mm. Probably the remains of a pendant or strap-end of indeterminable form. Not illustrated. 2. 5004 (SF014). Small phallic mount with very angular modelling. Length 24mm. There are the remains of one rectangular stud in the centre of the back. Cf Corbridge (Allason-Jones 1988, 179 no.168); Exeter (Allason-Jones 1991, 249 no.62), although neither is a very close parallel (Figure 47). 3. 1299 (SF016). (Not illustrated). Sturdy functional ring with a groove decorating its outer edge and a slight collar around the inner edge, on one face. Diameter 33mm, thickness 7mm. Given the other finds from the site a function as part of horse harness would seem a reasonable suggestion (Bishop 1988, 94). 4. 1228 (SF047). Mount with a slightly upward curving phallus projecting out of it. Length 16mm, max. diameter 9mm. The base plate of the mount is slightly damaged but appears to have been D-shaped, 16mm by 16mm. The bottom edge of the plate is notched and worked to give a crude representation of testicles. There is a short circular stud, with flanged end, in the centre of the back: length 4mm, diameter 5mm. Cf Wroxeter (Mould 2000, 123 no.22), for a mount of similar general form. 5. 1252 (SF050). Flat block, one end of which is curved and scalloped. Length 29mm, width 17mm, thickness 6mm. Both faces have grooved lines running in from the scalloped edge, converging at a central point, below which there is a circular peroration, apparently contain a rivet. The presence of iron corrosion over the hole on one face suggests that this rivet originally had an iron shaft. The other end of the block is flat with a small, central, circular hole, apparently containing traces of iron. The size and form would suggest a small key handle but the presence of the rivet and the fact that the hole in the end seems small to take the shaft of a key, suggests that a decorative terminal or mount is probably a better interpretation. 6. 1713 (SF054). (Not illustrated). Plain flat handle with faceted edges. Length 94mm, max. width 26mm, thickness 7mm. It tapers slightly before expanding into a spayed terminal, which has a central circular perforation. The other end is decorated by a groove on both faces and has the remains of an iron tang inserted in it. 7. 1592 (SF068). (Not illustrated). Globular headed stud with the remains of a washer still attached. Diameter of head 9mm, length 12mm. 8. 1841 (SF072). (Not illustrated). Harness junction loop fragment with twin rivet holes each surrounded by an incised circle. Length 30mm [broken]. Little of the plate survives beyond these first rivet holes but it appears to be of Bishop type 3b (Bishop 1988, 157). 9. 1796 (SF076). (Not illustrated). Roughly semi-circular plate with a short rectangular projection from the middle of the straight edge. Width 45mm, length 41mm [broken], thickness 2mm. There is a central, large circular perforation, diameter 12mm. Incomplete and poorly preserved but possibly the remains of a side-plate for a bit. Cf Hod Hill (Brailsford 1962, 2 nos. A25-27); Vindonissa (Unz & Deschler-Erd 1997, 50 no.1958). 66

Nidum. Excavations at the Roman Auxiliary Fort 2010 to 2012: Neath, Neath Port Talbot CBC

Site 716 10. 6045 (SF006). (Not illustrated). Rod expanding into what appears to be a broken and battered blade. Length 54mm [broken], max. width 14mm. Possibly the remains of a razor or one arm of a pair of shears. 11. 6652 (SF026). (Not illustrated). Narrow plate with a raised transverse band. Length 35mm [broken], width 9mm. Both ends are broken but one appears to be notched. A stud projects from the back near the notched end. Possibly the remains of a harness fitting, with the notch suggesting the fixed part of a hinged female strap fastener. Although on the narrow side, Bishop (1988, 168 type 6b) lists an example from Hofheim of similar width. Lead Site 677 12. 1338 (SF024). Narrow strip with a rod projecting from one end, probably part of a pottery repair rivet. Length 29mm [broken], width 6mm, thickness 4mm; rod 9mm long, diameter 4mm. Not illustrated. 13. 1338 (SF046). (Not illustrated). Circular sectioned rod, flattened into a circular, perforated, terminal at one end. Length 112mm, diameter of rod 15mm, diameter of terminal 30mm, thickness of terminal 6mm, diameter of perforation 10mm. 14. 1551 (SF053). (Not illustrated). Oval sectioned rod broken off across a rounded perforation, at one end. Length 47mm [broken], width 26mm, thickness 24mm. 15. 1110. Rod flattened at one end, towards which it is also slightly tapered. Length 69mm; wider end 10 by 10mm; narrower end 8 by 4mm. Not illustrated. 16. 1228 (SF042). (Not illustrated). Crude rod tapering to a point at one end. Length 44mm, diameter 6mm. 17. 1338 (SF025). (Not illustrated). Short T-shaped rod. Length 15mm, diameter 5mm; width across arms 10mm. 18. 1592 (SF065). (Not illustrated). Semi-circular block with cut marks along its straight edge suggesting it was originally a circular disc that has been cut in half. Diameter 58mm, thickness 18mm. Site 716 19. 6023 (SF011). (Not illustrated). Crude rod of roughly D-shaped section, with a circular perforation, diameter 5mm, near one end. Length 78mm, width 20mm, thickness 13mm. 20. /6652 (SF022). (Not illustrated). Bar with a small, 6mm square, perforation near one end. It expands and thins out away from the hole. Length 87mm; narrow end 17 by 13mm; wider end 26 by 8mm.

Nidum. Excavations at the Roman Auxiliary Fort 2010 to 2012: Neath, Neath Port Talbot CBC

7. The Roman ironwork (Figures 48-53, Plates 23-25) By William Manning Site 716 spearheads Nos 1-3 comprise a group of three spearheads from context 716/6663 (Figure 48, Plate 24 H911, H912). All are similar in their general form and size with long and relatively narrow blades, which are lentoid in cross-section, and narrow sockets. The three were clearly buried as a group and have corroded together. Although mineralised wood remains in the sockets it is unlikely that the hafts were complete when they were buried. In a discussion of the patterns of deposition seen in the Flavian fort at Newstead the writer has argued that spears with broken hafts were routinely returned to the fabrica for repair and that on occasion, particularly when a fort was closed, they could be discarded (Manning 2006). Their form and the dimensions of the blades place them in Type 3 of the classification proposed by the writer for the spearheads from Newstead (Manning 2006, 81, fig.4.5). 1. 6663 Spearhead 1: Length 171mm to the end of the surviving socket; 212mm to the end of the fragment of the wooden haft which is probably the original length of the spearhead. Length of the blade 126.4mm; width of the blade 33.0mm. 2. 6663 Spearhead 2: Length 202mm. Length of the blade 130+/-mm (The precise length is obscured by corrosion); the width of the blade is uncertain. 3. 6663 Spearhead 3: 213.4mm. Length of the blade 164.5mm; width of the blade 36.6mm. Similar spearheads come from Caerleon (Chapman 2005, 29, 30 no. Da07), the Corbridge hoard (Allason-Jones & Bishop 1988, 11, no. 3, fig. 10; no. 9, fig. 11; 13, no. 27, fig. 13; no. 34b, fig. 15; no. 38, fig. 16; 17, no. 39, fig. 16; nos. 45 & 46, fig. 18), Dragonby, Lincolnshire (May 1996, 290, no. 17, fig. 11.32), Housesteads (Rushworth 2009, 457, no. 316, fig. 14.14), Richborough (Bushe-Fox 1926, 48, pl. XVI, 38; 1949, 153, pl. LVIII, 277 & 278) and Vindolanda (Bidwell 1985, 135, fig. 48, 19). Continental examples include examples from Butzbach (MĂźller 1962, 59, Taf. 16, 10), Hofheim (Ritterling 1913, 159, Taf. XVII, 13), Neuss/Novaesium (Simpson 2000, 91, pl. 29, 14), Stockstadt (O.R.L. B III, Kastell 33 (1914), 54, Taf. IX, 22), Carnuntum, Austria (von Groller 1901, 123, Taf. XXII, 11-13), and 282, Taf. 17-18) and Vindonissa, Switzerland (Unz & Deschler-Erb 1997, 20-21, nos.261-263, 273- 282, Taf. 17-18). 4. 6714 (SF 030) Fragment (? of a spearhead). (Figure 48, Plate 25 - H905). Length 67.5mm. It tapers from a thin, relatively wide, broken edge to its tip, thickening as it narrows to create a blunt, rod-like point. It may be the tip of a spearhead of Newstead Type 4, some of which have intentionally blunt tips (Manning 2006, 81, 84, fig. 4.6). Spearheads with tips of this type come from Newstead (Curle 1911, 188, pl. XXXVI, 1, 2. 5, 7), Richborough (Bushe-Fox 1949, 153, pl. LVIII, 286), the Corbridge hoard (Allason-Jones & Bishop 1988, 11, no. 3, fig. 10) and Vindolanda (Birley 1996, 33, no. 105, fig. 16). Continental examples are known from Weissenburg (O.R.L. B VII, Kastell 72 (1914) 39, Taf. VIII, 28), Zugmantel (O.R.L. B II 1, Kastell 8 (1937) 98, Taf. XIV, 9), Niederberg (O.R.L. B I, Kastelle 68

Nidum. Excavations at the Roman Auxiliary Fort 2010 to 2012: Neath, Neath Port Talbot CBC

2a (1937), 8, Taf. VII, 14) and Vindonissa, Switzerland (Unz & Deschler-Erb 1997, 20-21, nos. 264-266, 287.) Site 677 Boltheads (Ballistae) 5. 1187 Bolthead. Length 87.0mm (Figure 49, Plate 25 - H906). The mouth and seam of the socket are damaged; originally it was somewhat longer. The blade is relatively narrow with slightly asymmetrical edges. Its maximum width was near its midpoint. The point is missing. 6. 6714 (SF 028) Bolthead. Length 74.8mm (Figure 49, Plate 25 - H906). Socket and fragment of the blade. The socket is relatively narrow with no obvious signs of the seam; it still retains the remains of the wooden shaft. There is a nail close to the mouth. The blade was probably narrow but only the lower part survives. Originally it probably resembled no.5. The size of these two missiles heads suggests that they were intended to be fired from a ballista rather than being spears or arrowheads. Although they are not a common type, similar ones are known from Hod Hill (Manning 1985, 175, V254-8), and the same fort has produced large numbers of functionally similar heads which have simple flanged sockets (Manning 1985, 176, V259-279). Other boltheads Site 716 7. 6257 Bolthead. Length 96.1mm (Figure 49, Plate 25 - H908). It has a relatively long, narrow socket with no visible seam, now filled with mineralised wood and corrosion products. A short, rod-like neck connects it to the head which widens to a bi-pyramidal head, with its widest point near the mid-point. The tip is lost. Site 677 8. 1338. Bolthead. Length 93.3mm (Figure 49). It has a relatively long, narrow, almost cylindrical-sectioned socket and a long, conical head, broken at its tip, with slight, level shoulders. The distortion of the head is probably the result of corrosion. 9. 1333. Bolthead. Length 104.3mm (Figure 50, Plate 25 - H906). It has a long, narrow, subrectangular sectioned socket, damaged at its mouth, which retains part of the wooden shaft. There are very slight shoulders between the socket and the long pyramidal head. 10. 1091 (SF 021). Bolthead. Length 79.0mm (Figure 50, Plate 24 - H910) The socket, which is relatively narrow, is slightly damaged at its mouth. There is no sign of the seam, although it may be concealed by corrosion. The short, symmetrical bi-pyramidal head is only a little wider than the socket. Large numbers of boltheads of this type come from Hod Hill (Manning 1985, 170-175, V141- V249 where other examples are cited.) Examples from Newstead are discussed in Manning 2006 (86, fig. 4.7) 11. 1142. Socketed head. Length 48.2mm (Figure 50, Plate 25 - H909). Relatively long, cylindrical socket, damaged at its mouth, with a short neck between the socket and the short, rounded, pyramidal head. Mineralised wood remains in the socket. 12. 1106=1343. Socketed Head? Length 67.4mm (Figure 50, Plate 25 - H906). It has a long, cylindrical socket broken at its mouth and still filled with mineralised wood, which runs into 69

Nidum. Excavations at the Roman Auxiliary Fort 2010 to 2012: Neath, Neath Port Talbot CBC

an asymmetrical head. The sides of the socket continue into the head with scarcely any widening; only on one side does it expand through a distinct shoulder into a marked projection, above which it narrows on all sides to a rounded tip. 13. 1731. Socketed Head. Length 50.2mm (Figure 50, Plate 25 - H905). It has a short, wide, conical socket with a short, almost globular head. There is no neck between the socket and the head. Fittings of this type have traditionally been identified as the ferrules of bolts fired by ballistae, but there are considerable problems with this suggestion. Other possible functions are discussed at some length by the writer in the forthcoming catalogue of the iron artefacts from the fort at Newstead (Manning forthcoming) where it is suggested that they could be the heads of ballista bolts used in training. Unlike the usual bolt these would not have penetrated a wooden target and could have been recovered for reuse. Examples are known from Caerleon (Chapman 2005, 57, Jc01), Caersws (Jones 1993, 75, nos. 2, 5-8, fig. 24; Jones 1996, 28, no. 12, fig. 11; Britnell 1989, 70, nos. 1-5, fig. 35), Carlisle (Howard-Davis 2009, 712, fig. 362, no. 25), Bremetenacum/Ribchester (Buxton & HowardDavis 2000, 267, fig. 70, nos. 49-59), Cataractonium/Catterick (Wilson 2002, 99, no.2, fig. 278), Corbridge (Bishop & Dore 1988, 193, no.13, fig. 90), the Corbridge hoard (Allason-Jones & Bishop 1988, 17, nos. 51, fig. 19; 52, fig. 10; 53, fig. 19; 58-59; 60-61, fig. 20; 22, no. 63, fig. 20), Ham Hill (Webster 1960, 83, fig.5, 129), Housesteads (Manning 1976, 21, no. 280), Leucarum (Lougher) (Marvell & Owen-John 1997, 284, pl. XX, 2nd right; Chapman 2005, 57, Jc05), Usk (Manning, Price and Webster 1995, 31-32, no. 18, fig. 33; Chapman 2005, 57, Jc03) and Vindolanda (Birley 1996, 23-4, nos. 43-66, figs. 9 & 10). In Germany examples come from Hofheim (Ritterling 1913, 160, Taf. XVII, 57-66), Hüfingen (O.R.L. B V 2, Kastell 62a, (1937), 44, Taf. XII, 56-9 & 66-71), Kösching (O.R.L. B VII, Kastell 74 (1914), 23, Taf. IV, 16), Kastell Oberstimm (Schönberger 1978, 166, no. 57, Taf. 18), Rheingönheim (Ulbert 1969a, 52, Taf. 46, 24 & 25), Risstissen (Ulbert 1959, 107, Taf. 67, 1) and Wiesbaden (O.R.L. B II 3, Kastell 31 (1915), 100, Taf. XI, 30-1),. Others are known from the Magdalensberg (Dolenz 1998, 70, M82-83, Taf. 9) and Vindonissa in Switzerland (Unz & Deschler-Erb 1997, 23-4, nos. 399-406, Taf. 21). Knives Site 716 14. 6447. Knife. Length 95.0mm. (Figure 51, Plate 25 - H909). The back of the blade arches up from the plate-like tang before curving down to a broken tip. The edge is slightly convex with a distinct step between it and the tang. It is an example of a Manning Type 14 knife (Manning 1985, 115, fig. 28, where other examples are cited). 15. 6443. Knife. Length 174.0mm. (Figure 51, Plate 24 - H913). The back, which continues the line of the top of the flat, tapering tang, rises in a shallow curve before curving down to the broken tip. It has a slightly curved heel. It is an example of a Manning Type 18a knife (Manning 1985, 117, fig. 29, where other examples are cited).

Nidum. Excavations at the Roman Auxiliary Fort 2010 to 2012: Neath, Neath Port Talbot CBC

Key and Padlock Site 677 16. 1077. Slide Key (Type 2). Length 69.6mm. (Figure 51, Plate 25 - H909). The handle widens slightly from its junction with the neck to a rounded top below which is a round eye. There are slight shoulders between the handle and the neck. The rectangular bit is set at right angles to, and on one side of, the neck. There are three slits running across the face of the bit; of these those adjacent to the neck and near the end of the bit stop short of the front edge of the bit. Both are about 2.7mm deep. Between the two is another cut which only runs for half of the width of the bit. It is a common type which is discussed in the report on the artefacts from the fortress at Usk (Manning, Price and Webster1995, 267, nos. 13-16, fig. 82), and in greater detail in the catalogue of the ironwork from the fort at New stead (Manning, forthcoming). 17. 1081. Lever Padlock Length 72.4mm. (Figure 52, Plate 23 and Plate 24 - H910). Large fragment of a padlock, consisting of the upper part of the case and the moveable Ushaped bar or shackle. The lower part of the case and most of one face are missing revealing a significant part of the mechanism. The top and sides of the case were flat with the sides splaying out from the top. X-ray photographs suggest that within the corrosion the top is bent into a shallow V-shape, probably the result of damage, perhaps resulting from an attempt to force the padlock open. How much of the case and the mechanism is missing is uncertain, but sufficient remains for the principles on which it worked to be clear, and the X-ray photographs help to clarify the details, although inevitably much remains obscure.

Plate 23. X-ray of padlock 1081

In the X-ray photographs the curving body of the shackle appears to be hollow, a feature which probably indicates that it was formed by welding two bars together; the apparent central hollow being the result of corrosion along the weld. The X-rays reveal that the left-hand end of the shackle (as seen in the illustrations) is still inside the case, although it has been pushed up from its original position by the distortion of the top of the case. In the X-ray photographs the top of the case appears to end just to the right of the shackle, no doubt indicating the edge of the hole 71

Nidum. Excavations at the Roman Auxiliary Fort 2010 to 2012: Neath, Neath Port Talbot CBC

through which the shackle entered the case. The end of the shackle divides around a pivot which is visible on the face of the lock and which can be seen as a light disc in the X â&#x20AC;&#x201C;ray photographs. Although its primary function was as a pivot it will also have acted as a rivet holding the two faces of the case together. Another rivet can be seen both in the X-rays and on the surface in a similar position at the other end of the case. This will be discussed below, but it too will have acted as a structural rivet holding the case together. The other end of the shackle is visible on the X-ray photographs running through a bolt-hole in the top of the case; the left-hand edge of which is clear, as is the edge behind the shackle. Inside the case the end of the shackle thins to end in a relatively wide plate, which can be seen inside the case as well as on the X-rays.. Immediately behind the end of the shackle where it enters the case is the second rivet referred to above, in a position which suggests that it may also have acted as a guide to control the movement of the end of the shackle as it was pushed into the case. The use of large rivets to hold a padlock case together is seen even more clearly in a padlock from Fishbourne (Manning 1971) which has six such rivets. The end-plate of the shackle must have had a hole through it to receive the end of the bolt when it was locked, but this is not visible on the X-ray photographs. When discarded the lock was closed but not locked. The bolt is clearly visible in the broken case. It is held by two rectangular collars which are presumably attached to the back of the case. The X-ray photographs suggest that the right-hand collar has an irregular feature at its base which may have been related to one of the missing elements of the mechanism. The bolt is heavily corroded but its general form is clear. Viewed from the side the section at its left-hand end, the tail of the bolt, takes the form of a right-angled triangle, its top continuing the line of the top of the bolt but with its underside sloping down from a square end to a distinct heel. Some way to the right of this heel a short V-shaped lug projects from the underside of the bolt. The heel of the tail of the bolt and this lug are the 'stops' which the bit of the key pushed against to move the bolt when the key was turned in the lock. Between the heel and the lug the underside of the bolt rises in a concave curve which will have allowed the bit of the key to rotate. To the right of the V-shaped lug the underside of the bolt slopes up to meet the back in a squared tip. The X-ray photographs show the back of the bolt to be slightly dished. It is generally similar to the bolt of a padlock from the Fishbourne villa (Manning 1971, 143, fig. 65) rather than the alternative form, seen in locks from Caerleon (Manning 1968) and the Lullingstone villa (Anstee 2001), which have a slot in the bolt to receive the bit of the key rather than the lugs seen here. The X-ray photographs reveal a strip of metal which curves down from a point slightly above heel of the bolt to run along the top of its tip. This is clearly a spring intended to hold the bolt in position while allowing it to move to and fro when the lock was opened and closed. A rectangular fitting, open at its centre, is attached to the top of the case just to the left of the righthand collar. The details are obscured by corrosion but the underside of this fitting appears to touch the top of the spring. The most likely explanation is that it ensured that the end of the spring remained in contact with the bolt when the latter was moved. In the X-ray photographs the open centre of this rectangular component resembles a capital F, suggesting that it too originally had an additional function, although that was not obvious from the surviving fragments. Although the details are not absolutely clear on the X-ray photographs there appears to be a second, shorter spring curving down below the main one to which it is attached at its left-hand end. The other end of this spring touches the top of the bolt at its midpoint. Its 72

Nidum. Excavations at the Roman Auxiliary Fort 2010 to 2012: Neath, Neath Port Talbot CBC

function will have been to prevent the bolt rising as it moved forward when the lock was closed. The rear ends of both springs must have been attached to the body of the lock but this part of the case is missing.

Plate 24. X-ray plates of iron artefacts (H910, H911, H912, H913)

The fragment of strip below the bolt, and a loop-like feature by it are almost certainly not in their original positions. The â&#x20AC;&#x2DC;loopâ&#x20AC;&#x2122; may be the distorted remains of one of the wards around the, now missing, keyhole. Originally there will have been more springs to support and move the bolt and, to judge by other padlocks of this date, a tumbler which will have secured the bolt, but if so, they were lost along with much of the case including the key-hole and the wards which would have ensured that the lock could only be opened with the correct key. It is an example of a relatively common form of Roman rotary lever lock, all of which differ in the details of their mechanism. Examples are known from Caerleon (Manning 1968), the Fishbourne villa (Manning 1971), the Lullingstone villa (Anstee 2001), and the Brading I.O.W 73

Nidum. Excavations at the Roman Auxiliary Fort 2010 to 2012: Neath, Neath Port Talbot CBC

villa which may be from a door rather than a padlock (unpublished). A bolt which appears to be generally similar to that in the Neath lock, and which was probably found in one of the Gloucestershire villas excavated by Samuel Lysons, is in the British Museum (Manning 1985, 95, O65). Horseshoe. Site 677 18. 1000. One side of a horseshoe. Length 102.7mm. (Plate, 25 - H905 Not illustrated). A nail hole survives near the outer edge. The original end is not thickened. Roman horseshoes from the fortress at Usk are discussed in Manning, Price and Webster 1995 (42-44, fig 16, 23-26) where other examples are cited. Loop spike. Site 677 19. 1339. Double-Spiked Loop. Length 51.4mm. (Figure 53, Plate 25 - H906). The loop head has parallel arms one of which is turned out through the right angle to end in a short, pointed tip. The end of the other arm is broken. A very common type cf. Manning 1985, 130, R34-48 and Manning, Price and Webster 1995, 287, nos.36-45, fig. 92. Joinerâ&#x20AC;&#x2122;s dog. Site 677 20. 1338. Joiner's Dog. Length 142.1mm. (Figure 53 Plate, 25 - H908). It has a long, rectangular-sectioned body, which tapers slightly from its midpoint to the ends which are turned down through a right angle to form short spikes, one of which is largely complete, the other broken. The body is now bent into a shallow U-shape which is probably the result of bending when it was removed from the wood to which it was attached. A common type cf. Manning 1985, 131, B52-3 and Manning, Price and Webster 1995, 289-291, nos. 54-69, fig. 93. Plate. Site 716 21. 6588. Reinforcing Plate (?) Length 68.9mm. (Plate, 24 - H913Not illustrated). Thick, slightly irregular plate with one slightly convex edge and rounded ends. It is pierced by one certain nail, and there may have been another but its condition prevents certainty. It probably joined two pieces of wood, although it is surprisingly thick for such a purpose. Fitting. Site 716 22. 6479 SF 009 Pelta-shaped Fitting. (Figure 53, Plate 25 - H905). Length 60.4mm. Lentoidsectioned bar which curves to form two-thirds of a circle, while narrowing in width from its mid-point to the tips of the arms. It appears to be complete but its function is not obvious, although it is clearly not a shoe for an animal or from the heel of a leather shoe. Pin. Site 677 23. 1338. Pin with Discoidal. Length 106.4mm. (Not illustrated). Pin of slightly rounded, rectangular section with a wide discoidal head set on one side of the stem. Its function is not obvious but it is probably too heavy to have been worn on the person.

Nidum. Excavations at the Roman Auxiliary Fort 2010 to 2012: Neath, Neath Port Talbot CBC

Plate 25. X-ray plates of iron artefacts (H906, H906, H908, H909)

Fragments and nails. Site 677 24. 1081. Fragment. Length 69.5mm. (Not illustrated). Fragment of thick rectangular-sectioned bar which widens slightly from one end to the other. The widest end, and probably the other, is broken. A short, tapering arm, which curves downwards and away from the bar, has been welded to one side immediately below the narrower end. It is clearly a fragment of a larger object but too little survives for it to be identified.

Nidum. Excavations at the Roman Auxiliary Fort 2010 to 2012: Neath, Neath Port Talbot CBC

Site 716 25. 6443. Fragment of Rod. Length 135.3mm. (Not illustrated). Fragment of square-sectioned rod with chamfered corners. It is almost certainly a fragment from an artefact rather than a simple rod, but a certain identification is impossible. 26. 6440. Hobnails. Lengths 38.6mm & 42.7mm. (Plate 25 - H906 (Plan and side views)). Two groups of hobnails now corroded together. Their arrangement indicates that they were still attached to a backing, presumably a leather shoe, when they were discarded. The diameter of the heads is c.9.5mm and the length of the stem c.12.4mm. 27. 6537. Tacks and Fragments. (Plate 25 - H909). A small group of tacks, most of which are broken, and other fragments. The most complete tack is 13.5mm long. The heads of the more complete tacks suggest that they are examples of Manning Type 7 tacks (Manning 1985, 135, fig. 32).

8. The metalworking residues (Tables 7-9, Appendix A) By T P Young Summary The two sites produced approximately 31kg of archaeometallurgical residues and associated materials, 26.4 kg of which were from stratified Roman contexts (Table 7). The collection was dominated by residues from iron-working. These included 15.8kg of smithing hearth cakes, 6.9kg of closely related gravelly lining slags, 2.6kg of vitrified hearth lining and 1.6kg of other lining slags. A further 2kg of slags were ‘indeterminate’; in practice mainly materials that were too fragmented to be assigned to one of these classes. The smithing hearth cakes (SHCs) show a weight-frequency distribution that is very similar to those from other sites engaged in the end use of iron (blacksmithing). The Roman smithing was almost entirely undertaken with charcoal fuel – just three examples of possible coal residues within slags were identified. Some 20 Roman contexts yielded small quantities of coke or clinker, indicating the coal was being used – but perhaps mainly for non-metallurgical purposes. The evidence for early iron smelting was extremely limited, with just 200g of possible bloomery iron smelting slags, of which only 14g was in a stratified Roman context. Post-medieval/modern contexts produced a variety of materials, including more common coke and clinker, as well as small amounts of blast furnace slag, alongside residual Roman materials. A programme of more detailed analysis was undertaken on a series of iron-working residues (from context 1502) and on a single piece of problematic dense slag from context 1676 (See Table 8 for details of the samples selected, and Appendix A for the detailed analysis report). The main purpose of the analytical phase was to examine evidence for a genetic relationship between these and more conventional smithing hearth cakes (SHCs). The smithing residues proved to be mostly a coherent suite of materials, the composition of which could be modelled through simple mixture of iron oxide with assimilated hearth lining. One of the examined iron-working slag cakes, an iron-rich, hollow example, did not appear to 76

Nidum. Excavations at the Roman Auxiliary Fort 2010 to 2012: Neath, Neath Port Talbot CBC

fit this model. It appears likely that deep weathering of this piece has led to preferential loss of the interstitial phases and that the iron-rich bulk composition is due to secondary iron oxides formed after the weathering of much metallic iron. It is likely, but not certain, that this slag cake also originated within the same chemical system as the other examples; other interpretations are discussed. The dense slag fragment from within the fort (c1676) is very unusual in being dominated by wustite and becoming more highly oxidised and less iron-rich towards one margin. Firstly, iscorite and magnetite overgrow/replace the wustite dendrites, then there is a zone with magnetite dendrites and finally slag with equant euhedral magnetite. The structure of the slag is layered, with particular horizons demarcated by oxide crusts. These resemble the chilled, iron oxide-rich, margins of flow lobes within a tapped slag, but are thicker (up to 200Âľm) and here may be more indicative of oxidation of growth stages of the slag. This highly unusual slag can be paralleled only by two examples from elsewhere, both interpreted as being derived from bloom-smithing or bloomrefining. The microstructure is suggestive of the slag having formed through oxidation of metallic iron, but is unlike textures observed around iron particles in conventional bloomsmithing slags or adjacent to the bloom in a bloomery. It appears likely that the slag was produced in a form of remelting hearth akin to that described by Evenstad in the 18th century. Such hearths may be used to control the content of carbon (e.g. in hearth steelmaking) and phosphorus (e.g. for dephosphorising phosphoric iron) in iron, but also forms a useful approach to recycling iron, particularly the iron formed in a bloomery but not incorporated into the bloom, or lost from the bloom during processing. Methodology All materials were examined visually with a low-powered binocular microscope where required. Most of the materials were not subjected to any high-magnification optical inspection, not to any form of instrumental analysis, and therefore the identifications of materials in this report are therefore necessarily limited and must be regarded as provisional. For the methodology of the analytical phase, see Appendix A. Results The following sections describe the residues and their distribution. Materials in the sample determined not to be, or not likely to be, of metallurgical origin (e.g. iron artefacts, concretions and natural rock) have been excluded from this descriptive section, but are listed in full in the catalogue (Tables 1 and 2, Young 2013c). Description of materials Coke: these materials are highly porous, low-density, carbonaceous materials. They are the dominantly organic residues of incomplete combustion of coal. It is likely that they will be mainly residues from the use of coal, rather than deliberately manufactured coke. A total of 230g of coke was recovered, with just 56g from contexts interpreted to be of Roman date. Friable red clinker: this material occurs mainly in sheet-like form, up to about 40mm thick. The materials are very friable and contain a high proportion of un-fused shale fragments. It typically has a bright red colour. The material is suggestive of clinker formation during the relatively low-temperature combustion of coke. Such materials are not necessarily metallurgical 77

Nidum. Excavations at the Roman Auxiliary Fort 2010 to 2012: Neath, Neath Port Talbot CBC

in origin – but may arise from uses such coal combustion in steam boilers or during the manufacture of town gas. A total of 202g of this facies of material was recovered, with just 20g of red-coloured clinkers certainly coming from Roman contexts. Other clinker: these materials are more conventional clinkers, showing a higher degree of partial melting of the coal shale inclusions, indicative of higher temperature or more prolonged burning of coal. They typically show a matrix of black glass, containing pale, often bloated, partiallymelted relicts of coal shale and have a maroon to purple surface colour. Clinkers of this type may be the residues from a wide range of processes, both metallurgical and non-metallurgical. In a metallurgical context, they indicate a process where the metal contributes little to the residue, which is composed almost entirely of the fuel residue. This may be the case in much light blacksmithing work, from which forge clinkers may be almost indistinguishable from those formed in steam boilers for instance. A total of 1217g of conventional clinker was found, of which 129g came from contexts of Roman age. Gravelly lining slag: this facies of residue is formed of rounded masses of partially melted ceramic and stone bound by dark (usually), sandy, lining slag. In some instances, the stone clasts are glazed with a thin transparent green glass and similar isolate glazed stones were found. The green glaze is indicative of involvement of only wood-ash as a flux. This material is related to, and gradational with, the low-iron varieties of SHC described below – except that they have not amalgamated into the characteristic morphology of an SHC or tongue. Some 6.9kg of this class of residue was recorded, approximately 23% of the overall collection. Smithing hearth cakes (SHCs): SHCs provided the largest class of material from the collection with a total of 15.8kg. There were 47 examples of SHCs which were complete, or sufficiently so for the original weight to be estimated. They ranged from 74g to 630g, with a mean weight of 244g. Both the complete examples and the substantial collection of fragments showed a very wide range of morphologies and textures. At one end of the spectrum the SHCs embraced examples similar to the ‘tongues’ described by Young (2012a), which form as pad-like slag cakes extending forward of the blowhole and comprising an upper siliceous, often glassy, layer derived largely by melting of the hearth wall. As this slag mass reacts with iron oxides derived from the workpiece, it generates more fluid melts (of a broadly fayalitic composition) which may drip from the tongue and form pendent, descending, iron-rich, slag prills from its lower face. In the case of the low-iron SHCs from Dwr-y-Felin, the melting of the hearth wall led not only to the formation of a viscous pad, but also to the covering of the pad by the less easily melted gravel-grade stone clasts from the hearth walls. These gravel clasts variously bloat and melt, leading to some extremely inhomogeneous textures on the upper face of these slag cakes. These tongues are closely related to the ‘gravelly lining slags’ described above – except that these lack the characteristic tongue/SHC morphology. As the iron content of these tongues/SHCs rises, so the textures and morphologies tend towards more conventional SHC forms (usually oval in plan and with a plano-convex form, but sometimes concavo-convex if the dense bowl is not filled with slag, or biconvex if slag 78

Nidum. Excavations at the Roman Auxiliary Fort 2010 to 2012: Neath, Neath Port Talbot CBC

accumulates above the bowl). The fluid slags form a discrete, more-or-less well formed, lower bowl, usually with a dense crust. In some cases, these bowls are filled with material similar to that of the tongues, in others much more iron rich–slag, eventually leading to examples where the SHC comprised a very fluid slag ‘puddle’. The controls on the formation of these different forms of SHCs are very poorly understood. Most will derive from forging at different temperatures, for different lengths of time, with different sizes of work-pieces. The tasks involved in the forging operation may involve a greater or lesser involvement of forge welding. All of these factors will feed into the variety of SHCs produced during the process. Most SHCs showed clear evidence of charcoal as fuel, but one fragment (context 2148) shows a surface coating of shale fragments, possibly derived from the use of coal fuel. Several pieces of SHC (most notably from context 1076) show basal tool marks (expressed as steep side ridges on the base of the slag cake), where a poker has been used to manipulate or remove the slag. Many pieces show strong deformation from being removed from the hearth when hot, including being folded in half. Vitrified hearth lining: 2.6kg of fragments of vitrified oxidised fired lining were found. The collection includes several significant pieces from the blowhole area, and taken in conjunction with the evidence from vitrified hearth lining adhering to the end of some of the SHCs, some evidence for the blowhole can be obtained. Interestingly evidence suggests that in some cases the blowhole was, at very least, on a ‘boss’-like protrusion from the wall (e.g. context 6577; demonstrated by the convexity of the face of the wall near the area of attachment of the SHCs). Two other pieces show evidence for the sharp margins of the protrusion (contexts 1333 and 2148) which show the ‘boss’ may have been about 100mm across – and indeed may just possibly have been in the form of a ceramic tuyère. Other lining slag / FAS: there was 1.6kg of material in the form of slags largely derived from the hearth lining, which did not fall into the above categories. This was mainly material analogous to the ‘gravelly lining slags’, but simply lacking the partially meted stone clasts. Some lining slag pieces have a curved, sheet-like morphology, suggesting, like some of the SHCs, that they were pulled from the hearth whilst hot and plastic. Tapped bloomery iron-smelting slag: possible bloomery iron smelting slag provided just 199g of the assemblage – and just 14g of that was in a Roman context. Some of the material in post-Roman contexts may be derived from other industrial processes as was suggested for some material from Dwr-y-Felin site 674 (Young 2013b). The material from the current site is mainly present as very small prills and lobes, and whilst consistent with an interpretation as tapped bloomery slag is not certainly so; other processes can generate fluid fayalite slags with maroon surface oxidation. Other dense slag: this category has been used here for just a single piece of slag (58g; context 1676), which appears to comprise parallel dendrites of iron oxide (wustite?) and little else. Such materials may form as residual crusts on the interior of some industrial furnaces, but this piece was recovered from a Roman context. Identification of its origin must await further investigation. 79

Nidum. Excavations at the Roman Auxiliary Fort 2010 to 2012: Neath, Neath Port Talbot CBC

Blast furnace or cupola slag: 109g of glassy slag and pale crystalline slag are compatible with an origin in a blast furnace (post-medieval iron smelting) were recovered. Several of these pieces appeared water-worn and all were recovered from modern contexts. Indeterminate slag: material in this category (2kg) lacks sufficient diagnostic features to be able to be attributed to a particular class of residue. Much of this category is probably smithing slag lacking in evidence for its occurrence in an SHC morphology. Distribution of the residues The distribution of residue classes by context is given in Table 7. There is a widespread background scatter of slag across the site and, at present, that appears to show little pattern. There are, in addition, several contexts with rather richer assemblages: 1. The intervallum area and in particular the makeup for the via sagularis: this area shows a consistent low level of slag – probably residual material incorporated into the make-up layers of the road (1.4kg). However, other sources may be present and the significant amount (1.8kg) of slag present in context 1278 (fired clay deposit part of oven 1023) is suspicious. The 1.2kg of slag present in context 1414 (a yellow silty clay) is also noteworthy. 2. The ‘mansio’ area has several contexts with important assemblages, including contexts 1076, 1111, 1276, 1387 and 1502 totalling 13.8kg. This is slightly more than half of all archaeometallurgical residues recovered from stratified Roman contexts (a total of 26.4kg). This raises the likelihood of their having been a smithy in this area of the site (despite the fact that 13.8kg is still a fairly small assemblage, it is still indicative). 3. The assemblage from posthole 1008 is also noteworthy for an assemblage of almost 1kg of smithing debris. Another observation from the residue distribution is that although coke and clinker both occur in Roman contexts (there are 20 contexts with either or both of these, totalling 56g of coke and 129g of clinker), there is no particularly strong relationship to the contexts containing other metalworking evidence. Just three slag samples apparently contained coal residues – an SHC from context 2148 with a scatter of shale inclusions, an incipient slag ‘tongue’ from context 1387 and a fuel ash slag from context 1844 (which is not necessarily metallurgical). Thus although coal may have been used in a minor role as a smithing fuel, it seems likely that it may have been used more frequently in other roles. Material from context 1075 may be worthy of note, since it is an assemblage that is not markedly different from other smithing assemblages on this site. The context has been provisionally labelled as pre-Roman, which in the basis of the similarity of evidence may be unlikely (although not impossible). Interpretation The assemblage from the site is almost entirely of residues from iron-working (blacksmithing). Other residues include some material probably from bloomery iron smelting, that was residual in later contexts, with just a single 14g piece from a Roman context (and certain identification of that as bloomery tapped slag is not possible on such a small fragment).

