In the Anglo-Celtic system, the term cousin is used to refer to the children of a person's (ego's) father's sisters and brothers, and of their mother's sisters and brothers. In other words, the system merges under the term cousin all the children of ego's parent's siblings. Note also that the Anglo-Celtic term cousin is neutral with respect to sex. Anglo-Celts call the children of their own father and mother either brother or sister, depending on their sex. The local Indigenous system is completely different. Local people call the children of their mother's brother MBC and the children of their father's sister by a different term, FZC. A MBC may also simultaneously be a MMBDC, and the same term is applied to both categories of relative. Like cousin, these terms are neutral with respect to sex. Put another- 56 - way, the system distinguishes two kinds of cross-cousins: matrilateral (MBC) and patrilateral (FZC). Local people call the children of their mother and her sisters and of their father and his brothers B if they are male and Z if they are female. In other words, in the local system siblings and parallel cousins are merged. Like the Anglo-Celtic terms for siblings, these too are differentiated by sex. Fig. 3.3 illustrates the differences between the two systems for the set of terms for siblings and cross and parallel cousins.

People raised in the Anglo-Celtic system think of son and daughter as 'natural' categories. Children are defined, as it were, with respect to their parents' marriage: both parents use the same terms for their offspring. The children of ego's brothers and sisters are merged under the term nephew for males and niece for females. The local Indigenous system operates according to a different set of principles, which appear just as 'natural' to local people. A woman calls her own children and those of her sisters ZC, and those of her brothers by a different term, BC. A man calls his own children and those of his brothers BC, and those of his sisters ZC. Children are here being defined not with respect to their parents' marriage, but with respect to their lineage: BC means 'child of my patriline' and ZC means 'child of my matriline'. These terms for children (as with Anglo-Celtic cousin), are not differentiated according to sex. Fig. 3.4 illustrates the differences between these sets of terms in the two systems.

- 57 -

All kinship systems have terms that are 'classificatory' in the sense that they classify people together according to a set of underlying structural principles. The term cousin is probably the most classificatory of the Anglo-Celtic kinship terms (although uncle, aunt, grandparent, and the category in-law are quite complex as well). But the local Indigenous system applies more (and more abstract) principles of classification than does the Anglo-Celtic system. For example, males in the generation above ego in ego's patriline (including ego's own father) are F, and all females in the generation above ego in ego's mother's patriline (including ego's own mother) are M. But these terms have even wider application: they apply also to kin in generations other than the parental generation.

- 58 -

If one thinks of ego's own generation and the generations of their grandparents and grandchildren as 'even' and ego's children's, parents' and great grandchildren's generations as 'odd', the term M applies to any female member of ego's mother's patriline who is in an 'odd' descending generation with respect to ego (see Fig. 3.5). No M is any more or less of an M than any other, just as no cousin is any more or less of a cousin in the Anglo-Celtic system.^[15] People have been told that M means 'mother', and so that is how M was often translated by the enumerators, whether the person was, in Anglo-Celtic terms, a mother, a daughter-in-law, a nephew's wife, a great granddaughter-in law, or a brother's great granddaughter-in-law. The last two, significantly, are scarcely kin terms at all in the Anglo-Celtic system. Sometimes the enumerators, realising that some categories of M do not count as mother in the Anglo-Celtic system, attempted to substitute the 'correct' Anglo-Celtic term—with varying degrees of success, as we shall see in the case of one dwelling (see Fig. 3.7 and the accompanying discussion). None of the Anglo-Celtic terms for the kin comprising the 'nuclear family' are directly translatable into local Indigenous kinship terms. And vice versa: none of the core terms, let alone the non-core terms, of the local Indigenous system are directly translatable into Anglo-Celtic kinship terms. The Anglo-Celtic reader, who is probably suffering from 'kinship fatigue' at this point, should bear in mind the plight of the Indigenous enumerator, who is faced with exactly the same problem (seen from the other side), in a real-time interviewing situation.

The local Indigenous and Anglo-Celtic systems differ in another very important way. In the Anglo-Celtic system (as it operates today), people are rarely kin to one another before they get married. A marriage brings together, in a set of in-law relationships, two previously unrelated kindreds. Their only point of intersection is the married couple—the husband and wife—and the connection is then carried down into the couple's descendants.

