Demographic transition in a hunter-gatherer population: the Tiwi case, 1929-1996.

Peterson, Nicolas ; Taylor, John (English pop musician)

Populations that have been isolated for extended periods and that have a well-documented contact history are rare and those for which there is reasonable demographic information are even rarer. This paper examines the evidence for demographic transition in what is one of the better documented such population isolates.

In Australia there have been three known population isolates, all on islands. The largest isolate was the 4,000 or so people who made up the population of Tasmania at any one time and who were completely cut off from all external contact for at least 10,000 years. Unfortunately, however, there is no detailed information on the demographic structure of this population at the time of contact and subsequently the population was subjected to genocidal reduction. A second group of people, who stand in strong contrast to the Tasmanians, were the Kaiadilt of Bentinck Island in the Gulf of Carpentaria. At a known maximum they numbered only 123 people in 1942 (Tindale 1962b, 297) and they appear to have had only very limited contact with outsiders for the previous 140 years or more (Evans 1995, 17, 24-27; Tindale 1962a). For these people there is reasonable demographic data collected in 1960 (see Tindale 1962a, 1962b). The third population is that of Melville and Bathurst Islands, which lie in the Arafura Sea off the coast of the Northern Territory (see Hart and Pilling 1960).

As far as can be established, the thousand or so Tiwi who lived on these two islands were completely isolated from the mainland of Australia for at least the last 6,000 years until the eighteenth century. Since then there is a documented history of interaction with outsiders.(1) In 1929, CWM Hart completed a comprehensive census of the whole population. At this time over three-quarters of the population were supporting themselves off the bush. Given the availability of subsequent population counts, and using certain assumptions about the closure of the islands' population, it is possible to trace the demographic transitions that have taken place over the subsequent 60 years as people have moved from an independent hunting and gathering way of life to dependency on the state.

This period covers three successive phases of government policy for Aboriginal people. Hart's material relates to the protectionist era which began at the turn of the century and lasted until the Second World War. The second period was the assimilation era which ended at the beginning of the 1970s. The data for this period comes from the annual returns of the Catholic mission on Bathurst Island to the Northern Territory administration and later returns from the government settlement at Snake Bay on Melville Island. The third period of, so-called, self-determination was ushered in with the election of the federal Labor government in 1972 and is covered by the inclusion of Aboriginal people in the national census from 1971.(2) In addition to these data, there is a comprehensive demographic study of Bathurst Island by Lancaster Jones (1963) and an excellent anthropologically informed study of both islands by Jane Goodale from 1962 (1971).

From April to November 1928 and May to October 1929, CWM Hart carried out fieldwork with the Tiwi. He published seven papers based on this fieldwork and a book, The Tiwi of North Australia (1960), co-authored with AR Pilling who worked with the Tiwi in 1953-54. The census was a major aspect of Hart's fieldwork: it was based on contact with almost everybody covered by it and on detailed genealogical records covering the whole population of the two islands on which the Tiwi live. This census, which was never published and which was only partially worked up by him, forms the stimulus and basis for this paper.

We will begin with a brief history of the Tiwi, followed by a sketch of the relevant features of their social organisation and a consideration of some general aspects of Hart's census before presenting the details. This will be followed by an examination of changes in population size, age structure and sex ratios since 1929 in the light of changes in government policy.

A brief history of the Tiwi

By 8,000-10,000 years or so ago, Bathurst and Melville Islands, on which the Tiwi live, were separated from the mainland of Australia as a result of rising sea levels. There was probably no contact with the mainland until the islanders gained access to dugout canoes and metal axes through contact with Indonesian fishermen.(3) Exactly when these fishermen first started arriving regularly is subject to debate. Campbell Macknight (1976, 8) argues that regular visits to the northern coast of Australia to harvest sea cucumber between December and March probably did not begin until the 1700s, which would be the earliest time the Tiwi could have obtained the means to visit the mainland.(4) However, it may not have been until a buffalo hunter, Joe Cooper, arrived in the 1890s that the people obtained substantial numbers of dugout canoes or the skills to make them, for both Matthew Flinders passing by in 1803 and the officers at Fort Dundas (see below) reported that the dugout canoe was unknown to the Tiwi, although a dugout was seen in use in the northern part of Melville Island in 1877 (Hingston 1938). Indonesian fishermen were probably almost annual visitors to the islands until 1907 when prohibited by the Australian government. Relations between the islanders and the fishermen do not seem to have been particularly friendly (Searcy 1909, 46) nor intense as the trepang beds around the islands were not particularly rich.

There is some scanty evidence that the Portuguese may have raided the islands for slaves in the 1700s (see Hart and Pilling 1960, 97-98), and in 1705 employees of the Dutch East India Company visited the islands over a period of two months during which there was both fraternisation with the Tiwi and conflict (Forrest 1995). The first sustained contact with outsiders was not until 26 September 1824 when the British founded Fort Dundas on the northwestern part of Melville Island near what is now Pularumpi (formerly Garden Point). Relations with the Tiwi were poor and the settlement was abandoned on 31 March 1829. Although the exact nature of the impact that this settlement must have had on the population is unclear, there is every reason to suppose that it was not so great that, in the 68 years before the establishment of the next permanent outsider presence, it had largely been mitigated. There was no wholesale slaughter of Tiwi nor reports of numerous deaths from new diseases.

While various ships visited parts of the islands, ran aground or sank off them (see Hart and Pilling 1960, 98-99), it was not until 1895 that another attempt at settlement on the islands was made when a buffalo shooter, Joe Cooper, established a base on the southwest of Melville Island at Paru. Cooper left the island in 1897, it seems, when his brother was speared, returning around 1900 with an entourage of Aboriginal people, mainly Iwaidja speakers from the vicinity of the Cobourg Peninsula, and stayed until 1916 (Krastins 1972, 41, 43).(5) The impact of Cooper and his workers, particularly on his return, seems to have been more marked, especially because he armed his Aboriginal workers, some of whom made their guns available to people identifying as Mandiimbula to attack the Yeimpi (Pilling 1958, 196). There is, however, no evidence about the numbers or sex of those killed.

Permanent settlement began in June 1911 when a Catholic mission was established on the southeast coast of Bathurst Island by Father Gsell. By 1923 the number of Tiwi at the mission averaged 50, only a small fraction of the total population on the island, but in 1921 Gsell initiated a practice which was to have a substantive impact on the people's way of life and family structure. He started purchasing the right to bestow young girls and with this gained the right to place them in his convent to be brought up by the mission. Between 1921 and 1938, one hundred and fifty girls were placed in the convent where they remained until their mid-teens (Gsell 1956, 90).

From the late nineteenth century, Indonesian fishermen were joined by Japanese pearlers initially working the pearl beds between Darwin and the two islands and in the Apsley Strait between Bathurst and Melville Islands (Krastins 1972, 39). They established trading relations with the Tiwi, first on the south coast of Melville Island and later on the north coast, in which various goods were exchanged for access to Tiwi women.(6) On 27 March 1939 the government established a presence at Garden Point close to the site of Fort Dundas to regulate relations with the Japanese taking place there and elsewhere (Commonwealth of Australia 1938-39, 23), but it was not until the outbreak of war that these came to an end.