Nidum. Excavations at the Roman Auxiliary Fort 2010 to 2012: Neath, Neath Port Talbot CBC

Post-medieval/Modern contexts produced a variety of materials including clinker (not necessarily of metallurgical origin) and a few fragments of blast furnace slag. The majority of the assemblage was, however, of blacksmithing residues. Those found in later contexts was similar to that from stratified Roman contexts – and therefore it seems likely all the blacksmithing residues are Roman. One parameter allowing comparison of smithing residue assemblages is the weight-frequency distribution of the SHCs. Summary statistics are presented in Table 9, where they are compared with those from other contemporary assemblages. The Dwr-y-Felin assemblage of SHCs is very similar to those from a group of sites (Carmarthen, Crew 2003; Bulmore, Young 1999; Marsh Leys Farm, Young 2005; Cowbridge, Barford 1996) which appear to have derived from general purpose blacksmithing activities. Comparative data from sites engaged in bloomsmithing are sparse, but evidence from sites at Kingstone, Herefordshire (Young 2012c), Cardiff Castle (Young & Kearns 2011), Dymock, Gloucestershire (Young & Kearns 2010b) and Miskin (author’s unpublished data) all suggests that SHCs from bloomsmithing commonly range up to several kg in weight, with a similarly marked rise in the average weight of the SHCs. Further examples of bloomsmithing slags are probably to be found in the examples cited by Allen (2009, 2010) but which he erroneously classified as furnace bottoms from non-slag tapping bloomery furnaces. Bloomsmithing may therefore be discounted as a significant activity at Dwr-y-Felin. The Dwr-y-Felin smithing assemblage is particularly interesting because the slag cakes span the range of morphologies of conventional SHCs, of lining slag cakes similar to those termed ‘tongues’ by Young (2012a), and of the irregularly-shaped gravelly lining slag lumps. These gravelly lining slags have been observed in other Roman assemblages (Caergwanaf, author’s unpublished data; Caerleon Priory Field, Young 2013a; Caerleon Endowed Schools, Young & Kearns 2010a; Exeter, Young 2011c; Calstock, Young 2012b; Neath, Young 2013b), but in isolation have not been able to be distinguished from residues of non-ferrous metalworking. Insufficient Roman civilian sites have been studied in detail to allow comment on whether the occurrence of the gravelly lining slags on all these Roman military sites is of particular significance. Where these slags have been recorded previously, there has been a degree of uncertainty over whether these iron-poor residues were the result of ferrous or non-ferrous metalworking. The spectrum of compositions at Dwr-y-Felin strengthens the argument that these are iron-poor iron-working residues. The relative paucity of evidence for coal-fuelled smithing is a slight distinction from some of the other comparative assemblages from South Wales. At site 674 at Dwr-y-Felin (Young 2013b), the general scatter of smithing debris was similar to that in the present study and mainly charcoal-fuelled, but there was a relatively late hearth that provided good evidence for coalfuelled smithing. At Caerwent (Young 2006), the late (4th century) phase of smithing within the basilica appeared to be entirely coal-fuelled. The 2nd century smithing at Bulmore was largely coal-fuelled (Young 1999). The evidence from the blacksmithing activity on the site at Cardiff Castle (Young & Kearns 2011) also suggests the major, or sole, use of coal. Even on the late Roman rural site at Trowbridge (Young 2009a, 2009b), the smithing appeared to have been entirely coal-fuelled. 81

Nidum. Excavations at the Roman Auxiliary Fort 2010 to 2012: Neath, Neath Port Talbot CBC

Perhaps the most similar pattern of fuel use to that of the present site within the area is the military iron working within the fortress at Caerleon (Endowed Schools field, Young & Kearns 2010a; Priory Field, Young 2013a), which shows evidence for the use of a mixture of fuels, but with somewhat stronger evidence for coal use than at Dwr-y-Felin. In cases where thorough and intensive sampling for microresidues has been undertaken within Roman military sites (e.g. Calstock and Exeter, St Loyes, Young 2012, and 2011c respectively), the pattern of distribution of hammerscale within the interior of the forts has suggested locations of ironworking that would not have been recognised on the basis of macro-residues alone. Workspaces inside the fort would have been kept clean, with macroscopic slags moved elsewhere for disposal. This should be borne in mind when interpreting the scatter of macroresidues from the internal areas at Dwr-y-Felin. Whilst it is currently assumed on the known pattern of distribution that ironworking took place outside the fort, in the â&#x20AC;&#x2DC;mansioâ&#x20AC;&#x2122; area, the possibility of external disposal of waste created inside, should also be considered.

Nidum. Excavations at the Roman Auxiliary Fort 2010 to 2012: Neath, Neath Port Talbot CBC

677 u/s

716 u/s

110 2

1000

Machine clearance finds

1001

Topsoil

1002

Subsoil

1003

Intervallum

Finds in topsoil over (via sagularis) 1004

1004

Intervallum

Internal road (via sagularis)

1006

Intervallum

Mid brown silty clay deposit

1009

Rampart

Fill of posthole 1008

1021

Rampart

1029 1030

184

total

indeterminate slags 32

221

214

6 338 102

Roman

218

Mid brown silty clay

Roman

Intervallum

Upper fill of 1026, orange-brown silty clay. Pit group 1038

Modern

Intervallum

Upper fill of 1025, dark brown/black silty clay loam

Modern

6 26

blast furnace slag

other dense

other lining/FAS

lining

SHC

gravelly lining slag

other clinker

friable red clinker

coke

Provisional age

Context notes

Area

Context

Table 7. Summary of residue classes by context (weight in g)

507

120 40

18 79

480

66 1

255

109

914 1

118

119

780

122

total

indeterminate slags

blast furnace slag

other dense

other lining/FAS

lining

SHC

gravelly lining slag

other clinker

friable red clinker

coke

Provisional age

Context notes

Area

Context

Nidum. Excavations at the Roman Auxiliary Fort 2010 to 2012: Neath, Neath Port Talbot CBC

1031

Intervallum

Lower fill of 1026, dark brown/black silty clay loam. Pit group 1038

Modern

122

1044

Trench 2 ditch

Ditch fill, clay between 1043 and 1045 Trench 2

Roman

1046

Trench 2 ditch

Ditch fill

Roman

1052

Intervallum

Charcoal rich clay deposit (? fuel oven 1023)

Roman

1053

Intervallum

Clay and pebble deposit between oven 1007 and 1052

Roman

1054

Intervallum

Dark grey-brown silty clay deposit . Pit group 1037

Modern

1065

Intervallum

Charcoal rich clay deposit (? fuel oven 1023)

Roman

1073

Extramural

Shallow depression -possible posthole/rooting (fill 1073)

Natural

1075

Extramural

Sandy clay subsoil occupation level

Pre-Roman

1076

Extramural

Dark brown silty clay deposit, near mansio

Roman?

1077

Extramural

General finds number between road 1061 and the mansio

1081

Extramural

Clay deposit fill of 1085

Modern

1083

Extramural

Stone deposit, mansio?

Roman

142

172 20

208

114

402

382

3078

170

3696

364

396

828

154 76

total

indeterminate slags

blast furnace slag

other dense

other lining/FAS

lining

SHC

gravelly lining slag

other clinker

friable red clinker

coke

Provisional age

Context notes

Area

Context

Nidum. Excavations at the Roman Auxiliary Fort 2010 to 2012: Neath, Neath Port Talbot CBC

1084

Extramural

Clay loam, post-med ditch fill

Post-med

1086

Extramural

Clay and loose stone deposit (mansio area)

Post-med

1088

Extramural

Stone linear, wall?

Post-med

1089

Extramural

Stone spread wall?

Post-med

1096

Extramural

Pebble and silty clay deposit

Roman

1104

Extramural

Clay deposit over NW/SE road 1061

Post Roman

1110

Extramural

Stone linear near mansio

Roman?

1111

Extramural

Charcoal rich clay between 1061 and 1110

Roman?

1112

Extramural

Cut for drain 1103

Modern

1117

Intervallum

Charcoal rich clay

Roman

1128

Intervallum

Timber slot

Roman

1131

Intervallum

Fired clay charcoal and stone

Roman

1141

Extramural

NW/SE road same as 1061

Roman

1142

Extramural

Clay deposit overlying road 1141

Roman

1170

Intervallum

Clay deposit underlying 1053 (ovens area)

Roman

1201

Intervallum

Light brown clay deposit

Roman

1227

Intervallum

Dark brown silty clay underlying 1006

Roman

62 1

274 2

274

140

256

164 62

1134

102

84 48

18 1

1540

18 1

134 4

404

404 104

104 6

306

1228

Intervallum

Finds number (area between 1101 and oven 1052)

Roman

1247

Extramural

Finds number for clay silts between road 1061, post-med drain and excavation edge

Post Roman

1270

Extramural

Posthole/ditch/depression (fill 1271)

Roman

1276

Extramural

Clayey sand and stone, walling/structure part of â&#x20AC;&#x2DC;mansioâ&#x20AC;&#x2122;

Roman

492

816

1278

Intervallum

Fired clay deposit part of oven 1023

Roman

296

1336

1320

Extramural

Clay deposit under foundation 1100

Roman

1332

Extramural

Mixed stone and sediments within depression 1331

Roman

348

1333

Extramural

Orange brown silty clay and cobble deposit

Roman

194

1338

Extramural

Mid brown grey silty clay deposit overlying 1337

Roman

1346

Extramural

Linear stone scatter in clayey silt

Roman?

1387

Extramural

Linear deposit of stone - wall structure mansio related?

Roman

1414

Intervallum

Yellow silty clay

Roman

1439

Extramural

Yellow clay loam deposit

Roman

60 20

112

170

1590 176

396 284

128

370 4

444 246

1844

474

220

total

indeterminate slags

blast furnace slag

other dense

other lining/FAS

lining

SHC

gravelly lining slag

other clinker

friable red clinker

coke

Provisional age

Context notes

Area

Context

Nidum. Excavations at the Roman Auxiliary Fort 2010 to 2012: Neath, Neath Port Talbot CBC

444 194

154

142

1210

1440

Extramural

Yellow brown sandy loam deposit

Roman

1494

Extramural

Dark brown clay deposit

Roman

1502

Extramural

Orange/brown silty clay deposit. Iron waste slag in deposit similar to ditch 1481 fill (i.e. 1076)

1506

Extramural

1550

total

indeterminate slags

blast furnace slag

other dense

other lining/FAS

lining

SHC

gravelly lining slag

other clinker

friable red clinker

coke

Provisional age

Context notes

Area

Context

Nidum. Excavations at the Roman Auxiliary Fort 2010 to 2012: Neath, Neath Port Talbot CBC

247

288

6954

142

Roman

1640

3800

796

430

Cleaning context southern corner site (same as 1494)

Roman

Extramural

Grey-brown clay

Roman

1551

Intervallum

Yellow brown clay deposit

Roman

128

1555

Recess

Grey clay deposit, fill

Roman

1676

Intervallum

Mid brown silty clay fill levelling up/oven debris

Roman

1708

Intervallum

? decayed turf deposit underlying 1707

Roman

1731

Intervallum

Spread of oven clays and charcoal

Roman

1779

Intervallum

Gravel surface

Roman

1822

Intervallum

Fill of pit 1840

Roman

1844

Intervallum

Clay deposit overlying surface 1841

Roman

2148

Intervallum (section)

Light brown clay with yellow/grey hue, Fe stained, fill of pit 1712

Roman

90 74

164

330 6 58

24 38

38 2

326

404

6000

External works unstratified finds number

124

total

indeterminate slags

blast furnace slag

other dense

Roman

other lining/FAS

Sandy clay fill of pit 5036

342

lining

Trench 5 extraction pit

SHC

5006

gravelly lining slag

Roman

other clinker

Dark brown clay loam and gravel upper pit fill

friable red clinker

Trench 5 extraction pit

coke

Context notes

5001

Context

Area

Provisional age

Nidum. Excavations at the Roman Auxiliary Fort 2010 to 2012: Neath, Neath Port Talbot CBC

592

6023

Intervallum

Clay deposit between via sagularis and rampart

Roman

6027

Intervallum

Clay deposit in pit 6025

Roman

6030

Intervallum

Clay deposit in pit 6025

Roman

148

6042

South strip

Clay loam topsoil - playing field

Modern

6044

East strip

Subsoil similar to 6046

6045

South strip

Clay and stone deposit. General soil build up

Post-Roman

6049

Internal

Clay deposit, pit 6512

Roman

6053

Internal

Clay and pebble deposit

?Post Roman

6066

Intervallum

Clay deposit with pebbles and sandstone, same as 6067

Roman/Post Roman

16 16 14

2 32

14 2

116

149

164

total

indeterminate slags

blast furnace slag

other dense

other lining/FAS

lining

SHC

gravelly lining slag

other clinker

friable red clinker

coke

Provisional age

Context notes

Area

Context

Nidum. Excavations at the Roman Auxiliary Fort 2010 to 2012: Neath, Neath Port Talbot CBC

6067

Intervallum

Clay deposit with pebbles and sandstone, same as 6066

Roman/Post Roman

6070

Internal

Metalled surface adjoining the via sagularis

Roman

6089

Intervallum

Clay deposit with pebbles, charcoal and slag within via sagularis make-up

Roman

6095

Intervallum

Deposit making up part of the via sagularis

Roman

6096

Intervallum

Clay deposit within via sagularis

Roman

6110

Internal

Silty clay deposit, fill of gully 6111

Roman

6182

Extramural

Unstratified finds number extramural area

Mixed, Post Med/Roman

6197

Extramural

Greyish orange silty clay deposit

Roman

6226

Extramural

Grey clay deposit

Roman

6240

Extramural

Grey-brown clay deposit

Post-Med

6245

Extramural

Grey clay deposit

Roman

6255

Extramural

Charcoal rich clay deposit

Roman

6258

Extramural

Charcoal rich silty clay deposit

Roman

6259

Extramural

Stone fill in ditch 6284

Post-Med

496

126

364

1036

48 176

316

10 22

402

193 2

17 1

total

indeterminate slags

blast furnace slag

other dense

other lining/FAS

lining

SHC

gravelly lining slag

other clinker

friable red clinker

coke

Provisional age

Context notes

Area

Context

Nidum. Excavations at the Roman Auxiliary Fort 2010 to 2012: Neath, Neath Port Talbot CBC

6277

Extramural

Brownish-orange silty clay deposit fill of ditch 6284

Post-Med

6280

Internal

Clay deposit

Roman

6422

Internal

Brown-grey clay deposit with pebbles.

Roman?

6437

Internal

Clay and stone fill of ditch 6578

Post Roman

6438

Internal

Sandy clay deposit fill of posthole 6439

Post-Med

6444

Internal

6445

Intervallum

Mid grey-brown silty clay deposit

Post Roman

6448

Internal

Charcoal deposit in ditch 6514

Roman

6479

Intervallum

Sandy clay fill of pit 6035

Roman

6508

Internal

Dark brown silty clay underlies road 6059

Roman

6533

Internal

Dark brown silty clay deposit

Roman

6537

Internal

Silty clay fill of a gully 6746

Roman

6547

Intervallum

Metalled surface

Roman

6557

Internal

Lower deposit of 6059 via sagularis

Roman

6577

Internal

Wall/structure. Linear platform of stone

Roman

6588

Internal

Light brown grey sandy clay

Roman

4 2

2 8

1 54

156

282

254 282

1 1

190

191

278

278 38

6621

Internal

Stone and silty clay, fill of posthole 6620

Roman

6773

Internal

Mid greyish brown deposit (? occupation layer)

Roman?

7059

Intervallum

Dark brown, pebbly, silty clay deposit with charcoal

Roman?

total

indeterminate slags

blast furnace slag

other dense

other lining/FAS

lining

SHC

gravelly lining slag

other clinker

friable red clinker

coke

Provisional age

Context notes

Area

Context

Nidum. Excavations at the Roman Auxiliary Fort 2010 to 2012: Neath, Neath Port Talbot CBC

20 74

74 12

totals

230

202

1217

6892

15755

2600

12 1585

199

109

1962

30809

Table 8. Metalworking residues: Details of samples selected for further analysis. Sample Context

Wt of Spec.

Specimen description

Subsample notes

NTH1

1502

22g

hearth lining

Bulk chemistry subsample: unvitrified section of wall

NTH2

1502

130g

maroon-surfaced gravelly tongue attached to vitrified lining

SEM subsample: dense slag with edge of lining material, Bulk chemistry subsample: bulk tongue

NTH3

1502

218g

curved gravelly sheet-like tongue with pale glass

SEM subsample: clast/slag boundary, Bulk chemistry subsample: bulk

NTH4

1502

106g

well-formed tongue (or incipient SHC) with dark glassy top, rusty base

SEM subsample: grey slag, Bulk chemistry subsample: bulk

NTH5

1502

270g

SEM subsample: lower crust Bulk chemistry subsample: bulk

NTH6

1502

200g

pale green glassy-topped SHC, rusty hemispherical base, minor explosion from iron corrosion, fuel dimples in top. Hollow. part of pale dense SHC with possible basal tool marks

NTH7

1676

56g

dense angular slag fragment

Bulk chemistry subsample: bulk

SEM subsample from lower part â&#x20AC;&#x201C; does not include the pale lining-influenced material on top, Bulk chemistry subsample: bulk

Nidum. Excavations at the Roman Auxiliary Fort 2010 to 2012: Neath, Neath Port Talbot CBC

Table 9. Weight distribution of SHC assemblage from Dwr-y-Felin, with selected comparative assemblages. Weights in g.

Blacksmithing only Dwr-y-Felin

Carmarthen

Bulmore

Neath

With bloomsmithing

Marsh Leys Farm

Cowbridge

Plot 331,

Cardiff Castle

Dymock

Kingstone

677 & 716 This project

Crew 2003

Young 1999

Young 2005 & 2011b

Barford 1996

Young 2012c

Young & Kearns 2011b

Young & Kearns 2010b

No. of SHCs

136

Min weight

100

127

175

230

193

112

Max weight

630

820

770

824

700

1035

1090

3885

Average weight

244

227

270

333

403

565

626

1032

<500g

89%

94%

77%

47%

40%

60%

<1000g

100%

93%

80%

86%

>1000g

20%

14%

>3000g

% of SHCs with weight:

Nidum. Excavations at the Roman Auxiliary Fort 2010 to 2012: Neath, Neath Port Talbot CBC

9. Archaeometallurgical residues from environmental samples Site 716 (Table 10) By D D Dabal The material described was recovered from the residues of the palaeo-environmental samples collected from excavations of Dwr-y-Felin Comprehensive School, Neath - Port Talbot. The examined materials formed a small assemblage (approximately 1228.2g), which were selected from 31 contexts. The majority of them dated to Roman times, apart from the materials found in one Post-medieval context 6440. The archaeometallurgical finds clearly demonstrate that metallurgical activity took place on the site in Roman times. Within the collection, there are residues which might be linked with blacksmithing. One example of a low-iron SHC was washed from sample 055 (context 6790) and small quantities of flake and spheroidal hammerscales were present in half of the analysed contexts. Also, one context (6089) yielded small fragment of tap slag that suggests iron smelting with use a bloomery furnace, probably slag-tapping furnace. Yet, such a small amount of the smelting residue (9.8g) indicates that it took place somewhere else. Method The materials described were recovered from the residues of the palaeo-environmental samples from site 716. All specimens were inspected visually, using 40x magnification microscope where necessary. Examined materials were described and recorded into a database. The materials were not subjected to any form of instrumental analysis, therefore their identification and interpretation are limited and must be regarded as provisional. The following paragraphs describe the materials and their distribution. Residues which are not of metallurgical origin (i.e. natural concretions or rocks) have not been discussed below. However, they all are listed in the full archive catalogue as a list arranged by sample number (Project archive), and secondly with the same information organize by context number and residues classes (Table 10). Results The examined materials, weighing approximately 1228.2g, were derived from 31 contexts. They were divided into the following classes of residues and described using the terminology adopted by Young (2013c) (Appendix A). This report supplements that report. Iron smelting slag In general, the most characteristic remains after iron smelting with use of a boomery furnace are slags which were tapped out of the furnace and subsequently freeze outside. In context 6089 (sample 005) one small fragment of such slag was found (9.8g). It was dense with characteristic lobbed top and charcoal marks at the base. This fragment of bloomery tapped slag was similar to the slags recorded in other excavated deposits: 5001 site number 677 (Young 2013c).

Nidum. Excavations at the Roman Auxiliary Fort 2010 to 2012: Neath, Neath Port Talbot CBC

Iron smithing slag The most distinctive evidences of iron working are smithing hearth cakes (SHC). They vary in size and weight as well as in texture. They are usually plano-convex to concavo-convex in section and circular or oval in shape. ne example of a low-iron SHC was examined from sample 055 context 6790 (131g, dimension: 70 x 50 x 40mm). It had oval shape, plano-convex section, partly glassy surface with incompletely melted stone debris, and a fragment of burr attached to the side. At the base the slag showed evidence of charcoal used as a fuel. It also had dents at its bottom and side, most likely caused because it was pulled out of the hearth when hot. It is generally agreed that, the size of the smithing slag depends on the amount of forged iron and the small SHC are product of secondary smithing, which doesnâ&#x20AC;&#x2122;t cause large loss of metal, unlike bloom-smithing. In the case of the low-iron SHCs, it is believed that they have been largely formed by the melting material from the hearth walls during light iron-working. Similar SHCs were found in several contexts of the Roman forts. These slags ranged from 74g to 630g and were poorly developed masses with various morphologies and textures (Young 2013b). These SHCs characterised by inhomogeneous textures with top surface showing less easily melted gravel-grade stone clasts and often glassy siliceous layer (Young 2013b). Lining slag This class of material is related to the low-iron type of SHC, which had not developed to the characteristic form of SHC (Young, 2013b). In sample 005 from context 6089 many fragments of iron-poor lining slags had been found (total weight 202g). This material occurred in the form of low density rounded masses of slags (usually dark grey in colour) which had been formed by partially melted ceramic from the hearth walls. The largest fragment had diameter of approximately 57mm. Most of these poorly developed slags had a grainy texture caused by melting of silica rich material. Also, many slags had marks indicative of charcoal use as a fuel and small rusty inclusions. Some slag fragments had a glassy surfaces in white/green/dark green colour. The green glaze may suggest that wood-ash was used as a flux. Vitrified lining Generally this class of materials form in iron smelting furnaces and smithing hearths, but also can form in hearths where non-ferrous metals are worked. In sample 005 context 6089 six fragments of fired clay with vitrified surfaces (total weight 12g) were found. They were characterised by red oxidised or grey reduced clay with mineral temper on one side and dark glossy glaze on the other surface. These finds were likely to be parts of a metallurgical hearth or kiln. However, the fragmentation of the material didnâ&#x20AC;&#x2122;t allow any additional information to be obtained as the largest piece had diameter of just 25mm. Indeterminate slag Many fragments of broken slags or slag prills were too small to allow their identification (total of 87.2g, sample 005 context 6089). The largest piece had a dimension no greater than 20mm. Most of these slags had low density and were characterised by a slightly glassy surface with 94

Nidum. Excavations at the Roman Auxiliary Fort 2010 to 2012: Neath, Neath Port Talbot CBC

partly melted silica rich material. It is very likely that some of these indeterminate slags were fragments of poorly developed SHCs or lining slags formed within a smithing hearth. Microscopic smithing residues Smithing micro-residues (fines) are produced when metal is forged or welded. On the one hand these fines can form when metal is forged or welded on the anvil and they pile up on the floor or outside the workshop. On the other hand these fines can accumulate within the hearth where iron is heated up by a smith. The most characteristic form of the smithing residues are two types of hammerscale: flake hammerscale (HSF) and spheroidal hammerscale (HSS). Flake hammerscale is an iron oxide scale which forms on the surface of worked iron and it is broken off from the metal by hammering, brushing or simply tapping. Spheroidal hammerscale is formed by liquid slag particles which are expelled by hammering when two fragments of metal are forged together (Mills & McDonell 1992). Small quantities of flake and spheroidal hammerscale were found in sixteen out of thirty-two samples from Dwr-y-felin site 716. The majority of the hammerscale was present in the form of flakes which had been found in twelve samples. Most of the flakes were very small (<2mm) and thin (<0.5mm). Within two samples, 005 (context 6089) and 019 (context 6472), larger hammerscale flakes were recorded (max. 4mm). Small amounts of various size hammerscale spheroids, made of air chilled droplets of slag or expelled molten iron oxides, were found in seven samples (contexts: 6026, 6030, 6038, 6079, 6089, 6170, 6683). Most of them were rounded particles with a metallic lustre and diameter between 1.5mm - <0.5mm. These materials were classified as spheroidal hammerscale, yet examination of their surfaces did not eliminate the possibility that they might be slag droplets expelled from iron when in was worked. Indeterminate microscopic iron-working residues A large part of the magnetic residues recovered from the paleo-environmental samples were formed from naturally magnetic rocks particles and fragments of various slags, which were too small to allow their precise interpretation. Slag flakes and blisters were found in eight contexts (6021, 6079, 6089, 6104, 6434, 6544, 6558, 6790). Most of them had a dark colour with metallic lustre or were rusty on one side and showed broken vesicles on the other. Some were characterised by white/green/dark green glaze, others had black/maroon surface colours (sample 005 context 6089). These thin slag films were slightly thicker than hammerscale with a maximum diameter of 0.6mm. The largest amount (0.6g) was recorded in the context 6089. Slag droplets, some rounded and the other slightly dented had been found in thirteen contexts (6021, 6030, 6037, 6038, 6079, 6089, 6115, 6431, 6434, 6544, 6558, 6595, 6790). They ranged in diameter from 1.5mm to 4mm. The largest amount of slag spheroid was found in sample 005 context 6089. In general, they showed a variety of surface colours and lustres. Slag spheroids and sub-spheroids are larger than spheroidal hammerscale. They might be interpreted as isolated droplets of slag, which didnâ&#x20AC;&#x2122;t connect with main SHC which formed in the hearth, and cooled down in the fuel bed of the smithing hearth. However, very similar residues are also 95

Nidum. Excavations at the Roman Auxiliary Fort 2010 to 2012: Neath, Neath Port Talbot CBC

formed by interaction of fuel with hearth walls in various non-metallurgical contexts. As had been recorded, spheroidal micro-residues were found within ash from cereal-drying kilns (Young 2005, 2015, 2016). Therefore it is possible that some of the micro-residues may have come from kilns or hearths where a high temperature was generated. Distribution of the residues The described residues were from Roman contexts, apart from the materials found in context 6440, dated as Post-medieval (the distribution of material classes is presented in Table 1). Most of the deposits were recovered from the earliest forts - 1st century timber forts (Tuck & Crawford 2010). Contexts 6258 and 6287 were located in the vicus, a town or village adjacent to a fort, under a military control. The largest amount of material was found in context 6089 which was a clay deposit with pebbles, charcoal and slag within the make-up of the internal road via sagularis. Interpretation The archaeometallurgical residues recovered from the environmental samples formed a very small collection of finds which clearly demonstrated that metallurgical activity took place close to the excavated part of the site. Within the collection there were materials which might be linked with blacksmithing. One example of a low-iron SHC was washed from sample 055 (context 6790) and small quantities of flake and spheroidal hammerscales were present in half of the analysed contexts. Mills & McDonnell (1992) noted that micro residues stay close to a smithy or slag dump, and are not discarded or reused like other iron production wastes. The small amount of hammerscale detected within the samples suggested the existence of a nearby blacksmithing workshop. It is worth noting, that the low-iron SHC and low density lining slags are similar to the materials recovered in other deposits on the site, previously described by Young (2013b). Also, at the other Roman military sites (like Caerleon, Caergwanaf Uchaf in Miskin and Cardiff) comparable slags were found (Young and Kearns 2011, and Young 2013). One context (6089) yielded a small fragment of tap slag that suggested iron smelting with use a bloomery furnace, probably slag-tapping furnace. But, such a small amount of the smelting residue (9.8g) indicated that it took place away from the site. The materials described above as indeterminate slags and other micro-residues could be of metallurgical origin and perhaps were formed in a smithing hearth. Unfortunately, the majority of materials, particularly various slag fragments, are too fragmented to allow their precise interpretation.

Nidum. Excavations at the Roman Auxiliary Fort 2010 to 2012: Neath, Neath Port Talbot CBC

Roman

6037

Intervallum

Clay fill deposit in pit 6035

Roman

Intervallum

Charcoal rich deposit on north edge of via sagularis

Roman

6038 6079

Intervallum

6089

Intervallum

Clay and pebble deposit, Fe stained, fill of 6093 Clay deposit with pebbles, charcoal and slag within via sagularis make-up

6104

Intervallum

Clay deposit in 6025, lining?

Roman

6115

Intervallum

Silty clay deposit, fill of pit 6121

Roman

6115

Intervallum

Silty clay deposit, fill of pit 6121

Roman

6170

Internal

Mixed yellowish-white deposit of silty clay

Roman

6177

Intervallic

Charcoal rich deposit possibly part of 6176

6258

Extramural

Charcoal rich silty clay deposit

6287

Extramural

6338

Intervallum

Charcoal rich clay deposit fill of 6321 Mid-grey silty clay drain fill of drain 6341

5 <1?