- 59 -

In the local Indigenous system, the preferred marriage is between people who are already in a kinship relationship: a man marries his (actual or classificatory) matrilateral cross-cousin: his MBD. A woman thus marries her (actual or classificatory) FZS. Marriage in the Indigenous system does not create bonds of kinship: it reinforces and reaffirms already existing kin relationships. So a man's MBD may be his wife, but she is also a kind of cousin in the Anglo-Celtic system, and a man's male ZC (actual or classificatory sister's son) is his nephew in Anglo-Celtic terms, but may also in addition be his daughter's husband, or son-in-law. Thus it is perfectly possible for a woman with no daughters and an unmarried man to refer to each other as FZDC 'son-in-law' and MMBD 'mother-in-law' respectively, as happened on one SIHF and the related SIPFs.

The Indigenous kinship terminology encompasses categories of people on which the Anglo-Celtic system is silent. It distinguishes and covers seven patrilines that are related matrilineally through the marriage system. The Anglo-Celtic kinship terminology focuses on the individual and their direct ancestors and descendants, and merges patrilineal and matrilineal kin at every level. The system fades off very quickly into cousins and then non-kin as soon as it leaves the realm of ego's nuclear families of origin and procreation. In the Anglo-Celtic system there is no term for (Z)DDFZC, who is the person (or the sister of the person) who potentially marries your (Z)DD (daughters' daughter from a female point of view, and sister's daughter's daughter from a male point of view). E1 and I had several conversations about how such a person should be characterised in English. The option of classifying a (Z)DDFZC as a 'friend', or as 'unrelated', did not enter the frame.

For Q. 4 the enumerators were often successful in assigning the 'correct' term—in Anglo-Celtic terms—with core kin. In at least one case, a father's brother, who in local terms is another F ('father') was put down as 'uncle', and the use of the terms 'nephew' and 'niece' corresponded to Anglo-Celtic usage. The enumerators' own superficial knowledge of the Anglo-Celtic system and the training provided by the CFO thus had some effect. But it did not penetrate very far into the system.

It should now be clear that in the context of local Indigenous kinship, Q. 5 is unanswerable. Even if one is operating according to the Anglo-Celtic kinship system it is difficult to know how one might answer it. Who is closer to ego: father or mother? In a three-generation household, who is closer to ego: parent or child? Is the relationship to a spouse closer than to a child? Is the relationship to a sibling closer that to a child? What precisely is meant by 'closely related'? Is it a question about biology, or a question based on unexamined assumptions about what constitutes closeness in terms of a particular system of kinship? It is not surprising that this question was left blank on the majority of SIPFs. E1 commented at one point: 'this is a real [white man's] question', by which he meant that it made no sense in local Indigenous terms.

Anglo-Celtic kinship, then, is not the ideal idiom for attempting to elucidate the structure of local Indigenous households. But what of the implicit model of the household that lies behind the census questions? The definition of the household given in the 2001 Census Dictionary (ABS 2001) allows for the possibility of more than one 'household' in a dwelling,- 60 - but not for households whose membership is spread across more than one dwelling. Two major types of 'household' are identified: those whose members are 'related' (family-based households), and those whose members are 'unrelated' (group households).

In the ABS definition of the family, 'the basis of a family is formed by identifying the presence of either a couple relationship, lone parent-child relationship, or other blood relationship … other related individuals (brothers, sisters, aunts, uncles) may be present in the household' (ABS 2001; emphasis added). Although the ABS does not use the term 'nuclear family' it is clear from the definition that this is what is meant by a 'family', since other 'related individuals' may be associated with it, but are not part of it. Daly and Smith provide a succinct account of how the ABS categorise the data obtained from 'multi-family' households: '[T]he nuclear family of parents and children is taken as the base around which all family types are constructed, and other families within the household are placed in relation to this "primary" family … if there are more than three family types in a household, the adults in any additional families are "disbanded" as a family type, reclassified as related individuals, and assigned to the "primary" family' (2000: 12).

