Previous Next

An Archaeology of Early Christianity in Vanuatu


Erromango: Archaeology and the Martyr Isle

Erromango is the fourth-largest island in Vanuatu, and the largest in the southern TAFEA province, with a surface area of 855km2 (Figure 1.1; TAFEA is an acronym consisting of the first letters of the five islands in the province: Tanna, Aniwa, Futuna, Erromango, Aneityum). The island has a high volcanic plateau that has formed over the last 5.5 million years, with most of the volcanic activity occurring within the early Pleistocene period and continuing until the end of the Pleistocene. The volcanic cones that formed Traitor’s Head on the east coast of the island are the most recent formations. The coasts are characterised by uplifted limestone reef terraces, especially in the northern and western areas of the island (Colley and Ash 1971).

The early archaeological sequence of Erromango is better understood than on the other TAFEA islands. Initial human settlement of Erromango is well attested by the presence of Lapita pottery, which is the marker of initial colonisation in the western islands of Remote Oceania (Green 1991). The sites of Ponamla and Ifo are the primary Lapita sites on Erromango, dating to roughly 2800BP (Bedford 2006: 32–39) (Figure 2.1). The Ponamla site is notable for the presence of possible 2,700-year-old stone structures (Spriggs 1999). Within a few centuries of settlement, there is a notable change in ceramic decoration style to incised and fingernail-impressed wares, as well as red-slipped and plain ceramics. Ceramic technology disappears from Erromango by about 2000BP. After this point, archaeological material culture consists of distinctive stone and shell artefacts, both tools and objects of personal adornment (Bedford 2006). Rock art sites are widespread on Erromango, with a prevalent tradition of black linear paintings and drawings (Wilson 1999, 2002).

First European contact on Erromango occurred during Cook’s second expedition in 1774. This was a very brief encounter. Cook went ashore at a place he called Polenia Bay, now Port Narvin (a European orthography of ‘Potnarvin’, literally ‘sandy place’). After handing out some ‘trinkets’ and receiving some food and fresh water, Cook determined to return to the main ship. At this point, the local people attempted to try to keep Cook and the ship’s boat on shore. In the ensuing scuffle, Cook ordered his men to fire and it appears that four Erromangans died in the attack. Cook and his men returned to the ships while local people continued to fire arrows and throw stones as they retreated (Beaglehole, ed. 1969: 477–480). After this, the ships weighed anchor and headed south for Tanna (see Chapter 3). The main legacy of this encounter is the name Erromango from Cook’s hearing local people saying, ‘armae n‘go’, ‘good food’ when offering yams. The original name for the island was Nelokompne (Huffman 1996: 129), though I will use the more common name Erromango here.

It would be another 50 years before Europeans would return to Erromango in earnest. The discovery of sandalwood, a valuable trade item, in the area known locally as Umpongkor, now called Dillon’s Bay, in 1828 attracted the first traders to the island (Shineberg 1967: 16 –28). In a single decade, one trader removed 1,600 tons of sandalwood from the island, with a gross value of £64,000, a monumental sum for that time. Within 50 years, Erromango’s uplands were largely deforested and sandalwood almost entirely wiped out. By one estimate, £175,000 worth of sandalwood (over £19,000,000 in 2016 estimated equivalent currency; Morley 2016) was removed from Erromango (Robertson 1902: 34, map insert). As we will see, tensions between Erromangan people and sandalwood traders would have some impacts on the missionary experience.

With sandalwood largely exhausted, traders in the New Hebrides sought other sources of profit. Particularly important on Erromango was the labour, or ‘Blackbirding’ trade (Docker 1970; Palmer 1871; Shineberg 1999), which resulted in hundreds of young Erromangan men travelling to Queensland and Fiji, primarily to work on sugarcane plantations. Some would not return. Missionaries were vehement in their descriptions of the negative impacts of the labour trade (e.g. Kay, ed. 1872), though these accounts should be tempered against the fact that they were watching potential converts disappear from the islands. Many Erromangans and others likely boarded labour vessels voluntarily, though there was some trickery, kidnapping, and abuse during the Blackbirding era (Moore 1992). The labour trade combined with epidemic diseases to decimate the Erromangan population. By one estimate, the mid-19th-century population of about 5,000 (possibly 7,000; Gordon, ed. 1863: 134), which may already have been down because of earlier contacts, was reduced to a low of about 600 inhabitants in 1967 (Colley and Ash 1971: 2–3). One of the early ethnographers on the island believed that population decline was to some extent responsible for loss of cultural traditions and general ‘despondency’ or ‘apathy’ on Erromango (Humphreys 1926: 123, 129, 133, 148; see also Speiser 1922).

Today, Erromango remains sparsely populated, but the ‘despondency’ asserted as relating to cultural traditions was certainly overstated. Erromangan people continue to have a keen interest in cultural traditions and arts, many of which are currently undergoing something of a renaissance (Carillo-Huffman et al. 2013; Christidis et al. 2009; Huffman 1996; Lawson 2001; Naupa, ed. 2011).

Kastom on Erromango

What was life like on Erromango in the decades leading up to European, and particularly missionary, arrival on the island? To understand this, we can turn to a variety of sources, including early missionary accounts (Robertson 1902) and ethnographies (Humphreys 1926). Both of these sources come with their own kinds of biases, and must be read critically (Douglas 2001). On the other hand, they also represent early recordings of traditions not currently practiced or largely transformed. It is worth ‘triangulating’ the common observations from various sources that may reflect some aspects of cultural reality. What follows is an account of Erromangan kastom that is by necessity incomplete, fragmented, and non-authoritative. It is a preliminary attempt to organise this information in a way that may explain patterns relevant to the historical archaeology of missionary encounters that follows. Future research, and particularly the observations of Erromangan scholars (e.g. Naupa, ed. 2011) can and should refine these observations. This basic outline offers a starting point for the broad patterns and ‘deep structures’ that may have shaped Erromangan experiences through the colonial era.

Traditionally, Erromango was divided into six districts (Figure 2.1). The districts were called lo (canoe), and consisted of collections of villages, though an early ethnographic account indicates that tribal organisation was not strictly geographic in nature (Humphreys 1926: 128). There were six or seven languages on the island originally, though only three are spoken today, and the primary language, Syé, is quickly absorbing the remaining speakers of the other two. Language boundaries and district boundaries did not overlap precisely across the island.

Figure 2.1 Map of lo (districts) on Erromango, and sites mentioned in text. Note that division boundaries are approximate only and meant to be for general reference.

Source: James Flexner

Chiefs were called Fan lo, and women of the chiefly class referred to as Nasimnalan. Each village had a Fan lo, and there was a higher-ranking Fan lo in charge of each district. Commoners were referred to as Taui natimono. Chiefly women were required to marry men from the Fan lo class, but in some cases, male chiefs could marry women of the commoner class, particularly as Fan lo were allowed multiple wives. Chiefly titles were hereditary rather than achieved, passed primarily through the male lineage, though chiefs could also nominate an appropriately ranked successor as they neared the end of their lives (Humphreys 1926: 128–134; Spriggs and Wickler 1989: 83–85).

Erromangan subsistence used the typical suite of Oceanic crops, including yams, taro, banana, coconuts, various fruits and ‘island cabbages’, as well as marine resources (Figure 2.2). While the garden soils of the west coast were less productive overall, each lo offered enough resources for surplus agricultural production in normal years. Early observers recorded 26 varieties of yams, 17 of taro, 13 of banana, as well as a great variety of other crops. Food was either roasted in the coals of an open fire, or prepared in earth ovens. Dishes included roasted tubers, fish, and pigs, and various sorts of starchy puddings, including neoki, which was made with yam or papaya, coconut, and prawns. People ate fruit or small amounts of roasted vegetables or fish during the day, with the main meal taking place in the evenings (Humphreys 1926: 138–140; Robertson 1902: 376–381). Pigs were raised on the island, but were used primarily in feasting.

