Volume 61 1952 > Volume 61, No. 1 + 2 > Recent Maori occupation of Notornis Valley, Te Anau, by Roger Duff, p 90-119
                                                                                             Previous | Next   

- 90

SPECTACULAR as it was, Dr. G. B. Orbell's rediscovery in November 1948 of living specimens of Notornis hochstetteri in a sub-alpine valley in the Te Anau area, was followed in late 1949 by the discovery in the same valley of evidence that recent Maori visitors—possibly in the post-European period—had encountered and killed a single individual of Megalapteryx didinus, the small forest-dwelling moa.

The circumstances were these:—

In the course of field work in Notornis Valley in October 1949, Mr. Ken Miers, a wild-life field officer of the Department of Internal Affairs, was searching the floors of the limestone cliffs on the south side of the valley for evidence of their use by Notornis as a winter retreat. Protruding from the sandy limestone dust on one floor (shelter B, over) he recognized a feather of the kakapo (Stringops habrop-tilus). Probing in the loose dust of the surface revealed other feathers, including those of Notornis and kiwi. These were in a confined area suggesting a human concentration rather than a moulting scatter, but the final evidence was the presence of a layer of tussock flooring which had been charred. Mr. Miers then made a closer inspection of a similar shelter floor (shelter A, over) about 100 ft. higher along the line of bluffs. Here he found midden bird bones on the surface, including a scorched vertebra of Megalapteryx and, as well as similar feather deposits, surface signs of fire.

I cannot do better here than quote verbatim Mr. Miers' account of the discovery, as taken down by me during an interview with him on 10 March, 1950:—

Shelter B (30 October, 1949); K. Miers.

“The first shelter observed was B. I saw a kakapo feather protruding from the dust, pulled it out, and several - 91 other feathers came with it. I scratched with my knife and hands, and received a handful of feathers including kiwi and Notornis.

“The upper dust covering carried a heavy layer of kakapo feathers from 1 in. to 3 in. down. Below this I found a layer of burnt Danthonia tussock with feathers, including kiwi and Notornis, mixed. The main concentration was in a small area (about 2 ft. by 3 ft.), beyond this just an occasional feather. I looked for cooking sign but failed to find it. The only artifact found was a badly preserved flax noose which may not have survived.

Shelter A (2 November, 1949); K. Miers.

“Three days later in making a second trip to B (with Frank Woodrow) I found A. Surface ash and charred wood were the first indication [scattered round general area of moa skeleton, in plan, over]. No feathers were then found. Later in the same day I found a moa vertebra that had been burnt and bones of small birds, probably kiwi. These bones were in the loose surface dust and were bleached white. The moa vertebra was on the surface. The bird bones were mainly kiwi tibia bones, well splintered. As by this time I was familiar with the cave deposit of naturally deposited moa bones on the opposite side of the valley I did not realize at this stage that the moa bone came from a bird that had been killed.”

Shelter A (2 December, 1949); K. Miers.

Mr. Miers returned as a member of the December 1949 expedition, and guided the party to the shelters on 2nd December.

“I returned to shelter A taking a kitchen shovel and a bush slasher as tools, and immediately found tussock bedding with a profusion of feathers. I did not note the depth from the surface. With Mr. L. Gurr I then dug where the bones had been found. We first found a profusion of bird bones in a concentrated layer. They would be about 5 in. or 6 in. down, mixed with charcoal. We discovered more odd moa bones, and then struck the main line of vertebrae still joined together, alongside the undisturbed line of tracheal rings. I don't think the skull was attached, but I believe part of the skull was found during the day.

- 92

“The vertebral column commenced about 3 ft. in from the lip of the shelter floor, running from the pelvis end inward. The main concentration of bones would be grouped over an area about 4 ft. square [indicated on plan]. Feathers were found from a careful sifting of the bedding, and those which I learned from Dr. Falla were probably moa came from the area [marked] adjacent to the bones.

“I have no record of the sandal, fire stick, etc., but they were found in the bedding, toward the small pit. Between the small pit and the rock wall Jim Ollerenshaw found deep down a heap of round droppings which Dr. Falla believed could be moa.

“I would say the loose surface dust varied from 5 in. toward the lip, where the moa bones were found, to 1½ in. toward the wall.”


The December visit referred to in Mr. Miers' narrative was the expedition led by Dr. R. A. Falla, Director of the Dominion Museum. During this visit the surface dust of both shelters was thoroughly raked through, some excavation carried out in shelter A, and the great bulk of the material evidence which suggests the recency of the Maori occupation, secured. While the significance of the cuts on the ischial process of the pelvis had not then been realized, Dr. Falla was satisfied from the shallowness of the human deposit, from the preservation of the feathers, the tussock bedding, the flax textiles, and the wood, that the occupation was recent.

He concluded his diary entry of 2 December with the following: “It would be hard to estimate the period that has elapsed since it [shelter A] was used—possibly within the last 100 years.” Writing on 17 January 1950, before the possible significance of the cuts on the pelvis had been pointed out by Dr. H. D. Skinner, Dr. Falla adds: “I feel that the date should be estimated at about 1840 or 1850. Like the inhabitants of Taylor White's deserted village at the mouth of the Upokororo, these people would appear to have had access to metal implements. The fire-stick has every appearance of having been cut with iron or steel.”

- 93

Dr. Falla's full diary entry for 2 December reads as follows (he stresses that these are to be regarded as preliminary field notes):—

“Today's expedition was to the dry shelter caves (A & B) recently found by Ken Miers. The upper cave (A) had a foot or two of dust, very loose for the first 5 in. At this level was charcoal, broken bones of Megalapteryx, kiwi and other birds. There were also dry tussock, charred tussock, and numerous feathers in good condition (a few charred). They were of kiwi (A. australis and A. oweni), of kakapo (the most plentiful), of Notornis, weka and a few owl and parrakeet. The latter were in a nest containing tufts of fur which might have been bat, rat or stoat (sample taken). There were also a few larger feathers with aftershaft, and they have been taken for examination (?moa).

