Volume 66 1957 > Volume 66, No. 1 > Field archaeology in New Zealand, by J. Golson, p 64-109
FIELD ARCHAEOLOGY IN NEW ZEALAND 1
FIELD WORK in archaeology consists of two branches: excavation, the recovery of the historical information that lies hidden in the ground, and field archaeology, the recording of the historic remains that are visible on it. This duality is well reflected in the titles of two recent text books on archaeological research: Crawford's Archaeology in the Field and Wheeler's Archaeology from the Earth. 2
Excavation is the basic technique of archaeological research. By recovering the surviving paraphernalia of past cultures in context it provides the framework for the ordering of archaeological evidence and sets the conditions for its interpretation. Field archaeology is a necessary preparation for and an essential adjunct to the work of excavation and provides for its practitioners a wide and important field of research without recourse to digging.
As defined by Crawford the subject data of field archaeology are the remains of the past that are visible on the surface or are indicated by superficial remains such as potsherds, flints, soil discoloration or the growth of crops and the task of the field archaeologist is their discovery, observation and recording. 3
Each site discovered is obviously a potential site for excavation and the choice of the ideal site for an excavation designed to answer specific problems can be the work of field archaeology on the highest level. Thus the excavations at the site of Star Carr, Yorkshire, 4 were the culmination of a plan to answer certain problems connected with the early postglacial settlement of England, in circumstances that would ensure the completest possible survival of the organic content of the cultural groups involved. Attention was therefore concentrated on areas where waterlogged peat deposits fulfilling the conditions for preservation were- i
- ii Page is blank- 65
present and the actual site chosen for excavation was revealed by the presence of distinctive flint tools in the bank of an artificial drain.
The discovery of evidence in the field is more, however, than the provision of sites for excavation. Field archaeology gives a distributional dimension to the information assembled by excavation at a single site. Archaeologists write the prehistory of any region in terms of a sequence of cultures, each characterised by the possession of distinctive items of material equipment. Sites identified in the field can sometimes therefore be attributed to their place in the culture sequence simply by the occurrence of distinctive artefacts whose cultural and chronological significance has been previously established by excavation. In New Zealand Duff's excavations on the Wairau Bar have isolated the material equipment characteristic of any early stage of Polynesian settlement, distinguishing that stage from the classic Maori culture seen by Captain Cook. 5 The frequent discovery of distinctive types of adze and ornament on eroding sand dunes from one end of the country to the other enables us to see the presence of Duff's moahunters and their archaeologically reconstructed way of life in areas where no excavation has been done. Field work based on the data of excavation therefore gives archaeology geographic information which the excavation itself does not provide and enables conclusions to be drawn in terms of habitat, subsistence activities and the like when the recorded sites are plotted on topographical, geological and soil maps.
Field archaeology of this type has been highly developed in some countries. In the United States the extensive collection of pottery from surface sites has been a major feature of archaeological research in the Mississippi Valley. By these standards the ubiquitous collector of Maori artefacts in New Zealand is not a true field archaeologist. He employs, frequently with great success, many of the field archaeological techniques, but his discoveries are rarely recorded even in the simplest fashion and artefacts collected without details of location are scientifically valueless.
Thus far field archaeology has been discussed with reference to sites where artefacts come to light in the course of the natural processes of erosion or of the activities of animals and men. The same considerations apply, however, to that large and important class of archaeological evidence, the visible monuments of the prehistoric past that have made a well-nigh permanent imprint on the landscape. Field monuments are artefacts as much as tools and weapons and the same principles of study are applied to both. The archaeologist is interested in their function and their form. Their function, the part they played in the society that made them, may be deducible from the nature of the surviving earthworks, from factors of locality and distribution, from contemporary descriptions, from excavations or from a combination of two or more of these. Their form, although, as with tools and weapons, partly dependent upon the function they were meant to serve, may to a considerable extent be determined by the particular traditions of the social group responsible for them. Not all prehistoric groups in - 66 England practised burial beneath piled up mounds of earth but amongst those that did morphological differences in mound structure often betray the presence of quite distinct social groups sometimes at great removes both in time and culture. 6 The final test of the cultural significance of earthworks such as these is excavation. But an earthwork once investigated and checked by excavation, some of the resultant information may be assumed for identical earthworks identified elsewhere. Thus the distinctive causewayed camps of southern England have been shown by excavation to belong to the earliest peasant culture of the British Isles. 7 Their surface features are so unmistakable that further examples can now be attributed, without excavation, to the people responsible for them and the record of their field occurrences gives additional information about the people in question and the range of their activities. 8
It is obvious that in order to apply excavated data in this way we must have a classification in terms of common characteristics for field monuments just as for the tool and weapon category of artefact. This the present study attempts to do for New Zealand.
Classification provides the mould in which the observations of the field archaeologist can be cast, but it is in observation itself that many of the difficulties appear. As Crawford points out, the historical document which is the cultural landscape has not only been written but erased and written over many times. 9 The processes of nature smooth the contours of banks and fill in ditchs, covering both with vegetation. The activities of succeeding generations often carry these processes to the point of obliteration; in temperate Europe a prime agent of destruction has been the plough. The skill of the field archaeologist lies in his ability to detect the hand of man in the most negligible irregularities of contour and to relate the fragments to the structures of which they are the surviving parts. His technique consists of the frequent reexamination of a suspected area under different seasonal conditions and at different times of day. The presence of virtually obliterated mounds and banks can often be detected by the long shadows caused by the low light of the rising or setting sun. 10 Silted up ditches will sometimes be betrayed by the taller, darker crops growing above the moister and more fertile soil of the silting. 11
These techniques of the archaeological field worker have been immeasurably extended by the development of aerial photography. 12 The aerial view not only gives clarity and unity to what to the field archaeologist is often a blurred and fragmentary jumble of earthworks, but it also discloses sites which to the ground observer are not - 67 visible at all. The majority of these latter have been in the form of soil and especially crop sites. The discoveries made in England over the last two generations have been numerous and important. Amongst other things the balance so long weighted against the survival in the archaeological record of the organic component of past cultures has been righted to a small but significant degree. The timber circles 13 that partnered Stonehenge and the wooden farmsteads of the lowlands 14 that matched the stone huts of moorland England have been found because the filled-in post holes and the silted ditches have left their precise pattern in the darker green of more luxuriant wheat growth.
Field work itself has been streamlined with the recruiting of the aeroplane to archaeology. Especially in barely populated or infrequently visited areas aerial reconnaissance for archaeological purposes or the study of aerial photographs taken for other needs has lightened the burden and increased the output of discovery. The archaeological interpretation of aerial photographs is, of course, a task for those with field experience and to avoid mistakes the site discovered from the air needs checking on the ground.
The classification of New Zealand field evidence offered here is in the main functional. This functional approach is possible, despite the fact that the body of excavated data is negligible, because some of the types of site and monument to be described were in active use when Europeans first visited New Zealand and during the early years of European contact. Contemporary writers not only describe the role of the more permanent features still diversifying the landscape today. They record the nature of the impermanent features of sites which in other countries it is the laborious business of excavation to reconstruct:and they give evidence of the living role of monuments in the contemporary culture in ways at which archaeological evidence can barely hint. Ethnologists in more recent times have had occasion to criticise these early writers for the generalities and frequent vagueness of their descriptions and especially for their failure to record the common things they saw. 15 However, sufficient contemporary information has been recorded on some topics for us to be able to interpret our field evidence in its light.
Deficiencies of information from the European side have in the past been made up by data from Maori informants. The value of such evidence varies considerably 16 but it has proved of some importance in the interpretation of the ritualistic monuments of prehistoric New Zealand.
Sometimes contemporary information is wholly lacking and the material has been organised typologically.- 68
Within each functional or typological class the field evidence is summarised, variant forms described and some data on distribution given. So little excavation has been done on the relevant sites that it is not possible to set many of the monuments described into a culture sequence which itself, because of the nature of the prehistoric settlement and the character of the surviving evidence, is only clearly defined as yet in its earliest and its latest stages.
The field evidence reviewed consists both of field monuments and of sites commonly exposed by natural action. The data have been supplied by the author's field work and by scattered information, largely of a vague descriptive character, culled from a variety of sources.
The value of these latter has in many cases been slight. Field evidence has not gone unnoticed in New Zealand ethnological literature but has rarely been accorded investigation in its own right. The descriptions attempted of field monuments have been largely unsystematic, ambiguous and inadequately provided with plans and photographs. None of these strictures applies to Elsdon Best to whom we owe the only really detailed monograph in field archaeology that New Zealand has yet produced. The Pa Maori, 17 despite the often wayward organisation of the material, discloses Best as a field worker of uncommon perception. He describes fully every feature of his sites, gives measurements and provides plans and cross sections. He is conscious of the importance of variations in the form of monuments of the same general class and notes regional differences in this respect. Towards the end of his life Best turned his attention to the field study of coastal middens in the Wellington district. 18 This work, which is a model of its kind, has been extended by Adkin 19 and Palmer, 20 with the result that the Wellington coast, superficially so poor in field evidence, is one of the best served areas from the field archaeological point of view. 21 In the South Island comparable work has been done in Otago and Southland particularly by Teviotdale and Lockerbie, though here the main emphasis has been on excavation. 22
The paucity of similar studies is somewhat surprising in view of the ubiquity and variety of the field evidence in New Zealand. The present classification is designed to provide a framework for field recording and to suggest lines along which research might be directed. The data to be collected are just as important for the study of New Zealand prehistory as the adzes and ornaments in museum collections. The field record holds evidence for prehistoric patterns of population distribution and the extent of environmental determinants on them. The classification of field sites and a much fuller typological study of them than is possible at present will provide vital evidence for comparisons with the Polynesian homeland, for the assessment of the adaptations of the ancestral culture made by the Maori in New Zealand - 69 and for the study of cultural differentiation developed regionally within New Zealand itself.
Unlike adzes and ornaments which can be taken away for study in museums, field monuments are not portable and must be studied on the spot. 23 This is the work of many hands. The success of field archaeological research depends on the informed and active interest of local archaeologists and their associations.- 70
1. Sites of Defence.
The Maori pa is the most ubiquitous, most striking and best known of prehistoric monuments in New Zealand. Examples were seen and described in full operation by early visitors, 24 at Mercury Bay for instance by Cook and his companions on the first voyage (1769), 25 and at the Bay of Islands by the Frenchman Crozet (1772) 26 and later by the missionaries Marsden and Nicholas (1814-15). 27 By the time of the missionaries, however, the musket had made its baleful entry into tribal warfare and the descriptions we have from the following decades are largely of pa modified in certain respects to meet the requirements of the new weapon. Maori construction and experience in defence found a final and most effective expression in the fortifications erected in the wars against the Europeans, heralded by the war in the north associated with the Nga-puhi chief Hone Heke (1845-6) and closed by the campaigns against Te Kooti (1868-72). To the same troubled period belong the stockades built by the Europeans, which, in so far as their imprint is on the countryside today, are part of the field archaeology of New Zealand.
