War shields of the Torricelli Mountains, west sepik province, Papua New Guinea.

Craig, Barry ; Lewis, Gilbert ; Mitchell, William E. 等

INTRODUCTION

Researchers at Field Museum, Chicago, have analysed the distribution of over 6,000 ethnographic objects collected mainly by A.B. Lewis during 1909-10 along the north coast of New Guinea, from Jayapura (West Papua) to Madang (Papua New Guinea) (see Terrell & Welsch 1990; Welsch 1996; Welsch & Terrell 1994, 1998; Welsch, Terrell & Nadolski 1992). The geographic region examined by the Chicago team's project is relatively uniform, being primarily coastal with small nearby islands and hills hinterland. Although linguistically diverse with 55 languages belonging to 5 unrelated language phyla along 700 kilometres of coastline, the people share a more or less homogeneous material culture complex, having a common pool of resources, material products, and cultural practices.

The focus of the Chicago team's analysis was on how trade and inherited friendship networks contributed to the interdependence of people on a far wider scale than one might suppose given the linguistic diversity of these societies. They have stated (Welsch, Terrell & Nadolski 1992: 587):

 We question the value of parsing New Guinea's people into
 'ethnolinguistic' groups, as if differences in language
 automatically translate into differences in culture. Simply knowing
 that two neighbouring communities speak unrelated languages does
 not allow us to assume that there are important and meaningful
 differences in their respective cultures.

Until the Chicago study, information about the extent of the correlation of material cultural and linguistic differences has been haphazard and scattered throughout the literature, with examples suggesting both high and low correlation. The Chicago study is therefore a welcome introduction of rigour into study of the subject. But there have been critics questioning how the data were analysed and, in particular, the statistical methods used (Moore & Romney 1994, 1995, 1996; Roberts, Moore & Romney 1995). A more recent study based at the South Australian Museum (the Upper Sepik-Central New Guinea Project--see http://uscngp.sai.net.au) is in the process of testing the conclusions reached by the Chicago team utilising data from some 10,000 objects collected in the upper Sepik and central New Guinea regions but results are not yet available.

A recent survey of Melanesian war shields (Beran & Craig 2005) brings together all the types of shields then known to the authors and editors. Implicit in that survey is the question of the correlation of material culture and language, so the locations of all the shields covered by the survey are linguisticly identified. A typology of shields based on formal characteristics is presented (ibid.:19-25) and the distribution of the types indicated (ibid., Map 2). But the shields were not examined in sufficient numbers or detail to contribute substantially to the debate raised by the Chicago study. What does seem to be emerging is that conclusions reached from the results of the Chicago study, questionable on the grounds of analytical categories and statistical method, may also turn out to be questionable on the grounds of the location and nature of the cultures examined. What might prove true for communities linked lineally by a vigorous maritime trade network might not be true for land-based communities scattered across two, rather than one, geographical dimensions. Further, the detail of material culture difference is certain to be crucial--differences obscured by analysis of categories defined too broadly (eg. particular types of string bag looping techniques obscured by analysis based on the presence or absence of looped string bags).

This paper, reviewing examples of certain war shields in the collections of the South Australian Museum, the Australian Museum, the PNG National Museum, the Museum der Kulturen in Basel and the American Museum of Natural History in New York, with reference to fieldwork in the Torricelli Mountains by all three authors, is an admittedly anecdotal contribution to the above debate. Certain formal and graphic design details of a number of well-provenanced war shields of the Lumi-Anguganak area of the Torricelli Mountains (see map, Fig. 1) will be examined and plotted against the distribution of languages in the region.

We propose an addition to the four types of Torricelli shields as reported in Beran & Craig (2005: 38-42), report on a looped string protective device worn by warriors in the vicinity of Anguganak, and describe a type of pigskin war shield to the south-east of Anguganak. Ethnographic data relating to the use of the shields are provided.

It should be noted that our considerations are only of the shields that we are aware of; there may be examples of shields from among the One speakers north-west of the Olo speakers, and from among the Elkei, Au, Yil and Alu speakers to the south-east of the Olo speakers, of which we are unaware.

THE TORRICELLI REGION

The Torricelli Mountains are inhabited by populations speaking many different languages, mainly of the Torricelli Phylum, with a few Sepik languages located at the south-east end of the range (Table 1). (2) The largest language group is Olo, with speakers numbering over 10,000, called the Pai on the northern fall of the Torricelli Mountains, and the Wape (3) on the southern fall. There are almost 5000 Urat speakers and 4000 Au speakers, but most of the other languages are spoken by only a few hundred people each (Laycock 1973).

[FIGURE 1 OMITTED]

The region is characterized by a highly dissected landscape rising abruptly from the north coast to about 1900 metres, then falling away more gradually towards the plains and swamps of the upper Sepik to the south. Local and provincial government services and Christian missions are mostly based around the airstrip at Lumi, located near the western end of the road running from the East Sepik Province capital at Wewak, through Maprik in the Prince Alexander Mountains and then through Nuku and Anguganak in the Torricelli Mountains. Only two of the Torricelli groups have been intensively studied by anthropologists (for Gnau speakers, see Lewis 1975, 1980, 2000; for Olo speakers, see Mitchell 1978, 1988, 1990, 2004, and McGregor 1982 for a missionary account).

TORRICELLI SHIELD TYPES

Craig (Beran & Craig 2005:38-42 and Figs 3.14 to 3.17) identified four types of shields (4) in the Torricelli Mountains. The first, third and fourth types need not detain us here. (5) The second of the types identified by Craig is the well-known 'Lumi' shield, hereafter referred to as the Olo shield type as it is made, so far as we know, only by Olo speakers. A newly recognized fifth type is represented by several examples from Rauit and Wamil, to the south-east of Anguganak, and is hereafter referred to as the Gnau shield type.

