Other formats

    TEI XML file   ePub eBook file  


    mail icontwitter iconBlogspot iconrss icon

Ngā Tohuwhenua Mai Te Rangi: A New Zealand Archeology in Aerial Photographs

3 — Pā

page 41


Pā are one of the most common features, of the New Zealand historic landscape and provide an extraordinary range of opportunities for aerial photography. Archaeologists always use the term 'pā' to refer to earthwork fortifications; the term as used by archaeologists does not refer to a modern Māori settlement. The term 'papakainga' rather than 'pā' better expresses the sense of a modern Māori settlement long associated with former ancestral villages or living areas. Examples were illustrated in chapter 1, where their common association with pā in the archaeologist's sense was evident. Most pā went out of use somewhat less than 200 years ago, primarily in response to the desire to live near European settlements and trading harbours, although forms of pā for gunfighting were built up until 1880. Prominent and well recognised in local landscapes, pā are frequently mentioned in traditional recitations on the neighbouring marae, or on more distant marae when the local people are visiting there.

There was early European interest in the description of pā, not least from the earliest navigators who, aware of the risk of being attacked and overwhelmed by surprise or by numbers, carefully surveyed Māori defensive as well as offensive techniques. The most important of the descriptions are those of James Cook in Mercury Bay on the Coromandel in 1769 and Roux in the Bay of Islands in 1772. 1 For full and accurate detail of a significant site which I have been able to photograph usefully, Wharetaewa, the pā described by Cook at Mercury Bay, is the most important:

... on the East side is a high point or peninsula. ... in some places quite inaccessible to man and in others very difficult except on that side which face'd the narrow ridge of the hill on which it stands, here it is defended by a double ditch a bank and two rows of Picketing—the inner row upon the bank but not so near the Crown but what there was good room for men to walk and handle their arms between the Picketing and the inner dich: the outer Picketing was between the two ditches and laid sloping with their upper ends hanging over the inner ditch, the depth of this ditch from the bottom to the Crown of the bank was 24 feet. Close within the inner picketing was erre[c]ted by strong posts, a stage 30 feet high 40 in length and 6 feet broad, the use of this stage was to stand upon to throw darts at the Assailants, and a number of darts lay upon it for that purpose. At right Angles to this stage and a few paces from it was another of the same construction and bigness. . .. the whole Village was pallisaded round with a line of pretty strong picketing run round the edge of the hill. The ground within ... they had divided it into little squares and levelled each of these; these squares lay in the form of an amphitheatre and were each of them pallisaded round and had a communication one with another by narrow lanes or little gate ways which could easily be stopped up, so that if any enemy had force'd the outer picketing he had several others to incounter before the place could be wholy reduced. . . . The Main way leading into this fortification was up a very steep part of the hill and thro' a narrow passage about 12 feet long, and under one of the Stages... 2

James Cook had learnt his surveying techniques as part of military operations on the Canadian coast, and his description is reliable and thorough. The features of Wharetaewa on the ground today are relatively simple in arrangement. The ditch and bank are clear, as are some of the terraces used for habitation. In the aerial page break
Views from different directions of the pā Wharetaewa and Wharekaho in Mercury Bay

Views from different directions of the pā Wharetaewa and Wharekaho in Mercury Bay

Left The view is to the east. Wharetaewa is the pā furthest out on the peninsula. Its large transverse ditch and bank described by Cook and Banks shows (just above centre) beneath two large pohutukawa and in line with the camera viewpoint. Several other pā including Wharekaho, not described by Cook, lie further inland along the cliff edge, bottom left. They may not have existed in 1769; alternatively, they may have been occupied earlier than 1769 and their wooden palisades had decayed or the earthworks were hidden in scrub at the time of his visit.

