Galatea Bay —The Excavation of a Beach-Stream Midden Site on Ponui Island in the Hauraki Gulf, New Zealand
John Terrell
By
[Received by the Editor, 18 August 1966.]
Abstract
Shell middens are common archaeological sites along the shores of New Zealand. Their occurrence suggests the importance of shore and marine resources in the daily life of the prehistoric Maori. An area excavation was conducted at Galatea Bay on Ponui Island in order to investigate the function of beach middens in Maori settlement and determine the range of evidence which has survived. In addition, an intensive laboratory study of the midden materials has made it possible to describe several techniques in midden analysis experimentally, and to interpret the non-artefactual evidence in natural and cultural terms.
Introduction
This report describes the excavation of a small shell midden site (N-43/33) at Galatea Bay on Ponui Island in the Hauraki Gulf, New Zealand. This work was carried out to recover information about the occupation represented by the beach midden in order to add to our knowledge of Maori settlement patterns. Before discussing the excavation and the results which have come out of it, an assumption about human behaviour upon which archaeology rests will be pointed out, and how this assumption influences archaeological research will be reviewed. These general introductory comments on archaeology, and a subsequent statement on objectives and methods are intended to emphasise the reasons why the Galatea Bay site was excavated and show how the excavation is related to current research on the prehistory of New Zealand.
The Significance of Cultural Patterns to Archaeology
In general discussions about archaeology, much emphasis is placed on the uniqueness of each archaeological site. It is claimed, quite rightly, that by digging, the archaeologist destroys the relationships which exist between the evidence he uncovers and also destroys a good part of the evidence itself. In other words, each time he excavates, he performs an experiment which can never be repeated. Thus, while like the natural scientist, the archaeologist can try to objectify his procedures and observations, unlike the natural scientist he cannot reproduce an identical set of conditions. It would appear that the results of an excavation have a uniqueness about them which cannot be verified by repetition. To stress such
a point of view too strongly, however, obscures the very discovery the archaeologist wishes to make: the discovery of regularities, of patterns, in the phenomena he observes.
The archaeologist believes, as does the natural scientist, that there are statements which can be made about the evidence he uncovers which go beyond the description of specific sites and the expression of unique historical events. To arrive at such general statements, he assumes that much of human action is patterned under recurrent relationships. It is this patterning of human behaviour into regularities which the anthropologist calls “ culture.” For the archaeologist, recurrent associations of prehistoric evidence are the “ cultures ” he studies, and the relationships between the elements in such associations are seen as the result of cultural patterning in the old human behaviour he is indirectly observing.
If this assumption about cultural patterning is correct, and most prehistorians would agree that it is, then two conclusions can be made about the testing of archaeological evidence. First, we can see there must be general characteristics in each archaeological site, as well as unique ones. Second, because these general characteristics can be interpreted as the result of the repetition of behaviour at similar sites belonging to the same cultural pattern, they avoid to a great extent the difficulties inherent in the unreproducible nature of each field excavation. These two conclusions have several implications.
First of all, general statements about a prehistoric culture which are based on the results of a single excavation are only “ inferred general statements This is the case because only by repeated experimentation at other similar sites can a common pattern be fully isolated from the unique aspects of a specific site. Moreover, it follows that conclusions founded on evidence from truly unique sites are only inferred generalities, and as such give the prehistorian very little to work with in his efforts to revitalise the past.
In addition, in talking about general statements a useful point can be raised about validation in archaeology which is a corollary of the observation already made. It is important to recognise that the archaeologist ought to worry about testing his conclusions just as much as the natural scientist must, although no one would pretend that the archaeologist can ever achieve the same degree of accuracy and reliability. General statements in archaeology, therefore, should follow from the evidence at hand and should suggest real ways in which they may be verified by further analyses and excavations. If such validating limitations cannot be applied, then the archaeologist must admit he is only speculating. He is not concluding.
Finally, if the archaeologist is to discover regularities in his data, he should look upon excavation as deliberate experimentation. It is not merely observation. The experimental method seeks to observe isolated phenomena by controlling known variables, and no single experiment ever pretends to handle all possible variations and contingencies. Thus, in archaeology research must be directed toward stated aims and be conducted in a systematic, clearly formulated manner. For example, selection of a site must take into cognisance the kinds of questions to be investigated and the actual contingencies which can be known prior to excavation.
Now it may be objected that excavation is not experimentation because the archaeologist cannot manipulate the factors involved. In as far as this objection goes, it is quite acceptable. The archaeologist does not manipulate past events. But he does manipulate the surviving traces of such events and it is here that he must exercise control over what he investigates. How a site is selected, what excavation techniques are used and how the results are analysed and interpreted: in all these aspects of research he must be objective, explicit and purposeful.
On the following pages the excavation at Galatea Bay will be described and its objectives, methods and results detailed. The overall purpose of the excavation can be briefly stated: by this initial excavation of one beach-stream midden it was planned that we could arrive at certain inferred generalities about this kind of site in New Zealand, the validity of which would rest on further testing at other similar sites. In the various sections of the paper the following subjects are discussed :
1. History of the research. 2. The location of the site and the reasons for its selection. 3. The objectives of the research and the methods used. 4. The observations made and how these were analysed. 5. The range of conclusions which can be proposed. 6. Some of the broader implications for the objectives and limitations of archaeology in New Zealand which can be raised because of the present research.
History of the Research
The Galatea Bay site was first recorded on 13 March 1965, when the author visited the locality. On 4 April arrangements were made with the owner, Mr G. Chamberlin, for an excavation there by the Department of Anthropology and the Archaeological Society at the University of Auckland. On 11 April Mr K. M. Peters of the Department of Anthropology and the author arrived at the site to set up field operations and prepare the site for excavation. In accordance with the research programme which had been formulated for Galatea Bay, they conducted a test excavation in the south half of Square D-l which established an initial knowledge of the stratigraphy.
The full labour force of twenty-five student and amateur excavators worked from 16 to 19 April under the direction of the author and Mr W. Shawcross of the Department of Anthropology. Afterwards, a small labour force remained until 25 April to complete the recording, photographing and mapping of the excavation.
A final visit to the locality was made on 28 July when Mr Peters, Mr A. Michael of the Department of Zoology, University of Auckland, and the author returned to take samples of the living shell fish population from the shore of the bay.
Location of the Site and the Reasons for its Selection
Ponui Island is situated in the Hauraki Gulf some twenty miles directly east of Auckland (Fig. 2). Less than a mile to the north-west lies Waiheke Island. Ponui, also called “ Chamberlins ” after the family which has owned it since the middle of the last century, is five miles (eight kilometres) long from north to south, and 2.3 miles (3.7 kilometres) wide near the centre from east to west. The land rises over 560 feet (170 metres) above sea level toward the middle of the island. The topography is rolling but moderately steep. Even at the water’s edge the land is frequently over 100-150 feet (30-45 metres) in height where it drops off abruptly in sharp cliffs of exposed rock or falls rapidly in valleys to small bays with short beach fronts. The soil, technically a yellow-brown earth which has been strongly leached (Pohlen, 1965), has developed on an old sedimentary base formed during the deposition of greywacke sands and muds in the New Zealand Geosyncline during the Permian-Jurassic geological periods (Ballance, 1965). Most of the island today is covered in grass and is used for livestock farming. In the past, however, the island was probably covered in forest. Detailed research on the pre-European vegetation has not yet been done. It is known that when the French explorer Dumont d’Urville sailed between Waiheke and Ponui on 27 February 1827, he was impressed greatly by the islands in this part of the gulf. He reported, unfortunately not by name, that some of the islands “ were lofty and mountainous, covered with magnificent forests, others lower and only covered with more ordinary vegetation” (d’Urville, 1950: 163). By extrapolation, it is proposed at least two of the islands “ lofty and mountainous ” must have been Waiheke and Ponui.
The Galatea Bay site (N-43/33) lies beside a small, shallow bay locally known by that name on the north-west side of Ponui Island. Directly to the north is a similar embayment known as Crescent Bay, and to the south, another called Rabbit Bay. All three have sandy beaches in front of erosion valleys which are separated from each other by ridges ending in steep cliffs at the sea. The beaches at Galatea and Rabbit Bay are roughly the same in length, approximately 150-175 yards (135-160 metres) long. The beach at Crescent Bay, however, is over 400 yards (365 metres) long. Both Crescent and Rabbit Bay have an area of flat land behind each of the beaches before the land rises to the ridges over 200 feet (60 metres) high. There is very little flat land behind the beach at Galatea Bay for the steep sides of the small valley fall almost directly to the sea.
Galatea and Rabbit Bay each have a single valley stream. Crescent Bay has two. During April, 1965, only the streams at Crescent Bay had flowing water. The stream at Galatea Bay carried no surface water, while that at Rabbit Bay held ponds of stagnant water with only a slight flow at the beach front. It seems reasonable to assume, however, that if the island was forested in the past, water flowed in all the streams at least part of the year. In spite of aggradation of the stream bed due to erosion and the probable lowering of the water table because of deforestation, surface water was observed in the stream at Galatea Bay during the winter when the site was revisited in July, 1965.
Beside each of these three bays are extensive deposits of shell midden with cooking or hangi stones, ash and charcoal. These shell deposits are well known to the local inhabitants who report that skeletal remains, said to be human, were pulled from the midden at Rabbit Bay several decades ago. Because the shell middens in all three valleys are directly associated with bays where fish and shell fish could be obtained and also lie next to fresh water streams, these two important associations have been combined in calling these midden sites “ beach-stream midden sites ”.
At Galatea Bay, shell midden can be seen in the raised beach front on both sides of the stream (Fig. 3). Each midden grades out toward the stream bed. Thus there are two separate midden areas along the beach front because while the stream separates the two middens, it does not bisect them.
It is the midden layer to the north of the stream which is of particular interest for it is here that excavation was conducted. The deposit rests below the valley side on a sloping bench of land 10-13 feet (3-4 metres) above the high water mark and roughly 115 feet long by 33 feet wide (35 x 10 metres). The midden is well preserved. The bench face has been stabilised by grass except for a few exposed patches, and slumping has been minimal.
The midden stratum on the south side is 8 inches (20 centimetres) thick. It is buried beneath a sandy turf 7 inches (18 centimetres) in depth, and rests on a moderate slope, which may have been artificially terraced, at the foot of the valley. The natural section at the beach front is highly eroded and affected by slump action. The section shows several distinct stratigraphic zones below the midden layer, but these cannot be adequately interpreted without detailed pedological analysis. None of these lower zones appears to be cultural in origin. The lowest level visible is a well sorted sand without shell, which suggests it may be an old dune formation.