Roman

2.6

<1?

Clay

Clay deposit in pit 6025

Char-coal

Intervallum

Coal/ Coke

6030

Total:

Concretion

Roman

Other magnetic material

Charcoal deposit defining edge of pit 6025

Slag microresidues

Intervallum

SSS

6026

SFB

Roman

HSF

Orange grey clay deposit rear of rampart

Other slag

Rampart

Clinker

6021

Vitrified lining

Period

Lining slag

Context notes

SHC

Location

Bloom. slag

Context

HSS

Table 10. Microresidue classes by context Dwr-y-Felin School, Neath (Site 716)

3.2

14.6

17.8

2.2

17.8

<1?

0.6

319.6

0.6

4.6

Roman

9.8

202

87.20

<1 4

Roman

<1? <1

Roman

6433

Internal

Fired clay deposit

Roman

6434

Internal

Mid brown sandy clay deposit fill of posthole 6454

Roman?

6440

Intervallum

Charcoal rich clay

6449

Internal

Roman

6472

Internal

6544

Internal

6558

Internal

Fired clay deposit Mid grey brown sandy clay at the base of a posthole 6472, contained by 6454 Grey brown clay and stone fill of slot 6587 Dark brown grey clay, fill of slot 6615

6564

Internal

Charcoal deposit

6593

Intervallum

6595

Intervallum

Grey silty clay and charcoal fill of channel 6592 Grey silt clay, fill depression/channel 6594

6597

Intervallum

Fired clay deposit

Roman

6683

Internal

Charcoal deposit associated with 6682 (burned wood beam/plank)

Roman

6722

Internal

6790

Intervallum

Total:

Fired clay (centre of hearth) Clay deposit with pebbles, charcoal and slag within via sagularis make-up

Post-Med

Roman

179

107

Clay

Fired clay deposit

Char-coal

Internal

Coal/ Coke

6431

Total:

Concretion

Roman

Other magnetic material

Fired clay deposit

Slag microresidues

Internal

SSS

6418

SFB

Roman

HSS

Fired clay deposit

HSF

Internal

Other slag

6417

Clinker

Period

Vitrified lining

Context notes

Lining slag

Location

Bloom. slag

Context

SHC

Nidum. Excavations at the Roman Auxiliary Fort 2010 to 2012: Neath, Neath Port Talbot CBC

180 107

4 <1

9.4

141

3.2

Roman

1.2

Roman

Roman 9.8

131.4

40.6

131.4

258.6

<1 12

1.2

87.2

0.6

8 5.4

719

182

1228.2

Nidum. Excavations at the Roman Auxiliary Fort 2010 to 2012: Neath, Neath Port Talbot CBC

10.

Assessment of graphitic crucible sherds

By D D Dabal Summary Three sherds of graphitic crucible (total weight of 46g) were found in context 1002 during the excavations at Dwr-y-Felin School (site 677). The crucible fragments were found among materials ranging from the 16th century to modern times. The sherds carried a strong resemblance to the Post-medieval â&#x20AC;&#x2DC;dark cruciblesâ&#x20AC;&#x2122; made out of graphitic clay outcrops in Bavaria in Germany, Bohemia or Upper Austria. Method The sherds were inspected visually, using a low powered microscope but not subjected to any form of instrumental analysis. Therefore, identifications of the material and its interpretation is limited and must be regarded as provisional. Results The three sherds of graphitic crucible had a total weight of 46g. The crucible had regular concave inner shape and flat base (approx. 65mm diameter). Unfortunately, it was impossible to reconstruct the crucible height and shape of its rim as only one fragment of the base and two fragments of walls had been found. The walls were 7mm thick apart from where they joined the base to a maximum of 9mm thick, the base itself was also 9mm thick. The object was characterised by its dark grey colour and metallic appearance of the fabric. It had low density and showed small linear cracks with sporadic voids after fine organic temper. The inner surface of the pot showed no visible residues. It was very smooth and sparkly/metallic as a result of the large graphite flakes (grain size up to 1.5mm). The outer part of the vessel had rather coarse texture with black, very lightly burned/vitrified areas of the surface. On the base there was a partial stamp comprising of a part of the lower right cartouche corner with fragment of indeterminate rounded symbol or letter. Distribution of the residue The crucible fragments were recovered from the subsoil, context 1002. In general this context yielded small quantities (480g) of various metallurgical residues, such as smithing hearth cakes, lining, clinker, indeterminate slag, coal and coke (Young 2013c). None of these finds suggested that specific metallurgical activity was carried out on the site. The dating evidence found in the context, pottery and glass, ranges from the 16th century to modern times. The largest group of finds were of 17th, 18th, and 20th century origin. Interpretation Many industrial and laboratory processes require containers which withstand high temperatures. Graphitic crucibles are high quality objects well known for their unique properties. These vessels are infusible as well as unreactive (Newcomb 2009). In other words, they do not melt or react with substances contained in them, like metals or slags, and are more resistant to chemical attack of corrosive charge. Also, the graphite content enhances thermal

Nidum. Excavations at the Roman Auxiliary Fort 2010 to 2012: Neath, Neath Port Talbot CBC

conductivity of the pot saving time and fuel while fired in a furnace (Martinón-Torres and Rehren 2009). Moreover, graphite grains make the container resistant to a thermal shock, so it is very unlikely that it suddenly breaks and starts to leak. For the above reasons the graphitic crucibles were, and still are, highly valued and have been used since Medieval times up to the present day in many industries. In general, these type of ceramic vessel were broadly used for chemical and metallurgical operations. Alchemical laboratories, metallurgical and jewellers workshops, mints and glass workers were using them for mixing, melting, calcining, reducing, evaporating various substances under high temperatures (Martinón-Torres 2010). Numerous archaeological finds and historical documents confirm that, at least since Medieval times, crucibles made of ‘iron clay’, so named because of its metallic appearance, were produced in few places on the European continent where outcrops of the naturally rich in graphite inclusions clays could be found. The crucible sherds found during the excavation of Dwr-y-Felin School have a strong resemblance to ‘the dark crucibles’ from Bavaria in Germany and similar black wares made in neighbouring regions, such as Bohemia and Upper Austria, where deposits of graphitic clays exists (Martinón-Torres and Rehren 2004). The most popular producers of graphitic crucibles in Post-medieval times were potters from Obernzell in Upper Bavaria, Southern Germany (Martinón-Torres and Rehren 2009). Many of the vessels made there were exported across much of the world between the 16th and 19th century. These vessels are easy to identify as most of them show a complex stamp pressed in the base of the pot: ‘a symbol resembling a number 4, a cross and two initials inside an elongated cartouche’ (Martinón-Torres and Rehren 2004 21). Unfortunately, the stamp on the exterior of the crucible base shows only a square corner of a frame/cartouche and part of an indeterminate symbol or letter which could be a stylised anchor or letter ‘S', if this is indeed the original Bavarian product. If it is not, it could be a fragment of the letter 'O' or 'B'. The crucible fragments were among other metallurgical residues scattered across the playing fields of the modern school. Ordnance Survey mapping gives no evidence of any buildings existing on the site before the year 1919 and during Post-medieval times the whole area was farmland. Modern historical records however mention a school and an adjacent Mining and Technological Institute which opened its doors in 1931 (The Independent 2013). The Ordnance Survey map from 1935 depicts the buildings of Neath Grammar School and Cwrt Herbert playing fields as part of the area belonging to it. In 1949 the school became Neath Technical College. Therefore, it is most likely that the crucible was brought to the site sometime in 20th century. The crucible could have been part of the Mining and Technological Institute collection of laboratory equipment or, it was somehow related with copper, tin or steelworks industries which all were operating in Neath Port Talbot from the 17th century onwards.

100

Nidum. Excavations at the Roman Auxiliary Fort 2010 to 2012: Neath, Neath Port Talbot CBC

11.

The Roman brick, tile and daub (Table 11)

By Edith Evans Altogether, there was a total of 1774 fragments, weighing just over 64kg, over all three sites (677, 716, 740). Brick and tile fragments from each context were sorted by type and the numbers of fragments and total weight of each type were recorded. For most context/type groups, an electronic scale graduated in grams was used1 but for larger amounts, normally above 1 kilogram, the equipment used was a standard domestic kitchen scale marked in divisions of 0.025kg. Types were identified using the criteria laid down in Marvell and Owen-John 1997, 189. Surfaces were examined under a raking light for markings. Across all sites, the brick and tile was fragmented and had suffered heavily from erosion. In some instances, erosion was so heavy that pieces of brick or tile had been converted into pebbles, but even where this had not occurred surfaces were so soft that in some cases damage had resulted from the brushes used in the normal finds-washing regime. Given the poor state of preservation and the small size of the fragments, it was not always clear whether all fragments categorised as Roman tile actually were this – some could be burnt daub or other fired clay types, and some fragments (particularly roofing tile with a curved profile, and brick) could be post-medieval. The daub fragments were generally separated from the Roman brick and tile on the basis that they had either no inclusions of sand or other mineral filler in the fabric, or the amount was lower than is normal in brick and tile. However, in some cases it was not possible to come to a definite conclusion, and these fragments were left in with the undiagnostic brick and tile. Most of the material categorised as ‘daub’ could in fact be derived from any type of clay walling – mud brick or cob as well as actual daub. Only one piece showed any evidence for wattle impressions enabling it to be identified definitely as daub. This was a fragment (overall dimensions approach 0.07x0.06x0.05m, weight 0.176) from site 716 context 6000, which exhibited a marked asymmetrical ridge where it appears to have been squeezed between two wattles with a max width of 25mm at the base and height of 13mm.2 Fabric Given the poor state of preservation of the assemblage it was not considered that a detailed study of the fabric would be profitable, so examination was limited to inspection of hand specimens generally under no more than x3 magnification. A certain amount of variation was noted even within a single specimen, depending on the mix, with differential levels of firing making further variations in the colour between one specimen to another. There was no 1

This gave a variable, but fluctuating, margin of error of up to about 10g which was significant for very low weights, but this was judged to be of negligible importance in view of the fact that all the material thus affected was undiagnostic. 2 Fragments originally categorised as brick/tile in 6071, 6178, 6229, 6508, 6530, 6533, 6623, 6622, 6652, 6663, 6707 and 6773 are probably daub rather than tile. A cursory examination of the fragments categorised as burnt clay showed that most of these were probably daub as well, but in view of the fact that none of them will provide any useful information, it was not considered worth producing revised figures.

101

Nidum. Excavations at the Roman Auxiliary Fort 2010 to 2012: Neath, Neath Port Talbot CBC

evidence that would suggest that tiles had been sourced from anywhere outside the South Wales Upper Coal Measures, overlain by alluvium in the river valleys and elsewhere by glacial till. The feel in the hand was frequently silty, suggesting that it was alluvial clay that was principally being used, and quartz sands were a common component. The colour was mainly mid orange, but there were occasional streaks of cream-coloured clay that had not been properly mixed in. Inclusions were mainly fairly sparse orangey-brown platey fragments of what appeared to be mudstones, tiny black water-worn pebbles, probably coal, an occasional small lumps of coarse sandstone. There was also possibly also a little grog. Form A number of different types of Roman tile were present (Table 11). These were brick, used in the construction of walls and floors; tegulae and imbrices, designed for a roofing system that consisted of flat units laid edge-to-edge (tegulae), with the joints between them covered by halfround units (imbrices); and box tiles which were wall-cladding units designed to take hot gases from underfloor hypocausts up to vents in the eaves. Table 11. Table of Roman tile forms

Site no Brick

Tegula

Imbrex

Box

no. weight no.

Undiagnostic

no.

weight.

no.

weight

weight.

no.

677

11.418kg

2.910kg

0.282kg

1.600kg 503

17.696kg

716

3.862kg

3.704kg

0.240kg

0.332kg 770

11.653kg

740

0.484kg

0.264kg

1.004kg

Total

15.764kg

6.614kg

0.522kg

2.196kg 1287

weight.

30.353kg

Brick This is the only type where thickness alone has been taken as the diagnostic feature. Tile with a thickness of 40mm or more is taken to be brick, provided it is clearly not a tegula flange. Thinner objects have also been accepted as being brick where at least two sides of over 10mm in length are present. None of the fragments of brick presented any particular points of interest, most being extremely weathered. Thickness ranged from 38mm to 65mm. Two of them had partial signatures (see below). Tegula Flanges and/or cutaways were regarded as diagnostic features, provided that the length of the flange is at least 13mm. Thickness alone is not taken as a diagnostic criterion, as there are other types of tile of this thickness. The possibility that some of these might be half-box tiles (see Brodribb 1987 65-7) has been discounted as to date very few certain examples of this type have been noted in South Wales. The only complete dimensions for tegulae were thickness (in the

102

Nidum. Excavations at the Roman Auxiliary Fort 2010 to 2012: Neath, Neath Port Talbot CBC

21-27mm range) and flange height and width; the height was 25mm to 35mm, and the width generally 30mm to 35mm although one was 40mm. There was one cut-out (site 677), from the top of a tile, of at least 45mm, but it was so heavily eroded that it was not possible to say whether this was its full size. One unstratified tegula from site 677 had a nail hole. Imbrex Curvature is taken as a diagnostic feature, provided the thickness of the tile is neither larger than 30mm nor smaller than 10mm, the outer surface is smoothed and the inner surface sanded. Imbrex fragments from all sites were so small and weathered that no useful information was present. Box tile Two types of box tile have been recorded from South Wales. Box tile Type 1 was made by bending a slab of clay around a former; Box tile type 2 was made by assembling four slabs of clay within a mould, sometimes with the addition of strips of clay added to the corners for reinforcement (Evans and Stapenhorst 1984). The only fragments present from any of the sites came from box tiles of Type 2. The following features are taken as diagnostic of this type:  

sanded outer surface combined with manipulated inner surface; corners formed by the junction of two slabs of clay with either straight or bevelled edges, or with additional clay applied to the inside of the corner; thin strips of clay of rounded triangular section;  scored lattice,3 combing or some other form of keying on the outer surface;  cut-out vents. On the whole, the box tiles from all sites had suffered least from weathering because they had been harder fired than the other types of tile. Most of the fragments came from corners; where the slabs had come apart at this point, it could be seen that in most of the examples the edge of the slab had been bevelled to provide a better fit. Reinforcing strips were generally in place, in some examples pressed so firmly into the corner that it was difficult to see the join. None of these tiles preserved any of the original dimensions except for thickness, with a maximum of 20mm which was subject to the usual irregular ridges caused by manipulating the inner surface by hand. None of the fragments were marked with lattice on either face. One tile preserved evidence for a cut-out vent, apparently curvilinear in shape although so little was left that it is difficult to be certain. The edge of the cut-out was 48mm from the exterior of the corner which, allowing for the thickness of the adjoining slab and the reinforcing piece, is as close to the corner inside as could be contrived. Markings Two signatures were noted on bricks from site 677, neither of which survived in anything like a complete form. One (from 1089) had part of a signature which had been drawn using two

Note that a scored lattice on this type of tile cannot be considered an indication of an early date: it appears on the box tile assemblage from the late 3rd century building at Cold Knap, Barry (Evans, Dowdell and Thomas 1985). 3

103

Nidum. Excavations at the Roman Auxiliary Fort 2010 to 2012: Neath, Neath Port Talbot CBC

adjacent fingers, but with the lines joggled; this had not survived complete. The other consisted of two parallel lines (from 1061). A fragment of unidentifiable tile (from 6612) also preserved part of a signature, the apex of a â&#x20AC;&#x2DC;Vâ&#x20AC;&#x2122; with both arms cut short by the break. There was also a fragment of unidentifiable tile (from 1061) that preserved the edge of the footprint of a hobnailed sole; only a single row of hobnails was visible at the break. Two fragments from site 677 (part of a brick from 1075, and an undiagnostic fragment from 5001) had an indistinct closely set criss-cross impression. It was though that these might have been the impressions of coarse fabric or fine matting, but closer inspection by a textiles specialist failed to reveal any clear evidence of weave, and it seems more likely that they were caused by being laid an unstructured spread of hay or grass stems to dry.

12.

The stone objects (Figures 54-56)

by Jonathan Parkhouse The stone material from the most recent investigations at Dwr-y-Felin is similar to that recovered during earlier work (Parkhouse 1992). A number of items within the assemblages were difficult to identify with confidence, and in some cases it was difficult to be certain that they have been modified by human agency. Whetstones were particularly problematic; wear which is the result of deliberate use rather than some natural form of abrasion may be difficult to distinguish, and within the local gravels there will have been copious quantities of stone available for ad hoc use (eg Nos 6, 14), although in some cases (eg No 9) the identification is reasonably secure. In the case of some of the other items submitted for inspection, natural phenomena, such as the erosion of inclusions (eg Nos 8, 17) and the ripple effects along bedding planes may be confused with deliberate modification. Even where deliberate working seems probable, fragments may be so small as to make identification problematic, whilst some items may be waste from working (eg Nos 7, 20). Unsurprisingly stone from local sources in the Upper Coal Measures (Pennant) and Old Red Sandstone predominates. The only exceptions are the fragments of basalt lava quern (Nos 2 and 10), but given the frequency with which this material has been recorded on both military and civilian sites in South Wales its occurrence at Neath is unremarkable. The nearest sources of this material are within Cenozoic lava outcrops in the Eifel mountains and at Volvic in the Massif Central, both of which were worked during the Roman period, and whilst it is not possible to differentiate between lava from these two sources macroscopically, the Eifel is the most likely source for most, and probably all, of the material from Roman Britain. The poor preservation of the basalt lava from Dwr-y-Felin is typical of many sites, particularly those on gravel substrates, where slightly acidic ground water and the vesicular nature of the lava make it vulnerable to decay (Roe 2002). Site 677 1. 1000. Fragment of sandstone, the maximum dimension being 113.5mm (Figure 54). It appears to be a small part of a flat slab some 10mm thick, with a flange c28mm wide and 31mm high defining a curved edge. Although the edges of the piece are abraded, it appears to have

104

Nidum. Excavations at the Roman Auxiliary Fort 2010 to 2012: Neath, Neath Port Talbot CBC

been deliberately worked; if this is the case it may possibly be part of the base of a stone vessel. Weight 0.271kg. 2. 1048 (SF 009). Quernstone fragment. Fragment (approximately 25%) of a lower stone in basalt lava from the Eifel. Diameter originally approximately 44cm: spindle hole (apparently slightly off centre) 30.5mm diameter. Maximum thickness 71.6mm. The edge is very worn, but there are traces of the characteristic vertical striae along part of the outer edge. The grinding surface has typical radial grooving. There has been abrasion following breakage, as well as post-depositional chemical attrition of the surfaces; lava is prone to such weathering, especially on sites situated on glacio-fluvial gravels such as at Neath, as its vesicular nature enables slightly acidic groundwater to penetrate more readily. 3. 1264 (SF030). Quernstone fragment. Slightly less than 50% of upper stone in medium/coarse grained quartzitic sandstone; diameter originally around 38cm, with central hopper 112mm diameter defined by a shallow flange, generally c18mm wide. Maximum thickness 90 mm. Traces of opposing rynd chases 23mm long and 2 18mm deep (width unmeasurable due to fracture). The rynd was evidently replaced in a different position as there is an additional rynd chase across the top of the hopper and extending below the flange, 36 mm long x 22mm wide x 18mm deep. Large handle chase, wedge shaped, c 96mm long, 17 mm deep and varying in width from 51 mm near the outer edge to c68mm near the hopper flange. The handle was also replaced and repositioned, as there is evidence for a second handle, probably a simple straightsided slot c 65mm long and c20mm deep; the stone has fractured along this feature. Typical radial dressing, reasonably worn. 4. 2148. Long pebble, broken at one end, in fine-grained grey sandstone, maximum length 91mm. The cross-section varies from rounded sub-trapezoidal profile to oval. There are no flat surfaces, but some parts of the surface appear more polished than others and there are some faint shallow striations at right angles to the axis of the stone which may represent sharpening of fine metal or bone points. The stone would have been suitable for honing a long blade; the balance of probability is that this was used as a whetstone. Weight 0.212kg. 5. 1002. Two fragments â&#x20AC;&#x2DC;slate pencilâ&#x20AC;&#x2122;, 26.3 and 19.9 long; 3.9 mm diameter. Post-medieval. Weight 0.004kg. 6. 1046. Broken pebble in fine-grained sandstone, with sub-rectangular cross-section 33.6 x 23.7mm max. The narrower sides appear slightly more worn than the other surfaces, and the shape of the object would not be inconsistent with use as a Roman whetstone, but its identification as such is not certain. Weight 0.094kg. 7. 1061. Roughly rectangular sandstone fragment, 59 (max) x 68 x 25.6 mm, apparently broken from larger bar-shaped stone. The narrow faces appear to be slightly polished. Prominent Vprofile groove/cut mark along one face, 3.2 mm deep and c9mm wide; on the opposite side and parallel with the groove the profile is 'stepped' by some 4mm, possibly an attempt to cut or split the stone. The stone has clearly been worked, but its function is uncertain; possibly an offcut. Roman. Weight 0.224kg.

105

Nidum. Excavations at the Roman Auxiliary Fort 2010 to 2012: Neath, Neath Port Talbot CBC

8. 1075. Sandstone pebble, maximum diameter 77mm; one side convex, the opposite side concave; one edge fractured. No obvious signs of deliberate modification; the hollow on the concave side may be the site of a pebble or other inclusion, much eroded. Roman. Weight 0.098kg. 9. 1228. Two items. (a) Fragment of rectangular cross-sectioned fine-grained pinkish grey sandstone pebble, Maximum length 75mm (broken both ends); 54.6 x 41.4 mm cross-section. All four surfaces show signs of polish, suggesting use as a whetstone. In addition one narrow face has evidence of a roughly pecked band c14mm wide and c1.0mm deep. Roman. (b) fragment of gritstone c130 x 190mm and up to 42mm thick; no evidence for deliberate modification. Weight 1.595kg. 10. 1293. Small rounded nodule basalt lava. 49.3mm maximum length. Presumably derived from a quernstone; similar nodules were recovered from the earlier work at Dwr-y-Felin (Parkhouse 1992, 285). Roman. Weight 0.068kg. 11. 1506. Elongated pebble of medium-grained grey sandstone, maximum dimensions 128 x 28 x 53mm. Generally worn, but no distinctive evidence of use as a whetstone. Roman Weight 0.420kg. 12. 1086. Split pebble of fine-grained sandstone; no diagnostic evidence of modification or use. Weight 0.336kg. 13. 1111. Elongated fine-grained grey sandstone pebble, broken one end, oval cross-section with rounded corners, maximum dimensions 129 x 26 x 43mm. No differential wear patterns, or any other evidence for use as a whetstone. 14. 1117. Elongated sandstone pebble with two approximately parallel sides, maximum dimensions 136.5 x 43.7 x 23.1m. Whilst of a suitable size and profile for use as a whetstone, there are no obvious signs of such use. Weight 0.246kg. 15. 1143. Rounded sandstone pebble, maximum dimensions 79 x 48 x 42mm; no evidence of differential wear or other sign of use as a whetstone. Weight 0.234kg. 16. 1337. SF 051. Irregularly-shaped piece medium-grained grey sandstone with two flat edges at right angles, probably a broken fragment of a larger block. One of the flat faces has six variously distinct traces of grooves, some straight and some slightly curved, c 2.5mm wide and up to .0.8mm deep. There are a further nine similar but wider grooves on one of the other surfaces which is otherwise smooth and almost flat. It is difficult to be certain whether or not this is an area of deliberate carving or random post-depositional abrasion such as plough-strike. A second fragment of an irregular but generally flat slab of similar stone, 50mm thick, has traces of similar grooves. 17. Unstratified. Fragment of what is apparently an elongated pebble, maximum 107mm long (broken) x 77mm wide, with â&#x20AC;&#x2DC;hollowâ&#x20AC;&#x2122; on inside. No definite evidence for use/modification; the hollow may be the site of an eroded pebble/inclusion, Weight 0.268kg. 18. 1022. [no (SF no.] Approximately 20% of upper quernstone in medium grained sandstone. Original diameter in region of 38cm; maximum thickness 91.3mm. There is a hopper with a 106

Nidum. Excavations at the Roman Auxiliary Fort 2010 to 2012: Neath, Neath Port Talbot CBC

diameter originally in the region of 11cm; this was encircled by a roughly dressed flange c21mm wide and 5mm high. There is a possible vestige of a rynd chase on the inside of the hopper. Radial grooves on the worn grinding surface. Site 716 19. 6445. Near-spherical sandstone pebble. No evidence for modification or deliberate use. Pebbles of similar size and shape have been retained by excavators on other Roman sites, including Biglis and the previous excavations at Dwr-y-Felin, in the belief that they may have been slingstones, but there is no convincing evidence to prove such an attribution in most cases, especially in the case of single finds. See Parkhouse 1988, 61 and fig 25 (Biglis); Parkhouse1992, 285 (Dwr-y-Felin). Weight 0.034kg. 20. 6286. Elongated bar-shaped fragment fine-grained grey sandstone, 100mm long, broken either end, with rectangular cross-section, max 36.7 x 25.8mm. One edge has vestiges of an apparent longitudinal ‘rib’ approximately 9mm wide; the surface either side of the rib appears to be worn and the rib is incomplete, as if deliberately removed, leaving only a shallow vestige. Possibly an off-cut from a piece of architectural stone. Modern. Weight 0.190kg. 21. 6067. Small roughly rectangular fragment sandstone, 26.5 x 30.3 x 11.4mm. One ‘face’ has an apparent narrow shallow groove near three of its edges (the fourth edge is fractured), but it is difficult to be certain whether this is a deliberate incision. The opposite face is flat, but not obviously worn. Weight 0.016kg. 22. 6178. Several abraded fragments of stone or possibly cement. Weight 0.138 kg.

13.

Animal bone (Tables 12-14)

By Hannah Bowden Introduction Archaeological investigations at the Roman fort of Nidum, Neath (GGAT sites 677, 716 and 740) produced 397 animal bone fragments hand collected from 91 Roman to Modern contexts, 17 (0.071 kg) fragments from 5 contexts of mixed and unknown date and 18 (0.006 kg) fragments from unstratified deposits. The contexts comprise of a range of charcoal-rich deposits, depositional spreads, metalled surfaces, ditch, drain, gully, latrine, oven, pit and road fills and stone deposits. The following report comprises an assessment of the faunal assemblage and its potential for future analysis. Method The faunal remains were identified to species or species group with reference to known specimens and publications (Cohen and Serjeantson, 1996; Hillson, 1992; Schmid, 1972). The distinction between sheep and goat and the identification of bird remains to species level was not made for the assessment phase. When species could not be identified, bone was categorized to species size, consisting of large mammal (domestic cattle or horse), medium mammal (domestic and wild ruminants and non-ruminates) and small mammal (domestic and wild

107

Nidum. Excavations at the Roman Auxiliary Fort 2010 to 2012: Neath, Neath Port Talbot CBC

carnivores, insectivores, lagomorphs and rodents). Ribs and vertebrae have been classed to species size only, unless they are associated with an articulate skeleton. Bone fragments were counted using the NISP methodology (Lyman, 2008). The preservation of the bone remains was classed on a scale of 1 to 5: Scale 1 2 3 4 5

Preservation of bone Only enamel has survived. The surface of the bone is extremely pitted, cracked and rough, with the tendency to splinter even further. The various layers of bone are notable. The surface of the bone is slightly pitted, with faint cracks. The surface of the bone is fairly intact, with a chalky texture or the bone has been burnt. The surface of the bone is in very good condition, with a smooth surface.

The degree of fragmentation of the bone remains was classed on a scale of 1 to 4: Scale 1 2 3 4

Fragmentation of bone Complete element surviving. Near complete element but slightly fragmented. Fragmented but with some identifiable features on the bone. Fragments are either medium/small, small or unidentifiable.

Results From the assessment of the faunal assemblage, 5.32 per cent of the remains were identified to species level, 20.60 per cent to large, medium and small mammal size and 74.07 per cent were indeterminable (Table 12).

Species Cow

NISP 12

% 2.78

Sheep/goat

2.08

Horse

0.23

Avian

0.23

5.32

Large mammal

17.36

Medium mammal

2.78

Small mammal

0.46

Indeterminable

320

74.07

409

94.68

431

100.00

Total

Table 12. Animal bone NISP and % values of Species recovered (Sites 677, 716, 740).

108

Nidum. Excavations at the Roman Auxiliary Fort 2010 to 2012: Neath, Neath Port Talbot CBC

Species Representation. Site 740. One sheep/goat metacarpal was recovered from Post-medieval/Modern floor deposit 022, which was associated with rubble deposit 020 which may have been the remains of a possible building. The metacarpal had fine cut marks along the shaft (diaphysis) of the bone and was fairly well preserved at scale 3. Sites 677 and 716. The majority of the faunal assemblage recovered from Dwr-y-Felin (Sites 677 and 716) is comprised of unidentified species at 94.68 per cent (409 NISP). This is largely comprised of indeterminable bone fragments at 74.07 per cent (320 NISP) followed by large mammal sized bone remains at 17.36 per cent (75 NISP). The rest of the unidentified species consists of medium and small mammal remains. The quantity of the identifiable faunal remains in comparison is relatively small at 5.32 per cent (21 NISP) and consists of cattle, sheep/goat, horse and bird (Table 13). Large mammal The large mammal bones were mostly recovered from Roman deposits (56 NISP) which ranged from channel, gully and pit fills to charcoal and oven deposits. These consisted of 47 indeterminable bone fragments from contexts 1111, 1521, 1731, 1800, 1861, 6479, 6545 and 6593, six limb bone fragments from 1131 and 1710 and three rib fragments from 1800 and 5007. Large mammal bone was also recovered from Post-medieval ditch fills, with two indeterminable fragments from contexts 1088 and 6240, three metapodial fragments from 1089 and one distal humerus fragment from 1084. Seven indeterminable fragments were recovered from Modern deposits 1000 and 1081 and undated context 1077 contained three limb bone and three indeterminable fragments. Medium and small mammal size bone Out of the 12 medium mammal bone fragments that were recovered, seven fragments were from Roman contexts. These consisted of indeterminable fragments from oven deposit 1675, pit fill 1822 and structure deposit 6286. Two ankle bones were recovered from oven deposit 1919, a limb bone diaphysis fragment from road deposit 6117 and a rib fragment from occupation/destruction deposit 1113. The remaining five medium mammal remains were limb bone fragments from Post-medieval ditch fill 1088. Two small mammal limb bone diaphysis fragments were recovered from Post-medieval charcoal rich deposit 6440. Cattle A mineralized cow metapodial which was fragmented in two was recovered from Roman ditch fill 1465 and the remaining fragments consisted of teeth, six from Post-medieval deposits (1082, 1084 and 1089), three from Modern deposit 1002 and one from undated context 6189. 109

Nidum. Excavations at the Roman Auxiliary Fort 2010 to 2012: Neath, Neath Port Talbot CBC

Identified Species Location Ditch, trench 1 (Site 677) East Strip (Site 716)

Extramural (Site 677 & 716)

Internal (Site 716) Intervallum (Site 677 & 716)

Rampart (Site 677 & 716)

South Strip (Site 716) Trench 5 extraction pit (Site 677) U/S (Site 677 & 716) 022 (Site 740)

Unidentified Species

Period

NISP

Cow

Sheep/goat

Horse

Bird

Large mammal

Medium mammal

Small mammal

Indeterminable

Roman

0.46

Unknown

0.23

Modern

0.69

Post-medieval

8.33

Post Roman

2.55

Roman

19.44

Mixed

1.16

Unknown

2.55

150

34.72

Roman

12.73

Modern

5.32

Post-medieval

0.46

Roman

121

164

37.96

189

43.75

Modern

0.93

Roman

0.93

1.85

Post Roman

0.46

Roman

1.39

N/A

4.17

Postmedieval/Modern

0.23

432

100

Total

Table 13. Animal bone NISP and % values of the faunal remains from Dwr-y-Felin (Sites 677, 716 and 740) by Location and Period

110

Nidum. Excavations at the Roman Auxiliary Fort 2010 to 2012: Neath, Neath Port Talbot CBC

Sheep/goat One sheep/goat molar was retrieved from Roman depositional spread 1006 and the rest of the sheep/goat remains consisted of a limb bone diaphysis fragment from Post-medieval context 1084, four tooth fragments from Modern deposits (1001 and 1029) and two fragments of a distal humerus from undated context 6182. Horse The horse 1st phalange was recovered from undated context 6183 but contained black spotting on the bone similar to the bone remains that were recovered from Post-medieval ditch fills 1084, 1088 and 1089. Bird There was only one bird bone that was recovered from the site which was a Bantam-size bird femur from Roman context 1006 located within the intervallum area. Preservation. The preservation of the faunal assemblage was generally poor with the remains highly fragmented and ranging from scale 2-4, with 81 per cent of the assemblage at scale 4 which consisted of burnt bone. 69 per cent of the burnt bone was recovered from Roman deposits. The colour of the burnt bone was mainly white with some fragments containing patches of blue/grey colouration and a few of the indeterminable limb bone fragments from Roman pit fill 6037 were burnt black on the interior surface and white and blue/grey on the exterior. Six indeterminable limb bone fragments and one indeterminable bone fragment which were recovered from 6037, all contained a cracked surface texture which is a typical modification of the surface of the bone which has been subjected to burning at high temperatures (Nicholson, 1993; 417). The colour and texture of burnt bone can indicate the approximate temperature of the fire it was burnt in. Bone is reputed to turn black when burnt at temperatures above 300C and white when burnt at temperatures ranging from approximately 700C to 1000C (Lyman, 1994, Nicholson, 1993; 414 and Shipman et al, 1984; 312-313). An assessment of the colour of burnt bone can therefore provisionally indicate the category of burning activity. Temperatures of camp-fires can range from approximately 400°C to 700°C in which the colour of burnt bone can consist of red-brown, dark brown and blacks to blue-grey, light grey and white, with the structure of bone displaying evidence of cracking (Lyman, 1994; 386). The majority of the bone recovered from Dwr-y-Felin could therefore possibly indicate that they were the remains of domestic waste which had been placed in a fire or hearth in preparation for disposal then deposited along with the rest of the refuse on site. The unburnt bone which made up 19 per cent of the faunal assemblage (83 NISP) was mainly comprised of durable bone elements such as teeth, mineralised and heat affected bone fragments. The exceptions were the indeterminable fragments from Roman road deposit 6294 and Post-medieval contexts 1084 and 1088, the medium limb bone fragment also from 1088 and the large mammal distal humerus and sheep/goat limb bone fragment from 1084. The teeth

111

Nidum. Excavations at the Roman Auxiliary Fort 2010 to 2012: Neath, Neath Port Talbot CBC

consisted of a sheep/goat molar from Roman depositional spread 1006, cow molars from Postmedieval deposits 1082, 1089, 1084 and modern deposit 1002 and a sheep molar from Modern context 1001. The molars from context 1082, 1002 and one from 1084 were all fragmented into three pieces which refitted. The fact that the teeth recovered were highly fragmented, even from the modern deposits, demonstrates the harsh burial conditions of the site. The unburnt bone recovered from the Post-medieval ditch fills (1088, 1089 and 1084), contained black and brown spotting on the bone surface which may possibly be staining caused by the presence of manganese within the soil. The remains consisted of a sheep/goat limb bone, a cow molar and third molar from 1084 and a cow upper molar and indeterminable limb bone fragments from 1089, a large mammal sized and an unidentified indeterminable bone fragment and a medium mammal limb bone diaphysis fragment from 1088. There was evidence of orange and brown iron concretions on burnt and unburnt faunal remains recovered from the extramural, intervallum and internal areas of the site. The deposits that the bone was recovered from consisted of channel (6593), drain (1465), gulley (6545) and pit fills (1800, 6023 and 6479). The concretions were also present on the faunal remains from road deposit (6178), oven deposits (1201, 1710), depositional layers (6324, 6286, 6590, 6023, 6668, 6672, 6678 and 6588) and metalled surfaces (6707 and 6709). The majority of the deposits consisted of grey clays which indicate a waterlogged environment and the presence of ironstones were noted. The staining of the bone is probably a result of the leaching of the iron oxides from the soil and ironstones and the absorption of the minerals into the faunal remains. The mineralisation of the bone fragments is only evident on the bone recovered from the Roman deposits. No. of measurable bones (NISP)* Cow Sheep/goat Horse

Teeth (NISP)** Cow Sheep/goat 1 2

Context Bird 1006 1084 1465 1 6183 (if context can be dated) 1 *Only complete bone elements which include both fused proximal and distal epiphysis will be measured, or the latest fusing epiphysis of a bone element. ** The tooth wear stage is recorded for individual cheek teeth and mandible tooth rows. Table 14. No. of Measureable bones and teeth (NISP) from Dwr-y-Felin (Sites 677 and 716)

Ageing and measurable bone. The cattle metapodial from Roman context 1465 was the only complete and phased bone recovered which can be measured but is heavily mineralised. Due to the poor preservation of the faunal assemblage, metrical analysis cannot be carried out for this site but measurements of the metapodial and the tooth wear stage of the individual teeth recovered from the site can be recorded for future reference (Table 14).

