skip to content

Department of Archaeology


In May 2007 a small team, under the direction of R. Rabett, carried out preliminary observations of known historical remains in the park (dating from the 10th to 16th century AD) and, in particular, began excavation of a shell mound in the mouth of Hang Boi, a cave in the north-west of the park. Hang Boi has a spectacular underground chamber and is an intended fixture for future tourism in the park. This work revealed >2 m of cultural deposits from human occupation at Hang Boi, spanning the period from c.12,300 to c.10,500 calendar years ago (Rabett et al., 2009a).

In November 2007, the project undertook a short two week survey field-season to ascertain the likelihood of additional evidence of early settlement that might be comparable to that which we were uncovering at Hang Boi. Three sites were investigated (Rabett et al., 2009b). The top of the undisturbed cultural material-bearing deposits at one of these, a cave c.1 km from Hang Boi, called Hang Trống, was radiocarbon dated to 12,700 years ago. Access to Hang Trống was tortuous, however, making any immediate excavation of the site problematic, and leaving the 2008 field season focused on the developments at Hang Boi.

The 2008 season ran for three weeks (Jun.-Jul.), during which time we were able to successfully identify and track discrete stratigraphic layers of crushed and whole shell within the Hang Boi midden, allowing us to implement single context recording for the first time (the 2007 excavation had followed a spit digging strategy). Work undertaken during this season was key to creating a reliable picture of site formation processes at Hang Boi (Rabett et al., 2011). The age of the sequence was extended back to c.12,600 years ago, while the provision of additional AMS 14C dates for the season revealed an unusually fine-grained sequence and outstanding potential to retrieve information about how early human groups adapted to the major environmental changes that were occurring during this period. The base of the midden, though, had not been reached. Preliminary efforts to determine the viability of using AAR dating on shells from the site proved promising after the 2007 season and were followed up by further samples this year. Although, ultimately, the provision of absolute dates was not possible, due to high variably in shell chemistry, a relative sequence was able to verify the general picture of site antiquity and highlight that occupation may have persisted into the Holocene; a consideration we had initially discounted.

In Nov. 2009 Lucy Farr continued to head-up excavation in Trench 1 and a radiocarbon sample from the then bottom of the sequence produced a date of c.13,600 cal. BP. Whilst excavation continued in our main trench, at Hang Boi, a second trench was opened close to the wall of the cave by Jo Appleby. The objective here was to explore the possibility of human burials – known to have been placed close to cave walls at other sites of comparable age and character – and to determine the make-up of the midden's periphery. Although no human remains were uncovered, we anticipate that information generated will prove important to our understanding of how the midden formed.

By 2009, the park developer had constructed a path to Hang Trống, allowing us to begin excavations in this cave. A small team under the site direction of Chris Stimpson proceeded to uncover a cultural sequence through the upper 2 m of deposits that bore some similarities but also some notable differences to what we had been finding at Hang Boi, including in the kinds of lithics that were emerging. The suite of radiocarbon samples collected in 2009 produced one modern date at the top of the sequence, suggesting the surface layers are somewhat mixed. The three dates taken from levels beneath this, however, confirmed that human occupation here is almost entirely of late Pleistocene antiquity, with dates from c.15,400-18,500 cal. BP. Use of this site appears to have ended not long before that at Hang Boi began. In 2009 a programme of high-resolution geoarchaeological and isotopic work was started at both caves by Dr. Mike Morley (Oxford Brookes University) and Natalie Ludgate (the Open University) with the aim of gaining detailed understanding of differences in their respective site formation processes.