In addition to these two major midden areas, there are also several other archaeological features in the immediate vicinity. On the valley slope directly above the south midden area soil slumping has exposed a scatter of shell. On the ridge is an artificial pit. At the end of the ridge by the sea there is a small flattened area which may be the vestige of a terrace. Just inland behind the terrace are several pit-like depressions. If these are rectangular pits, little remains of them, because erosion has carried much of the cliff edge down to the waters below. The large pit and possible vestigial pits and terrace have been designated by a separate site number (N-43/34) in accordance with the practice of the New Zealand Site Recording Scheme. The relationship of the pits and terrace to the beach middens is unknown.
The last remaining feature in Galatea Bay which may have archaeological importance is the peninsula outcrop which runs out into the gulf on the south side of the bay. Today the sides are steep and very weathered, and there is little flat land along the top. K. M. Peters of the excavation staff attempted to investigate the possibility of traces of occupation there. The ascent proved to be impossible without special equipment. Mr G. Chamberlin, the owner of the land at Galatea, reports, however, that a ditch-like feature exists on the crest.
These are the features of Galatea Bay and its location. Figure 1 shows the appearance of the locality today. There are a number of reasons why the bay was selected for excavation. Here was a well-preserved midden site of small dimensions which was physiographically very well defined and delimited by the small valley itself. The middens showed a variety of shell fish, and evidence for at least one domestic activity, that of cooking, existed not only in the presence of charcoal and burned cooking stones, but also in a naturally sectioned hangi or cooking pit on the south side of the beach front. The bay suggested the source of the shell fish. The stream bed suggested a probable source of fresh water. With only a limited amount of time available for excavation, Galatea Bay seemed to offer an excellent place to investigate the question of the function of at least one kind of shell midden site. It is to the further examination of this question of the function of middens in Maori settlement, as well as to the specific objectives and techniques adopted at Galatea Bay, that it is now necessary to turn.
Objectives and Methods
During the development of New Zealand archaeology much attention has centred on sites belonging to the “ Moa-Hunter Period ” in Maori prehistory (Duff 1956: 13, 16-17), a theoretical phase also referred to by some as the “Archaic Phase of New Zealand Eastern Polynesian Culture” (Golson, 1959: 36-37). In the last six years, however, great interest has grown in the study and excavation of the Maori hill-forts, the pa, which are usually assigned to a later period termed the “Classic Maori” phase or culture (Golson, 1959: 47-48, 54). Major excavations at pa sites have been carried out. This research has uncovered sequences in terms of certain constructional features and some artefactual materials. It has only been in the last two years that attempts have been made to excavate enough area inside two pa to give an idea of the range and spatial distribution of archaeological elements within the fortified locations (Groube, 1965; Shawcross, 1964), although work by Smart (1962 b) along the same lines had somewhat earlier indicated the advantages of wide area excavation.
Principally because of theoretical discussions by R. C. Green and W. Shawcross of Auckland University (Green and Shawcross, 1962; Green, 1963 a) and ethnohistorical study by L. M. Groube of Otago University (Groube, 1964 a, 1964 b), the interests of pa site archaeology have been extended to include the more general problem of what were the prehistoric and protohistoric settlement patterns in New Zealand. For example, Groube has raised the question of the actual function of the hill-forts (Groube, 1964 b; 138-140). By re-examining the earliest European accounts, he has demonstrated how imprecise are our ideas about the Maori use of pa. At a more general level, he has described evidence which contradicts former concepts of Maori settlement patterns.
In the older point of view variously articulated by scholars such as E. Best, P. Buck and R. Firth, the prehistoric settlement pattern in New Zealand consisted basically of two elements. The first of these was the fortified village, the pa. The second was the unfortified village, the kainga. Some scholars have held these were two contemporary kinds of settlement (Best, 1952: 254; Firth, 1959: 91-93; Golson, 1957: 71-72; Vayda, 1960: 10-11). Others have declared that with an increase in the occurrence of warfare in prehistoric New Zealand, the kainga gave way to the pa (Buck, 1962: 137-139). In addition to these two kinds of villages, Firth, for example, also notes the existence of temporary camps (1959: 93). The difficulty with this settlement pattern picture lies in the definition of “ village ” in each case.
According to Firth (1959: 91-94) the prehistoric Maori lived in “village communities ” composed of a number of huts of various sizes and degrees of workmanship and containing also a plaza or marae, a large village meetinghouse or whare runanga, storage facilities and other defining features. An individual village consisted of a number of household groups in the form of nuclear or extended families (1959: 105, 110-111). In general, a village was inhabited by one extended kinship group or hapu, which may be designated loosely by the term “ clan,” which was made up of a number of related households, although a large village might have had several hapu residing together (1959; 113),
Firth has clearly explained what he considers to have been the relationship between fortified and unfortified villages (1959: 92). He states that pa differed from kainga virtually only in the presence of defences around the former: “pa and kainga exhibited the same essential features, although the fortifications of the former were of necessity so laid out as to conform to the exigencies of the site, and no consistent shape of settlement was in vogue. In regard to social life and institutions, apart from war, one may speak equally well of either ”. As indicated above,
this kind of interpretation is often encountered in the literature, both in archaeology and anthropology (for example, Piddington, 1963: 167-168), It is, however, not the only one.
Best, in his extensive description of Maori hill-forts (1927), presents a more complicated picture of Maori settlement than the one often encountered. While he does draw attention to the existence of “ open hamlets ” or unfortified villages as well as “fortified hamlets” (1927: 4), he records the Maori had names for a number of kinds of fortified sites which ranged from a term for any kind of fortified place to a term for simple retreats for non-combatants during a war, places frequently not fortified in any artificial way. Included in this wide range are the terms pa kokori and pa korikori which “ were applied to any defences of an inferior nature, such as a few huts surrounded by an ordinary type of palisaded barrier. Such places were often constructed at cultivation grounds away from the fortified village, and at fishing camps on the coast” (1927: 14). Most important, however, is Best’s interpretation of the function of pa:
Nor must it be supposed that the Maori village community lived permanently in the fortified village. When no attacks were feared from enemies the people often lived outside the fort, and even moved away from it to live for a while on the sea coast, or in the forest, when engaged in fishing or bird taking operations, etc., or at their cultivation grounds when planting, tending, or lifting crops. In many cases the community lived in a village protected merely by a fence or stockade, but had a strongly fortified pa in the vicinity to retreat to when danger threatened (1927: 19).
The late prehistoric settlement pattern now proposed by Groube (1964 b; 82107) after an examination of the accounts of early explorers most resembles that given by Best, but differs in the definition of occupation within pa and undefended sites. Settlement at the time of first European contact in 1769 may have consisted basically of scattered undefended “ hamlet sites ” of only three or four houses associated with separate cooking sheds and a communal dump. Moreover, these hamlets may have been used only by one or two extended families and may have served only particular specialised functions such as seasonal activities (fishing or cultivation). Thus, they cannot be called kainga in the traditional sense because of their limited size and shifting utilisation throughout the year. Settlements with a number of huts, which might correspond with what ethnographers working with the historic Maori have termed kainga, seem to have been rare. In addition, pa may only have been fully occupied by entire communities during periods of crisis or during particular seasons. Groube suggests:
The pa then is not necessarily permanently occupied, but seems to be the centre of a more extensive settlement pattern of which it is the citadel. Seemingly in the majority of cases, it was occupied only during crisis with only ‘ a few remaining ’ on the pa. Everyday economic activity, at least during summer, was carried out in dispersed hamlets or huts (Groube, 1964 b; 102).
Two important points can be drawn from this brief discussion of the various conceptions of the prehistoric settlement pattern in New Zealand. First of all, in view of these conflicting interpretations of the ethnographic data, it is apparent that there is great need for an archaeological examination of settlement evidence in the hope that by such study, patterns can be determined both for the late prehistoric Maori which may be compared with the ethnographic data, and also for earlier times in New Zealand. Second, and following from this point, to arrive at such patterns it is necessary that sites other than pa be excavated as well, if it is to be decided whether or not undefended kainga or specialised hamlets in fact once did exist.
Only a few late prehistoric sites* other than pa have been excavated. Most of these have been middens (for example: Dawson, 1949; Dawson and Yaldwyn, 1952; Fomison, 1963; Green and Pullar, 1960; Hunt, 1962; Nicholls, 1963 and 1964; Skinner, 1953; Spring-Rice, 1963; Trower, 1962; and unpublished excavation by the Department of Anthropology, University of Auckland, at Smuggler’s Cove, Whangarei, in 1964). Other work has been done on the rectangular pit sites which are alternatively interpreted as storage areas or house sites (Batley, 1961; Green, 1963 b; and Parker, 1962), on above-ground houses (Buist, 1962), and on cave sites (Batley, 1961; Hosking, 1962; and Wright and Bennett, 1964). Systematic study along these lines, however, has only begun.
As Elsdon Best long ago commented (Best, 1927: 98), beach middens are extremely common along the shores of New Zealand. He interpreted them as refuse dumps near the cooking sheds of “ old time villages ”. Although a limited amount of excavation has been carried out on beach middens not attributed to the Archaic phase of Maori culture, no site, prior to the one under discussion, had been systematically excavated in order to demonstrate its function in Maori settlement and determine the range of archaeological evidence which such a site contained. Thus, it seemed desirable to excavate a beach midden to gain these kinds of information to complement in part that available from pa excavations.
Once the Galatea Bay site had been chosen for study, careful consideration was given to the selection of the most suitable archaeological methods by which the desired information might be obtained. An excavation limited to a small area of the site would give only a stratigraphic sequence. Therefore, it was imperative
that a wide enough area be opened to secure evidence on the full range of activities at the site, the spatial distribution of that evidence, and, in turn, the general function of the midden. The method of excavation finally chosen was area excavation of all of the site which it was feasible to examine.
The size of the excavation grid (18m x 6m) was settled upon after field examination of the site (Fig. 3).* The southern end was limited by the presence of the root system of the large pohutukawa tree (Metrosideros tomentosa) . Study of eroded areas at the bank suggested the midden probably graded out over or beneath the old slope slump to the north, a prediction which proved to be correct. The northern end was arbitrarily determined to include only part of the midden where it seemed to grade out. The western edge was set back from the raised bank to prevent erosion and allow for a causeway for spoil transport to the dump established north of the site. A similar causeway was left at the foot of the slope on the east side of the excavation. Time limited work to the midden alone.
Stratigraphy of the Test Excavation, Square D-l (Fig. 4)
The southern half of square D-l was excavated before the rest of the site to gain an impression of the stratigraphy of the midden which could be used to predict to some degree that in the rest of the area to be studied. Five stratigraphic zones were distinguished:
Layer A: 5-1 Ocm thick. A dark brown humic sandy turf. No other constituents.
Layer B: “ Upper ” 10-22.5 cm thick.
Concentrated shell in a black sandy matrix. The shell, mostly pipi (Amphidesma australe ), was largely unbroken. Other constituents were bone (mostly fish bone), small lumps of charcoal, and burned fire stones (hangi stones), most of which were small, well-broken and angular, except for whole stones actually in position in a hangi at the bottom of Upper Layer B in the south-west corner of the square. The midden contained no visible smaller subdivisions or lenses, although during excavation, small concentrations of shell with little matrix were sometimes encountered which suggested small dumps of shell within the relatively homogeneous midden deposit.