“There is no doubt that many birds lived in the shelter for, in addition to the chewed fibre bundles characteristic of kakapo, there were abundant droppings of kakapo, Notornis, ? moa. (These [last] were in a deep layer at the back of the shelter and may be human, but seem to be composed of fine vegetable fibre).

“Remains of the human hunters consisted of flax snares, a shaped peg, a well-worn flax sandal, and a lashed bundle of weka wing feathers (snare lure).

“In the lower cave (B) there was more char, more flax nooses, remains of another sandal. This cave, however, had larger masses of kakapo and Notornis feathers in a good state of preservation.

“This dust is extremely fine, completely dry to a depth of about 10 in., and uniformly cold. It would be hard to estimate the period that has elapsed since it was used—possibly within the last 100 years.”

These records by Mr. Miers and Dr. Falla establish the basic facts: The dry dust floors of the shelters had been used by a transient hunting party or parties; the midden remains and artifacts were found essentially in the shallow surface layer of loose dust, approximately 5 in. deep; the occupation was recent enough for the feathers, the textiles, the tussock bedding, and the wood to be well preserved; the midden skeleton of Megalapteryx was found in the floor of shelter A, with this bird's presumed feathers mixed with the tussock bedding.

- 94

As a consequence of Dr. Falla's report, Dr. H. D. Skinner and I made a follow-up visit to the valley in February 1950, in the hope of establishing the stratification of the shelter floors and in the course of a three-day examination of the shelters, made a careful examination of the floor section of shelter A. I am indebted to Dr. Skinner for this privilege of recording the results of our investigations, to Mr. Miers and Dr. Falla for the use of their field notes, and to Dr. Falla for permission to describe the important midden material and artifacts recovered by his expedition. To Mr. V. C. Browne, to expedition members J. Ollerenshaw and C. A. Fleming and to the National Publicity Studios thanks are expressed for permission to reproduce photographs.

The shelters, which we first visited on 2 February, and worked on throughout the three following days, were typical of those in South Canterbury, and elsewhere, in being created by the base erosion of limestone cliffs, the floors composed of limestone and sandstone dust kept perpetually dry by the immense overhang of the cliffs at this point. This perfect dryness is an essential element in any consideration of the age of the remains. In this matrix tough flax textiles, tussock, even feathers, should last as long as, or longer than, on museum shelves. As elsewhere the north facing bluff on the south side of the valley was preferred, and similar natural shelters in the south-facing cliff opposite have not been used.

They differ in their picturesque sub-alpine setting (shelter A is approximately 3,220 ft. above sea level) and in the beech forest covering of the limestone cap overhanging them and the slopes leading steeply down to the bare valley floor and lake. The general setting of the valley and the shelters may be gathered from the aerial panorama (plate 1) of the mountain hinterland of the west shore of Te Anau, with the shore of the south arm running off to the left. The 6,000 ft. tops of the Murchison Range rise in the background above a forest line of 3,500 ft. with watercourses such as the Point Burn (centre) and the Tunnel Burn (right) forming valley flats which at first run at a gentle slope. At a line equivalent to the lower limit of the bare mountain ridge to the left, they cut sharply down through an extensive downward-tilting limestone cap to plunge in

- i
V. C. Browne, photo., Aerial view of western shores of Te Anau showing Point Burn (centre) and Tunnel Burn (right). The dotted line indicates the probable Maori route into Notornis Valley.
- ii
View of valley looking down toward limestone bluffs (left and right) which herald the descent of the Tunnel Burn.
- iii
Limestone wall on south side of valley. Crosses show position of shelters (shelter A on right).
Kakapo droppings, near shelter A.
- iv
Closeup of shelter A. Position of Megalapteryx remains indicated by Gurr (kneeling, right). Area of tussock bedding between Falla (centre) and Miers (left).
- v
Detail, trench 2 (shelter A), taken through outer rim showing maximum depth and deposit of charcoal.
- vi
Shelter B, showing best preserved of rock drawings (retouched on print).
- 95

gorges to the lake shore. Such is the course of the Point Burn (centre) and the Tunnel Burn (right), but the latter, originating in a higher valley, has provided the more extensive tussock flat which supported the colony of Notornis which provided for us, and the Maori, the incentive to find a way in. In plate 2, the limestone bluffs left by the downward cutting course of Notornis Valley, may be seen at the lower end of the tussock flat heralding the plunge of the Tunnel Burn to the lake. The shelters occur toward the upper limit of the limestone cliff to the right of the plate; the higher south-facing bluffs to the left have produced nothing to date but a cave with an extensive accumulation of bones of Megalapteryx, apparently geologically old and deposited by underground water flows.

The setting of the shelters at the base of the limestone cap overlooking Notornis Valley from the south, makes it unlikely that the Maori would find their way in by the roundabout route followed by our, and other, scientific parties. This route (see plate 1) follows along the north ridge of the Tunnel Burn to a point about 2,500 ft. above lake level, where it descends sharply through the northern limestone cap, drops through forest to the lower valley near the tarn outlet, and thence climbs steeply up again through the opposite forest slope to the southern limestone. I believe the Maori would approach along the shield between the Point Burn and Tunnel Burn, and I have tentatively marked the route as outlined by Lawson Burrows who has found this line of approach practicable and marked by odd totara trees whose bark had been stripped off to provide containers for the birds (see route marked on plate 1). This route would take the Maori directly to the crown of the shelter bluffs, down which they could doubtless scramble, camp for the night, and spread their hunting cordon around the valley next day. It is by no means certain that the Notornis and weka were run down only in the flat below. From the heaps of undisturbed droppings it is certain that Notornis and kakapo sheltered here for long periods, and it is probable that the Notornis still retreat here when snow falls have driven them from the tussock. The other flightless quarry, kiwi and laughing owl could also be captured in the immediate neighbourhood. It is possible that the Megalapteryx was taken the same way. - 96 The 18th century fowling party, making the first (?) human penetration into the valley, found sheltering beneath the cliff the bird for which they possibly had no name, but killed and ate nevertheless. Until man and his introduced predators disturbed them we can regard the dry floors as moulting and dusting shelters used by Notornis, kakapo, kiwi and the occasional Megalapteryx moa. It is important to note in passing that Megalapteryx did not nest in the shelters; the absence of any fragment of eggshell in the dust sufficiently establishes this.