THE OLO SHIELD TYPE

Perhaps the earliest notice of the Olo shield type was by E.A. Briggs (6) whose expedition from the University of Sydney was the first to collect artefacts, and to record cultural activities by photographs and cine film, among the Olo speakers of the Torricelli Mountains. (7) In November 2004, in the course of doing other research in the collections of the Australian Museum, Craig came across two shields from the Torricelli Mountain, which had been acquired from the University of Sydney (via the Institute of Anatomy in Canberra) with the provenance recorded as 'Epiyu, Wapi, New Guinea' (Figs 2, 3). He realised that these two shields would most likely have come from Briggs; it is highly unlikely that anyone else from the University of Sydney would have collected. artefacts from Epiyu. (8) Epiyu was one of the villages Briggs visited and where he stayed overnight on 15 and 16 January 1926. One of Briggs's photographs held by the National Geographic Society shows two shields (reproduced in Benitez & Barbier 2000:182--see Figs 4, 5). (9) In his over-dramatised, even derogatory, published account of the expedition, Briggs wrote (1928: 266):

 These primitive savages of the Sepik Valley arm themselves with
 long bows and bone-tipped arrows, and carry large wooden shields
 that cover the body from head to knees. The outer surface of the
 shield is deeply incised with a curious spiral design, which also
 appears in various forms in all their carvings and crude paintings.

The first to identify the Olo shield as a type was Frank Tiesler in his 1970 survey of Tragbandschilde ('carrying-strap shields') of northern New Guinea. Unfortunately, none of the six shields he illustrates (1970: 201-2, Abb. 36-41) is located to village; all are in the collections of the Rijksmuseum voor Volkenkunde in Leiden and are simply described as from the Lumi area. They were purchased in 1967 by Adrian Gerbrands in the town of Lae (Morobe Province) without specific provenance (ibid.: 188).

The Olo shield is called paraku (sing.), parakel (pl.). It is roughly oblong in shape, usually with curved edges and described by some as 'bean-shaped', (10) with a central, vertically-oriented and generally symmetrical design carved as incised lines; paint is rarely used. The design, with some variation, consists of vertical parallel lines with a pair of spirals at each end, and often a pair of spirals at the centre. Although many researchers have asked about the 'meaning' of the designs on these shields, informants have insisted that there is no meaning. Mitchell, who was based at Taute (seven kilometres south-west of Lumi), interviewed old men who had made and carried such shields as young warriors and was never able to elicit a meaning. His main informant, when pressed, said 'the design is just to make it look nice'. Craig asked for meanings in 2002 but also was unsuccessful.

Stone adzes would have been used to cut and shape the plank from the buttress root of a tree and the design carved with a small sharp tool such as a hand-held flake, bone chisel or boar tusk (although this has not been confirmed).

Mitchell was told that shields were stored under the house thatch where they developed a rich black patina from the smoky hearth fires; this would also protect them from insect damage and rot and, being very dry, make them lighter to carry. The carrying device is a horizontally-aligned bast strap so the shield can be hung from the shoulder, thus freeing the warrior's hands to use his bow and arrows. This strap is secured to two pairs of closely-spaced holes near each vertical edge of the shield; sometimes there is a short second strap running from the middle of the horizontal strap to a loop passed through a pair of closely-spaced holes at the top centre of the shield.

To consolidate the identification of this Olo shield type, we draw attention to several other examples. These are one in the South Australian Museum (Fig. 6) collected by Kenneth Thomas in 1928 at Kapoam, an Olo village located on the north fall of the Torricellis overlooking Aitape (Craig 2002, Fig. 13.12); (11) another recently donated to the South Australian Museum (Fig. 7), collected in 1970 at the village of Buru'um; an old shield (Fig. 8) that Craig photographed in 2002 at the village of Wilium (Craig 2006, Fig. 80); and two (Figs 9, 10) collected by Mitchell in 1971 in the village of Taute. Two in the Australian Museum, collected by H.D. Eve in the 1930s, are from Sabite (Figs 11, 12). Another twelve Olo shields (Figs 13-24) are in the collections of the Museum der Kulturen, Basel, all collected by Meinhard Schuster during his 1965-67 upper Sepik and central New Guinea field-work from five villages west of the Lumi government station and airstrip.

GNAU SHIELD TYPE

During his field trip in 2002, it became apparent to Craig that the shields of the Gnau, who live in three villages several kilometres south-east of the Anguganak airstrip, are significantly different from Olo shields. He photographed one at Rauit (Fig. 25) and subsequently located two at the PNG National Museum, collected by Gilbert Lewis during 1968-69 at Rauit (Figs 26, 27).

The Gnau shield is called lita'at (sing.), latagap (pl.). (12) The wood used for the shields came from the buttress roots of trees called lu lat^t, or lu bawanit. Before the availability of steel tools, the plank would have been trimmed and shaped with a stone adze. While the size is similar to that of Olo shields, the shape is more rectangular than are the Olo shields (13) and the carrying device is somewhat different. Although there is a horizontal strap as for the Olo shields, it is secured to one pair of holes on either side rather than to two pairs of holes on either side. In addition, there is a vertical strap secured between a pair of holes at the top and a pair of holes at the bottom of the shield. The design carved on the Gnau type of shield is also different in that the spirals are much larger than for the Olo shields, so that the design occupies much more of the surface of the shield.

The Gnau word for the incised design on a shield is matat, meaning 'design' or 'pattern'. Although there was no explanation of what the design meant or represented, it was often reproduced in the paintings on sago palm spathes traditionally used in ceremonies to line the underside of the roof of the men's house (see Briggs 1928: 269).

When completed, the shields were covered with black paint (gablit). This black paint was made by burning a particular vine (lambat gablit) so that the soot collected on the inner surface of a clay pot inverted over the smoke; the soot was then mixed with water and the sticky sap of a breadfruit tree. The phrase laga'ai gablit ('they cook up the black paint') also meant by association 'they are preparing to fight' because the black paint was used to smear over the faces and skin of men before fighting.