Right. Close-up view of the area of ditch and bank on Wharetaewa described by Cook and Banks. The interpretative drawing (below) shows how the key features such as the palisade and fighting stages relate to the archaeological features such as the ditch.

page 44 photograph, the natural slope defences can be seen to be very steep. Of some interest also is that there are now three pā on the same point, Wharetaewa being only the pā at the very tip of the point. It is possible that these pā did not exist in 1769, since Ngāti Hei, the hapū of this locality, were 'besieged in Whare-taewa' as late as 1800. 3 Clearly, care is needed not to assume that all the features presented in a surface view are the same age.

In the earlier part of this century, significant work was done to protect and record pā by S. Percy Smith and W.H. Skinner, both surveyors by profession. 4 They recorded accounts of traditions relating to pā, while other surveyors not infrequently recorded the location of pā in their field notebooks. The greatest historian of pā was Elsdon Best who published his book, The Pa Maori, 5 in 1927. Best's book is kept in print because it remains the most comprehensive and accessible source for the names and history of individual pā in many regions of New Zealand. His characterisation of the nature and function of defensive features is also still valid. A more recent work by Aileen, Lady Fox, Prehistoric Māori Fortifications in the North Island of New Zealand 6 provides a contemporary view of the range of pā in New Zealand, and brought (to the debate about function) useful insights from her extensive studies of Iron Age hillforts in England and Wales.

The tactical objective of the fortification was to prevent the attackers from being able to engage in hand-to-hand fighting, at the same time exposing them to spear thrusts through the palisade, or darts and stones thrown from above. Attacking forces could not get into the pā, or their entry was slowed down by the fortifications. The weapons used in pre-European Māori society were of a particularly devastating hand-to-hand type, although spears, darts and sling stones were also used. Clubs of both shorter reach (mere, patu and others) and longer reach (taiaha) were important; 7 they were described in some detail by Cook:

The Arms they use are Long spears or lances, a Staff about 5 feet long, some of these are pointed at one end like a Serjeants Halbard others are round and sharp, the other ends are broad something like the blade of an oar [taiaha]; they have another sort about 41/2 feet long, these are shaped at one end like an Axe and the other is made with a sharp point [tewhatewha]; they have short Truncheons about a foot long, which they call Pattoo Pattoos [patu], some made of wood some of bone and others of stone, those made of wood are variously shaped, but those made of bone ai stone are of one shape, which is with a round handle broadish blade which is thickest in the middle and tape to an edge all round, the use of these are to knock mei brains out and kill them outright.... 8

Such weapons would have been used at close quarte with timing, great muscular force and requiring psychological conditioning of both attacker and attacked.

As we have seen from Cook's description, pā he earthwork ditches and banks and wooden palisades I protect the occupants against attack. 9 The ditches and banks comprised a number of elements: single, doub and even triple lines; intricate entrance ways which e: posed the attackers to spear thrusts; and finally, to the rear of the bank and palisade, fighting stages: tall pla forms built above the rear palisade line, or between palisade lines, at key points of weakness. The defender could occupy these stages with relative impunity, and from them rain down darts and stones (and even water to make the ground slippery) on an attacking party.

As seen today, the earthworks of a pā usually consi of defensive ditches and banks surrounding the natural] steep summit of a hill or a crest in a ridge or, more rarely, cutting off a point in a river. The location of pā tends to follow the natural defensive advantages of the landscape ridges and the summits of hills offered many opportunities for defence, with the defenders taking advantage ( naturally steep slopes. Pā were commonly built in loc; tions where a sudden steepening or narrowing of a ridge reduced the labour of constructing a ditch and bank. In hilly country, particularly on the coast or by lakes, headlands or the ends of ridges could be simply defended by a single length of ditch and bank. In areas where a longer occupation might be expected, this could be elaborated page 45
Te Awamate, a swamp pā on the swampy margins of the Rangitikei River flood plain, Manawatū

Te Awamate, a swamp pā on the swampy margins of the Rangitikei River flood plain, Manawatū

In this winter view in light, overcast conditions, the pā is marked by the bare branches of young poplar trees. Surrounding the pā is a raupō swamp. The linear marks in the foreground are probably eel channels, dug in seasonally dry sands to concentrate the passage of migrating eels. The pā was occupied by hapū of Ngāti Apa in the late eighteenth and early nineteenth centuries. The pā is about 30 by 10 m in area; the view is to the south-east.