At the beginning of the Second World War, the population of the mission was told to return to the bush to support themselves from hunting and gathering (Jones 1963, 22), and a military base was established on the southwest side of Snake Bay on the north coast of Melville Island. In 1945 with the cessation of hostilities, this was turned into a government-controlled settlement for the Tiwi. In the year following the base becoming a settlement, over 250 Tiwi visited it, mainly people living a free life in the bush up until that time (Commonwealth of Australia 1946, 26).

Since the end of the war, there have been no Aboriginal people living a completely independent life in the bush and today almost all of them live at one of the three settlements, all of which have been given Aboriginal names. Bathurst Island mission is now known as Nguiu, Garden Point as Pularumpi and Snake Bay as Milikapiti. There are two other settlements, both on Melville Island: Paru, opposite Nguiu, where Cooper had his camp and now occupied by a small number of older non-Christian Tiwi; and Pickertaramoor, which was formerly a forestry camp but which with the demise of the forestry industry now has a limited population. In this analysis the whole population is covered, the Paru population being included with the Bathurst Island population and the Pickertaramoor population with the Snake Bay population.

Although these contacts clearly had an impact on the population, there is no reason to believe that any of them have so radically altered the structure and size of the population that it bears no relation to the pre-1824 population. The population at that time could, undoubtedly, have been impacted on already by disease introduced from Macassar; however, comparative population densities with other areas in Australia (see below) do not suggest that, at the time Hart was carrying out his census, there was radical depopulation. Events known to have had a dramatic impact on the death rate, such as the 1918-19 flu epidemic, are evident in the 1929 age structure (see below), yet their demographic impact appears to have been short-lived.

Social organisation and the census

Little need be said here about Tiwi social organisation as it has been well documented by the several ethnographers who have worked on the islands (Brandl 1971; Goodale 1971; Grau 1983; Hart 1930; Pilling 1958), except to comment on how it is related to Hart's census.

The two islands are divided into a number of `countries'. All ethnographers are agreed that country membership is primarily patrilineal and, according to Goodale's informants, people were expected to marry within their country, at least before Europeans arrived (1971, 18). The names of these countries and the numbers of people affiliating with each of them has varied over time (Table 1). In 1981, the number of countries was renegotiated, primarily as a result of land rights legislation, resulting in a current list of seven (Venbrux 1993, 28-29).

Table 1 Names and numbers of people associated with Tiwi countries, 1929-88

Hart 1929 Goodale 1962

Malauila 111 Malau 101
Mingwila 56
Rangwila 66 Uranggu 146
Tiklauila 213 Tikalaru 173
Munupula 170 Munupi 102
Wilrangwila 84 Wileranggu 44
 Palauwiunga 29
Turupula 67 Turupi 41
 Yanganti 18
Yeimpi 152 Yiempi 23
Mandiimbula 142 Mandiupi 81
 Unassigned 39

Brandl 1969 Venbrux 1988

Malauwila 202 Malau

Rangkuwila 321 Rangku
Tikilauwila 258 Tikelaru
Munupula 147 Murnupwi
Uluruangku 86 Wulurangku

Tirupula 33

Yeimpi 50 Impinari
Mantiumpila 175 Mandiupwi
Unassigned 111

Note: Goodale and Venbrux use the names of the countries while Hart and Brandl mainly use the name of people associated with the countries as marked by the suffix -ila.

Sources: Hart 1929 (unpublished field notes); Goodale 1962 (Goodale 1971, 16); Brandl 1969 (Brandl 1971); Venbrux 1988 (Venbrux 1993, 285)

Almost all of Hart's census material was compiled by him in terms of these countries. Given that Hart apparently spent much of his time living in the bush, particularly in the two northern countries of Malau and Munupi but visiting all of the countries on both islands for varying periods (Hart et al 1988, 150, 155),(7) it can be assumed that he was quite well informed about the countries, their populations and their significance for understanding Tiwi life. Within the countries, the population spent more than three-quarters of the year living in small groups often made up of several households which Hart called establishments. There were seven such establishments, or bands as they would now be called, in Malau and about nine in Munupi in 1928 (Hart et al 1988, 155).

Nevertheless, the significance of country affiliation as a unit of analysis is not entirely clear since Hart writes that `the best generalisation [about country affiliation--that is, what Hart calls `horde' membership] is perhaps to say that a woman retains all her life a connection with the horde into which she is born no matter where patrilocal marriage may lead her, while a man on the other hand belongs to whatever horde he is living with at the moment, this being normally but not at all invariably his father's horde' (1930, 176). Thus, when Hart writes of Malauila males and females, for example, he is not writing of the total population of Malauila country but about the notional membership based on males and females of Malauila fathers resident in the country, the children of Malauila males and Malauila females living in other countries. The actual population of Malauila country would include women from other countries who had married Malauila men and moved to live with them. Clearly, too, there has to be some category of visitor, so that `living with' is different from `temporarily resident in' Malauila country. It can be presumed that it took some time before a man became identified with a new country after moving to take up residence there.(8) Thus, the listings by country are not groups but categories. While this makes the fluctuations in their sizes over time more comprehensible, it makes their use for demographic analysis puzzling. We have, however, kept his categories since they were clearly perceived as important by Hart and are for that reason of interest, but they are problematic for demographic analysis if only because of their small size. The more reliable figures, given the size of the population as a whole, are the aggregate figures for the two islands.

The terminology Hart uses for these countries is also confusing: he refers to them as band or horde countries (1930, 173), when these terms would today be used in respect of the districts or sub-countries into which they were divided (Goodale 1971, 14). Each of these sub-countries was occupied by a group of families which Hart sometimes calls an establishment. The establishments, or bands, were composed of households and the households of the older men were large because they were highly polygynous, with one man having as many as 29 wives, if those promised plus those actually living with him are included (Hart and Pilling 1960, 17). People also belonged to matrilineal descent groups. Goodale records the names of 23 but only 21 had living members (1971, 20-21).(9) None of Hart's data deals with matrilineal descent group membership.

Hart's data was collected over an extended period and based on genealogical evidence rather than a count on a particular day, although he states explicitly that he `met and talked to the whole tribe' except for the younger men living in Darwin. Working from genealogies has some obvious dangers, in particular that people who are dead may be counted as living and that some people may be counted twice or not at all. It is evident that Hart was aware of the problem of distinguishing the deceased from the living as some of his tables make evident. He was also well aware of the problem of being confused about people's identity, especially because of the Tiwi practice of renaming all of a woman's children each time she remarries and prohibiting the use of the former names (Hart 1931; Hart et al 1988, 157-58). This does not, of course, eliminate the possibility of error but indicates an awareness of the problems. When combined with the fact that he travelled all over the two islands and met most adults, it seems that we can be reasonably confident of his figures. He wrote of his genealogical work: `Generalisations made on the basis of a complete genealogical record of a people are the nearest equivalents anthropology can offer to the objective and experimentally checkable generalisations of the natural sciences and are the least influenced by that unconscious bias towards such and such a point of view which seems inherent in even the most disinterested and objective fieldworkers'.(10)

Another aspect of Hart's census is that everybody is given an age. One problem with this is that it is not certain what system he was using, but, assuming that he was following conventional practice by which we call somebody a particular age only when they have completed the year in which they reach that age, his listing would require some people to be assigned an age less than 1. Nobody, however, is assigned an age of 0. This prevents an assessment of infant mortality. A second issue is how he arrived at the absolute ages assigned to every single person. Aging people is notoriously difficult in any context, not least in a cross-cultural one. He does not outline his methodology fully but he does state that all Tiwi could rank themselves relative to everybody else and that this was the cornerstone of his methodology.(11)