“ Lower” 3-20 cm thick.
Less concentrated shell in a black sandy matrix. The shell in this zone was more broken and less concentrated than the shell in Upper Layer B, although the other constituents were the same. Because of the higher proportion of charcoal black sand to shell, this layer gave the appearance of being a darker black-grey in colour than Upper Layer B.
Features:
The portion of a large hangi found in the south-west corner of the square consisted of a large shallow basin approximately 150 cm in diameter and 20cm deep which was filled with a compact mass of large, angular cooking stones in a loose, black, sandy matrix. In square D-l the basin had been cut through the Lower Layer B zone and slightly into the underlying Layer D. The basin with the contained cooking stones was itself filled with the debris of Upper Layer B.
The two other features in the test square which are shown in Figure 5 contained no concentrations of cooking stones. The feature in the north-west corner was a basin 12—13 cm deep which was filled with a sandy dark grey matrix and appeared to have been cut from Upper Layer B. The other feature was a depression at the base of Lower Layer B which held patches of concentrated charcoal.
Layer C: 8-20 cm thick.
A layer of yellow sand in the east half of the test square could be differentiated from the underlying Layer D by the presence of shell, mostly broken, small lumps of charcoal, pebbles and fragmented stones.
Features:
Removal of Layer G exposed two basin-shaped pits in the south-east and northeast corners of the test square which were 10— 12cm deep. These pits lay at the base of Layer G and were filled with the materials which made up that layer. The fill of the pit in the north-east corner had a visible band of charcoal within it (Fig. 4: “ash layer”). Moreover, the surface of the underlying sand, Layer D, between the two pits was red in colour, probably because it had once been heated by a fire. For descriptive purposes, pits of this kind on the site have been called “ash pits” (Fig. 6).
Layer D:
Natural yellow beach sand with water-worn shell fragments, small rounded pebbles, and small weathered fragments of the bedrock of Ponui Island.
Interpretation of the Stratigraphy in the Test Excavation
The stratigraphy in the test square was interpreted in terms of a sequence of events which the subsequent area excavation was able to test and augment. This initial reconstruction can be briefly detailed.
It was felt that Layer G was sufficiently dissimilar to either subdivision in Layer B to suggest that it probably denoted at least a distinct phase in the utilisation of the site, if not a different period of occupation. The two pits seemed to be the most significant aspect of Layer C. While they contained some charcoal and a few fragmented cooking stones, they were quite unlike the hangi in Layer B. Yet the presence of charcoal and stones in the fill and a red fire zone did suggest they were small cooking pits of some kind.
Because the fill of the pits was the same as the material of Layer G, it appeared the only activity represented by the layer was that associated with the pits themselves. Certainly this would have been the case if Layer G was a natural accumulation after the use of the pits. An alternative interpretation, of course, was that the pits had been deliberately filled in.
Layer B was more complex than Layer C. Three different phases could be interpreted. First there was the deposition of Lower Layer B. Two explanations for this zone seemed possible. Either it represented the formation of a midden dump which merely contained less shell or it was composed of material raked out of cooking pits. In the latter case, an associated midden dump might be elsewhere on the site.
Second was the construction of the cooking pit in the south-west corner. It definitely appeared to post-date the formation of Lower Layer B, and date before the deposition of at least most of Upper Layer B. Third was the formation of the concentrated shell midden of Upper Layer B which filled and covered the cooking pit. Since Upper Layer B was explained best as a dump of debris created by cooking, it seemed probable cooking pits would be found in or adjacent to this level, unless these had been eroded away by the sea.
General Stratigraphy of the Area Excavation (Fig. 7)
The prediction of the general stratigraphy made on the basis of the evidence from the test excavation was so successful that it is unnecessary to alter the details of each layer already given. The only difficulty in excavation arose during the tracing of Layer C: construction of the cooking pits of Layer B seems to have destroyed much of the underlying layer and it is likely it may not have been originally a continuous deposit.
Layer B: “ Upper ”
The concentrated shell midden of Upper Layer B extended over p the excavated. The thickness was not uniform: ranging from between 10-25 cm m souares B-2 to E-2: 15-22.5 cm along the east wall of squares B-l to Ed, 10-15 cm along the west wall of squares B-l to E-l; and grading out at the north endTthe site to 8-12.5 cm snd at the south end to 5-10 cm While the amount of shell in the midden varied with each square, it was m |f k ''xhe'' const the northern limit of the excavation (Fig. 10: midden sample 2). Th centration of charcoal in the matrix decreased toward both ends of the site, umy five cooking pits were found within the layer (Fig. 9a) No. .significant midden sub-divisions were discriminated although a poss:ihle Icycl within Uppi La B denoted by a scatter of ash, fragmented shell and cooking stones, was found around the large hangi in square E-l han a depth of approximately 10cm below the surface of the layer (Fig. 5).
“ Lower ” Lower Layer B was less homogeneous than the upper division. It varied between the black-grey matrix described in the test excavation to a grey sandy matrix. The distribution of the layer shows that it did not extend into squares A-l and X-2 and that at the other end of the site, the layer was a thin greyish zone 8-12 cm thick on the south side of square F-l which graded out entirely in the eastern half of square F-2. No concentrated shell midden was uncovered m associ ation with this layer.
Examination of the stratigraphy supports the interpretation of the deposit as one formed during the utilisation of the cooking pits Figure fi this explanation. There the deposit is linked directly to a hangi. Conceivably the dark grey sandy material was formed during the raking out of ash an charcoal after the stones were heated. Such a practice is recorded for the historic Maori (Best, 1924: 417). When the pit was abandoned, shell was dumped back into the cooking pit with the stones still in place. Subsequently, the area was
covered with the shell midden of Upper Layer B. The hangi shown in the west square on the North 625 cm section (Fig. 7) also had a similar spread of blackgrey sand to one side of the pit. This characteristic was clearly observed for the hangi on the North 1525 cm section of square F-2 and the South 575 cm section of square B-l.
Layer C;
Layer C proved to consist of two distinct but contiguous deposits. In square E-l, E-2, D-2 and G-2 the layer was the same as in the test excavation, in square D-l. Six “ash pits” were found at the bottom of the layer in square G-2. An additional pit was uncovered in square D-2, but it could not be stratigraphically linked directly to the other pit features. The small “ fire basin ” in square E-l also belonged to this level. This feature was a shallow (scm) basin containing white ash. The sand at the bottom was red.
The contiguous layer was a deposit of charcoal and ash 2.5-Bcm thick in the northern half of square G-l, 2.5-scm thick along the northern wall of square G-2 and the southern quarter of square B-2; and 2.5 cm thick in the test area of square B-l (Fig. 6). Removal of the baulk and midden block between squares C-2 and B-2 showed that this zone was directly connected to the Layer G deposit in square G-2 (Fig. 7). Moreover, in squares B-l, B-2, G-l and G-2 the charcoal zone was covered by a layer of yellow sand 2.5-15 cm thick (Fig. 7) which lensed out toward the middle of squares G-l and G-2. Thus, in these squares Layer G was stratigraphically separated from Layer B by a layer of sand. Lying within the charcoal zone were two concentrations of cooking stones which were not in cooking pits (Fig. 6 and 7). In square B-2 where the zone was less extensive, a circular patch of charcoal was found at the same level. The surface of the sand below was red. This latter characteristic was seen clearly also at the bottom of the charcoal zone in square G-l.
Layer D:
Beach sand lay at the base of the excavation over the entire area excavated. A small excavation made in the north-east corner of square A-2 to a depth of 90cm below the surface revealed no lower cultural horizons below the midden area excavated.
Features:
Layer B: Twenty-two hangi pits containing cooking stones in varying concentrations were uncovered in the excavation. All but three were well defined basins roughly 15-25 cm deep. In the remaining three, the basins were more difficult to distinguish from the surrounding matrix. On the composite plan of the features in Layer B (Fig. 5), concentrations of cooking stones found in the midden have also been indicated. These concentrations were not associated with cooking pits, and seem to have been merely dumps of stone in the midden layer. Most of the stones used were angular lumps of the local greywacke and could have been obtained easily from the nearby outcrops. In addition, there were rounded beach pebbles of a coarser-grained sedimentary rock which may also be of local origin, although they could also be pebbles of sandstone from the Waitemata Group in the Hauraki Gulf. The conclusion important to archaeology is that none of the hangi stones was imported to Ponui Island. Today, even the pebbles can be picked up from the beach in limited numbers.
Stratigraphic analysis of the hangi pits (Fig. 9) indicates that all but five lay under the midden deposit of Upper Layer B. This stratigraphic distribution supports the inference that Lower Layer B was created by the activity of cooking itself and suggests that at any one time the cooking area was distinct from the dumping area. Most of the hangi were found directly beneath the midden (Fig. 9b), but a few seem to date somewhat earlier than the rest because they were found either in or below the accumulated debris of Lower Layer B (Fig. 9c). From this evidence it can be concluded that at the beginning of the Layer B occupation, at least most of the area excavated was used as a cooking site. Any midden build-up must have been elsewhere, perhaps to the front of the cooking pits where the site is now eroded. Subsequently, the cooking area must have shifted, and the old hangi were covered with midden debris.
GALATEA BAY N-43/33 Layers
STRATIGRAPHIC POSITION OF FEATURES
Features within layer B (upper)
Features directly below layer B (upper )
Features in or at bottom of laverß (lower)
Layer C: Nine “ ash pits ” and one “ fire basin ” were found at the bottom of Layer G. The ash pits were approximately 10-15 cm deep and 50-65 cm in diameter. Interpretation of their function is uncertain. Because they were associated with a contiguous charcoal area with two distinct concentrations of cooking stones, it is likely the pits may have been a form of hangi which Best (1924: 419) has said the historic Maori were known to have sometimes favoured. In this variety of hangi, called umu konao, the stones were heated in a separate fire, and then placed in a cold pit. The rest of the cooking process was identical with that for the more traditional form. If this interpretation is correct, it would be the first known time this cooking technique has been identified in an archaeological excavation. This interpretation does not explain the red sand zone between the two ash pits in square D-l. Since this red zone was on the surface and not actually in the pits, it may very well have been formed before the construction of the pits and be unrelated to them.
Post holes: 88 post holes were found in the excavation. They were 3—lscm in diameter, with an average of roughly 7cm. The holes seem to have been made by small stakes driven a short way into the sand. It was impossible to trace them in the midden layers. The only time they were distinct was when they appeared in the underlying sand. Examination of the eight post holes which were found on the sections in squares E-l and E-2 (Fig. 6) shows that four appeared to be cut from Lower Layer B and the other four from the bottom of Upper Layer B. They form no recognisable pattern. The most likely interpretation is that they represent former cooking sheds made of light poles, such as those described by ethnographers (Best, 1924: 419, 1952: 254; Buck, 1962; 120; and Firth, 1959: 93).