page break
Otumoana, Urenui River, Taranaki

Otumoana, Urenui River, Taranaki

The steep-sided, narrow ridge is defended by ditches, steepened scarps, banks and palisades (no longer showing) constructed across the ridge, often taking advantage of the naturally steep slopes. The pā may originally have been defended only by ditch, bank and palisade across the ridge. The pā lies in three main segments (O, M, K, on the interpretative figure, following Elsdon Best's illustration in The Pa Maori), divided by the two clear transverse ditches and banks. At some time after the initial defences were created, a ditch and/or artificially steepened scarp were created along the side of the ridge (showing on the west of platform 'O'). These connected with the defences running across the ridge.

into double or triple ditches and banks. Where the sides of a ridge were not sufficiently steep, additional defences would be created. These would be either a narrow terrace and scarp created along the side of the ridge, joining up with the ditch and bank that ran across the ridge, or an extension of the ditch and bank. Where the ditch is more or less continuous around the whole of the pā site it is known as a 'ring-ditch', although it is seldom perfectly circular. A defensive advantage was also sought by locating pā in swamps, and occasionally on artificially raised islands, many examples of which are known from the Waikato, Bay of Plenty and Manawatū/ Horowhenua regions.
In chapter 2, I stressed the importance of river or page 47
Katikatiaka, a headland pā on the high terrace north of Whitecliffs, north Taranaki

Katikatiaka, a headland pā on the high terrace north of Whitecliffs, north Taranaki

The site takes advantage of a point created by the intersection of the cliffs and a valley. It is defended by at least three, possibly four lines of defence. The interior of the pā is segmented, reflecting either social divisions or the need to be able to retreat from the outer defended parts of the pā to the inner. The view is from the north-west, taken in evening light.

coastal terraces in Māori settlement. In such localities, the edges of river terraces or the river banks themselves could be fortified. A narrow point formed between the main river and a side stream could be simply cut off by a short straight length of ditch and bank. The edge of a terrace could be enclosed by a rectangular length of defensive ditch and bank. A ditch and bank forming a dogleg in plan could be used to cut off a point where a simple transverse ditch and bank would be too long.
In recent years there have been significant original descriptions of the localities, defensive setting and forms of pā, notably those of Geoff Irwin on the Pouto Peninsula, North Kaipara Harbour, Bruce McFadgen at Tauranga, my work on the East Coast and at Whakatāne, and Nigel Prickett in Taranaki. 10 Such studies indicate considerable regional variation in the nature of pā. Irwin reached the striking conclusion that almost all the Pouto pā belonged to the few decades before 1820. Although this result need not apply to other regions, it suggests that warfare in Māori society might have been a spo- page break
A ring-ditch pā on the Iahar plain of the lower slopes of Mount Taranaki

A ring-ditch pā on the Iahar plain of the lower slopes of Mount Taranaki

Where the ridge or hilltop was rounded and not very steep, as on this lahar mound, the ditch, bank and palisade might encircle the whole of the hilltop, giving rise to the 'ring-ditch'. This type of pā is particularly common in Taranaki where there are soft, easily worked soils. At the time of occupation the flat country between the lahar mounds would have been swamp. Such ring-ditch pā also occur frequently in the Bay of Plenty and Northland.

radic but, once begun, fierce movement that swept up all the related groups of a region.

The number of pā is considerable—as many as 6,000 in total is a commonly accepted estimate. 11 There were as few as 100,000 people in New Zealand at the time James Cook's visit in 1769, although higher estimates have been made, 12 indicating about one pā for every 20 people at that time. Although some, perhaps a high proportion, would not have been occupied at that time, many of them would have been occupied at some point in the preceding 100 years, as I discussed in chapter 2. The number of pā therefore suggests an intense rivalry between groups as small as 20 to 30 people in the pre-European period.

