opposite The portal dolmen at Trellyffaint, north Pembrokeshire. We are looking east, with the Mynydd Preseli (Preseli Mountains) in the background.

Trellyffaint: how excavating a Pembrokeshire portal dolmen illuminated Neolithic dairy farming in Wales

A project that initially set out to study rock art on a c.5,000-year-old monument has uncovered what could be the earliest direct evidence of dairy farming in Wales. George Nash, Carol James, and Thomas Wellicome from the Welsh Rock Art Organisation (WRAO), and Julie Dunne and Emmanuelle Casanova from the University of Bristol, explain more.

Start

Standing a few metres east of Cardigan Bay in north Pembrokeshire, Trellyffaint is a Neolithic portal dolmen, a type of chambered tomb that represents what is thought to be the earliest architectural style in western Britain (with a probable date range of c.3800-2800 BC). Its name, which might be translated as ‘Toad’s Hall’, stems from the early medieval historian and geographer, Giraldus Cambrensis (Gerald of Wales), who believed that its occupant was a chieftain who had been devoured by toads. Even without such colourful legends, though, Trellyffaint is special. This area of Wales boasts the largest concentration of Neolithic burial- ritual monuments, numbering around 60 extant sites with a further 70 surviving as ruins or place-names, but among these Trellyffaint is unusual in that it may have two chambers, something paralleled at only a handful of sites in western Britain. The monument is also one of only nine Neolithic chambered tombs in Wales that has rock art on its capstone.

opposite The portal dolmen at Trellyffaint, north Pembrokeshire. We are looking east, with the Mynydd Preseli (Preseli Mountains) in the background.
The portal dolmen at Trellyffaint, north Pembrokeshire. We are looking east, with the Mynydd Preseli (Preseli Mountains) in the background.

On a local level, Trellyffaint is one of eight extant stone chambered sites that occupy the ridges and coastal fringes of the Nevern Valley, all of which retain unique architectural traits though generically they are either portal dolmens (such as Trellyffaint and Llech y Dribedd) or hybrids associated with the Cotswold-Severn Group of monuments, such as Pentre Ifan. One of its closest neighbours is Trefael, once classified as a Bronze Age standing stone, which also has cupmarks gouged on to the upper surface of a probable chamber capstone (see CA 276). The prevalence of prehistoric monuments is not surprising: archaeological evidence from this valley and its estuary shows that this slightly undulating landscape was exploited by both Mesolithic and Neolithic communities, who utilised the rich food resources available and gathered beach pebbles to make flint and chert tools, although evidence for settlements is still lacking.

As for Trellyffaint itself, today the burial monument is much-ruined thanks to many years of farming that has ploughed out the cairn and earthen mound that once covered the stone chambers. The remnants of a mound edge can still be seen from above, however, and it is possible to reconstruct something of its original layout. It seems likely that the main chamber had been rectangular in plan, with a small chamber located to the west, now surviving as three upright stones. The capstone over the main chamber appears to have been oriented north-west/south-east and measures 2.1m by 1.8m, but has been moved at some time in the recent past, and on top of the capstone are up to 75 cupmarks, along with several gouged lines.

left Trellyffaint is one of only nine Neolithic chambered tombs in Wales that has rock art on its capstone: it has up to 75 cupmarks, as well as several gouged lines. The meaning of these markings remains obscure, and it was to examine these features that the project described in this article had originally set out. Instead, they discovered what could be the earliest direct evidence of dairy farming in Wales.
Trellyffaint is one of only nine Neolithic chambered tombs in Wales that has rock art on its capstone: it has up to 75 cupmarks, as well as several gouged lines. The meaning of these markings remains obscure, and it was to examine these features that the project described in this article had originally set out. Instead, they discovered what could be the earliest direct evidence of dairy farming in Wales.

These distinct features were first published in the Royal Commission inventory in 1925, and formed the original focus of our own project almost a century later. Between 2015 and 2019, our research programme initially focused on the capstone rock art, though we also carried out an earthwork survey of the monument – which not only recorded the in situ stone elements but also revealed the denuded extent of the mound – traced the capstone, and undertook geophysical surveys, exploring some of the buried anomalies that this last technique revealed c.15m north of the monument through targeted excavation. While our project initially set out to investigate the ancient artwork, though, we would uncover evidence of activities that were at once more everyday and more surprising, revealing what could be the earliest direct evidence of dairy farming in Wales.

above Although the dolmen is in a ruinous state due to may years of farming that has ploughed out its cairn and earthern mound, this aerial view of the monument shows that traces of a mound can still be seen from above.
Although the dolmen is in a ruinous state due to may years of farming that has ploughed out its cairn and earthern mound, this aerial view of the monument shows that traces of a mound can still be seen from above.