Because of their numbers, outstandingness and importance, pa sites have received more attention than any other type of field evidence in New Zealand. In the main this has been a by-product of the study of tribal history. The result has been the identification in the field of many named pa important in Maori tradition, rather than the study of sites in their own right. 28 General papers relating to the field evidence have been contributed by W. H. Skinner 29 and Raymond Firth, 30 but by far the most important work has been done by Elsdon Best. 31 Best adduces contemporary European evidence and material from Maori informants to reconstruct the features and functioning of the pa but he adds to this an exhaustive analysis of a large number of sites studied in the field. No significant development of this work has taken place since the study was published. Good surveys of individual sites have been made by Firth near Auckland 32 and by Delph and Archey in the Wai-kato. 33 Kelly has done a series of interesting and important little descriptions of pa sites famous in connection with early European contact or Maori tribal history without, however, employing the classificatory framework provided by Best. 34- 71
The consequence is that really full material for a comparative survey of pa sites is available to the writer only for northern Taranaki (Best), Auckland city and the Bay of Islands (Best and fieldwork) and parts of the Coromandel peninsula (fieldwork). Except for individual sites the Waikato, the Bay of Plenty, the East Coast including Poverty Bay and Hawkes Bay, and the Wanganui district are ill-known. Wellington is a little better served, but the South Island is a virtual blank.
The general features of Maori fortification in the musket era have been dealt with by Best, 35 with the aid of good contemporary descriptions. In his exhaustive study of the Maori-European wars Cowan has treated in detail the fortified sites in use during the period. 36
THE PRE-MUSKET PA.
Like the hill forts of Iron Age England, the Maori pa was essentially a tribal citadel situated wherever possible in a position of natural strength and equipped with artificial defences to supplement the natural ones. It is usual to contrast the pa with the kainga, the open village sited for considerations other than defence and provided at most with a timber palisade. Archaeologically this is a convenient distinction since the field expression of both will in most cases be different; indeed in the case of the kainga often non-existent. A difficulty has been suggested by Best for the Wellington district where the scarcity of fortified sites may reflect not merely low density of population and the relative security of the region from hostile attention, both arguable on other grounds, but also the impossibility of constructing earthworks because of the rocky nature of the terrain. 37
Whatever the archaeological convenience of the distinction between kainga and pa, the social implications of that distinction are less easy to assess. It is probably axiomatic that inhabitants of unfortified villages would have a pa at hand to which to retire in times of danger. 38 At the same time it is usually taken for granted that some pa at least were permanently inhabited. 39 Information on this point, as on many others in New Zealand culture history, will undoubtedly be provided by a reinterpretation of historical and traditional evidence in the light of archaeological interests and considerations. At the same time some light can be thrown on the question by the record of the field occurrence of fortified sites in respect of size, complexity, density and location.
Difference of status is obvious between the hill fort of Maungakiekie (One Tree Hill), Auckland, with its 100 acres of earthworks and subsidiary features, and the small pa near New Plymouth recorded by Best 40 with its single line of defence enclosing an area 25 yards by 12. Similarly the refuge pa of Maru, 41 high up in the bush fastnesses on the western slopes of Mt. Egmont, is in a different category from the - 72 elbow-jostling forts of the Auckland isthmus where defensible positions were contiguous to plentiful and varied food resources. In areas such as the latter, though we have as yet no evidence as to dates of occupation beyond that provided by traditions and genealogies attaching to particular sites, the multitude of pa would lend support to the view that some of them at least were permanent centres of settlement. Certainly their numbers give evidence of the denseness of the prehistoric population in certain favoured regions as also of the prevalence of warfare. No distribution maps of pa sites can yet be drawn to define these areas in detail but the general evidence points to the Taranaki coast, the Bay of Plenty, particularly in the area of Tauranga and Whakatane, the Auckland region and the North Auckland peninsula. Upwards of 35 pa are to be found within Greater Auckland alone. 35 are listed by Diamond for the Waitakere Ranges, the majority strung down the 16 miles of coast-line from Muriwai beach in the north to the Manukau Heads in the south. 42 Best plots 26 pa in a small area round the village of Urenui, Taranaki, between the Onaero and the Mimi rivers. 43 Nearly 100 sites are known in the vicinity of Whakatane and the Whakatane Valley alone 44 and the majority of these have been plotted by Best. 45 The increasing infrequency of fortified sites as one goes south reflects the growing unfavourableness of the country for Polynesian settlement, climatically and to some extent topographically, until midway down the east coast of the South Island we pass beyond the limit of agriculture. 46 Lack of recorded data prevents documentation of this process.
Field Archaeology of the Pa.
The major consideration in the siting of a Maori pa was the strength of the natural defences in conditions of warfare where missile weapons were almost totally absent and limited in range when employed. 47 Artifice, stealth and surprise seem to have been, with siege, more effective and more commonly employed against pa than any form of direct attack other than incendiarism. 48 In these circumstances the two desiderata of defence were height and difficulty of approach. The need for height was purely local in the sense that it was required to give the advantage to defenders with long spears and stones against an enemy equipped with no better weapons, immediately beyond the outermost defence. In the pre-musket era a hill of greater altitude commanding a site, but at a distance of 200 yards and more, constituted no danger. 49 Thus the Ngati-maniapoto pa of Mangatoatoa,- iii
PLATE 1.- iv
Wiri Mountain (Manurewa), a volcanic cone near Auckland: a terraced hill pa with associated rubble walls.
PLATE 3.- v
This headland at Shoal Bay, Tryphena, Great Barrier Island, is cut off by a single, modest ditch and bank.
Small and simple headland pa at Sarah's Gully, near Mercury Bay, Coremandel Peninsula, with single lateral scarp.
PLATE 5.- vi - 73
Steep inner scarp of ditch across a spur on Mount Camel, near Houhora, Northland, forming part of the defences of a pa.
on the Puniu west of Te Awamutu, was commanded by higher ground immediately on the other side of the river. 50
Both the requirements of Maori defence were fulfilled to perfection and in countless instances by the specific features of the landform dominant in those areas of New Zealand where the pa was most needed, long low spinal ridges of uneven contour falling steeply and in numerous spurs into river and coastal flats. By far the greater number of Maori pa have upland sites. In certain areas of flat land, however, such as the Hauraki Plains and Horowhenua, other topographical features were pressed into service.
Since the topographical conditions are the constants of the defensive situation, the soundest basis for the classification of pa sites would appear to be in terms of topography. Best's classification, made in just these terms, 51 has been adopted with some modifications here.
The main distinction is between upland and flat land pa.
Class 1—Upland Pa.
Owing to the diversity of local situation it is necessary to distinguish three varieties within this class.
Class 2—Flat Land Pa.
This class exhibits a similar gradation from all round to partial protection by nature.
Variety A comprises sites like low islands in swamp and lake with natural protection on all sides.
Variety B is the equivalent of the promontory pa of the upland category in that natural protection on three sides is afforded by ox-bow meanders, the confluence of rivers and the like.
Variety C will include all those sites where the natural situation is less advantageous in point of defence and where the proportion of artificially to naturally protected perimeter is high.- 74
The artificial works which supplemented the natural defences consist of timber palisades, 52 ditches, banks and escarpments. These are widely known throughout New Zealand but their use varies. Best isolates four kinds of defensive technique, palisading only, ditch, bank and palisade, scarping and palisade, and a combination of ditch, bank, scarping and palisade. 53 He sees some degree of connection between the defensive technique employed and the defensive needs of the site in question, 54 but it is likely, as Best himself suggests at times, 55 that factors of regional and chronological variation also are at work. The data are hardly sufficient at present to document this but the evidence can at least be reviewed.
CLASS 1—UPLAND PA.
(figs. 1-5, plates 1-4)
This class sees the employment of every type of Maori defensive technique.
Variety A—Hill Pa (figs. 1-2, plates 1-2)
The specialised type of cliff, rock and rocky island stronghold is often so well protected by nature that the artificial defences consisted of palisading only. Some remarkable places of refuge, a number of which would not even require the addition of palisades, are recorded by Best for North Auckland. 56 Other hardly less spectacular sites are known from Taranaki 57 and the Waikato. 58 The lofty rock column of Paritutu and adjacent rocky islands just west of New Plymouth are excellent examples of the type. 59
Field evidence for the utilisation of sites such as these often consists only of the signs of habitation to be described later. In itself this evidence is often no different from that of the kainga or unfortified village, but the nature of the situation will in most cases enable the proper distinction to be made. It is likely that most sites of this character were temporary centres of habitation in times of danger. A different impression is certainly given by the descriptions of Waimate and Te Namu pa in Taranaki but these sites were seen in the 1830's during the unsettled era of musket warfare. 60
Some of the less precipitous islands, however, resemble the generality of hill pa proper in requiring artificial defence additional to stockades.- 75
The earthwork defences of the hill pa consist of scarps associated either with terraces or ditch. 61
1. The Terraced Hill Pa (fig. 1, plates 1-2)
In this category scarping, the artificial steepening of the hill slopes, is, with palisading, the only defensive device employed. The terraces are areas artificially levelled into hillside or on hilltop for purposes of storage and habitation. The practice of arranging the areas of habitation to serve the needs of defence is a particular feature of the Maori pa and the complexity and ingenuity of the arrangements made is nowhere better illustrated than in the class of fortification under discussion. Though desirable from the point of view of defence the isolated hill was often initially unsatisfactory from the point of view of habitation. The hilltop was frequently rounded or broken in contour and often provided insufficient room for the community which it was the aim of fortification to shelter. Artificial levelling of the topmost areas, therefore, took place to provide platforms for settlement. These vary in shape from square to rectangular and in size, depending on the original contour of the hilltop. The arrangement, besides preventing useless toil, was of defensive advantage in that each area became a separate unit of defence, protected by scarps 62 and palisades.
The levelling was taken as far down the sides of the hill as was compatible with the population of the pa and the requirements of defence and resulted in the formation of terraces of varying length and width, separated by scarps of varying height and steepness, which give the hill pa its distinctive stepped appearance. Each was protected frontally by scarp and palisade 63 and backed by the scarp to the next terrace; laterally there might be a break to a terrace at a higher or a lower level or the terrace might merge into the natural slope of the hill. Ranged one above the other, they provided with the segmented summit area defence in considerable depth. Innermost within the defences would be the tihi, the citadel of the pa and the abode of the chief. The complexity of the arrangements for its protection can be no better illustrated than by reference to the North Auckland pa of Nga Puke-pango at Ohaeawai 64 (fig. 1B), where on one side as many as seventeen terraces are ranged above each other up a spur leading to the cunningly disposed summit platforms.