112

Nidum. Excavations at the Roman Auxiliary Fort 2010 to 2012: Neath, Neath Port Talbot CBC

Butchery. There was evidence of butchery comprising of three cut marks on a medium mammal rib fragment from Roman context 1113 located within the extramural area of the site and a cut mark on a large mammal limb bone or rib fragment from Roman deposit 6479. There was also a possible saw mark on a cattle 3rd molar from Post-medieval context 1084.

14.

The charred plant remains (Table 15)

Wendy J. Carruthers Introduction During excavations by Glamorgan and Gwent Archaeological Trust (GGAT) in 2010-2012 on the site of the Roman Fort in Neath soil samples were taken from a range of features and deposits for the recovery of environmental information. The samples were processed by GGAT staff using standard methods of floatation. A 300 micron mesh was used to catch the flots and a 300 micron mesh was used to retain the residues. Because of the clayey nature of the soil double floatation was implemented, i.e. the dried residues were re-floated in order to recover charred material that was reluctant to float during the first floatation. Most of the flots and residues were sorted by GGAT Environmental Archaeologist Hannah Bowden, although some of the flots were sorted by the author. Sorted items and flots were sent to the author for analysis. Method Out of a total of 80 samples processed and assessed by Hannah Bowden (GGAT) 58 samples (from 38 contexts) and one hand-picked item were sent to the author for analysis. The 49 productive samples described in this report came from an evaluation trench (Site 674) in the internal part of the later stone fort built around AD 117-120/125 (samples from 7 contexts), and from excavations of the earlier Flavian timber fort built and initially occupied around AD 80 – 90 (Site 716). The samples from Site 716 came from features and layers in the internal (samples from 16 contexts), intervallum (samples from 13 contexts) and extramural areas (samples from 2 contexts). Whilst the material from site 674 was not obtained as part of the excavations described in this report it is included here as previously there has been little, if any analysis, of plant remains from Roman Neath. An Olympus SZX7 stereo microscope was used to sort flots and identify charred plant remains. The author’s own reference collection alongside a number of identification guides (primarily Cappers et al 2006) were used to assist in the identification. Together the evidence provides important insights into the local and other plant resources that were being brought into the forts at different periods. Results The results of the analysis are presented in Table 13. Nomenclature follows the traditional classification system presented in Zohary et al (2013), and for the remaining plants Stace (2010) has been used. For the sake of simplicity the term ‘seed’ will be used in the discussion where in some cases, botanically speaking, the items are fruits, for example the propagules of sedge

113

Nidum. Excavations at the Roman Auxiliary Fort 2010 to 2012: Neath, Neath Port Talbot CBC

(Carex sp.) are technically fruits or nutlets but in laymanâ&#x20AC;&#x2122;s terms may be better understood when described as seeds. Some notes on the identification of crop plants A very narrow range of crops with confirmed identifications (based on the presence of specific characteristics) was recovered from the samples. The identifications below draw on information from Jacomet (2006) and Zohary et al (2013); Hulled six-row barley (Hordeum vulgare), confirmed due to the presence of twisted lateral grains (only found in six-row barley) as well as traces of palea and lemma adhering to the slightly angular grains. The ratio of twisted to straight grains was roughly 1 to 4 instead of 2 to 1, as occurs in the ear. However, this is only to be expected in what is likely to have been fully processed grain being transported to the fort to supply the army. Not only are twisted grains sometimes hard to differentiate from straight grains, but more significantly they tend to be smaller than the straight, central grains so can be reduced in number by being sieved out of the crop during processing. Spelt wheat (Triticum spelta), confirmed due to the presence of very low numbers of glume bases (only two identifiable glume bases out of total of three glume bases from 49 samples). These identifications do not rule out the possibility that other cereals might have been present, for example emmer wheat, two-row barley or naked barley, though it is very unlikely that these other cereals were deliberately being brought onto the site. Emmer had largely been replaced by spelt by the Roman period in southern Britain though it continued to be grown in some areas of northern England and Scotland and may have persisted as a weed in many other areas. Unfortunately emmer and spelt grains cannot be reliably differentiated (Jacomet 2006) so the scarcity of chaff fragments at Dwr-y-Felin makes it difficult to know whether emmer was present. Two-row barley and naked barley have not been confirmed as being cultivated at this time. However, sporadic mutations can occur producing, for example, occasional naked barley grains. No evidence of this was observed in the Dwr-y-felin samples. It appears that some free-threshing wheat, most likely bread wheat (Triticum aestivum) was being brought to the site. However, this can only be confirmed by identifying well-preserved rachis fragments (Jacomet 2006) and none were recovered. Wheat grain morphology is very variable so the few shorter, more rounded grains that had the characteristics of free-threshing wheat were recorded only as possible bread wheat (Triticum cf. aestivum). The occurrence of free-threshing wheat in the Roman period is discussed further below. Oat grains (Avena sp.) were relatively frequent, particularly in samples from Site 674. As these were very much more productive than the small samples from Site 716 it is difficult to compare the two sets of samples. Since only grains of oat were recovered, and the preservation of whole florets (or floret bases) is needed to differentiate between cultivated and weedy oats, it is impossible to say whether oats were deliberately being brought to the fort or whether they were a more serious crop weed at the time of the stone fort. Although it was not possible to identify the oats beyond Avena sp. they had some characteristics of bristle oat (Avena strigosa) or wild oat (A. fatua) in being slender with a narrow apex (rather than plump and blunt-ended) and less

114

Nidum. Excavations at the Roman Auxiliary Fort 2010 to 2012: Neath, Neath Port Talbot CBC

than 5mm in length. Common oat grains (A. sativa) can reach 6 or 7mm in length. Oats are valued as being a high-energy fodder for horses so their presence at a fort is quite likely. Description of the samples by site, area and feature Site 716 Internal features/layers Hand-picked charred item from context 6684 – Just over half a charred date fruit (Phoenix dactylifera; flesh and stone) was recovered from a deposit located in the south-east corner of the trench. The date flesh and stone were in a remarkably good state of preservation, suggesting that charring had taken place fairly slowly under reducing conditions. This imported luxury food is discussed further below. Post-hole 6454, context 6472 (sample 19) – This post-hole was located in the south-east corner of the trench, close to where the charred date was found. The sample (taken from the base of the post-hole) contained no cereal remains or weed seeds, only three charred fragments of sloe stone (Prunus spinosa) and one of hazelnut shell (Corylus avellana). Whether the fragments slipped into the post-hole during its period of use, after it fell into disuse or were deliberately placed in the base of the hole as some sort of votive offering to bring success to activities taking place in the area cannot be determined. Scoop 6587, context 6544 (sample 23) – This small sample from a scoop close to the location of the date produced only a single barley grain, an indeterminate cereal grain and a possible Ericaceous leaf. Although the identification is tentative due to poor preservation evidence of heathland in the vicinity of the fort was recovered from pollen samples assessed by Sarah Jones (University of Wales, Lampeter; see below). Charcoal-rich and fired clay deposits 6431 (samples 26, 31, 36, 37), 6449 (samples 27, 30), 6433 (samples 29, 32, 46), 6683 (samples 40 to 44) – Fourteen samples from these four deposits of fired clay and charcoal produced some of the highest concentrations of charred plant remains from the site. The area of burning was located close to the southeast corner of the site, the most productive area of the early fort. Although a moderate number of cereal grains was recovered (12 emmer/spelt, 2 cf. free-threshing wheat and 36 barley grains plus one emmer/spelt glume base) and frequent charcoal fragments (samples 31 = 75ml, 29 = 35ml and 30 = 15ml), the principal components of the assemblages (71% of the total) were plants of grasslands and possibly marsh. The diverse range of grassland species included buttercup (Ranunculus acris/bulbosus/repens), clover-type (Trifolium/Medicago/Lotus sp.), ribwort plantain (Plantago lanceolata), sheep’s sorrel (Rumex acetosella), selfheal (Prunella vulgaris), yellow rattle (Rhinanthus sp.), frequent small-seeded grasses (Poaceae) and autumn hawkbit (Leontodon autumnalis). Grass-sized stem fragments were present in two of the flots. Indicators of wetter, marshy grassland include spike-rush (Eleocharis subg. Palustres), sedges (Carex spp.), cf. club-rush (cf. Schoenoplectus sp.), water-pepper (Persicaria hydropiper/mitis) and cf. cotton-grass (cf. Eriophorum sp.). Together with a fragment of bracken pinnule (Pteridium aquilinum) and some possible ericaceous leaves, both of which would have grown on poor, acidic rough grassland or heathland, the charred remains indicate that hay, marsh hay and

115

Nidum. Excavations at the Roman Auxiliary Fort 2010 to 2012: Neath, Neath Port Talbot CBC

perhaps heathland vegetation was being brought in from the surrounding area for use either as bedding, fodder, flooring or building materials, or fuel for ovens and hearths. In the context of an area of fired clay deposits some sort of industrial activity involving the burning of hay/marsh hay appears to have been taking place. However, it is possible that the burning represents a conflagration in the fort (Martin Tuck, pers. comm.) so the interpretation is uncertain. Similar (but less concentrated) assemblages were recovered from samples 47 and 6 in the Intervallum area (as well as those described below). Several fragments of hazelnut shell were also recovered from this group of samples (12 fragments) perhaps deriving from foods being consumed as snacks. Fired clay deposits 6418 (sample 28) and 6417 (sample 35) – These two adjacent deposits located further to the north than the above large group of samples produced similar small assemblages consisting of occasional cereal grains (3 barley and 1 emmer/spelt grain in total) and a few grassland taxa (buttercup, clover-type, sheep’s sorrel, grasses). A single charred bramble seed was recovered from sample 28. The samples appear to be very similar in character to the larger group of fired clay deposits described above, though less productive. Presumably similar activities were taking place in these areas involving fires fuelled with hay and marsh hay, either accidental or deliberate. Charcoal-rich deposits 6564 (sample 47) and 6561 (sample 48) – These layers were located a little further south than samples 28 and 35 though probably derived from similar activities to the other layers described above involving the burning of hay and marsh hay. A possible freethreshing wheat grain, two barley grains, four poorly preserved indeterminate grains, three fragments of hazelnut shell and frequent charcoal fragments were recovered from sample 47. Grassland, marsh and disturbed ground weed seeds were present in low numbers in both samples including clover-type, knotgrass (Polygonum aviculare), dock (Rumex sp.), sheep’s sorrel, chickweed (Stellaria media), sedges (Carex spp.) and grasses. The presence of chickweed and knotgrass indicate that the vegetation being burnt may have come from a more disturbed, nutrient-enriched area. However, since cereal grains were also recovered from most of the samples a fairly mixed type of waste was probably been burnt as fuel/kindling in all of these fired clay areas. Hearth 6923, context 6722 (sample 54) – This sample came from the centre of a fired clay hearth located close to the deposit producing the charred date in the south-east corner of the trench. The only item recovered from the sample was a small fragment of hazelnut shell found in the residue. It probably represents snacking waste thrown onto the fire. Pit 6860, context 6790 (sample 55) – This sample was taken from the dark grey silty clay primary fill of pit 6860. It contained a sparse assemblage of an emmer/spelt grain, a fragment of hazelnut shell, and a few seeds from grassland or disturbed ground plants (clover-type, dock and sedge with traces of grass-sized stems). Low numbers of remains of this type often represent background charred waste that was being blown and trampled around the site. However, the presence of frequent charcoal fragments suggested that ash from a hearth or oven may have been deposited in the pit, containing a little domestic waste. The pit was located close to the fired clay deposits in the south-east corner of the trench so the burnt waste may have been cleared from that nearby activity area. 116

Nidum. Excavations at the Roman Auxiliary Fort 2010 to 2012: Neath, Neath Port Talbot CBC

Gully 6771, context 6772 (sample 59) – A gully containing a light greyish brown silty clay fill produced a single indeterminate cereal grain and a fragment of hazelnut shell. As with the previous sample, these remains probably derive from general background waste. The gully was located close to the south-east corner of the site. Layer 6678 (sample 60) – The only charred items were two fragments of hazelnut shell. From an early deposit sealed by the internal road. Intervallum features/layers Pit 6025, context 6030 (sample 2), context 6026 (sample 6), context 6104 (sample 7), context 6124 (sample 11) – This large pit with clayey fills and charcoal-rich deposits possibly representing clay linings (contexts 6104 and 6124) produced only traces of cereal remains (two barley grains). However, frequent charred seeds from grassy and marshy vegetation were recovered from an upper charcoal-rich layer (context 6026) defining one edge of the pit. Unusually, the dominant taxon was water-pepper (Persicaria hydropiper/mitis), a plant of damp meadows and shallow water on acidic soils. Forty-nine seeds from this plant were present in the sample, which on their own could suggest they may have been collected for use as a flavouring since the ground seeds can be used in a similar way to pepper (Mears and Hillman 2007, 258). However, other rough, damp grassland plants were also well-represented, including buttercup, vetch/tare (Vicia/Lathyrus sp.), clover-type, knotgrass, dock, sheep’s sorrel, sedges and grasses, suggesting that the deposit contained hay cut from wet meadows that had been burnt as fuel or kindling. Since charcoal fragments were also frequent (125ml) the latter explanation; use of hay for kindling, might be the most likely. In addition, a single fragment of hazelnut shell was present in context 6026. The only other sample to contain a few charred remains was context 6104 which also included water-pepper and a grass seed. The pit was located along the north-eastern border of the trench, close to the rampart of the 1st century fort. Scoop 6093, context 6079 (samples 3a and 3b) – Two samples from this scoop, located close to the via sagularis, produced a single fragment of hazelnut shell only, extracted from the residue. Pit 6035, context 6037 (sample 8) - This sample came from a feature located along the northeast margin of the trench, fairly close to the via sagularis. It contained a fragment of possible free-threshing wheat, two barley grains, a sedge nutlet and a small grass seed. As suggested for many of the other small samples, the remains probably represent low-level background burnt waste that was likely to have been distributed around the fort. Pit 6121, context 6115 (sample 9) – South of pit 6035, pit 6121 produced only a single sheep’s sorrel achene. Sheep’s sorrel is a plant of acidic grasslands. Layer 6177, (sample 12) – Sample 12 came from a layer located underlying the via sagularis towards the southern end of the trench. Context 6177 consisted of a mid-grey clay deposit similar to adjacent clay 6176. An emmer/spelt grain and three barley grains were the only items recovered from the small sample.

117

Nidum. Excavations at the Roman Auxiliary Fort 2010 to 2012: Neath, Neath Port Talbot CBC

Pit 6332, context 6336 (sample 17) – The sample was taken from the basal fill of a c. 1.1m pit located next to the north-east margin of the site. The only charred remain from the very small sample was an indeterminate cereal fragment. Fill of cut 6483, context 6336 (sample 18) – This deposit consisted of charcoal-rich clay. It was located at the south-western end of the trench, not far from the via sagularis. A large quantity of charcoal was present in the flot (75ml; mixed species) as well as 15 barley grains, five indeterminate cereals, four fragments of hazelnut shell, a clover-type seed and four small grass seeds. The deposit is likely to represent ash cleared out of a hearth or oven – perhaps one of the large ovens located between the ramparts and via sagularis to the north-east of the feature. Layer 6038 (sample 21) – This charcoal-rich clay deposit located on the north edge of the via sagularis produced no cereal remains, but the seeds from a few grassland plants (buttercup, sheep’s sorrel, grass) and frequent charcoal fragments (100ml) were recovered. In addition a couple of weeds of cultivated and waste ground were present (redshank/pale persicaria (Persicaria maculosa/lapathifolia); many-seeded goosefoot (Chenopodium polyspermum)). This material may represent ash cleaned out of a hearth or oven, such as the large ovens located between the rampart and via sagularis. Channel 6592, context 6593 (sample 24) – The grey silty clay fill of channel 6592 contained a small amount of charcoal and a few remains from plants of grassy and disturbed ground. These included clover-type, dock, grass, black nightshade (Solanum nigrum) and ribwort plantain. Black nightshade often grows in soils that have high nitrogen levels, such as occur in farmyards. However, charred plant remains are not necessarily found close to the places that they have grown, particularly in a context shows that it could represent vegetation gathered for use as kindling or fuel. Fired clay deposit 6597 (sample 45) – This small sample from the fired clay area produced one barley grain and three indeterminate cereals, as well as a fragment of hazelnut shell, a small grass seed and several fragments of grass-sized stem. It is similar to the other small assemblages of mixed food and kindling debris that probably represent background waste. Extramural layers Pit [6321], context 6258 (sample 14) and context 6287 (sample 15) – The two fills of this shallow pit (or scoop) both consisted of charcoal-rich silty clay soil. The primary fill, context 6258, produced thirteen hulled barley grains, five of which were twisted lateral grains demonstrating that six-row hulled barley was represented (Hordeum vulgare). Three indeterminate grains, a spelt glume base (Triticum spelta) and two clover-type seeds were also recovered. The deposit above, context 6287, contained three hulled barley grains and a brome grass fragment (Bromus sect. Bromus). The assemblages represent a small amount of grain with traces of chaff and seeds from common arable/disturbed ground weeds. With such a small amount of data from small soil samples it is difficult to interpret the assemblages with any degree of certainty in terms of whether cleaned grain and crop processing waste was present or simply mixed domestic waste. The concentration of barley grains was relatively high in this feature, compared to other samples from within the Flavian fort, but as this was the only 118

Nidum. Excavations at the Roman Auxiliary Fort 2010 to 2012: Neath, Neath Port Talbot CBC

extramural feature examined this could be due to the chance factors involved in preservation by charring. Site 674 Ditch 140, context 141 (sample 31), context 160 (sample 32) –Charcoal fragments were frequent, particularly in context 160. The ditch was the most archaeobotanically productive feature analysed for this report, although the large sample size for sample 32 meant that the two samples did not produce the highest concentrations of charred plant remains (13.2 fragments per litre and 6.5 fpl respectively). The primary ditch deposit, context 141 (sample 31) produced the highest concentration of cereals though as the sample size was small the quantity was not high; 37 barley grains, plus three emmer/spelt grains. Context 160 (sample 32) produced 203 barley grains and an emmer/spelt grain from 40 litres of soil. Apart from a number of indeterminate cereal grains, the only other items present were two fragments of hazelnut shell and four seeds from plants of grassy and marshy habitats; two spike-rush nutlets, one sedge nutlet and one small grass seed. All of these items were present in context 160. These two assemblages differ slightly from the five samples taken from the upper layers in this trench. Firstly, no oat grains were present, whilst all five samples in the later deposits produced oat grains. Secondly, no free-threshing wheat grains were recorded, while three of the five later samples contained free-threshing wheat. This is discussed further below. The dominance of barley in these productive samples and possible trace of hay from damp meadows could be said to indicate the presence of fodder in the ditch. However, the general dominance of barley over wheat in the Dwr-y-Felin samples as a whole deserves closer examination, as discussed below. Slot 109, context 108 (sample 8) – A possible beamslot that cut a road surface was examined to investigate late activity on the site. The dark brown sandy clay loam fill (context 108) produced a small assemblage of five emmer/spelt grains, an indeterminate wheat grain (possibly free-threshing wheat), three poorly preserved barley grains, six indeterminate oat grains and a few poorly preserved cereal fragments. A fragment of sloe stone and two fragments of hazelnut shell represent burnt waste from other foods being consumed. A pale persicaria achene (Persicaria lapathifolia) and a small grass seed were the only other items present. The assemblage appears to represent a small amount of burnt food waste. The presence of moderate amounts of charcoal, slag, fired earth and bone indicate that mixed waste had been used to backfill the feature. Slot 121, context 122 (sample 15) – A larger soil sample (40 litres) was examined from fill 122, producing a higher number of cereal grains than most of the samples from Dwr-y-Felin, though not one of the highest concentrations (1.25 charred fragments per litre). Six possible free-threshing grains were recovered – the largest number from any of the samples, in addition to five emmer/spelt wheat grains. Barley grains were quite scarce (five grains) but oat grains were frequent, dominating the assemblage (25 grains). The frequency of the grains suggests that bristle oat rather than wild oat may have been present, in which case the assemblage could represent a mixture of grains destined for both human and livestock consumption. The only other charred plant remains were oat awn fragments and a hazelnut shell fragment. The presence of a high proportion of stones in the backfill (75%) and no tip lines suggests that the feature may have been infilled in a single event (GGAT sample description) with stones, clay and burnt 119

Nidum. Excavations at the Roman Auxiliary Fort 2010 to 2012: Neath, Neath Port Talbot CBC

charred waste including charcoal fragments and cereal grains. Sloe (Prunus spinosa) charcoal was identified at the assessment stage (C J Griffiths unpublished assessment report 2012). Industrial deposit, context 133 (sample 23) – This dark, ashy deposit contained isolated iron fragments and was thought to have possibly been a hearth-like structure, perhaps having been used as a ‘smithy’, though there were no signs of heat-affected deposits at the base (GGAT sample description). Black and silver slaggy-type heat affected materials were frequent in the flot, as well as hammerscale, and oak charcoal was present. The 40 litre soil sample produced 81 cereal grains comprising 64 barley grains, six oat grains, four emmer/spelt grains and a possible free-threshing wheat grain. The oats were similar in form to those recovered from sample 15 (slender grains, acute at the apex) suggesting that they also might have been bristle oat, although this is not a confirmed identification. As with sample 15, the only other items present were two hazelnut shell fragments. If the area had been used for metalworking the grain could represent food and fodder waste being burnt kindling/fuel or thrown on to the fire as a votive offering to ensure that the work went well. Layer 148 (sample 26) – This clay deposit overlay the pebbled road surface 147, possibly representing a period of abandonment. The small sample contained burnt bone, slaggy material, charcoal and coal fragments. It produced only two oat grains, an indeterminate cereal grain and a spelt glume base. The remains probably represent burnt waste that was blowing around the site at the time of abandonment. If so, it may be significant that the only cereals identified were oats and spelt wheat (see the discussion below). Internal floor surface or floor make-up, context 111 (sample 27) – This deposit consisted of a dark clay with pebbles which appears to have been a floor surface or floor make-up layer. One barley grain and four oats were present, as well as two indeterminate grains, a fragment of hazelnut shell and four weed/wild plant seeds (dock, Chenopodium rubrum, C. polyspermum and a small grass seed). Depending on the nature of the building the remains could have derived from spillage/waste from the preparation of food, or animal fodder. Evidence from environmental assessments carried out by Archaeological Services, University of Wales, Lampeter Environmental samples taken during excavations carried out on Site 677 and Site 674 were sent to the Archaeological Services department at the University of Lampeter for assessment in 2012. The assessments of the charred plant remains were carried out by C.J. Griffiths and based on 17 sub-samples from ditches, layers, ovens and pits in the area between the via sagularis and the ramparts in Site 677, as well as 8 samples from Site 674, six of which were analysed for this report (see above). Because the Site 677 samples were not taken further the assessment information is briefly noted here. Samples were assessed from within and around four large ovens. Although the flots were relatively charcoal-rich no cereal remains were recovered from the ovens to help explain their function. Charcoal from the small assessment samples included oak, hazel, ash and alder. Charcoal from these species was also recovered from the other features, together with a little birch and sloe. The only charred plant macrofossils to be recovered were an indeterminate

120

Nidum. Excavations at the Roman Auxiliary Fort 2010 to 2012: Neath, Neath Port Talbot CBC

cereal grain, a sedge nutlet and a straw node. Because of the scarcity of charred plant remains no further work was carried out on the samples from Site 677. In addition an assessment of three pollen samples taken from ditch fills was carried out by Sarah Jones (See pollen report in Section 15 below). Similar results were obtained from the three ditch samples indicating that open herbaceous grassland dominated the surrounding landscape although tree and shrub pollen, in particular alder pollen, was growing nearby, most likely as areas of alder carr on wet ground close to the fort. Hazel/oak woodland with some birch and ivy would have occupied drier ground a little further afield. The presence of Ericaceae pollen (heather family) indicated that some heathland existed in the area, or perhaps heathy vegetation was being brought into the fort. The single cereal pollen grain in one of the samples could have been brought in with processed grain or chaff/straw being brought in for fodder and bedding. It is also possible that arable cultivation was taking place in the area, although in view of the scarcity of cereal pollen, probably not in the immediate vicinity of the fort. Discussion The results from analysing samples from 25 features and layers from the first century timber fort, one external first century feature and six features / layers from the second century stone fort in Neath (Nidum) provide an insight into foods and other plant resources being brought into the fort during different periods of its use. Although most of the small samples produced low quantities of charred plant remains, by drawing the information together and comparing it with other Roman forts in Wales and further afield a picture emerges which can be built on if further excavations take place in the area in the future. Use of the data- Because the soil samples varied greatly in size from less than a litre to 40 litres (see the bottom of Table 15) analyses were carried out either by using presence / absence of plant taxa per sample (e.g. the presence of the different cereals in samples) or by calculating occurrences to taxa per litre of soil, so as to reveal concentrations of taxa across the site. This enabled comparisons to be made to other sites such as Caernarfon fort. Comparative sites â&#x20AC;&#x201C; Although several other forts have been excavated over the years it is hard to find comparable sites that have been extensively sampled for charred plant remains and have been fully published so that the raw archaeobotanical data is accessible. Ideally comparisons would be made with forts in Wales, since in parts of the country where the local soils and growing conditions are very different, comparisons could be affected by which crops are best suited to the geology and climate. One of the nearest forts is Caerleon (Isca) and although an important assemblage of charred malted spelt was recovered from a hut outside the fort (Helbaek 1964) no widespread excavation and sampling has been carried out (see the more detailed description below). The main site for which the full data was available was Caernarfon (Segontium), located on the Menai Straights in north Wales. The two Welsh forts are located in similar coastal locations on similar low-fertility soils; freely drained acid loams and more clayey, slow draining loamy soils (www.landis.org.uk/soilscapes/). In addition, some of the forts along Hadriansâ&#x20AC;&#x2122;s wall have been extensively sampled, for example Birdoswald (Bana) (Huntley 1997) and Annetwell Street, Carlisle (Huntley 1989) and these have provided valuable comparative data, particularly with regard to the frequency of barley grains and high occurrence

121

Nidum. Excavations at the Roman Auxiliary Fort 2010 to 2012: Neath, Neath Port Talbot CBC

of grassy and marshy materials at Annetwell Street, Carlisle. These assemblages are described below. Cereal remains on Roman military sites - The most notable factor concerning the charred assemblages from Dwr-y-Felin is that barley grains dominated samples from both phases of the fort and all areas of the fort investigated. In total 82% of the identified grains were barley, compared with 5% emmer/spelt (most likely spelt), 2% free-threshing wheat and 11% oats. The type of barley being brought into the fort was six-row hulled barley (Hordeum vulgare), a globally important crop plant that has been cultivated since Neolithic times and can grow on the widest range of soil types under a variety of climatic conditions. Chaff fragments are rarely found in significant numbers from this crop as it is easily separated from the ear, though remains tightly enclosed in the thin but protective husks (palea and lemma) until ‘hummelled’ (rubbed or pounded) to free the grain. The fragile waste from hummelling rarely survives charring. While barley can be a useful fodder, the whole grains are less digestible than oats and some horses dislike the taste. There is some debate as to the role that barley played in the military diet, as discussed by Britton and Huntley (2011, 42). While some authors note that barley bread was disliked by Roman soldiers and given to them as punishment rations (Davies 1971), classical documentary sources such as Pliny the Elder (Natural History) describe its value for increasing muscular strength, its use in recipes such as polenta and its medicinal role in making poultices (Book 18, Chapter 18; ‘The Nature of Barley’). Pliny does also say, however, that barley bread had “fallen into universal disrepute” and was mainly used for animal fodder (Natural History Chapter 17, translation by Bostock & Riley) so it appears to be use in bread that was a problem rather than general consumption. A recent paper by Britton and Huntley (2011) investigated the question as to whether barley was being consumed by the Roman army, rather than being present in archaeobotanical assemblages as the remains of animal fodder. The authors analysed bran in waterlogged faecal deposits from two Roman military sites on the Roman frontier in northern England, Birdoswald and Carlisle, as well as a contemporary civilian site and were able to confirm that significant amounts of barley were being consumed, with barley dominating one of the seven military samples and one of the three civilian samples. The medium to large size of the bran fragments suggested that the grains were probably being primarily consumed as whole or crushed grains in soups and stews, whilst the small wheat/rye bran fragments indicated that these cereals had been milled into flour to be consumed as bread. Interestingly, while oat does not appear to have been eaten at Birdoswald, two of the four samples from Carlisle contained small amounts of oat bran, both of which were second century in date. No oat bran was present in the two first century samples at Carlisle. This slight change in diet might also have taken place at Dwr-yFelin as no oats were recovered from the first century fort samples but five of the seven second century stone fort samples (Site 674) contained oat grains, oats amounting to 11% of the identifiable cereal remains recovered from this phase. Although previously thought to have primarily been used for fodder, being particularly valuable for providing energy to draught animals and cavalry horses, Britton and Huntley’s study suggests that they also played a role in the Roman diet in the later phase at Carlisle fort. At Dwr-y-Felin at least in samples from the later stone fort, the quantity of oat grains suggested that they were not present as crop