Within the North Island, there are relatively few pā in the regions south of about Whanganui, including the lower part of the Wairarapa. 13 Those that do exist a sculpted to only a limited extent into the hill slopes. Hawke's Bay and on the East Coast generally north Cape Runaway, pā are common but they too are sculpted little into the ground surface. Soils on ridges in the regions are shallow, the rocks are close below the surface: as a result the sides of ridges may be steep, and need little fortification. 14 Pā on river terraces are more common on the East Coast than in many other parts with the exception of the Whanganui region and parts of the Waikato. This reflects the existence of large rivers with extensive and wide valley floors in these regions.

The regions with the highest density of pā and also the most elaborate and deeply dug defences are Whanganui, Taranaki, the Bay of Plenty, the Waikato, and Auckland and Northland. In the Bay of Plenty and Taranaki the hill forms are often rounded in profile, 15 for example on a lahar mound, and as a result the distinctive ring-ditch pā is common. 16 It was once argued that this similarity between the types of pā reflected a common cultural influence: people had migrated from Northland south into the Bay of Plenty and Taranaki, taking the ring-ditch 'style' of pā with them. 17 Today, it is generally accepted that common environmental factors, the presence of deep volcanic soils and softer landforms are the direct cause of this apparent cultural similarity. Adjacent regions can have very different styles of pā simply because of variation in the relative ease of working the soil.

page 49

Pā near Pakipaki, inland Hawke's Bay. The pā has strong lateral and transverse defensive ditches. A vertical aerial photograph of this pā is on pages 30-31.

page break

Pāwerawera, the pā of Māhaki, on the high Pleistocene terrace of the Waihuka River, East Coast. The pā has several defensive scarps on the steep slope to an old meander, and is effectively isolated as a terrace 'island' in the centre of the valley.

page 51

Storage pits on slump hillocks above the flood plain of the Whangaehu River.

page break

Pā with many kūmara storage pits, near Tūpāroa, East Coast.

page 53

Manuaitu, a pā on old dune country, Aotea Harbour in background. A ditch and bank fence cuts into the lateral defences at right nearest the camera viewpoint. A vertical aerial photograph of this pā is on page 118.

page 54

Middens on eroding dunes at Aotea north head. The middens show as white scatters; the brown patches are ash soils.

page 55

Rangihoua, outer Bay of Islands. The terraces of the pā are prominent, and horticultural trenches (some accentuated by tunnel erosion) run down the slope to the stream.

page 56

Trenches down lower slopes of a hill near Te Oihi (Marsden Cross), outer Bay of Islands. They were probably a form of irrigation for taro, grown in the groundwater trickling dowr the base of the trench. Kūmara were probably grown in the well drained ground between the trenches.

page 57
Pā with many storage pits near Tirau, Waikato

Pā with many storage pits near Tirau, Waikato

The density and regularity of kūmara storage pit construction is remarkable. The defensive perimeter is defined by ditch and bank. At either end of the prominent row of pits on the narrowest part of the ridge, at a slightly lower level and on the wider levelled parts, are house floors—large, shallow rectangular depressions distinctly different from those of the pits. The group that lived in a pā such as this may have been a single hapū with as few as 50 people, adults and children. Each pit (a minimum of 30 is visible) is about 4 by 2 m in plan and may have supported one or two people through the winter and early summer. A generally frosty inland climate (necessitating underground storage of the crop), a very warm summer growing season and large crops may explain the numbers of pits. The view is to the north.

Living in the Pā

The function of pā in the wider settlement pattern of a region or locality has been much debated. 18 The debate has been about whether the pā was a village for everyday living or a 'citadel' to which people went only when under threat. In times of threat, people may have gone either from an 'open' undefended settlement in a river valley or coastal plain to the pā on the hill country or at the edge of the terrace country, or they may have congregated there from more dispersed, smaller settlements, in the pattern described by de l'Horme in 1769 in Doubtless Bay. Some pā are clearly 'refuge' pā, being located far inland away from more closely settled areas. This pattern is common in Samoa or Hawaii; the relatively few fortifications there typically lie well inland from the page 58 coastal strip where the main population lived. In New Zealand such pā may have been a refuge for a hapū or whānau temporarily out of favour with more senior or powerful hapū of its iwi.