Total population characteristics, 1929

The estimates made of the original population of the two islands are highly variable. Despite their limited area, Bathurst Island being approximately 2,071 sq km and Melville Island 5,697 sq km, amounting to only 7,768 sq km in all, people have found it hard to arrive at a common estimate of the population. Reconstructing the population in 1954, Pilling estimated that at the time of the settlement of Fort Dundas in 1824 there were between 200 and 400 Tiwi all up (1958, 39). This would give an average of one person to between 39 and 19 sq km. The former figure would seem to be very conservative. Basedow, who visited the area in 1911, estimated the population of Bathurst Island at 500 (1913, 291), which would suggest an overall population of around 1,500 for the two islands based simply on area. However, it seems that Melville Island, though double the size of Bathurst Island, may not have been as hospitable an Environment.(12) Father Gsell also erred on the optimistic side, believing that there were 1,000 people on each (1956, 40-41; see also Ritchie 1934, 24, who makes the same estimate). Fry, a doctor, visited the islands in 1913 and estimated the population at around 650 (1949).

Hart himself found it hard to work out what the population was before he completed working over his census material. His first estimate was that the total population `might be anything between 1000 and 1500 although I tend to think the latter figure nearer the mark.(13) In August 1929 when he completed his census of all living Tiwi, he counted 1,068 distributed as follows: 2 in Sydney, apparently permanently; 34 in Darwin; 22 working on Japanese luggers; and 1,010 on the two islands, giving a grand total of 1,068.

The subsequent figure used for the population was 1,062, presumably because of the balance of births and deaths by the time he closed off his count (see, for example, Hart et al 1988, 8), but in this paper we use a total of 1,061 as one record is missing. If Pilling's higher estimate of the population in the 1820s were accepted, it would suggest that the population had grown by two-and-a-half times in the 100 years between the settling of Fort Dundas and Hart's fieldwork. This is substantially higher than the near-stationary growth levels postulated for the Aboriginal population prior to contact with Europeans, although achievable in this period of time (Gray 1985). It is also in spite of the fact that from the turn of the century, at least, new factors were added to the death rate: liaisons with the Japanese introduced various diseases while, as has been mentioned, some of the Iwaidja speakers made guns available to some Tiwi which were used in feuding (Pilling 1958, 196). Thus, overall, the evidence suggests that the original population was a good deal higher than 400 prior to the establishment of Fort Dundas.

The age and sex distribution shown in Table 2 has several features of interest. The sharp decrease in males aged 10-14 years and females aged 15-19 years relates to the period between 1910 and 1919 when two major epidemics were reported, one in 1913 (see Fry 1949, 79-80) and one in 1919. Although Fry reports that 56 men, 87 women, 17 boys, 25 girls, and one male and one female infant died in the 1913 epidemic, 187 in all, this high number seems somewhat unlikely, as Jones suggests (1963, 23). There could be many sources of error, including the methodology and the lumping together of deaths from a number of years especially since the majority of people were then living in the bush away from the mission.(14)

Table 2 Age and sex distribution of Tiwi population, 1929

Age Male Female Total

0-4 58 73 131
5-9 78 64 142
10-14 48 70 118
15-19 51 23 74
20-24 74 40 114
25-29 44 65 109
30-34 46 66 112
35-39 30 36 66
40-44 20 34 54
45-49 21 22 43
50-54 12 22 34
55-59 15 12 27
60-64 10 11 21
65-69 6 4 10
70-74 3 1 4
75-79 1 0 1
80+ 0 1 1
Total 517 544 1,061

Note: for raw data, see Table 10

Source: Hart 1929 (unpublished field notes)

The marked discrepancies between the numbers of males and females in the age brackets 15-19 and 20-24 years, with a reversal between the sexes for the age groups of 25-29 and 30-34 years, may simply be the effects of differential mortality rates or they may indicate that Hart had difficulty estimating the age of the younger women, consistently seeing them as older than they were.(15) It seems clear, however, given the substantial excess of women over men that there was no systematic infanticide of female children at this period, although it may be that at specific periods in the past it was practised in order to secure desired sex ratios.(16)

Effects of contact on the regional populations of the Tiwi

This issue seems to have interested Hart the most, largely because he was working in a strictly functionalist paradigm and keen to locate the most `pristine' population. He divided the regional populations into degrees of `contamination' by outside influences. The populations of Malauila and Munupula he describes as `uncontaminated', although Fort Dundas had been established on Munupula country 100 years previously. The Wilrangwila, Turupula, Yeimpi and Mandiimbula he describes as populations influenced by Cooper and the Japanese, although it seems surprising that the Mandiimbula directly opposite the mission on Bathurst were not influenced by it: this is presumably because of the influence he attributed to Cooper. The third population is made up of Tiklauila, Rangwila and Mingwila as a group influenced by the mission. The validity of these groupings has to be accepted, although from a formal demographic point of view they are not big enough to make the variation between them statistically meaningful. There are some other problems too.

For instance, the Turupula are the most spatially distant from the areas where contact is known to have taken place with the Japanese, yet the sex ratio for this group is more unbalanced than for the Mandiimbula who were adjacent to Buchanan Island where much of the contact is known to have taken place (Table 3).(17)

Table 3 Percentages of Tiwi males and females, by country, 1929

 Males Females
Populations with
little contact

Malauila 45.9 54.1
Munupula 44.7 55.3
Total 45.3 54.7

Populations influenced by
Cooper and Japanese

Wilrangwila 56.1 43.9
Turupula 60.6 39.4
Yeimpi 56.3 43.7
Mandiimbula 45.7 54.3
Total 54.7 45.3

Populations influenced by
the mission

Tiklauila 44.4 55.6
Rangwila 41.8 58.2
Mingwila 49.1 50.9
Total 45.1 54.9

These figures appear to indicate that the sex ratios in the countries with little contact and the mission are broadly similar, marked by a substantial predominance of females, while those in the countries with contact with the Japanese and Cooper have a predominance of males, with the exception of Mandiimbula which fits the former pattern.

The figures in Table 4 on the number of wives tell a similar tale to those in the previous table. The countries influenced by Cooper and the Japanese come out lower on each index on average. Interestingly, however, those people in the vicinity of the mission are the most polygynous and also have the greatest number of living children, possibly because of the rations and health care available from the mission, as well as having the highest fertility.

Table 4 Average size of Tiwi families per married man, by country, 1929

 Wives Children
 (living) (living)

Populations with little contact

Malauila 2.76 3.06
Munupula 2.03 2.53

Populations influenced by
Cooper and Japanese

Wilrangwila 2.00 1.99
Turupula 1.31 1.44
Yeimpi 2.11 2.90
Mandiimbula 2.43 1.46

Populations influenced by
the mission

Tiklauila 3.20 3.43
Rangwila 2.90 3.10
Mingwila 2.55 2.10

 Wives Children
 (living (living
 & dead) & dead)

Populations with little contact

Malauila 4.17 4.53
Munupula 2.72 3.84

Populations influenced by
Cooper and Japanese

Wilrangwila 3.17 3.18
Turupula 2.00 2.20
Yeimpi 3.20 4.61
Mandiimbula 3.86 3.20

Populations influenced by
the mission

Tiklauila 4.60 5.46
Rangwila 3.80 4.40
Mingwila 3.10 4.90

In Table 5, the average sizes of families of married men over 50 are considered. For this group, the number of living wives increases as does the number of children, both doubling for the population as a whole. The average number of living wives for men over 30, using Hart's figures, is 3.73 in 1929. Thirty-three years later, it had dropped to 1.93 (see Goodale 1971, 58, xix). These average figures cover a great range of variation in Hart's time when men with ten or more wives were still common (Hart and Pilling 1960, 17).