Figure 2.2 A traditional Erromangan garden.

Source: James Flexner

The main type of feast on Erromango was the nisekar (Humphreys 1926: 180–182; Naupa, ed. 2011: 30–32; Robertson 1902: 390–391). These feasts, presented by one village to another through the Fan lo, often began with a sham fight, followed by an exchange of gifts, and feasting could last for several days. Often a feast presented by one village would be reciprocated by the other, sometimes almost immediately. A large ceremonial structure, called the nevsem, was often constructed to display agricultural produce at nisekar and other events, such as a chief’s funeral. The feasts and nevsem have been interpreted both as a means of keeping the peace on the island (Naupa, ed. 2011: 24–26) and as a means of competition among rival chiefs (Spriggs and Wickler 1989: 84–85). There is no reason to take these as mutually exclusive proposals, as competitive feasting provided an alternative to competition through warfare, though it has been suggested that feasting and fighting were carried out in alternate seasons (Humphreys 1926: 181).

Another significant structure in Erromangan life was the Siman lo, a large ceremonial house structure inhabited by young and unmarried men (Figure 2.3). It was a shared sleeping and eating house, but also served a number of important social functions. The Siman lo was a symbolically dense, important meeting place where the Fan lo would meet to make community decisions. It was also the place where men would meet in the evenings to drink kava, the intoxicating beverage made from the root of the Piper methysticum plant (or there may have been a separate, but nearby kava house of similar form; Humphreys 1926: 156–158, 178; Naupa, ed. 2011: 26–30). A mid-19th-century Siman lo was measured as 100 ft (30m) long, 20 ft (7m) wide, and 25 ft (8m) high (Robertson 1902: 375). Erromangan people made a variety of objects out of wood, plant materials, bark, shell, and stone, which will be described at greater length in Chapter 5. Star-headed carved wooden clubs (telugohmti), stone money (navela), barkcloth (nemasitse) and cowrie shells (numpuri) were important ceremonial and exchange items. The subsistence system and chiefly system provided the structure for everyday life on Erromango in the centuries surrounding the arrival of missionaries and other Europeans.

Figure 2.3 Siman lo (men’s house) at Dillon’s Bay, Erromango.

Source: James Flexner

Missionaries who arrived in Erromango settled on an island with already rich supernatural beliefs, legends, and stories, many of which have been lost because of demographic and religious change on the island. Others have survived through the present, and in early texts. Erromangans believed in a creator being, Nobu, who made the island, the first people, the navela (stone money), and in some stories the surrounding islands, but was generally not otherwise involved in the world after that. Much more important supernatural beings consisted of ghosts and spirits (natemas). Ghosts of deceased ancestors were thought to roam the island at night, and if encountered, could cause illness, bad fortune, or even death, and were to be propitiated with offerings of food (Humphreys 1926: 165–167; Robertson 1902: 389).

There were powerful practitioners of magic on Erromango, called Tavuwa. Tavuwa were capable of both beneficial and detrimental magic. The beneficial kinds involved calling the rains and causing crops to grow. The same kinds of magic could be used to call cyclones or damaging storms and damaging crops to limit growth for reasons both apparently bad (attacking a rival district) and good (reducing crop waste, though it should be noted that this reasoning is probably a bit tenuous, as excess crops could always be used to raise more pigs for feasting). Perhaps equally influential was the ability of Tavuwa to cause illness or death. The method is remarkably similar to that used on Tanna (see Chapter 3), where something that had come into contact with a man’s body, especially food waste such as banana peels or sugarcane mast, was tied in a bundle with a magic stone (natemas evai) with malicious power. In fact, it has been suggested that the magic stones themselves as well as these practices came originally from Tanna. Missionaries and early ethnographers suggest all deaths were believed to be caused by natemas or Tavuwa (Humphreys 1926: 167–170; Robertson 1902: 400–402), though of course such an observation should be read as partly reflecting European biases about native superstitions.

Erromango is known to have had strong connections with neighbouring islands Tanna and Aniwa. Besides the natemas evai, which indeed could originally have been brought to Erromango from Tanna, there were more regular material exchanges. Black manganese and red ochre, used as pigment, clubs, bows and arrows, and brides were exchanged with neighbouring islands. In return, Erromango received pigs, kava, and shell valuables (Spriggs and Wickler 1989: 84). Thus when missionaries arrived on Erromango in 1839, they landed on an island that already had a rich cosmology and human social ecology. They also were not the first Europeans to visit the island, and this may be partly responsible for the tragic shape of mission history on Erromango.

The First Martyrs: John Williams and James Harris

The basic events surrounding the deaths of John Williams and James Harris at Dillon’s Bay on 20 November 1839 are generally well agreed upon. Williams was a London Missionary Society missionary who had success on Samoa, the Cook Islands, and Tahiti, having worked in Polynesia since 1817. Harris, a young man who had travelled to the South Seas for health reasons, was working as his secretary. With a contingent of recent Samoan converts who had volunteered to expand the mission field westward, Williams and Harris travelled to the New Hebrides in the mission brig Camden to attempt to establish new mission stations. They arrived on Tanna on 18 November, dropped a few Samoan teachers off at Port Resolution, and then went on to Erromango. On the morning of the 20th, Williams and Harris landed on the beach at Dillon’s Bay in the hopes of meeting a chief. Despite the reluctance of local people, Williams and Harris insisted on travelling inland. Subsequently they were chased and attacked by local warriors, and both men died trying to escape back to the mission ship (see Gordon, ed. 1863: 99–101; Robertson 1902: 47–56).

News of Williams’ and Harris’ deaths travelled fairly quickly, solidifying the impression among Europeans that Erromango, and the New Hebrides more generally, were dangerous islands ruled by black magic and death (American Sunday School Union 1844; Copeland 1878). Ironically, this reputation if anything seemed to further encourage potential missionaries hoping to ‘bring light to dark isles’. A week after this event, the HMS Favourite arrived and collected the bones of Williams and Harris (note that the bodies had allegedly been butchered and cooked, so bones rather than bodies were offered). Williams’ body was returned to Samoa, and is currently buried on the foreshore in Apia (Figure 2.4).

Figure 2.4 Marker of John Williams’ grave, Apia, Samoa.

Source: James Flexner

The European version of these events is missing several key plot points. H.A. Robertson, a later missionary, was able to ascertain from talking to several eyewitnesses that the reasons for the attack stemmed partly from outrages committed by previous white visitors, one of whom ‘had stolen a chief’s daughter’ (Robertson 1902: 56). Local people today remember that two sons of a local chief had been shot during a feast just before Williams’ and Harris’ arrival. It is possible that this was the feast relating to the circumcision ritual (Humphreys 1926: 177–179). It would have been an unimaginable insult to mar such an event with a violent act. Regardless, it is clear that the people at Umpongkor (Dillon’s Bay) had good reason to be wary of the intentions of the foreigners. It was suggested locally that after these events, Erromangans decided to block any European settlement on their island.