A closer view of the shelter bluffs is shown in plate 3, which looks south from the tussock flat to include the lower (east) end of the lake and its forested, steeply-climbing southern slopes. The arrow marks the approximate position of shelter A; the second shelter is a little lower and to the left. Closer views, plate 4, convey the immense overhang and height of the cliffs above them, the steepness of the slope below, and the close approach of the forest cover. The narrow but continuous zone of tussock immediately separating the cliff base from the forest was sufficient to provide warm bedding for the campers and presumably a standby diet for storm-sheltering or wintering birds. (Accumulations of undisturbed Notornis and kakapo droppings are found along the cliff base, the former probably including recent additions to the deposit. While there is no evidence of the continued presence of kakapo, their occupation is recent enough to have left accumulations such as the undisturbed, unposed deposit shown in plate 5.)


The area of the natural floor is small—approximately 13 yd. in length and, on a zone varying from 5 yd. to 3 yd. from the rock wall, bounded by a low natural rim of limestone rubble which the Maori had deepened by dumping spoil and rubble to clear a sleeping area. As proof of this, a cross section of the rim at trench 2 revealed that the charcoal layer, which in the dust floor proper (point B) was only 3 in. below the surface, was there almost 12 in. down. As in cutting our trench we worked in from the rim, this deep charcoal zone at first gave rise to hopes that it represented an older occupation, but the continuation of the trench demonstrated that it was continuous with the sub-surface - 97 charcoal deeper in. The charcoal layer was inconsiderable—just sufficient to mark the direction of the stratum, and to register on a photographic plate (plate 6). A very few fires would have accounted for it, and the few stones recovered suggested only one small earth oven. Outward the charcoal zone terminated below the crest of the rim, inward it ran out at point B which marked the limit of tussock bedding. At trench 1, where the floor was deeper, Dr. Skinner found no traces of charcoal. While a longitudinal section might reveal otherwise, all the evidence to hand suggests that the fire and cooking area was concentrated in the small zone coinciding with the widest part of trench 2. Here the major recovery of Megalapteryx bones was made, and the greatest concentration of the never numerous midden bones of smaller birds. The surface indications which led Dr. Falla's December party to select this point for excavation apparently represented the major, if not the only, midden area.

Working outward from the wall along the section of trench 2 we can now follow the history of the stratification:

The original base was provided by the bench-like continuation of the shelter wall. On this there gradually accumulated debris in the form of limestone blocks which fell from above, and limestone dust which consolidated. This is the sterile base stratum (layer D) which varies from a depth of 2 in. toward the wall to over 6 in. toward the rim. At all stages in its history this was capped with a loose dust which had not consolidated (layer A). This runs continuously toward and over the rim. While at our visit its depth varies from 1 in. toward the wall to 3 in. at the rim, we were dealing with a stratum which had been thoroughly raked over, and it is probable that from point B to the base of the rim it was originally 5 in. as estimated by Falla and Miers.

This simple original sequence—surface dust, consolidated dust, rock—is represented by section C. It was disturbed by the arrival of the Maori who spread tussock bedding over the surface from the wall to a point short of B. Between this point and the rim they lit fires and deposited midden bones, superimposing on A the brief and shallow charcoal stratum (now represented by layer C). Most of the midden bones, notably those of Megalapteryx, were deposited where they might be expected, immediately above - 98 the charcoal, and toward the rim. On the arrival of Miers in October 1949 the surface dust had enveloped them to a depth of only 5 in., but so imperfectly that one charred and bleached vertebra of Megalapteryx was still exposed. This sequence—surface dust with bones, charcoal, consolidated dust, and rock, is represented by the section at point B.

Most of the Megalapteryx bones found had been dropped in the concentrated area indicated by the circle of crosses in the plan, and by the group of crosses in the section. Despite our notions that every scrap of animal flesh would be required in that inhospitable setting the hunters followed the practice, revealed at Shag Point and other coastal camps where the large Euryapteryx moa was hunted on a vast scale, of severing the vertebral column and neck and casting them aside. Miers and Gurr found a number of the vertebrae in natural articulation, with the adjacent line of the tracheal rings (see line of crosses on plan). From this apparent waste the cranium was excepted, shattered to extract the brain. Some of the remains were flung toward the wall (the rib process, quadrate and cranial fragment indicated on the plan), and some toward the rim (the gizzard, rib and vertebra—see plan). In the section it will be seen that the latter lay in a new stratum, layer B.

The intrusion of layer B represents the third and final stage of the stratification of the shelter. While clearing the floor to strew the bedding and lay the fires the occupants heaped the rubble and surface dust toward the rim. This buried the charcoal layer to a maximum depth of 8 in., embedding midden bones in the process, and creating an exaggerated impresssion of the time lapse since the charcoal was laid down. In turn the surface dust blew over this to a depth of 3 in., tussock grew on it and held it. Point A of trench 2 represents this final stratification—surface dust, consolidated spoil, charcoal, consolidated dust, rock.

Whether the moa was skinned or plucked, some of its feathers were embedded in the adjacent tussock flooring (see area marked on plan). I follow Miers for the localized concentration, although Skinner records some from trench 1. These as well as similarly embedded feathers of Notornis, kakapo, kiwi and weka were sealed in by the drift of the surface dust, to be revealed by the December expedition with - 99 the well-preserved tussock, just beneath the surface. In addition the flax sandal, the wooden fire-stick, the weka snare noose and lure, and other artifacts were sealed in with the flooring. No record of the area covered by the flooring was taken at the time, but from the recollections of Miers and Ollerenshaw I place it as indicated on the plan. Skinner and I found faint traces of embedded tussock, with occasional feathers, extending beyond this zone across trenches 1 and 2, as indicated. The depth generally represented the lower margin of layer A. Midden bones were scarce at trench 1. They included one uncinate rib process of Megalapteryx, bleached white, but with visible traces of the dessicated flesh and tissue (as noted sporadically on the small bird bones). The parrakeet feathers appeared too bright in colour to have contemporary with the occupation, and I follow Falla that they came from a nest of the introduced stoat (Mustela). The only mammalian remains were a few bones of the Maori rat, and a patch of unidentified fur. The colours of the undoubtedly contemporary feathers were still visible though faded: they might best be described as shop-soiled.