The steepness of the scarps between the terraces depended on the nature of the excavated subsoil 65 and in instances the slope was virtually- 76 - 77
perpendicular. 66 Their height was governed by features of topography and the number of terraces required. 3 ft. scarps are found on the gentler slopes of One Tree Hill, Auckland, 67 but most scarps are more than 6 ft. high and in some cases attain 40 68 and 50 ft. 69 The width of terrace varies from 6 ft. in examples on Taylor's Hill, Auckland, and 10 ft. recorded by Best on Pouerua volcanic cone pa, Ohaeawai, North Auckland, 70 to 45 ft. on Maungaturoto, a similar pa in the same area, 71 and, exceptionally, 45 yards on Mount Eden, Auckland. 72 Width is evidently related to the natural slope of the hill, but it is interesting to note that one terrace excavated on Taylor's Hill was partially built up to give extra width. Length varies in like manner. Best records a short terrace of 10 ft. on Pouerua 73 and terraces of 80 yards on Mount Eden. 74 Some hill pa show terraces even longer (plate 2). Limitations of terrace length are not entirely, at times not at all, due to topographical considerations. In discussing the volcanic cones of the Ohaeawai region and the Auckland isthmus Best comments on the frequently discontinuous nature of terrace formation, characterised by abrupt changes in level presenting steep lateral defensive scarps at the point of transition. 75 He suggests that the aim was to increase the obstacles to attack by segmentation into a large number of units of habitation and defence. The practice is particularly well illustrated by the pa on One Tree Hill. Occasionally, where the transverse break in terrace level produced a scarp of insignificant height, the deficiency was made good by an artificial bank, today 3 to 4 ft. high, above the scarp and in one case by a small ditch at its foot as well. 76 Best records the occasional use of banks and ditches as secondary features on terraced hill pa, largely to divide the summit areas on volcanic cones in North Auckland and the Auckland isthmus, where only a short stretch of earthwork was, because of the topography, required. 77
Within the general class of terrace, however, it seems possible to isolate a variety of small terrace formation somewhat different from the highly integrated interrupted formations described above. In form these are half moon-shaped platforms into the hillside, never very large, straight at the front of the platform, a rear curve formed by the scarped hillside rising above. Without doubt they served as areas for storage and habitation and their presence on the same pa as the more developed terraces may mean that they constituted only a slightly variant form - 78 in the integrated terrace complex. On the other hand there are sites where they are the predominant feature. Best quotes the hill pa of Mangere, Auckland, as an example where such small platforms, irregular in size and arrangement, abound, though he admits that the prevalence of rocky outcrops may be the reason. 78 However, he attempts to link the type with two traditionally ancient pa of the Napier district, Heipipi 79 and Otatara, 80 both characterised by a similar irregular disposition of small terraces or “linchets” as Best terms them, 81 and with a number of pa in the Waiapu Valley with the same features, which are reputed to be of ancient origin. 82 For Best these would constitute the beginning of a typological series culminating in the developed terraces of the Auckland series, with the earthworks of Mangere forming an intermediate stage. 83
Our present interest in them is that at times their occurrence may be more realistically explained as a device to provide flat ground for construction than as necessarily a part of the defensive system as such.
The type of terraced pa has been largely described in terms of its most developed examples on the Auckland isthmus and in the Ohaeawai region of North Auckland. Similar and less complex examples are, however, known elsewhere. They have been recorded for Whakatane, 84 Taranaki, 85 and Te Kuiti, 86 and seen in the Coromandel Peninsula. They are most numerous, however, in the districts from Auckland City north. 87
2. The Ditched Hill Pa (fig. 2A).
This category of hill pa is characterised by the use of the ditch as a major defensive feature, not transversely over a negligible length, but encircling the whole hill or as much of it as was necessary.
The type might be illustrated by a diminutive pa near New Plymouth recorded by Best. 88 The summit of the hillock appears, from Best's section, to have been flattened. The upper slopes of the hill have been scarped and the scarp terminates, not in a terrace, but in a ditch with outer lip rising 7 ft. above the ditch bottom. These defences continue round the curved side of the hill to impinge upon a bluff at either end. Best describes the outer lip as a rampart and the context shows that artificial construction is implied. In this example as in others it would appear likely that the outer lip is a baulk of the original- 79 - 80
hillside left undug during the construction of the ditch, heightened by the material derived from its digging.
The features here described are reproduced elsewhere. On Matuaaewe pa at Otumoetai, overlooking Tauranga Harbour, a steep knoll on the coastal ridge has been virtually encircled by a ditch 15 ft. wide halfway up its slopes. The hillside above the ditch has been scarped to a steep angle and the summit area has been levelled into two platforms, one irregularly rectangular in shape, flanked on three sides by the other 4 to 8 ft. lower.
On a neighbouring site overlooking the Wairoa River near Te Puna a ridge peak has been treated in similar fashion (fig. 2A). The summit area, stepped into three broad terraces separated by scarps 4 to 6 ft. high, falls 20-25 ft. by a steep scarp into an encircling ditch provided, on most of the circuit at least, with a heavy built-up rampart on the outer lip. The ditch is so placed that it serves to cut off the knoll at the level of the ridge spur above which it rises. A downward trending spur to the north and a small side spur to the east have been stepped into small terraces which may be subsidiary habitation areas outside the defences or, less probably, integrated into the main defensive system. The behaviour of the ditch is interesting because at one point, for a space of 20 yds., it levels out into a simple terrace. This type of change from ditch to terrace has been recorded by Best for Taranaki and will be discussed at a later stage.
The ditched hill pa seems to be of frequent occurrence round Tauranga Harbour and on the small volcanic hills in the neighbourhood of Pungarehu, Taranaki, but information on other parts of the country is too meagre to extend the distributional survey much further. It has not been noted in the field in Coromandel or North Auckland, though Karewa Ki Runga pa, Hokianga, would appear from McDonell's description to be of this type. 89 The unsatisfactory description of Te Ika a Maru pa, Wellington, may conceal a further example. 90
Since there does exist in New Zealand a type of stronghold wholly terraced and another wholly ditched, it may be that two quite distinct traditions in defence met and mingled during the course of New Zealand prehistory. On the other hand it is equally possible that the two defensive techniques co-existed within the same culture, to be employed, alone or in combination, as circumstances of topography and the density of population required. Only extended fieldwork will provide the answer.
Whatever the ultimate explanation, there are hill pa where the two traditions or techniques are employed together, at times apparently interchangeably. On the pa of Otahuhu (Mt. Richmond), Auckland isthmus, a lateral ditch 91 of considerable length is a quite exceptional adjunct to the typical scarped defences of this volcanic cone. In his description of the Ramaroa pa, Poverty Bay, Halbert mentions “a semi- - 81 circular wall of earth, surrounded by a deep ditch” with terraces above and below. 92
More interesting is Okoare pa near New Plymouth on its isolated, elongated hill 93 (fig. 2B). The shorter ends are defended by scarps, the slope on the northern side by terraces and scarps. On the southern slope the defensive scarp falls steeply into a ditch over 130 yards long with an outer rampart. Above the high steep inner scarp stands a built-up parapet, in places 5 ft. high when seen by Best.
The summit area of the pa is particularly interesting. It is divided into four segments, at slightly different levels, by the construction across the hill of three rather massive parallel ditches flanked by banks on both sides. This is a variant method of subdivision to that by scarps described in connection with terraced hill pa, but the purpose was the identical one of providing independent units of defence integrated within the complete fort. The reason here may have been that the topography of the hilltop made scarping impracticable.
The new features here described, lateral ditches with inner scarps surmounted by a parapet of earth and transverse ditches and banks employed to subdivide the habitation area and provide defence in depth, are commonly employed on the more complex ridge and promontory forts, to which the elongated form of Okoare pa bears some topographical similarity. The distribution of the features concerned will not, therefore, be described until the sum of their occurrences has been reviewed.
Another topographically intermediate pa is Whare-kaho at Mercury Bay, sketched and briefly described by Kelly. 94 The site occupies a hill top abutting the cliff edge and forming a slight wide-based point into the bay. The hill top has been scarped for habitation and cut off on the landward side by a long ditch across the base of the point. The site might therefore be alternatively classified in, and certainly serves to emphasise the connection between, the ditched variety of hill pa and the category of promontory pa now about to be discussed.
Varieties B and C—Promontory and Ridge Pa (figs. 3-5, plates 3-4).
These varieties may be treated in the same section since the type of promontory pa consisting of ridge spur or ridge end easily passes over into the ridge pa category, while certain problems of defence are common to both varieties.
The simplest form of promontory pa consists of a fosse cut across the neck of land forming the only easy approach to the site and backed by bank or stockade or both (plate 3). Ideally nature provides adequate defence, possibly supplemented by palisades, for the remainder of the pa in the form of cliffs, bluffs or precipitous slopes into sea, lake, swamp, river or flat ground. The archetype of the ridge pa has natural - 82 defence provided laterally by steep hill slopes and artificial defence frontally and in the rear by fosse, bank and palisade. Most of the sites that fall into these classes, however, are characterised by a number of elaborations of defensive equipment.
In the promontory class the basic type would be represented by the small pa on Stokes Point, Northcote, now obliterated by the northern foundations of the Auckland Harbour Bridge. The ditch on Stokes Point showed no indication of having been accompanied by an inner bank, even though the ground sloped slightly downhill into the interior of the pa. It was presumably felt that sufficient height was given to the defenders and sufficient obstacle presented to attack by the addition of a palisade to the lip of a ditch that could be dug deep and sheer into the soft sandstone subsoil.
A similar practice is to be observed on the promontory pa of Paeroa, Moturoa Island, Bay of Islands (fig. 3A), of which a plan, made during the ill-fated visit of Marion Dufresne in 1772, is reproduced by Kelly. 95 Here, however, the ground sloped upwards towards the interior of the pa so that the inner scarp of the ditch was heightened by the steepening of the hillside above it. It is notable that on the whole group of pa described and figured by Kelly from the small area of the Bay of Islands associated with the activities of Marion Dufresne not one fosse carries any trace of inner bank 96 (fig. 3A and C). This suggests a local practice.
A. The Bank.
Despite these examples, an inner bank, invariably sited directly above the inner scarp, was a common concomitant of a ditch, especially where, as on many promontory pa, the defended area was at the same level as or lower than the ground outside. Even in the case of ridge pa where low saddles were often selected as the point for defensive works and the ditch, cut at the base of or at some point up the slope, presented a considerable inner scarp to the attackers, an earthen rampart was sometimes added to heighten the slope. The practice resulted in a defensive arrangement identical with the lateral scarp with ditch at base and rampart at crest described for Okoare pa above. 97
An occasional feature is the presence of a bank on the outer as well as the inner lip of the defensive ditch. This is recorded by Best for Taranaki 98 and apparently for Poverty Bay. 99 The writer has observed- 83 - 84
the same practice on Popoia pa, Poverty Bay, 100 and on Great Mercury Island, Coromandel. In the few cases where information is available the outer bank is inferior to the inner one. 101 It recalls the built-up outer rampart described for lateral fosses on pa like Okoare. Perhaps the feature was in the nature of glacis designed to bring the attacker within range of the long spears wielded from the top of the ramparts or from fighting stages 102 set up behind them.