During the field work in Nov. 2010 we reached the base of the shell midden in Hang Boi, c.0.5 m below the previous year's bottom of excavation. Indications are that the midden rests on the large fragments of limestone blocks from an earlier roof collapse event. The limited nature of the excavated area precluded further exploration of or beneath this debris. Although further work at this cave would be undoubtedly fruitful, at this point we concluded our investigation here. Meanwhile work continued at Hang Trống under Chris Stimpson with the opening of two further trenches in addition to the one we were already digging in the centre of the cave. Trenches 2 and 3 were excavated adjacent to the eastern and western walls, respectively by Jo Appleby, Nguyêń Cao Tâń, David Simpson and Marc Verhoeven. Evidence of localised burning (most likely hearths) was encountered during excavation of all three trenches, with a particularly good series appearing in Trench 3, on the eastern side of the cave; an area that is currently comparatively sheltered. Preliminary field observations appear to show that at least some of these features at Hang Trống are in their primary position, as evidenced by reddened substrates and corroborated by other diagnostic, internal macro-structural features. Excavations in Trench 3 also produced a small collection of bone, shell and lithics that, from their discrete position within a rock niche, were almost certainly deliberately placed during the course of one of these episodic occupations. In Trench 1 we broke through the base of the shell midden and into a distinctly more compact and stony yellowish silt-rich colluvium containing few shells. These deposits continued to yield a small amount of cultural material, including lithics and animal bone, and one of several fragments of charcoal recovered gave us a date of c.24,400 cal. BP, just prior to or a very early phase of the Last Glacial Maximum; making climatic deterioration a possible cause behind this marked change in deposition. Geoarchaeological analysis of these layers is currently (June 2011) in progress. Their study will also form part of a wider macro-botanical examination of the local modern vegetation and its prehistoric counterparts by Jasminda Ceron (University of the Philippines), which was instigated during this field season. The final component of the 2010 fieldwork was the creation of three-dimensional LiDAR scans of both Hang Boi and Hang Tràng by John Meneely (Queens University, Belfast). This highly accurate spatial data and visual representation of both caves will be a valuable aid in reconstructing site formation and in presentation, both scientific and public.


In November 2011 targeted archaeological excavation and geomorphological field sampling of cultural deposits was carried out at Hang Moi cave. This new site had been surveyed briefly at the end of the 2010 field season, at which point it was thought (on account of its low elevation and evidence from surface scatter) a good prospect for tracing the Early to Mid-Holocene occupation of this landscape, and responses to the Mid-Holocene high-stand. Two trenches were opened during the 2011 season. Trench 1 (2x1 m) was situated at close to the western wall and furthest from the entrance of this small cave. This produced a complex sequence of hearth deposits and proved quite rich in terms of cultural material, including faunal remains, highly fragmented corded ware pottery of known local early Neolithic affiliation (Da But), and a beautiful polished jade axe. A further significant find in this trench emerged late in the season as we began to uncover what proved to be a series of small stake holes around a pit and probable hearth further down in the excavated sequence - possibly from some kind of tripod structure or other hearth architecture. One of the stake holes was half-sectioned; recovered sediment was retained for laboratory analysis and charcoal from the hole was submitted for 14C dating to Queen's University, Belfast. The resultant date of c. 5,500 cal. BP confirmed our expectations raised by the ceramics about the antiquity of this part of the site.

Trench 2 (a dog-leg 2x2 m) was located against the north wall of the cave. This trench produced a very different but fortuitously complementary picture to that uncovered in Trench 1. Overlying 'recent' cave sediments quickly gave way to a thick more or less homogeneous grey ashy deposit. This was particularly rich in faunal remains (molluscan, crustacean and vertebrate) as well as quite large fragments of the same corded (Dabutian) pottery. The greater size of fragments, together with the rich faunal assemblage, suggested to us that this was a midden deposit, and one that had probably been situated away from the main area of activity in the cave. One charcoal sample taken from near the base of the excavation for this year in Trench 2 produced a radiocarbon age almost exactly comparable to that from Trench 1.

Overall, the archaeological evidence from the three caves that have made up the Tràng An Archaeological Project overlap incredibly well creating an almost unbroken sequence of human activity spanning c.20,000 years: clear testament to the richness of the prehistoric record in Tràng An. In each cave, recent and historic use appears to have been surprisingly slight, leaving the records of their ancient use perhaps uniquely intact as well as readily accessible. In light of this and the fact that the majority of the park's numerous caves have yet to be thoroughly investigated, the likelihood for extending the range and detail of this record is very high, while the value that can be attached to the archaeological significance of the park cannot be underestimated. Partly in light of these discoveries, Tràng An is currently the subject of a UNESCO World Heritage Site bid.


Last updated on 17/10/2012
Information provided by Ryan Rabett