The deaths of Williams and Harris had deep reverberations in the Erromangan as well as European communities. The ‘event landscape’ relating to Williams’ and Harris’ deaths is remembered to this day (Flexner 2014c: 8–11), encapsulated in place on the west Erromangan coast. Walking through the landscape, Erromangan people can point to the places and material traces that relate to 20 November 1839. At the time of Williams and Harris’ arrival, there was a nisekar happening at a place in the uplands called Nokiyangouwi, where a nevsem had been erected. Auwi-Auwi, the chief who was hosting the nisekar, decreed that the visitors were to be welcomed so long as they remained on the beach (points 1 and 3 in Figure 2.5). Despite warnings by local people (who did not speak English just as Williams did not speak Erromangan), Williams and Harris insisted on proceeding inland, and the events proceeded as recorded above.

From here, the event landscape again picks up some of the threads not present in the documentary accounts (Figure 2.5). As the crew of the Camden retreated, the bodies of Williams and Harris were dragged up the river. Where some drops of Williams’ blood fell, red-leafed lompot (Cordyline fruticosa; Bislama nangare) plants sprouted, and the same kinds of plants still grow on the spot today (point 4 in Figure 2.5). Williams’ body was placed on a large stone, and cupules were pecked to mark the location of his head, feet, and hands (point 6 in Figure 2.5; Robertson (1902: 58) suggests local people believed the missionary was short and stout). From there, the bodies of Williams and Harris were taken to the river, where they were divided among local chiefs, and taken away to be ritually eaten (point 7 in Figure 2.5). Or, as Robertson (1902: 58) recounts, the bodies were taken to a village about three miles inland to be exchanged for pigs. There is some question in local reckoning about whether Williams and Harris were cannibalised. Several studies suggest that scepticism is a reasonable stance when presented with such cases (Arens 1980; Barker et al., eds 1998; Goldman, ed. 1999; Obeyesekere 2005). We should simultaneously be careful to take the equally erroneous revisionist stance that suggests cannibalism never happened in the past in Melanesia. One way to find out for certain would be to examine the bones of John Williams for evidence of butchery or cooking, though obviously this would be a politically and socially complicated process involving many stakeholders.

Regardless of whether it involved cannibalism or not, John Williams’ death had a major impact on Erromangan consciousness. Erromangan people’s guilt was so intense that in 2009 a reconciliation ceremony was held at Dillon’s Bay with descendants of John Williams and descendants of those responsible for his death (Mayer et al. 2013; Naupa, ed. 2011: 92–108). It was believed that Erromango had been ‘cursed’ by the death of John Williams, and it was necessary to make amends before the population of the island could grow to its pre-colonial level. The name of Dillon’s Bay was changed officially to Williams Bay (though Dillon’s Bay remains in common use outside of west Erromango, and was used historically so will be used elsewhere here).

Erromango remained largely without resident missionaries in the years immediately following this event. A few Samoan teachers were present on the island in 1840, dropped off by the Camden, but they had little success and had to be essentially rescued from the island in 1841 (Liua‘ana 1996: 60–63). In 1848, the Canadian missionary John Geddie established the first successful mission to the New Hebrides on Aneityum. A little over a decade later, George Gordon and his wife Ellen Powell Gordon landed at Dillon’s Bay, becoming the first resident European missionaries on ‘the martyr isle’.

Figure 2.5 Event landscape covering the death of John Williams and James Harris (see text for explanation of features).

Source: James Flexner

George and Ellen Gordon at Dillon’s Bay

George Gordon grew up on Prince Edward Island in the Lower Provinces of British North America. Gordon’s grandfather had emigrated from Scotland after serving in the British Army during the American Revolution and later in India. George Gordon served as a city missionary in Halifax before volunteering for the foreign mission and being appointed to the New Hebrides, for which he departed in 1855. On his way to the mission field, he met Ellen Powell in London and they were married before departing together for the South Seas. The couple settled at Dillon’s Bay in June 1857 (Gordon, ed. 1863).

The Gordons’ mission work progressed slowly but steadily. They were supported by Samoan teachers who were resident on the island. Early victories included the conversion of the young Erromangans Yomot and Usuo, who became important supporters of the church. The Gordons lived at Dillon’s Bay for slightly less than four years. During this time they won some native converts, and their work is credited with the long-term success of the Presbyterian Mission on Erromango (Robertson 1902: 66–71). They are best remembered as Erromango’s second martyrs, but their final moments were only a tiny fraction of the time the Gordons spent at Dillon’s Bay.

There is some documentation of the Gordons’ lives on the island, which mostly revolved around learning language and customs and attempting to convince Erromangan people to convert (Gordon, ed. 1863: 132–184). However, we know little of the form of everyday life for the missionaries. Further, if conversion work was a process of material entanglement as much as transformation of spiritual belief, we must look to the spaces where such interactions took place. For this, we turn to the archaeological record, which has proven to be an invaluable primary source of interaction for understanding mission life in Oceania (e.g. Lydon 2009; Middleton 2008; Smith et al. 2012, 2014).

The Gordons built their house high on a cliff overlooking the Williams River, and fairly far from the main population centre in the river valley. This was apparently to escape the ‘bad air’ and miasmas that caused tropical illnesses, particularly malaria (Gordon, ed. 1863: 199–200). Indigenous settlement in the river valley was said to be sparse, with depopulation resulting largely from violent contacts with sandalwood traders and introduced diseases (Gordon, ed. 1863: 137). As is the case on Tanna and elsewhere in Vanuatu, the inland areas of Erromango were likely much more densely populated during the past than they are today. Coastal aggregation of settlement took place later, partly as a result of missionary and other colonial activities.

We excavated a single test-pit at a site near the Williams River called Undam, where a local villager had found a few sherds of red-slipped pottery, probably made around 2,000 years ago (Bedford 2006: 158). The unit yielded an assemblage of shells, fire-cracked rock, and charcoal typical for the later prehistoric sequence of Erromango. While the pre-European assemblage seems poor in material compared to the mission sites discussed below, it should be noted that most of the objects used by the people who lived along the Williams River would have been made of organic materials unlikely to preserve archaeologically (Chapter 5). A 19th-century map of the area of Dillon’s Bay south of the Williams River shows a high density of indigenous place names around where the Gordons settled (Vanuatu National Archives Land Record 46 S.I. 3), indicating the rich cultural landscape within which the missionaries lived.

G. Gordon House Domestic Space and Stratigraphy

The remains of George and Ellen Gordon’s house (referred to as the G. Gordon House below, Figure 2.6) today consist of a long, low wall running perpendicular to the line of the cliff. The site was cleared of vegetation some time in the 1990s and, during this time, the Erromangan people who did the clearing built a small memorial circle of stones at the northern part of the site. Like so many missionary sites, G. Gordon House has been incorporated into the local pantheon of customary sacred places (Flexner and Spriggs 2015), and is periodically visited as such. Apparently some of the surface artefacts were removed and thrown over the cliff during cleaning of the site, or taken as souvenirs by local people. It is also likely that the Gordons dumped their rubbish over the cliff, though reconnaissance of the area below the house has yet to locate any midden deposit.

Figure 2.6 Plan of G. Gordon House.

Source: James Flexner

To better understand the arrangement of domestic space, and to obtain a sample of material culture from G. Gordon House, we excavated a series of test pits ranging from 1x1m to 2x2m in size around the site (see Appendix B for a list of excavation contexts). The excavated area totals 14m2. Across the site, we revealed a single, shallow occupation layer with artefacts dating to the middle of the 19th century. All of the pre-1870s mission houses display this kind of pattern, as they were generally inhabited briefly (less than five years), and abandoned suddenly. Some of the recovered artefacts were likely deposited in the course of everyday life, while others remain from the abandonment deposit as the house was left to decay following the Gordons’ deaths in 1861.