The size of pā varies greatly. Even where the actual boundaries can be determined, there may be obvious restrictions on the area. A cliffed hilltop would have been readily defended, but may have been uncomfortably small, or used only as a refuge for a wider settlement. Many pā are tiny (as small as 10 by 15 m) and have very restricted access to water. Would this have limited the number of people and the length of time they could stay in the pā? These factors have a bearing on the issue of whether pā were occupied from time to time when needed, or whether they were centres of settlement. What was the role of hilltop settlements as symbols of territorial occupation? We do not really know.

The number of people who may have lived in pā has proved to be controversial. Figures as high as 1,000 people have been offered for the very largest pā such as the Auckland volcanic cones. 19 On the other hand, the very small pā will have held social groups as small as a single whānau. The usual estimates based on measures of the size of pā (mainly the defended area) are probably too high. An added difficulty is that it is often difficult to determine just what is the extent of the pā. In the Bay of Plenty, it is common for there to be small pā, say 30 by 10 m in plan, which are clearly part of larger settlements. One hundred people packed into a pā of this size would have very little living space. In some cases, pā are obviously built or occupied with a 'concertina' effect in mind. There may be a central fortification in a system of ridges, with inner and outer fortifications. If the outer defensive line was broken through, the defenders could retreat to an inner line of defences. At the same time, running down the ridge towards the river, there will be a series of terraces extending over perhaps 400 m of ridge line. What is the function of these terraces— for houses, or gardens? Do the scarps (outer slopes) of the terrace represent additional defences?

A further difficulty lies in considerations of the fortified perimeter, the actual length of the defensive ditch and bank. How much work would it take to create a given length of such a ditch and bank? How much time did the inhabitants have to dig it? Was it all done in emergency conditions, or was it a more leisurely process? In emergencies, small numbers of people would explain quite large fortifications. More importantly, how much of the defensive perimeter had to have people behind it to be effective? What were the attacking tactics of their opponents? The answers to these questions are not known. Looking at the question more broadly, it is unwise to apply our modern levels of comfort, inability to withstand privation, and lack of sheer physical capacity for work and endurance to the people who built and occupied pā. In pre-European times, life could be very tough indeed. As a general rule, then, it is advisable to 'think small' when considering the numbers of people occupying any single fortification even if it appears to require large numbers.

Within the pā there would have been storage pits and houses. Communal open spaces ('marae') were apparently rare in pre-European sites, the only known examples being from a Waikato swamp pā 20 and a pā recorded in 1772 by the French on Moturua Island in the Bay of Islands, although smaller open areas are sometimes recognisable. 21 The surfaces of some pā are filled with closely spaced storage pits, suggesting that pā in some areas may function simply as defended foodstores. 22 Houses or more strictly house floors are relatively difficult to detect from the surface. 23 House floors are marked by shallow (less than 40 cm deep), rectangular depressions, and may be mistaken for storage pits in some cases. House floors tend to be more regular in their rectangular outline and have a distinctively flat bottom. In the central North Island, where they are common, the depressions are shallow, fairly large (5 by 3 m in plan), and have a distinctive rim around only three sides. The rim was an earth wall built against the exterior walls and beneath the low eaves of the house. The rimless side indicates the location of the porch. Such houses will show in aerial photographs if lighting and ground conditions are right.