A shield collected in the 1930s by geologist H.D. Eve from the Wiaki-speaking village of Wamil (= Wemil, Womil) (Fig. 28) is entirely consistent with the three from Rauit and therefore belongs to this type. Wiaki speakers are eastern neighbours of the Gnau, and Wamil is only 8 kilometres from Rauit. Further, Lewis, while recording genealogies, was told that a Rauit group moved to Wamil eight or nine generations before that of the older people alive in 1969. Lewis saw wooden shields like these also at Yemereba and Yulem, respectively some five and nine kilometres south of Wamil. These villages speak Beli, a language of the Maimai Stock as is the Wiaki language, whereas Olo and Gnau are of the Wapei-Palei Stock of Torricelli Phylum languages. Also, Lewis was told, but did not observe, that the men of Laeko and Libuat villages, speakers of a different Family of the Maimai Stock of Torricelli languages but neighbours of Yemereba and Yulem only five kilometres to the west, also used the Gnau shield type. Speakers of several languages classified in different Families and two Stocks therefore shared the same Gnau shield type. The information that a Rauit group moved to Wamil several generations ago may suggest one of the reasons why artefacts characteristic of one language group are made also by neighbouring language groups.

A third shield collected by Lewis at Rauit is notable for its different carved design (Fig. 29). The size, shape and carrying device are consistent with the other Gnau shields, but the simpler design is carved with wider relief bands than for the other Gnau shield design, and by removing more of the shield's surface from around the relief bands. Lewis was told that the design on this unique Rauit shield had been carved by Walbasu of Rauit for a man named Dangen, both of whom lived four or five generations before the 60-year old man who told him about it in 1969. Lewis never saw any other like it during his time in the field. The design recalls certain Awun, Namie and Abau (upper Sepik) designs (eg. respectively: Kelm & Kelm 1980, Plate 29; Craig 1975, Fig. 25; Kelm 1966, Abb. 201, 206) rather than Gnau or Olo shield designs; however, the upper Sepik shields are different in size, form and carrying device (compare Tiesler 1970, Abb.52 with Craig 1990, Fig. 8).

It seems best to consider this unique Rauit shield as belonging to the Gnau type, on the basis of size, shield form and carrying device, but with a different carved design on its front surface. Of course, there are many areas in New Guinea where variation in the carved and/or painted graphic designs does not disqualify objects from belonging to the same 'type'.

ANALYSIS OF OLO AND GNAU SHIELD DESIGNS

Twenty-three Olo shields and five of the Gnau type with good location data have been identified. This provides an opportunity to examine the graphic designs carved on the outer face of the shields to determine whether there are any non-random distributions in design characteristics.

The presence of a pair of spirals at the top and at the bottom of the design is consistent for all the designs but the design elements at the centre vary, some utilising a pair of spirals, concentric circles, lozenges or 'banana' shapes, and others a zig-zag form. No pattern could be discerned in the geographical distribution of these central elements of the design.

Another variation is that the incised design is mostly carved into the flat surface of the shield but sometimes it is carved into an area of relief, with most of the shield's surface removed a few millimetres below the level of the design. Again, no pattern could be discerned in the distribution of this trait.

Several shields are carved with a motif like a pair of horns, above and below the top and bottom pair of spirals. There were three examples from Wilkili, two from Buru'um and one from Wilium; thus most from villages immediately west of Lumi.

Although there are marked differences between Olo and Gnau shields, the Wamil shield is not significantly different from most of the Rauit (Gnau) shields even though the people of Wamil speak Wiaki, a language of a different Stock to Gnau.

[FIGURE 2 OMITTED]

[FIGURE 3 OMITTED]

[FIGURE 4 OMITTED]

[FIGURE 5 OMITTED]

[FIGURE 6 OMITTED]

[FIGURE 7 OMITTED]

[FIGURE 8 OMITTED]

[FIGURE 9 OMITTED]

[FIGURE 10 OMITTED]

[FIGURE 11 OMITTED]

[FIGURE 12 OMITTED]

[FIGURE 13 OMITTED]

[FIGURE 14 OMITTED]

[FIGURE 15 OMITTED]

[FIGURE 16 OMITTED]

[FIGURE 17 OMITTED]

[FIGURE 18 OMITTED]

[FIGURE 19 OMITTED]

[FIGURE 20 OMITTED]

[FIGURE 21 OMITTED]

[FIGURE 22 OMITTED]

[FIGURE 23 OMITTED]

[FIGURE 24 OMITTED]

[FIGURE 25 OMITTED]

[FIGURE 26 OMITTED]

[FIGURE 27 OMITTED]

[FIGURE 28 OMITTED]

[FIGURE 29 OMITTED]

[FIGURE 30 OMITTED]

[FIGURE 31 OMITTED]

[FIGURE 32 OMITTED]

[FIGURE 33 OMITTED]

The distribution of how the spirals are drawn and carved also reveals a pattern. Fig. 30 shows two ways of representing a spiral: A, a double-interlocking spiral and B, a single spiral. All the spirals on a particular shield are not necessarily carved the same way. On the map (Fig. 1) we have plotted the presence of each of these two ways of carving spirals with no attempt to indicate frequencies, bearing in mind that the total number of shields reviewed for this purpose was only twenty-eight. The number of shields for individual villages is small (one to four) so the results are indicative rather than conclusive.

What emerges is that over the whole Lumi-Anguganak area, the type A spiral is ubiquitous on shields, regardless of language and shield type. Only among the western Olo villages of Buru'um, Taute and Wilkili west of Lumi, and at the Olo village of Kapoam on the northern fall of the Torricellis, were there shields carved with type B spirals. Nevertheless the type A spiral is by far the most common. By contrast, Craig observed in 2002 that the spiral carved on slit-gongs, most hand drums and house posts, and painted on sago petioles, are predominantly of type B, the single spiral. It is not at all clear why this should be so.