It is worth recording at this stage that palisades by no means invariably surmounted the earthen banks built by the defenders to increase the height of their made scarps. The evidence of Cook, Nicholas and Maori informants agrees on this point. 103 The absence of stockades was designed to provide room for manoeuvre on the part of the defenders. Best, apparently arguing from traditional rather than field evidence, mentions the presence of two kinds of bank in pre-European pa 104:
Certainly great and significant differences are apparent in the size and to some extent the form of earthen ramparts despite years of natural and animal erosion, but whether these observable differences fit the facts adduced by Best is at present uncertain. The pa near Herbertville described by Taylor White and mentioned by Best 105 has two ramparts 5 to 6 ft. high, perpendicular on both faces. The pa of Rua a Rehu at Puha, Poverty Bay, 106 strictly speaking of the flat land variety discussed in a later section, has a remarkably wall-like bank, 10 to 11 ft. wide, standing in places 6 ft. high above a ditch nowadays up to 4 ft. deep (plate 6).
As might be expected, the defensive banks tend to be higher where the interior of the pa is at the same level as or lower than the ground outside. Such is the case with the Taranaki promontory pa of Kumara-kaiamu and Te Kawa where Best records inner ramparts of 4 to 6 ft. for the former 107 and up to 6 ft. for the latter. 108 Where the upward slope of the ground into the interior of the pa provides for the ditch an inner scarp of some size, the surmounting parapet where present, tends - 85 to be more modest, up to 4 ft. for example on the first pa in the Tapa-huarau series in North Auckland. 109
To any of these genearlisations exceptions can be found, the significance of which is at present obscure. Thus we have seen, in the case of the Northcote pa, a ditch across flat ground with no accompanying bank, and some fosses in similar topographical circumstances have most modest ramparts (plate 3).
The general impression is gained that banks in Taranaki are larger than those in North Auckland whatever the circumstances, but documentation is lacking.
B. The Ditch (plate 5)
A measure of relationship is to be expected on general grounds between the size of the bank and the size of the ditch from which the material to form it was won. The depth of the ditch may be limited by the nature of the subsoil and the accompanying bank may be similarly small. Such appears to be the position at Tapuariki pa near Paeroa, Hauraki Plains, 110 and the Wellington pa discussed by Best. 111 But large ditches occur with little or no bank and small ditches where no limitation in size is imposed by the nature of the underlying rock (plate 3). Some formidable inner scarps result where ditches are cut on sloping ground; examples up to 25 ft. are quoted by Best from Te Koru pa, Taranaki, 112 and Nga Potiki a Rehua, Waitotara, 113 the outer scarps at both places averaging 10 ft. Some of the fosses cutting across fairly level ground on Taranaki promontory pa are of considerable size; together with superimposed ramparts of 4 to 5 ft. the ditches at Kumara-kaiamu and Te Kawa present inner scarps of 16 ft., 114 while the largest of the big complex of ditches at Te Rerenga is 30 ft. from brow to brow and 15 ft. deep. 115
Again the general impression is that the ditches of Taranaki pa tend to be large. Certainly it was to Taranaki and not to the neighbouring King Country that Kelly turned for a parallel to the huge fosse, 30 to 50 ft. deep and 25 to 30 ft. wide, which he recorded on the Ngati-maniapoto pa of Orongokoekoea on the upper Mokau. 116 The evidence is by no means full enough for a comparative survey and detailed information is particularly lacking for the Bay of Plenty and the East Coast. Amongst the data available Kelly records a 6 ft. inner bank above a 6 ft. ditch at Totara pa, Thames, 117 a 6 ft. bank above a 4 ft. ditch at Taupiri- 86 - 87 - 88
pa, lower Waikato, 118 and a 5 to 6 ft. combined bank and ditch on the pa of Kahuwera, Bay of Islands. 119 These modest dimensions are in line with Best's figures for the cross-rim trenches on the cone pa of Pouerua, North Auckland, 120 and for the neighbouring pa of Mawe on a promontory into Lake Omapere. 121
Cook, however, records inner scarps for ditch and bank from Mercury Bay of 22 ft. and 24 ft. 122 The main ditch of Piraunui pa near Putaruru has a 12 ft. outer scarp and an 18 ft. inner one including the rampart, 123 while the extraordinary stepped fosse on Oneonenui (Kore-kore) pa, Muriwai, upwards of 60 ft. wide, shows an inner scarp of 30 ft. 124 These two last pa are built on rock somewhat softer than usual, rhyolite in the former case, sandstone in the latter. Whatever its effect on size, the nature of these particular subsoils has enabled very steep scarps to be constructed and preserved, that at Oneonenui being virtually perpendicular. In so far as surface observations are reliable, the cross section of Maori fosses ranges from rectangular in soils of this character to trapezoid in less stable materials 125 (plate 5).
On many sites transverse defence of the type described is multiplied. Sometimes two or three ditches cut off a promontory at a single point, the bank separating any two of them usually being simply the baulk of undug earth. In the majority of cases, however, the defences are staggered, the result being, as we have had occasion to notice before, the subdivision of the pa interior into separate areas of defence and habitation (fig. 3C).
In the case of the ridge fort of Otumoana, near Urenui, this practice resulted in the creation of seven fortified posts within the one pa 126 (fig. 4). A similar arrangement is to be seen at Te Rerenga pa near New Plymouth. 127 The outer defence consists of a ditch presenting, with superimposed rampart, an inner scarp of 10 ft. which cuts off a tongue of high land surrounded by bluffs. The interior is divided into two sections by a triple ditch system of much more massive dimensions, over 50 ft. wide and 15 ft. deep at the deepest point. The outer section is obviously a subsidiary area with less elaborate defences, while the inner area is the real core of the pa. This gradation in defensive - 89 strength is even more apparent at a site like Piraunui near Putaruru, where the terraced habitation area outside the deep fosse which cuts off the main pa at the headland's end is lightly protected by a trench 15 ft. wide and 5 ft. deep including the banks heaped up on either side. 128
The practice of multiple cross-trenching is, however, best exemplified by some of the ridge and ridge end forts. The pa of Nga Potiki a Rehua, for example, 129 is strung out along six hundred yards of narrow sandstone ridge overlooking the Waitotara River. Before the main terraced living area is reached at the ridge end, seven ditches must be crossed. These ditches are cut where the ridge is narrowest and the areas between them, although often exceedingly narrow, have been levelled off for use. It is interesting to note that the earthen banks are all on the west side of the ditches they accompany. This indicates that attack was expected from one direction only, so that the western most area was the citadel of the pa. We may contrast this with the situation at Okoare pa (fig. 2B) where the internal ditches were banked on both sides to guard against the possibility and feasibility of attack from two directions. 130 The same arrangement is apparent in the layout of Oneonenui (Korekore) at Muriwai. 131 Topographically this site bears some resemblance to Nga Potiki a Rehua, in that the narrow approach westward to the ridge end is crossed by three ditches with the intervening areas levelled off. However, the fact that the banks accompanying the first and third of these are on different sides of their particular ditches, the sides nearest each other, means that the focal point of the pa was the intervening area and not the ridge end whose extremity was comparatively lightly defended. 132
The multiple cross-trenching on a pa like Nga Potiki a Rehua is obviously related to the particular topographical features of the site. It is interesting to note, however, that on other pa with topographical features of the same order cross-trenching is less conspicuous and may indeed in certain cases be completely absent, while the desired result is achieved by scarping. No better example could be quoted in this respect than the North Auckland pa of Nga Puke-pango 133 (fig. 1B) where the narrow spurs which radiate from the central hill have been elaborately and exclusively cross-scarped for defence.
The distinction which I am trying to make here is perhaps more apparent on sites with wider summit areas than those discussed above and where cross-ridge earthworks are of necessity on a larger scale. The choice in defensive subdivision between scarp and ditch may be dictated wholly by the topography of the summit area, on the argument that flattish ground precludes scarping while rising ground makes it possible. On the evidence available, however, it would be equally arguable that the distinction is at least partly regional in that the cross ditch was preferred in Taranaki and the cross scarp, very occasionally - 90 surmounted by an earthen parapet, in North Auckland. 134 A comparison of Best's plans and cross sections for a number of selected Taranaki and North Auckland pa, e.g. Okoare, 135 Otumoana 136 and Okoki 137 on the one hand and Taka-poruruku, 138 the Tapa-huarau complex 139 and an unnamed pa near Kawakawa 140 on the other, would bring the point home (figs. 4 and 5 here). Best selects this same feature of North Auckland pa as a point of contrast, not with the Taranaki practice, but with that of the Bay of Plenty. 141 Recorded information about pa in this area is negligible, but a preliminary field survey suggests the prevalence of ditched defences there.
One minor point is worthy of note before this section is concluded. It is the practice, exemplified by a number of pa in the Bay of Islands district, of extending cross-ridge ditches some distance down the hill-side 142 and even of constructing special ditches in this position 143 (fig. 5). Best suggests it as a local custom to prevent a fortified position being outflanked. Besides the instances mentioned by Best we seem to have the same feature represented on some of the pa recorded by Kelly for the same area. Thus the innermost fosse on Tangitu pa (fig. 3C) not only cuts across the ridge but continues down the side of the cliff, 144 the main fosse on the unnamed pa at Otehei Bay, Urupukapuka, cuts down from the ridge top almost to the waters edge, 145 while the two ditches on Paeroa pa (fig. 3A) are extended down to the sea on one side. 146 Something similar in the way of downhill extension has been noted on a promontory pa at Shoal Bay, Typhena, Great Barrier Island and also on Great Mercury Island.
If fosse and bank defended the ridge and promontory pa at the points of easy access, the common form of earthwork defence laterally, where such was needed to supplement nature and palisades, was the scarp. As on sites of the hill pa category, the scarp was normally associated with either terrace or ditch.
A. Lateral Terraces (fig. 3A)
These performed the same function as platforms for storage and habitation as the terraces of the hill pa. The need for them was - 91 governed by the extent of habitable ground available in the interior of the pa and their disposition was dependent upon the circumstances of topography. Best comments on the regularity of terrace formation on the ridge forts of the Ohaeawai district, as contrasted with the shortness and irregularity of the terraces on the neighbouring cones of Pouerua and Maunga-turoto. 147
On some promontory pa with strong natural protection the disposition of the terracing is such that the sole purpose would appear to be to provide level ground for habitation, with integration into the defence system a subsidiary concern, if intended at all. Such would appear to be the case with the scattered half-moon “linchets” that cling perilously to the precipitous slopes of the pa on Opito Point, north of Mercury Bay.