With the exceptions that the surface dust was deeper, and the charcoal layer absent, Skinner records the same general sequence from trench 1 (see notes, which follow). In interpreting them, in conformity with the reconstruction of trench 2, I would suggest that Skinner's surface strata do not correspond wholly to the “surface dust” (layer A) of my sequence, but include “compacted dust” (layer B).

Section, at end of trench 1, adjacent to pit (corresponds to C, trench 2): Surface stratum, 9 in.: “Nine inches conveying several layers each representing a separate visit. Dust, tussock, beech leaves. Many droppings of ?kakapo, ?moa and Notornis. Many feathers.

Second stratum, 5 in.: “Coarse sand full of small lumps of limestone—nothing else.”

Third stratum, 4 in.: “Brownish layer apparently due to decomposed tussock. No human material.”

Base stratum, 9 in.: “Fine greenish-yellow sand. Apparently wind-blown.

Rock, at 2 ft. 6 in.

Section, mid-way toward rim trench 1 (corresponds to B, trench 2): Surface stratum (A), 8 in.: “... in confusion through raking but containing most of artifacts - 100 found, e.g., piece of scutched flax, two pads [of ribbonwood inner bark], butts of tussock with cut blades, butts of tussock with blades chewed (?Notornis), leaves of Celmisia, which does not grow in shelter, many feathers, principally kakapo.

Second stratum (B), 2 in.: “Compact dust. Contained bedding material of tussock and beech leaves, charcoal, kakapo feathers.”

Base stratum (C), 10 in.: “Sand with blocks of fallen limestone.”

Rock at 1 ft. 6 in.

Section near rim, trench 1 (corresponds A, trench 2): Surface stratum, 8 in.: “Ashy layer, but no charcoal. Bedding material. Feathers of parrakeet, kakapo, moa. Droppings of several undetermined species.”

Base stratum, 8 in.: “Yellow sand layer. No organic material, many limestone flakes.”

Before proceeding to describe the textiles and artifacts, the rock drawings discovered in shelter B, and the origin of the cuts on the ischial pelvic process of Megalapteryx, we should note the following conclusions which emerge from our investigations of the stratification:

The occupation was confined to one period of time, it was slight, its traces were confined to a shallow surface stratum. In shelter A the traces were found wholly in the surface dust, not lower than the embedded tussock which marked its downward limit toward the wall, or the charcoal zone toward the rim—at most a depth of 5 in. (Where the dumping of spoil at the rim had deepened it, the charcoal zone, as the lower limit of finds, was depressed to 12 in.). Furthermore, in this shelter all the charcoal traces and most midden sign were confined to an area a few feet square (in trench 2). The signs could well be accounted for by the single visit of a large hunting party, or at most by repeated visits of the one smaller party within a brief cycle of years.

In shelter B, with its much more extensive and deeper floor, no excavation was done by Skinner and myself, and only surface raking by Falla. However, surface indications suggest a generally similar history to that of A. Charcoal traces were even scarcer than in A, and midden bones fewer. It is important to note that neither the presumed Megalapteryx feathers, nor its bones, were found in shelter B.

- 101

The occupation of shelter B was further marked by wall drawings in charcoal, which in pigment, method of application and character were strongly reminiscent of those of South Canterbury and North Otago. We can imagine the compositions lightly traced on the uneven wall face with a charcoal stick picked from the fire. Unfortunately the subjects selected throw little light on the occupants of the shelters. One is a stylized human figure (line figure 3) whose outstretched elongated arms suggest that he is a composite conception of bird and man, and generally resemble the rendering of the bird “deity” at Frenchman's Gully, Pareora. A second is a group of three heart-shaped designs, linked by means of lines, as shown in plate 7. While at first glance these seem non-Maori, a closer examination makes it clear that each “heart” is made up by symmetrically pairing the S-shaped spirals of the Maori puhoro design. Of the three remaining designs there was only one I was able to recognize sufficiently to copy, and my interpretation of it (line figure 5) may well be hopelessly out. The adjacent double spiral was clearly recognizable, and clearly Maori. Being quite unable to recognize the two final designs I made no attempt to copy them.

The traces were so faint that the December party in the course of some hours' occupation of the shelter had failed to notice them.

Their chief significance is not so much the light they throw on the hunters of Notornis Valley as the link they establish with the drawings from other South Island shelters. They agree in form—notably the degree to which at first glance they appear non-Maori and on analysis reveal well established Maori motifs. They also agree in their state of preservation, the rapid scaling of the wall leaving their outlines as a series of discontinuous (black) dots contrasting feebly with the weathered grey limestone.

Assuming that the present theory of the recent occupation of Notornis Valley is sustained, the drawings might be attributed to an intrusive 18th century hapu of Ngaitahu or a surviving one of Ngati Mamoe. These are both modern Maori tribes in a culture-historical definition of the term Maori.

- 102

The theory that the shelters were occupied in the late 18th or early 19th century is supported by a combination of wholly circumstantial evidence. Without the evidence of the cuts on the bone, the state of preservation of feathers, textiles, fire plough and tussock could not eliminate the early 18th, or even the 17th centuries as possibilities. I would agree that the occupation must be confined to post-Fleet tribes, but had the shelters not been discovered until 2050 I believe that the feathers and textiles would have scarcely altered, provided the protective dust canopy remained in being.

Everything depends then on the interpretation of the cuts on the Megalapteryx pelvis. In the absence of an objective method of determining their origin we have nothing but subjective judgments to go on. Until experiments 1 on cooked butchers' bones have demonstrated what a stone edge cannot do I feel we should be restrained by caution from postulating the metal origin which the depth and sharpness of the cuts suggest.