Composition of the Midden
A small block was left unexcavated in the north-west corner of each square to facilitate the taking of samples of the shell midden for analysis of the composition. Roughly the upper 10cm of the midden in each block was later removed for study. No sample was taken from the block in square A-l because of the low concentration of the midden there. Instead, a sample was removed from the baulk between squares A-l and A-2 (Fig. 5). In addition, two samples were taken from Layer G (Fig. 6).
Analysis of the midden was conducted for a number of reasons: 1. To determine the composition of each layer, the range of variation between the samples, and the difference between the two layers; 2. To interpret the findings in natural and cultural terms; 3. To examine and objectify the analytical methods used; and 4. To evaluate some of the uses of quantitative midden analysis in New Zealand.
A.— lnitial Study
Research began with an analysis of the composition and variation of the midden. The methods used were those described by New Zealand authors (Ambrose, 1963; Davidson, 1964 a, 1964 b, 1964 c; and Smart, 1962 a). Difficulty was encountered in ascertaining the size of the sample necessary for study. No empirical test was known by which a minimum sample could be determined. Arbitrarily, a sample of IOOOgm was taken from eleven of the field samples under controlled conditions to ensure that each was representative of the total sample from which it came. Because of the presence of large stones in sample 9, which Davidson proposed might affect the success of analysis (1964 b: 148), 1500 gm were removed from that field sample. Further, because of the lower concentration of the samples from Layer C, 2000 gm were extracted from each field sample from that layer.
Following in part the procedures already outlined by Davidson (1964 b: 148161), each sample was dried, sifted through £in, £in, £in and sieves, and the resultant fractions isolated separately. Then the composition of the £in and fractions was analysed and weighed. No attempt was made to analyse the shell constituents by the number of each species present. The graph of the composition shows that each sample varied in terms of the amount of each constituent present (Fig. 10). The most abundant single constituent was pipi shell fish (Amphidesma australe) which had an average weight of 32% of the total composition weight. Cockle shell fish (Chione stutchhuryi) averaged only 4%, while the remaining species of shell fish combined averaged 5%. Stone, bone and charcoal combined amounted to an average of 6%. The remaining weight was taken up by the £in, and less than fractions, with the latter having an average of 31.5%. The data are presented in a table at the end of the paper (Appendix).
During this initial study it was observed that while the midden samples did lose weight upon dehydration, as one would naturally expect, the loss had a mean of only 4% (Fig. 11). Although Davidson (1964 b: 147) considered such weight loss to be of significant analytical importance, it had not been demonstrated by empirical study that this loss varied differentially from constituent to constituent. In other words, it would only affect the relative weight of each midden constituent if certain constituents were relatively heavier than others because of their contained moisture. Moreover, it had not been made clear that dehydration was not a steadystate. If drying was a useful procedure, then the samples would have to be maintained at the same degree of dehydration throughout the period of analysis. Otherwise, they would naturally absorb moisture from the air once drying was halted. For example, it was noted that control samples during the drying process gained sgm during the night when the heating element in the drying rack was turned off. Because of these observations, subsequent research was designed to examine and objectify techniques in midden analysis and began with a study of dehydration.
B. —An Experimental Approach to the Quantitative Analysis of Shell Middens 1. The Importance of Dehydration Four test samples, two from square B-2 (sample 4) of IOOOgm and 2000 gm and two from square E-2 (sample 10), also of IOOOgm and 2000 gm, were used
to study the effect of dehydration. These samples were sifted and analysed before drying. It was noted that in the process of sifting as much as 1.6% of the total weight was lost. In other words, an error of as much as 1.6% was introduced into the analysis by sifting alone.
The analysed portions of each sample were then dehydrated until they showed no further weight loss, and the total loss in each case was computed. Next, the individual loss of each constituent was calculated as a percentage of the total loss. The results of these four tests were consistent (Fig. 12). Half the weight lost was given up by the less-than fraction. The rest was more or less evenly divided between the other constituents. Therefore, the only significant differential weight
due to contained moisture occurred in the less-than fraction. Analysis by fractional combustion of a sample of this fraction from square E-l (sample 9) showed it was composed of:
Charcoal 7.8% Organic matter 9.4% Sand and silt 82.8% 100.0%
Thus, as one would expect, the greatest amount of moisture was held by the fraction composed of small particles. To test further the possibility that different kinds of shell fish might hold significantly varying amounts of moisture, two test samples were collected of the most common shell fish in the midden, pipi and cockles. Because of the low concentration of cockles, the samples were of limited size; 80gm of each species were taken from a fraction of field sample 13, and 50gm from the fraction of the same sample. The pipi shells lost 0.8% of their weight in drying and the cockle shells lost 1.5%, a difference of only 0.7%. Conclusion : On the basis of these results, for at least most purposes it is not necessary to dehydrate midden samples before they are analysed to determine their composition.
2. The Size of the Sample
To determine the effect sample size has on the results of quantitative midden analysis, eleven test samples were analysed and the results compared (Fig. 13).* The size range of the samples was 500-2000 gm. Three of them were ones previously analysed in the initial composition study. Three came from square A-2 where the initial composition analysis showed an unusually low shell concentration. Comparison of the results shows a relatively high degree of consistency between the results, regardless of the sample size. While she did not detail her evidence, Davidson arrived at a similar conclusion (1964 b: 149). Although the number of test samples in the present study was not large, several important points may be drawn:
(a) In analysing a portion of a larger field sample, one is, in effect, “ sampling a sample Thus disagreement between analyses of the same field sample must be expected, and is not, therefore, only a function of the size of the sample. Sampling error must be taken into account. This error can be observed by comparing the results of the two IOOOgm samples from square B-2, and also the two from square E-2. Moreover, sampling error seems to be the most logical way of explaining the total disagreement of all four of the less-than tVin sample fractions from square E-2. (b) While not truly conclusive, the results suggest that 500 gm test samples are too small to obtain a reliable estimate of all of the constituents in a midden. On two occasions, constituents were totally absent from these samples which were identified in larger ones (Fig. 13). On the other hand, even the 500 gm samples gave a generally consistent estimate of the constituents present in them. Therefore, for many purposes a 500 gm sample may, nevertheless, be adequate for ascertaining midden composition. (c) Comparison of the IOOOgm samples in square B-2 and E-2 shows that the overall error due to sampling is less than 2%. In Figure 13 sample disagreement of more than 2% has been identified. Even here no clear picture relative to sample size emerges.
Conclusion : Sampling error and not sample size accounts for most of the discrepancies found in a comparative study of eleven test samples.
3. The Effect of Sifting
The use of sieves in midden analysis simplifies research by sorting the constituents in terms of size. It is a convenient procedure. It must be asked, however, what effect this arbitrary sorting has on the results of analysis. Because sifting divides the midden into fractions, then theoretically the mathematics of fractions must be taken into account. That is, if the largest fraction is the one left in the sieve, then analysis of the smaller and fractions should not alter greatly the picture of the composition obtained from the -|in fraction alone.
Study of the midden composition table in the appendix shows that for the midden at Galatea Bay, the largest fraction was that left in the fin sieve. Comparison of the results for shell fish in the midden obtained from the fin fraction with the combined results of both the fin and fin fractions supports the theoretical prediction made above in terms of the mathematics of fractions (Fig. 14). It can be seen that in both graphs, the general pattern of variation remains more or less the same. The range of variation from sample to sample is only moderately affected. Moreover, the mean percentages also change only slightly with the addition of the fin fraction. Even for the largest single shell constituent, pipi shell fish, this change is only 4%.
Conclusion: Within the range of accuracy indicated, at least when it is the major fraction present in the midden, analysis of only the fraction gives a reasonable estimate of the relative proportions of the constituents which it contains.
Unfortunately, no attempt was undertaken during the present study to qualify the phrase: “the major fraction”.
4. Reconstruction of the Archaeological Shell Fish Population
Thus far, three procedures in analysing shell middens have been experimentally studied. It remains to consider the interpretation of the midden evidence in terms important to prehistory. One of the immediate questions concerns the kind of prehistoric activity which resulted in the deposition of the midden. Specifically, can it be shown that the shell fish were selected for a certain size, or were they collected without selection?
At Galatea Bay, only pipi shells existed in sufficient number to permit an estimate of the size range of the shells in the midden. From the initial twelve test samples used to determine the composition of Upper Layer B, the lengths of all possible pipi shells were measured. Next a graph was constructed showing both the actual number of shell fish for each unit of length, and also a generalised length curve based on the percentage of the total number of specimens occurring in arbitrary scm groups (Fig. 15).
The graph indicates that for pipi shell fish, no size selection was operative, because the size distribution curve well approximates what one might expect a normal population curve for pipi to be, although it does show negative skewness. Moreover, since it will be subsequently demonstrated that the food value varies with size, it seems difficult to believe that in practice much nutrition would have been gained from the smaller specimens in the midden. Because these are present, it seems even more likely no size selection was practised. As will be described below, this interpretation was tested against a study of the living shell fish population in the bay at Galatea.
5. The Use of Differences in Shell Fish Proportions to Distinguish between Cultural Layers
Figure 14 shows what appears to be a marked difference between the mean proportions for pi pi in Upper Layer B and Layer C. Although these layers were distinguished on other grounds, it should be possible to use this observed difference in shell fish proportions to further discriminate these two layers. This may be done without asking directly for any explanation for this difference.
In accordance with statistical practice, it was assumed there was no significant distinction between the two means. It was asked: what is the possibility that the two means fall within the same statistically probable range of variation? If it could be shown that this was unlikely, then it would be permissible to use the observed difference to distinguish the layers. It was further assumed that the standard deviation for Layer G was the same as that for Upper Layer B. This assumption was necessary because of the limited sampling from Layer G. Even then, it was necessary to deal with the small number of samples from Upper Layer B, and for this reason, the Student’s T statistical formula was used. Calculation showed that the likelihood that the two means belonged to the same range of variation was improbable. In other words, the observed difference between the mean for Upper Layer B and Layer G was highly significant.*
It should be emphasised that in contrasting cultural layers in terms of a quantity which varies within each layer, it must be demonstrated that the differences used are, in fact, statistically significant ones. In studying the composition of shell middens in the Kauri Point locality, Green failed to take variation into account, as Davidson showed (1964 b: 108). Therefore, his seriation of the shell middens is not acceptable (Green, 1963 a: 147-150).
6. The Living Shell Fish Population
Although some New Zealand authors have concluded it is not necessary to have a detailed knowledge of the structure of the living and archaeological shell fish populations to interpret shell middens (Davidson, 1964 b: 167-168, 182-184), the present study has produced evidence which makes it impossible to concur with this opinion. Without an understanding of living and archaeological shell fish populations it is impossible to assess much of the evidence obtained from midden analysis, and impossible to decide whether natural or cultural factors should be used to explain the analytical results. This latter point is particularly important, because until natural explanations can be ruled out, it is unwise to assume the validity of cultural ones.