Table 5 Average size of Tiwi families per married man over 50, by country, 1929

 Wives Children
 (living) (living)

Populations with little contact

Malauila 4.40 7.40
Munupula 3.30 5.30

Populations influenced by
Cooper and Japanese

Wilrangwila 2.75 5.00
Turupula 1.25 3.00
Yeimpi 3.85 7.40
Mandiimbula 3.57 3.57

Populations influenced by
the mission

Tiklauila 5.62 5.62
Rangwila 3.30 5.00
Mingwila (no figs, only one case)

 Wives Children
 (living (living
 & dead) & dead)

Populations with little contact

Malauila 8.00 10.20
Munupula 6.30 10.10

Populations influenced by
Cooper and Japanese

Wilrangwila 5.25 8.00
Turupula 2.50 4.00
Yeimpi 6.57 10.90
Mandiimbula 7.10 8.29

Populations influenced by
the mission

Tiklauila 8.40 9.70
Rangwila 5.66 7.00
Mingwila (no figs, only one case)

Such high levels of polygyny could only be maintained by most of the male population remaining unmarried. Table 6 indicates the percentage of married men by country and shows that, overall, 64 per cent of the male population was unmarried. It is interesting to compare this figure with Goodale's figure, collected 33 years later in 1962, where she found that only 30 per cent of the male population was unmarried (Goodale 1971, 58).

Table 6 Percentages of married and single Tiwi men, by country, 1929

 Married Single Under 15

Populations with
little contact

Malauila 33.0 23.5 43.5
Munupula 42.1 22.3 35.6

Populations influenced by
Cooper and Japanese

Wilrangwila 39.1 32.6 28.3
Turupula 40.0 35.0 25.0
Yeimpi 30.2 32.6 37.2
Mandiimbula 43.1 35.4 21.5

Populations influenced
by the mission

Tiklauila 31.6 17.9 50.5
Rangwila 35.9 14.1 50.0
Mingwila 32.2 10.7 57.1

Table 7 provides details of the average number of children, living and deceased, of women over 40 years of age.(18) For such women, who were almost certainly near or at the end of their child-bearing years, the average number of births was 4.17. The group that Hart counted as the least contacted, the Malauila, had eight women over 40 years with 23 living and dead children between them, giving an average fertility rate of 3.1. Hart believed that this reflected fertility levels under a precontact regime.

Table 7 Average number of children of Tiwi women over 40, by country, 1929

 Children
 (living & dead)

Populations with little contact

Malauila 3.10
Munupula 4.80

Populations influenced by
Cooper and Japanese

Wilrangwila 3.85
Turupula 3.50
Yeimpi 3.43
Mandiimbula 4.66

Populations influenced by the mission

Tiklauila 5.40
Rangwila 5.20
Mingwila 4.00

Table 8 sets out the percentage of the population in each age group in terms of the same three nature-of-contact aggregations. It can be seen that the proportion of the population at the mission aged 10 years and under is high, pointing to high growth there, but it could well have been inflated by parents who lived in the groups that had contact with the Japanese sending their children to the mission. This possibility is suggested by the fact that only 19.6 per cent of the group contacted by Cooper and the Japanese were children under 10 years of age.

Table 8 Percentage of Tiwi population in each age group, by nature of contact, 1929

Age group Least contact Cooper, etc Mission

0-10 years 33.25 19.60 37.80
11-25 years 23.60 33.60 29.00
26-35 years 21.20 18.80 16.30
36-50 years 15.20 18.20 11.20
51+ years 6.80 9.70 5.60

By 1929, Gsell's practice of `purchasing' young females for the convent had been running for eight years, with 25 already bought by June 1925 (Commonwealth of Australia 1925, 17), but, as Table 9 makes clear, this policy does not appear to have appealed to the Cooper and Japanese contact groups as only one person from these countries had a mission marriage. For these mission marriages, the average age of the husband was 28.3 years and of their wives, 23.7 years, making a difference between husband and wife of only 4.6 years. This is a huge change from the age difference between husbands and wives for the rest of the population at that time. It is not possible to calculate what this was from Hart's figures at this stage,(19) but Goodale's figures for age differences between husbands and wives at Snake Bay in 1962 indicate that it was still 21.8 years (1971, 64).

Table 9 Tiwi mission marriages, by country, 1929

 Marriages Children
 Live Dead

Malauila 2 4 1
Munupula 1 4 0
Wilrangwila 1 1 1
Tiklauila 12 17 7
Rangwila(*) 2 8 3
Mingwila 3 1 0
Total 21 35 12

(*) Hart in his field notes has Wrangwila. We assume this is Rangwila.

At the time Hart compiled his census in 1929, over three-quarters of the population was living an independent life in the bush, although all of the population had substantial contact with either the mission or Japanese pearlers. Despite the fact that there had been some permanent non-Aboriginal settlement on the island since 1900, it does not appear to have had a dramatic effect on the population overall. The transition from a population structure related to a self-sufficient hunting and gathering way of life to a population encapsulated by the state and welfare-dependent took a considerable period to establish itself. While it would be wrong to treat the population structure as if it were a pristine image of the precolonial situation, it is as close as we are likely to get for an entire Aboriginal population. As will be seen from the material to be presented, it took 50 years before the size and age structure of the population started to change radically.

At 1,061 the population density was 7.3 sq km per person or, on an island basis, 4.6 sq km per person on Bathurst and 9.3 per person on Melville Island, confirming the general impression that Melville Island was less hospitable. This is a high population density compared with the sort of overall figure Warner calculated for eastern Arnhem Land at around 20 sq km per person (1958, 16), but the Tiwi economy was clearly a coastal one although not fed by large and rich freshwater rivers such as in the Anbara area where population densities rise to 2 sq km per person (Meehan 1982, 15). The Tiwi population density is close to that calculated by Sharp for coastal western Cape York where the actual land used was estimated at 6 sq km per person (1940, 487).

The overall sex ratio with its strong preponderance of women (100 males to 106 females) does not suggest a population grossly impacted upon by colonisation, which often results in a lower preponderance of women since they frequently bear the brunt of new sexually transmitted diseases which not only affect their health but reduce their fertility. Nor does it suggest the prevalence of female infanticide. Such a sex ratio stands in marked contrast to the figures from western Cape York. There the overall sex ratio was 100 males to 76 females (Sharp 1940, 490).

The age structure of the Tiwi population also tends to confirm the lack of colonial impact. Although it is not possible to calculate an infant mortality rate, it is evident that the regime was one of high fertility and high mortality, as the age distribution in Table 2 indicates. Again, the comparison with western Cape York is instructive. There, the average number of children per woman was 2.3 (Sharp 1940, 501), while among the Tiwi the average was 2.5 per woman, with the range varying by country from 1.44 to 3.43 (see Table 4).