B. Lateral Ditches (figs. 3B and 5)
The lateral ditch is identical with the ditch of the ditched hill pa, being provided with a built-up outer rampart. Frequently, though by no means invariably, the inner scarp is crowned by an earthen parapet. This feature, which has not been noted on the ditched hill pa, 148 emphasises the similarity between the lateral and the cross-ridge fosse.
On the small island of Rangiwae in Tauranga Harbour is a promontory pa on a flat ridge washed by the sea at its apex and flanked by two small creeks, which appears to be protected round its entire perimeter by a fosse. It is cut off from the main by a long and rather massive ditch, 20 ft. wide with a 5 ft. outer scarp and an inner one of 15 ft. including an inner rampart which stands above the interior of the pa at heights ranging from 2 to 6 ft. At both ends this cross ditch turns at right angles to follow the flanks of the promontory and on these lengths it is provided with an outer rampart. Towards the extremity of the pa it becomes choked with bramble but it has every appearance of completing the circuit.
Doubtless other such examples of complete ditching remain to be discovered in the field. Certainly there are none recorded in the literature. There are some Taranaki sites described by Best which approach the Tauranga example in some degree. At Kumara-kaiamu the frontal fosse turns along the west side of the pa to become the lateral defence for half its length. 149 It then changes its character, widens out, flattens, loses its outer rampart and continues the rest of the way as a normal terrace. 150 This change from fosse to terrace has been noted in connection with the ditched fort near Te Puna (fig. 2A) discussed in the hill pa section and may be found elsewhere in the western Bay of Plenty. It is therefore interesting to note that it is a not uncommon feature of the lateral defences of recorded Taranaki pa. At Te Kawa a broad residential terrace is transformed into a defensive ditch through the - 92 medium of a narrowing terrace which assumes an outer parapet. 151 The same three stage transition of fosse into terrace and back to fosse is present also on Parihamore pa near New Plymouth. 152
The lateral fosse is also known in North Auckland. Kelly figures it as the sole defence on the western flank of the headland pa at Otehei Bay, Urupukapuka, 153 and as a main lateral feature on the southern side of Pikiore's pa at Te Hue, Bay of Islands, where it appears, from Kelly's sketch, to finish as a terrace. 154 At Okuratope pa near Ohaeawai the fosse on the southern and eastern flanks seems to give on to a terrace at one if not both ends, while a parapet crowns its inner scarp 155 (fig. 3B). On the more elaborate ridge pa inland of the Bay of Islands, however, the lateral fosse is very much subsidiary to scarp and terrace formation. On the pa called by Best Tapa-huarau no. 1, a cross-ridge fosse at the east end of the site curves round the hill flank for 15 yards on the north side before transforming itself into a normal terrace. 156 On the adjacent Taka-poruruku pa the lateral fosse is used in short discrete stretches to supplement the scarped defence where subsidiary spurs offer flank access from the flats below. 157 The change of fosse into terrace noted in Taranaki is in these circumstances not well represented. The three instances already quoted, on Okuratope, on Pikiore's pa and on Tapa-huarau no. 1, are quite simple and show none of the flexibility of the Taranaki examples. There are a number of cases of lateral terraces stemming from the ends of cross-ridge fosses 158 but this practice, which occurs in Taranaki also, 159 is not the one at issue. Finally the terrace with outer rampart, seen to be part of the same complex in Taranaki, is not a feature of North Auckland pa.
Few examples of the lateral ditch have been recorded from other parts of the North Island. Kelly notes traces of one on the north side of Taupiri pa in the lower Waikato. 160 The suggested reconstruction of Popoia pa, Poverty Bay, 161 shows a fosse the length of the south side of the fort. On the ground today this interpretation is less sure and the earthwork has the appearance of a true terrace, at most with parapet above the scarp. A number of other terraces on Popoia show this latter feature unambiguously. It is recorded also on Ihupuku and Nga Potiki a Rehua pa near Waitotara 162 and Delph and Archey would appear to mean the same thing when they talk of rubble walls at Piraunui near Putaruru. 163 Something similar may be referred to by Kelly in his description of the Ngati-maniapoto Ngakuraho pa near Te Kuiti. 164- 93
Though North Auckland apparently shows no evidence of the parapeted terrace, it is characterised by the occasional appearance of the parapet on scarps subdividing the summit areas of pa. Its occurrence as recorded by Best 165 would suggest its use as an additional protection on the more accessible sections of hill flanks. In this it would replace the lateral ditch formation that might have been used in similar circumstances in Taranaki, where the relevant scarp was at times crowned by an identical rampart. 166
A rare feature noted on North Auckland pa is a combination of lateral with frontal scarp parapets as though to mark off an area for special protection. 167
A somewhat exceptional use of rampart is reported by Kelly from Orongokoekoea pa on the upper Mokau, where an earthen wall runs parallel to one of the scarps of the summit area but is set back a little from its crest. 168 A similar arrangement is recorded for the Taranaki pa of Okoare, where a bank accompanying one of the internal dividing ditches is set 10 ft. back from the scarp 169 (fig. 2B).
Two points might be made in relation to upland pa in general before discussion of them is concluded.
CLASS 2—FLAT LAND PA.
(fig. 6, plate 6)
In this class stockades, ditches and banks were the devices employed for defence. Scarping, the artificial steepening of hill slopes, is for obvious reasons absent.
In Variety A, where there are natural obstacles to approach on all sides, palisading was the sole defensive method used. Into this category fall low islands in swamp and lake. Recorded swamp pa include the natural island site of Oruarangi on the Hauraki Plains, 180 an allegedly artificial island pa in the Whakatane Valley, 181 and two sites in Horowhenua, one a natural, 182 the other an artificial island. 183 Low island pa in lakes include the artificially constructed sites in the Horowhenua region, 184 where only is the practice well attested though alleged elsewhere.
The field identification of palisaded pa of this character is, because of the absence of earthworks, apt to be difficult. When field evidence is present it is likely to consist of the signs of habitation and may not materially differ from the field evidence of open settlement, except perhaps in so far as circumstances of location are concerned.
Varieties B and C, with open approaches over a smaller or greater distance, can be treated together since the same type of defence is applied to both. This consists basically of a ditch backed immediately by a bank, thus creating for the defenders the advantage of superior height where nature failed to provide them. There may be a single bank and ditch, as on the three forts of the Muriwai complex in bends of the Maraetaha River near Gisborne, multiple defences at a single point, as at Mangatoatoa pa near Te Awamutu 185 (fig. 6A), or multiplied defences spaced over a distance and serving to segment the area into separate habitation areas, as at the Waikato pa of Matakitaki near Pirongia 186 (fig. 6B).
Variety B is the lowland equivalent of the promontory pa of the upland category and recorded examples occur within river loops and at the confluence of streams. Thus the small but important pa of Mangatoatoa near Te Awamutu is situated in a deep bend of the Puniu River and is protected by a double fosse now filled in, respecting which Kelly makes no mention of ramparts 187 (fig. 6A). Raupa pa near Paeroa is- 95 - 96
situated at the old junction of the Ohinemuri and Waihou Rivers and was cut off by three ditches and banks, the main, innermost bank still 4 to 6 ft. high despite levelling. 188 The Matakitaki pa near Pirongia is situated on a point of land where the Mangapikopiko River joins the Waipa 189 (fig. 6B). Before their confluence the streams flow parallel for some distance and at points in their winding course approach each other closely. At three such points a ditch and bank have been constructed from one river to the other, dividing the pa into three living areas each with its own name. The outermost fosse has been virtually obliterated but the two others are still visible, 20 ft. wide at the bottom and showing an inner scarp of 12 to 13 ft. including the height of the bank. In the final segment, Matakitaki proper, there is a narrow ridge of higher land which has been levelled on the top and protected at both ends by a transverse ditch and bank.
Variety C is a wide category including those pa where little defence is given by nature and artificial works in the form of ditch and bank comprise the greater part of the circuit (fig. 6C). A good illustration of this is provided by Rua a Rehu pa, Puha, Poverty Bay, where the defences were anchored to a straight stretch of the precipitous bank of the Waikohu River. 190 The pa which was laid out from this base-line was regular to a degree unusual in pre-European strongholds, with straight sides and right-angled corners. There are two features worthy of note in the plan of Rua a Rehu. One is the rectangular projection of the frontal defences of the pa, quite like the flanking bastion of the post-musket pa (plate 6). The other is a remnant of earthen wall which Best describes as dividing the interior of the pa in two. Since there are traces of a ditch in front of this the more likely explanation would be that it constituted the original line of the pa on one side, later extended to its present limits. The defences in the main consist of double bank and ditch but the outer series has been formed along some stretches only. This may be purposeful and not, as Best suggests, an indication of the unfinished state of the earthwork. The inner bank and ditch are the more imposing. When Best saw them the ditch was in places 6 ft. deep and the bank 6 ft. high. Today representative dimensions are 4 ft. for the ditch, 6 ft. for the bank, the former 6 to 7 ft. wide at the top, the latter 10 to 11 ft. at the base. The bank is remarkable for its wall-like form, especially considering the nature of the material of which it is built.
Another example of this variety of pa is Manu-korihi at Waitara, Taranaki 191 (fig. 6C). Here three massive banks and ditches straddle in rough semi-circular fashion a considerable distance to enclose an area of flat land protected on the two remaining sides by the bluffs of the Waitara River. Each defence is separated by an area for habitation. - 97 The innermost ditch varies in width from 30 to 40 ft., its outer scarp from 5 to 9 ft. high, its inner scarp from 10 to 16 ft, the inner bank accounting for from 5 to 9 ft. of this. The outermost defence is the heaviest with an inner scarp of from 10 to 20 ft., including 6 to 12 ft. of bank.
Notable features on this site include definite traces of a small parapet at points along the outer lip of the middle ditch, 192 and also the considerable width of the banks at the point where the middle and innermost lines of defence run side by side. 193 The relevant dimensions here, 6 to 8 ft., are proof to Best that the defenders occupied the tops of the defensive ramparts in pre-European times.
It is apparent from the discussion of ramparts in Maori defensive works of both upland and flat land type that use was made in pre-European times of rampart, bank or wall only in connection with scarp and fosse. It is hard, therefore, to judge the significance in terms of defence of the rubble walls or banks running up the hillsides to the base of the terraced defences of the volcanic cone pa of Puketutu in the Manukau Harbour, Auckland, and associated with other walls along the base of the hill. 194 Fairfield notes that they are subsidiary to the main defences and suggests that the areas they define were intended to quarter a large population in time of war. There is no evidence, however, that they were of a defensive nature. An alternative explanation could be made in terms of agriculture though the steepness of the slopes involved might be an objection. However, an identical feature is to be seen at Wiri Mt., near Auckland, where the rubble walls in question spread down the hillside from the terraced defences of the pa well on to the surrounding plain (plate 1). It seems certain that they must be interpreted as the signs of agriculture and not defence.