The advantage of working in a small locality becomes apparent in studying the living shell fish at Galatea Bay, Until evidence arises to the contrary, the assumption seems valid that the small bay was the source of the shell fish in the midden. Living or dead specimens of all the species found in the excavated midden could be picked up from the beach. Therefore, it is only necessary to deal with one small collecting locality.
Sampling of the shell fish in Galatea Bay was carried out using procedures developed by Mr A. Michael of the Department of Zoology, University of Auckland.* These procedures were experimentally devised during a current study of pipi and cockle shell fish populations in New Zealand. No special equipment is required. All shell fish in an area 50cm square are collected. Analysis of samples from larger areas has shown they are not statistically more reliable.
Four samples were obtained from the bay, one at the low water line and the rest at four-metre intervals up the beach until no further living shell fish were
encountered. The results are presented in Figure 16.* For comparison with the reconstructed archaeological populations, the four samples also have been combined in a fifth graph, because there is no reason to assume that the prehistoric shell fish were collected from only one level of the beach. The following inferences can be made precisely because of the evidence gained from this study of the living population:
(a) Examination of the distribution curves reveals that each is dominated by a single size class, except Sample 1 where there are two dominant size groups. Research by A. Michael has shown that this dominance indicates the numerical superiority of a single year group. Other year groups, as a result, have not been able to become as well established in the beach. This dominance is maintained over the life span of the year group (which is estimated to be four-five years at Galatea Bay, although elsewhere in New Zealand it has been found to be as high as seven-ten years).
(b) The size class of the specimens in the dominant group in each sample decreases from sample to sample with increasing height above the low water line. This decrease in size with increased height can be explained by taking into account the shorter length of time under water for each higher sample. At a given height there is a restriction on the maximum size of the shell fish which can exist at that level in the beach because of the decreasing length of time for feeding due to tidal fluctuation.
(c) While the other samples form roughly normal distribution curves, Sample 1 is bimodal. It appears there are two dominant year groups. The most reasonable explanation for this observation is that competition with the cockle population in the bay, which was only well represented in this sample, has divided the pipi population here more sharply into year groups. The presence of cockles at this level of the beach can be accounted for, in part at least, by the ecological change at this level from a sandy to a more muddy one.
(d) The presence of cockles in Sample 1 is significant for another reason. In Upper Layer B of the midden the ratio of pipi to cockles is 90.2% pipi to 9.8% cockles. In Sample 1 it is 88.3% to 11.7%, also by weight. Only a single additional cockle specimen was found in the other samples (No. 2). The combined ratio for all four samples is 96.2% to 3.8%. Although in this latter ratio the frequency of cockles is lower than in the midden, it is similar enough to suggest the ratio in the midden is a natural one. In other words, the proportions do not indicate a prehistoric cultural preference for pipi shell fish over cockles. This conclusion is supported by other evidence.
(e) First, as it was earlier concluded, the pipi shell fish in the midden were gathered without selection. The cockles could have been gathered up along with the pipi without deliberate selection for them. Sample 1 supports this likelihood. Second, analysis of the midden has shown the minor presence of certain other shell fish species ( Cominella glandiformis, C. adspersa and Zeacumantus subcarinatus) which were found also in Sample 1 and 2 along with cockles and pipi. It is difficult to conceive of deliberate selection for at least the two smaller species (C. glandiformis and Z. subcarinatus). This evidence supports the conclusion that all shell fish from the beach were simply gathered or dug in mass without selection either for size or species. Third, sampling of the living population was conducted at only one place on the beach. It was observed that the occurrence of cockles was higher toward the southern end of the beach, but the tide prevented sampling from this area. Nevertheless, since there is no reason to assume shell fish collecting took place only in the area sampled, this observation alone explains why the ratio between pipi and cockles in the combined beach samples does not match precisely the ratio in the midden: the microecology varies from place to place in the bay and with it, the frequencies of the two species.
(f) The ratio of pipi to cockles in Layer C, based on the limited evidence of only two samples, is 54.8% pipi to 45.2% cockles.* Although perhaps not as reliable, the reconstruction of the pipi population curve for Layer G (Fig. 17), and the presence of the minor species of shell fish (Appendix: Samples 13 and 14) both again suggest unselective gathering of shell fish from the beach. It appears, however, that the conclusion may be acceptable that “ cultural preference ” instead of “ natural ratio ” better explains the pipi- cockle ratio in Layer G. Some caution must be given on drawing a conclusion of this kind even when there is a major disagreement between midden proportions and present-day ones. There are a number of reasons for this caution. The length of time separating the occupation of Layer G and Layer B is unknown. Although it seems unlikely, the ecology of the bay may have changed. Alternatively, according to A. Michael, shell fish exploitation may have altered the relationship between the two species. In reference to this last possibility, it must be remembered that the midden to the south of the stream bed was not studied. It may possibly represent exploitation of the bay prior to the deposition of Upper Layer B in the midden to the north. Indeed, unlike that for Upper Layer B, the population curve for Layer G shows dominance by one size class, a situation comparable to that in the present population. There is no reason to assume there is at present any significant exploitation of the shell fish in the bay. Since Layer G is stratigraphically earlier than Layer B, this similarity between the two populations may also indicate there had not been any recent exploitation of the beach prior to the deposition of Layer G and that the cockle population was formerly more vigorous. The effect of exploitation on shell fish populations will be discussed at greater length below.
(g) If the proportions in the midden layers are naturally derived ones, the microecological variation in the bay suggests that further experimentation will show that the proportions of cockles and pipi are different for each bay on Ponui Island. Should this be the case, then it would be impossible to interpret differences in these proportions in the middens of the island as indicative of change over time. Thus, it would not be possible to attempt any kind of chronological seriation of the middens similar to that proposed by Green at Kauri Point (Green, 1963 a: 147-150). Indeed, these data suggest that before any chronological use is made of the coc\de-pipi ratio, it must be demonstrated that wide-scale ecological change or cultural preference was operative in the particular locality under study. Moreover, even if seriation is attempted for proportional differences between shell fish from different kinds of selection areas (e.g.; sandy shores and rocky shores), it must be shown that natural explanations such as the accessibility of one selection zone over another do not more economically explain midden differences than the shifting of cultural preferences with time.
(h) Comparison of the archaeological population in Upper Layer B with the composite distribution curve for the beach samples (Fig. 16) indicates the range in size of the population has not changed since the time of the deposition of the midden. However, while the living population shows clear dominance by one size class, the curve for the archaeological population is less steep and reflects a higher occurrence of smaller specimens. If this difference between the two curves is not explicable in terms of a methodological error in the archaeological reconstruction, this disagreement must be explained.
(i) The divergence of the archaeological curve from the pattern of the present population may mean shell fish from some other population were brought to the site. Although this interpretation cannot be ruled out, it is not necessary to rely on it alone. Lack of dominance by a single year group can also be explained by the effects of exploitation and the passage of time. Heavy exploitation of the shell fish bed could have altered the population by permitting other year groups to settle in the beach in high numbers. But the effect of exploitation, however, would take two or three years to become obvious in the distribution curve. This implication suggests the beach, if not the midden, would have had to have been used for more than one year to allow the observed structural change to appear in the midden. On the other hand, use of the beach for a number of years, even without heavy exploitation, also could have resulted in the archaeological curve. In other words, the distribution curve may reflect the cumulative effect of sampling the beach over a period of years. Thus, even if exploitation was not heavy enough to alter the structure of the beach population, sampling from the same dominant year group over its life-span would have produced the high frequency of several size classes. In either case, use of the locality in more than one year is indicated, but it is not possible to specify how much longer it took to form the midden, because it could reflect the cumulative effect of the size growth of more than one year group. On the evidence at hand, it cannot be decided whether exploitation or cumulative sampling best explains the archaeological curve.
(j) The amount of meat in an adult shell fish varies with the reproductive cycle. Sampling of the population was done only once. Although it was not possible to take this seasonal variation into account, the living pipi population was used to construct a curve showing the amount of flesh to shell in each size class (Fig. 18).* The curve shows that the proportional amount of meat decreases as the size of the shell increases. In reading the graph, for example, the amount of meat in shell fish 4.0-4.4 cm long is found to be 25% of the weight of the shell in that size class. In the paper by W. Shawcross on the economic interpretation of the midden, it has been possible to estimate the amount of meat represented in the midden by calculating with this graph and the graph of the reconstructed archaeological population.
G.—Summary of Findings on Midden Analysis
This section has sought to do more than describe the composition of the midden. Three analytical procedures previously recommended have been studied experimentally to objectify the effect each has on analysis. It should be possible to use a knowledge of these effects to determine what variations on these procedures should be adopted in a specific research programme. The conclusions given have been based on the assumption that great accuracy in midden analysis contributes little to the actual archaeological interpretation of the evidence. It has been implied that for archaeologists to derive statements about prehistory from midden evidence, they are going to have to rely on truly significant midden characteristics.
Study of the living shell fish population has shown the importance of this kind of comparative information. The conclusion that the pipi- cockle ratios in the midden layers are naturally derived casts doubt on the utilisation of this ratio for the chronological seriation of midden sites in New Zealand. This conclusion emphasises the point already made that it is necessary to rule out natural explanations before cultural ones are proposed. From the comparison of the reconstructed shell fish populations in Upper Layer B and Layer C with the structure of the present population, a number of inferences have been suggested about the length of occupation represented by each layer.
In this study no economic interpretation of the midden has been given. This topic, which contributes much to an understanding of the past activities at Galatea Bay, is covered separately in a paper by W. Shawcross.*
Artefacts
The number of artefacts found in the excavation is quite small: an unfortunate fact which seems to be quite characteristic of the later prehistoric beach middens in New Zealand.
Upper Layer B
Square A-l: mid-section fragment of an adze (Fig. 19, No. c). A mid-section fragment of a quadrangular adze was found at the bottom of Upper Layer B and resting on the surface of the underlying beach sand of Layer D. The fragment, made of a light brown greywacke possibly from the Waiheke-Ponui area, is 4.2-5.6 cm long, 3.3 cm thick, s.lcm wide on the larger face and approximately 4.ocm wide on the lesser face. The type cannot be determined. The specimen shows signs of fire-cracking and probably had been used as a cooking stone.
Square B-2: pumice block (Fig. 19, No. f). An irregular block of pumice 10.8 cm long, 5.7 cm wide and 3.ocm thick at the mid-section was discovered in the shell midden layer. There are no indisputable signs of artificial shaping.