The mission impact on Tiwi social structure was still limited in 1929. The best single measure of this is the fact that 63.6 per cent of the male population was single, although many of these people were involved in marriage contracts yet to be realised. After eighteen years of missionary work, only 21 couples had been married by the mission. Among the western Cape York population, which did not have anything like the same precolonial levels of polygyny (Sharp 1940, 497), only 26.7 per cent of the male population was single (Sharp 1940, Table 1 and 6b).

Tiwi demographic transition

The availability of Hart's historic data provides an apparently unique opportunity to chart the changing demographic structure of a discrete Aboriginal population as it has moved from an independent hunting and gathering mode of existence to incorporation into wider social and economic networks characterised by growing dependency on state institutions. Much depends in this exercise on an, admittedly tenuous, assumption regarding closure over time of the Tiwi population on Bathurst and Melville Islands.

While it is known that migration to the settlement of Darwin on the neighbouring mainland continued following Hart's census, such data as are available suggest that this movement was neither large nor overly selective in demographic terms and the vast majority of Tiwi appear to have remained on the islands, at least up to the 1970s (Brandl 1971; Jones 1963, 22). Furthermore, much of the earlier movement away from the islands has been described as temporary in nature and characterised by spells of employment in Darwin of around three months duration. By the mid-1950s it was estimated that the number of Tiwi resident in Darwin was no more than 150 and most of these had migrated many years earlier while some were reportedly moving back to the islands (Jones 1963, 24). The most recent, but now very dated, estimate of the number of Tiwi resident in Darwin is provided by Brandl (1971, 55) who reported a total of 188 in 1969. While it is most likely that this figure increased in subsequent years, such migration data as are available from the five-yearly national population census point to only limited net outflow of population from the islands since the 1970s. The fact is that movement in both directions is recorded and there is a very real sense in which Darwin now forms part of the social and economic space of Tiwi Islanders, with frequent movement occurring between the islands and the mainland for a host of reasons.

Despite the fact that Darwin-based individuals may be seen as an integral part of the Tiwi population, no statistical data are available to enable their inclusion in an analysis of Tiwi population change over time. For this purpose, only data for those recorded as resident on Bathurst and Melville Islands at particular points in time are available. Thus, in constructing a framework to provide a preliminary sketch of Tiwi demography, it is simply assumed that population change data for these islands maintain a semblance of closure over time and that the main factor controlling the construction of a time series is data availability rather than its internal consistency.

Data sources

Although an annual Aboriginal census was conducted by the Northern Territory administration from 1921 to 1944, population figures for Bathurst and Melville Islands were not identified separately from the larger Census District of Darwin until a new annual count of state wards commenced after the Second World War. This coincided with a shift in government policy aimed at the ultimate assimilation of Aboriginal people into mainstream society. Central to this goal was the encouragement of Aboriginal residence in government and mission settlements and an annual population count of such localities became part of the machinery of welfare administration. Thus, for the assimilation period of the 1950s and 1960s, data are available indicating the total numbers of Aboriginal people either resident at, or supervised from, the Bathurst Island mission and the government settlement of Snake Bay on Melville Island.(20) While these data are disaggregated by sex, no age breakdown is available beyond a crude distinction between adults and children, although a detailed analysis of the population of Bathurst Island mission throughout the 1950s provides a partial exception to the data shortfalls of this period (Jones 1963).

Much more detailed data are available for the Tiwi during the period of self-determination which began in 1972. Since the census of 1971, Aboriginal people have been included in the full enumeration of the Australian population every five years by declaring their Aboriginal origin on the census form. Although this introduced a definitional shift in the official construction of Aboriginal identity and involves a degree of error most probably tending towards underenumeration (Martin and Taylor 1996), it enables the most detailed demographic description of the Tiwi population of both islands to be made since Hart's efforts of the late 1920s and extends the analysis of population change to the present day.

The numbers of Tiwi recorded as resident on Bathurst and Melville Islands at regular points in time between 1929 and 1996 are plotted in Figure 1. This reveals three distinct phases of population growth: a period of population decline from the late 1920s to the early 1950s; a period of fluctuating growth from the mid-1950s to the early 1970s; and a period of sustained growth through the 1970s and 1980s. The broad features of demographic change during each of these periods are discussed below.

[Figure 1 ILLUSTRATION OMITTED]

Population growth 1929-50

Adopting the practice of including in the population total those Tiwi who were temporarily absent, we can accept a figure of 1,061 for the Tiwi from Hart's data of 1929. This being the case, it would appear that the population slumped during the 1930s and 1940s by almost 1 per cent a year to reach a figure of 832 in 1950. Whether this is an accurate portrayal of the population trend for the whole of this period is difficult to establish in the absence of separate annual data for the Tiwi. It is worth noting, however, that overall Aboriginal population growth in the Northern Territory is considered to have been only roughly stable around this time (Smith 1980, 179), which does allow for the possibility of decline in some regions, as indicated for Melville Island.

It comes as little surprise, therefore, that the distinctly pyramidal structure of the population evident in 1929 was less obviously so in 1951, at least as far as age/sex data for the Bathurst Island mission drawn from Jones (1963, 28) indicate (Table 10 and Figure 2). However, as with the broad base of the age pyramid in 1929 (26 per cent under 9 years of age), the relatively large proportion of children in 1951 (23 per cent) is indicative of a population experiencing renewed growth after a period of decline. In both instances, it seems that the introduction of mission health services may have played a vital role in this revival.

[Figure 2 ILLUSTRATION OMITTED]

Table 10 Age and sex distribution of Tiwi population, 1929-96

 1929 1951

Age group M F M F

0-4 58 73 36 61
5-9 78 64 29 38
10-14 48 70 27 23
15-19 51 23 27 34
20-24 74 40 33 42
25-29 44 65 34 48
30-34 46 66 33 23
35-39 30 36 21 22
40-44 20 34 15 19
45-49 21 22 12 22
50-54 12 22 19 19
55-59 15 12 16 23
60-64 10 11 8 11
65-69 6 4 2 4
70+ 4 2 1 5
Total 517 544 313 394

 1971 1996

Age group M F M F

0-4 99 100 83 96
5-9 104 76 88 94
10-14 60 71 98 92
15-19 37 59 120 92
20-24 32 53 94 120
25-29 40 36 96 85
30-34 26 31 79 75
35-39 29 40 61 68
40-44 31 48 58 50
45-49 44 42 34 53
50-54 28 23 39 35
55-59 23 19 17 17
60-64 6 9 11 15
65-69 8 13 5 10
70+ 16 18 7 13
Total 583 638 890 915

The other feature of this period of population decline is a dramatic fall in the number resident on Melville Island and a concurrent rise in the population of Bathurst Island (Figure 1). Indeed, shortly after Hart's census of 1929, it seems that Bathurst Island became the primary focus of residence for the Tiwi, a position that it has held ever since. The main reason for this inter-island transfer of population would seem to have been the drawing power of the Bathurst Island mission.

Population growth 1950-76

In line with the experience of Aboriginal people generally in the Northern Territory, the beginnings of sustained growth among the Tiwi population can be traced to the early 1950s, with two distinct periods of growth discernible over the subsequent 25-year period to the mid-1970s (Figure 1). The first of these periods is characterised by a steady rise in the level of population growth during the 1950s, particularly on Bathurst Island. According to Jones (1963, 34), this can be attributed to the establishment of a hospital at Bathurst Island mission in 1946 which had the almost immediate effect of halving the previously high infant mortality rate while fertility levels remained high and even increased?