USE OF OLO SHIELDS IN FIGHTS

Mitchell has published an account of local fighting (1978: 87-93) by Taute village men in the 1930s before extensive Western contact. Taute was the enemy of Lau'um and Buru'um villages, also Olo speakers, who resided along the same ridge above Sibi River. Traditionally these ridge villages were stockaded to impede direct attack. (14) Although enemies, the villages intermarried but there was one important marital caveat--it was forbidden for a warrior to kill his wife's close relatives.

Mitchell recounts an extended conflict between Taute and another enemy, the people of Kamnum village who spoke Autu, a Sepik language. Kamnum is about two hours walk south of Taute and the conflict involved an opportunistic series of back and forth killings, often separated by years. There were six old men, still alive in Taute when Mitchell was there, who had taken part in this feud. The shield in Fig. 10 was carried by one of these men, Tongol, and was carved by his father's brother.

 As young warriors they had decorated their bodies with paint and
 bright leaves to strike fear into their victims. Setting out on an
 ambush or raid, they carried long black bows, viciously barbed
 arrows that ripped into an enemy's body, and large black carved
 shields to crouch behind when an arrow came their way. In their
 armbands were cassowary-bone daggers for hand-to-hand combat
 (ibid.: 90).

Despite their bold appearance, the strategy was to attack a hamlet by surprise, kill the first human being in view (as often a child as an adult) and flee triumphantly back to their village. All that mattered was 'a life for a life', it did not matter whose. Finally, a dysentery epidemic during WWII caused so many deaths the two villages were forced to deal with what they believed to be the cause--the vengeful ghost of one of the raid victims. Reciprocal friendship feasts were held to placate the ghost's anger. That, abetted by the arrival of Western patrol officers and missionaries after WWII, resulted in a lasting truce.

PRACTICAL AND RITUAL ASPECTS OF SHIELDS AMONG THE GNAU

In past times, Gnau shields were kept with bows and arrows on racks above the beds in the men's house (gamaiyit). Patrol Officer P.E. Fienberg in 1951 (Lumi Patrol Report 4/51-52) reported:

 ... an interesting feature, noticed in all the villages, of a
 central arsenal, in which is kept an assortment of weapons, spears,
 shields, bows and arrows, skulls, other
 bones and trophies of war and of the hunt. Even spare bowstrings
 and arrowheads are ready to hand. A continual guard is maintained
 over these buildings, which are usually situated in a central
 position in the villages.

The shield was carried on the left shoulder if the man was right-handed as nearly all were. The left hand held the bow. The bow (wanggapa), as for the Olo speakers, was made from black palm and was about 1.8 metres long, taller than most men. The right arm drew the bow, leaving the right side of the body exposed. To provide additional protection, a Gnau warrior wore a rectangular apron called golugi gnggi ('stringbag-blood'). These were not in fact bags but consisted of a single layer of fabric made from string using the same looping techniques as for string bags or 'bilums'. These aprons hung from the right shoulder under the armpit and were variable in size (c. 60 x 40 cm). Complex patterns were worked into them by dyeing the string as it was looped (Fig. 31). They were said to be made for a man typically by his mother or by his sister. They gave some protection from wounding. If the arrow struck obliquely, it might be deflected or tangled in the fabric, but a direct shot could pierce through. The only other instance of a protective fabric 'shield' reported for New Guinea is from the northern side of the Huon Peninsula in Morobe Province, but it is made of woven strips of bast (Beran & Craig 2005: 56-7, Fig. 3.29).

At Rauit, they referred to the past as the 'time of arrows' (wibasam dap sigap). (15) The danger was from surprise attack, ambush and raid rather than from formal battles. It is obvious that many of the skills and qualities required for successful hunting and warfare are similar--alertness, cunning, stamina, and skill with the bow and arrow. Success in hunting would bring success in war, they said. If a man killed a pig, and the appropriate ritual was performed, he would become a better fighter. If he killed a man and the rites for homicide (Lyig^t--'ginger') were done, then he should go to hunt for he would have success. The theme of reciprocal enhancement was suggested by many of their comments (Lewis 1980: 168; 1995: 31-3), by their explanations for the elaborate system of food prohibitions (Lewis 1980: 158-64), and by the criteria for the right to wear the special hunter's string bag (lagil bawug), the splay-winged mounted hornbill (wamalan) at his back (visible in the photographs in Lewis 2000:115, 156), and the pig-tusk torque (dalyiwug).

Given the small size of village communities (250-350 people with perhaps 60 able-bodied men), any man might find himself obliged to fight. The enemies were usually known and identifiable individually. The Gnau were not particularly warlike but the boys and young men had to be prepared for the possibility of raids occurring at any moment, in which they might depend directly on a next-door neighbour or the man in the nearest garden. At Rauit, the puberty and male initiation rituals enhanced male bonding through shared seclusion and the revelation of male secrets.

Certain rituals involving war shields were specifically intended to prepare boys and young men for warfare (Lewis 1995: 33-5). Harrison, in his book on Avatip (middle Sepik) warfare (1993), points out very convincingly that men had to be made capable of extreme violence, ready to over-ride the normal social and psychological inhibitions on killing. This was clearly the case for the Gnau. If someone was shot close enough to the village, young boys or youths were brought by their fathers to shoot or jab an arrow into the body, to experience for the first time what it was like. Selaukei, quite elderly when he talked to Lewis, recalled his first experience when he was a boy of about eleven. His father fetched and pulled him to the body of a man shot near the edge of the hamlet, calling out to Selaukei's mother's brother to bring his shield. His father made him jab an arrow into the body even though Selaukei felt sick and had not wanted even to look at it--it was such a mess with arrows sticking into it. But he shot it with an arrow, then was made to lie down on the ground while his mother's brother covered him with the shield and crushed him down with it. The next time this ritual was performed, he was fed a bit of fight-aroid (a species of wild taro) and again crushed down by his mother's brother sitting on the shield. He shouted and shouted but his mother's brother demanded he keep still, explaining that he was giving him the wild taro and crushing him down so that he would grow strong, steady and fearless, and would fight, not run away when faced by an enemy. This ritual was repeated and reinforced on subsequent occasions.