122

Nidum. Excavations at the Roman Auxiliary Fort 2010 to 2012: Neath, Neath Port Talbot CBC

contaminants, although it is not possible to be certain that they were destined for human consumption rather than for horse fodder. Other cereals represented in the samples were hulled wheat, with spelt (Triticum spelta) being confirmed by the presence of just two glume bases, and free-threshing wheat, most likely bread wheat (Triticum cf. aestivum-type). The latter was indicated by the presence of shorter, more rounded grains of wheat although no rachis fragments to confirm the identification. Wheat grains are morphologically too variable to identify to species level (Jacomet 2006). No emmer chaff was recovered but the presence of chaff was so scarce that consumption of emmer cannot be ruled out. It is clear from comparative reports of charred archaeobotanical assemblages from Roman forts that charred waste was not present in large amounts around these establishments, apart from when accidental or deliberate fires were set, particularly when buildings were abandoned. Reports from well-sampled sites such as Birdoswold (289 moderate-sized soil samples), Carlisle (205 moderate-sized samples) and Caernarfon (88 moderate to large soil samples (15l to 120l)) state how charred plant remains were only present in low quantities throughout the phases of occupation. Comparing the number of charred fragments per litre of soil processed at Dwr-y-Felin with Caernarfon, there were on average 4.8 fragments per litre compared with 2.0 fragments per litre at Caernarfon â&#x20AC;&#x201C; both very low levels when compared with most Roman settlement sites. At Carlisle, only 16% of the productive samples produced over 51 items, even though sometimes large volumes of soil were processed. The percentage at Dwr-y-Felin was 10% (percentage of analysed samples with >51 items), and in most cases the richer samples were dominated by grassy taxa rather than food remains. It is likely that military establishments would have been kept clean and tidy in comparison with civilian settlements, with wellorganised systems for waste disposal being provided, including toilet blocks. However, possibly a more important difference between domestic settlements and military establishments is the fact that foods such as cereals were almost certainly being brought into the forts as clean, fully processed grain. At Birdoswald, Caernarfon and Dwr-y-Felin very little chaff and few weed seeds specific to arable fields were recovered. At Carlisle a few samples produced relatively high numbers of chaff fragments and straw nodes but these could have been brought in for bedding and fodder. Probably the most frequent causes of charring plant material in the Roman period was a) parching hulled wheats prior to de-husking the grain and b) using cereal processing waste as fuel. Because cereal processing was not taking place in forts there was little opportunity for cereals, in particular hulled wheats, to become charred. It is likely that in many cases grain was being brought in already ground into flour, which might explain the relative scarcity of spelt wheat and bread wheat grains compared to barley grains. As suggested by Britton and Huntleyâ&#x20AC;&#x2122;s analysis of waterlogged bran fragments mentioned above, whole grains of barley were probably being used in soups and stews, as their thin husks make barley more suitable for this purpose than hulled wheats. In addition, although Pliny the Elder suggested that barley bread was no longer being consumed (Pliny died c. 79 AD), mixed flour breads using both barley and wheat flours are lighter, tasty and nutritious. Unlike the large deposit of sprouted spelt from outside the fort at Caerleon (Helbaek 1964), there were no signs of deliberate sprouting of grain for the production of malt from Dwr-y-Felin

123

Nidum. Excavations at the Roman Auxiliary Fort 2010 to 2012: Neath, Neath Port Talbot CBC

or any of the other forts used as comparanda in this report. Occasional sprouted grains or detached cereal sprouts found on some sites are likely to indicate poor storage conditions or damp animal fodder rather than malting (e.g. Caernarfon, Nye 1993). If fermented drinks were being consumed by some of the occupants it is likely that ale and wine would have been brought into the fort ready-made. Evidence for non-cereal aspects of the Roman military diet - There was no evidence for the consumption of pulses such as peas, beans or lentils at Dwr-y-Felin, although occasional traces of peas, beans and lentils have been recovered from forts such as Birdoswald (Huntley 1997), Carlisle (Huntley 1989) and Caerleon (Helbaek 1953). Davis (1971) suggests that these highprotein foods would have been staples of the Roman military diet. However, foods such as pulses, leaf vegetables, root vegetables, fruits and spices have a low chance of becoming charred and surviving in a recognisable form because they do not require parching prior to cooking. Because of this, evidence for the consumption of imported, exotic foods from Roman forts appears to be largely limited to sites that produce waterlogged features such as wells and ditches, such as Carlisle and Caeleon. Mineralised faecal deposits also have the potential for providing evidence of exotic fruits, pulses and flavourings but no examples from forts are known to the author. No waterlogged deposits were excavated at Dwr-y-Felin so it was simply due to good fortune and the sharp eyes of an excavator that half of a complete date (Phoenix dactylifera; flesh and stone) was recovered from a deposit (context 6684) from the internal part of the earlier timber fort. This was the only imported food item recovered from the excavations – an energy-rich food originating in the Near East and North Africa. In comparison, waterlogged and charred deposits at Carlisle contained a range of exotic flavourings, fruits and nuts including coriander, dill, nigella, fig, grape, olive, walnut and lentil (Huntley 1989). At Birdoswald where only a few waterlogged features were excavated the only exotic remains were a few fig seeds, although the dating of these remains was uncertain (Huntley 1997). No waterlogged deposits were excavated at Caernarfon fort (Segontium) (Nye 1993) and the most ‘exotic’ foods preserved by charring were a couple of strawberry seeds which may have been from native wild strawberry plants. Other native fruits and nuts commonly found in charred assemblages from forts in low numbers and also present in the Dwr-y-Felin samples were from brambles (blackberries), sloes (or possibly cherries at Caernarfon) and hazelnuts. Amongst these remains, hazelnut shell fragments are most likely to become charred by being thrown as waste onto hearths. None of the forts examined for this report produced high numbers of nutshell fragments in any features so the consumption of hazel nuts appears to have been merely as an occasional snack rather than a major food. This may also be the case for the exotic foods, since even sites with good waterlogged assemblages such as Annetwell Street, Carlisle (Huntley ibid.) produced low numbers of these remains. According to a recent review of ‘new plant foods in Roman Britain’ (van der Veen et al 2008) only four records of dates have been found, three of which were from military sites (2 intramural and 1 extramural). The fourth was from a burial. This compares with 91 records of fig, 28 of walnut and 17 of olive. A total of 31 ‘new foods’ were recovered from military sites across the British Isles amounting to 230 records. However, far greater numbers of records were recovered

124

Nidum. Excavations at the Roman Auxiliary Fort 2010 to 2012: Neath, Neath Port Talbot CBC

from major towns, in particular London (458 records). Most of the evidence (70% of the records) came from waterlogged deposits from features such as wells and ditches. The extramural military record came from Colchester, Lion Walk, where a wide range of imported foods were preserved by charring during the Boudiccan rebellion (Murphy 1984). It is calamitous events such as this (i.e. deliberate or accidental conflagrations) which provide ‘snapshots’ of the types of foods available to people during the Roman period, albeit the wealthy civilian population in this example rather than occupants of forts. Materials being brought into the fort as fuel, kindling, flooring or fodder – The largest area sampled from the early timber fort (Site 716) was located in the south-east corner of the excavation and consisted of layers of fired clay, some of which were rich in charcoal. A couple of alternative interpretations have been given for these deposits, one being that it was an area of industrial activity, although hearths, hammerscale and slag were not abundant. However, blacksmithing appears to have taken place in this area, being fuelled by charcoal and at a later date coal (Young, 2013c). It has also been suggested that the area may have suffered some sort of conflagration. Some of the samples from these layers produced the highest concentrations of charred plant remains, the dominant category of taxa being plants of grassy and marshy habitats with traces of heathland vegetation. The presence of hay meadow species such as yellow rattle, sheep’s sorrel and autumn hawkbit suggests that hay cut from acidic meadows, ranging from well-drained to marshy land, had been strewn around the area. This could represent kindling/fuel for smithing, burnt fodder, bedding or dung for livestock/horses, or floor coverings, thatching materials or turf roofing if buildings had been burnt down. Similar concentrations of grassy and marshy taxa were found at Carlisle and to a lesser extent at Birdoswald and Caernarfon. At these two forts there was also a little evidence for use of heathland or moorland resources and in the wells at Caerleon there was strong evidence of vegetation from a salt-marsh community being deposited. These remains demonstrate the extensive use of local resources around the forts for fodder, bedding, flooring, kindling and possibly a variety of other purposes as described below. It is important to take this into consideration when measuring the effect that Roman military establishments had on the local population. Comparable plant assemblages from Roman forts - Few forts in Wales have been excavated and sampled on a wide enough scale for charred plant remains to make useful comparanda. Excavations at the legionary fortress of Isca at Caerleon in 1958 and the 1980’s were localised and quite different in the archaeobotanical information they provided. The former excavation produced a large deposit of charred spelt grain and possibly rye, with traces of possible bread wheat and barley, lentil and horsebean plus weed seeds (Helbaek 1964). The deposit, recovered from a destruction layer in a timber hut, was thought to date from around AD 80-130 and was located around 150m south-west of the south-west gate of the fortress beneath the metalling of a road. The importance of this large deposit was that almost all of the spelt showed signs of sprouting, providing evidence for the preparation of malt for brewing. In addition, the presence of frequent seeds from several leguminous weed species native to Mediterranean and southern European countries indicated that imported grain was being brought to the fortress. Later excavations in the legates house (Scamnum Tribunorum) at Caerleon involved the taking of bulk samples from waterlogged fills of two wells. A wide range of waterlogged plant remains 125

Nidum. Excavations at the Roman Auxiliary Fort 2010 to 2012: Neath, Neath Port Talbot CBC

were recovered from the samples, providing evidence for vegetation from salt-marsh, wetlands, drier grasslands, arable crops (wheat chaff and barley grains) and weeds, hedgerows and woodlands and a few exotic foods (Caseldine and Busby 1993). However, these samples do not provide comparative information about the principal foods being consumed on a daily basis. Even so, it is useful to see evidence suggesting that resources from areas such as the Severn Estuary were being exploited, as well as obtaining evidence for the consumption of dill, coriander, grapes, fig and date at Caerleon. The date (Phoenix dactylifera) evidence consisted of a single waterlogged date stone from well 1. Explanations for the possible uses of salt-marsh hay were that the remains may have been brought in in animal dung or fodder, or that (as suggested for a Roman well in York (Hall et al 1980)) material had been brought in with shellfish or in ships ballast. The fort at Caernarfon produced very similar, sparse assemblages to Dwr-y-Felin although spelt was the dominant cereal (70% of the identified grain) and some emmer glume bases were recovered. Barley was present in fairly low concentrations amounting to 28% of the identified grain. The difference between spelt and barley percentages, however, were largely due to three samples that were rich in spelt-type grains. At Dwr-y-Felin the percentages were much more skewed towards barley, accounting for 82% of the grain overall, with similar percentages in both phases of the fort (Sites 716 and 674). No obvious changes through time were observed in the charred plant assemblages from Caernarfon, suggesting that the environment and economy were stable from the later 1st to later 4th century AD. Despite being located hundreds of miles away on Hadrianâ&#x20AC;&#x2122;s Wall, the Roman fort at Birdoswald (Banna), Cumbria, was widely sampled and analysed for charred plant remains so provides data that can be used for comparative purposes. A total of 289 samples were processed from granaries, ditches, ovens, a toilet block and other structural features such as ditches and turf walls (Huntley 1997). The majority were taken from the granaries but these were found to contain very little charred plant material, presumably because no major accidents/burning events had occurred. This contrasts with the abundant charred grain deposits (over 30 thousand grains) from granaries in the fort at South Shields which provided evidence for the storage of bread wheat, spelt and smaller amounts of barley in the 3rd century AD (van der Veen 1994). The possibility of the bread wheat having been imported was raised but not confirmed. The study at Birdoswald, however, is useful as a comparandum in that sampling was widespread across the fort and statistical analyses were undertaken (Huntley 1997). In one way, however, the results might differ slightly from Dwr-y-Felin â&#x20AC;&#x201C; at Birdoswald only the >1mm fractions of the flots were fully sorted and quantified, and the 0.5 to 1mm fractions were scanned. At Dwry-Felin the fine fractions were important in providing strong evidence for the bringing in of quantities of hay from acidic and damp grasslands. Many of these remains (including small grass seeds, clover-type seeds etc) might have been missed by scanning. The principal observations at Birdoswald in relation to Dwr-y-Felin were as follows; i) Relatively low numbers of charred plant remains were recovered. ii) Wheat (spelt and free-threshing wheat) and hulled six-row barley were present in almost equal quantities.

126

Nidum. Excavations at the Roman Auxiliary Fort 2010 to 2012: Neath, Neath Port Talbot CBC

iii) Evidence for â&#x20AC;&#x2DC;exoticâ&#x20AC;&#x2122; foods was scarce, consisting of a single charred Celtic bean and occasional waterlogged fig seeds (the latter not securely dated). iv) Arable weeds, ruderal plants, grassland and wet ground plants were present but not abundant in any samples and chaff was scarce. The low occurrence of charred plant remains and scarcity of exotic foods applies to both forts. The recovery of slightly greater proportions of wheat at Birdoswold could be due to the sampling of different areas of the fort, in particular the fact that the majority of samples came from the granaries (75%) while at Dwr-y-Felin most samples came from the internal area of the 1st century timber fort (52%). Very different assemblages are likely to be recovered from each of these areas, with stored grain most likely contributing to the majority of the granary samples and activities possibly relating to smithing having taken place in the excavated area of the internal part of the fort at Dwr-y-Felin. However, the granaries at Birdoswold were fairly unproductive, so spelt grains were only slightly more frequent than barley grains. The fort at Annetwell Street, Carlisle produced a much wider range of evidence because of the presence of frequent waterlogged deposits. It is interesting to note that at this fort barley was the dominant cereal, even though on this site a wider range of samples were taken across different areas of the fort when compared with Dwr-y-Felin. Spelt was the second most frequent cereal, and bread wheat plus a little emmer were recovered. The small numbers of rye and oat grains were thought to represent weeds. As noted above, a range of exotic fruits, nuts and flavourings were recovered from the waterlogged samples. Some changes were observed through time but these mainly related to changes in use of specific structures such as furnaces. Unlike at Caernarfon, Dwr-y-Felin and Birdoswold, a few samples from Carlisle produced frequent chaff and some contained arable weed seeds such as corn cockle. It is possible that some crop processing was taking place at this fort, or that cereal processing waste was being brought in to use as fodder, bedding, fuel, kindling as well as other purposes. Conclusions The small samples from limited areas of the Roman fort at Neath have provided evidence for the use of barley, spelt and (probably) bread wheat in the fort, with the possible addition of oats in the later period, although it is uncertain whether these remains were destined to be consumed by humans or livestock/horses. Although barley was notably dominant in all areas and phases, and evidence for the consumption of barley by soldiers has been demonstrated in other forts, this report notes that caution should be observed when translating frequencies of charred cereal grains into importance in the diet. Many foods would have been brought into the fort in forms that would have little chance of being preserved as identifiable charred plant macrofossils. By chance, an almost complete date fruit became charred, providing evidence that some luxury imported foods were being consumed, though possibly only by the higher ranks. The scarcity of cereal chaff and arable weed seeds indicates that cereals were being brought to the fort as fully processed, clean grain. A range of other materials were being brought into the fort from local woodlands and hedgerows (fuel wood, nuts and berries), meadows and marshes (hay and marsh hay) and heathand (heather for use as fuel, fodder, bedding, thatch). Comparisons with other sites (some of which were sampled extensively and produced very productive

127

Nidum. Excavations at the Roman Auxiliary Fort 2010 to 2012: Neath, Neath Port Talbot CBC

waterlogged assemblages) were useful to help determine which aspects of the data were similar to other forts across the British Isles, which appear to differ significantly and which differ because a relatively small area has been sampled at Dwr-y-Felin and only charred assemblages were available for study.

128

Nidum. Excavations at the Roman Auxiliary Fort 2010 to 2012: Neath, Neath Port Talbot CBC

Table 15. Analysis of the charred plant remains

KEY TO TABLE 15 Feature types: C=channel; D=ditch; G=gully; L=layer; P=pit; PH=post-hole; S=slot; Location: EXT=external; INVM=intervallum; INT=internal Habitat preferences: A=arable; C=cultivated; D=disturbed; E=heath; G=grassland; H=hedgerow; M=marsh/bog; P=pond/ditch; S=scrub; W=woods; a=acidic soils; d=damp; n=nutrient-rich; o=open habitats; s=sandy; w=wet/damp *=imported food SITE

674

716

SUB-DIVISION

40-44

29,32,46

27&30

26,31,3637

3&3b

sample

S129

D140

INT

160

L INT

141

S121 INT

111

S109 INT

148

L INT

133

L INT

122

G6771 INT

108

P6860 INT

6684

H6923 INT

6678

L INT

6772

L INT

6790

L INT

6722

L INT

6561

L INT

6564

L INT

6683

L INT

6417

L INT

6433

S6587 INT

6418

PH6454 INT

6449

L INVM

6431

C6592 INVM

6544

L INVM

6472

L6483 INVM

6597

P6332 INVM

6593

S6176 INVM

6038

P6121 INVM

6440

P6035 INVM

6336

6079 S6093 INVM

6177

6124 INVM

6115

6026 INVM

6037

6104

6030 INVM

INVM

6278 EXT

6258

context

P6025

L6321

feature

EXT

CEREAL GRAINS

Triticum cf. aestivum-type (cf. free-threshing wheat grain) 1

Triticum dicoccum/spelta (emmer/spelt grain) 1

Hordeum vulgare ssp. vulgare (hulled six-row barley grain, twisted grain)

Hordeum vulgare (hulled barley grain) 1

129

Nidum. Excavations at the Roman Auxiliary Fort 2010 to 2012: Neath, Neath Port Talbot CBC

Hordeum sp. (poorly preserved barley grain) 1f

153

Avena sp. (wild/cultivated oat grain) 2

Indeterminte cereals 3

CEREAL CHAFF

T. spelta L. (spelt glume base) 1

T. dicoccum/spelta (emmer/spelt glume base) 1

Avena sp. (oat awn frag.) +

WEEDS, WILD & OTHER ECONOMIC PLANTS

Pteridium aquilinum (L.) Kuhn (bracken pinnule frags) GaE 1

Ranunculus acris/bulbosus/repens (buttercup achene) DG 1

Vicia/Lathyrus sp. (<2mm small vetch seed) CDGH 1

130

Nidum. Excavations at the Roman Auxiliary Fort 2010 to 2012: Neath, Neath Port Talbot CBC

Medicago/Trifolium/Lotus sp. (medick/clover/trefoil seed) GD 2

Prunus spinosa L. (sloe stone) HSW cf. 3

Rubus sect. Glandulosus (bramble seed) DHSW 1

Corylus avellana L. (hazelnut shell frag.) HSW 1

Persicaria lapathifolia (L.)Gray (pale persicaria achene) CDw 1

Persicaria maculosa/lapathifolia (redshank/pale persicaria achene) Cwo

Persicaria hydropiper/mitis (water-pepper/tasteless waterpepper achene) GwaMP

cf.2e

11+3 8e

Polygonum aviculare (knotgrass achene) CD 14

cf. 1

Rumex sp. (dock achene) CDG 3

Rumex acetosella L. (sheep's sorrel achene) EoGCas 2

Stellaria media(L.) Vill. (common chickweed seed) Cno 1

131

Nidum. Excavations at the Roman Auxiliary Fort 2010 to 2012: Neath, Neath Port Talbot CBC

Chenopodium rubrum L. (red goosefoot seed) CD 1

Chenopodium polyspermum L. (many-seeded goosefoot seed) CD

cf.1

Solanum nigrum L.(black nightshade seed) CDn 1

Plantago lanceolata L.(ribwort plantain seed) Go 1

Phoenix dactylifera L. (whole date fruit) *

1 wh ole

Prunella vulgaris L. (selfheal nutlet) GDWo 2

Odontites vernus/Euphrasia sp. (red bartsia/eyebright seed) ADG

Rhinanthus sp. (yellow-rattle seed) G

Lapsana communis L. (nipplewort achene) DHWo 1

Leontodon autumnalis L. (autumn hawkbit achene) G 1

Eleocharis subg. Palustres (spikerush nutlet) MPw 1

132

Nidum. Excavations at the Roman Auxiliary Fort 2010 to 2012: Neath, Neath Port Talbot CBC

Carex sp. (trigonous sedge nutlet) MPw 1

Carex sp. (lenticular sedge nutlet) MPw 4

Cyperaceae cf. Schoenoplectus sp. (cf. club-rush nutlet) MPw 1

Cyperaceae cf. Eriophorum sp. (cf. cottongrass nutlet) Mw 3

Lolium perenne/rigidum (ryegrass caryopsis) GD 4

Bromus sect. Bromus (brome grass caryopsis) AD 1f

Poaceae (small-seeded grass caryopsis) CGD 1

101

cf. Poaceae NFI (3mm thin cf. grass seed) 1

Poaceae (grass-type stem frag.) CGD +

cf. Ericaceae leaves (cf. heather leaves) Ea 15

Indeterminate plant remains 2

133

Nidum. Excavations at the Roman Auxiliary Fort 2010 to 2012: Neath, Neath Port Talbot CBC

TOTAL 19

102

112

1.66

1.88

1.11

0.25

8.4

3.42

3.6

1.35

1.1

5.95

2.75

8.28

0.5

3.6

3.5

2.9

1.9

4.4

3.6

1.5

7.5

6.1

4.7

3.9

186

2.84

7.8

2.7 5

0.4

3.2

2.5

23.8

0.3

1.7

261

4.8

1.6

ha nd pic ke d

6.44

3.4

3.2

3.4

0.4

1.3

3.73

1.3

2.1

0.9

3.8

13.2

6.5

soil sample volume (litres) 1.24

1.18

2.2

0.09

3.9

0.96

2.4 8

15.3

3.4

33.3

26.2

0.4

charred plant remains per litre of soil processed

134

Nidum. Excavations at the Roman Auxiliary Fort 2010 to 2012: Neath, Neath Port Talbot CBC

15.

The pollen samples (Site 677) (Table 16)

By Sarah Jones Introduction Three samples from ditch fill features from Dwr y Felin School, Site 677, Neath were selected for pollen assessment. This report outlines the laboratory procedures, and offers an assessment of the range of taxa observed, state of preservation and pollen concentration. Sample/context

025/1378

026/1379

031/1397

Trees & Shrubs

Alnus

Pollen type

Betula Corylus t.

Hedera

Quercus

Herbs

Cyperaceae und.

Lactuceae und.

Poaceae und.

Caryophyllaceae und.

Cerealia t.

Ericaceae

Filipendula ulmaris

1 1

Lonicera

Aster t. Plantago lanceolata

Ranunculaceae

Rumex t.

Urtica

TLP Pteropsida

102

113

105

Polypodium

Pteridium

Sphagnum

Pteropsida monolete undiff. Indeterminate grains (Total) Degraded

Crumpled Lycopodium

Table 16. Pollen data from Site 677, Dwr-y-felin School, Neath, expressed as actual pollen counts.

Laboratory procedure Volumetric samples of 1 ml were taken for initial assessment of pollen identification, concentration and preservation. Samples for pollen analysis were prepared using standard 135

Nidum. Excavations at the Roman Auxiliary Fort 2010 to 2012: Neath, Neath Port Talbot CBC

techniques (Moore et al., 1991), including treatment with HCl to remove carbonates, HF digestion to remove silicates and acetolysis to digest organic matter. Prior to chemical digestion, however, all samples were passed through microsieves (aperture 10Âľ), a procedure that has been shown to be especially effective in concentrating pollen grains in sediments of low primary pollen. A known quantity of Lycopodium spores were added to each sample to enable the calculation of pollen concentrations within the samples (Stockmarr, 1971). The residues were mounted in silicon oil and analysed under a Leica DMR microscope at a magnification of x400, with critical identifications at x630 and, where necessary, under oil at x1000. As the analysis was to assessment level, the aim was to achieve a sum of 100 total land pollen (TLP), and this was achieved in all of the samples. Pollen was identified using standard pollen keys (e.g. Andrew, 1980; Moore et al., 1991), and type collections. Plant nomenclature follows Stace (1997). The state of pollen preservation is referred to using the categories identified by Jones et al (2007), e.g. corroded (biochemical deterioration) and degraded (chemical deterioration). Results Pollen concentration and preservation The results of the initial assessment are presented in Table 16, and are expressed as actual pollen counts. In all of the samples, pollen grains were abundant and an assessment level pollen count of 100 TLP was easily achieved. Pollen preservation was variable, with relatively low numbers of indeterminate pollen grains recorded in Sample 025 (Context 1378). Deterioration of pollen grains and differential accumulation of more robust pollen types did not appear to be significant in any of the samples, and although some thinning of the pollen grain exine and mechanical damage (e.g. crumpled and broken grains) were noted. Sample 031 Context 1397 A similar pollen assemblage was recorded in this sample, being dominated by Poaceae, Alnus and Corylus t. Other tree pollen included Quercus and Betula (birch). This sample contained a similar variety of herbaceous pollen types to that of the other assemblages, including Lactuceae und., Plantago lanceolata, Ericaceae, Aster, Filipendula ulmaria (meadowsweet) and Rumex (dock). One grain of Cerealia t. pollen was recorded. Spores of Polypodium and Pteridium were also noted. Discussion Some preliminary palaeoenvironmental inferences can be made from the pollen assessment from Dwr y Felin school. a) The taphonomy of pollen incorporated in ditch fill sediments is complex and the data may be difficult to interpret, as pollen and spores of such features may come from a variety of sources, including secondary external sources. However, the pollen evidence from all three samples suggests that during infill of the various ditch features, the landscape was dominated by open herbaceous grassland. The representation of tree and shrub pollen, especially that of Alnus, suggests alder carr growing on wetter substrates nearby and mixed woodland stands further afield.

136

Nidum. Excavations at the Roman Auxiliary Fort 2010 to 2012: Neath, Neath Port Talbot CBC

b) The herbaceous pollen types recorded, such as, ribwort plantain (Plantago lanceolata), dandelion types (Lactuceae), buttercups (Ranunculaceae) and docks (Rumex), are more usually indicative of grassland and pastures than of arable, disturbed or bare ground (Behre, 1986). Cereal pollen was rare, but recorded in Sample 031, Context 1397, possibly representing arable farming within the vicinity of the site. Small quantities of meadowsweet (Filipendula ulmaria), sedges (Cyperaceae) and pinks (Caryophyllaceae) are indicative of some wetter grassland habitats. c) Due to the limited range of pollen types identified, it is difficult to suggest a chronology for the ditch features. The pollen assemblages from all three samples are broadly comparable and may indicate a similar depositional age for the ditch fill features. 16.

Plant macrofossils (Site 677) (Tables 17-18)

By C J Griffiths Introduction Following the excavations at Dwr y Felin School, Neath, 17 sub-samples from larger bulk samples (site 677) were received at the palaeo-environmental laboratories, Trinity Saint Davidâ&#x20AC;&#x2122;s, Lampeter. The aim was to assess the environmental material from the samples should carbon dating be required and to provide information about the economy and/or environment of the site and its landscape. Method The samples were sieved using a simple wash-over technique, given the small quantity of material available for the assessment per sample; the flot and residues were collected on 250Âľm mesh and air dried. All the flots and residues were examined using a Wild stereoscopic microscope. Attention was given to whether charred plant macro fossils were present. A small proportion of the wood charcoal was identified using a LEICA DMR microscope with incident lighting to provide material suitable for radio carbon dating. Identification was by reference to Schweingruber (1978) and Schloch et al (2004). Nomenclature follows Stace (1997). Results Overall the samples contained a variable amount of wood charcoal, with very few identifiable charred plant macrofossils (Table 17). Sample 010, context 1052: A charcoal rich sample, with both ring porous and diffuse porous wood charcoal present. One bone fragment was present. Sample 014, context 3005: A silty, stony sample, with a small quantity of wood charcoal, including a large fragment of oak (Quercus sp.) charcoal, water logged seeds were also present. Sample 020, context 1329: A red-brown stony sample, with wood charcoal present; there is also a possibility that charred weed seeds are present. Sample 025, context 1378: A white-grey stony sandy sample with a small quantity of wood charcoal; a fragment of straw with node was also noted.

137

Nidum. Excavations at the Roman Auxiliary Fort 2010 to 2012: Neath, Neath Port Talbot CBC

Sample 026, context 1379: A small quantity of wood charcoal was present, with a small quantity of indeterminate charred organic material. Sample 031, context 1397: A stony sample, with no charcoal or other plant remains present. Sample 033, context 1465: A charcoal-rich sample, with both ring porous and diffuse porous wood charcoal present. Non-charred bone fragments were noted. Sample 034, context 1338: The sample contained principally pottery fragments which included rim sherds. A small quantity of wood charcoal was also present. Sample 038, context 1510: A small charcoal-rich sample, with both ring porous and diffuse porous wood charcoal present. Sample 041, context 1553: A charcoal-rich sample, with both ring porous and diffuse porous wood charcoal present. Also present was a charred sedge nutlet, (Carex sp.) and indeterminate charred organic material. Sample 049, context 1597: A charcoal-rich sample, with ring porous and diffuse porous wood charcoal; some round wood fragments were also present. Small fragments of burnt bone were also noted. Sample 053, context 1726: A small charcoal rich sample; some burnt bone fragments were also present. Sample 054, context 1759: A charcoal-rich sample, with both ring porous and diffuse porous wood charcoal present. The presence of probable metal working waste was noted; coal fragments were also present. Sample 055, context 1753: A charcoal-rich sample with both ring porous and diffuse porous wood charcoal present. Sample 063, context 1869: A charcoal-rich sample, with both ring porous and diffuse porous wood charcoal present, including round wood, twiggy material and some tree buds. A wheat/barley (Triticum/Hordeum) grain was also present. Sample 064, context 1863: A charcoal-rich sample, with both ring porous and diffuse porous wood charcoal present. Sample 065, context 1851: A dark gritty sample with 2 or 3 charcoal fragments; a small quantity of water logged sedge (Carex sp.) nutlets were present along with fragments of wood. Discussion All the samples examined provided only a very small amount of flot and residue, which fell broadly into two categories: the samples from the ovens were relatively charcoal rich and those from the ditch samples contained waterlogged material. The oven samples contained no evidence of cereal crops, with only one or two identifiable seeds present: a bedstraw type (Galium sp.) found in context 1329, a charcoal spread from oven 1023 and a sedge (Carex sp.) nutlet from a charcoal spread found amongst the ovens. The only charred cereal grain from the assemblage was from context 1869 and comprised a wheat/barley (Triticum/Hordeum) grain.

138

Nidum. Excavations at the Roman Auxiliary Fort 2010 to 2012: Neath, Neath Port Talbot CBC

The ditch samples provided small quantities of waterlogged monocotolydenous material with intermixed charcoal, including a charred straw fragment from context 1378 from trench 2. A small quantity of charcoal was identified from the samples which contained relevant material (Table 18). A list of charcoal fragments suitable for C14 dating was compiled. The species identified from the contexts included oak (Quercus sp.), Ash (Fraxinus excelsior), hazel (Corylus avellana), birch (Betula sp.), alder (Alnus glutinosa) and blackthorn (Prunus spinosa). The assemblage suggests that fuel for the fort and especially the ovens was collected locally possibly from the edge of woodland, the oak and ash may also be from woodland local to the fort. The overall assemblage of plant macrofossils from the site is disappointing, the charred plant remains provided evidence of Spelt (T. Spelta) wheat, both twisted and straight hulled barley (Hordeum sp. grains and hazel (Corylus avellana) nut shell fragments. The remains derived from the submitted plant macrofossil sub-samples considered here can only be interpreted as incidental detritus.

139

Nidum. Excavations at the Roman Auxiliary Fort 2010 to 2012: Neath, Neath Port Talbot CBC

Period (General)

Sample Context

Feature

Volume/ litre

Charred Plant remains

Waterlogged Plant remains

Charcoal

Bone frags

Roman

010

1052

Charcoal spread adjacent (east ) of oven 1007

0.5

Roman

014

3005

Ditch fill T3

0.5

Roman

020

1329

Charcoal spread oven 1023

0.4

Cf Bestraw (Galium sp.) seed present

Roman

025

1378

Ditch fill T2

0.9

Straw frag

Roman

026

1379

Ditch fill T2

1.0

Indeterminate charred organic material

Roman

031

1397

Ditch fill T2

1.0

Roman

033

1465

Rampart drain

0.5

Roman

034

1338

Clay deposit over hard standing 1137

0.3

038

1510

Charcoal deposit oven 1040

0.5

Roman

041

1553

Charcoal spread amongst ovens

0.6

Roman

049

1597

Charcoal deposit. Early feature

0.5

053

1726

Fill of pit 1712

0.5

054

1759

Charcoal from oven 1039

0.5

055

1753

Charcoal filled timber slot

0.2

Roman

063

1869

Lower clays

0.3

Early Roman /prehistoric

064

1863

Early pit fills

0.6

Roman

065

1851

Clay base of T4 extraction cut/recess

0.6

Table 17. The Plant macrofossil remains from Dwr y Felin School, Neath

140

Comments

Pottery fragments present

Sedge (Carex sp.) and indeterminate charred organic material

Charred wheat/barley (Triticum/Hordeum) grain

Sedge (Carex sp.)

Nidum. Excavations at the Roman Auxiliary Fort 2010 to 2012: Neath, Neath Port Talbot CBC

Sample

038

041

Context

1052

1329

1759

1510

1553

1753

1869

1863

1597

1726

1465

1338

1851

Feature

Oven 1007

0ven 1023

Oven 1039

Feature 1040

Timber Slot

Lower clays

Early Pit fills

Early Feature

Pit 1712

R. Drain

Clay deposit1137

Base T4 cut/ extract

Quercus spp. (Oak) Fraxinus excelsior (Ash) Betula spp. (Birch) Alnus glutinosa (L) Gaertner (Alder) Corylus avellana L. (Hazel) Prunus spinosa L. (Blackthorn) Total

Charcoal Spread ovens -

Table 18. Charcoal identification from Dwr-y-Felin, Neath - Site 677

141

Total

Appendix A:

GeoArch Report 2013/27: Residues from Neath

Archaeometallurgical residues from Dwr-y-Felin School,Neath (GGAT 677 & 716) Dr T.P. Young

Abstract This report details analytical investigations undertaken on a series of ironworking residues (from c1502) and on a single piece of problematic dense slag (from c1676), all from excavations in and near the Roman fort at Neath. The smithing residues proved to be mostly a coherent suite of materials, the composition of which could be modelled through simple mixture of iron oxide with assimilated hearth lining. One of the examined iron-working slag cakes, an iron-rich, hollow example, did not appear to fit this model. It appears likely that deep weathering of this piece has led to preferential loss of the interstitial phases and that the iron-rich bulk composition is due to secondary iron oxides formed after the weathering of much metallic iron. It is likely, but not certain, that this slag cake also originated within the same chemical system as the other examples; other interpretations are discussed. The dense slag fragment from within the fort (c1676) is very unusual in being dominated by wustite and becoming more highly oxidised and less iron-rich towards one margin. Firstly iscorite and magnetite overgrow/ replace the wustite dendrites, then there is a zone with magnetite dendrites and finally slag with equant euhedral magnetite. The structure of the slag is layered, with particular horizons demarcated by oxide crusts. These resemble the chilled, iron oxide-rich, margins of flow lobes within a tapped slag, but are thicker (up to 200Âľm) and here may be more indicative of oxidation of growth stages of the slag. This highly unusual slag can be paralleled only by two examples from elsewhere, both interpreted as being derived from bloom-smithing or bloomrefining. The microstructure is suggestive of the slag having formed through oxidation of metallic iron, but is unlike textures observed around iron particles in conventional bloomsmithing slags or adjacent to the bloom in a bloomery. It appears likely that the slag was produced in a form of remelting hearth akin to that described by Evenstad in the 18th century. Such hearths may be used to control the content of carbon (e.g. in hearth steelmaking) and phosphorus (e.g. for dephosphorising phosphoric iron) in iron, but also forms a useful approach to recycling iron, particularly the iron formed in a bloomery but not incorporated into the bloom, or lost from the bloom during processing.