Hints of henges

As far as we are aware, there has been little research and no intrusive archaeological investigations on the site previously – although one cannot rule out any unrecorded antiquarian activity before the site was designated a Scheduled Ancient Monument in 1927. While we do not know of any antiquarian excavations, we do know that the site attracted the interest of early scholars, who published engravings of the monument; these have proved useful for our team to identify any damage that may have occurred over the past 150 years. Rather more modern forms of imaging also proved vital to our research: between 2016 and 2018, we carried out two geophysical surveys around the vicinity of the monument, covering an area of 40m by 40m in four 20m2 grids. Their results were illuminating, revealing extensive buried traces of activity including several circular anomalies, the largest of which measured around 12m in diameter. The plots revealed a possible trackway, too, running to the east, outside the scheduled area. At that time, the dates for these anomalies were unknown, but they clearly warranted further investigation.

below Geophysical surveys at the site revealed a number of anomalies that warranted further investigation, including the large circular feature shown on this plot of the results. It was interpreted as a possible henge or hengiform monument, and the location of the two trenches opened over its outline are shown as white boxes. The other photo shows the project team at work, using magnetometry.
Geophysical surveys at the site revealed a number of anomalies that warranted further investigation, including the large circular feature shown on this plot of the results. It was interpreted as a possible henge or hengiform monument, and the location of the two trenches opened over its outline are shown as white boxes. The other photo shows the project team at work, using magnetometry.

The circular features were of particular interest: their size and their proximity to the Trellyffaint monument suggested that they could be henges (or hengiforms), a type of buried monument that appears to be a common feature within Neolithic landscapes, especially where extant stone monuments stand close by. The inner circle looked like a perfect candidate to explore further, and so we submitted a formal request to Wales’ national heritage agency, Cadw, to excavate. Particularly strong anomalies had been identified over the south-western section of this circle, and so there we opened two trenches separated by a narrow linear baulk. Both of these revealed a series of ploughmarks running north-east to south-west, and after removing these historic deposits (including the plough soil), about 0.6m below the modern ground level, we exposed a possible Neolithic land surface spanning Trench 1. This was scattered with sherds of abraded pottery and worked flint, and a circular shallow pit measuring 0.80m by 0.75m cut into it. While the upper fill of this pit contained no recognisable cultural artefacts, its lower layers yielded the fragmentary remains of a ceramic Neolithic Grooved ware vessel, which we will discuss further below.

above This shallow pit was found within the north-west area of Trench 1. Its lower fill yielded more than 30 fragments from a Neolithic Grooved ware vessel, which would provide vital evidence of dairy products being used on the site. Here we see the pit fully excavated and the pottery carefully bagged and ready for laboratory examination. ABOVE RIGHT In Trench 2, the team found the remains of a linear ditch or gulley, as well as this pit or post-hole whose stone-lined upper section could be seen in the south-west corner of the trench.
This shallow pit was found within the north-west area of Trench 1. Its lower fill yielded more than 30 fragments from a Neolithic Grooved ware vessel, which would provide vital evidence of dairy products being used on the site. Here we see the pit fully excavated and the pottery carefully bagged and ready for laboratory examination.

The larger Trench 2, which was put in immediately to the north, revealed a circular pit or post-hole, as well as a linear ditch or gulley up to 80cm deep, which measured 2.4m by 0.4m.

In Trench 2, the team found the remains of a linear ditch or gulley, as well as this pit or post-hole whose stone-lined upper section could be seen in the south-west corner of the trench.

Although heavily truncated by later farming activity, this latter feature could clearly be seen running north-south in the western and central sections of the trench, and a small number of abraded pottery fragments were recovered from its whole length. Adding to this picture of Neolithic activity on the site, some 14 pieces of flint and chert were recovered from five archaeological contexts across the two trenches. These materials were readily available locally; the main flint source would probably have been from the nearby beaches of Cardigan Bay – but there were hints of further-reaching connections too. Among this assemblage were three worked pieces identified as blades, and two of these are thought to have originated from outside the locality, based on their colour and lustre. It was the pottery that would prove particularly illuminating, however.

left A retouched flint blade from Trench 1. Three were recovered in all, two of which (including this one) are thought not to be local to the area.
A retouched flint blade from Trench 1. Three were recovered in all, two of which (including this one) are thought not to be local to the area.