The model does not fit the Indigenous facts on the ground. Indigenous households 'who make common provision for food or other essentials' (ABS 2001) are often spread across more than one dwelling (for example in the case of a man and his co-wives, as discussed elsewhere in this paper). Moreover, previous ethnographic work on Indigenous household structures led Daly and Smith to conclude that: 'the nuclear family is not the most common residential form … indigenous households in the 1990s were characterised by considerable compositional complexity, porous social boundaries and large size' (1999: 2). These generalisations continue to hold true for the households at community A in 2001.

Table 3.1 shows the 'usual residents' of dwelling J as they were listed on the SIHF for the dwelling. The local Indigenous kinship term by which person 1 actually addresses each person is given in the last column. Superficially, this looks like a four generation family, consisting of person 1 and his wife, their daughter and her husband (who is also person 1's sister's son, and hence nephew), two of the daughter's children, and person 1's wife's mother. A perfectly 'normal' pair of related nuclear families, plus one mother-in-law, one might think, although the fact that person 1's wife and his daughter appear to be the same age gives grounds for suspicion that all is not as it seems on the surface.

And indeed it is not, as Fig. 3.6 shows. Person 2 is indeed person 1's current wife (and also, as it happens, his actual MMBDD), and person 3 is indeed her mother. But person 4 is the daughter of person 1 and his deceased first wife, and the two 'grandchildren' far from being person 4's children, are people who would not even be classified as kin in the Anglo-Celtic system. They are the great grandchildren of person 1's mother-in-law's deceased husband's other wife. In the local Indigenous system, these children are considered kin to person 1. Their mother is his classificatory BC ('daughter'). He looked after her when she was a child, and now her children live in his household.

- 61 - - 62 -

Another possible scenario in cases where a dwelling contains people who are kin in the Indigenous system, but not according to the Anglo-Celtic system, is demonstrated by the case of dwelling K. A partial genealogy for this dwelling is given in Fig. 3.7. This dwelling had 11 'usual residents'. In one case the Anglo-Celtic kin term entered in response to Q. 4 differed from the one entered on the SIHF: person 5v (a woman) was put down as 'uncle' on the SIPF and as 'daughter' on the SIHF. Person 10 (an 8-year-old boy) was put down as 'sister's son-in-law'. Person 8 was put down as 'brother-in-law' and his wife, person 9, as 'granddaughter'. Person 6 was put down as 'great-granddaughter'. In terms of the kinship diagram, and the Anglo-Celtic system, these responses seem to be not only wrong, but also incomprehensible. However, if the Indigenous system is taken into account, sense of a kind emerges. These answers represent an attempt, from a local Indigenous viewpoint, to translate from the Indigenous system to the Anglo-Celtic.

In the Indigenous system, person 5v is person 1's M. The enumerators realised, however, that this person would not be classified as mother in the Anglo-Celtic system. On the SIHF, in attempting to solve the problem, the enumerators seem to have inverted the relationship: 'daughter' is probably a translation of ZC, which is what person 5v calls person 1. On the SIPF another solution was adopted: 'uncle' is the English term that locals most often use to translate MB (mother's brother). Person 6 is another M. Again, the enumerators realised that mother was not the appropriate Anglo-Celtic term. They aimed for an Anglo-Celtic term that picks out an M of a lower generation—but overshot by two generations.^[16]

- 63 -

'Sister's son-in-law' (as person 10 was described on his SIPF) was an attempt to translate the term (Z)DCFZC. The enumerators nearly got it right, but again they missed a generation. One possible rendering of (Z)DCFZC is 'sister's daughter's (potential or actual) son-in-law'.^[17] Person 11 is person 1's sister's daughter. Person 11's as yet non-existent daughter would be person 10's MMBDD, and thus also his MBD, or potential wife.

In the case of person 9, the enumerators, perhaps suffering themselves from kinship fatigue at this point, attempted to opt for a 'straight' translation between the two systems. Person 8 is person 1's wife's brother, and so like her is person 1's MBC. This was 'correctly' translated as 'brother-in-law'. In the local system, the wife of a male MBC (i.e. person 9 in this case) is MM—a type of 'grandmother'.

Every single enumeration that I observed produced results of the kind described above. The census data, if coupled with the ethnographic data, offer a fascinating insight into the local Indigenous kinship system and principles of household formation, and into how local people think about and abstract principles from their kinship categories. But as raw data on household structure they are unusable, for two reasons. Firstly, the incommensurability of the two kinship systems results in 'relationship' data that reflects neither system, and which cannot be used to construct 'families' within households. Secondly, the implicit model upon which ABS household structures are predicated—the nuclear family—is a bad model for local Indigenous households in particular and, it could be argued, for Indigenous households in general.