One of the duties of the mother's brother towards his sister's son was to give him his shield and to help with the Lyig^t rites for him. These rites were carried out after a homicide and involved the mother's brother and senior men feeding the killer the Lyig^t ritual substances. The preparation included the wild taro, a very hot ginger, and blood or bowel contents off arrows retrieved from the body of an enemy, pounded together in a herb preparation with scrapings from a boar's tusk, a shield, an ancestral skull or teeth, and hardened coconut flesh. The mixture was administered by spitting it onto the homicide's body and rubbing it in with fierce stinging nettles. Simultaneously, fierce nettles were rubbed on, and stuck into, his ears. They described how the returning homicide was first acclaimed in triumph, ushered into the gamaiyit (men's house), relieved of his weapons, then put through the Lyig^t rites. These were said to carry the risk of sending the man into a fierce mindless state in which he might shoot at anything. He was pinned down with a slit-gong (bung'gei) until his eyes showed that his mind had come back to its right place. This was repeated for two or three days, during which at times he might be held down with a war shield or a big hand drum instead of with the slit-gong, until it was judged that he had returned to his senses and had been strengthened and protected by the treatment. While he stayed inside the gamaiyit after the killing, he could eat only 'dried out things' (see Lewis 1980: 140, 148-151,157).

PIGSKIN SHIELDS

Lewis obtained information worth reporting for the light it sheds on the techniques of using shields and on the circumstances under which an artefact type characteristic of one group of people, in this case the pigskin shield, comes to be adopted by another group of people.

Yemereba and Yulem (about 13 and 15 kilometres south-east of Rauit respectively) were near the limit of the Rauit people's old range of recognised connection and rare visits. On a trip to those villages, Lewis went with an old man who had long ago gone to Yemereba to help in a fight there; they received a warm reception. They talked about fights in the past and that was the context in which they showed Lewis a big oval pigskin shield, the skin fastened with split rattan onto a thick rattan oval frame. Lewis does not recall the carrying device.

A Yemereba man said that they, being on the boundary between bow versus spear people, used to fight with the Sepik-speaking, spear-using people of Yawa, Rapaw and Wosd to the south who had the oval pigskin shields. The people of Yemereba used the Gnau type of wooden shields, and bows and arrows. The Rauit and Yemereba people used shield and bow in the same way. They did not line up with shields as a wall with bowmen behind but the bowmen advanced carrying their shields, using a particular foot-shifting step (in Pidgin: sam-sam, in Gnau: laga'alp) that kept the feet and legs protected by constant movement. The Rauit people also had throwing spears but these weapons were less often mentioned in talk about past fighting.

The Yemereba man said he had disliked fighting with the spear people because they would hide behind their shields, peeking through the gaps in the binding around the edge, advancing implacably until they got close. The problem for bow and arrow people was that they carried only a few arrows. The spear people would wait until a man had shot all his arrows and then they would rush him and tear his shield aside to spear him.

Lewis's Yemereba and Yulem informants said that in the past they sometimes combined the use of bows and wooden shields with using spears and pigskin shields when they fought with the Yawa-Rapaw and Wosapom people. (16) They said the older men fought in the front with spears and pigskin shields and the younger men backed them up with bows and arrows and wood shields. (17) The men of Yemereba said that they copied the use of pigskin shields so they could fight the men of Wosapom more successfully. In the case of the Wamil shields, cited above, the shield type may have been adopted from an immigrant friendly group; in this case, the pigskin shield was adopted from an enemy group.

In 1970, while looking for a suitable research site, Mitchell came across a pigskin shield (Fig. 32) at Klaplei III, about 14 kilometres south-east of Nuku Patrol Post and 27 kilometres east-south-east of Yemereba. (18) The people of Yemereba speak Beli, a language of the Maimai stock of Torricelli languages, whereas the people of Wosapom speak Pahi; Klaplei III is a Mehek-speaking village. Pahi and Mehek languages are of the Tama Stock of Sepik languages and distantly related to Kwanga, spoken by the people of Bongos (11 kilometres east of Klaplei III), and to Kwoma, near Ambunti. Both Bongos and Ambunti are areas where pigskin shields were made and used (Beran & Craig 2005:91 and Fig.4.22). Equally significant is the geographical contiguity of these locations where pigskin shields are used, between the Sepik River and the southern foothills of the Torricelli Mountains.

Thus, there seems to be a clear boundary between bow-and-arrow people speaking Torricelli languages and spear people speaking Sepik languages correlating with the distribution of wood versus pigskin shields, with an example of diffusion across the boundary at Yemereba and Yulem.

THE COMMODIFICATION OF LUMI SHIELDS

In many parts of Papua New Guinea, artefacts following traditional forms were, and still are, produced for sale to tourists and other outsiders. This industry has been most successful in areas readily accessible to tourists and where there was always a rich tradition of carved forms, such as along the Sepik River. The Lumi-Anguganak area is less accessible and its restricted inventory of religious objects quickly succumbed to the influence of Christian missions. This left only the old shields available for sale.

The German colonial administration attempted to prevent warfare by destroying the people's weapons. As reported elsewhere (Craig 2002: 203), Patrol Officer Kenneth Thomas noted that this meant shields were no longer made. 'Hence the only ones obtainable [were] those few which were safely hidden away when most weapons were destroyed'. But the Germans did not penetrate far into the Torricellis and there must have been a greater number of hidden and/or unused shields in the numerous villages of the area than Thomas supposed. No doubt, at first people were reluctant to sell them to outsiders because, as Thomas believed, 'the shield perhaps having been "plank belong pappa" [sic], since deceased, the son would not part with it'; secondly, they may not have been convinced that the ban on warfare by the succeeding Australian administration would hold.