A second phase is apparent during the 1960s, with a slowing down of overall population growth due to fluctuating fortunes on Bathurst Island as the population of Melville Island continued to expand throughout this period. Given continued high fertility, the main cause of this variability in population growth has been attributed to rising infant mortality on Bathurst Island contingent upon poor environmental health in the increasingly congested mission settlement. A high incidence of gastro-enteritis and respiratory disease among Aboriginal children was a common feature of this period (Packer 1962, 67), while Jones (1963, 47) noted the growing incidence of childhood deaths at the Bathurst Island mission due to diseases commonly associated with overcrowded and unhygienic living conditions. Notwithstanding this rise in childhood mortality, the persistence of high fertility is reflected in a further broadening in the base of the age pyramid by 1971 (Table 10 and Figure 2).

Population growth 1976-96

The current phase of Tiwi population growth from the mid-1970s to the present is remarkable for the continuation of high growth rates at a time when fertility levels have declined (Figure 2). Between 1976 and 1996, the Tiwi population grew at an annual rate of 2.5 per cent, which is substantially higher than the rate of 1.4 per cent sustained over the previous 25 years. At the same time, however, fertility levels showed clear signs of decline, with total fertility rates (TFRs) among the Tiwi falling from 5.7 in the period 1967-71 to 4.4 between 1972 and 1976 (Gray 1985, 25). While fertility data for the Tiwi have not been analysed beyond this date, fertility decline continued in subsequent years among Aboriginal people generally in Australia (Gray 1990; Gray and Tesfaghiorghis 1993), while in the Northern Territory the TFR in 1996 had fallen to 2.7 (Australian Bureau of Statistics 1996a, 56). The steeper age profile for the Tiwi and greater concentration of population in older age groups revealed by the 1996 Census suggests that the island populations had shared in this fertility decline (Table 10 and Figure 2).

Whether this was sufficient to account for the abrupt levelling off of population growth observed between 1986 and 1991, however, seems unlikely. More plausible explanations may be consistent with doubts raised by the Tiwi Land Council about the accuracy of census counts (Commonwealth of Australia 1992), particularly given that the Aboriginal population generally in the Northern Territory continued to rise at around 3 per cent a year over the same period. A related possibility could have been the renewed out-migration to Darwin during the late 1980s as the city became a regional focus for increased expenditure and job opportunities in Aboriginal affairs.

The overriding cause of recent Aboriginal population growth generally has been a significant decline in infant mortality rates, mainly due to a marked decline in post-neonatal deaths (Thomson 1983). Overall in the Northern Territory, infant mortality rates declined from 84 per thousand live births in 1972 to as little as 29 per thousand in 1988 and 19 per thousand by 1996 (Australian Bureau of Statistics 1996b, 41). While separate rates for the Tiwi have not been calculated, analysts of Aboriginal mortality decline are unequivocally of the view that comprehensive public health care intervention over the past 20 years, rather than improvements in the social and economic conditions of Aboriginal communities, have been responsible (Gray and Tesfaghiorghis 1992). At the same time, however, increased intervention by the state in Aboriginal communities such as those on Bathurst and Melville Islands across the whole gamut of social services has had major implications for housing and community infrastructure as well as education, employment, and income levels.

Precisely what links exist between these socio-economic changes and the manifest demographic transition among the Tiwi remain to be fully established. Whatever these may be, it is certainly unlikely that, when projecting the population of Bathurst Island mission, Jones (1963) could have foreseen the scale of state intervention in the daily lives of the Tiwi that has occurred since the early 1970s in such a way as to contemplate building this into his analysis. In this context, it is interesting to note that the combined population of the two islands in 1996 was not much greater than Jones' lowest projected population for Bathurst Island mission alone in 1986 of 1,636. This projection was based on assumptions regarding sustained high fertility and somewhat lower mortality, excluding the option of increased out-migration (Jones 1963, 55-57). Thus, the lower than expected population growth that eventuated suggests that all, or some, of Jones' assumptions failed to hold. To discover precisely which ones, and why, requires further analysis based on the collection of household-level data regarding recent demographic processes.

Conclusion

Hart's census provides a unique window on to the demographic structure of a hunting and gathering population just prior to it undergoing major changes. The initial changes came from two main sources. The deliberately intended consequences of social engineering by the mission, with its interventionist strategy to create monogamous families based on marriage between partners of similar ages, and the unintended consequences of settling people down. At the time Hart made his census, over 90 per cent of the people were self-supporting and living almost exclusively by hunting and gathering for most of each year. There is every reason to suppose that the general structure and size of the population bore a direct relationship to the size and structure of the precolonial population.

The 20 years following Hart's census saw the population switch from hunting and gathering to becoming entirely settled. With the change went an increase in mortality and a decline of approximately 20 per cent in overall numbers. Missions, in particular, were poorly funded prior to the late 1960s and it is unsurprising that the move to a sedentary life brought with it increased ill health with its effects on mortality rates. Following the war, a hospital was established at the mission, which appears to have had a dramatic impact on the survival rate of infants, halving their mortality rate.

It is uncertain what caused the slowing down in the growth rate during the 1960s, but it seems that crowded living conditions and lack of adequate sanitation were the root cause. On the other hand, it is more certain what affected the growth in population from the end of the 1960s. It was only in 1968 that the Tiwi, like most other Aboriginal people in remote Australia, fully entered the cash economy.(22) All government social security payments from that year on were made in cash directly to Aboriginal people instead of to the mission or government superintendent on their behalf. Between 1968 and the mid-1970s, there was a real growth in Aboriginal incomes despite the fact that fewer and fewer of them were working. Increased income came to them from eligibility for a wider range of social security and other payments, ending up with access to unemployment benefits which had been long denied. Added to this was a dramatic increase in funding for Aboriginal communities in all spheres of life. Undoubtedly, housing and diet greatly improved for Aboriginal people once the Labor government was elected in 1972, as did the delivery of health care. By the end of the 1970s, however, average incomes among the Tiwi ceased growing and their per capita income plateaued at one-third of the Australian average (Stanley 1983, 33-35), locking people into dependency on the state, a situation that persists in the Northern Territory (Taylor and Roach 1998). Just what ongoing demographic impact this is having for the Tiwi population remains unclear.

While this paper has provided a preliminary sketch of the transformations that have taken place in the Tiwi population over the last 60 years, when it comes to explaining the figures much of the discussion here has of necessity consisted of informed conjecture. In order to provide firmer conclusions about the causes of demographic change, what is needed is a return to Hart's field methods of some 70 years ago. This would yield a data base of individuals as a foundation for longitudinal analysis of their demographic histories. This paper lays the framework for such a detailed study.

ACKNOWLEDGEMENTS

Peterson's interest in Tiwi demography arose out of work on band organisation (Peterson and Long 1986) and the hope that Hart's census would provide details of the composition of residential groups beyond what he had published. Unfortunately, it is probably not possible to relate his data to specific bush living groups. Shortly before Hart died, Peterson met him over a long and memorable lunch which was to be Hart's last scholarly exchange, during which there was an extensive discussion about Hart's work on the Tiwi including the topic of this paper. Following his death, Hart's colleague and friend, Dorothy K Billings, gave Peterson access to his papers. Jane Goodale provided a most useful review of the paper; and John Early and Eric Venbrux both commented helpfully on an early draft. Doreen Bowdery and Jin Liu provided valuable research assistance.