Figure 2.7 Closing plan photograph and stratigraphic profile of TU1/2/3, G. Gordon House. Note that the vertical scale of the profile drawing has been distorted to make the layering clearer.

Source: James Flexner

Three test units (TU1/2/3) were excavated across the northern part of the wall feature that forms the western boundary of G. Gordon House. The stratigraphic sequence from these units consists of sediment accumulated from slope wash during the time after the site was abandoned, with few artefacts found, overlying the main occupation layer, and a red clay subsoil with high percentages of limestone inclusions from the underlying uplifted Pleistocene-age reef (Figure 2.7). Excavations revealed that this stone wall was likely unmortared, and was probably built as a stone retaining wall to maintain a level terrace for the Gordons’ house.

Elsewhere on the site the accumulated sediment is not present, owing to the absence of stone wall features. A unit excavated in the northeastern area of the site (TU4) revealed very shallow deposits and relatively low artefact density, suggesting this area was part of the yard or garden. Significantly, there was no window glass, or indeed glass of any kind in TU4, and relatively few nails, though the quantity of lime mortar was comparable to the other units. The southern and eastern limits of the house are defined by a linear stone feature that was centred in TU5 (see Figure 2.6), and a diffuse linear stone feature uncovered running north–south in TU6 (visible in the north and south profiles; see Figure 2.8). We did not uncover any intact walls or other architectural features. This indicates that G. Gordon House was constructed quickly and with relatively minimal labour input, which is typical of the early New Hebrides mission structures. However, the presence of lime mortar throughout the site and the stone alignments suggests a modest stone footing for the house walls.

Figure 2.8 Stratigraphic profiles, TU6, G. Gordon House.

Source: James Flexner

Figure 2.9 Stratigraphic profiles, TU5, G. Gordon House.

Source: James Flexner

To the north of TU6, the deposits again become very shallow and artefact densities decrease, as was evident in TU7. The overall impression is of a site that was briefly occupied and left a relatively light impact on the environment, just as the Gordons made a light but apparently significant impact on the religious landscape of Erromango. In TU5, the occupation deposit overlay not red clay subsoil as elsewhere in the site, but a transition to mottled soil containing charcoal and a small amount of 19th-century material. Excavation revealed cross-bedded deposits that went down to roughly 30cm below the ground surface, where they overlay the red clay subsoil found elsewhere on the site (Figure 2.9). A few mortar fragments and some charcoal came from this sediment. These deposits are interpreted as possible infilling of this area of the site to create the level terrace on which the Gordons built their house. Thus while the building itself does not appear to represent a major labour investment, the Gordons did apparently spend some time shaping the space upon which the house was built. It is quite likely that they engaged local labourers in doing so (Gordon, ed. 1863: 136–137).

G. Gordon House Artefact Assemblage

A total of 1,651 artefacts were recovered from G. Gordon House, which had the lowest find density of any of the mission houses excavated on Tanna and Erromango (see Appendix C for an overview of recovered finds). The assemblage consists primarily of architectural materials. Lime mortar fragments (3.7kg were recovered across the site) represent a European material type that was produced locally by burning coral and shells, almost certainly with help of indigenous labour. At least 67 per cent of the nails recovered were hand-wrought, with most of the remaining nails too corroded to identify. Seven of these nails appear to have been intentionally bent into hooks and loops to hang clothes or household items, or to run a rope through. Two round wire nails were found, but likely relate to more recent activities around the site as they came from surface deposits.

Window glass occurred in relatively small amounts at G. Gordon House. Just 45g of window glass were recovered, and no excavation unit contained more than 20g. This suggests the house had few windows, which is significant as again it appears that the house was built as simply as possible. Imported materials like window glass would have been an expensive, difficult to transport luxury, and it is likely that most of the openings of G. Gordon House would have been covered with wooden shutters instead. That said, the house was one of the first on Erromango to have window glass of any kind, which may have made it something of a source of curiosity for local people.

The rest of the domestic assemblage consists primarily of small amounts of olive bottle and colourless vessel glass, and refined earthenware ceramics. The olive glass bottle is a nearly ubiquitous object of 19th-century domestic sites, associated generally with alcohol consumption, though also re-used for many other purposes. Notably, G. Gordon House yielded the highest quantity of olive bottle shards of any of the excavated mission houses (though even then not a huge quantity at 120g from 14m2 of excavation units). This suggests that the Gordons at Dillon’s Bay may have consumed more alcohol than their fellow missionaries. It has been suggested that alcohol consumption was one of many class markers viewed negatively in the CMS missions in New Zealand (Middleton 2007). It is not clear whether the same was true of the Presbyterian missions in the New Hebrides. Further, glass bottles were likely kept as water storage containers after their original contents were consumed, especially considering the remoteness of the Gordons from regular capitalist trade networks.

A very small number (N=37) of ceramic sherds were recovered at G. Gordon House. They are primarily the type called ‘Whiteware’, which was a hard-bodied refined earthenware produced from 1820 to the present, though many of the sherds show the ‘bluing’ suggesting a transitional form from the earlier ‘Pearlware’ (see Majewski and O’Brien 1987). The decorated refined earthenware sherds, of which there were 11, are all decorated with blue transfer-printed designs. Identifiable motifs are of the ‘Oriental’ type, including bamboo, Chinese-style bridges, and pagodas (Figure 2.10). The ‘Chinoiserie’ and ‘Willow’ patterns that exemplify the Oriental style are extremely common on mission sites elsewhere in Oceania, and were generally popular worldwide during this period (Lydon 2009: 141–143; Middleton 2008: 202–209; Smith et al. 2012: 58–67, 2014: 30–40). Intriguingly, the only sherd of Chinese porcelain recovered from a mission site on Tanna or Erromango came from G. Gordon House.

Figure 2.10 Decorated ceramics, including transfer-printed whitewares and Chinese porcelain (second from right on bottom row).

Source: James Flexner

One each of clay tobacco pipe bowl and pipe stem fragments were recovered at the house. Ceramic pipes and tobacco were extremely important trade items in the mission setting, as objects valued by Melanesian people but seen by missionaries as less problematic when compared with even more valuable alcohol or firearms. Four slate pencil fragments were recovered. Literacy was a massively important technology, especially for the Calvinist-derived Presbyterians for whom reading scripture was an absolutely necessary Christian practice (see Keane 2007). Their low density at G. Gordon House suggests the Gordons were only beginning to invest time and material in this practice, and they struggled with convincing local people to attend, or allow others to attend lessons (Gordon, ed. 1863: 142).

One of the recovered slate pencil fragments was modified, and local people indicated the form was likely that of a large ‘needle’ used in the manufacture of numplat or grass skirts (see Lawson 2001). In addition, we recovered two sea urchin spine abraders at the site. Other sea urchin spine and coral fragments are likely also intentional manuports considering the distance from the site to the sea. A fine-grained basalt flake and a core uncovered during excavations around the stone retaining wall are likely also markers of indigenous craft and labour around the Gordons’ house (see also Gordon, ed. 1863: 136–137). While small in number, these artefacts are important reminders of the early material entanglements between missionaries and Melanesians. Such objects likely wound up in the G. Gordon house deposits as indigenous people helped with house building activities, or began visiting, trading, and engaging with the missionaries.

In addition, the Gordons would have been almost entirely reliant on local people for their food. Faunal preservation at the site was poor, suggesting acidic soils. Only 25 shell fragments were recovered, but significant food taxa were identified, including Arcidae, Conidae, Nerita sp., Turbo sp., and Tridacna sp. No mammal remains were found aside from a few cattle bones on the surface, which probably post-date habitation of the site. Likely the vast majority of protein for the Gordons came from local marine resources. Likewise, small amounts of flour or rice were probably dietary supplements to the staple yams, taro, and local fruits and vegetables.