From the 1830s there was a steep decline in Māori population and many hill country settlements were abandoned for the trading advantages of areas, generally flatlands, close to European anchorages and trading points. This was a far-reaching change in settlement pattern, but it has been too mechanically reported in some accounts. There was extensive pre-European settlement on flatland, for example, on the East Coast, and hills remained occupied after European contact. One hill pā known to have been occupied in the 1820s and 1830s, Rangihoua in the Bay of Islands, illustrated in chapter 16, appears not to have used its earlier pre-European fortifications, 24 but was still occupied because it was near page 59 a mission settlement. The technical detail of the debate about the use of hill country versus that of lowland settlement need not concern us here. 25 There is no reason why a mixture of such settlement was not possible with a different pattern, more dependent on lowland settlement, beginning at the time of initial European settlement in the 1830s. From that time, depopulation caused by disease, coupled with the advantages of trade in settlements accessible to the anchorages of European schooners and co-location with the rapidly growing European settlements, were powerful influences and few Māori settlements remained on hilltops.

Pā—age and origins

Te Rangihiroa, Sir Peter Buck, author of The Coming of the Māori, 26 the standard 1940s account of Māori migration and history, wrongly thought that pā were invented in New Zealand. Les Groube and Roger Green have investigated the Polynesian origins of pā. 27 Roger Green's conclusions about the wider Polynesian origins of pā are now generally accepted by archaeologists. He concluded that pā were in universal use throughout Polynesia, and would not have been invented or independently developed in New Zealand. Similar fortifications were well known in the nineteenth century, for example in Fiji, where European colonial forces were often forced to attack them. From archaeological sources, they are known in Fiji, Rapa, Hawaii, the Marquesas, Samoa and Tonga. 28 They are not known in the Cook Islands, nor Tahiti, nor the wider Society Islands. The use of fortifications in the nineteenth century and earlier in Fiji is well documented. An example of a fortification is the 'square ring-ditch' fortification, named Navuso, on the Rewa River delta, Fiji. 29 The photograph shows the pā lying on the higher flood-free bank of the river amongst a modern crop probably of sugar cane. Despite the greater antiquity of settlement and population densities of any of these islands compared with New Zealand, fortifications are not as frequent in the landscape as they are in New Zealand.

Roger Green argued that the similarity between Fijian, Tongan, Samoan and New Zealand fortifications resulted from a common ancestral tradition of warfare from or within fortifications. The absence of such fortifications in the Cook or Society Islands appears to be an anomaly, because they have the greatest cultural affinities of any Polynesian society with New Zealand Māori.

On the other hand, other relatively close neighbours of New Zealand and the Society Islands, such as Rapa in the Austral archipelago, 30 are famous for their hilltop fortifications.

Were pā present from earliest times in New Zealand, or were they a response to population pressure and conflict over resources later (as is sometimes assumed) in the pre-European period? Theory suggests that the conflict may be about resources, but as we have also seen it may be about conflicts over mana expressed in the control of abundant resources. If the latter, we would expect pā in earliest times in New Zealand. Much of the argument revolves around the empirically established age of pā compared with other forms of archaeological site. If it could be conclusively proven that the oldest pā were younger in age than other types of site, for example middens, then a case could be made for independent invention of pā in New Zealand—an empirical problem without an easy resolution.

The dating of fortifications by radiocarbon is not a simple task. In the case of a meal of shellfish or moa, the animal has died not long before it was eaten, and the refuse of the meal itself (shell, bone) is 'directly' dated. Such direct dating is not possible when taking radiocarbon samples from the fortifications of a pā. Instead charcoal or shell from layers either preceding or sealing off (later than) the fortification must be dated. This is not always possible because of the way layers are created in archaeological sites and revealed in the course of archaeological investigation. The date is invariably either before or after the feature needing to be dated. Furthermore, pā are often reworked, particularly in favoured locations, destroying earlier evidence of age of a more direct kind. The case for an earlier age for fortification can be made indirectly, however. Where pā have been dated satisfactorily—all too rarely—there are occasionally dates very much earlier for features of the site not positively related to fortification. It is not unreasonable to accept that these are related to settlement on a spot chosen in part for its defensive potential. A recent critical review of the radiocarbon data has settled on an age of about 1500 (i.e., 450 years B.P.) for the earliest pā, 31 some 300 or more years later than first settlement of New Zealand.