Contents Abstract .............................................. Methods .............................................. Results .............................................. Distribution of materials ................................. Description of materials General ............................................. Sample details ......................................... Interpretation .............................................. Iron working .............................................. Dense iron slag from c1676 ............................ Conclusions .............................................. References .............................................. Figure Captions .............................................. Glossary ..............................................

1 2 2 2

Appendix A: other bulk analyses ....................... 30 Table A1: .............................................. 31 Table A2: .............................................. 32 Appendix B: EDS microanalyses ...................... 33 Table B1: .............................................. 34

2 3 5 5 6 7 8 9 10

Appendix C: image archive ................................ 42 Archive plate index ........................................... 43 Archive plates .............................................. 44

Table 1: Details of samples................................ 12 Table 2: Major element analyses by XRF ........... 13 Table 3: Trace element analyses by ICP-MS ..... 14 Table 4: Major elements, iron-free normalisation 15 Table 5: Entrained hearth ceramic compositions 16 Table 6: Glass composition in entrained clasts .. 17 Figures:

.............................................. 18

GeoArch Report 2013/27: Residues from Neath

been constructed with elements expressed as oxides in weight% calculated stoichiometrically, except for mineral structure calculations, where the measured oxygen has been used.

Methods All archaeometallurgical materials from the project were examined visually (with a low-powered binocular microscope where required) and the summary catalogue of examined material was presented by Young (2013, Table 1). The material included a range of iron-working residues, including relatively low-iron ‘gravelly’ slag masses. The main purpose of the analytical phase was to examine evidence for a genetic relationship between these and more conventional smithing hearth cakes (SHCs). To this end, a piece of hearth lining and a suite of five slag cakes of differing iron contents were selected from a single context (c1502; samples NTH1 to NTH6; Table 1). The second aim of the analysis phase was the investigation of an unusual iron oxide-dominated slag (NTH7; from deposit c1676).

Throughout this report standard mineral terminology is applied to both natural and anthropogenic materials – although artificial phases are no longer strictly considered to be minerals. This report is formatted in two sections: the main report is followed by appendices for the archiving of all data not included in the body of the main report. This project was commissioned by Martin Tuck, of the Glamorgan-Gwent Archaeological Trust.

The selected samples were slabbed on a diamond saw and subsamples used firstly for preparing a polished block for use on the SEM and secondly for crushing for preparation of a whole-sample chemical analysis, Polished blocks for investigation on the SEM were prepared in the Earth Science Department, The Open University. Electron microscopy was undertaken on the LEO S360 analytical electron microscope in the School of Earth and Ocean Sciences, Cardiff University. Microanalysis was undertaken using the system’s Oxford Instruments INCA ENERGY energy-dispersive x-ray analysis system (EDX).

Results Distribution of materials The investigated smithing materials (a piece of hearth lining and a suite of five slag cakes of differing iron contents; samples NTH1 to NTH6; Table 1) were selected from amongst the assemblage from a single context (c1502). This context was one of several from the area of the ‘mansio’ (including contexts 1076, 1111, 1276, 1387 and 1502) which together yielded a smithing residue assemblage totalling 13.8kg. This was slightly more than half of all archaeometallurgical residues recovered from stratified Roman contexts during this project (a total of 26.4kg) which, although still a relatively small quantity, raises the likelihood of there having been a smithy (at least for a period) in this area of the site.

Chemical analysis was undertaken using two techniques. The major elements (Si, Al, Fe, Mn, Mg, Ca, Na, K, Ti, and P) were determined by X-Ray Fluorescence using a fused bead on the Wavelength- Dispersive X-Ray Fluorescence (WD-XRF) system in the department of Geology, Leicester University (this also generated analyses for S, V, Cr, Sr, Zr, Ba, Ni, Cu, Zn, Pb and Hf). Whole-specimen chemical analysis for thirty six minor and trace elements (Sc, V, Cr, Co, Ni, Cu, Zn, Ga, Rb, Sr, Y, Zr, Nb, Mo, Sn, Cs, Ba, La, Ce, Pr, Nd, Sm, Eu, Gd, Tb, Dy, Ho, Er, Tm, Yb, Lu, Hf, Ta, Pb, Th, U) were undertaken using a sample in solution on the ThermoElemental X-series Inductively-Coupled Plasma Mass Spectrometer (ICP-MS) in the School of Earth and Ocean Sciences, Cardiff University (this also generates lower quality results for Fe, Mn, Ti, P that are used mainly for QA purposes).

Description of materials General The specimens of slag sampled for this project vary from tongue-like cakes to more typical morphologies of smithing hearth cake (SHC). ‘Tongue’ (in the original usage a ‘pro-tuyère tongue’) was a term introduced by Young (2012a) to describe small, tabular slag cakes forming below and in front of, the blowhole of a ceramic tuyère, in some cases above a larger slag cake in the base of the hearth, at Clonfad, Co. Offaly. The examples from Neath they have a similar morphology, but are likely to have been the only slag cake formed in the hearth. They lack the strongly convex base of the classic morphology of SHC, reflecting the fact that they contained little iron-rich, fluid, slag. They can be regarded here as a sub-group of smithing hearth cakes (SHCs). Specimens NTH2 (130g, but incomplete) and NTH3 (218g) may be classed as tongues. Both show considerable surface evidence for inclusions of ceramic and sandstone. Individual inclusions, or groups of inclusions, appear to be responsible for the ‘gravelly’ texture in hand specimen. The sample from NTH2 for examination on the SEM spanned the contact between a partially melted clast and a lobe of dense grey slag. The sample for SEM examination from NTH3 proved to have most of its surface occupied by a large, partially melted sandstone clast.

The results of the chemical analyses are presented in Table 2 (major elements by XRF) and Table 3 (trace elements by ICP-MS). The lower quality analyses are present in Appendix A as Tables A1 (trace elements by XRF) and A2 (major elements by ICP-MS). Locations of EDS analyses are presented as sample-areaanalysis (e.g. NTH2 SOI2 #3). The area of the sample in a particular image is referred to as SOIn (SOI = Site Of Interest in the terminology of the INCA microanalysis software) and individual analyses (points or areas) are labelled #m. The microanalytical data are presented in Appendix A, Table A3. Images of all SOIs are included in Appendix B, including, where appropriate, details of the analysed points/areas. All EDS analyses were collected with all elements analysed (including oxygen, but not carbon; all samples were carboncoated). Analytical totals were frequently far from 100%, because the analytical system is designed to provide totals of 100% from spot analyses in the centre of the field. The area analyses required for this project are not standardised in the same way and will diverge from a total of 100% (either above or below, depending on the location of the area with respect to the centre of the field).In order to make the microanalytical results simply comparable across materials (and also sites), no attempt has been made to adjust for the oxidation state of elements with variable valency. The figures employed in the report have therefore

Specimen NTH4 is a small (106g) SHC with a convex base, which is rather pale and has deep fuel dimples. It measures 50mm x 70mm x 40mm (of which the bowl is 20mm deep). Its upper surface is slightly concave, but with large raised lobes of lining-influenced slag at the proximal side, and includes numerous pale, raised blebby lobes. These lobes are largely formed of partially-melted ceramic material, with

GeoArch Report 2013/27: Residues from Neath

a superficial dark glaze, very similar to the upper surfaces of the tongues. The main body of the slag, forming the convex bowl, is about 20mm deep and is formed of a more uniform grey iron slag.

and 2-3mm in length. Where the olivine is coarse it may have a clear interstitial glass, but generally the glass contains later, finer generations of olivine. In places where the olivine is fine, it may form a very fine-scale texture with visible glass at all (e.g. Figure 2a upper right). These extremely fine-grained areas correspond to zones appearing to be glass in hand-specimen. The olivine shows quench textures, varying from careous textures in the coarse crystals through to complex interstitial dendrite to fine, feathery, eutectic textures.

Specimen NTH5 is a rather more substantial and wellformed SHC, weighing 270g and measuring 70mm x 95mm x 45mm (of which the bowl is 35mm deep). The top of the SHC is formed of translucent green material with ‘micaceous’ olivine crystals, but also with deeply impressed by fuel. There are large cavities below top layer – with individual rounded cavities of up to 30mm across. The bowl is formed by a thin layer of vesicular slag, 6-10mm thick, around the base and surrounding the large cavities. The sample for SEM examination was taken from this thin bowlcrust.

The section shows clasts (these are probably not attached to the wall) of fragments of bloated ceramic and quartz grains bound by glass. The clasts have abrupt margins with the enclosing slag. One clast shows development of a strongly aluminous interstitial glass to the remnant quartz grains.

The densest smithing slag specimen is NTH6. This 200g piece is probably a slightly incomplete SHC, much of which is of a dense, moderately coarse-grained slag, with large vesicles (including tubular ones arising from the basal surface) that overall resembles the burr region of a larger SHC. The upper surface is very degraded, but includes surviving patches of a dark surfaced, coarse slag with fuel dimples in places, whereas in others the slag is pale, glassy and finely vesicular. This suggests an uneven distribution across the upper surface of silicic material derived from the hearth lining. Below a thin layer on the upper surface, the slag appears dense, homogeneous and uniformly iron-rich. The sample for SEM examination was taken from this homogeneous slag.

Where analysed, the fayalite showed cores of approximately Fa96Fo4 and margins of Fa98Fo2 with 1% calcium substitution. The bulk silica:alumina ratio of the bulk sample is 9.4 (by weight). The U:Th ratio is 0.30. The sample shows marked enrichment in Cr (541 ppm) and Ni (836ppm). The Upper Crust-normalised rare earth element (REE) profile (Figure 1) is flat (but with slight relative light REE depletion) and rather ‘saw-toothed’, with elevated relative abundance of Eu, Dy and Sm. It is approximately parallel to that of NTH 1. ΣREE is 63.4ppm.

Sample NTH7 was a 56g angular fragment of extremely dense slag. The surface was weathered and had become etched to reveal long linear dendrites of iron oxide.

NTH3 (Figure 2 d) The slag comprises sheaves of fayalite crystals. The slag is in contact with a partially-melted sandstone clast (Figure 2d). Close to the contact, the fayalite is delicate and fine. The fayalite-rich slag in these areas sometimes bears small irregularly-shaped fine crystal aggregates, of which the mineralogy is uncertain because of their very fine crystal grain-size, but they possibly contain feldspar and may possibly include pyroxene.

Sample details NTH1 (Figure 2 a-c) This sample was taken from the unvitrified rear face of a 22g fragment of hearth lining.

The regions of relict sandstone are quite coarse-grained (remnant cracked quartz grains are up to 250µm). Where the regions are embayed, the zones of interstitial glass are coated with layers rich in feldspar (probably mainly potassium feldspar) up to 100µm in apparent thickness. Adjacent more electron-dense areas are too fine-grained for analyses of single crystals, but it is possible that some of the very fine iron-rich minerals are a pyroxene rather than olivine. Serial analyses across the slag-ceramic interface show gradational elemental compositions even where the ceramic is not embayed (Figure 3). This clearly indicates a degree of chemical transfer across the margins of the sandstone fragments.

The bulk chemical analysis shows an iron content (16.97% expressed as wt% Fe2O3) which is moderately elevated in comparison with what might be expected from a local alluvial clay. Other South Wales furnace clays range from 2.5-5.5% Fe2O3 where unaltered; the higher level in NTH1 may suggest this sample was more contaminated than suggested by its external appearance. The ceramic is relatively low in calcium, magnesium and titanium. It only shows rather moderate levels of potassium. The bulk silica:alumina ratio is 7.3 (by weight). The U:Th ratio is 0.33. The Upper Crust-normalised rare earth element (REE) profile (Figure 1) is flat (but with slight relative light REE, LREE, depletion) but rather ‘saw-toothed’, with elevated relative abundance of Eu, Dy and Sm. Total REE (ΣREE) is 103.4ppm.

The bulk silica:alumina ratio of the bulk sample is 7.3 (by weight). The U:Th ratio is 0.39. The Upper Crust-normalised rare earth element (REE) profile (Figure 1) is flat (with slight relative light REE depletion) but rather ‘saw-toothed’, with elevated relative abundance of Eu, Dy and Sm, closely parallel to that of NTH1. ΣREE is 69.6ppm.

NTH2 This specimen is a slightly incomplete gravelly, tongue-like cake, with a section of attached vitrified hearth lining. The tongue is typically 10-20mm thick, but expands to over 30mm near the wall contact. The sample for SEM examination was taken from this thicker zone. The exposed surfaces of lobes of dense slag show a maroon surficial oxidation layer.

NTH4 (Figure 4 a-d) This sample has a heterogeneous microstructure (Figure 4 a) with some patches with early wustite dendrites, some with sheaves of elongate olivine similar to NTH2 and NTH3, others with coarse olivine and all with scattered former vesicles. The vesicles are commonly rimmed by leucite, with a lining of ingrowing wustite-cored fayalite in a possible cotectic with further leucite (Figure 4 b). The pore fillings

This SEM sample shows a felted mass of near eutectic olivine in glass (Figure 2 a,b). The olivine forms, at its coarsest, complex elongate crystals up to 100µm in width

GeoArch Report 2013/27: Residues from Neath

comprise wustite-cored fayalite, minor Ti-bearing hercynite and glass.

The interstitial areas are mostly of glass bearing extremely fine olivine dendrites, but locally there are cotectic intergrowths of olivine and leucite (Figure 5c).

Where the texture is more even, it comprises bundles of elongate fayalite crystals, up to 4mm in length and 600µm wide (Figure 4 c). The interstitial areas are dominated by leucite and a leucite-wustite cotectic, with some hercynite, but are badly weathered across much of the specimen. Where analysed, the fayalite showed cores of approximately Fa99Fo1 and margins of Fa100 with up to 1% calcium substitution.

The main coarse olivine ranges from Fa97Fo3 in the core to Fa100 on the outside, typically with 1% calcium substitution on the margins. The olivine in the cotectic with leucite is Fa100 with 2% calcium substitution. The silica:alumina ratio of the bulk sample is 8.0 (by weight). The U:Th ratio is 0.38. The sample shows marked enrichment in Cr (12201ppm) and Ni (2709ppm).

The bulk silica:alumina ratio of the bulk sample is 7.47 (by weight). The U:Th ratio is 0.34.

The Upper Crust-normalised rare earth element (REE) profile (Figure 1) is flat (but with slight relative light REE depletion) but rather ‘saw-toothed’, with elevated relative abundance of Eu, Dy and Sm. The profile is approximately parallel to that of NTH1 and particularly similar to that of NTH3. ΣREE is 57.1ppm.

The Upper Crust-normalised rare earth element (REE) profile (Figure 1) is flat (with slight relative light REE depletion) but rather ‘saw-toothed’, with elevated relative abundance of Eu, Dy and Sm. It is approximately parallel to that of NTH1, but is very slightly more depleted towards the light REE. ΣREE is 81.0ppm.

NTH7 (Figure 5 d,e; Figure 6) This sample is unusually dense and possesses a strongly layered internal structure (Figure 6). This layering incudes areas bounded by thick (up to 200µm) crusts (Figure 6), mainly formed of magnetite, which resemble flow-lobe margins as seen in tapped smelting slags particularly, but also in flow slags and prilly smithing slags. This oxidised lobe margins are also accompanied by a markedly oxidised character to the slag throughout the upper half of the specimen.

NTH5 (Figure 4 e,f) NTH5 showed a heterogeneous texture (Figure 4e). The coarsest areas showed large olivine crystals formed of subequant, porous, grains up to 1.2mm. These however, may have been components of larger crystals. The large olivine crystals contained sparse blebs of iron. The large olivine crystals were of fairly constant Fa98Fo2. The interstitial areas showed a few medium-sized olivine crystals (up to 800µm long and 80µm wide) with cotectic wustite, but mostly showed a sparse primary dendritic wustite, followed by fine-grained olivine (up to 400µm long and 50µm wide) and with an interstitial glass. The fine olivine grains showed a composition of Fa99Fo1 to Fa100 with up to 1% Ca substitution. Other areas show finer textures, but with essentially the same microstructure on a finer scale. Parts of the specimen showed heavy weathering.

Layers towards the base of the sample (as orientated in Figure 6; the original orientation is unknown) are dominated by tightly-packed dendritic growths of wustite (Figure 5 d,e). These bear a few small former vesicles, now filled by silicate-rich slag (Figure 5e), but otherwise these layers are greater than 80% wustite. Such layers constitute Zones 1-4 of the section examined in detail (Figure 6). Zones 1 and 2 are separated by Crust 1, an approximately 60µm thick lobe margin of wustite and magnetite. Approximately 2.5mm of Zone 1 was observed within the section studied in detail, but it was significantly thicker than this elsewhere. Above Crust 1, the wustite-rich zones 2-4 were approximately 9.5mm thick.

The sample contains a few effectively unreacted wall clasts, with abrupt contacts with the enclosing slag (Figure 4f). The bulk chemical sample of this specimen showed the highest iron content of the SHC specimens. The bulk analysis shows a very significant loss on ignition (approximately 1.5%) compared with the gain on ignition (negative loss on ignition) showed by the other residues. This loss indicates the loss of volatiles; in this context, most likely hydroxyl/water from hydrated iron oxy-hydroxides present as weathering products. Very high levels of secondary weathering products are usually the result of the corrosion of metallic iron, rather than of the slag itself. The silica:alumina ratio of the bulk sample is 7.8 (by weight). The U:Th ratio is 0.30.

Above Zone 4 lies a layer approximately 600-800µm in thickness, comprising Zones 6 and 7, in which the strongly dendritic nature of the underlying layer is replaced by a coarse, blebby or pseudo-dendritic structure in which the wustite is replaced with, and overgrown by, magnetite and iscorite (Figure 6, enlarged section). The iron oxides here comprise 60-70 of the texture and they are followed by elongate olivine crystals of up to 600µm in length. Where analysed, these olivine crystals had cores of Fa96Fo4 with 2% calcium substitution, apparently grading to Fa97Fo3 with 5% calcium substitution. Interstitial areas were poorly preserved but appear to show, at least in part, an intergrowth of leucite and a calcic fayalite (Fa100 with 17% calcium substitution).

The Upper Crust-normalised rare earth element (REE) profile (Figure 1) is flat (but with moderate relative light REE depletion) but rather ‘saw-toothed’, with elevated relative abundance of Eu, Dy and Sm. The profile is noticeably relatively depleted towards the light REE compared to those of the other SHCs (NTH2,3,4,6) and the lining (NTH1). ΣREE is 25.6ppm.

The overlying Zones 7 and 8, together approximately 1mm thick, show a return to a dendritic structure, which becomes slighter (although as much as 1mm in length) and more sparser upwards, decreasing to approximately 15% of the microstructure at the top of this layer. The wustite dendrites are partially altered to magnetite, with minor iscorite plates. At the top of the layer a few euhedral magnetite grains of approximately 100µm appear. The oxide dendrites are followed by olivine in elongate form (up to 600µm in length).

NTH6 (Figure 5 a-c) This sample shows some areas with a microstructure very close to that of sample NTH5 (Figure 5a). The majority of the sample, however, shows a slightly more quenched texture, in which the larger equant olivine grains (up to 500µm) show continuity with some of the interstitial dendritic crystals (Figure 5b). In some areas the olivine is more elongate, with crystals greater than 1.5mm by 500µm.

These partially oxidised zones are overlain by Crust 2, a 100µm thick layer including some relict wustite, but mainly formed of magnetite.

GeoArch Report 2013/27: Residues from Neath

Above the oxide crust, a third lobe (1.5 to 3mm thick) comprises Zones 10, 11 and 12. These variously comprise euhedral equant magnetite or magnetite dendrites with wustite relicts and minor iscorite. The oxide grains comprise 30-50% of the microstructure and are followed by olivine of up to 800µm length.

feldspar (Figure 11). The data form a very poor fit to the fields of the FeO-SiO2-Al2O3 ternary diagram (Figure 10), because the relatively high proportion of sodium, potassium and calcium with respect to aluminium (so that a phase diagram with hercynite as the main aluminous phase is not appropriate). Utilisation of the FeO-SiO2-anorthite pseudoternary (Levin et al. 1964, p. 288, fig. 869; as employed by Morton & Wingrove 1969, 1972) is also not particularly appropriate, since the concentrations of sodium and potassium are high with respect to calcium.

The fourth lobe continues to the top of the piece. It is separated from the lobe below by Crust 3, which is up to 200µm thick and formed of magnetite. The lobe is approximately 3mm thick and contains two discontinuous, possibly partially resorbed crusts – Crusts 4 and 5. The fourth lobe coincides with Zone 13, in which equant, euhedral magnetite of up to 150µm is the primary phase, followed by olivine of up to 800µm length.

The approach adopted here is recast the analyses in terms of a pseudo-ternary plot with poles FeO, SiO2 and (KAlSi3O8 + NaAlSi3O8 + CaAl2Si2O8, i.e. total feldspar). Plotting the fields of the FeO - SiO2 - CaAl2Si2O8 phase diagram provides some approximation by which the data may be interpreted. On this diagram, the slag microanalyses for samples NTH2, NTH3 and NTH6 plot close to the location of the feldspar-fayalite eutectic in the FeO - SiO2 CaAl2Si2O8 diagram, whereas those from NTH4 plot closer to the position of the hercynite-fayalite eutectic (hercynite was recorded as a late phase in NTH4).

The whole-sample bulk analysis is dominated by iron oxide. When the major element analysis is recast and normalised on an iron-free basis (Table 4), then several elemental oxides (SiO2, Al2O3, Na2O, K2O and TiO2) are in approximately the same ratio as they are in the main suite of SHCs and the lining (NTH1-6). Other elemental oxides (MnO, MgO, CaO, P2O5) are enriched by 5 to 8 times their concentrations in the lining sample (NTH1).

Fragments of lining incorporated as clasts into the slag cake were observed in NTH2, NTH3 and NTH4, as well as NTH5. Examples of the incorporated material were analysed by EDS (Table 5). The clasts incorporated into NTH2, NTH4 and NTH5 appeared to be bloated and vitrified ceramic, with abundant quartz sand, now comprising relict cracked quartz grains in a glassy matrix. In sample NTH3, clasts of bloated sandstone were recorded, rather than of hearth ceramic.

The silica:alumina ratio of the bulk sample is 7.9 (by weight). The U:Th ratio is 2.55. The Upper Crust-normalised rare earth element (REE) profile (Figure 1) is moderately flat (but with significant relative light REE depletion) and rather ‘saw-toothed’, with elevated relative abundance of Eu, Dy and Sm. The profile is this distinct from those of the other samples in this project. ΣREE is 17.8ppm.

Most observed examples of such clasts possessed apparently abrupt contacts with the encasing slag (e.g. Figure 2b, Figure 4f). Serial EDS analyses across the margins showed, however, that compositional gradients exist, demonstrating the assimilation of the clasts into the melt (Figure 3). Such assimilation processes must generate the bulk composition of the slag cakes through the overall reaction between iron oxides lost from the workpiece and the melted hearth wall. The opposite side of that process, the assimilation of the iron oxides, may be observed in NTH4 and NTH6 SOI7 (Figure 5a). The abruptness of the contacts between the vitrified clasts and the slag may be because the initial vitrification and partial melting of the clast takes place at high temperatures. The liquidus temperature for compositions equivalent of the typical interstitial glass of the clasts may be as high as 1400C and for the marginal feldspar-rich zones seen in NTH3 about 1300C according to the FeO - SiO2 - CaAl2Si2O8 phase diagram. The actual liquidus temperatures may have been slightly lower (because of the influence of iron, sodium and potassium), but they are likely to be significantly higher than the c. 1100C liquidus temperature for the main slag phase. The textures now observable in the slag cakes must reflect the final solidification of the slag, at a much lower temperature than (and therefore long after) the solidification of the partial melt in the clasts.

Interpretation Iron working residues from c1502 With the exception of sample NTH5, the smithing residues (NTH2, NTH3, NTH4, NTH6) proved to be a coherent suite of materials the composition of which may be modelled through simple mixture of iron oxide with assimilated hearth lining. The bulk chemical compositions of the samples form a reasonably linear array between iron and a ceramic close in composition to that of NTH1. The similarity of the samples is demonstrated by a comparison of the major element composition, when recast and normalised on an iron-free basis (Table 4). Additional evidence for the ceramic composition can be found in the composition of ceramic clasts within the slags, as measured by EDS (Table 5). The close relationship between the samples is shown by the REE profiles (particularly when normalised against NTH1; Figure 7), the similarity of the silica:alumina and uranium:thorium ratios, and the various bivariate plots (Figures 8, 9).

The interstitial glass in the clasts represents a partial melt phase derived from the clasts (Table 6). The melt composition shows a shift away from the SiO2 pole on the ternary diagrams (Figure 11), with a modest increase in iron content (probably indicating a mild fluxing effect in the original environment of the clasts). For NTH2 and NTH4 these plot (on the FeO-SiO2-feldspar diagram, using the fields of the FeO - SiO2 - CaAl2Si2O8 diagram) on the silica side of the silica-anorthite eutectic, but for NTH3 the analysed interstitial glasses lie in the anorthite field. On examination of the analyses (Table 6) and when plotted in various binary combinations (Figure 12) the composition of the interstitial glasses can be seen to have systematic variation. There is co-variation in the contents of sodium, potassium and calcium of the interstitial glass in the partially melted clasts. For the apparently relatively unaltered clasts,

As would be expected from such a dataset, the array is not perfect, with, for instance, NTH2 having a higher silica:alumina ratio than the other residues, presumably due to a slightly different ceramic composition (a higher proportion of sand or sandstone grains). The nature of the assimilation of the hearth lining varied from complete reaction through to entrainment of almost compositionally unaltered (if vitrified) clasts. The further detail of the assimilation process can be obtained through consideration of the microanalysis of the microstructural components by EDS. This is illustrated by the ternary diagrams of FeO-SiO2-Al2O3 (Figure 10) and FeO-SiO2-total

GeoArch Report 2013/27: Residues from Neath

calcium contents are low, commonly below detection. These relationships suggest that although external fluxing by wood ash rich in calcium and potassium may have been involved in some of the samples, for sample NTH3, the marked proximity of the composition of the partial melt to the feldspar pole of the FeO – SiO2 – feldspar ternary plot suggests that melting of detrital feldspar grains within the sandstone is likely to be the major control on melt composition in this instance.

the bulk sample is derived from the furnace lining (as is the usual case in bloomery iron making; Thomas & Young 1999a and 1999b). In this case, if all the alumina in the bulk residue was derived from the lining, then the residue would be approximately 10.6% lining. Such a value would be in line with those calculated from mass-balance modelling of bloomery furnaces smelting Bristol Channel Orefield ores. If this proportion of the value for the lining is deducted from each elemental concentration in the residue, then a model composition for the ore under these assumptions. Such a model would require the ore to have extremely low trace element contents. Ores of such a composition do occur within the Bristol Channel Orefield, but no examples have been recorded in which an ore of this type has developed a structure or mineralogy similar to NTH7 on partial reduction.

One of the examined iron-working slag cakes (NTH5) did not appear to fit this model. It differed in being a hollow slag cake, with an iron-rich composition. It appears likely that deep weathering of this piece has led to preferential loss of the interstitial phases and that the iron-rich bulk composition is due to secondary iron oxides formed after the weathering of a large proportion of original iron metal.

A second possibility is provided by the occurrence of iscorite-bearing slags amongst the residues produced by post-medieval techniques for oxidising the impurities (carbon, phosphorus, sulphur, silicon…) from cast iron as part of the process for the production of wrought iron. Iscorite has been recorded in residues (macro- and micro-) from both refining (Young in prep.) and puddling (Young 2009, 2011a, 2011b). All these residue types, however, typically have elevated levels of phosphorus (usually to several wt%) and often sulphur, neither of which is a feature of the present material. There is no evidence that the specimen was intrusive into its originating context and its degree of weathering is similar to other Roman residues from the site.

Microanalysis of fresh slag shows compositions close to those observed in sample NTH4. Areas of weathered slag show a range of compositions extending to a point on the FeO-SiO2 line (Figure 11) that would correspond to fayalite with minor wustite. This is interpreted as the compositional trend with removal of the interstitial phases. It is likely, but not certain, that this slag cake also originated within the same chemical system as the smithing residues. The observed differences are interpreted as resulting from the weathering of the slag and of metallic iron inclusions. There are, however, other possible interpretations of this piece. The development of a large cavity below the upper surface and the presence of metallic inclusions are properties that have been observed in slag cakes produced from hearth refining/remelting of iron in order to control phosphorus and carbon contents (author’s pers. obs.; Sauder 2013), including small scale steel making. These processes are described in more detail in the following section (the discussion of sample NTH7). It may possibly be significant that the REE profiles for both NTH5 and NTH7 (Figures 1 and 7) show a slight relative depletion of the LREE compared with the profiles of the other samples. Unfortunately, because of both the poor preservation of this cake and the current poor understanding of hearth steel making slags, it is not possible to discriminate between these possibilities.

The third possibility, having dismissed the previous two as unlikely explanations, is suggested by a very few occurrences of similar slags elsewhere. In particular a comparison might be drawn with a sample from Plot 331 of the Brecon-Tirley Gas Pipeline (near Kingstone, Herefordshire) of latest Iron Age or early Roman age. This site yielded an assemblage of SHCs, interpreted to be residues from bloom-smithing, some with very high wustite contents and one of which (Young 2012b, Sample K; p. 4 and plates 1a, 1c, A27, A28 and A29) showed features similar to the present material. The EDS analyses of Sample K had a slightly lower average iron content than the bulk analysis NTH7 (71.3wt% expressed as FeO), but like NTH7 it displayed iron oxide crusts separating lobes that had differing mineralogy and iron content. Iscorite occurred in a small area of the specimen. Sample K was taken from a small dense SHC, measuring 110mm x 110mm x 40mm and probably originally weighing around 900g (854g was surviving after sample preparation).

Dense iron slag from c1676 This slag piece was unusually rich in iron oxides, possessed a somewhat layered structure, contained iron oxide crusts defining lobe boundaries and also showed a gradient in the oxidation state of the iron (with the unusual occurrence of iscorite).

A second example of an SHC containing iscorite was recorded from the Viking settlement at Woodstown, Co. Waterford. Sample WTN1 (Young 2009, p. 4-5 and Plates 1 and 2) was taken from a 300g fragment from a SHC of unknown overall size. The slag cake appeared to have been produced from a deep, fluid slag puddle. The microstructure of a section through the SHC shows that, although the bulk iron content is relatively low (56.6wt% expressed as FeO), it shows a similar style of stratification to NTH7, with localised zones characterised by the content and morphology of wustite, magnetite and iscorite. The sample does not, however, show the development of a lobate texture or internal iron oxide crusts. This sample was somewhat different to the other residues from Woodstown that were interpreted as from bloom-smithing and was tentatively proposed to be from an oxidative decarburising process.

These features place the specimen outside the normal range of Roman slag compositions and textures. The compositional gradient away from dense wustite is suggestive of a progressive oxidation of iron metal. There are three possible interpretations of the material. The first possibility is that the piece represents a fragment of partially reduced ore (or reduced and re-oxidised ore). In this model, the ore would need to be very high grade (since the non-iron content of the slag forms a moderately good match to the hearth ceramic seen in the smithing residues described above). Although incompletely reduced ore is not uncommon on bloomery smelting sites, none has been recorded with a layered structure, with iron oxide crusts or bearing iscorite. The potential ore composition could be modelled by subtracting the influence of furnace lining from the bulk composition of the sample. The overall similarity of the silica:alumina ratio for the bulk sample with those from analysed smithing slags and the general form of its REE profile, would suggest much of the non-iron component of

The less strongly reducing environment that can be interpreted to have prevailed during the formation of these two examples parallels that for NTH7. Such conditions might be entirely accidental, but it would appear likely that they are the result of a deliberate process (or processes). Slags interpreted as bloom-smithing residues may contain significant quantities of metallic iron, but these tend have

GeoArch Report 2013/27: Residues from Neath

been observed as apparently dissolving into a fayalitic melt (e.g. Young 2012a, Clonfad, samples CFD6b and CFD13b). Similar textures, with mainly relatively low-iron content slag and only a limited development of wustite-rich microstructures, surrounding iron particles have been observed in poorly consolidated bloom in furnace slags (e.g. Young 2012a, Ballykilmore, sample BYK5; Young 2012b, bloom fragment, sample D2, Plot 430 of the Brecon to Tirley Gas Pipeline; Young 2012c, bloom from Cannop). It is possible that the oxidation of a bloom-smithing slag could have been produced accidentally, but this seems unlikely

Ireland (e.g Woodstown, Young 2009), the tongue morphology is seen in slag cakes with an iron content (expressed as FeO) of less than approximately 45%. In the present assemblage NTH2 and NTH3 have the tabular tongue morphology whereas NTH5 and NTH6 are more classic SHCs. NTH4 is borderline, retaining the blebby texture of a tongue, but developing a more roundedly convex base. Both tongues and classic SHCs developed a fayalitic slag. The various examples differ in the proportion of fayalitic slag (lower in the tongues, higher in the SHCs), but also in the composition of the slag. The fayalitic slag of the tabular tongues NTH2 and NTH3 was iron poor compared with that in the convex-based cakes. NTH4 and NTH6 show slag that is slightly more iron-rich on average than that in the tongues, but which is very heterogeneous. The heterogeneity of these samples appears to be controlled in part by iron-rich areas (assimilated iron oxides?), but also because of the role of the vesicles in controlling composition.