G. Gordon House is the earliest mission site on Tanna and Erromango. It represents the initial foothold on the furthest frontier of global Christianity in the late 1850s. Unsurprisingly, then, the site has a relatively small archaeological footprint. The household deposits are shallow, and the artefact density relatively diffuse compared with the other sites. There are some valuable but fleeting hints at interactions with local people. While the overall amount of imported material at the site is small, it does contain a few objects that would have been particularly valued, such as the Chinese porcelain vessel. Limited investment in the physical mission house may also relate to the fact the George Gordon spent much of his time itinerating, visiting other areas of Erromango as he gauged possible interest in missionary presence among various communities (Gordon, ed. 1863: 138–139). Perhaps if he had found a more willing community than the one at Dillon’s Bay, the Gordons would have moved to another place, in which case there might have been some sense that the house was to be temporary anyway.

The Second Martyrdom

The Gordons’ work on Erromango was cut short abruptly on 20 May 1861. A group of warriors from Bungkil and Unepang to the south came to the mission house asking for Mr Gordon. They claimed there was a sick man and they wanted his help treating him. Gordon accompanied the men, who ambushed him on the road, killing the missionary as he attempted to flee. The men then returned to the mission house and killed Mrs Gordon in turn. After their deaths, the house was abandoned and the Erromangan converts fled the island for Aneityum, fearing they too would be the targets of violence (Gordon, ed. 1863: 182–183; Robertson 1902: 73–77).

Like the earlier martyrdom, the Gordons’ deaths left an ‘event landscape’ (Flexner 2014c) that cements the social memory of this moment in place (Figure 2.11). Walking on the trail towards Bongkil, there is a boulder outcrop with a slight discoloration (point 2 in Figure 2.11). This is said to be George Gordon’s ‘bloody footprint’, and significantly, it’s thought to be a reference to Matthew 16:18: ‘on this rock I will build my church’. Thus Gordon in his moment of death is thought to have made a statement of the future of the church on Erromango (see also Robertson 1902: 126). Further south, there is a natural stone basin that collects rainwater, where the warriors who killed the Gordons are said to have washed their weapons while returning home (point 3 in Figure 2.11).

Figure 2.11 Event landscape covering the deaths of George and Ellen Gordon (point 1: G. Gordon Mission house; point 4: ‘guard’s house’, apparently belonging to a native guard meant to protect the house; point 5: Vedavil Stream, the nearest water source; see text for explanation of features 2 and 3).

Source: James Flexner

Unlike Williams and Harris, the Gordons had a relatively long residence on Erromango, learned enough of the language and kastom of the island to survive, and had even managed to win a few converts. So why were they suddenly attacked and killed? The answer to this question is something of a refrain in the story of early missions on Erromango and Tanna (see Adams 1984: 64–65). Before the attack, there had been a number of destructive tropical storms, and more significantly repeated outbreaks of introduced disease, mostly measles, some of them extremely severe. The native population was dying in unprecedented numbers, which coincided with the arrival of Europeans, and the missionaries were especially close with local populations. George Gordon was even believed to have poisoned an Erromangan chief when he visited the dying man to administer medicine (Gordon, ed. 1863: 184–203).

As mentioned above, there was no natural cause of death in Erromangan kastom. It was the work of Tavuwa, powerful sorcerers. When missionaries arrived, and people began dying in a way they never had before, the association was more than most people could overlook. The cosmology of the missionaries did not help their case. Germ theory was still decades from being invented, let alone gaining acceptance in European communities. For the missionaries, disease was caused by bad air or humoural imbalances, but whether people got sick, lived, or died was completely in God’s hands. In expressing this to Erromangan people, the missionaries cast themselves as a new kind of Tavuwa, with God as another, more dangerous natemas (spirit) than had existed on the island before. When disaster struck, it was only natural that some Erromangan people blamed the missionaries and acted to protect their island.

Figure 2.12 Burned bottle glass shards from G. Gordon House.

Source: James Flexner

There is a final archaeological legacy of this story. Of 256 glass artefacts recovered from the site, 68 per cent (N=175) showed evidence of burning (Figure 2.12), as do a few of the ceramic artefacts. When James Gordon visited the site a few years after his brother’s death, he found the house was no longer standing (Robertson 1902: 126). The burned artefacts indicate that the G. Gordon House may have been burned shortly after the Gordons were killed. Perhaps this was meant to cleanse the site of any missionary natemas that may have remained. Yet the event still remained heavy on the hearts of Erromangans, and in 2012 another reconciliation ceremony was held, this time with Gordon family descendants.

James Gordon at Potnuma

George’s brother James would follow to the New Hebrides, settling on Erromango in 1864 (Robertson 1902: 124–130). In the first part of James Gordon’s tenure on Erromango, he focused on the western side of the island. Land records indicate that he purchased large tracts on behalf of the Presbyterian Church at Dillon’s Bay and Bongkil from the trader Andrew Henry in 1865 (Vanuatu National Archives Land Record 33 S.I. 10, 34 S.I. 2). Gordon sought to continue his brother’s work, protecting converts and looking for new areas with good potential as mission stations, not only on Erromango, but also further north, particularly on Espiritu Santo (Robertson 1902: 135, 141–142).

Figure 2.13 Mission building, Cook’s Bay.

Source: James Flexner

Another Canadian missionary couple, James Macnair and his wife, joined James Gordon in 1867. Macnair helped to expand the mission field on Erromango, establishing a station at Cook’s Bay, specifically near the harbour at a place called Unōva. Lifu Nokilian sold the land to Macnair. The young chief Potnilo was instrumental in the establishment of this settlement according to local traditions. While it wasn’t a major focus of mission activity, a small church with stone footings was built at Unōva (Figure 2.13). A stone-lined grave in the center of the church is said to hold the remains of Narainabuo, the first male convert from Cook’s Bay, along with those of Neri, the first female convert. Macnair died of illness, probably malaria, in 1870. His wife left the New Hebrides and later married George Turner, an important LMS missionary in Samoa (Robertson 1902: 132–135).

When Macnair settled on the island, James Gordon shifted his focus and primary mission station to the more populous eastern side in 1868, settling at Potnuma, just north of the large village in Port Narvin (Robertson 1902: 134). Potnilo, who may have been the district Fan lo of Roviliau considering the extent of his influence, was also key to allowing Gordon to settle at Potnuma. Novolu and Netai were likewise early allies of the mission cause. However, the fact that Gordon settled on the border between the lo of Roviliau and Numpunaraipau may have made his situation a somewhat precarious one. James Gordon became the last martyr missionary of Erromango in 1872. Here we focus once more on the patterns of everyday life around the mission house and in the surrounding Melanesian communities rather than the final moments and their ongoing presence in the landscape, though these will also be discussed below.

Around Potnuma

Before turning to the mission house itself, there are a few significant features in the surrounding area (Figure 2.14). Rock art is an important feature of the archaeological landscapes on Erromango (Wilson 1999, 2002). In the area between Port Narvin and Potnuma, there are several major rock art sites, including a cave site called Netngonavon with painted motifs, and two coastal sites with petroglyph fields pecked into volcanic rock outcrops, Malap and Bomtal. Motifs from these sites were initially recorded by Wilson (2002: 24–25), and the petroglyph fields were more completely mapped for this project (Flexner 2013: 21, 2014c: 12).