Square C-2: shell fish-hook point (Fig. 19, No. d). A small pointed arch of shell approximately 2.ocm long from point to point was found. This specimen has been shaped from a piece of the outer lip of the smaller species of Struthiolaria, S. vermis. In type it is identical to two intact specimens found by F. G. Fairfield near the Manukau Heads (Fairfield, 1933: 151-153). They resemble Polynesian ruvettus hooks (Anell, 1955: 228-237). The shell points are larger than the Ponui specimen and are made from S. papulosa shells. The similarity with Fairfield’s specimens is so great that except for size, the Ponui hook would be indistinguishable from one of them (Auckland War Museum No. 31706, Fairfield’s Figure 15, and present Figure 19, No. e). The type of fish caught with these hooks is unknown. Square D-2: small adze fragment (Fig. 19, No. b). A small fragment made of dark greywacke with two polished surfaces which meet at a right-angle: probably from the edge of an adze. Layer C
Square B-2: adze (Fig. 19, No. a). An adze of dark greywacke, possibly of local Waiheke-Ponui origin, was found in the cluster of cooking stones in Layer G. The specimen is a wholly unusual type, made by flaking, pecking and polishing. It is two-ended. The cutting edge at one extremity has been formed by the removal of a large flake and by subsequent polishing. The cutting edge at the other extremity has been shaped by retouching with some polishing. While the adze has a crude quadrangular cross-section, it cannot be classified as belonging to one of the common New Zealand adze types (Duff, 1959). It is 11.2 cm long, s.Bcm wide and 4.ocm thick at the mid-section. The most striking characteristic of this very small artefact assemblage is the virtual absence of diagnostic types. Only the small fish-hook point can be related to similar specimens found elsewhere, and this is the first time this type is known
to have been found in an excavated context. Moreover, as a type, it is rare in New Zealand with a distribution limited to the area between Auckland and the North Cape.* Fairfield (1933: 153) has dated the Manukau examples to an unspecified “pre-European era”. Anell (1955: 235), on the other hand, has expressed the opinion they are not very old, and could have been derived from “ isolated loans from tropical Polynesia at a rather late stage ”.
Possible Shell Artefacts
No flakes or worked artefacts of obsidian were found in the excavation. This absence is unusual for a New Zealand site. As a result, it is impossible to draw any of the inferences about trade and the age of the site which this kind of evidence often allows one to make (Green, 1964). The total lack of obsidian tools suggests either cutting tools were not required in the preparation of food stuffs at the site, or, which seems more likely, some other cutting material was used. During midden analysis, twelve possibly utilised pipi shells were identified (Fig. 19, Nos. g-p), All but one came from the samples from Upper Layer B. These may have been the functional alternatives to the more common obsidian tools.
The use by the Maori of shells for cutting instruments is historically well attested. The first recorded archaeological discovery of possibly utilised shells in a New Zealand site was at Paterangi pa south of Thames on the Hauraki Plains (Shawcross and Terrell, 1966, and their figure. As in the Pater angi examples, the most suggestive signs of use on the Ponui specimens are shallow notches on the ventral margins. Some of the margins also show smoothing and straightening, but as previously stated for the Pater angi specimens, these additional indications of use are even more difficult to distinguish from natural abrasion in the midden deposit.
The existence of utilised shell tools in New Zealand archaeological sites has been proposed so recently that at this stage in research all that can be concluded is that the pipi shells could have been used as cutting tools. The necessary experimental study with fresh shells remains to be done which should establish empirical criteria by which wear by cutting can be distinguished clearly from wear by accidental abrasion.
Interpretation : Historical Events and General Statements
When the results of the excavation are brought together to describe the history of the site, the sequence of events is found to be far more complex than casual observation, unaided by excavation, might suppose. Most notably, two occupations were discovered and each was in several ways unlike the other. During the first occupation (Layer C) of the raised beach front, which is itself probably the remnant of an earlier strand line, fires were lit on the sand in one restricted area of the site. Behind this hearth area, small pits were dug into the sand. If the interpretation of these basins is correct, these unusual features were hangi pits of an uncommon type. Instead of fires being lit in them, the cooking stones were heated in the fires of the hearth area, and only then brought to the pits to cook the food placed in the hangi with them. The discovery of this practice of cooking in such umu konao is a totally unexpected result of the excavation, and
is the first time this type of hangi has been described in an archaeological context. No distinct midden dump was found near the cooking area. If a dump at one time existed, it seems not to have survived. Because of the difficulty in tracing the post holes, it cannot be said whether any structures were associated with this layer.
The date of this first use of the site is not known. As yet, no radiocarbon dates are available. The only artefact found in Layer C is undiagnostic. It does seem, however, the occupation was of a shorter duration than that in Layer B, although the range of activities, fishing, shell fish collecting and cooking, is the same. This inference is derived from the general impression given by the smaller area of the site used and the lesser number of pit features. In addition, because the reconstructed shell fish population curve shows dominance by a single size class, it can be inferred on somewhat more substantial grounds, in fact, that the occupation took place in only one year.
After the Layer C occupation, the old hearth area became covered with sand and the pits filled in. How long the site was abandoned is unknown. Eventually, however, the terrace of the raised beach front was reoccupied. This time the cooking pits were of the traditional form usually described for the Maori.
Layer B marks what must have been a major occupation of the bay lasting at least several years according to the inference based on the comparison of the archaeological and living shell fish populations. During the first phase (Lower Layer B) at least most of the site was used as a cooking area. Cooking sheds were probably constructed around the hangi. An associated midden dump may have existed in front of the cooking area which has been subsequently eroded away by the sea. At a later date (Upper Layer B), the cooking area must have shifted, and the old location was turned into a dump. During this later phase when the concentrated shell midden was deposited, a few hangi were constructed in the old area, but even these were later covered with midden.
The inferred general statements which emerge from the excavation are several; (a) Cooking areas at these times in Maori settlement were separate from midden areas. Excavation of a midden may miss the associated cooking site; (b) The later prehistoric Maori merely gathered or dug shell fish from beach collecting zones without selection either for size or species; (c) The range of activities attested in the excavation for this kind of site is quite restricted and specialised. Only evidence for fishing, shell fish collecting and cooking was found. Manufacture of durable artefacts such as adzes and fish-hooks did not take place on the site. Except for the adze chip, the two adzes found had both been used for an entirely different function than the one they originally had been intended for: they were used finally as cooking stones;
(d) This restricted ratine in the archaeological evidence indicates the site represents only one aspect of occupation at the bay. There must have been at least associated areas where sleeping houses stood because, unless the post holes found should be interpreted as indicating houses instead of cooking sheds, which seems improbable, there were no traces of houses. The inference is that these existed outside the area excavated. This seems to be the case whether they were rough shelters erected by travellers who used the site only temporarily, or more substantial living houses used at least seasonally; and (e) With the single exception of one fish-hook, the small artefactual assemblage seems so undiagnostic as to be unsuitable for use in relating this site with any other site in terms of them.
Evaluation
Because of the excavation at Galatea Bay, it is possible to offer an evaluation of the importance of this kind of site in the detailing of New Zealand prehistory. It can be seen that beyond interpreting the site as a specialised cooking and later dumping area, little more can be said here without additional evidence about what other kinds of occupation areas may be associated with it. As the information stands, the site conforms well to what one might expect a cooking area to be like, either in terms of the pa-kainga concept of the Maori settlement pattern, or the pa-hamlet pattern. What the excavation has been able to accomplish is an archaeological description of one cooking site and an indication of the range of evidence which such a site can offer. Ideally, what now must be done is to relate this area to any associated occupation areas, and specifically, to some complex of house structures suggesting either a small hamlet or a true kainga.
From an examination of the locality, the presence of a true kainga complex of numerous houses and a marae seems unlikely. It is more probable that farther up the small valley, for example, traces of former huts may exist indicating that the cooking area was part of a small hamlet of the kind described by Groube (see above, page 38) . There were no surface indications of such huts, but one would not really expect there to be any. On the other hand, the possibility cannot be ruled out that the midden was associated with the archaeological features on the ridge to the south of the site, or even to a fortified pa on the peninsula outcrop.
The major difficulty in establishing which, if any, of these possibilities was the actual case has been brought out by the excavation. Little evidence was obtained upon which one could draw the necessary connection between the midden area and a cluster of huts, or other features, in another part of the locality. Short of a direct stratigraphic link or close spatial propinquity, no artefacts were found which could be used to equate the midden occupations with occupations elsewhere on the basis of close assemblage identity. Dating by radiocarbon would be of little help, because even identical Carbon 14 range dates alone will not assure that separate sites were used at the same time by the same people.
Thus, the conclusion is that it will be difficult, if not impossible, to relate the excavated site to other sites in the locality. Yet, middens are among the most easily identified sites in New Zealand. If middens and other specialised activity areas cannot be related empirically to each other except by direct stratigraphic correlation, it will be extremely difficult to determine the characteristics of settlement in different parts of New Zealand at different times. With reference to middens, the problem lies not only in finding associated occupation areas such as hut clusters, but also in proving the reality of the very association.
The lack of common diagnostic assemblages to help draw temporal and cultural connections between different sites constitutes a serious limitation on archaeology in New Zealand (Terrell, 1965). Because this limitation makes it so difficult to correlate occupation sequences at separate sites even in the same small locality, it may be that only by extensive excavation at a large number of similar sites will it be possible to demonstrate a connection between the various settlement elements at any one site by identifying a common pattern recurring at many of them. Yet, in spite of this restriction, during the present research it has been possible to derive a number of inferences about Maori culture and settlement solely from the internal, non-artefactual, evidence gained from the excavation of a single midden site. Although from the point of view of the traditional archaeologist, who is most often
concerned with the classification and comparison of artefacts, the Galatea Bay site is discouraging for very real reasons, from the point of view of an archaeologist who is interested in reconstructing prehistoric cultures in all their details, the site has proven to be fruitful, especially in respect to prehistoric economics, as described in a separate paper.*
Acknowledgments
Without the assistance and encouragement of Mr W. Shawcross and Mr K. Peters of the Department of Anthropology, University of Auckland, this research would not have been possible. It is only because of the kindness and help of Mr G. Chamberlin, Mr A. Logan and Miss O. Chamberlin of Ponui Island, that the actual excavation was successful. Mr and Mrs E. Winstone, Mr B. Winstone and Mr K. Winstone made possible the initial recording of the site and greatly assisted in making the arrangements for the excavation. In addition, the author wishes to thank the following individuals for their technical assistance; University of Auckland—Dr P. Bergquist and Mr A. Michael (Zoology), Miss E. Crosby and Dr R. Green (Anthropology), Mr J. Blackford (Mathematics), Mr T. Wilson (Geochemistry), Mr L. Wright (Geography), Mrs K. Shawcross (History); University of Otago—Mr L. Groube (Anthropology); and the Auckland War Museum —Mr V. Fisher (Ethnology), and Dr A. Powell (Conchology).