The stone facing of terrace scarps is sometimes erroneously referred to as stone walling for defensive purposes. 197 Such would appear partly to be the case in the short archaeological description of the Poor Knights Islands given by Fraser. 198 However, the caption to one photograph appears to claim the presence of a 6 ft. high free-standing wall of stone cutting off a headland and perhaps of defensive character. If a free-standing wall is meant, the feature, though it would appear to be unique for a pa, is comparable with the stone boundary walls recorded by Adkin for eastern Palliser Bay and interpreted by him as marking off a sacred, not a defensive site. 199- 98
ENTRANCE AND ACCESS.
The entrances that were to admit friends inside the pa and from area to area within it had at the same time to be adequately protected against foes. From historical and traditional data Best is able to describe the variety of devices employed in this respect. 200 The actual gateway might be a small opening made in a large carved post set in the stockade or cut through the palisades themselves. The entrance might be blocked by a slab of wood, by a kind of hurdle or by a sliprail arrangement.
The defensive elaborations included 201:—
The entrances in the different lines of defence which went to make up the complex pa were commonly staggered to make access to the interior areas as difficult as possible and to expose the flank of any hostile party which attempted to force entrance by the gateways.
A particularly vivid picture of the obstacles opposed to an invader attempting to storm the gateway of a pa is given in the description of the French attack on Paeroa pa, Bay of Islands, after the death of Marion Dufresne. 202 The gateway at which the French attack was aimed was an opening only 2 ft. square, situated halfway along one side of the promontory. The approach was by a narrow path, wide enough only for one man at a time to pass, with the line of palisades on one side and cliffs on the other. During the attack the defenders threw water on the pathway to render access even more hazardous. A similar gateway and approach path seem to have existed on the opposite side of the headland.
Much of the ingenuity shown in entrance defence has left no record in the field today, but there are sites that illustrate very well the general effect aimed at. At Te Rewarewa pa near New Plymouth, for example, two of the entrances were attained by long narrow approaches flanked by bluffs, presumably palisaded, on the one side and the main earthwork defences of the pa on the other. 203 At Nga Puke-pango, near Ohaeawai, the elaborately terraced hill displays a variety of modes of access and a complex system of communication through a multitude of terraces to the summit. 204
The field evidence that does exist shows that certain practices were standard throughout the country, others less extensively represented. They can best be described in relation to the type of defensive earthwork through which they were designed to allow access.- 99
Access at Fosses.
Some of the apparent entrances at the flat land pa at Manu-korihi (fig. 6C) may not be original, but the practice is well-established here of forming an entrance way by leaving a narrow undug causeway over the ditch with a corresponding break in the bank. 205 The same features reappear at one of the seven spaced cross-ditches on the long upland promontory pa at Nga Potiki a Rehua, the causeway here being only 2 ft. wide though standing upwards of 8 ft. above the bottom of the ditch. 206
The device of the undug causeway is, indeed, of quite frequent occurrence, especially, if we are to judge from the nature of these occurrences, where no inner bank is present and where the areas to be linked show no great differences of level. Examples might be quoted from Te Koru, Taranaki, 207 and Mawe pa, North Auckland. 208 The causeway is used also as a connecting device between inner areas of a pa sundered by a fosse. Instances are to be found on the North Auckland ridge pa near Kawakawa described by Best 209 (fig. 5, between N and FO), on Tapa-huarau no. 2 in the same region 210 and on the Taranaki pa of Kai-tangata 211 and Okoki. 212
Best records one instance of an undug causeway across a ditch where the areas linked are at very different levels. 213 It is found on the North Auckland pa of Taka-poruruku and the causeway in question ends against a scarp rising 10 ft. above the ditch. No means of passing this scarp are apparent and it is probable that a notched pole was employed.
In the case of fosses where no form of crossing survives we may assume the use of wooden bridges, but this assumption is unsatisfactory where inner banks occur and no breaks are visible in them. It is possible that the notched pole was used here also, or perhaps the access path outflanked the fosse, as in the case of Paeroa pa described above. Alternatively the fosse itself may have been used as the approach way. This seems a reasonable explanation in those cases where a transverse fosse gives directly on to a lateral terrace or where it turns through a right angle and becomes a lateral fosse. Field evidence for some such practice would be provided by Urenui pa, where an upward sloping pathway at the south end of the site enters the inner of two cross trenches which continues to serve as an access lane for some distance round the west flank of the pa. 214 Any assailant entering the pa by this route would have a 10 ft. wall on his left and a 16 ft. scarp on his right. A somewhat similar arrangement is to be observed on Kumara- - 100 kaiamu 215 where the outer wall of the lateral fosse swings out a little to admit and protect entrance at the point of transition from fosse to open terrace. An identical feature has been noted on the pa on Te Puna peninsula near Tauranga (fig. 2A). A final example may be taken from Kelly's description of Whare-taewa, Mercury Bay, where the main cross-ridge fosse of the pa turns at right angles down the steep flank of the ridge and gives access to and from the flats below. 216
Access at Scarps.
Frequently on pa we find a few terraces so disposed that they grade into each other. This is the case on the old pa near Kawakawa described by Best 217 (fig. 5, terrace H), on Urenui pa, 218 at Te Koru 219 and at many other sites. The terrace thus takes on the aspect of a graded roadway but the practice is possible only where the differences in level are not too great for the distances involved.
A true graded roadway is present at Te Koru pa. It is 8 ft. wide and slopes up from the rivers' edge to the main entrance of the fort. 220
Access from terrace to terrace was often gained by direct sunken way and occasionally by raised causeway or by a diagonal approach cut out of the scarped face. 221
Recorded examples of the sunk way include the North Auckland pa of Nga Puke-pango, 222 One Tree Hill, Auckland, 223 and the Taranaki sites of Okoki, where the excavated passage is 4 ft. wide and 4 to 8 ft. deep, 224 and Te Koru. 225 At Te Koru the graded road which gives access to the defended area from the river bank becomes a sunk way as soon as it enters the defences and continues as such for more than 30 ft., cutting through two scarps and attaining a depth of 8 ft. As Best points out, these narrow passages would pass beneath the stockades that lined the terrace scarps and could easily be blocked.
There are but few recorded instances of raised causeways. The line of the sunk way described above at Te Koru is continued as such a causeway on the higher terraces 226 and other examples occur within the Taranaki area at Urenui 227 and at Kai-tangata. 228 The causeways are raised in the sense that the ground is left undug during the scarping and terracing of the hillside.- 101
The diagonal pathway is represented on the slope of the cone pa of Maunga-turoto, 229 where it gives access almost to the very summit, the majority of the terraces through which it passes breaking their level at the point of passage. A similar pathway leads up the main 30 ft. scarp of Orongokoekoea pa on the upper Mokau. 230 Kelly describes a low earth wall at the bottom of this path, running parallel to the scarp and defining and protecting the only possible access to the track.
A hybrid form between the diagonal pathway and the raised cause-way is represented on Nga Puke-pango, where access to the summit area up a scarp 12-17 ft. high is gained by means of a sloping embankment. 231
It is likely that on some scarps there were removable instruments of ascent. Notched poles were seen in use by Europeans on cliff pa in Taranaki. 232 On rocky and difficult sites like this steps were sometimes cut: examples have been recorded on Piraunui pa near Putaruru 233 and are present on Whitianga Rock, Mercury Bay.
THE GUNFIGHTERS' PA.
With the opening of New Zealand to European contact in the early 1800's came the musket and with the musket an efficient long-range weapon was put into the hands of the Maori warrior for the first time. The musket's increasing use meant adaptations in traditional methods of defence. However, the changes observed in pa construction over the course of the nineteenth century are to be explained solely in terms of the advent of the gun at latest for the period up to 1840, the final epoch of intertribal warfare.
With the declaration of British sovereignty in 1840 came European settlement and European influence on an increasing scale. Land was bought, a new economy established and the forms of government introduced. The tremendous changes that came over Maori society in this period involved the pa no less than other aspects of the traditional culture. No longer the tribal stronghold in conditions of intertribal warfare, the pa became the strictly military instrument of often temporary and expedient defence against European arms wherever friction burst forth into open warfare.
The Final Period of Tribal Warfare.
It was during the final period of tribal warfare, however, that the Maori seems to have evolved under the stimulus of firearms, rather than to have adopted, a whole series of defensive devices that rendered the redoubts which he opposed to the European “superior to any constructed … by either Imperial or Colonial troops.” 234
A major change alleged for the introduction of the musket was the - 102 abandonment of the hill pa and the construction of new defended settlements on flat ground. 235 Certainly some hill positions were given up, as the one at Kaipara pointed out to Marsden in 1820. 236 An upland fort strong in terms of pre-European offensive equipment might well become untenable in the musket era due to its being commanded at a short distance by higher ground. European observers in the thirties, however, still describe upland pa in occupation. 237
The most radical effect of the musket was to drive the defending Maori from the rampart and force him to take cover in the ditch, 238 while the ditch, taking on the new function of a rifle pit from which to fire, took on new dimensions. 239 The deep fosse and large ramparts common in the pre-European period were replaced by ones of smaller dimensions, 240 the ditch being rarely more than 5 ft. deep.
How quickly and in what ways this change was accomplished it is impossible from the evidence to judge, but from the shift in emphasis from rampart and scarp to ditch stemmed a whole series of adaptations and modifications of distinctly “European” look, which, however, appear fully formed in the strongholds erected in the north during 1845 and 1846 in the very first war against the Europeans.
No better example of the developments involved can be quoted than Ruapekapeka pa near Kawakawa 241 (fig. 7A). These consisted of:—
On contemporary sites a few supplementary features are represented. The Okaihau pa in the same area, for example, seems, from the virtually obliterated earthworks seen by Best in 1914, to have had an outwork of rifle pits, 3 to 4 ft. deep, with earthen parapets of 1 to 2½ ft. on the outer lip and no communication between them. 243 The apparent development of the rifle pit at so early a stage is interesting. It is a- 103 - 104
particular feature of the later pa of the Maori Wars period, where it served an identical function with the ditch which it tended to replace. Typologically, however, the step from ditch with traverses to rifle pit is not very great.