Malap displays classic motifs of ‘contact’ rock art (Figure 2.15). These include European sailing ships, bottles, an axe, and what appear to be muskets (they might also be stylised fish). The documented petroglyphs also include a single serif letter ‘T’. The petroglyphs continue below the current low tide line, suggesting that the sea level may be rising on this side of Erromango. Alongside the introduced motifs are stone cupules, which were used as a counting device, and pecked circles, which were male and female symbols. Bomtal is primarily a cupule site, though a possible sailing ship is present (Figure 2.16). Significantly, the site is associated with a local oral tradition about Sou Sou, an early Christian convert who apparently defended his faith with a club (each cupule is said to represent a battle he fought; Flexner 2014c: 11–12).

Figure 2.14 Archaeological features in and around Potnuma.

Source: James Flexner

Figure 2.15 Petroglyphs at Malap.

Source: James Flexner

These sites have been interpreted as reflections of ‘social upheaval’ during the contact era: ‘[K]een resistance to social transformation may have prompted a resurgence in the production of traditional symbols, as a way of maintaining a former social order. Acceptance of incoming ideas, in contrast, may have seen the incorporation of new motifs into the rock-art record’ (Wilson 2002: 140). Not surprisingly, what is apparent is not an either/or tendency towards purely traditional and purely introduced motifs, but overlap and blending of the two. Traditional patterns of cupules and concentric ovoids form the structure within which European things are placed.

In the other direction, to the north of Potnuma, there are several kastom sites that were recorded during archaeological survey of the area (Figure 2.14). At the boundary between the two lo, at a place called Potnepko, lie two natural tubes in the uplifted limestone that were used as places to dispose of the dead. The inland tube was used specifically for sacrificial victims and to ‘block’ travel on the road along the coast (presumably during times of war, because of fears about natemas). Further along is a stone with natural impressions said to be the footsteps of Vetemanu, a ‘giant’ who was killed and his head sent to Goat Island (a small islet off Traitor’s Head). A rock with numerous egg-shaped ground cupules called Nduwe is said to have been used to increase the rate at which chickens laid eggs (a classic case of ‘sympathetic magic’; Frazer 1922: 14–62). Finally, there is a small rockshelter with a stone wall enclosure in the front said to be where Nialnowre, the first woman to keep pigs, lived at Ntue. This landscape of storied places would also have likely been densely inhabited, though the population had decreased dramatically by the time James Gordon arrived at Potnuma.

Figure 2.16 Petroglyphs at Bomtal associated with oral traditions about Sou Sou.

Source: James Flexner

J. Gordon House Domestic Space and Stratigraphy

When James Gordon chose to relocate to the east coast of Erromango, there is some evidence that he was becoming increasingly withdrawn from the main hierarchy of the Presbyterian Church (Church of the Lower Provinces of British North America 1869–1871). He built a compound consisting of a two-storey mission house and a mission church, both of which had substantial limestone footings and walls. Of the church, the walls on the eastern side, which was where the sanctuary was located, are particularly well-preserved (Figure 2.17). The lime mortar is a European material, but produced locally. Construction of mission buildings almost certainly involved significant inputs of local labour, which would have been organised by the local chiefs.

Figure 2.17 Plan map of James Gordon’s church at Potnuma.

Source: James Flexner

The mission house (called J. Gordon House below) had lower footings of lime mortar and probably a timber upper storey (Figure 2.18). There was a scattering of surface artefacts at the site, primarily refined earthenware dishes but also some glass (Figure 2.19). Three test units ranging in size from 1x1m to 2x2m were excavated around the site. The excavations revealed the extent of the house, which is estimated as covering an area of 10x5m (32x16ft). The remains of the lime mortar walls are mere stumps, but the openings in the front of the feature suggest the front (sea-facing) wall contained a central front door with two windows to either side.

Figure 2.18 Plan of J. Gordon House.

Source: James Flexner

Figure 2.19 Surface artefacts from J. Gordon House.

Source: James Flexner

Three test units covering a total of 7m2 were excavated at J. Gordon House. Stratigraphy across the site is uniform, with a single, shallow layer containing the majority of the artefacts. There is a large population of burrowing crabs on the site, and their burrows have caused a great deal of vertical displacement of materials. In the front of the house, TU2 showed that the lime mortar footings rested directly on the clay soil of the site (Figure 2.20). TU1 was located directly along the southern part of the western wall of the house. This unit showed the shallow stratigraphy, and contained refuse possibly disposed through the rear of the house (Figure 2.21). Further west, TU3 was located in the shallow midden deposit. In each case, a single, slightly darker layer containing materials from the short habitation of the site overlay the natural brown clay of the coastal flat.

Figure 2.20 Stratigraphic profiles, TU1, J. Gordon House.

Source: James Flexner

Figure 2.21 Stratigraphic profiles, TU2, J. Gordon House.

Source: James Flexner

J. Gordon House Artefact Assemblage

Artefact density at J. Gordon House was roughly double that of G. Gordon house at Dillon’s Bay (N=1753 from 7m2 vs N=1651 for 14m2). The assemblage was also a much richer one, carrying a wider variety of imported material culture. As the Erromango Mission was better established and more secure, more materials were brought in by the Presbyterian Church. Erromangans may have read the increasing material affluence of the mission in positive terms. This was not just a matter of increased trading opportunities, but also a means of expanding prestige in Melanesian terms, as chiefly converts had access to a material world not so easily available for rival Fan lo who avoided the mission.

Despite the surface evidence for mortared stone architecture, the amount of lime mortar at J. Gordon House was quite small compared to other Erromango sites (1.3kg versus 3.7kg at G. Gordon House). Possibly this relates to preservation conditions at the site. The compact clay soils and rainy environment, combined with ongoing bioturbation means that softer materials, like mortar, are less likely to survive. Other architectural materials were abundant. Hand-wrought nails were still the most common identifiable type (N=70, or 53 per cent of the total nail count), but there were a few machine-made cut (N=5) and round wire (N=15) nails as well. Window glass was abundant on the site, with a total of 265.7g recovered, the largest amount from any of the Erromango Mission houses. A few shards of window glass (N=5) had linear or cruciform discolourations etched into the glass, which may be evidence for screens that have since disintegrated (Figure 2.22a, b, c).

Figure 2.22 Glass artefacts from J. Gordon House.

Source: James Flexner

J. Gordon House yielded a number of other glass finds. Perhaps most remarkable were 45 shards of hand-blown colourless glass vials, as well as one complete example (Figure 2.22k). These vials likely came as part of a medical kit. Also of note is a frosted glass stopper (Figure 2.22n), which probably came from a perfume bottle. Colourless drinking glass fragments (Figure 2.22f, g) included etched stemware fragments that may have had a vine motif, probably indicating alcohol consumption. However, there was relatively little olive glass, at just over half the density found at G. Gordon House (34.6g from 7m2 versus 120.2g from 14m2). This suggests that James Gordon may have primarily consumed alcohol as part of European social rituals when hosting white visitors.

Refined earthenware ceramics also relate to European dining rituals. The most common type found at J. Gordon House was Berlin Swirl Ironstone (Figure 2.19, 2.23), produced by Liddle Elliot & Son in Staffordshire around 1864 (Dieringer and Dieringer 2001: 66). Berlin Swirl was sent to Australasia in large quantities when commercial markets were disrupted by the American Civil War (Brooks 2005: 57–60). There was a single sherd of a transfer-printed vessel with an Oriental motif on the surface. The other transfer-printed vessel (Figure 2.24) was identified as ‘Minerva’ pattern, which was produced by Podmore, Walker & Co., an English Staffordshire pottery, between 1834 and 1859 (Godden 1991: 501; Williams and Weber 1978: 72). The pattern featured the Roman goddess of wisdom seated alongside classical urns in the centre, with urns and columns in cartouches around the rim. Another moulded refined earthenware plate with grape vine motifs around the rim is likely a ‘cheese plate’ form, with a radiating ribbed surface to allow the oil of a cheese to drip down. These kinds of plates would have been important for Victorian dining rituals. The Minerva plate specifically may have served a further purpose as a storytelling object when describing deeper European traditions to local people.