Processes involving remelting of iron in a strong air blast might be suitable candidates for providing such an environment. Remelting pig iron to decarburise it is the basis of the finery process (as well as the refinery, puddling and other related processes) and small-scale decarburising of cast iron as early as the Saxon period has been suggested by Mack et al. (2000). More recently a small scale steel-making process described by Aristotle (in Meteorologica) involving the opposite of fining, the remelting of a low carbon bloomery iron to produce a high carbon steel, long assumed to be fanciful, has been rehabilitated first, theoretically by Wagner (1990) and later, practically by Sauder (2013). This process has been shown to produce steel, with a controllable carbon content, and a low proportion of slag inclusions. The renewed interest in the ‘Aristotle process’ has also led to renewed interest in the hearth processes described by Ole Evenstad in Norway in the late 18th century (Jensen 1968; Wagner 1990). There is currently much experimental interest in the ‘Evenstad process’ which appears to employ slightly different hearth configurations (all with a strong horizontal air-blast into a very hot hearth) to first de-phosphorise iron in a first remelt, and then increase its carbon content to make steel in a second remelt. There is, therefore, now a growing recognition that remelting is likely to have played a part in many bloom refining and processing traditions alongside the more conventional forge-and-fold style of bloom-smithing. Unfortunately, there is as yet on published account of the residues produced by these different processes, although a limited amount of analytical work has now been undertaken (author’s unpublished work). It is not yet possible to attribute the sorts of residues described here to any particular process, beyond the overall observation that NTH7 appears to have been formed in a compositional gradient suggesting the residue may have formed in contact with metallic iron and that the environment. At least on occasion, was insufficiently reducing to maintain all the iron oxides as wustite.

Dominantly sharp margins to the observed ceramic inclusions in the slags, and rather gentle compositional gradients in the slag away from these margins, suggests that the ceramic clasts reacted with the iron slags rather quickly and homogenised rapidly. The textures, composition and morphology of the ‘gravelly’ slags is thus likely to be due to a low rate of iron supply during the formation of the slag mass, rather than these cakes representing incipient SHCs limited by short work periods. Processes with a low rate of iron loss from the workpiece might include simple forging tasks (such as nail production) and repair work (such as retempering tools). Sample NTH5 showed slag of a markedly different composition (low in aluminium and alkalis), perhaps suggesting loss of the interstitial phases during weathering. Poor preservation of this specimen prevents certainty as to whether this cake was similar to the main suite of smithing residues or whether it represented a rather different process. The most iron-rich slag, NTH7, resembles materials on other sites where bloom processing is believed to have been taking place. It is not clear what process, precisely, was responsible for the production of this residue, but it was possibly one of the family of remelting techniques (hearth remelting processes may be employed for a variety of purposes, including adjusting the composition of iron, such as for small scale steel-making, and recycling scrap).

Conclusion The evolution of the smithing residues can be interpreted as a two stage process. Firstly loss of material from the wall may have taken place in a largely particulate manner, with fragments detaching from the wall (although the particles themselves may have been very viscously fluid), rather than a progressive flow of low viscosity partial melt (although that may play a role too). A significant degree of fluxing of the wall probably took place in situ, with both wood ash (particularly rich in calcium and potassium oxides) and iron oxides acting as fluxes. The partially melted particles apparently varied greatly in size, perhaps up to about 10mm, with small stone clasts sometimes being rather larger. It is these partially melted clasts that give the ironpoor slags their ‘gravelly’ appearance. The compositional difference between the tongue-like gravelly slag cakes and the denser iron-rich conventional SHCs sheds light on the development of the smithing slags. The difference is principally in the proportion of iron they contain. In both the present examples and those from

GeoArch Report 2013/27: Residues from Neath

References WAGNER, D.B. 1990. Ancient carburisation of iron to steel: a comment. Archeomaterials, 4, 111-117.

JENSEN, N.L. 1968. "A treatise on iron ore as found in the bogs and swamps of Norway and the process of turning it into iron and steel", Abridged translation of Evenstad 1790. Bulletin of the Historical Metallurgy Group, 2, 61-65.

YOUNG, T.P. 2009. Archaeometallurgical residues from Llynfi Vale Ironworks, Maesteg, GeoArch Report 2009/18, 23 pp.

LEVIN, E.M., ROBBINS, C.R. & McMURDIE, H.F., 1964. Phase Diagrams for Ceramists, Columbus, Ohio, American Ceramic Society.

YOUNG, T.P. 2009. Ferrous archaeometallurgical residues from Woodstown 6, GeoArch Report 2009/22, 66 pp.

MACK, I., McDONNELL, G., MURPHY, S., ANDREWS, P. & WARDLEY, K. 2000. Liquid steel in Anglo-Saxon England. Historical Metallurgy, 34, 87-96.

YOUNG, T.P . 2011a. Archaeometallurgical residues and associated materials from the M74 Completion, Glasgow. GeoArch Report 2010/22, 274 pp.

MORTON, G.R. & WINGROVE, J. 1969. Constitution of bloomery slags: Part 1: Roman. Journal of the Iron and Steel Institute, 207, 1556-1564.

YOUNG, T.P. 2011b. Some preliminary observations on hammerscale and its implications for understanding welding. Historical Metallurgy, 45, 1, 26-41.

MORTON, G.R. & WINGROVE, J. 1972. Constitution of bloomery slags: Part II: Medieval. Journal of the Iron and Steel Institute, 210, 478–488.

YOUNG, T.P., 2011c. Archaeometallurgical residues from the N7 Castletown to Nenagh scheme, Camlin 3 (E3580), Co. Tipperary. GeoArch Report 2011/23. 62 pp.

SAUDER, L. 2013. Making steel in the ‘Aristotle furnace’. In: D. Dungworth and R.C.P. Doonan (eds) Accidental and Experimental Archaeometallurgy. Historical Metallurgy Society, Occasional Publication No. 7.

YOUNG, T.P. 2012a. Appendix 3 -. Exploiting the bog -: iron production and metalworking, pp. A3.1 – A3.60 In: P. Stevens & J. Channing, Settlement and Community in the Fir Tulach Kingdom. National Roads Authority and Westmeath County Council.

SCHAIRER, J.F. & YAGI, K. 1952. The system FeO- Al2O3SiO2. American Journal of Science (Bowen volume), 471512.

YOUNG, T.P. 2012b. Archaeometallurgical residues from the Brecon to Tirley gas pipeline. GeoArch Report 2012/18, 75pp.

TAYLOR, S.R. & McLENNAN, S.M. 1981. The composition and evolution of the continental crust: rare earth element evidence from sedimentary rocks. Philosophical Transactions of the Royal Society, A301, 381-399.

YOUNG, T.P. 2012c. Investigation of a bloom fragment from Cannop, Forest of Dean (Glos HER 37920). GeoArch Report 2012/21, 5pp.

THOMAS, G.R. & YOUNG, T.P. 1999a. A graphical method to determine furnace efficiency and lining contribution to Romano-British bloomery iron-making slags (Bristol Channel Orefield, UK). In: YOUNG, S.M.M., BUDD, P.D., IXER, R.A. and POLLARD, A.M. (eds). Metals in Antiquity, British Archaeological Reports International Series, 792, 223-226. Archaeopress, Oxford.

YOUNG, T.P. 2013. Assessment of archaeometallurgical residues from Dwr-y-Felin School, Neath (GGAT 677 & 716). GeoArch Report 2013-11, 31pp. YOUNG, T.P. in prep. Archaeometallurgical residues from Ynysfach Ironworks, Merthyr Tydfil. GeoArch Report 201329,

THOMAS G.R. & YOUNG, T.P. 1999b. Bloomery furnace mass balance and efficiency. In: POLLARD, A.M. (ed) Geoarchaeology: exploration, environments, resources, Geological Society of London, Special Publication, 165, 155-164.

GeoArch Report 2013/27: Residues from Neath

Figure Captions Figure 1: REE profiles, normalised to Upper Crust (normalisation factors after Taylor & McLennan 1980)

Figure 5: BSEM Images NTH6, NTH7 a. NTH6 SOI7. Scalebar 1mm. minor primary wustite dendrites (bright), with wustite scale-like particle in lower left, followed by large olivine crystals with complex quenched morphology in interstitial glass (dark). b. NTH6 SOI3. Scalebar 1mm. Rather equant olivine (pale) associated with a finer elongate olivine in the interstitial glass (dark). Black areas are large porosity. Vesicle partially infilled by secondary weathering products at top. c. NTH6 SOI9. Scalebar 100µm. Detail of an interstitial area, showing the two complementary textures – firstly fine late dendritic olivine in glass (right) and secondly an olivineleucite cotectic (left). d. NTH7 SOI7. Scalebar 2mm. Densest zone of wustite (bright) showing layering and large vesicles, partially filled by a slag with fine dendrites of wustite, elongate olivine (mid grey) and glass (dark grey). e. NTH7 SOI9. Scalebar 300µm. Detail of (d) showing margin of vesicle.

Figure 2: BSEM Images NTH2, NTH3 a. NTH2 SOI1. Scalebar 1mm. Typical olivine-rich microstructure. b. NTH2 SOI5-6. Scalebar 600µm. Abrupt contact of olive-rich slag with the margin of a deeply vitrified inclusion. Dark cracked grains are quartz, set in a variably grey glass with elongate dark crystals of mullite and bright specks of Fe-Ti oxides. c. NTH2 SOI8. Scalebar 1mm. Bloated ceramic clast. Vesicles are black where filled by the mounting resin, but pale and smoothappearing where unfilled. Quartz grains are midgrey in a paler glass. d. NTH3 SOI8-11. Scalebar 600µm. Complex contact between olivine-rich slag and partially melted sandstone clast. Vesicles are black where filled by the mounting resin, but pale and smooth-appearing where unfilled and white where filled by secondary iron oxides. The dark grey marginal zone with elongate crystals is dominantly feldspar..

Figure 6: Details of sample STH7. a) Montage of BSEM images to form a transect across the sample. b) Montage of BSEM images corresponding to area outlined in red in (a). This area covers the change from wustite replaced and overgrown by magnetite and iscorite towards the base, through magnetite replaced elongate wustite dendrites to euhedral equant magnetite towards the top above the crust of wustite, replaced and overgrown by magnetite. c) Explanation of (a) with zones and crusts labelled. Black areas are vesicles; green areas are early vesicles now filled by fayalitic slag.

Figure 3: EDS analyses from NTH3, placed in an approximate location with respect to the clast:slag interface. This demonstrates the compositional gradients present in the slag. Solid (upper) symbols are EDS analyses of representative analyses of typical areas. Open (lower) symbols indicate EDS analyses (points and areas) of the interstitial glass between the quartz grains of the clast. The patterns of enrichment of CaO and K2O in the interstitial glass of the clast suggests their enrichment from fuel ash before incorporation into the iron slag. Figure 4: BSEM Images NTH4, NTH5 a. NTH4 SOI3. Scalebar 1mm. Heterogeneous texture, with large olivine crystals (pale), with vesicles (dark) and weathered patches (mid grey), surrounding a vesicle with a discontinuous leucite rim (dark) and a fill of finer olivine in glass (dark grey). b. NTH4 SOI4. Scalebar 100µm. Detail of (a) showing leucite rim to vesicle (dark), overlain by an inward-growing cotectic of leucite and wustite-cored fayalite. The main pore filling shows wustite (bright), overgrown by olivine (pale grey) and minor hercynite (mid-grey) with interstitial glass (dark grey). c. NTH4 SOI5. Scalebar 1mm. Sheaves of olivine (pale) with interstitial glass (mid grey) d. NTH4 SOI6. Scalebar 1mm. Bloated ceramic clast. Vesicles are black where filled by the mounting resin, but pale and smoothappearing where unfilled. Contact with very finegrained slag to right. e. NTH5 SOI5. Scalebar 1mm. Heterogeneous texture with variable size and morphology of olivine (pale). Vesicles partially filled by weathering products (dark grey). f. NTH5 SOI6 Scalebar 1mm. Small vitrified ceramic inclusion with sharp margins in heterogeneous slag.

Figure 7: REE profiles, normalised to ceramic NTH1, using analyses recast on an iron-free basis. Figure 8. Plot of Al2O3 against SiO2 for normalised EDS data and XRD bulk analyses. Figure 9. Plot of total iron expressed as Fe2O3 against SiO2 for normalised EDS data and XRD bulk analyses. Figure 10. FeO-SiO2-Al2O3 ternary diagram. Fields simplified after Schairer & Yagi 1952. Figure 11. FeO-SiO2-feldspar pseudo-ternary diagram. Feldspar pole constructed from total feldspar based on contents of Na, Ca, and K, with available Al distributed between them. The fields shown are those of the FeO SiO2 - CaAl2Si2O8 pseudoternary (after Levin et al. 1964) and provide only a general guide to field locations; the liquidus fields would actually depend on the Na:K:Ca proportions. Figure 12: EDS analyses of glass formed by partial melting of ceramic clasts and associated materials. All values in wt%.

GeoArch Report 2013/27: Residues from Neath

Glossary End-member: The limits of solid solution in a mineral system. Used in cases where all the possible sites which a particular element can occupy in a crystal are occupied by atoms of the element. An example would be fayalite, Fe2SiO4 being the iron end-member of the olivine group, whereas forsterite, Mg2SiO4 is the magnesium endmember.

Aristotle process/furnace: a modern name for a technique for increasing the carbon content of iron to produce steel, by remelting it in a small enclosed furnace. The ‘Aristotle Furnace’ is a small furnace to produce a small ‘puck’ of steel, invented by Skip Williams, following accidental steel production in a hearth, inspired by a mention of remelting in Aristotle’s Meteorology Book 4 (supposed to be of the 4th century BC). The tuyère angle and bed depth can be adjusted to control the carbon content of the product (Sauder 2013).

Evenstad process/hearth: a modern name for a process involving remelting iron in an open hearth, with a strong horizontal air blast. The technique was described by Evenstad in 1790 in his account of traditional ironmaking in Norway (Jensen 1968; Wagner 1990). Evenstad’s account is not without problems, but seems to imply that iron smelted in a bloomery could be remelted twice with the hearth in different configurations. Modern commentators have generally assumed the first remelt was to control (reduce) the phosphorus content and the second to make carbon steel. Modern experimentalists and smiths use variants of the process to make ‘hearth steel’ (c.f. the Aristotle process) and as a means of recycling iron-rich debris from bloomery smelting.

Blacksmithing: the working of iron and steel. Usually restricted to the secondary smithing – i.e. the smithing of iron to produce or repair artefacts, as opposed to the smithing involved in the production of billets or bar iron from raw iron blooms. Bleb: a small rounded particle or textural component, often a droplet or prill. Bloom: The raw block of iron produced during smelting in the bloomery process. Blooms require further compaction and reworking (bloomsmithing) in order to remove voids, charcoal and most of the slag entrained in the raw bloom. The remaining slag forms elongate inclusions in the iron, which gives bloomery iron its toughness. Many of the objects described as blooms have undergone at least some compaction. Fully compacted blooms (but not yet worked down to usable bar iron) are termed billets.

Euhedral: a crystal shape in which the crystal has developed its faces, indicating its growth was unobstructed by previously formed phases. Fayalite: the iron-rich end member of the olivine group, Fe2SiO4. Forging: the process of hot-forming a metal by beating it.

Bloomery: a furnace for smelting iron from ore in which iron is produced as a solid material. The bloomery process was employed mainly prior to the introduction of the later blast furnace from the late 15th century.

Forsterite: the magnesium-rich end member of the olivine group, Mg2SiO4. Hercynite: an iron-aluminium member of the spinel group of minerals: FeAl2O4

Bloom-refining: Either (1) an alternative term for bloomsmithing, or (2) bloom processing techniques other than physical working (such as re-melting).

Interstitial space or interstice: the space between the main generation of crystals in a material. On cooling the main phase of crystals will solidify whilst the interstitial spaces are still occupied by molten material, which will solidify at a lower temperature.

Bloomsmithing: The process of reworking a raw bloom, through repeated reheating and hammering, to reduce its content of slag, to remove unwanted inclusions and to draw out the remaining slag into elongate inclusions. The end product may be a bar or billet.

Iscorite: the informal name for a phase (artificial mineral) to date only observed in metallurgical slags. The chemical composition is Fe2+5Fe3+2SiO10. The mineral is typically encountered in contexts where wustite is oxidised in the presence of fayalite.

Blowhole: A hole through a furnace or hearth wall through which air is blown. Burr: the zone of interaction between slag and the hearth or furnace wall just below the tuyère/blowhole. Typically seen as a dense lunate slag, with attached highly indurated ceramic, forming a protuberance on the margin of an SHC or FB.

Leucite: a potassium-bearing mineral KAlSi2O6 Magnetite: an iron oxide member of the spinel group, Fe3O4.

Cotectic: crystallisation of a liquid to produce two phases at the same time. Dendrite: a branched crystal form, often associated with rapid growth.

GeoArch Report 2013/27: Residues from Neath

Olivine: a group of silicate minerals of the form (M2+)2SiO4 where M can commonly be iron, magnesium, calcium (up to half the M2+ ions) or manganese. Includes the end-members fayalite (Fe2SiO4), forsterite (Mg2SiO4), tephroite (Mn2SiO4), kirschsteinite (CaFeSiO4) and monticellite (CaMgSiO4). The complex substitutions are described in this report by using the convention of describing the Fe-Mg as the fayalite-forsterite proportion (for instance Fa95Fo5, where Mg is 5% of the total of Mg+Fe) and then describing the Ca and Mn concentrations as percentage substitutions of the forsterite-fayalite. Prill: a small aggregate of a material, either a spheroidal droplet or a runnel, formed from a melted liquid and either occurring as a discrete particle or as an inclusion within another material. Quench: to cool rapidly from high temperature. Remelting: a group of techniques (including those now commonly known as the Aristotle and Evenstad techniques) in which iron is melted in a small hearth or furnace to adjust its chemical composition, or to recycle small fragments into a fresh bloom. Smithing: the activity involved in forming a metal object, including, but not limited to, forging metal and joining metal by welding. Smithing hearth cake (SHC): the slag cake formed in a smithing hearth by reaction of iron oxides lost from the workpiece with silicate material from hearth ceramic and/or fuel, below and in front of the tuyĂ¨re/blowhole. They are very variable in form, but are typically plano-convex in shape (although concavo-convex and biconvex forms are common), frequently with a dense lower crust and a less well consolidated upper part. In some cases the top is formed of glassy slag. Spinel: a mineral group with the general formula X2+Y3+2O4, which includes, amongst many others, the minerals hercynite and magnetite. Tap slag (or tapped slag, sometimes tapslag): slag that has been tapped from a furnace as a liquid to solidify outside the furnace. Tapping: the process or act of allowing a liquid to flow from a furnace. In the bloomery process it is the slag that may be tapped; in a blast furnace both slag and iron are tapped. Tongue: a slag mass that formed in a similar manner to a smithing hearth cake, but with a tabular form, typical silica rich and often glassy above, with pendent more iron-rich prills below. Valency: a measure of the number of bonds formed by an atom of a given element Vesicle: a void or pore, usually rounded and formed as a preserved gas bubble in a solidified melt. Wustite: an iron II oxide FeO.

GeoArch Report 2013/27: Residues from Neath

Table 1: Details of samples selected for further analysis. Sample

Context

Wt of Spec.

Specimen description

Subsample notes

NTH1

1502

22g

hearth lining

Bulk chemistry subsample: unvitrified section of wall

NTH2

1502

130g

maroon-surfaced gravelly tongue attached to vitrified lining

SEM subsample: dense slag with edge of lining material, Bulk chemistry subsample: bulk tongue

NTH3

1502

218g

curved gravelly sheet-like tongue with pale glass

SEM subsample: clast/slag boundary, Bulk chemistry subsample: bulk

NTH4

1502

106g

well-formed tongue (or incipient SHC) with dark glassy top, rusty base

SEM subsample: grey slag, Bulk chemistry subsample: bulk

NTH5

1502

270g

pale green glassy-topped SHC, rusty hemispherical base, minor explosion from iron corrosion, fuel dimples in top. Hollow.

SEM subsample: lower crust Bulk chemistry subsample: bulk

NTH6

1502

200g

part of pale dense SHC with possible basal tool marks

SEM subsample from lower part â&#x20AC;&#x201C; does not include the pale lining-influenced material on top, Bulk chemistry subsample: bulk

NTH7

1676

56g

dense angular slag fragment

Bulk chemistry subsample: bulk

GeoArch Report 2013/27: Residues from Neath

Table 2: Major element analyses of samples by XRF, expressed in oxide wt%. LOI = loss on ignition

NTH1 NTH2 NTH3 NTH4 NTH5 NTH6 NTH7

SiO2 wt%

Al2O3 wt%

Fe2O3 wt%

Mn3O4 wt%

MgO wt%

CaO wt%

Na2O wt%

K2O wt%

TiO2 wt%

P2O5 wt%

SO3 wt%

LOI wt%

Total wt%

66.60 61.40 42.61 38.90 16.18 29.99 7.68

9.13 6.56 5.84 5.21 2.07 3.75 0.97

16.97 25.30 45.96 51.50 79.01 61.69 88.94

0.25 0.14 0.22 0.21 0.09 0.15 0.20

0.72 0.62 0.47 0.49 0.15 0.45 0.54

1.34 1.42 1.11 1.02 0.05 0.90 1.37

0.78 0.83 0.55 0.49 0.05 0.34 0.09

2.43 2.88 1.90 1.85 0.18 1.28 0.37

0.77 0.50 0.50 0.46 0.24 0.33 0.10

0.262 0.295 0.320 0.445 0.164 0.297 0.280

0.041 0.040 0.067 0.062 0.123 0.112 0.039

0.44 -1.40 -2.84 -3.36 1.44 -4.11 -6.16

99.86 100.08 99.71 100.73 99.81 99.37 100.63

GeoArch Report 2013/27: Residues from Neath

Table 3:Trace elements analysed by ICP-MS, expressed in ppm.

NTH1 NTH2 NTH3 NTH4 NTH5 NTH6 NTH7

Sc ppm

V ppm

Cr ppm

Co ppm

Ni ppm

Cu ppm

Zn ppm

Ga ppm

Rb ppm

Sr ppm

Y ppm

Zr ppm

Nb ppm

Mo ppm

Sn ppm

Cs ppm

Ba ppm

7.8 5.0 6.2 5.9 2.8 5.0 2.0

81.2 76.1 79.8 75.4 39.8 70.6 39.6

47.8 541.2 46.2 34.8 97.5 1220.9 17.8

75.5 120.7 249.5 281.7 442.9 331.7 504.5

34.2 835.9 34.1 27.7 250.1 2708.5 43.6

265.7 270.5 416.2 362.4 518.6 522.6 589.9

101.1 220.4 280.8 83.6 106.5 281.2 131.4

11.1 5.7 6.7 8.3 6.7 7.6 4.7

54.7 50.2 35.8 41.8 5.4 27.4 4.3

131.1 128.9 128.7 121.2 23.0 103.0 53.1

19.9 13.9 16.9 14.7 5.6 10.5 4.6

313.2 261.7 292.6 234.1 139.1 211.0 95.6

10.37 7.53 7.39 6.46 2.21 5.77 1.70

3.19 4.02 6.05 2.62 36.20 5.46 9.61

3.31 4.93 5.04 1.46 1.14 2.68 0.85

2.48 1.79 1.46 1.55 0.29 1.10 0.15

324.9 273.4 215.2 250.0 57.4 248.9 118.4

La ppm

Ce ppm

Pr ppm

Nd ppm

Sm ppm

Eu ppm

Gd ppm

Tb ppm

Dy ppm

Ho ppm

Er ppm

Tm ppm

Yb ppm

Lu ppm

Hf ppm

Ta ppm

Pb ppm

Th ppm

U ppm

21.57 13.65 14.34 16.80 5.32 11.98 3.88

42.96 26.19 28.69 33.19 10.37 23.31 6.54

5.26 3.27 3.48 4.05 1.28 2.88 0.87

17.78 11.01 12.37 14.19 4.42 10.04 3.17

3.51 2.01 2.23 2.69 0.81 1.91 0.71

0.82 0.50 0.61 0.66 0.19 0.50 0.19

2.77 1.43 1.74 2.23 0.66 1.51 0.49

0.50 0.32 0.35 0.40 0.15 0.29 0.12

3.30 2.04 2.29 2.68 0.88 1.88 0.78

0.59 0.38 0.42 0.50 0.17 0.34 0.12

1.75 1.03 1.19 1.48 0.51 1.00 0.40

0.30 0.19 0.23 0.25 0.10 0.18 0.07

1.97 1.17 1.41 1.67 0.66 1.14 0.39

0.29 0.18 0.20 0.25 0.09 0.16 0.06

7.08 5.11 5.28 5.94 3.10 4.73 2.05

2.28 1.49 1.62 1.75 0.44 0.89 0.68

13.84 20.29 337.47 17.36 3.75 30.74 58.22

6.49 4.40 4.03 5.04 2.17 3.18 1.01

2.11 1.30 1.56 1.71 0.66 1.19 2.58

GeoArch Report 2013/27: Residues from Neath

Table 4. Major element analyses (by XRF) normalised on an iron-free basis.

NTH1 NTH2 NTH3 NTH4 NTH5 NTH6 NTH7

SiO2

Al2O3

Mn3O4

MgO

CaO

Na2O

K2O

TiO2

P2O5

SO3

wt%

80.9 82.2 79.5 79.2 83.8 79.8 66.0

11.1 8.8 10.9 10.6 10.7 10.0 8.3

0.3 0.2 0.4 0.4 0.5 0.4 1.7

0.9 0.8 0.9 1.0 0.8 1.2 4.6

1.6 1.9 2.1 2.1 0.3 2.4 11.8

0.9 1.1 1.0 1.0 0.3 0.9 0.8

3.0 3.9 3.5 3.8 0.9 3.4 3.2

0.9 0.7 0.9 0.9 1.2 0.9 0.9

0.3 0.4 0.6 0.9 0.8 0.8 2.4

0.0 0.1 0.1 0.1 0.6 0.3 0.3

GeoArch Report 2013/27: Residues from Neath

Table 5: Hearth ceramic and entrained clast compositions. Analyses expressed in wt% oxide and normalised to 100%, with all iron expressed as Fe2O3. < = below detection Sample

Area

NTH1

bulk

NTH2 NTH2 NTH2 NTH4 NTH5

SOI6 SOI8 SOI9 SOI6 SOI6

1 1 1 1 1

NTH3 NTH3

SOI4 SOI4

1 2

Type

Nature

Na2O

MgO

Al2O3

SiO2

P2O5

K2O

CaO

TiO2

Mn3O4

Fe2O3

XRF

bulk analysis of NTH1

0.79

0.73

9.19

67.10

0.26

2.45

1.35

0.77

0.25

17.10

EDS area EDS area EDS area EDS area EDS area

bloated ceramic bloated ceramic bloated ceramic bloated ceramic bloated ceramic

0.49 0.85 0.68 0.90 0.84

1.07 0.88 0.67 0.58 0.54

12.90 11.23 10.24 12.40 10.89

74.88 78.66 80.87 71.88 74.46

0.27 < < 0.86 0.00

3.14 2.37 2.34 1.49 3.81

0.32 < < 0.00 0.52

0.80 0.78 0.62 0.83 0.68

< < < < <

6.13 5.23 4.59 11.07 8.27

EDS area EDS area

bloated/vitrified sandstone bloated/vitrified sandstone

0.55 0.54

0.30 0.20

9.93 10.96

81.08 81.87

< 0.26

0.96 0.95

0.17 0.00

0.43 0.39

< <

6.59 4.82

GeoArch Report 2013/27: Residues from Neath

Table 6: Composition of interstitial glass in entrained clasts. Analyses by EDS, expressed in wt% oxide and normalised to 100%, with all iron expressed as Fe2O3. < = below detection. Sample

SOI

Type

Na2O

MgO

Al2O3

SiO2

P2O5

K2O

CaO

TiO2

Mn3O4

Fe2O3

NTH2 NTH2 NTH2 NTH2 NTH2

SOI6 SOI6 SOI8 SOI8 SOI9

2 3 2 3 2

area area area area area

glass with mullite? glass with Fe-Ti oxides? ‘bright’ glass ‘medium’ glass ‘bright’ glass

0.71 0.69 1.14 1.17 1.15

1.59 1.79 2.58 1.54 1.25

20.34 19.65 12.63 20.23 8.74

61.95 58.99 61.90 66.36 58.37

0.31 0.40 1.00 0.30 0.75

4.03 3.59 3.06 3.01 3.60

0.57 0.53 1.21 0.00 3.69

1.16 1.21 0.91 1.08 0.60

< < < < 0.25

9.35 13.15 15.57 6.30 21.58

NTH3 NTH3 NTH3 NTH3 NTH3 NTH3

SOI6 SOI6 SOI11 SOI11 SOI11 SOI11

5 7 1 4 5 6

area point area area area area

glass glass glass glass glass glass

1.98 1.74 1.70 1.72 1.54 1.37

1.16 0.59 1.18 0.36 0.76 0.64

17.13 15.22 16.76 14.18 12.77 14.03

55.61 54.27 55.23 53.47 57.90 55.67

0.45 0.00 0.41 0.38 0.41 0.48

6.25 5.07 5.78 4.69 5.34 5.51

1.39 1.21 2.16 1.56 2.00 1.76

1.03 0.78 1.08 0.71 0.72 0.72

< 0.20 0.30 < 0.17 <

15.01 20.91 15.38 22.92 18.39 19.82

NTH4 NTH4

SOI6 SOI6

3 5

point point

glass near margin of clast glass

0.88 0.83

1.10 1.27

16.51 13.88

65.34 70.60

0.69 0.68

2.88 2.82

< 0.25

1.34 1.62

< <

11.25 8.04

Figure 1

NTH1 NTH2 NTH3 NTH4 NTH5 NTH6 NTH7

0.1 La

Figure 2 a

Figure 3

wt% 100

clast slag

}SiO2 10

}K2O }Al2O3 }Fe2O3

}CaO

0 -1500

-1000 -500 0 500 distance from clast-slag interface (mm)

1000

Figure 4 a

Figure 5 a

(a)

(b)

(c)

Figure 6

1mm

Crust 5

1mm

Zone 13: equant euhedral magnetite (c.30%; up to 150mm), followed by elongate olivine (to 800mm) Crust 4 Crust 3 Zone 12: equant euhedral magnetite (c.30%; up to 150mm), followed by elongate olivine (to 800mm) Zone 11: dendritic magnetite (c. 50%; dendrites to 800mm), with some iscorite, followed by elongate olivine (to 800mm) Zone 10: equant euhedral magnetite (c.30%; up to 150mm), with some iscorite, followed by elongate olivine (to 800mm) Crust 2 Zone 9: dendritic wustite, overgrown by iscorite and altered to magnetite (c.25%; dendrites to 1mm), followed by elongate olivine (up to 600mm) Zone 8: blebby dendritic wustite, heavily altered to magnetite (70%), followed by elongate olivine (up to 600mm) Zone 7 blebby dendritic wustite, heavily altered to magnetite and overgrown by abundant iscorite (80% total), followed by elongate olivine (up to 600mm) Zone 6: blebby dendritic wustite, partially altered altered to magnetite, rare iscorite, followed by elongate olivine Zone 5: badly corroded zone, probably originally equant magnetite (<20%) followed by elongate olivine

Zone 4: coarse dendritic wustite (60-80%; dendrites to 2mm), coarsest and densest in centre, fines towards edges, followed by olivine, best seen in occluded early vesicles

Zone 3:coarse dendritic wustite (80%; dendrites to 1mm)

Zone 2: coarse dendritic wustite (90%), well formed dendrites to 2mm in lower part, more blebby in upper

Crust 1

Zone 1: fine dendritic wustite (90%)

Figure 7

NTH2 NTH3 NTH4 NTH5 NTH6 NTH7

0.1 La

Figure 8

Al2O3 (wt%)

0 0

100

NTH3 slag w/clots NTH3 clots NTH3 dark fringe NTH2 clast glass NTH3 clast glass NTH4 clast glass NTH7 bulk NTH6 bulk NTH5 bulk NTH4 bulk NTH3 bulk NTH2 bulk NTH1 bulk NTH5 ceramic NTH4 ceramic NTH4 ceramic glassy NTH3 ceramic NTH2 sandstone NTH7 slag NTH6 slag NTH5 slag NTH4 slag NTH3 slag NTH2 slag

SiO2 (wt%)

Figure 9

Fe2O3 (wt%) 100

NTH3 slag w/clots NTH3 clots

NTH3 dark fringe NTH2 clast glass NTH3 clast glass

NTH4 clast glass NTH7 bulk 70

NTH6 bulk NTH5 bulk

NTH4 bulk NTH3 bulk NTH2 bulk

NTH1 bulk NTH5 ceramic 40

NTH4 ceramic NTH4 ceramic glassy

NTH3 ceramic NTH2 clastic NTH7 slag

NTH6 slag NTH5 slag 10

NTH4 slag NTH3 slag

NTH2 slag 0

100

SiO2 (wt%)

Figure 10 NTH2 slag NTH2 ceramic NTH2 glass in ceramic NTH2 bulk NTH1 bulk NTH3 bulk NTH3 slag NTH3 ceramic NTH3 glass in ceramic NTH3 feldspar fringe NTH3 clots NTH3 slag with clots NTH4 slag NTH4 bulk NTH4 ceramic NTH4 glass in ceramic NTH5 bulk NTH5 slag NTH5 ceramic NTH6 bulk NTH6 slag NTH7 bulk

0.8

0.2

0.4

Fe O

te ali fay

0.6

0 130 0 0 2 1 1150 00 12 00 13 50 11 0 te 140

2 SiO

y trid

0.6

mullite

0.4

corundum

n rcy

0.8

0.2 00

sti

0.2

0.4

0.6

Al2O3

0.8

Figure 11

0.8

liqu

ids

0.2 two

NTH2 slag NTH2 ceramic NTH2 glass in ceramic NTH2 bulk NTH1 bulk NTH3 bulk NTH3 slag NTH3 ceramic NTH3 glass in ceramic NTH3 feldspar fringe NTH3 clots NTH3 slag with clots NTH4 slag NTH4 bulk NTH4 ceramic NTH4 odd ceramic NTH4 glass in ceramic NTH5 bulk NTH5 slag NTH5 ceramic NTH6 bulk NTH6 slag NTH7 bulk NTH7 slag

cristobalite

0.4

0.6

Fe O

SiO

0.6

0.4

tridymite

fayalite

0.8

0.2 anorthite hercynite

wustite

0.2

0.4

0.6

0.8

KAlSi3O8 + NaAlSi3O8 + CaAl2Si2O8

Figure 12 2.50

4.00

CaO

Na2O

3.50

2.00

3.00 2.50

1.50

2.00 1.00

1.50 1.00

0.50 0.50 0.00

0.00 0.00

2.00

4.00

6.00

K2O

8.00

0.00

7.00

4.00

K2O

CaO

2.00

4.00

6.00

K2O 8.00

3.50

6.00

3.00

5.00

2.50 4.00 2.00 3.00 1.50 2.00

1.00

0.50

Al2O3

0.00 0.00

5.00

10.00

15.00

20.00

25.00

100.00

Al2O3

0.00 0.00

5.00

10.00

15.00

20.00

25.00

2.50

SiO2

Na2O

80.00

2.00

60.00

1.50

40.00

1.00

20.00

0.50

Al2O3

0.00 0.00

5.00

10.00

15.00

20.00

25.00

Al2O3

0.00 0.00

5.00

10.00

Bulk areas of ceramic clasts (NTH2, NTH4, NTH5) Bulk areas of sandstone clasts (NTH3) NTH2 i/s glass NTH3 i/s glass NTH4 i/s glass NTH3 marginal feldspar-rich zone

15.00

20.00

25.00

GeoArch Report 2013/27: Residues from Neath

APPENDIX A: Other bulk analyses

GeoArch Report 2013/27: Residues from Neath

Table A1: Trace element analyses of samples by XRF, expressed in oxide wt%.