Dr. Skinner's statement that they were produced with a metal implement should therefore be regarded with the respect one accords to the judgment of a connoisseur of long experience, but not as evidence. By contrast H. S. McCully of Timaru who has spent a lifetime studying (and making) Maori flake tools is equally positive that scarfs of that section could be produced by a stone flake of a commonly found type.

The pelvic process was forwarded to Dr. Skinner in January, 1950, by Dr. Falla who was concerned with the origin of the animal gnawing along its distal margin. “The chewing,” reported Dr. Skinner, “seems to me human rather than canine or rodent. But much more important are the marks of cutting along the lower edge obviously intended to cut the tendons attached along this margin. These are the work, not of a stone flake, but of a heavy steel knife. To me - 103 this constitutes convincing evidence that the species of bird to which the ischium belongs was living when European influence was felt among Maoris even in remote places.”

Scarfs generally resembling those across the neck of the Notornis Valley ischium appear on a stone age ischial process from Wairau, but there is none resembling the finer crest cuts. Nor from the hundreds of moa bones recovered by this Museum from ancient stone age sites along the east coast of the South Island can I recall comparable cuts. This is the broadest and strongest evidence in favour of Dr. Skinner's hypothesis and was doubtless the basis of his judgment, but it remains negative evidence.

If the Te Anau occupants carried flint knives for this purpose they were normally expendible and one would expect both complete flakes or fragments to be left behind. The scarcity to date (written January, 1952) of remains of stone tools from the shelters is due primarily to their brief and limited human occupation, but there is sufficient to establish that the hunters used ground stone adzes and imported flints. Dr. Falla found in Shelter A an undoubted fragment of a ground adze of grey baked argillite (? of Nelson-Marlborough origin). The traces of bruising on the surface make it certain that it was originally prepared as an adze blade. In a second visit in December, 1951, during which I dug most of the untouched floor of Shelter A as left by Skinner and myself in January, 1950, I recovered another fragment, presumably of the same adze, and a waste flake of imported flint doubtless struck from a larger knife. Assuming that the Maori occupation of the shelters was of the brief period of years indicated by the stratification of the floors, it will be readily conceded that the evidence of a stone culture during any part of it is of considerable significance. While the introduction of the first metal knives need not mean the immediate abandonment of stone adzes and knives, we still have no positive evidence of the presence of metal.

Writing before these last finds, of December, 1951, I considered that the evidence was not sufficient to justify acceptance or rejection of Skinner's metal tool hypothesis, so that estimates of time remained balanced between a late stone era (closing theoretically at Cooks' visit to Dusky - 104 Sound in 1773), and a pre-European settlement metal age 2 lasting into the eighteen forties. (By either reckoning the continued survival of Megalapteryx in the fastnesses of Fiordland was not a fantastic possibility).

I feel now that the scales are tipped against the metal hypothesis, with all that this implies regarding the possible continued survival of the Megalapteryx moa. As stone age occupation of the shelters allows a greatly extended time scale for the capture of the moa whose remains were found in Shelter A, its backward limit being quite indefinite, its upward limit not necessarily defined by Cook's sojourn in Dusky Sound, but capable of extension into the early eighteen hundreds.

On the western lakeshore, about 400 yards south of the Tunnel Burn, there is a second growth forest-clearing on which Lawson Burrows noted traces of a long ditch and low earth rampart. This would be even more conveniently sited as headquarters for the expedition, and no less as a secluded stronghold of the retreating Ngati Mamoe from which they were driven into Notornis Valley and the wild hinterland of the Murchison Range. Mr. Burrows reports Dr. Holloway's estimate of the age of this second growth forest as 150 years, which would accord with either a stone era or metal age hypothesis.

This, and the general question of Maori knowledge of the habitat of Notornis, is a little outside the scope of this paper and might provide an interesting exercise for local students. In the meantime I would note that the apparently definitive references to Notornis in Beattie's The Maoris and Fiordland (1949) were published after Dr. G. B. Orbell's discovery of the colony. In this book Beattie applied the previously recorded, but unlocalized placename, “Kohaka - 105 takahea" [nest or nesting-place of the Notornis] to Notornis Valley because it fitted the description, thus setting up problems with “Wai o pani” which had pre-emptive rights!

The following passage from Mr. Beattie's letter to me of 20 February 1950 should make clear the peculiar difficulties of this method of allocating names:

“The allocation of traditional placenames is extremely difficult and Hall-Jones, Lawson Burrows and I all differ as to the locality of Wai o pani, my surmise being recorded on p. 53 of The Maoris and Fiordland. The same difficulty exists in regard to Hapopo kakapo (p. 52) and Kohaka takahea (p. 53) both west of Te Anau, and I waited years before allocating either of them. If Takahe Valley should prove to fulfil the description it might be the Wai o pani, and then Kohaka takahea would have to be applied to another little lake where the birds lived.”

A second claim in the same book that in the 1880's MacKinnon and Richard Henry found deserted tussock huts still standing in the valley would be of the greatest importance if it rested on better evidence than hearsay.

“Reverting once more to the cave traditions, I am told that these came to the ears of Quinton MacKinnon and Richard Henry over 60 years ago, and that they set out to verify the Maori statement [of the existence of large limestone caves in the lower Tunnel Burn] ... Following up the burn they made an even more astonishing discovery, for above the bush-line they came on some deserted tussock whare still standing. These had been left by Maoris hunting the takahea, or Notornis, and bore out their claim that this was one of the homes or breeding-places of that rare bird, and consequently a place to chase the young birds in the correct season for such a procedure.”

In his letter (above) Beattie comments on the source of the first part of this account as follows:

“I do not know if the explorations and discoveries of Q. MacKinnon and R. Henry were all recorded in print, but my account came from the Murrell family [of Lake Manapouri] who have been identified with the district since the early 'sixties.”