The site yielded one sherd from a vessel bearing a currently unidentified transfer pattern signified by a patent mark. There were also sherds from a porcelain measuring cup. This object reflects the increasing standardisation of everyday life in the industrial era, a pattern that apparently extended all the way to the remote corners of the New Hebrides. The final ceramic vessel identified at the site is represented by 11 sherds of a red-slipped coarse earthenware recovered from TU2. Indigenous ceramic production ceased on Erromango around 2000BP (Bedford 2006: 158). There is evidence that on his last trip to Espiritu Santo Island in 1869 James Gordon was trading for curiosities, using the Erromangan convert Novolu as an indigenous middleman to get better prices (Robertson 1902: 141). It is reasonable, then, to interpret this object as a souvenir collected by Gordon on one of his trips north. Petrographic analysis of the red-slipped vessel indicates not Santo, but Ambae, a neighbouring island some 400km north of Potnuma (Dickinson 2014).

Evidence for more local interactions with Melanesian people were less clear in the artefact assemblage from J. Gordon House, which lacks the coral or sea urchin spines common on other sites. We recovered 17 slate pencil fragments from TU1 and TU2 (Figure 2.25). These objects ranged in diameter from 0.46cm to 0.58cm. Pencils from the two different test units in which they were found have different diameters, further bolstering the argument that teaching occurred throughout the domestic space at Potnuma over time. The pencils were grey to dark grey in colour. The fragments are generally quite small, averaging just 1.7g each. It is quite likely that most of these fragments were broken by Erromangan students in the process of learning to write, including both unintentional breakage and regular sharpening as the pencil tips dulled from repeated use. Slate pencils were crucial technological implements for mission work, as literacy was one of the necessary disciplines required of Christian converts.

Figure 2.23 Ceramic manufacturers’ marks and other markings, J. Gordon House.

Source: James Flexner

Figure 2.24 ‘Minerva’ transfer-pattern ceramics, J. Gordon House.

Source: James Flexner

Figure 2.25 Slate pencil fragments, J. Gordon House.

Source: James Flexner

The faunal assemblage at J. Gordon House probably suffers from poor preservation conditions. Of the shell remains recovered, the primary taxa were Turbo sp. and Nerita sp. As elsewhere on Erromango, James Gordon would likely have relied on local fish and shellfish as his primary protein source at Potnuma. The ceramic dishes noted above also likely held yams or bananas more often than rice or bread. The only mammal remains found at J. Gordon House consist of a single, isolated human molar recovered in TU1 (PN34). This is interpreted as possible evidence of historical dentistry work, though whether the molar is Gordon’s or a Melanesian convert’s is a matter for future research to discover. Certainly missionaries regularly used their medical ‘expertise’ as a way of engaging with local people, though the state of medical practice was fairly primitive before formally trained medical missionaries began arriving in the 1880s (Miller 1986: 16–25).

Erromango’s Final Martyr

Despite his considerable success in winning converts, and greater access to imported things, James Gordon became the last missionary to suffer a martyr’s fate on Erromango (ironically, he published a biography of his brother and sister-in-law called The Last Martyrs of Eromanga shortly before departing Canada). On 7 March 1872, Gordon was ambushed in his house by Nerimpau and Naré, men from a nearby village. As noted above, the Cook’s Bay people were much more sympathetic to the mission cause, so Gordon’s presence at Potnuma for nearly four years was always a threatened one, as noted after the fact by Yomot, one of Gordon’s supporters. The event set off a series of revenge killings, following in some ways existing patterns in Erromangan warfare. Notably, the killer Nerimpau was tracked and killed (Robertson 1902: 156–163). He was beheaded, and his body hung from a tree that still stands today (Figure 2.26). The tree is still tabu, considered to be poisoned because of Nerimpau’s evil deeds and not to be touched.

Figure 2.26 Tabu tree where Nerimpau’s body was hung.

Source: James Flexner

Missionary accounts suggest that the motivation for the killing was simply hatred of missionaries and the mission cause. However, local social memories suggest that Gordon had accidentally brought poisonous reef fish to a local feast, and that he was killed for causing illness among the local population (another possible example of a missionary being identified as a dangerous sorcerer). It was also apparently rumoured that Gordon was keeping a large amount of gold in the house, a tantalising hint at an early version of later beliefs that associated capitalist wealth also with foreign sorcery or magic (e.g. Lindstrom 1993; Taylor 2016).

Erromangan people marked James Gordon’s death in several ways. Gordon died in the front of his house. An upright stone was later erected to mark the spot where his blood had apparently run, over 25m away towards the sea. To the southwest, local people also constructed a memorial grave enclosure of lime mortar and stone (Figure 2.27). Significantly, this monument takes the form of a sea turtle. Turtles were both male symbols in Erromangan cosmology, and known for making long sea voyages. The head of the turtle faces towards the northwest, which is in the general direction of the Presbyterian homeland in Scotland, and Gordon’s homeland in Canada’s Maritime Provinces. The rectangular concrete slab over the grave marker, which appears to be a later addition, integrates a headstone flanked with a brown glass jar and green glass bottle, and a ceramic water filter with the mark ‘[A]ND SONS PATENT LIMESTONE FILTER’, which was probably salvaged from J. Gordon House and repurposed as a flower urn for the memorial. This resting place shows that missionaries had begun to be integrated into Melanesian societies. The objects are foreign, but the overall form of the memorial uses distinctively local references. The process is one of Melanesians integrating European people, things, and religious beliefs into their world, rather than the other way around.

Figure 2.27 Sea turtle-shaped memorial to James Gordon, Potnuma.

Source: James Flexner

The Robertsons at Dillon’s Bay

Hugh Angus Robertson and his wife Christina settled at Dillon’s Bay mere weeks after James Gordon had been killed on the other side of the island. Robertson was another Nova Scotian missionary, who had initially sailed with James Gordon to the New Hebrides on the Dayspring in 1863, arriving in the islands in 1864. Like James Harris before him, Robertson was travelling for health reasons as warm climate and fresh air were thought to improve the constitution (Robertson would have infinitely more luck than Harris in the New Hebrides). Inspired by what he saw on Aneityum, Robertson determined to study to be a missionary when he returned to Nova Scotia. He was ordained and licensed, and returned to take up the assignment on Erromango in 1872 (Robertson 1902).

Upon arriving in Dillon’s Bay, the Robertsons set about resuming mission work in the area, having some reason to be optimistic despite James Gordon’s recent death. However, the situation on the island was far from stable. The Robertsons would themselves narrowly escape possible attacks on several occasions (Robertson 1902: 195–196, 315–318). The Robertsons lived on Erromango for over four decades, and managed to convert large portions of the population. The archaeology of their mission station reflects material processes of labour and settlement patterns that were instrumental to that apparent success.

The Robertson Mission

Unlike previous missionaries on Erromango, the Robertsons built their mission in the middle of the indigenous village at Dillon’s Bay. Their house became a major attraction for parties of Erromangan visitors almost immediately. In January 1873, the house was blown down in a hurricane, an event that would have been interpreted in terms of supernatural causes. Indeed, Robertson notes that he was warned to keep doors closed for fear of attack particularly after the storm. Robertson embarked immediately on a project of rebuilding the house, which he describes as 62x28ft (19x9m) with a 6ft (2m) veranda. The house had wooden walls, plastered on the interior, and a thatched roof. Yomot, an early friend of the mission, ‘worked splendidly’ on the construction of the house (Robertson 1902: 193–208).