Some of these innovations were already in use by the end of the 1830's. In the account of the voyage of the French frigate Venus is a description of a pa near Kawakawa which mentions loopholes, flanking angles and covered ways. 244 The Bay of Islands may indeed have been the area where these developments in defence took place, for it was here that the musket had been longest known. The missionary Taylor tells a story which indicates one way in which the innovations may have become more widely known. In 1845 there was an unsuccessful British attack on Ohaeawai pa, a stronghold of the type of Ruapekapeka and in the same district. Taylor, in the inland districts of the North Island far away to the south, saw neatly constructed models of the pa and its defences exhibited and explained by messengers from the battlefield. 245 Certainly identical features of these northern pa were being employed by 1846 at the other end of the North Island, near Wellington. The descriptions of Te Rangi-haeata's pa of Matai-taua 246 near Porirua mention a ditch 4 to 5 ft. deep crossed by numerous traverses, a double palisade in front, flanking defences, and underground communication beneath an inner bank between the ditch and huts within the pa. One of the contemporary authorities specifically mentions the similarity of the pa to those of the northern series. 247
Ruapekapeka pa and its partners, however, are examples not only of the adaptations made to meet the challenge of the musket in the final stages of intertribal warfare, but of the innovations called forth by the use of cannon in the new era of European warfare. Huts and shelters in the interior of the pa went underground and were roofed with timber and earth. Underground access was, as we have seen, provided to the firing trenches. 248 The published cross sections of the bomb proof pits include a type identical with the bell-shaped storage pit of pre-European times 249 (fig. 7A).
The excellence of the fortifications erected against Europeans at the very beginning of the European wars was a matter for comment from all military observers. Of the attack on Ruapekapeka pa Best says “The British troops opened on the above pa with three 32 pounders, one 18 pounder, two 12 pounders, and seven brass guns and rocket tubes. At the end of ten days the troops took this pa by assault—after the enemy had obligingly vacated it.” 250- 105
The Final Pa.
The later pa belong to the period of the Maori Wars in Taranaki, the Waikato and the western Bay of Plenty, and the Hauhau campaigns in the broad stretch of country running across the centre of the North Island from east coast to west. The sites are exhaustively reviewed by Cowan 251 and a few general remarks only will be offered here.
In the main the strongholds display features already described in connection with the northern pa of Hone Heke's war, with the particular development, however, of the rifle pit, the underground passage and the subterranean shelter in the face of cannon. Less emphasis seems to have been placed on palisading in these same conditions. 252 Many of the positions occupied at this period were never intended as more than temporary earthworks thrown up to meet particular military situations. Such were the famous pa at Rangiriri 253 and at Orakau 254 and the Gate pa near Tauranga, 255 perhaps the best known of all. The fortifications at Rangiriri consisted of a high curtain wall and broad fronting ditch, protected by rifle pits, strung across the main south road between Waikato River and Waikare Lake. The main feature was a large bastion connected by tunnel passage to the rear of the defences. The Gate pa was an intricate network of rifle pits and underground passages surrounded, in Gudgeon's works, “by a mere cobweb of palisade.” The surviving earthworks at Paratui pa near Te Kuiti, 256 constructed by the Kingites in 1864 after their defeat at Orakau, show a triangular area with sides 50 to 70 yards long, surrounded by a ditch with bays and bastions, backed by a rampart with rifle pits on top, and connected by short communication trenches with a second ditch inside the rampart.
There are variations on the common defensive theme from place to place. Thus the Pourere redoubt erected near Roto-aira in 1869 (fig. 7B) had as its main defences an earthen wall broken into flanking angles at two opposite corners. 257 There was a ditch immediately outside, the inset entrance was protected by a detached ravelin of earth 258 on the line of the main rampart, and the interior of the pa was divided into two by an independent earthen wall so disposed as to cover the entrance at one end. An earthwork of similar but simpler plan stands near the summit of the road from Napier to Taihape. 259- 106
European Redoubts and Fortifications.
The description by Best of the redoubt built by European troops at Weraroa near Waitotara, 260 an earthen wall surrounded by a 5 ft. ditch and broken into possibly three flanking angles, will serve to emphasise the similarity of defensive works achieved by different military traditions in similar conditions of warfare. Speaking of Maori and English fortifications in general, however, Thomson remarks 261 that where in an English fort the ditch was deep and outside the defences, being designed to obstruct the enemy, in a Maori stronghold the ditch was shallow and inside the palisade, being intended to cover the defenders firing from it.
The contrast may be illustrated from two Taranaki sites, the earthworks of the old military redoubt at Oakura 262 and the Maori pa of Orongo-mahanga described by Colonel Mould. 263 The fort at Oakura is square and the earthworks massive. The outer ditch, 65 yards long on each side, has an outer scarp of 8 to 13 ft. and an inner one of 18 ft. to the top of the inner wall. There are flanking angles at each corner and a banquette or ledge on the inside of the wall 3 ft. above the interior level. The gunfighters' pa of Orongo-mahanga was the site of an old pre-European pa with ditch and bank still standing. In the re-utilisation, however, the bank was ignored as a defence, the old fosse was reformed into a 6 ft. deep ditch with traverses, a double stockade erected outside it and covered pits dug behind. A banquette was made, the idea perhaps taken over from the Europeans, the ledge itself, however, not behind the bank but on the outer scarp of the ditch immediately beneath the palisades.
For possibly the majority of cases, the distinction made by Thomson and illustrated here is valid. There are, however, examples like the Pourere redoubt described above (fig. 7B), where the Maori gunfighter sheltered behind an earthen wall, which would break the rule and conform more to the European practice. Luckily little confusion is likely to arise in field work. The mass of historical material relating to the Maori Wars, as also the recentness of their occurrence, made it possible for Cowan to do a well-nigh complete review of the incidents and places connected with them. 264 The attribution to Maori or European of relevant earthworks is in virtually all cases, therefore, discoverable, down to the most minor and transitory of defences like the small ditch with little flanking angles round the Presbyterian Church of Pukekohe East where the still visible remains record an incident in the South Auckland campaign against the Kingite Maoris in 1863. 265
(to be continued)- 107
1 I am indebted to the following for help in this study: to the University of New Zealand for a research grant to finance field work and equipment; to Mr. W. Ambrose for preparing the plans; and to many friends who with information, transport or accommodation have aided my work in the field; Mr. A. Adams of Pukenui, Northland, Mr. and Mrs. A. Delamore and P. Mizen of Great Mercury Island, Mr. and Mrs. A. J. Black of Whitianga, Mercury Bay, Mr. and Mrs. Murdoch of Hikutaia, Hauraki Plains, Mrs. and Mrs. R. A. Sanders of Otumoetai, Tauranga, Mrs. K. A. Gordon and Mr. L. Fowler of Gisborne, Mr. and Mrs. R. A. L. Batley of Moawhango, Taihape, Miss D. L. King of New Plymouth, Mr. and Mrs. J. G. M. Smart of Wanganui and Mr. and Mrs. J. B. Palmer of Wellington; also the members of the Auckland University College Archaeological Society.
2 Crawford 1953; Wheeler 1954.
3 Crawford 1953:36. See Field Antiquities 1952 for a discussion of types of field evidence.
4 Clark 1954:xvii-xxi; 1956:1-3.
5 Duff 1956: chapters 4-6.
6 Grinsell 1953:11-28.
7 Crawford 1953:132-134; Piggott 1954:18, 20.
8 Piggott 1954:20-21.
9 Crawford 1953:51.
10 Cookson 1954 plates 7-9 for the use of this principle in archaeological photography.
11 de Laet 1957: plate 2.
12 On aerial photography see Crawford 1953:45-50 and references there quoted. For New Zealand see Blake-Palmer 1947.
13 e.g. Woodhenge, Cunnington 1927 and plates opposite 92 and 94.
14 e.g. the Woodbury complex, Crawford 1929 and plates opposite 385 and 452.
15 e.g. Best 1927:20-21, 46.
16 e.g. Best 1927:67.
17 Best 1927.
18 Best 1918a, 1918b.
19 Adkin 1948:17-104; 1950; 1955.
20 Palmer 1956.
21 For other work see Duff 1956:276-279.
22 See summary in Duff 1956:249-281 for details.
23 The distinction between portable and non-portable types of archaeological evidence is made by Childe 1956:116-117.
24 A large number of references are collected by Best 1927: chapters 1 and 2.
25 Beaglehole 1955:197-200.
26 Kelly 1951:31-34.
27 Nicholas 1817a:335-341; Elder 1932:97-99.
28 See, for example, the works of Percy Smith 1897, 1910.
29 Skinner 1911.
30 Firth 1927.
31 Best 1927.
32 Firth 1925.
33 Delph and Archey 1930.
34 Kelly 1931, 1932, 1933a, 1933b, 1934a, 1934b, 1938a, 1938b, 1940, 1945, 1951, 1953.
35 Best 1927:271-307.
36 Cowan 1922a, 1922b.
37 Best 1919:11; 1927:3-4.
38 Best 1927:19, quoting contemporary evidence to this effect.
39 e.g. Firth 1927:78.
40 Best 1927:166-167.
41 Maxwell 1933.
42 Diamond 1955:310-313. See also pp. 305-306.
43 Best 1927:165.
44 Best 1927:2.
45 Best 1925b.
46 These points regarding prehistoric population distribution in New Zealand are made by Cumberland 1949 and Lewthwaite 1949.
47 Best 1927:149.
48 Best 1927:115-128. These points are confirmed and amplified by the very full study, not yet published, of pre-European Maori warfare by Dr. A. P. Vayda of Columbia University.
49 Best 1927:149.
50 Kelly 1933b:169. Kelly says the site was allegedly abandoned with the coming of the musket.
51 Best 1927:15-16.
52 For an exhaustive discussion of palisading in Maori pa, which of course largely falls outside the scope of the field worker, see Best 1927:40-78.
53 Best 1927:17.
54 Best 1927:17.
55 e.g. Best 1927:176, 219, 221.
56 Best 1927:260, 262.
57 e.g. the cliff pa of Waimate and Te Namu, Best 1927:254-260.
58 e.g. Pohatu-roa at Atiamuri, Best 1927:260.
59 Best 1927:252-255.
60 Best 1927:254, 256, 259-260.
61 Best 1927:167. Best's alternative description, unique to this reference, of the ditch as an earthen wall or parapet following the line of the hill, lays emphasis on but a single feature of the earthwork concerned.
62 On hilltops these are often no more than 2 ft. to 3 ft. high depending on the natural contour.
63 Palisades are generally assumed for the brows of scarps, e.g. Firth 1927:75. Incomplete excavations at the volcanic cone pa of Taylor's Hill, Glendowie, Auckland, suggest that they may not have been invariably present.
64 Best 1927:234-237.
65 Best 1927:40.
66 e.g. the scoria scarps uncovered by excavation at Taylor's Hill, Auckland.
67 Best 1927:229. See footnote 39.
68 e.g. Mt. Eden, Auckland, Best 1927:230.