Discussion of the geographical range and human association of Megalapteryx is again worthy of a separate paper. For the purposes of this account it is complicated by the discrepancy between Archey's (1941) and Oliver's (1949) system of classification of the genus. The bones from the natural cave deposit on the north side of Notornis Valley, as well as those of the more recent individual from - 106 the Maori shelter (opposite), agree with Megalapteryx didinus (Archey). The species was centred on the great western rain-forest zone of the South Island with northern extensions through Nelson to D'Urville Island, and to the south, spilling over the Fiordland ranges to Central Otago and possibly the Catlins coast. For no other genus are there such well-preserved remains—notably the naturally dessicated head and neck and limb bones from a Queenstown rock cleft, which Owen described (T.Z.S., XI, p. 257); a second and similarly preserved specimen from Cromwell in the Dominion Museum; the dried leg with ligaments, skin and feathers from Waikaia, described by Hamilton (1895); and now the bones from the Notornis Valley shelter.

Although the rapid and effective natural dessication of the examples cited might have preserved them for one or two millenia, the shelter bones are recent by the scale of decades or centuries. The remains represent a very incomplete set of bones which, with the possible exception of three ribs, are clearly from one individual. Sufficient departments of the skeleton are represented to make it clear that the Maori dealt with the whole carcase, and we may easily deduce that the missing bones were flung down the hill slope, where exposure destroyed them.

Pelvis: Base of left ischial process.

Left ischial process, wrenched away, deeply incised cuts on ventral crest; tissue remains. These fragments of pelvis include one distal fragment of pubis, one proximal fragment of ilium.

Ribs: Eight left ribs or rib fragments, two with uncinate process detached, three fragments of massive ribs with uncinate processes completely ossified (? same individual); one with tissue remains.

Six right ribs, two with uncinate processes detached, and with tissue remains.

Pelvic rib, wrenched (1).

Sternal rib processes (3).

Free uncinate processes (2).

Vertebral Column: Atlas and axis.

Twelve vertebrae (two found on surface by Miers; six sub-surface by Miers; Duff and Skinner four).

- 107

Five tracheal rings (Duff and Skinner). Mr. Gurr reports that in December the greater length of the trachea was represented by the rings found articulated.

Complete set of gizzard stones (Duff and Skinner).

Cranium: Part mandible.

Fragment right ramus mandible.

Cranial fragments.


Limbs: Top of femur.

Left tarsus, bleached and weathered.

Three phalanges, two claws and middle toe (? left).

Upper portion fibula.

The length of the tarsus (170 mm.) put it within the range ascribed both by Archey and Oliver to M. didinus. By comparison 12 adult tarsi from the older cave deposit on the north limestone wall, range from 167 mm. to 139 mm. Archey's M. didinus would embrace all these, while Oliver would classify three as M. didinus, the lower nine as M. hectori.

The bone of particular interest is the left ischial process of the pelvis, the upper of the two projections which stem from the femur sockets. Like all the unexposed bones of this skeleton it is a fresh creamy-yellow colour, obviously with a high percentage of organic matter retained, while dried remains of tissue on its inner surface are clearly visible to the naked eye as dark stains. The distal extremity shows signs of gnawing doubtless, as Skinner observes, by Maori teeth. To bring to bear on its lower crest the cutting tool which has left the traces, the pelvis would necessarily have to be laid on its back and the protecting pubic process wrenched away. Unless the ischial projection remained buttressed to the pelvis neither metal knife nor flint could have been applied with sufficient force to scarf it. The projection in turn was then wrenched from the pelvis.

The direction and nature of the cuts are shown in the ventral view of the ischial process (plate 8). Their general line of direction supports Skinner's hypothesis, that they represent a series of metal knife slashes from an operator chopping at right angles to the long axis of the ischium, and not with the helve directed along its axis as with an adze. - 108 The definitive cut is the third from the left. Its diagonal line suggests a badly directed slice with a knife rather than a chop with an adze. A stone flake would also have the same effect as a knife. The five closely grouped incisions on the left are the product of a blade thin enough to have driven to an average depth of 3 mm. into the crest of the bone without punching off a chip. The final two cuts across the neck of the ischium were so closely placed and deeply driven that they removed a noticeable scarf. The right side of the scarf is 7 mm. deep and 14 mm. long, the left 4 mm. deep and 15 mm. long. The width between the two faces of the scarf is 7 mm.

As Megalapteryx did not range over the open grasslands which supported Euryapteryx, and perhaps other large genera, whose slaughter is marked by the early moa-hunter Maori sites along the eastern seaboard of Marlborough, Canterbury and Otago, it is difficult to establish archaeological evidence of its association with man. The only record of the presence of Megalapteryx bones in a moa-hunter camp sensu stricto is from Papatowai (Catlin's) at the southern range of these coastal sites (Teviotdale, 1938). However, the number of genera and species recorded is so all-embracing that I question whether the bones submitted were all truly established as bona fide midden remains. It may therefore be regarded as doubtful. Oliver (1949) sensibly doubts the provenance of a similar association from Waingongoro (on the southwest coast of the North Island) where midden remains lay with naturally deposited remains in a confusion which neither Mantell nor Taylor was able to resolve (Archey, 1941).

The only undoubted association is from a cave burial at Strath Taieri, Central Otago, where a tiny insert of the feathered skin of Megalapteryx was recognized in an inner garment of sewn weka skins wrapped round the skeleton (Hamilton, 1893):

“Over one of these seams on the front I found a narrow strip of skin, placed apparently to cover the join, very much decayed, 3 in. long by ¼ in. wide, but still carrying a dark-grey down, and five or six double-shafted feathers of the moa.” [These feathers are preserved in the Otago Museum (D. 10.172A); they appear identical with the Notornis Valley specimens and may be regarded as Megalapteryx.]

- 109

From the archaic nature of the techniques of the burial wrappings—an inner garment of sewn skins, an outer cloak with thick warps of plaited partly dressed flax with feathers caught into its twists—this association is presumably older than the Notornis Valley find, and might be regarded as not later than the 17th century.