NOTES

A version of this paper was originally presented at the 7th International Conference on Hunting and Gathering Societies held in August 1993 in Moscow. Of the book written jointly with AR Pilling (1960), Hart wrote pages 9-95 and Pilling, pages 97-113 (Pilling 1962, 321).

(1.) Prior to 10,000 or so years ago Melville and Bathurst Islands would have been joined to the Australian mainland. As the sea levels rose they became islands 8,000 years or more ago, although somewhat larger than at present. As the sea level continued rising to a maximum of just over a metre above present levels around 6,000 years ago, the two islands would have been a little smaller than they are today. They would have reached their present size 3,0004,000 years ago. We are grateful to Kurt Lambeck of the Research School of Earth Sciences at the Australian National University for this information.

(2.) In 1975 when the Labor government lost power, the Coalition changed the policy to self-management. Although the two policies should be radically different, the actual differences between them are small at this stage. Until 1967, people legally defined as Aboriginal were not included in the national census figures although Aboriginal people have long been counted at census time. The reasons for this are complex but relate to the politics of federation, taxation, political representation and bureaucratic practice (see Smith 1980, ch 3). A comprehensive listing of the Aboriginal population of the Northern Territory, including the Tiwi, was published in 1957 and known as the `Register of Wards' (Northern Territory Government Gazette, no 19B, 13 May 1957).

(3.) Morris (1965) records accounts of raiding the mainland by Tiwi men who travelled via northwest Vernon Island. Pye (1977, 13-15) also records raids to the mainland to secure wives but this would seem to have been during the nineteenth century and so after the arrival of the dugout canoe. In 1953 six Melville Islanders were said to be descended from three captured mainland women. The maximum open-sea distance was only 16 km but it is an area of fast tidal flows. The distance between the eastern extremity of Melville Island and Cape Don on the Cobourg Peninsula is 27 km, and an even more difficult stretch of water, which seems to have been rarely crossed, although Sunter (1937, 35-36) records one such trip in the 1930s. He speaks of it as `daring seamanship', as indeed it was. The twelve-foot canoe with its five occupants and six inches of freeboard turned over midway through the voyage! The occupants were young men seeking tobacco.

(4.) Carbon 14 puts the dates for a number of so-called Macassan sites on the north coast of Arnhem Land back to around 1,000BP, but Macknight does not accept the association of these dates with Macassan remains. More recent archaeological work by Scott Mitchell (1994) on the Cobourg Peninsula supports Macknight's views.

(5.) The accounts of Joe Cooper's movements are conflicting, although all are in agreement that he left finally in 1916.

(6.) It is surprisingly difficult to establish, from the literature, exactly where these meeting points were but it seems that on the south coast the main one was on Buchanan Island in the mouth of Apsley Strait (Commonwealth of Australia 1937, 23). Although Goodale (1971, 11) reports that the northern base was on the northeast coast of Melville Island, this may be in error for the northwest coast of Melville Island near Garden Point which was a well-established area for contact in the 1930s as was Rocky Point on the northwest coast of Bathurst Island (Haultain 1971, 39-57). On the other hand, Goodale may have been referring to Sam Green's sawmilling operation, running from 1908 to 1918 in the northeastern area (see Pye 1977, 41). Japanese pearlers came from two sources: some were working on boats registered in Darwin and others were from boats originating in Japan.

(7.) This is evident from some of his manuscript writings.

(8.) In a stray typewritten sheet, Hart notes the following about four brothers of a single father but each with a separate mother: `In my time Tamboo was regarded as a Mingwila, Poodi as a Tiklauila, Domenico also a Tiklauila and Ki-in-kumi was the outstanding patriarch of the Malauila'.

(9.) Brandl recorded 20 in 1969 (1971, 74).

(10.) From page 16 of a manuscript entitled, `Chapter 1: Introduction'.

(11.) Fred Rose made very effective use of this technique in aging people on Groote Eylandt (see 1960, 42-51).

(12.) Ritchie, an agriculturalist who worked on the islands between 1930 and 1933, provides a detailed ecological description of Melville Island, making it clear that the southern portion of that island was much more hospitable than the north because it had ample freshwater and game. He describes it as `the real hunting-ground of the myall [unsophisticated/wild/uncivilised] blacks' (1934, 65). Jane Goodale comments that the Melville Islanders from the north consider that it is more bountiful than Bathurst Island, but it is generally the case that people see their own country as better than everybody elses regardless of how it looks to outsiders. Nevertheless, this means that our invocation of Ritchie's description of an `inhospitable environment' could be wrong and the low densities could be the result of the depredations wrought with Cooper's guns and of disease introduced by pearlers.

This description of the southern groups as the `myall blacks', while appearing to conflict with Hart's view that the northern groups of the Malauila and Munupula were the `uncontaminated' Aboriginal people, is not necessarily at odds with it. It may well be that, from a mission perspective, the use of `myall' for the south of Melville Island who were in regular contact with the Japanese referred not only to the fact that they were bush living but was also a moral judgement about their unacceptable drinking behaviour and sexual involvements with these outsiders.

(13.) A statement on page 6 of a manuscript on local organisation.

(14.) The methodology used by Fry (1949, 80) to arrive at the mortality figures was fraught with difficulties because he was unable to communicate directly with the people. He got them to notch sticks with different size notches for babies, children and adults, with separate sticks for males and females.

(15.) Rose (1987, 30) argues that Hart overestimated girls' age at marriage. However, the grounds for this suggestion are that the `traditional' system was corrupted by the time Hart did his fieldwork and that the figure of 14 years for the age of marriage would have been based on statements by informants. Rose's dichotomisation of history into traditional and non-traditional is a self-serving device to privilege certain sets of information, including his own. As is made clear here in the body of the text, there were diverse relationships between individual Tiwi and the mission. At least a quarter of the Tiwi population had, according to Hart, `little contact' with outsiders generally. Further, Gsell did not start his purchasing of girl children until 1921.

(16.) Venbrux comments (letter, September 1993) that, given the political importance of females in Tiwi society, female infanticide would be unlikely. He was told by Tiwi that, in the case of twins (a boy and girl without deformity and of equal strength), it was more likely that the boy would be killed. Women said it would be painful to breastfeed twins and, besides, that was something for dogs and not for humans. The practice of disposing of one of a pair of twins seems to have been abandoned in the 1950s. Senilicide was practised (see Hart et al 1988, 154).

(17.) It seems, however, that this may be due to the presence of a white saw miller, Sam Green, in the area from 1909 to 1918 (see Pye 1977, 27; Venbrux 1993, 45, and letter, September 1993).

(18.) This is the age chosen by Hart for the construction of this table. The convention in American demographic work is to take the age of 45 and in UN writings the age is 50. It is not possible to reconstruct a different table from the data.

(19.) It might just be possible to obtain this data after extended cross-correlation and analysis of Hart's data but it would be a huge task.

(20.) It would appear that these figures included Tiwi who were temporarily absent on the mainland (Jones 1963, 24).

(21.) Between 1952 and 1972, total fertility rates among the Tiwi averaged 5.2 and rose from 4.0 between 1952 and 1956 to 5.9 between 1962 and 1966 (Gray 1985, 25). Total fertility rate is an expression of the number of births that would occur to a woman who has experienced a particular set of age-specific fertility rates as she has passed through the reproductive period.