In due course, attention to things of the past became less concentrated (especially with Christian mission influence), the ban on warfare was seen to be effective, and a cash income became more desirable. This set the conditions for shields to be brought from storage and sold to outsiders. After some time, it may be that a few were made specifically for sale but most 'Lumi' shields in collections do seem to have been made, if not actually used, for warfare. What is unfortunate is that the collectors of Lumi shields usually have failed to record where the shields were made or purchased, other than 'Lumi'. The Torricelli terrain is difficult and it is likely that collectors and artifact dealers, in the interest of time and convenience, had villagers bring shields and other artifacts to the Lumi government station and mission, rather than collectors trekking to the villages. This was common practice elsewhere in New Guinea, often leading to incorrect or vague information about provenance, for example, in the case of the so-called 'Josephstaal' shields (Beran & Craig 2005: 109).

During a field trip to the Torricellis in 2002, Craig came across one old shield at the Olo village of Wilium (Fig. 8 and Craig 2006, Fig. 80). He made no attempt to purchase it and the owner volunteered the information that he and his relatives were not interested in selling it but wished to keep it as an heirloom to remind them of the past. By now there must be few such examples remaining in the villages of the Torricellis and Craig saw no evidence that shields were being manufactured specifically for sale. Earlier, on a revisit to Taute village in 1982, Mitchell photographed one of his earlier informants making a shield he hoped to sell. However, it was an isolated example of one man's entrepreneurship. More recently, a few shields have been carved as decoration for some of the Catholic mission churches in the area (Fig. 33). These recent shields are distinguishable from traditional shields by the coarser carving, with the design occupying more of the shield's surface, and by the liberal application of red, white and black pigments.

CONCLUSIONS

This paper identifies and describes a fifth (Gnau) type of wood shield used in the Torricelli Mountains of West Sepik Province, Papua New Guinea. Whereas the Olo shield type (on the basis of the available evidence) is confined to Olo speakers, the Gnau shield type crosses language boundaries. There was a possibility that this may have been facilitated by kin group migrations.

We report a unique looped string protective device worn by Gnau warriors in the vicinity of Anguganak and a type of pigskin shield is noted, south and east of the region where the Gnau type of wood shield was used.

The wood shields were used by speakers of Torricelli languages and the pigskin shields were used by speakers of Sepik languages located in the region between the southern Torricelli foothills and the Sepik River. An example of diffusion of the pigskin shield to two Torricelli-speaking villages was explained as a means for coping with the different fighting techniques of Sepik-speaking enemies.

An analysis of the incised designs on twenty-eight examples of Olo and Gnau shield types found no regularity in the geographical distribution of the various central elements of the designs. Of the two types of spirals carved on the shields, one was found to be ubiquitous but the other occurred only on Olo shields, and mainly from villages immediately west of Lumi; a 'horn-like' design element had a similar restricted distribution. These characteristics of shield design were therefore associated with the Olo language, but no explanation for their partial distribution in that language area is evident.

These results suggest the need for detailed analysis of greater numbers of the full range of artefacts from defined regions to clarify the differential effects of language and distance on variation in material culture. Thorough surveys and analysis of collections in museums and in private hands throughout the world are necessary to achieve this.

ACKNOWLEDGEMENTS

This paper has benefited by the constructive comments of anonymous peer reviewers, for which the authors are grateful. The map and the drawings of shields are by Barry Craig and he apologises for his mediocre skills. Drawings of shields in the Museum der Kulturen, Basel, were based on images kindly supplied by agreement with the Director Anna Schmid; of shields in the Australian Museum, on photographs by Barry Craig and Andrew Fyfe; of shields in the South Australian Museum, on photographs by Elizabeth Murphy and Tony Vlavogelakis; of the Rauit shields in the PNG National Museum, on photographs by Barry Craig; of the shield collected by William Mitchell, in the American Museum of Natural History, on the image accessed on the AMNH website; and of the shield in the possession of William Mitchell, on a photograph supplied by him. The Image Library of the Metropolitan Museum of Art supplied the images of the pigskin shield photographed by William Mitchell. Please be aware that the drawings of the shields are not to the same scale.

REFERENCES

BENITEZ-JOHANNET, P. and J.P. BARBIER. 2000. Shields: Africa, Southeast Asia and Oceania from the collections of the Barbier-Mueller Museum. Munich, London, New York: Prestel.

BERAN, H. and B. CRAIG (eds). 2005. Shields of Melanesia. Adelaide: Crawford House Publishing Australia.

BRIGGS, E.A. 1924-1926. Field Diaries. University of Sydney Archives. nd. [1926?] The Black Heart of New Guinea. 16 mm film, original at Screensound, Canberra; copy in Sydney University Archives.

1927. Filming in Cannibal-Land. The Wide World LIX, Nr 350 (May 1927).

1928. New Guinea: Land of the Devil Devil. The Australian Museum Magazine. 3: 265-73.

CRAIG, B. 1975. The Art Style of Yellow River, West Sepik District, New Guinea. Baessler-Archiv 23: 417-445.

1990. Is the Mountain-Ok Culture a Sepik Culture? In Nancy Lutkehaus et al. (eds.), Sepik Heritage: Tradition and Change in Papua New Guinea. Durham: Carolina Academic Press. Pp. 129-149.

2002. 'A Stranger in a Strange Land': Kenneth Thomas in the North Sepik Region of Papua New Guinea. In Anita Herle, Nick Stanley, Karen Stevenson and Robert L. Welsch (eds), Pacific Art: Persistence, Change and Meaning. Adelaide: Crawford House Publishing. Pp. 191-207.