A second possible association in the same general area is the camp reported by Hamilton (1895) from the snow tussock plateaux of the White Comb headwaters of the Waikaia, Central Otago. Continuing on from investigating caves which yielded a dessicated feathered leg and numerous disassociated feathers of Megalapteryx, Hamilton found:

“a space of fully two acres thickly strewn with chips and flakes of stones ... There is absolute evidence in this case of the use of the moa as food. The quantity of bones lying by the ovens in which they were cooked, and the comparative absence of any other inducements (weka being the only other possible prey) for hunters to visit and camp on such an inhospitable spot, all point to these being moa-hunters' encampments ... The caves and shelters in the huge mass of mountains close at hand have furnished some of the best-preserved specimens of the moa yet found, and were, no doubt, a summer feeding-ground of the moa, to which yearly expeditions were made.”

This record must remain only as a possible association because it is not perfectly clear that the midden bones were of Megalapteryx. If established, the association would clearly be older than the 17th century Strath Taieri burial wrapping, and much older than the Notornis Valley occupation which might be regarded as 18th century.

From the associations of the Strath-Taieri find, the period might be conjecturally dated as the 17th century, and the textiles reasonably ascribed to Ngati Mamoe. For Notornis Valley the evidence suggests rather the 18th century (by Skinner's hypothesis some date after 1773).

We are now faced with the problem, that although odd individuals of one of the smaller genera of the Dinornithidae were hunted in Otago-Southland, from one to three centuries ago, the recorded traditions of the Maori of this area give no more support for the recent existence of the “moa” than those from other areas of the South or North Island where everything favours the theory of an early extinction. There are two immediately apparent explanations—the first that the traditions perished with the 18th century subjugation of Ngati Mamoe by Ngaitahu; the second, and more probable, - 110 that Megalapteryx was not recognized as the large vanished bird of tradition, and was not called “moa.” Its name, one imagines, would paraphrase the term chosen by Haast to describe the genus—big kiwi (or big weka). Somewhere among the list of Southland bird names Megalapteryx should be found hiding under his Maori alias—I would suggest toko-weka as an immediate possibility. The modern application of this name to the Stewart Island race of the largest kiwi, the roa or Apertyx australis of the western South Island, in no way settles the question in advance. I agree with Dr. Falla that there is every reason to doubt that Maori who were familiar with Apteryx australis as roa in the main island would call it tokoweka on crossing Foveaux Strait.

The demonstration that one individual of a small species of the Dinornithidae was killed in Notornis Valley probably after the opening of the 18th century does not, in my opinion, invalidate the concept of a distinctive and early moa-hunter period of Maori culture (Duff, 1950) nor does it include its hunters in the term “moa-hunter” as defined above. As coined by Haast, and followed by myself, the term refers to the culture of coastal South Island camps where large genera were killed in thousands. The period when such large birds roamed freely over the eastern grasslands is reasonably established as pre-Fleet for the North Island and not later than 1500 for the South Island. In both islands the tribes concerned were of pre-Fleet origin, and their material culture retained striking survivals of Eastern Polynesian prototypes. Traditions in both islands strongly support the early extinction of the bird referred to most commonly as moa, but also as manu whakatau and kuranui. I believe only Euryapteryx was included in these terms, and it is unfortunate that moa has become, in popular terms, synonymous with Dinornithidae. The anomalous survival of Megalapteryx until the arrival of the Fleet tribes Ngati Mamoe and Ngaitahu in the interior of Southland and Otago, neither makes the first a moa, nor the second moa-hunters.

  • Fire-plough, wood: Length 4 ⅞ in./124 mm.; max. width ⅞ in./22 mm.; depth 9-16 in./14 mm.
- vii
Pelvic process of Megalapteryx; ventral crest showing deeply cut scarfs.
Fire plough.
- viii
Weka snare-lure.
- ix
Sandal (upper view).
- x
Sandal (sole).
- 111

One of the more important finds: a typical fire plough (kaiure) in perfectly sound condition, except for bleached area on back suggesting that it was exposed to air or in dry dust layer (A) when found. Three rubbing faces well blackened with use, and presumably rejected for this reason.

I do not agree with Falla's suggestion that it was shaped with a metal tool (plate 9).

  • Lure, for weka snare, comprising 25 well-preserved weka tail feathers, tied with strip of undressed phormium (plate 9).
  • Noose, of weka snare, undressed flax, with loop kept open by inserted broken twig of mountain beech.
  • Beech twig, broken ends, with remains of small noose of undressed flax.
  • Noose, undressed flax.
  • Noose, folded, very wide loop, undressed flax. Diameter of noose 12¾ in./318 mm.
  • Knotted cord, ? phormium fibre. Fragment ditto.
  • Collection of strips of undressed phormium, some tied.
  • Fragment of epidermis scutched from ? phormium.
  • Knotted phormium strips, probably laces of sandal.
  • Knotted strips of tomentum of a large alpine celmisia [cf. Hamilton's Taieri examples, 1896].
  • Pad [? tinder] of inner bark of ribbonwood (houheria). [Ribbonwood is present in valley.]
  • Fragment of dried flax strip, scored as for thrum of piupiu.
  • Fragments of bark of totara, presumably remnants of containers (poha) for preserving birds, apparently scarfed from trees on Point Burn route from Lake Te Anau.
  • Plaited flax sandal, and heel portion ditto (see below).
  • Fragment of greywacke spawl.
  • Fragment of ground adze of grey baked argillite, Nelson-Marlborough origin. 3
- 112

The most interesting of the artifacts recovered were the complete sandal found in Shelter A and the heel remnant of a second found in Shelter B. These agree in every way with the most common (one-over-one) plaited sandal (paraerae) so widely used in the South Island to protect the feet from frost and snow, and no less from stones and the thorns of the Discaria toumatou (Wild-Irishman or “mata-gowry”). As protection from stones or thorns is not required in Notornis Valley I imagine they were worn principally for warmth.