Literature Cited
Ambrose, W., 1963. “Shell Dump Sampling.” N.Z. Archaeological Assoc. Newsletter, 6(3): 155-159. Anell, 8., 1955. Contribution to the History of Fishing in the Southern Seas. Studia Ethnographica Upsaliensia, IX, Uppsala. Ballance, P. F., 1965. “The Geology and Physiography of the Auckland District,” pp. 8-19 in L. O. Kermode (ed.). Science in Auckland. (Handbook for the 11th New Zealand Science Congress.) Auckland.
Batley, R. A. L., 1961. “ Otaihape Field Group.” N.Z. Archaeological Assoc. Newsletter , 4(4): 3-4. Best,, E., 1924. The Maori (Volume 1). (Mem. Polynesian Soc., Volume V.) Wellington. Harry H. Tombs Ltd. and Tombs Ltd. — 1952. The Maori as He Was. Wellington: R. E. Owen. Buck, P., 1962. The Coming of the Maori. Wellington: Maori Purposes Fund Board and Whitcombe and Tombs Ltd.
Buist, A. G., 1962. “ Excavation of a House-Floor at Waimate Pa.” N.Z. Archaeological Assoc. Newsletter, 5(3); 184-187. Davidson, J. M., 1964 a. “ Concentrated Shell Middens.” N.Z. Archaeological Assoc. Newsletter, 7(2): 70-78. —1964 b. The Physical Analysis of Refuse in New Zealand Archaeological Sites. (Unpublished M.A. Thesis, University of Auckland.) Auckland. Newsletter, 7(4): 152-163. Dawson, E. W., 1949. “ Excavation of a Maori Burial, at Long Beach, Otago; With Notes on Associated Artifacts.” Journ. Polynesian Soc. 58(2): 58-63.
Dawson, E. W., and Yaldwyn, J. C., 1952. “ Excavations of Maori Burials at Long Beach, Otago [Part II].” Journ. Polynesian Soc. 61(2); 283-291. Duff, R. S., 1956. The Moa-Hunter Period of Maori Culture. Wellington: R. E. Owen. W. R. Geddes (eds.). Anthropology in the South Seas. New Plymouth, N.Z.: Thomas Avery and Sons Ltd.
d'Urville, D., 1950. New Zealand 1826-1827. (English translation by Olive Wright.) Wellington; Wingfield Press. Fairfield, F. G., 1933. “ Maori Fish-hooks from Manukau Heads, Auckland.” Journ. Polynesian Soc. 42(167): 145-155. Firth, R., 1959. Economics of the New Zealand Maori. Wellington: R. E. Owen. Fomison, T., 1963. “Excavations at South Bay Kaikoura —Site S 49/43.” N.Z. Archaeological Assoc. Newsletter, 6(2): 100-102. Golson, J., 1957. “Field Archaeology in New Zealand.” Journ. Polynesian Soc. 66(1); 64-109.
and W. R. Geddes (eds.), Anthropology in the South Seas. New Plymouth, N.Z.: Thomas Avery and Sons Ltd. Green, R. G., 1963 a. A Review of the Prehistoric Sequence in the Auckland Province. (Auckland Archaeological Society publication No. 1, and N.Z. Archaeological Assoc. Monograph No. 2.) Auckland. Review, 11(3): 143-156. logical Assoc. Newsletter, 7(3): 134-143.
Green,, R. G., and Pullar, W. A., 1960. “ Excavations at Orongo Bay, Gisborne.” Journ. Polynesian Soc. 69(4): 332-353. Green, R. G., and Shawcross, F. W., 1962. “ Cultural Sequence of the Auckland Province.” N.Z. Archaeological Assoc. Newsletter, 5(4): 210—220. Grouse, L. M., 1964 a. Archaeology in the Bay of Islands. Dunedin: University of Otago, Department of Anthropology. University of Auckland.) Auckland. Places Trust Newsletter, No. 9: 5-7.
Hosking, T., 1962. “Report on Excavation of Whakamoenga Cave, Lake Taupo.” N.Z. Archaeological Assoc. Newsletter, 5(1): 22—30. Hunt, C. G., 1962. “A Fishing Camp Site Near Raglan.” N.Z. Archaeological Assoc. Newsletter, 5(1): 32-34. Nicholls, M., 1963. “ Preliminary Report of Excavations on Ponui Island.” N.Z. Archaeological Assoc. Newsletter, 6(1): 19-23. Parker, R. H., 1962. “Aspect and Phase on Skipper’s Ridge (Opito) and Kumara-Kaiamo (Urenui).” N.Z. Archaeological Assoc. Newsletter, 5(4): 222—232. Piddington, R., 1963. An Introduction to Social Anthropology (Volume 1). Edinburgh: Oliver and Boyd.
Pohlen, I. J., 1965. “Soils of the Auckland District,” pp. 28-30 in L. O. Kermode (ed.), Science in Auckland. (Handbook for the 11th New Zealand Science Congress.) Auckland. Powell, A. W. 8., 1961. Shells of New Zealand. Wellington: Whitcombe and Tombs Ltd. Shawcross, F. W., 1964. “Archaeological Investigations at Ongari Point, Katikati, Bay of Plenty.” N.Z. Archaeological Assoc. Newsletter, 7(2): 79-98. Shawcross, F. W., and Terrell, J. E., 1967. “ Paterangi and Oruarangi Swamp Pas.” Journ. Polynesian Soc. (in press).
Skinner, H. D., 1953. “ Stratification at Long Beach, Otago.” Journ. Polynesian Soc. 62(4): 400-402. Smart, G. D., 1962 a. “ Midden Recording and Sampling in the Waikanae Region.” N.Z. Archaeological Assoc. Newsletter, 5(3); 160-169. N.Z. Archaeological Assoc. Newsletter, 5(3): 170-184. Spring-Rice, W., 1963. “ Harataonga—Gt. Barrier Island.” N.Z. Archaeological Assoc. Newsletter, 6(1): 25—27.
Terrell, J. E., 1965. “ Limitations on Archaeology in New Zealand.” N.Z. Archaeological Assoc. Newsletter, 8(4). 125-130. Trower, D., 1962. “ Opito Beach: Two Sites.” N.Z. Archaeological Assoc. Newsletter, 5(1): 43-46. Vayda, A. P., 1960. Maori Warfare. (Polynesian Society Maori Monographs, No. 2.) Wellington. Wright, K., and Bennett, 8., 1964. “ Excavations at Smugglers Cave.” N.Z. Archaeological Assoc. Newsletter, 7(3): 133.
John Terrell, Department of Anthropology, Harvard University, Cambridge, Massachusetts.
* By “ late prehistoric sites ” all that is implied, as a matter of convenience, are sites not clearly associated with moa remains or “Archaic ” types of artefacts.
* In designating the squares, the north-west corner of the grid was selected as the datum reference point (DRP on Fig. 3). On all the plans of the excavation, this point has been placed in the upper right-hand corner. On the site map (Fig. 3), however, this point is in the lower left-hand corner. Each section was recorded as located along a line east or south of the datum reference point at a distance measured in centimetres from that point, and at a measured height above an imaginary plane optically projected from the datum zero point (DZP on Fig. 3) located to the south of the site (Fig. 7).
* Note: For convenience, the percentages in Figure 13 were calculated on the basis of the original test sample weight, but the samples were dehydrated to bring them into agreement with the three original IOOOgm analyses. Therefore, failure to reach 100% can be explained by weighing error and loss due to sifting and to drying.
* Indicates samples used in the original composition study (Fig. 10).
* For his assistance in the statistical handling of these data, the author wishes to thank Mr J. Blackford, a student in mathematics at the University of Auckland.
* The author wishes to thank Mr A. Michael for generously making available the results of his unpublished research on shell fish populations in New Zealand.
* In the graphs, the frequencies for specimens under 1.5-I.9cm are under-estimated because the smaller specimens, including ones between 0.1-o.9cm, were removed for separate study by Mr A. Michael; they represent the current spat of young bivalve molluscs.
* This ratio, based only on weight, is not strictly comparable to the ratio in Layer B because the dominant size group for pipi shell fish in Layer G is that of specimens only 3.5-3.9 cm long. The ratio, therefore, probably over-estimates the weight of cockles in Layer C. A more accurate ratio would have to be based on the number of individuals of each species present.
* The ratio for the 1.5-I.9cm class is to be interpreted as a ratio of more than 0.50. The ratio for the 6.0-6.4 cm class is based on only one specimen.
* Shawcross makes an analysis of the vertebrate remains, which shows the presence of 108 snapper in the excavated area; these would have supplied about 136 kilos of meat and it is calculated that the same deposit represented 5224 kilos of shell fish meat. The fish bones appeared also to indicate that the site was occupied only seasonally and, on this evidence and taking into account errors due to cumulative variations of values it is concluded that the site might have been occupied seasonally by a small community over a total of some six years.
* Information given by Miss E. Crosby which is based on her extensive study of Maori fishing and fishing gear.
* “An Investigation of Prehistoric Diet and Economy on a Coastal Site at Galatea Bay, New Zealand,” by Wilfred Shawcross, to be published in the Proceedings of the Prehistoric Society, Volume 33.
Mt-* »Nm ‘Sr Sh 5' A SH B‘ B* *2. ■5' n cockle in other 5’ CO O S ■i r • cs T3 *5] 5‘ o o o E <T> in other in stone 5" Sis' A Sh s' 5' ’B. ■5' Jin cockle Jin other 5’ o 13 fD *+-■ i-«. S ’2. *5’ Jin cockle Jin other Jin stone SQUARE A-2 (sample 2) 2) 500gm % % % % % % % % % % % 12.6 11.2 48.8 5.0 4.8 0.4 f 3.0 2.0 0.4 1.5 lOOOgm* 11.2 13.0 .5 f 6.0 1.8f 0.5 5.0 3.3 2.1 0.4 1.0 2000gm 10.8 11.8 49.9 5.3 3.0 0.4 2.9 3.3 2.1 0.4 1.9 Range: 10.8-12.6 11.2-13.0 48.8-51.5 0 8-4.8 0.4-0.5 0 3. 3 2.0-2. 1 0.4-0. 4 9 SQUARE (sample 4) 500gm 10.4 5.8 31.6 15.0 1.5 t 16. 8f 6.8f 2.6 1.6 1.0 lOOOgm 11.0 6.4 31.6 14.4 2.0 1.0 11.6 9.0 2.7 1.9 1.5 lOOOgm* 10.4 6.0 29.5 17. 3.0 1.6 9.5 7.2 4.0 2.4 0.7 2000gm 10.9 5.7 27. 19. 8f 14.4—19.8 1.5 2.5 11.3 8.7 2.7 1.9 1.0 Range: 10.4-11.0 7-6.4 27.1-31.6 5-3.0 0.0-2. 9.5-16.8 8-9.0 6. 6-4.0 1.6-2 .4 0.7-1. 5 SQUARE E-2 (sample 10) 500gm 10.0 5.8 35. 29.5 0.2 l.Of 0.2f 8.0 1.8 2.0 0.1 lOOOgm 8.2 4.5 22. 7f 36. 1.4 7.0 2.6 7.2 1.4 2.3 0.5 lOOOgm* 9.5 5.5 25. Of 33. 2.5f 5.8 1.0 7.5 1.5 3.1 0.5 2000gm 8.9 5.4 30.3f 30.3 1.0 3.5f 1. 2.8 8.3 1.1 2.7 0.5 Range: 8.2-10.0 8 22.7-35.0 29.5-36.0 0.2-2.5 8 2-8.3 1.1-1. 8 2.0-3. 1 0.