The archaeological remains of the mission house (called Robertson House below) consist of the large stone and mortar foundation (Figure 2.28). Excavation of the area immediately in front of the front step (TU1) revealed a foundation measuring 140cm from top to bottom (Figure 2.29), which is extremely deep for a single-storey house, even on the sandy soil by the coast in Dillon’s Bay. In fact, it appears that Robertson engaged local people for major construction projects of all kinds, building a storehouse for his materials, and eventually a large timber church, called the Martyr’s Memorial Church, in 1880. This was a prefabricated timber church built from a kit from Australia, the first in the New Hebrides (Robertson 1902: 321–326). Around these buildings, Robertson had native labourers drag boulders from the beaches of Dillon’s Bay to construct a series of stone walls outlining the paths around the village (Figure 2.30).

Robertson’s ability to call on large labour parties is not simply a reflection of the increasing power of the mission on Erromango in the later part of the 19th century. Rather, Robertson’s organisation of labour parties may have been a causal factor in his long-term ability to continue mission work and win converts. The fact that Robertson had local people working for him while managing to avoid illness and death that might have been wished upon him by the Tavuwa would have offered a powerful sign about the efficacy of his spiritual prowess, and that of his natemas, Iesu, the Christian God.

Figure 2.28 Plan of Robertson House, Dillon’s Bay.

Source: James Flexner

Figure 2.29 Stratigraphic profiles, TU1, Robertson House.

Source: James Flexner

Figure 2.30 Map of stone walls in the Robertson Mission compound.

Source: James Flexner

In addition to excavating immediately next to the house, we excavated test units in the backyard area, and on the sea side (Figure 2.28). Three adjacent units (TU2/5/6) were excavated 10m west of the house foundation (Figure 2.31). The stratigraphic sequence of the southern unit (TU5) consisted of a thin topsoil layer overlying a dark brown to black sand with coral limestone pebbles, suggesting a recent pavement in the area, possibly a path leading to the house. This overlay a layer of volcanic cobbles and boulders, which in turn covered a layer of compact reddish-brown clay loam. This layer is not present to the north in the adjacent units (TU2/6). In these units, there was another layer with coral inclusions underlying the boulders (Figure 2.31).

Figure 2.31 Stratigraphic profiles, TU2/5/6, Robertson House.

Source: James Flexner

The sediment underlying the archaeological deposits in TU1 was a clean black beach sand. We excavated TU6 to a much greater depth than the other units, revealing a stratigraphic sequence that alternated between deposits of sandy loam, thin lenses with coral inclusions (thought to be paving layers), and deposits of cobbles and boulders (Figure 2.31). Excavations were halted approximately 1m below the surface in a layer of very dark brown sand with cobble inclusions. The last excavated level included a piece of iron strap. It appears that the annual tropical storms that flood the river regularly deposit sediment in this area, including sometimes larger cobbles and boulders. Further, there appears to be a good deal of downward movement of materials, considering the great depth at which the metal was found (a radiocarbon sample from the same level returned a 19th-century date; see Appendix D for radiocarbon dates).

Behind Robertson House, archaeological excavations revealed significant features relating to everyday occupation of the site (Figure 2.32). Removal of the topsoil revealed two ovoid features filled with stone rubble, shells, and animal bones. Excavation revealed that the features are almost certainly earth ovens, likely dating to the later part of the Robertsons’ time in Dillon’s Bay. Robertson was apparently a great admirer of Erromangan cuisine, which he describes at length in his autobiography (Robertson 1902: 379–381), so it is perhaps unsurprising that some earth ovens were discovered behind his house.

Figure 2.32 Plan and cross-section of earth oven features and stone alignment, TU4, Robertson House.

Source: James Flexner

Initially, the Robertsons had trouble obtaining local foods (Robertson 1902: 208–209), but once established on the island would have included quite a lot of Melanesian ingredients and dishes in their diet. Faunal remains from the Robertson House were abundant, and again indicate heavy use of local shellfish, including Turbo sp., Trochidae, Conidae, Tridacna, Arcidae, and Nerita sp. Robertson House was also the only Erromango Mission site to yield pig bones, another indicator of the prestige the missionaries were able to achieve on the island.

The expansion of steamship voyaging in the Pacific (Steel 2011), and the regular voyages of the mission ship Dayspring to the New Hebrides from the 1870s onwards, would have made it easier to bring imported goods to the mission at Dillon’s Bay. One result of the importance of the mission settlement to the Erromangan landscape was population aggregation, as people from villages with declining populations, especially inland, began to settle along the coast at the more populous mission stations. An unfortunate archaeological by-product of this is that many of the later missions are now large villages. What this means is that the archaeological records of these sites have been significantly impacted by contemporary activities.

Robertson House itself is partly overlain by a medical clinic built within the last 20 years. The artefacts recovered around the upper layers of the site likewise include various contemporary materials (plastic, automotive glass, glass marbles, a 1979 Nouvelles Hébrides Franc coin). Despite this, we did recover a small number of artefacts that could be definitively associated with the Robertson Mission. Three sherds of transfer-printed whiteware with a green floral motif were recovered, which represent a single vessel. In TU1, we recovered two 19th-century coins, one was an 1872 French five Centime piece, and the other was a British ‘Victoria Bun’ penny, though the date couldn’t be read as the reverse had worn down (Figure 2.33). Three white clay tobacco pipe bowl fragments were found. Buttons, glass paste jewels, and a copper alloy thimble found around the house reflect changing habits of dress in the missionary era. The female domestic activity of sewing was an important habit cultivated among converts by missionary women.

Figure 2.33 British (left) and French (right) coins from TU1, Robertson House.

Source: James Flexner

Overall, though, surprisingly little definite 19th-century material was found in excavations around Robertson House, particularly glass artefacts. In part, this reflects a need for further extensive excavations around the site. However, sampling issues do not suffice to fully explain this pattern. It is possible that a discrete rubbish disposal area has yet to be discovered. Further, it is likely that rubbish has been dispersed over time because of various activities and other post-depositional processes in Dillon’s Bay that have disturbed the archaeological deposits. Documentary evidence indicates that the Robertsons did have access to and bring in large amounts of imported material (Robertson 1902), but it appears mostly to consist of materials that would not preserve archaeologically anyway, such as cloth, flour, or wooden furniture. In addition, like other ‘successful’ missionaries, the Robertsons likely took much of their domestic assemblage with them when they retired from the mission field.

Despite the apparent dearth of artefacts, the Robertsons had a huge material impact on the landscape of Dillon’s Bay. The Martyr’s Memorial Church stood for over a century, before being destroyed in Cyclone Uma in 1987. The plaques from the church and salvageable timbers were recycled into a smaller church building that later served as a schoolhouse. The remains of the stone walls and the massive foundations of Robertson House are tangible reminders of the amount of indigenous labour used to divide the ‘ordered’ space of the mission from the surrounding ‘disordered’ Melanesian landscape (Flexner 2013: 16). After the Robertsons, Erromango was a Christian Island. Christianity, however, did not constitute a negation of kastom. In TU1 at Robertson House, burned fragments of a carved coconut shell armband hint at the possibility of continuity in native adornment (see also Lawson 2001), though why such an object was burned is a matter for speculation. The discussion of continuity and change in Erromangan material culture will be taken up further in Chapter 5.

Previous Next