CRAIG, B. (ed.). 2006. Living Spirits with Fixed Abodes. The Masterpieces Exhibition, Papua New Guinea National Museum and Art Gallery. Adelaide: Crawford House Publishing Australia.

FIENBERG, P.E. 1951. Patrol from 24-9-51 to 8-10-51. Lumi Patrol Report 4/51-52, Sepik District, Territory of Papua and New Guinea.

FOLEY, W.A. 2005. Linguistic prehistory in the Sepik-Ramu basin. In, Pawley A., R. Attenborough, J. Golson and R. Hide (eds), Papuan Pasts: cultural, linguistic and biological histories of Papuan-speaking peoples. Canberra: Pacific Linguistics, RSPAS, The Australian National University. Pp. 109-144.

GUDDEMI, RV. 1992. We Came from This: Knowledge, Memory, Painting and "Play" in the Initiation Rituals of the Sawiyan? of Papua New Guinea. PhD Thesis, University of Michigan.

HARRISON, S. 1993. The Mask of War. Manchester: Manchester University Press.

KELM, H. 1966. Kunst vom Sepik II. Berlin: Museum fur Volkerkunde.

KELM, A. and H. KELM. 1980. Sago und Schwein. Ethnologie yon Kwieftim und Abrau in Nordost-Neuguinea. Wiesbaden: Franz Steiner.

LAYCOCK, D.C. 1973. Sepik Languages--Checklist and Preliminary Classification. Pacific Linguistics Series B, No. 25. Canberra: ANU.

1975. Languages of the Sepik Region [Map]. Pacific Linguistics Series D, No. 26.

LEWIS, G. 1975. Knowledge of Illness in a Sepik Society. A study of the Gnau, New Guinea. London: Athlone Press; New Jersey: Humanities Press.

1980. Day of Shining Red: an essay on understanding ritual. Cambridge: Cambridge University Press.

1995. 'Payback and ritual in war: "New Guinea'. In War: A Cruel Necessity? The bases of institutionalized violence. Edited by Robert A. Hinde and H. E. Watson. Pp.24-36. London and New York: Tauris.

2000. A Failure of Treatment. Oxford: Oxford University Press.

MARSHALL, A.J. 1937. Northern New Guinea. The Geographical Journal LXXXIX, 6: 489-506.

MCGREGOR, D.E. 1982. The Fish and the Cross. Goroka: The Melanesian Institute. 2nd Ed.

MITCHELL, W.E. 1978. The Bamboo Fire. New York: Norton & Co.

1988. The Defeat of Hierarchy: Gambling as Exchange in a Sepik Society. American Ethnologist 15: 638-657.

1990. Therapeutic Systems of the Taute Wape. In Nancy Lutkehaus, et al. (eds.), Sepik Heritage: Tradition and Change in Papua New Guinea. Durham: Carolina Academic Press. Pp.428-438.

2004. Wape. In C.R. Ember and M. Ember (eds.) Encyclopedia of Medical Anthropology: Health and Illness in the World's Cultures. New York: Kluwer Academic/Plenum Publishers. Vol. 2, Pp.1009-1017.

MOORE, C.C. & A.K. ROMNEY, 1994. Material Culture, Geographic Propinquity, and Linguistic Affiliation on the North Coast of New Guinea: A Reanalysis of Welsch, Terrell, & Nadolski (1992). American Anthropologist 96, 2: 370-92.

1995. Commentary on Welsch and Terrell's (1994) Reply to Moore and Romney (1994). Journal of Quantitative Anthropology 5:75-84.

1996. Will the "Real" Data Please Stand Up? Reply to Welsch (1996). Journal of Quantitative Anthropology 6:235-61

ROBERTS, J.M., C.C. MOORE, & A.K. ROMNEY, 1995. Predicting Similarity in Material Culture among New Guinea Villages from Propinquity and Language. A Log-Linear Approach. Current Anthropology 36, 5: 769-88.

TERRELL, J. & R.L. WELSCH, 1990. Trade Networks, Areal Integration, and Diversity along the north Coast of New Guinea. Asian Perspective 29, 2: 155-65.

TIESLER, F. 1970. Tragbandschilde aus dem Hatzfeldhafen-Gebiet, dem Hinterland der Berlinhafen-Kuste und dem Lumi-Gebiet, Nord-Neuguinea. Jahrbuch des Museums fur Volkerkunde zu Leipzig 27: 185-217, Tafel XXXIII-LII.

WELSCH, R.L. 1996. Language, Culture, and Data on the North Coast of New Guinea. Journal of Quantitative Anthropology 6: 209-34.

WELSCH, R.L. & J. TERRELI, 1994. Reply to Moore & Romney. American Anthropologist 96, 2: 392-96.

1998. Material Culture, Social Fields, and Social Boundaries on the Sepik Coast of New Guinea. In M.T. Stark (ed.) The Archaeology of Social Boundaries. Washington/London: Smithsonian Institution Press. Pp.50-77.

WELSCH, R.L., J. TERRELL, & J.A. NADOLSKI, 1992. Language and Culture on the North Coast of New Guinea. American Anthropologist 94: 568-600.

Barry Craig

South Australian Museum

Gilbert Lewis

St John's College Cambridge

William E. Mitchell

University of Vermont

NOTES

(1.) Barry Craig is Curator of Foreign Ethnology at the South Australian Museum; Gilbert Lewis is Fellow, St John's College, Cambridge; William E. Mitchell is Professor Emeritus, University of Vermont. Craig did three weeks fieldwork in the area in 2002, Lewis spent two years among the Gnau in 1967-9 on a Social Science Research Council grant; Mitchell spent 18 months at Taute, south-west of Lumi in 1970-71 on a National Institutes of Health research grant. Both Lewis and Mitchell have revisited their fieldwork locations in subsequent years.