(22.) Wages at the beginning of 1969 were $4.50 a week but in August a major increase resulted when the government introduced a Training Allowance Scheme which resulted in men being paid $25-$36 a week and women being paid $19-$27 a week (Stanley 1983, 10). This was on top of social security recipients having their payments made to them in cash.

REFERENCES

Australian Bureau of Statistics 1996a Births, catalogue no 3301.0, ABS, Canberra.

-- 1996b Deaths, catalogue no 3302.0, ABS, Canberra.

Basedow, H. 1913 Notes on the Natives of Bathurst Island, North Australia, Journal of the Royal Anthropological Institute of Great Britain and Ireland 43, 291-323.

Brandl, M. 1971 Pukumani: The Social Context of Bereavement in a North Australia Aboriginal Tribe, PhD thesis, University of Western Australia.

Commonwealth of Australia 1925 Report of the Administrator of the Northern Territory, Government Printer, Melbourne.

-- 1937 Report of the Administrator of the Northern Territory for the Year Ending 30th June, 1937, Government Printer, Melbourne.

-- 1938-39 Report of the Administrator of the Northern Territory for the Year Ending 1938-39, Government Printer, Melbourne.

-- 1946 Report of the Administrator of the Northern Territory for the Year Ending 30th June 1946, Government Printer, Melbourne.

Evans, N. 1995 A Grammar of Kayardild with Historical-Comparative Notes on Tangkic, Mouton de Gruyter, Berlin.

Forrest, P. 1995 The Tiwi Meet the Dutch: The First European Contacts, Tiwi Land Council, Darwin.

Fry, H.K. 1949 A Bathurst Island Mourning Rite, Mankind 4(2), 79-80.

Goodale, J. 1971 Tiwi Wives: A Study of the Women of Melville Island, North Australia, University of Washington Press, Seattle.

Grau, A. 1983 Dreaming, Dancing, Kinship: The Study of Yoi, the Dance of the Tiwi of Melville and Bathurst Islands, North Australia, PhD thesis, Queen's University, Belfast.

Gray, A. 1985 Limits for Demographic Parameters of Aboriginal Populations in the Past, Australian Aboriginal Studies 1, 22-27.

-- 1990 Aboriginal Fertility: Trends and Prospects, Journal of the Australian Population Association 7(1), 57-78.

Gray, A. and H. Tesfaghiorghis 1992 Social Indicators of the Aboriginal Population of Australia, CAEPR Discussion Paper 18, Centre for Aboriginal Economic Policy Research, Australian National University, Canberra.

-- 1993 Aboriginal population prospects, Journal of the Australian Population Association 10(2), 81-101.

Gsell, EX. 1956 The Bishop with 150 Wives: Fifty Years as a Missionary, Angus & Robertson, Sydney.

Hart, C.W.M. 1930 The Tiwi of Melville and Bathurst Islands, Oceania 5, 167-80.

-- 1931 Personal Names among the Tiwi, Oceania 1,280-90.

Hart, C.W.M. and A.R. Pilling 1960 The Tiwi of North Australia, Holt, Rinehart & Winston, New York.

Hart, C.W.M., A.R. Pilling and J.C. Goodale 1988 The Tiwi of North Australia (3rd edn), Holt, Rinehart & Winston, New York.

Haultain, C.T.G. 1971 Watch off Arnhem Land, Roebuck Press, Canberra.

Hingston, J.P. 1938 Exploration of Melville Island [1877], Royal Australian Historical Society Journal and Proceedings 24(2), 157-63.

Jones, F.L. 1963 A Demographic Survey of the Aboriginal Population of the Northern Territory, with Special Reference to Bathurst Island Mission, Australian Institute of Aboriginal Studies, Canberra.

Krastins, V. 1972 The Tiwi: A Culture Contact History of the Australian Aborigines on Bathurst and Melville Islands 1705-1942, BA (Hons) thesis, Australian National University.

Macknight, C. 1976 The Voyage to Marege: Macassan Trepangers in Northern Australia, Melbourne University Press, Melbourne.

Martin, D. and J. Taylor 1996 Ethnographic Perspectives on the Enumeration of Aboriginal People in Remote Australia, Journal of the Australian Population Association 13(1), 17-33.

Meehan, B. 1982 Shell Bed to Shell Midden, Australian Institute of Aboriginal and Torres Strait Islander Studies, Canberra.

Mitchell, S. 1994 Culture Contact and Indigenous Economies on the Cobourg Peninsula, Northwestern Arnhem Land, PhD thesis, Northern Territory University.

Morris, A.J. 1965 A Brief Study of Tiwi-Larrakia Feuds, Historical Society of the Northern Territory of Australia, Darwin, 1-4.

Packer, A.D. 1962 The Health of the Australian Native, Oceania 32, 60-70.

Peterson, N. and J. Long 1986 Aboriginal Territorial Organization: A Band Perspective, Oceania Monograph 30, Sydney.

Pilling, A.R. 1958 Law and Feud in an Aboriginal Society of North Australia, PhD thesis, University of California.

-- 1962 A Historical versus a Non-historical Approach to Social Change and Continuity among the Tiwi, Oceania 32(4), 321-26.

Pye, J. 1977 The Tiwi Islands, Colemans Printing, Darwin.

Ritchie, P. 1934 North of the Never Never, Angus & Robertson, Sydney.

Rose, F. 1960 Classification of Kin, Age Structure and Marriage amongst the Groote Eylandt Aborigines: A Study in Method and a Theory of Australian Kinship, Akademie-Verlag, Berlin.

-- 1987 The Traditional Mode of Production of the Australian Aborigines, Angus & Robertson, Sydney.

Searcy, A. 1909 In Australian Tropics (2nd edn), Robertson & Co, London.

Sharp, L. 1940 An Australian Aboriginal Population, Human Biology 12, 483-507.

Smith, L. 1980 The Aboriginal Population of Australia, Australian National University Press, Canberra.

Stanley, O. 1983 An Aboriginal Economy: Nguiu, Northern Territory, North Australia Research Unit, Darwin.

Sunter, G. 1937 Adventures of a Trepang Fisher: A Record without Romance, Hurst & Blackett, London.

Taylor, J. and L. Roach 1998 The Relative Economic Status of Indigenous People in the Northern Territory, 1991-1996, CAEPR Discussion Paper 152, Centre for Aboriginal Economic Policy Research, Australian National University, Canberra.

Thomson, N. 1983 Aboriginal Infant Mortality, 1976-1981, Australian Aboriginal Studies 1, 10-15.

Tindale, N. 1962a Geographical Knowledge of the Kaiadilt People of Bentinck Island, Queensland, Records of the South Australian Museum 14(2), 259-96.

-- 1962b Some Population Changes among the Kaiadilt People of Bentinck Island, Queensland, Records of the South Australian Museum 14(2), 297-336.

Tiwi Land Council 1992 Tiwi Land Council: Thirteenth Annual Report, Tiwi Land Council, Darwin.

Venbrux, E. 1993 Under the Mango Tree: A Case of Homicide in an Australian Aboriginal Society, Catholic University of Nijmegen, Nijmegen.

Warner, L. 1958 A Black Civilization: A Social Study of an Australian Tribe, Harper, New York.

Nicolas Peterson

Nicolas Peterson is a Reader in the Department of Archaeology and Anthropology, Australian National University, Canberra.

John Taylor

John Taylor is a Fellow in the Centre for Aboriginal Economic Policy Research, Australian National University, Canberra.