The complete sandal is perfect except at the heel where the heel portion of the tie laces and the adjacent felts of the sole are missing. By the aid of Te Rangi Hiroa's excellent descriptions of sandal plaiting technique (Hiroa, 1924) I have been able to set it down as a line drawing. Subject to the changes necessary to describe a left foot sandal rather than a right, and to the employment of only four doubled wefts of flax, the Notornis Valley technique is exactly as described by Buck, and I do not propose to describe it. Figure 6 conveys the technique of the complete sandal with the doubled wefts, commencing from the great toe, numbered from 1 to 4, and their paired elements respectively lettered AB, VC, WD, XE. The laces were still tied, the technique again agreeing except that only two side loops were formed, secured through marginal wefts E2 and V1, and A1 and E1. Although imperfect at the heel tie, the laces were still braced by the upper heel-band. The under surface of the sole shows evidenc of hard war.

The sandal is small, approximately 11.1/5 in. (281 mm.) in length, and 4.5/16 in. (110 mm.) in minimum width; the wefts are unusually wide, averaging ⅞ in. (22 mm.).

Fortunately the missing heel portion in this sandal is sufficiently preserved in the fragment from Shelter B to enable the technique of the heel attachment of the laces to be represented (Figure 7).

The Maori sandal has probably changed little in technique since its first development, so that the resemblance of the Notornis Valley sandals to those made by the Kaiapoi Maoris for a Canterbury Exhibition in 1882 should be assessed cautiously in assigning dates. However the shelter sandals most closely resemble the one-over-one (takitahi) - 113 pair found in a cave in the upper Taieri (Te Rangi Hiroa, ibid.) together with other textile material described by Hamilton (1896). From the general textile technique and the ability of the Puketeraki Ngai-tahu to explain the items found, the Taieri textiles might be regarded as eighteenth or even early nineteenth century.


The trend of the evidence is that the Notornis Valley shelters were occupied after the opening of the eighteenth century by a small hunting party of Ngati-mamoe or Ngai-tahu. Their visits were infrequent, if indeed more than a single reconnaissance. As well as Notornis, kakapo, kiwi, weka, and laughing owl, the visitors killed and consumed one individual of Megalapteryx didinus.

At the moment the weight of evidence is against the hypothesis that the cuts on the Megalapteryx pelvis were applied by metal blades; if the hypothesis that they were produced with metal is sustained, the occupation would be later than 1800 and the hunters a party of Ngai-tahu.

  • ARCHEY, GILBERT—1941. “The Moa.” Auckland Institute & Museum Bulletin No. 1, p. 92-4.
  • BEATTIE, HERRIES—1949. The Maoris and Fiordland; 1950. Private letter (February 20th).
  • COWAN, JAMES—1906. “Maori Place Names.” T.N.Z.I., 38.
  • DUFF, ROGER—1950. “The Moa-Hunter Period of Maori Culture.” Canterbury Museum Bulletin No. 1.
  • HAMILTON, A.—1893. “Notes on Some Old Flax Mats Found in Otago” (moa feathers attached, p. 487). T.N.Z.I., 25, 486-8; 1895. “Feathers of a Small Species of Moa (Megalapteryx), Waikaia River.” T.N.Z.I., 27, 232-238; 1896. “Notes from Murihiku.” T.N.Z.I., 29, p. 174-5.
  • HENRY, RICHARD—1903 The Habits of the Flightless Birds of New Zealand.
  • OWEN, R.—1883. On Dinornis (Part XXIV). Transactions of the Zoological Society, London. Part XI, p. 257-261.
  • TE RANGI HIROA—1924. “Maori Plaited Basketry and Plaitwork (2).” T.N.Z.I., 55, p. 357-360.
  • TEVIOTDALE, D.—1938. “Further Excavations at the Moa-Hunters' Camp at Papatowai.” J.P.S., 47, No. 1, p. 32.
  • WHITE, TAYLOR—1893. “A Maori pa at Lake Te Anau.” T.N.Z.I., 26, 513-5.
- 114
FIG. 1.

Floor plan of shelter A, showing limit of outer rim, zone of embedded tussock, and line of trenches 1 and 2 (at left, section at trench 2).

FIG. 2.

Section along trench 2 (shelter A). Layer A (surface dust)—1 in. to 3 in.; layer B (consolidated spoil)—1 in. to 8 in.; layer C (charcoal and bedding)—½ in. to 4 in.; layer D (sterile sand).

FIG. 3.

Rock drawing, shelter B (? human figure or bird man).

FIG. 4.

Rock drawing, shelter B (decorative pattern).

FIG. 5.

Rock drawing, shelter B (decorative patterns, including double spiral).

FIG. 6.

Diagram of complete sandal, shelter A.

FIG. 7.

Diagram (based on heel fragment of sandal, shelter B) showing method of attachment of ties and heel bands.

- 115
FIG. 1.
- 116
FIG. 2.
- 117
FIG. 3.
FIG. 4.
FIG. 5.
- 118
FIG. 6.
- 119
FIG. 7.
1   Mr. C. A. Fleming and others have suggested the possibility of a spectrograph analysis to determine the percentage of iron in the bone. Unless the iron content of the soil matrix was negligible it would be impossible to determine whether the iron in the bone resulted from the fleeting contact of a metal knife or from absorption from the soil.
2   Taylor White's description (1893) of the deserted village on the eastern lake shore at the mouth of the Upokororo refers to the adaptation and use by the Maoris of hoop-iron knives, in a technique independent of the European agency by which the iron started its inland journeys from the Sounds or Foveaux Strait. The village was occupied as late as the eighteen forties, when the Maoris abandoned it with the advance of European settlement.
As we have no record of the length of occupation of the village before the hoop-iron era, its midden moa-bones, recorded in Richard Henry's recollections of 1903, can neither be confidently associated with nor separated from the possibility of metal age expeditions.
3   Dr. Skinner, and others, have raised the possibility of the argillite—grey, with black stain-lines—being of North Otago origin. I suggest Nelson-Marlborough because of the precise resemblance of the fragment to examples of known D'Urville Island origin. The precise point of origin is immaterial to the main issue—whether the hunters were equipped with stone or metal implements.