(2.) Laycock 1975; although the identification of Laycock's Sepik-Ramu Phylum has been revised by Foley (2005) into separate Sepik and Ramu phyla, the Torricelli Phylum seems secure.

(3.) Also spelt Wapi and Wapei.

(4.) We are aware that 'types' are constructs, arising from consideration of a number of variable characteristics. In our considerations, variations in form (size, shape and carrying device) are given more weight than variations in graphic design (on the front of the shield). This is consistent with the approach adopted in Beran & Craig (2005).

(5.) However, it must be noted that Craig was in error in locating his fourth type (Beran & Craig 2005:42 and Fig.3.17, collected by Father Regis in the vicinity of the Wassisi Mission Station) in the Nuku area where people speak the Mehek (upper Sepik) language. The people in the vicinity of Wassisi speak Wiaki, a Torricelli language.

(6.) Briggs graduated from the University of Sydney in 1912, obtaining his MSc in 1924 and DSc in 1929 from the same university. He commenced employment at the Australian Museum in 1912 but left to take up an appointment in the Department of Zoology at the University of Sydney in 1919 where he remained until he retired in 1955.

(7.) Briggs made two field trips to New Guinea, the first in the Prince Alexander Mountains of East Sepik Province (25 June to 12 July 1924) and the second in the Torricelli Mountains of West Sepik Province (19 December 1925 to 11 February 1926). From his field diaries (University of Sydney Archives), it is clear that he made significant collections of ethnographic material on both occasions. The University of Sydney Archives holds a print of the cine film made by Briggs, titled The Black Heart of New Guinea, but all Briggs's monochrome photographs are held by the National Geographic Society in Washington DC. it is not known where Briggs kept the collections when they came to Sydney. They could have become part of the collections held by the Department of Anthropology until around 1957 when that Department's collections were sent to the Australian Institute of Anatomy in Canberra. These collections have since been returned to Sydney into the care of the Australian Museum.

(8.) Also spelt Epeu, Epiou and Epeyou.

(9.) During his fieldtrip in 2002, Craig showed this image to people in several Torricelli villages and was informed that the site in the photograph appeared to be Epiyu. The only time Briggs mentions in his diary that he photographed shields was on 16 January at Epiyu. However, these two shields are not the two in the Australian Museum. Another Briggs photograph (photocopy obtained from Virginia-Lee Webb of the Metropolitan Museum of Art, New York) shows a man demonstrating the use of a shield but again it is not one of the shields in the Australian Museum. In Briggs's film Into the Black Heart of New Guinea, there is a sequence of a man demonstrating the use of bow and arrows and shield, but this shield does not appear to be one of those photographed by Briggs nor one of the two in the Australian Museum. Unfortunately, then, there is no indisputable evidence that the two in the Australian Museum came from Briggs; but it remains a high probability.

(10.) The plank was cut from the buttress root of a tree that had the right thickness and a curved upper edge. A similar shape is characteristic of some shields of Central Province, also cut from buttress roots of trees (Beran & Craig 2005:187-9 and Fig. 7.8).

(11.) It should be noted that, unfortunately, this shield was published upside down.

(12.) For transcribing Gnau words, Lewis has used a for the schwa, ^ for 'u' sound as in English 'up', and -ngg- for the nasalised -ng- as in English 'single'

(13.) Given that both Olo and Gnau shield types are made from buttress roots, it seems that the different shape of the Gnau shield must be a conscious choice and not determined by the shape of the buttress root.

(14.) The villages were generally built on narrow ridges with steep falls on either side; a barrier at each end of the village across the ridge was sufficient to impede surprise attacks (Marshall 1937:492 and plate opposite).

(15.) Lit. 'in past times [when] we [had] arrows'.

(16.) Lewis isn't certain that the Yemereba and Yulem men made their own pigskin shields but they did volunteer the information that they were made from boar hide left to harden in the sun. At Wosapom, Lewis was told the shields are made of boar skin, stretched and left to harden in the sun; the shield maker was forbidden to wash lest the hide fail to harden.

(17.) This was opposite to the way the Kwoma deployed pigskin and wood shields (see Beran & Craig 2005:91 and endnote 13 on p.110).

(18.) Craig also was told that there were pigskin shields in this region (Beran & Craig 2005: 42). Guddemi (1992: 144) reports the use of pigskin shields (futono) by the Sawiyano of the West Range, south of the Sepik River and adjacent to the Namie-speaking settlement of Panewai. The Sawiyano, who speak a language of the Left May Phylum, used bows and arrows for warfare, rather than spears.

Table 1. Villages and Languages of the
Torricelli and upper Sepik Region

Villages (in text) Language Family

Wilkili, Taute, Lau'um, OLO WAPEI
Buru'um, Ali, Sabite
Wilium, Epeyou, Kapoam
Rauit GNAU WAPEI
Wamil WIAKI WIAKI
Yemereba, Yulem BELI BELI
Laeko, Libuat LAEKO-LIBUAT LAEKO-LIBUAT
Wosapom PAHI TAMA
Yawa-Rapaw KALOU TAMA
Klaplei III MEHEK TAMA
Bongos KWANGA NUKUMA
Kwoma villages KWOMA NUKUMA

Villages (in text) Stock Phylum

Wilkili, Taute, Lau'um, WAPEI-PALEI TORRICELLI
Buru'um, Ali, Sabite
Wilium, Epeyou, Kapoam
Rauit WAPEI-PALEI TORRICELLI
Wamil MAIMAI TORRICELLI
Yemereba, Yulem MAIMAI TORRICELLI
Laeko, Libuat MAIMAI TORRICELLI
Wosapom TAMA SEPIK
Yawa-Rapaw TAMA SEPIK
Klaplei III TAMA SEPIK
Bongos MIDDLE SEPIK SEPIK
Kwoma villages MIDDLE SEPIK SEPIK