Defensive form has also played a role, historically, in the debate about pā origins and has never been satisfactorily put to the test—or to rest. Elsdon Best's theory, sometimes revived but generally not favoured, was that page break
Navuso, a 'square ring-ditch' pā on a levee of the Rewa River, near Suva, Fiji

Navuso, a 'square ring-ditch' pā on a levee of the Rewa River, near Suva, Fiji

Three lines of defensive perimeter, square-ish in plan, surround a former village site. The innermost defended area is about 80 m across. The complexity of the fortifications is greater than that of New Zealand pā, but the concept of an area defended by earth ditches and banks is the same. The western ditches were probably water-filled; a drainage ditch runs from the outer defensive ditch south-west to the meandering stream on the south-west flank. The Rewa River is to the right and the fortification can be seen to be on the slightly elevated levee of the river, the outer eastern perimeter taking advantage of the natural terrace rise from the river.

the earliest pā were defended solely by the scarps of terraces. 32 He thought terraces were simple precursors of more elaborate ditch and bank fortifications. The argument gains purchase by the fact that both the front slope and rear of a terrace, the 'scarps', are steeper than the general run of the slope into which they are cut, giving them some defensive value. Pā such as Ōtātara or Hikurangi near Taradale in Hawke's Bay (see chapter 11) have very small ditch and bank defences and were deemed to be earlier than the other pā with fuller ditch and bank systems.

Chapter 2 reviewed the ways in which human settlement is manifested in the historical landscape, and the importance and difficulty of recognising the pattern of smaller settlements, all too often neglected in field studies, were stressed. In this chapter, we have considered the largest units of settlements—pā and their associated features. As James Cook so clearly described, pā were constructed physically to keep out attackers from the defended area and its inhabitants. Māori had elaborate techniques for constructing ditches, palisades and fighting stages, and carefully selected places with strong natural defences and good outlooks to reduce the risk of surprise. The relationship of pā to the smaller settlements is not only difficult to depict in aerial photographs, but also virtually impossible to establish at a particular point in time. The origin, age and function of pā in settlement pattern have been a perennial theoretical problem in New Zealand archaeology. We have useful results from aerial photography about the defensive features of pā, their relationship to surrounding country, and in rare cases the arrangement of settlement within them.

1 Cook (1955); Banks (1958); Kennedy (1969).

2 Cook (1955: 197-199). Note that Banks (1958: 76) does not record a double ditch.

3 In Cook (1955: 200, footnote); see also Kelly (1953).

4 They created many historic reserves incorporating pā.

5 E. Best (1927).

6 Fox (1976).

7 Golson (1959); Davidson (1984: 100-103).

8 Cook (1955: 200).

9 See also Fox (1976); Davidson (1984: 181-194).

10 Irwin (1985); Prickett (1980); 1982a; 1983; 1990); McFadgen and Williams (1991); O'Keeffe (1991); K. Jones (1989b).

11 Fox (1976: 9).

12 Davidson (1984: 56-59).

13 Using the strict archaeological definition of a site with artificial defensive earthworks.

14 K. Jones (1989b).

15 Compared with other regions of New Zealand such as the East Coast.

16 Prickett (1980; 1982a).

17 Groube (1970); see also contrary view of Fox (1976: 21).

18 Groube (1964); Davidson (1984: 162-170).

19 Fox (1983). E. Best (1927) also offers various figures up to 1,400 people for pā in the inland Bay of Islands.

20 Bellwood (1978).

21 Kennedy (1969); Davidson (1984: 161-163).

22 Law and Green (1972).

23 B.F. Leach (1979); Prickett (1979; 1982b); Marshall (1991).

24 Davidson (1982a: 15).

25 Pool (1977); Groube (1964); Ballara (1979).

26 Buck (1950: 138).

27 Groube (1970); Green (1967).

28 Green (1967); S. Best (1993).

29 Parry (1977: map).

30 Hall (1875: 82). The Australs are south-east.ef the Society Islands.

31 Schmidt (1993).

32 For example, E. Best (1927: 293) on Heipipi, Hawke's Bay.