NTH1 NTH2 NTH3 NTH4 NTH5 NTH6 NTH7

V2O5 wt%

Cr2O3 wt%

SrO wt%

ZrO2 wt%

BaO wt%

NiO wt%

CuO wt%

ZnO wt%

PbO wt%

HfO2 wt%

0.011 0.003 0.006 0.008 <0.003 0.004 <0.003

0.005 <0.004 0.006 <0.004 <0.004 <0.004 <0.004

0.005 0.011 0.010 0.013 0.007 0.019 0.012

0.065 0.055 0.052 0.051 0.034 0.039 0.027

0.021 0.014 0.008 <0.006 <0.006 0.011 <0.006

<0.002 <0.002 <0.003 <0.003 <0.004 <0.003 <0.004

<0.002 <0.002 <0.002 <0.002 0.003 <0.002 <0.003

0.016 0.009 0.010 0.010 0.009 0.012 0.007

0.003 0.007 0.063 0.005 0.009 0.010 0.013

0.005 <0.004 0.004 0.006 0.010 <0.005 <0.005

GeoArch Report 2013/27: Residues from Neath

Table A2: Major element analyses of samples by ICP-MS, expressed in oxide wt%.

NTH1 NTH2 NTH3 NTH4 NTH5 NTH6 NTH7

SiO2 wt%

MnO wt%

Fe2O3 wt%

TiO2 wt%

P2O5 wt%

44.38 46.73 42.48 30.12 9.65 21.14 2.48

0.17 0.10 0.19 0.18 0.06 0.12 0.18

15.16 24.71 49.80 56.44 87.34 68.19 96.04

0.64 0.44 0.52 0.46 0.24 0.35 0.11

0.28 0.33 0.36 0.53 0.19 0.36 0.33

GeoArch Report 2013/27: Residues from Neath

APPENDIX B: EDS Microanalyses

GeoArch Report 2013/27: Residues from Neath

Table B1: EDS microanalyses expressed in elemental wt%. < = below detection Sample

SOI

Area/ point

31.50

0.66

0.28

3.05

15.76

0.23

1.38

1.46

0.17

NTH2

area

typical

NTH2

point

ol mid

29.86

0.78

13.78

0.17

53.15

97.75

NTH2

point

ol inner

29.89

0.91

0.11

13.60

0.18

53.03

97.72

NTH2

point

ol inner

29.41

0.78

13.49

NTH2

point

ol mid

29.75

0.79

0.14

13.58

NTH2

point

ol outer

29.06

0.48

0.19

13.35

NTH2

point

late ol mix

29.46

0.19

0.58

13.75

NTH2

point

late ol mix

29.95

0.28

0.43

13.90

NTH2

point

late ol mix

32.52

0.40

0.20

2.45

16.12

0.17

NTH2

point

glass

38.41

1.15

8.31

20.43

0.48

NTH2

point

ol margin

31.18

0.42

0.34

1.63

15.73

0.11

NTH2

area

typical

36.02

0.74

0.41

3.58

20.43

0.22

NTH2

area

typical

34.11

0.69

0.34

3.42

18.04

0.19

NTH2

area

typical

33.34

0.75

0.40

3.54

18.21

0.21

NTH2

area

typical

47.62

0.27

3.02

39.36

NTH2

area

typical

30.12

0.61

0.34

3.18

16.23

NTH2

area

typical

30.31

0.71

0.28

3.28

NTH2

area

typical

33.68

0.64

0.29

3.70

NTH2

point

bright grain in clast

22.97

0.15

NTH2

area

typical

43.46

0.33

0.59

6.20

NTH2

area

glass with mullite?

42.44

0.49

0.90

10.05

NTH2

area

glass with Fe-Ti?

43.20

0.51

1.07

10.25

27.19

NTH2

area

typical

35.87

0.43

0.37

4.09

25.34

NTH2

area

bright glass

34.01

0.63

1.17

5.01

NTH2

area

medium glass

44.33

0.81

0.87

9.99

0.13

38.54

0.10

0.09

Total

93.02

0.18

53.26

97.25

0.21

53.33

97.81

53.15

96.48

0.24 0.17

0.29

0.42

51.07

95.93

0.33

0.43

52.05

97.36

0.72

1.06

47.80

101.43

2.87

5.16

0.69

18.11

95.70

0.80

0.58

48.04

98.84

1.86

1.76

0.26

34.63

10<

1.61

1.70

0.16

37.21

97.49 91.68

1.70

1.57

0.26

31.68

1.56

0.15

0.19

2.24

94.68

0.19

0.07

1.53

1.43

0.18

28.59

82.47

16.84

0.15

18.54

0.23

0.63

1.39

0.23

31.80

0.11

2.37

0.21

0.44

3.90

89.38

27.05

0.13

3.12

0.38

0.65

6.11

91.32

0.17

2.94

0.37

0.72

9.07

95.48

0.32

2.52

70.30

21.68

0.33

1.90

0.65

0.41

8.16

73.95

28.96

0.12

0.60

4.11

92.13

1.61

1.42

0.20

1.89

1.52

0.22

0.10

0.40

1.35 <

2.33

< 2.08

30.46

85.27

30.96

91.66

50.62

78.56

GeoArch Report 2013/27: Residues from Neath

Sample

SOI

Area/ point

NTH2

area

typical

27.94

0.27

0.21

2.89

20.14

NTH2

area

bright glass

46.70

0.96

0.85

5.17

NTH3

area

typical

35.88

0.53

0.33

3.68

NTH3

area

dendritic clot

38.84

0.56

0.16

4.05

21.05

0.17

2.18

NTH3

point

dendritic clot

39.18

0.68

0.13

4.14

21.24

0.20

2.40

NTH3

point

elongate dendrites

36.65

0.72

0.40

3.82

19.92

0.17

1.68

1.26

0.21

NTH3

area

elongate dendrites

36.89

0.73

0.39

3.97

20.19

0.19

1.85

1.33

NTH3

point

i/s

37.80

0.57

0.44

3.68

20.15

0.21

1.53

1.45

NTH3

area

typical

36.75

0.68

0.31

3.62

19.67

1.51

1.08

0.28

NTH3

area

c400um from clast

34.12

0.56

0.23

3.39

18.16

0.16

1.41

1.06

0.22

NTH3

area

c300um from clast

34.75

0.58

0.29

3.36

18.47

0.19

1.41

1.03

0.21

0.14

Total

1.04

0.20

1.71

54.39

30.50

0.37

3.34

2.95

0.40

0.20

16.87

108.30

19.05

0.19

1.61

1.26

0.27

0.30

37.19

100.30

1.19

0.43

0.16

31.65

100.45

1.24

0.37

30.23

99.81

0.19

35.77

100.77

0.27

0.19

34.06

100.06

0.22

0.20

36.79

103.01

0.17

37.06

101.14

0.17

35.32

94.81

35.40

95.84

NTH3

area

c200um from clast

35.01

0.66

0.22

3.43

19.50

0.09

1.58

1.00

0.26

0.24

33.45

95.47

NTH3

area

c100um from clast

36.65

0.64

0.25

3.58

20.55

0.12

1.90

0.97

0.31

0.13

30.61

95.71

NTH3

area

c500um inside clast

26.17

0.20

0.09

2.59

18.72

0.39

0.06

0.13

2.28

50.63

NTH3

area

c1000um inside clast

40.83

0.31

0.09

4.57

30.16

0.09

0.62

0.19

2.66

79.53

NTH3

area

typical

30.44

0.74

0.19

3.38

15.08

0.16

1.37

0.97

0.21

34.79

87.33

NTH3

area

typical fine dendrites

34.28

0.71

0.31

3.80

18.75

0.19

1.66

0.99

0.29

0.28

33.40

94.65

NTH3

area

fine with dendritic clots

38.65

0.82

0.39

4.19

21.44

0.20

1.83

1.08

0.33

0.22

32.04

101.20

NTH3

area

inner bright dendrites

42.74

0.88

0.34

5.29

23.89

0.18

2.55

1.18

0.45

30.63

108.14

NTH3

area

dark dendrites

46.25

1.07

7.25

27.69

0.17

4.08

1.64

0.64

11.61

100.38

NTH3

area

glass in clast

44.40

1.49

0.71

9.21

26.42

0.20

5.28

1.01

0.63

10.67

100.02

NTH3

point

bright bleb

36.98

0.11

16.05

NTH3

point

glass in clast

43.89

1.35

0.37

8.42

26.52

4.40

0.90

0.49

0.15

15.29

101.79

NTH3

point

dark dendrites

36.08

0.83

6.91

23.84

0.24

4.01

1.07

0.73

14.14

87.83

NTH3

point

dark dendrites

42.45

0.85

7.08

25.18

0.18

3.45

1.27

0.61

13.18

94.25

NTH3

point

dark dendrites

46.72

0.79

7.35

28.11

0.21

3.41

2.00

0.69

12.03

101.42

NTH3

area

250um from clast

32.98

0.66

0.29

3.56

17.64

0.16

0.10

1.51

1.11

0.28

0.19

32.65

91.14

NTH3

area

150um from clast

33.85

0.78

0.32

3.73

18.35

0.20

1.77

1.05

0.25

32.44

92.74

0.90

0.11

53.72

107.76

GeoArch Report 2013/27: Residues from Neath

Sample

SOI

Area/ point

NTH3

area

NTH3

50µm from clast

35.40

0.74

0.44

4.17

0.14

2.31

1.10

point

dark dendrite

34.75

0.60

0.09

6.35

24.29

0.19

point

dark dendrite

35.26

0.71

6.04

23.67

0.15

0.08

0.20

3.15

0.12

3.40

point

bright in inner eutectic

40.51

0.73

0.29

4.47

22.75

0.11

2.44

0.87

0.30

point

dark in inner eutectic

42.92

1.11

6.90

26.85

0.15

0.11

3.52

1.88

0.54

point

in separate eutectic bleb

43.13

0.69

4.86

23.61

0.12

2.53

0.78

0.38

0.27

5.78

25.51

0.10

3.48

1.04

0.35

6.15

23.84

0.19

3.26

1.17

0.46

0.14

0.12

3.13

1.71

0.40

0.07

0.16

2.81

0.94

0.34

0.14

3.21

1.18

0.45

0.23

29.58

1.34

0.44

1.20

0.41

Total

94.10

8.87

80.26

10.32

81.35

0.20

32.84

105.52

13.31

97.30

28.08

104.18

0.21

21.18

100.83

10.53

83.78

13.39

95.37

0.17

16.33

60.04

11.67

81.78

NTH3

area

inner bright eutectic

41.71

1.22

NTH3

area

in outer dark dendrites

37.36

0.83

NTH3

point

glass? Between outer dark dendrites

42.73

0.82

0.13

6.89

25.92

NTH3

point

hole?

17.80

0.41

0.16

4.01

16.82

NTH3

point

dark dendrite

34.94

0.67

6.34

22.95

NTH3

area

typical

35.02

0.64

0.29

3.52

18.21

0.21

1.33

1.14

0.23

36.64

97.22

NTH3

area

100µm from clast

36.93

0.75

0.25

3.88

19.51

0.17

1.73

1.12

0.34

36.66

101.34

0.24

0.21

NTH3

area

300µm from clast

34.17

0.52

0.22

3.49

17.21

0.15

1.35

1.07

0.20

NTH3

area

100µm from clast

41.43

0.61

0.32

4.21

21.94

0.18

1.65

1.16

0.31

0.40

0.25

3.45

17.12

0.18

1.49

1.05

0.21

98.20 111.20

NTH3

area

500µm from clast

32.88

35.68

92.84

NTH3

point

vesicle fill

38.73

7.74

0.13

52.41

99.01

NTH3

point

vesicle fill

36.72

7.34

0.15

51.21

95.41

NTH3

area

glass in clast

42.62

1.25

0.70

8.74

25.43

0.18

NTH3

area

outer bright

48.31

0.76

6.00

23.53

0.38

NTH3

area

outer medium

43.38

0.97

0.43

5.17

24.59

0.22

2.31

1.32

0.43

NTH3

area

glass in clast

43.63

1.33

0.23

7.80

25.98

0.17

4.05

1.16

0.44

NTH3

area

glass in clast

46.90

1.23

0.49

7.26

29.09

0.19

4.76

1.54

NTH3

area

glass in clast

43.30

1.00

0.38

7.34

25.75

0.21

4.53

1.25

NTH4

area

inner dendritic

29.76

0.61

0.16

2.94

10.85

0.39

1.03

0.89

0.12

47.36

94.12

NTH4

area

inner dendritic with W

34.61

0.62

0.28

3.56

14.65

0.33

1.52

0.85

0.18

47.19

103.79

NTH4

area

outer dendritic

25.73

0.51

0.11

2.79

10.24

0.33

1.11

0.85

0.10

34.26

76.04

NTH4

area

typical mixed

33.97

0.61

0.15

3.24

13.21

0.29

1.24

0.96

0.22

43.88

97.77

0.09

4.73

1.52

0.64

2.58

1.46

0.48

0.15

39.58 39.38

0.21

10.60

96.62

22.86

106.46

0.26

32.71

16.66

101.46

0.46

0.13

13.82

105.87

0.43

13.71

97.90

111.77

GeoArch Report 2013/27: Residues from Neath

Sample

SOI

Area/ point

NTH4

point

main ol outer

NTH4

point

main ol outer

28.54

0.32

0.24

12.30

0.17

29.34

0.28

0.18

12.52

0.17

NTH4

point

bright bleb in pore lining

23.78

NTH4

point

outer ol in pore lining

30.51

NTH4

point

late olivine core

31.29

0.30

0.19

13.22

0.17

53.85

99.19

NTH4

point

late olivine outer

30.05

0.20

0.27

12.70

0.17

0.34

53.28

97.02

NTH4

point

o/g on w dendrite in pore

31.94

4.85

7.31

NTH4

point

bright in pore lining

23.90

0.94

0.33

NTH4

point

late on W in pore

31.29

1.26

15.61

4.39

NTH4

point

granular on late ol

31.73

19.60

0.27

NTH4

point

medium in pore lining

32.18

16.73

0.60

NTH4

point

bright bleb in main ol

22.15

0.33

0.41

NTH4

point

glass in pore

39.42

3.38

7.64

17.91

1.57

0.31

2.12

NTH4

point

glass in pore

40.57

3.79

8.40

18.42

1.70

0.27

2.72

NTH4

point

granular on late ol

34.98

17.92

1.65

0.22

NTH4

point

o/g on w dendrite in pore

31.35

6.45

7.56

NTH4

point

Intergrown with leucite

31.14

15.48

0.36

NTH4

area

typical

31.02

0.59

0.19

3.45

13.95

0.25

NTH4

area

typical clast

35.00

0.44

0.23

4.31

22.10

0.25

0.81

NTH4

area

adjacent slag

25.25

0.50

0.17

3.07

13.88

0.09

0.88

0.18

1.15

0.24

0.34

12.50

0.18 1.00

0.26

0.09

0.52

0.59 0.32

0.20

1.40

Total

50.84

92.59

51.95

94.44

71.33

97.49

53.02

97.43

2.31

52.46

99.46

73.04

98.64

1.86

1.48

39.36

95.92

0.17

1.83

40.91

94.50

0.15

2.95

44.99

97.60

0.41

70.95

94.25

7.73

0.22

16.45

96.75

7.21

0.28

15.14

98.51

0.79

2.25

41.67

99.48

0.65

1.38

51.51

99.11

2.45

45.95

95.52

41.38

93.36

0.13 <

0.31

0.12 0.36

0.86

0.27 0.33

5.09

68.56

0.55

0.25

18.89

63.52

1.86

NTH4

point

glass near margin of clast

36.76

0.51

0.52

6.80

23.76

0.24

0.63

6.12

77.18

NTH4

point

marginal v fine zone

29.99

0.63

0.22

4.06

17.65

0.09

1.37

0.61

0.28

17.48

72.38

NTH4

point

glass in clast

48.31

0.61

0.76

7.30

32.79

0.30

2.33

0.18

0.97

5.59

99.13

NTH4

area

matrix to rusty part

29.04

45.34

75.44

NTH4

area

dense clast in rusty part

33.58

0.21

NTH4

area

glassy clast

38.80

0.39

NTH4

area

glassy clast

50.53

0.85

NTH4

area

rusty pore fill

31.71

NTH4

area

adjacent slag

26.80

NTH5

area

typical

31.87

1.06 0.12

1.09

0.37

0.20

2.32

28.42

0.19

0.70

3.94

33.02

0.27

1.35

16.52

0.20

0.55

0.26

3.17

14.16

0.23

0.40

0.18

2.30

12.56

0.21

0.09

0.20

51.15

86.73

2.13

0.91

0.19

6.85

80.40

4.12

2.72

0.38

11.45

107.98

0.45

1.37

0.84

0.25

0.62

0.38

0.10

0.21

24.72

75.24

25.91

73.75

50.80

99.22

GeoArch Report 2013/27: Residues from Neath

Sample

SOI

Area/ point

NTH5

area

NTH5

point

NTH5

point

NTH5

NTH5 NTH5

finer ol between v coarse

35.36

0.76

4.97

13.39

ol inner

28.49

0.46

0.17

12.12

ol outer

28.78

0.45

0.13

12.48

point

ol margin

28.53

0.38

0.16

12.35

point

ol core

28.48

0.40

0.12

12.25

point

ol inner

30.12

0.58

0.16

point

ol outer

30.71

0.54

0.21

NTH5

point

ol margin

31.23

0.53

NTH5

point

late ol

35.33

0.15

NTH5

point

late ol

26.21

NTH5

area

typical

35.31

0.20

NTH5

point

bleb inside marginal coarse ol

NTH5

area

typical

28.63

NTH5

area

typical

29.36

0.61

NTH5

area

coarse. Low W

30.91

0.67

NTH5

area

elongate with W

32.35

NTH5

area

ceramic clast

37.13

NTH5

point

bleb

8.80

NTH6

area

typical

34.35

0.58

NTH6

area

typical

31.26

NTH6

area

typical

NTH6

area

NTH6

area

NTH6

area

0.32

0.46

1.35

0.82

0.32

Total

50.88

107.84

49.00

90.24

49.87

91.70 90.57

0.09

50.34

91.68

13.00

0.11

52.12

96.09

13.32

0.10

53.63

98.52

13.54

0.15

55.29

100.74

0.31

15.22

0.15

61.37

112.52

0.42

12.90

0.22

0.12

58.03

97.89

2.26

13.82

0.35

0.21

55.04

108.05

101.27

0.54

2.54

11.18

0.32

48.56

91.67

2.76

11.79

0.76

0.42

0.26

46.02

91.99

0.94

0.57

0.22

46.47

95.28

0.22

0.17

55.02

101.65

0.27

0.30

3.50

12.01

1.56

12.04

0.14

4.29

25.90

0.21

3.70

0.21

3.46

15.27

0.13

0.09

1.59

0.98

0.21

0.38

0.27

2.02

12.30

0.09

0.95

0.63

32.65

0.71

0.26

3.23

14.21

0.15

1.45

0.96

0.17

typical

37.58

0.87

2.26

1.63

0.32

weathering

23.09

typical

33.06

0.66

49.15

0.27 0.24

4.39

18.29

0.22

0.09

1.53

4.17

0.30

0.24

3.28

15.95

2.35

0.67

0.36 1.60

1.05

0.28

0.16

4.30

75.25

108.25

120.95

39.35

96.22

45.40

93.46

40.93

94.72

40.56

106.21

37.66

67.78

39.17

95.29

GeoArch Report 2013/27: Residues from Neath

Sample

SOI

Area/ point

NTH6

point

large ol core

33.41

0.52

0.12

14.82

0.14

NTH6

point

large ol inner

32.44

0.60

0.15

14.44

0.10

NTH6

point

large ol outer

32.13

0.73

0.11

14.14

NTH6

point

large ol core

27.74

0.56

12.28

0.13

NTH6

point

large ol inner

27.70

0.50

0.11

12.13

NTH6

point

large ol o/g

29.11

0.28

0.57

13.14

0.15

46.81

90.94

NTH6

point

late ol

26.87

0.47

11.84

0.16

45.83

85.17

NTH6

point

late ol

33.06

0.49

0.15

14.48

0.13

54.33

102.63

NTH6

point

leucite in cotectic

44.75

11.90

24.70

1.01

98.30

NTH6

point

ol in cotectic

33.26

0.33

12.91

NTH6

point

ol in cotectic

31.15

0.18

13.37

NTH6

point

large ol outer

29.27

0.19

0.12

NTH6

point

large ol inner

29.86

0.51

0.18

NTH6

point

large ol outer

29.48

0.16

NTH6

point

large ol outer

30.05

0.12

NTH6

point

large ol inner

30.17

0.54

NTH6

point

large ol outer

30.48

0.32

12.97

NTH6

point

late ol

31.87

0.37

13.51

NTH6

point

glass

41.34

10.68

20.71

0.40

NTH7

area

euhedral magnetite-rich lobe 1

30.62

0.29

0.70

5.58

0.51

NTH7

area

dendritic magnetite-rich lobe 2

40.03

0.46

0.37

1.07

9.62

NTH7

area

magnetite/iscorite rich lobe 2

31.98

0.51

0.22

0.65

7.11

NTH7

area

wustite/iscorite lobe 3

30.54

0.55

0.39

1.01

9.50

NTH7

area

wustite/iscorite lobe 3

21.63

0.31

0.20

0.67

5.30

0.76

0.86

41.09

70.81

NTH7

area

17.44

0.19

0.20

0.57

3.40

0.35

0.44

34.55

57.16

NTH7

area

31.62

0.37

0.46

0.94

7.06

0.91

1.08

63.59

106.02

32.55

0.30

0.42

0.86

6.26

0.65

1.12

69.68

111.84

26.52

0.35

0.23

0.77

5.72

0.70

0.89

54.33

89.52

NTH7

area

coarse magnetite-iscorite layer in lobe 3 coarse magnetite-iscorite layer in lobe 4 wustite-rich upper part of lobe 3

NTH7

area

overall

0.69

2.57

0.41

0.37

56.48 0.14 0.13 0.10

15.25 0.57

Total

105.49

54.57

102.44

53.47

100.58

47.18

88.02

47.09

87.54

95.97

0.56

51.32

96.58

12.69

0.26

49.30

91.84

12.60

0.11

48.97

92.23

12.37

0.30

49.44

91.74

0.17

12.85

0.28

50.80

94.27

0.10

12.96

0.08

50.47

94.33

0.13

0.12

0.29

0.14

49.69

94.01

0.10

0.50

0.14

51.80

98.28

6.05

3.50

0.51

12.97

98.72

0.52

58.18

96.51

0.87

1.10

72.49

126.02

0.81

0.94

60.67

102.88

0.10

1.42

1.62

51.33

96.47

0.14

48.52

0.24

GeoArch Report 2013/27: Residues from Neath

Sample

SOI

Area/ point

NTH7

area

NTH7

NTH7 NTH7

wustite-rich

24.95

0.32

0.36

0.72

5.41

area

wustite-rich

25.20

0.31

0.66

4.31

ares

dense wustite

19.03

0.32

0.42

2.16

area

wustite-rich

12.67

0.15

0.29

0.38

2.03

0.09

area

wustite dense

28.56

0.93

0.60

2.41

0.11

NTH7

area

silicate rich vesicle fill

39.35

1.15

1.06

2.36

15.74

0.74

0.11

NTH7

point

weathering crust

34.04

0.54

0.98

15.63

0.40

NTH7

point

ol outer

35.40

1.85

13.84

0.56

NTH7

point

weathering

35.27

0.94

0.96

18.38

0.21

NTH7

point

glass

NTH7

point

NTH7

0.09

0.46

1.21

0.40

0.81

0.16

50.66

84.33

1.45

54.79

87.12

0.24

1.02

46.49

69.68

0.27

1.34

27.72

44.93

0.35

1.42

0.36

72.44

107.18

3.32

11.36

0.35

35.08

110.63

25.56

79.77

27.26

97.60

0.27

1.54

0.27 0.44

0.46

0.33

1.46

22.58

8.46

4.88

17.00

4.67

9.14

2.71

0.18

4.47

18.82

1.51

late ol in i/s

39.02

1.34

0.59

3.76

17.39

0.86

point

ol margin

36.78

14.51

0.17

17.08

point

ol core

36.56

0.26

1.81

0.11

14.35

0.37

18.96

point

ol outer

35.81

0.26

1.52

13.97

0.49

point

ol outer

35.39

0.30

1.86

14.06

0.44

NTH7

point

ol core

35.72

0.20

2.00

14.10

0.65

NTH7

point

ol outer

35.68

1.64

14.27

0.22

NTH7

point

ol outer

35.96

1.55

14.34

0.29

NTH7

point

ol inner

36.47

0.55

NTH7

point

iscorite

NTH7

point

magnetite?

NTH7

point

wustite

NTH7

point

olivine

NTH7

point

NTH7

0.27

18.23

1.47

< <

Total

80.39 1.18

99.66

0.36

24.91

102.03

0.56

30.99

101.56

0.42

28.40

101.22

17.30

0.39

29.08

98.83

18.47

0.48

27.50

98.50

18.75

0.45

27.57

99.44

17.99

0.51

28.98

99.29

18.17

0.42

29.92

100.66

0.41

29.15

101.49

18.30

1.73

14.48

0.12

28.35

0.31

0.46

4.35

65.37

98.84

28.35

0.49

0.41

69.76

99.01

24.41

0.14

0.27

32.47

0.94

0.12

13.43

olivine

31.25

0.65

0.31

14.72

point

olivine

32.50

1.00

point

olivine

32.60

0.97

point

hole?/glass

37.58

NTH7

point

magnetite?

NTH7

point

iscorite

73.29

98.10

1.72

0.19

49.97

98.85

1.73

0.22

49.15

98.31

13.48

0.97

0.25

50.98

99.17

13.31

0.84

0.16

51.22

99.09

4.46

19.82

24.24

98.61

27.98

0.15

0.72

0.46

69.40

98.71

28.11

0.34

0.35

4.11

66.56

99.47

1.90

0.28

0.39

0.09

6.02

4.11

GeoArch Report 2013/27: Residues from Neath

Sample

SOI

Area/ point

NTH7

point

NTH7

point

NTH7

point

late ol?

36.25

iscorite

27.80

wustite margin

24.39

0.54

1.46

13.54

0.24

0.37

0.45

4.16

65.19

97.97

0.17

0.26

73.37

98.19

1.96

5.65

39.78

Total

99.42

GeoArch Report 2013/27: Residues from Neath

APPENDIX C: Image Archive

GeoArch Report 2013/27: Residues from Neath

Plate C1: Sample NTH2

Plate C8: Sample NTH6

a) NTH2 SOI1. Scalebar 1mm. b) NTH2 SOI2. Scalebar 60µm. c) NTH2 SOI3 Scalebar 600µm. d) NTH2 SOI4 Scalebar 600µm. e) NTH2 SOI5 Scalebar 600µm. f) NTH2 SOI6 Scalebar 600µm. g) NTH2 SOI7 Scalebar 600µm. h) NTH2 SOI8 Scalebar 1mm.

a) NTH6 SOI4 Scalebar 100µm. Plate C9: Sample NTH7 a) NTH7 SOI1. Scalebar 2mm. b) NTH7 SOI2. Scalebar 2mm. c) NTH7 SOI3 Scalebar 2mm. d) NTH7 SOI4 Scalebar 2mm. e) NTH7 SOI5 Scalebar 2mm. f) NTH7 SOI6 Scalebar 2mm. g) NTH7 SOI7 Scalebar 2mm. h) NTH7 SOI8 Scalebar 100µm.

Plate C2: Sample NTH2 a) NTH2 SOI9. Scalebar 1mm. b) NTH2 SOI10. Scalebar 3mm. c) NTH2 SOI10. Al element map. Scalebar 600µm. d) NTH2 SOI0. Ca element map. Scalebar 600µm. e) NTH2 SOI0. Fe element map. Scalebar 600µm. f) NTH2 SOI0. K element map. Scalebar 600µm. g) NTH2 SOI0. Si element map Scalebar 600µm.

Plate C10: Sample NTH7 a) NTH7 SOI1. Scalebar 300µm. b) NTH7 SOI2. Scalebar 600µm. c) NTH7 SOI3 Scalebar 600µm. d) NTH7 SOI4 Scalebar 600µm. e) NTH7 SOI5 Scalebar 60µm. f) NTH7 SOI6 Scalebar 600µm.

Plate C3: Sample NTH3 a) NTH3 SOI1. Scalebar 600µm. b) NTH3 SOI2. Scalebar 60µm. c) NTH3 SOI3 Scalebar 600µm. d) NTH3 SOI4 Scalebar 2mm. e) NTH3 SOI5 Scalebar 1mm. f) NTH3 SOI6 Scalebar 300µm. g) NTH3 SOI7 Scalebar 60µm. h) NTH3 SOI8 Scalebar 300µ m. Plate C4: Sample NTH3 a) NTH3 SOI9. Scalebar 600µm. b) NTH3 SOI10. Scalebar 600µm. c) NTH3 SOI11 Scalebar 600µm. Plate C5: Sample NTH4 a) NTH4 SOI1. Scalebar 1mm. b) NTH4 SOI2. Scalebar 1mm. c) NTH4 SOI3 Scalebar 1mm. d) NTH4 SOI4 Scalebar 100µm. e) NTH4 SOI5 Scalebar 1mm. f) NTH4 SOI6 Scalebar 1mm. g) NTH4 SOI7 Scalebar 2mm. Plate C6: Sample NTH5 a) NTH5 SOI1. Scalebar 1mm. b) NTH5 SOI2. Scalebar 1mm. c) NTH5 SOI3 Scalebar 90µm. d) NTH5 SOI4 Scalebar 1mm. e) NTH5 SOI5 Scalebar 1mm. f) NTH5 SOI6 Scalebar 1mm. Plate C7: Sample NTH6 a) NTH6 SOI1. Scalebar 1mm. b) NTH6 SOI2. Scalebar 1mm. c) NTH6 SOI3 Scalebar 1mm. d) NTH6 SOI4 Scalebar 1mm. e) NTH6 SOI5 Scalebar 1mm. f) NTH6 SOI6 Scalebar 1mm. g) NTH6 SOI7 Scalebar 1mm. h) NTH6 SOI8 Scalebar 1mm.

Plate C1

PlateC2

Plate C3

Plate C4

Plate C5

Plate C6

PlateC7

PlateC8

Plate C9

Plate C10

GeoArch geoarchaeological, archaeometallurgical & geophysical investigations Unit 6, Block C, Western Industrial Estate, Caerphilly, CF83 1BQ Office: Mobile:

029 20881431 07802 413704

E-Mail: Web:

Tim.Young@GeoArch.co.uk www.GeoArch.co.uk

Glamorgan-Gwent Archaeological Trust Ltd (Projects Department) QUALITY CONTROL Report Title: Nidum. Excavations at the Roman Auxiliary Fort 2010 to 2012: Neath Report Date: July 2018 Report Number: 2018/038

Report prepared by:

Martin Tuck

Position:

Senior Project Manager

Date:

30/07/2018

Illustrations prepared by: Position: Date:

Paul Jones

Senior Illustrator

30/07/2018

Report checked and authorised by: Position:

Andrew Marvell

Chief Executive

Date: 30/07/2018 As part of our desire to provide a quality service we would welcome any comments you may wish to make on the content or presentation of this report.

Glamorgan-Gwent Archaeological Trust Ltd. Heathfield House, Heathfield, Swansea, SA1 6LE Tel. 01792 655208; Fax. 01792 474469 Registered Charity no. 505609 Web: www.ggat.org.uk e-mail: projects@ggat.org.uk