Fig. 13.—The Effect of the Size of the Sample. ϯ over 2% disagreement with the other values in the same class or missing.
Name Location Occurrence 1. Amphidesma australe Amphidesma australe mud and sandy flats mud and sandy flats common pipi common pipi 2. Chione stutchburyi tidal mud-flats common cockle 3. Perna canaliculus rocky ground mussel, common in north 4. Crassostrea glomerata rocky shores Auckland rock oyster 5. Cominella adspersa rocks and mud-flats very common 6. Cominella glandiformis inter-tidal mud-flats very common 7. Neothais scalaris rocky ground common 8. Pecten novaezelandiae Pecten novaezelandiae mud-flats and deep water mud-flats and deep water common scallop common scallop 9. Lunella smaragda inter-tidal rocks common cat’s-eye periwinkle 10. Maoricrypta costata mussel shells and rocks at low tide common 11. Maoricrypta monoxyla Maoricrypta monoxyla on shells or inside dead ones on shells or inside dead ones common common 12. Zediloma subrostrata mud-flats common 13. Zeacumantus subcarinatus inter-tidal mud-flats and rock-pools common
Appendix I—Key to Shellfish in the Midden at N-43/33 (after Powell, 1961)
Sample weight pipi cockle other and iin Ain I.t. Ain and charcoal 4in •iVin l.t. f&in No. 1 lOOOgm i 203gm i 16gm i C. 17gm i stone 8gm 160gm 132gm 270gm i 78gm i 16gm Neothais scalaris 2gm i stone 8gm i C. glomerata P. canaliculus 4gm 4gm charcoal fish bone 2 bits 1 fragment 4 78gm 4 16gm Neothais scalaris 2gm 4 stone 8gm 4 C. glomerata P. canaliculus 4gm 4gm charcoal fish bone 2 bits 1 fragment No. 2 lOOOgm i 60gm i 17.5gm i canaliculus 2gm i stone 50gm 112gm 130gm 515gm 4 32.5gm 4 21gm i L. smaragda canaliculus 3gm 4gm 4 bone stone bone charcoal fragment 1 lOgm fragment 4 No. 3 lOOOgm i 230gm i llgm i P. Igm i stone 28gm 90gm 57gm 270gm 4 56gm 4 lOgm C. glomerata 188gm fish bone 1 unid. piece i glomerata C. 6gm i stone 5gm canaliculus P. C. glomerata 188gm 6gm fish bone 1 spine, 1 scale, 1 verte,, 1 frag, less than 0.5gm charcoal 6gm 4 fish bone stone 1 unid. piece 5gm C. glomerata 6gm fish bone 1 spine, 1 scale, 1 verte., 1 frag, less than 0.5gm charcoal No. 4 lOOOgm i 178gm i 30gm i P. 6gm i stone 93gm 104gm 60gm 295gm 4 72gm 4 40gm i C. glomerata P. C. glomerata scalaris lOgm 21gm 2gm Igm i bone stone fish bone charcoal 2gm 5gm 2gm 0.5gm No. 5 lOOOgm i 306gm i lOgm i none i stone 40gm 90gm 52gm 320gm i 80gm i 13gm i P. C. glomerata scalaris 3gm Igm Igm 4 stone bone charcoal 12gm 1 Igm No. 6 OOOgm 1 i 275gm i 14gm i P. canaliculus lOgm i stone 85 gm 95gm 80gm 250gm 4 55gm 4 20gm C. glomerata 4gm fish bone 2 fragments 4 glomerata C. canaliculus 20gm 4 stone 5gm canaliculus C. fragment fish bone 6 fragments charcoal 2gm 4gm 20gm i fish bone stone 2 fragments 5gm C. glomerata 1 fragment fish bone 6 fragments charcoal 2gm No. 7 lOOOgm i 195gm i 20gm i canaliculus P. 2gm i charcoal bit 115gm 75gm 325gm 4 85gm i 27.5gm C. 4gm stone 75gm shell 1 fragment i stone 7gm i M. monoxyla five fish 5 fragments canaliculus 15gm charcoal 2gm C. glomerata 2.5gm C. glandiformis 1 specimen No. 8 lOOOgm i 305gm i 50gm i canaliculus 2gm i stone 95gm 90gm 65gm 190gm 4 53gm i 16gm C. 87.5gm i stone 6gm i canaliculus 4gm charcoal Igm 7gm charcoal Igm C. 7gm
Sample weight pipi cockle shell bonestone, and 4in •&in l.t. iVin charcoal iin tVin l.t. Ain 9 1500gm 4 405gm 4 lOgm 4 C. glomerata 15gm 4 stone lOOgm lOOgm 65gm 615gm 4 40gm 4 9gm P. canaliculus C. glandiformis M. costata 5gm Igm 2gm 4 fish bone stone fish scales 1 fragment 45gm 6 M. monoxyla Igm charcoal 2.5gm 4 P. canaliculus C. glomerata C. glandiformis M. monoxyla M. costata Neothais scalaris L. smaragda 25gm 2.5gm Igm Igm 2gm Igm 1.5gm 9 10No. 1500gmlOOOgm i 4 405gm350gm i 4 lOgm25gm i 4 glomerata C. canaliculus P. 15gm15gm i 4 stonestone lOOgmlOgm lOOgm95gm 65gm55gm 615gm250gm i 4 40gm75gm i 4 9gm15gm canaliculus P. glomerata C. glandiformis C. M. monoxyla 35gm 0.5gm 4 stone fish bone 2gm Igm M. C. adspersa 5gm8gm charcoal 2gm Igm 2gm i fish bone stone fish scales 1 fragment 45gm 6 M. monoxyla Igm charcoal 2.5gm i 4 canaliculus P. canaliculus C. glomerata C. glomerata glandiformis C. M. monoxyla 25gm 5gm Igm M. monoxyla M. costata Neothais scalaris L. smaragda 25gm 2.5gm Igm Igm 2gm Igm 1.5gm No.No. lOOOgm 4 385gm 4 lOgm 4 P. canaliculus 5gm 4 stone 25gm 45gm 25gm 365gm 4 45gm 4 5gm 4 C. glomerata M. monoxyla P. canaliculus C. glomerata M. monoxyla 7.5gm Igm lOgm 2gm Igm 4 stone charcoal 5gm 3 bits 10 12 lOOOgmlOOOgm i 4 350gm320gm i 4 25gm20gm i 4 canaliculus canaliculus 15gm3gm i 4 stonestone lOgm45gm 95gm95gm 55gm50gm 250gm300gm i 4 75gm80gm i 4 15gmlOgm C. glomerata M. monoxyla 35gm 0.5gm i stone fish bone 2gm Igm C. adspersa 8gm charcoal 2gm i 4 canaliculus glomerata C. glomerata C. P. M. monoxyla C. glomerata 25gm 5gm Igm M. monoxyla 5gm 5gm 2gm Igm 4 charcoal stone fish bone charcoal 1 lump lOgm 1 verte. & 1 frag. & 2 fish scales 2gm No. 13 2000gm 4 105gm 4 65gm 4 P. canaliculus 30gm 4 stone 115gm 260gm 215gm 800gm 4 85gm 4 lOOgm C. glomerata 5gm charcoal 2.5gm P. novaezelandiae 2.5gm 4 stone 50gm L. smaragda 3gm fish bone 8 fragments 4 P. canaliculus 25gm charcoal lOgm C. glomerata Igm L. smaragda Igm Cominella adspersa 0.5gm Z. subcarinatus 0.5gm Z. subrostrata 0.5gm 4 stone 32.5gm 280gm 250gm 945gm charcoal 2.5gm 4 stone 35gm 11No. 1 OOOgm2000gm i 4 385gm95gm i 4 lOgm57gm i 4 canaliculus P. canaliculus P. 5gm15gm fish bone 1 fragment i stone 25gm 45gm 25gm 365gm i 4 45gm92.5gm i 4 5gm90gm i 4 C. glomerata canaliculus P. 30gm charcoal 5gm M. monoxyla C. adspersa Igm P. canaliculus C. glomerata M. monoxyla 7.5gm Igm lOgm 2gm Igm i stone charcoal 5gm 3 bits No. 12 lOOOgm i 320gm i 20gm i P. canaliculus 3gm i stone 45gm 95gm 50gm 300gm i 80gm i lOgm i C. glomerata P. canaliculus 5gm 5gm i charcoal stone 1 lump lOgm C. glomerata 2gm fish bone 1 verte. & 1 frag. M. monoxyla Igm charcoal & 2 fish scales 2gm No. 13 2000gm i 105gm i 65gm i P. canaliculus 30gm i stone 115gm 260gm 215gm 800gm i 85gm i lOOgm C. glomerata 5gm charcoal 2.5gm P. novaezelandiae 2.5gm i stone 50gm L. smaragda 3gm fish bone 8 fragments i P. canaliculus 25gm charcoal lOgm C. glomerata Igm L. smaragda Igm Cominella adspersa 0.5gm Z. subcarinatus 0.5gm Z. subrostrata 0.5gm i i stone charcoal stone 32.5gm 2.5gm 35gm 280gm 250gm 945gm No. 14 2000gm i 95gm i 57gm i P. canaliculus 15gm fish bone 1 fragment i 92.5gm i 90gm i P. canaliculus 30gm charcoal 5gm C. adspersa Igm
Appendix ll— Midden Composition (All weights under sgm are approximate)
Permanent link to this item
https://paperspast.natlib.govt.nz/periodicals/TRSGEN19670714.2.2
Bibliographic details
Transactions of the Royal Society of New Zealand : General, Volume 2, Issue 3, 14 July 1967, Page 31
Word Count
16,689Galatea Bay—The Excavation of a Beach-Stream Midden Site on Ponui Island in the Hauraki Gulf, New Zealand Transactions of the Royal Society of New Zealand : General, Volume 2, Issue 3, 14 July 1967, Page 31
Using This Item
In-Copyright Materials
In-copyright materials are made available under a Creative Commons Attribution 4.0 International licence. This means that you may copy, adapt and republish this material, as long as you attribute both the author and the Royal Society of New Zealand.
In-copyright taxonomic materials are made available under a Creative Commons Attribution No-Derivatives 4.0 International licence. This means that you may copy and republish this material, as long as you attribute both the author and the Royal Society of New Zealand.
For advice on reproduction of out-of-copyright material from this periodical, please refer to the Copyright guide.