69 Firth 1927:71.
70 Best 1927:222.
71 Best 1927:225.
72 Best 1927:230.
73 Best 1927:222.
74 Best 1927:230.
75 Best 1927:210.
76 Best 1927:229.
77 Best 1927:221 (Mangere), 222, 224 (Pouerua), 226, 227 (Maungaturoto), 229 (One Tree Hill).
78 Best 1927:221.
79 Best 1927:212-215.
80 Best 1927:216-217.
81 Best 1927:212. See photograph of Otatara, 217.
82 Best 1927:219. Not all of these would be described as hill pa in the classification adopted here.
83 Best 1927:221.
84 Ohae pa at Ruatoki and Rakei-hopukia at Te Teko, Best 1927:211.
85 Pohokura pa, Urenui, Best 1927:163-164; Puke-kari-rua pa, Mokau, Best 1927:211.
86 Kelly 1934a. The description suggests scarping as the only defence on Ngakuraho pa, near Te Kuiti.
87 Best 1927:211.
88 Best 1927:166-167.
89 McDonnell 1887:595.
90 Best 1919:9.
91 Lateral that is as opposed to transverse ditches over short lengths such as are met with on some terraced hill pa. See footnote 54.
92 Halbert 1932:28-29. This ditch may have been filled in as I saw no unambiguous signs of it on a recent visit and had classified the site as that of a terraced hill pa.
93 Best 1927:147-150.
94 Kelly 1953:387-388.
95 Kelly 1951: between 72 and 73. See Kelly 1933a:87-88 for his own plan and description.
96 Kelly 1933a:91-96, describing the pa of the Maori chief Pikiore at Te Hue, the headland pa of Tangitu and an unnamed pa at Otehei Bay, Urupukapuka.
97 p. 81 and fig. 2B.
98 Manu-korihi pa, Waitara, Best 1927:133, a flat land pa; Moe-ariki pa, Urenui, 245; and perhaps Kumara-kaiamu, Urenui, 241.
99 Rangatira pa, Best 1927:265.
100 Tairawhiti Maori Association 1932:11.
101 The ditch at Rangatira pa does not appear to have an inner bank but a substantial inner scarp seems to be implied, Best 1927:265.
102 Best 1927:78-87.
103 Best 1927:41.
104 Best 1927:43. See also p. 38.
105 White 1900:399; Best 1927:41-42.
106 This pa is briefly described by Best 1927:138-139 but not named. For the name and some details of history I am indebted to Mrs. Waioeka Brown of Puha, the owner of the site.
107 Best 1927:242.
108 Best 1927:243.
109 Best 1927:185.
110 Kelly 1945:207.
111 Best 1927:3-4. Short and unsatisfactory descriptions of Wellington pa with fosses are to be found in Best 1919:4 (Worser Bay), 8 (Mt. Victoria ridge), 10 (Owhariu). See also Best 1927:250-251 for two pa at Titahi Bay.
112 Best 1927:157.
113 Best 1927:205.
114 Best 1927:242, Kumara-kaiamu, 243, Te Kawa.
115 Best 1927:238.
116 Kelly 1938b:148. The specific parallel quoted by Kelly is the famous site of Turuturu-mokai, near Hawera, southern Taranaki.
117 Kelly 1945:203.
118 Kelly 1940:149.
119 Kelly 1938a:24.
120 From 4 to 10 feet, Best 1927:222, 224.
121 From 6 to 10 feet, Best 1927:246, 248, 249.
122 Beaglehole 1955:197, 198-100. Quoted by Best 1927:26-27; Firth 1927:70. See Kelly 1953:384-387 for a description of some of the relevant earthworks today.
123 Delph and Archey 1930:62-63.
124 Firth 1925:5; Best 1927:195-196. Best and Firth record somewhat different dimensions.
125 Firth 1927:69.
126 Best 1927:167, 168. We may compare the Waikato tradition of Maungakiekie pa on One Tree Hill, related by Kelly 1941:170, where the pa is described as being divided into seven sections.
127 Best 1927:237-238.
128 Delph and Archey 1930: plan opposite 64.
129 Best 1927:199-207.
130 Best 1927:148-150.
131 Firth 1925:3-5; Best 1927:192-197.
132 Best 1927:196.
133 Best 1927:234-237.
134 Compare the previous proposition that the terraced hill pa is most numerous in the districts from Auckland city north, footnote 64.
135 Best 1927:148.
136 Best 1927:168.
137 Best 1927:172.
138 Best 1927:182.
139 Best 1927:184, 188, 190.
140 Best 1927:177.
141 Best 1927:176. Indeed in contrasting the Bay of Plenty practice, Best is inclined to compare the North Auckland to the Taranaki fashion.
142 Best 1927:176, 179, two pa near Kawakawa.
143 Best 1927:176, old pa near Kawakawa, 188, Tapa-huarau no. 2.
144 Kelly 1933a:94, 95.
145 Kelly 1933a:95, 96.
146 Kelly 1933a:87, 88.
147 Best 1927:180. Compare p. 184, Tapa-huarau no. 1.
148 So few have been studied, however, that this statement is more a confession of ignorance than an expression of fact.
149 Best 1927:241. The same feature is present also on Otumoana pa, Best 1927:171. Fig. 4 here, west of T.
150 Best 1927:241.
151 Best 1927:243.
152 Best 1927:240.
153 Kelly 1933a:95, 96.
154 Kelly 1933a:92, 93.
155 Kelly 1932.
156 Best 1927:184, 185.
157 Best 1927:181, 182, 183.
158 e.g. on Tapa-huarau no. 3, Best 1927:189, 190, 191.
159 e.g. at the western end of Otumoana pa, Best 1927:168, 169. E on fig. 4 here.
160 Kelly 1940:150-151.
161 Tairawhiti Maori Association 1932:11.
162 Best 1927:161, 206-207.
163 Delph and Archey 1930:62.
164 Kelly 1934a:104.
165 e.g. on Taka-poruruku pa, Best 1927:181-183.
166 e.g. on Kumara-kaiamu pa, Best 1927:241, 242.
167 e.g. Best's area 9, small and square, on Tapa-huarau no. 1 pa, Best 1927:184, 186.
168 Kelly 1938b: 150.
169 Best 1927:150.
170 Perhaps it is this practice referred to in the proverb Ka horo ko pare oneone, ka tu ko pare-toka. Graham's translation, 1921:168, is “Earthen parapet will indeed crumble, but stone parapet endure,” which suggests banks above terrace and platform scarps. These, however, seem not to have been a feature of the Auckland area from which, specifically Three Kings, the proverb derives.
171 Best 1927:155-156, 158 (quoting W. H. Skinner).
172 Best 1927:158, a small pa near Ohaeawai.
173 Best 1927:159.
174 Best 1927:159.
175 Best 1927:159, Nga-Weka pa, Stony River.
176 Best 1927:248.
177 Best 1927:34. He mentions the possibility of its being a post-musket development.
178 Best 1927:157, 198.
179 Best 1927:189.
180 Teviotdale and Skinner 1947:341.
181 Best 1927:268.
182 Best 1927:268. This example is actually post-European but it illustrates the type.
183 Adkin 1948:32-33.
184 Eight examples are known, six in Lake Horowhenua, one in Lake Papaitonga and one in Lake Waitawa, Adkin 1948:32-35. See also Taylor 1875; Buller 1893.
185 Kelly 1933b:168.
186 Kelly 1931:35-38.
187 Kelly 1933b:167-170. Kelly says that the interior of the pa is 3-4 ft. higher than the ground beyond the defences.
188 Kelly 1945:207-211.
189 Kelly 1931:35-38.
190 Best 1927:138-139. See footnote 84 above.
191 Best 1927:129-135. Manu-korihi is notable as being the first pre-European Maori pa site to be accurately planned. Carrington's survey of 1842 is reproduced by Best 1927:130.
192 Best 1927:133.
193 Best 1927:133.
194 Fairfield 1938:122-123.
195 Best 1927:138-139.
196 See p. 96 above.
197 Best 1927:158-159.
198 Fraser 1926:10-12.
199 Adkin 1955:465-471.
200 Best 1927:88-97.
201 Best 1927:88-89.
202 Kelly 1951:70-72. See also plan between 72 and 73.
203 Best 1927:136, 137.
204 Best 1927:234, 235.
205 Best 1927:131, 132.
206 Best 1927:203-204.
207 Best 1927:158.
208 Best 1927:246, 247, 248.
209 Best 1927:177, 178.
210 Best 1927:187, 188.
211 Best 1927:198.
212 Best 1927:172, 174.
213 Best 1927:181, 182.
214 Best 1927:143, 144.
215 Best 1927:242.
216 Kelly 1953:385, 386-387.
217 Best 1927:177, 178, terrace H.
218 Best 1927:143, 144, Q.
219 Best 1927:157, O and N.
220 Best 1927:153. It is noted by Best as being an unusual feature. A similar roadway is present at Mawe pa, Best 1927:248.
221 Best 1927:237.
222 Best 1927:234-236.
223 Best 1927:229.
224 Best 1927:174.
225 Best 1927:153-154.
226 Best 1927:154.
227 Best 1927:143, 144, access from L to ditch below S.
228 Best 1927:198, access from G to K.
229 Best 1927:228.
230 Kelly 1938b:148-149.
231 Best 1927:236-237.
232 e.g. at Waimate, Best 1927:254.
233 Delph and Archey 1930:63.
234 Best 1927:127.
235 Maning quoted by Best 1927:271, and Tuta Nihoniho quoted 272.
236 Best 1927:274.
237 e.g. Marshall 1834 for Taranaki, Wilkes 1840 for the Bay of Islands. Quoted Best 1927:277.
238 Best 1927:46.
239 Best 1927:127, quoting Thomson, 289, quoting Wakefield.
240 Best 1927:284.
241 Best 1927:291-292. See also Cowan 1922a:73-87.
242 As at the contemporary Ohaeawai fortification, Best 1927:290, quoting Maning.
243 Best 1927:289.
244 Best 1927:274.
245 Quoted by Best 1927:291.
246 Best 1927:277-278, quoting Collinson and Last.
247 Captain Collinson, quoted Best 1927:277.
248 These underground passages call to mind the traditions of subterranean entrances and escape routes in pre-European pa. See Best 1927:90.
249 Best 1927:292. Cowan 1922a:77.
250 Best 1927:292.
251 Cowan 1922a and b.
252 Best 1927:271, quoting Gudgeon.
253 Best 1927:295, 297-298. Cowan 1922a:326-334.
254 Cowan 1922a:371-376, 406.
255 Best 1927:295, 296. Cowan 1922a:423-434, 440.
256 Kelly 1939.
257 Best 1927:281. Cowan 1922B:376-380.
258 This feature, though employed in European military engineering, may derive from the palisaded work that often protected the gateway of the pre-European pa. See Best 1927:88-89.
259 This apparently unrecorded example was pointed out to me by Mr. R. A. L. Batley of Moawhango. It presumably dates from the Hauhau campaigns of the sixties.
260 Best 1927:286.
261 Quoted by Best 1927:292. See also Cowan 1922a:54 note.
262 Best 1927:158.
263 Best 1927:281.
264 Cowan 1922a and b.
265 Cowan 1922a:273-283.