The Last Supper

Trench 1 yielded some 34 sherds of pottery, which were carefully excavated in order to maximise the potential for organic-residue analysis. These fragments all came from a single Neolithic Grooved ware vessel, the body of which was coil-constructed and contained visible quartz and stone fragment inclusions (known as grog); initial examination of the pottery suggests a coarse ware fabric with rudimentary decoration around the rim, neck, and base. Recovering any pottery at all was an exciting development, as the acidity of the local soil means that pottery (and bone) rarely survives in this part of Wales, but there was more to come. Traces of possible burnt-on food residues could be seen on the inner surface of many of the potsherds, indicating that this could have been a cooking vessel.

To find out more about how the pot had been used, the potsherd assemblage was sent after the excavation to Dr Julie Dunne at the Organic Geochemistry Unit, University of Bristol, for organic-residue analysis. It was hoped that she would be able to determine whether absorbed lipids – the fats, oils, and waxes of the natural world – were present in the vessel. After carrying out combined chemical and isotopic analysis of these absorbed residues, the results revealed that the vessel had been routinely used to process solely dairy products, such as milk, butter, and cheese, from ruminant animals, either cattle, sheep or goats. Lipid concentrations were high, suggesting that the vessel had undergone sustained use in the processing of these dairy fats.

left Pottery rarely survives in this area of Wales due to the acidity of the local soil, making the recovery of 34 sherds from Trench 1 particularly fortuitous. More exciting still, many of the fragments bore traces of burnt-on food residues (BELOW) containing evidence of what the pot had once held.
Pottery rarely survives in this area of Wales due to the acidity of the local soil, making the recovery of 34 sherds from Trench 1 particularly fortuitous. More exciting still, many of the fragments bore traces of burnt-on food residues (BELOW) containing evidence of what the pot had once held.

This provided the opportunity for Emmanuelle Casanova, also from the Organic Geochemistry Unit, to apply the revolutionary new compound-specific lipid radiocarbon dating method, recently developed at the University of Bristol, to provide the first direct dating of a Neolithic pot in Wales. Radiocarbon measurements on the lipids from one of the potsherds produced a radiocarbon date of 4339+28 BP (BRAMS-3041) which, after calibrating into calendar age, suggests that the vessel was in use between the 31st and 30th century BC (that is, 3100-2901 BC). These results represent the earliest known direct evidence of dairy production in Wales, and confirm that Neolithic groups there were exploiting dairy products by at least the 4th millennium BC (and very likely earlier), hinting at the importance of milk or products such as butter or cheese to these early Welsh farmers.

above Emmanuelle Casanova working in the radiocarbon facility (BRAMS). There, she was able to use a new compound-specific lipid radiocarbon dating method, recently developed at the University of Bristol, to provide the first direct dating of a Neolithic pot in Wales, and to confirm that dairy production was happening in Wales in at least 3100 BC.
Emmanuelle Casanova working in the radiocarbon facility (BRAMS). There, she was able to use a new compound-specific lipid radiocarbon dating method, recently developed at the University of Bristol, to provide the first direct dating of a Neolithic pot in Wales, and to confirm that dairy production was happening in Wales in at least 3100 BC.

The dates reflect a time when radical new ideas known as the ‘Neolithic Package’ were sweeping across the Channel to Britain from continental Europe (CA 290). This transformation saw the introduction of animal domestication, pottery production, new tool types, and agriculture, as well as sparking the construction of imposing stone monuments and a much more settled way of life than Britain’s nomadic hunter-fisher-gatherer communities had known. With animal husbandry came a more reliable source of food, as well as new foodstuffs including the dairy products that were being used at Trellyffaint. Why pots containing milk, butter, or cheese had been brought to the hengiform monuments is not known, but given that no domestic settlement has been identified in this otherwise ritualised landscape, it is possible that they may have had some kind of ceremonial role. Perhaps the dairy products were consumed as part of the rites performed at the site, or in communal feasting after these events – or they could have been intended as some kind of offering, literally buttering up the supernatural forces who may have been venerated beside Cardigan Bay some 5,000 years ago.

Further information
The Welsh Rock Art Organisation is a non-profit research group that operates in Wales and along the Border Lands. For more information, visit www.facebook.com/rockartwales. If you are interested in carrying out commercial residue analyses, Julie can be contacted on julie.dunne@bristol.ac.uk.
You can read more about compound-specific lipid radiocarbon dating in E Casanova, T D J Knowles, A Bayliss et al. (2020) ‘Accurate compound-specific 14C dating of archaeological pottery vessels’, Nature 580: 506-510, https://doi.org/10.1038/s41586-020-2178-z.
‘Science Notes’ in CA 336 also discusses compound-specific radiocarbon dating, though not specifically applied to lipids.