The problem lies in the assumption that the nuclear family is a 'natural' structure that forms the basic social and coresidential unit in all cultures. Anglo-Celtic cultures tend to take the nuclear family as the 'norm', and to describe all other household types as variations on, or deviations from that norm. The ABS, as an institution of the Australian Anglo-Celtic mainstream, reflects that tendency in its definitions. The Anglo-Celtic kinship system, with its unique reciprocal terms for the members of the nuclear family, reinforces the view of the nuclear family as somehow 'natural'. In Fig. 3.8, each interior box surrounds an ego. The terms within the box are those by which other members of the Anglo-Celtic nuclear family address ego. The nuclear family and its constellation of relationships only comes into being with a marriage: any ego is likely to be a member of more than one nuclear family in their lifetime, first as a child (family of origin) and then as a parent (family of procreation).

- 64 -

The local Indigenous kinship system, in contrast, privileges lineages, not nuclear families (Fig. 3.9). Ego and ego's siblings are not primarily constituents of a nuclear family, but a point of intersection between an already existing patrilineage and matrilineage.

- 65 -

The box in Fig. 3.9 does not represent an individual ego, but rather a contains a set of relationships that are constituted by the intersection of a patrilineage and a matrilineage in a particular generation. These relationships exist independent of any particular marriage because the FZC–MBC relationship between two people exists before a marriage does, and every person has many FZC and MBC. In Fig. 3.9 it is simply impossible to draw a box around a set of reciprocal terms that apply exclusively within a 'nuclear family'.^[18] The siblings in the bottom box are BC with respect to their patrilineal parent, and ZC with respect to their matrilineal parent. The terms for sibling (B and Z) lie within the intersection of the two lineages that is constituted by the marriage.

My first thoughts, as I sat observing, were that with more (and more theoretically informed) training it would be possible for local Indigenous people to produce results that would be interpretable in mainstream terms, at least for categories of kin that the mainstream system recognises as kin. Thinking in abstract terms about kinship is an everyday part of local life, and it would simply be a question of working out a systematic set of translation principles. But as I thought about it more, I began to question the premise upon which such a course of action would be based. What about kin who do not fit Anglo-Celtic categories? Would they be described as 'friends', or 'unrelated'—even if, in many cases they are addressed by the same kinship term as people considered to be kin in the Anglo-Celtic system? Would such a solution actually reflect the reality of the composition and dynamic of Indigenous households? It would not. It would simply represent an attempt to distort the Indigenous system by squeezing it into the mainstream Anglo-Celtic mould. And this particular local system is only one of many systems of kinship in Indigenous Australia. The same process would have to be implemented Australia-wide. This would be expensive, time-consuming, and logistically complex, if not impossible.

The better (and cheaper) course is to go back to first principles: for whom and for what purpose are these data on household composition being collected? The important generalisations that are being sought for comparative purposes concern the size of households in relation to the size of dwellings, their age and sex profiles, and whether or not people consider themselves to be related. It is not necessary to model this information in terms of nuclear family structure. Indigenous people tend to live in large multi-generational households, in which everyone is considered kin to everyone else: that is, in extended family households. This has been amply demonstrated in the literature (for recent and relevant data see Henry & Daly 2001; Musharbash 2001; Smith 2000a). The census cannot hope to uncover the difference and complexity of the local Indigenous kinship system, or any other Indigenous kinship system, in two questions, and it should not be attempting to do so. The number of persons in the household and the range of people's ages, would be enough to indicate whether or not Indigenous people are living in households that approach or diverge from mainstream norms.^[19] If more fined-grained detail about particular sectors of the population is considered important, then those specific subgroups of the population could be targeted. For example, if ascertaining whether children are living in the same households as their 'real' parent(s) is considered- 66 - important, then there could be questions targeted at those under 15 years old (and these could appear later in the form, after the question about age), asking whether their 'real' or 'actual' mother and/or father live in the dwelling.

Removing the emphasis on kinship would serve another valuable end. The case of the 'missing persons 1' highlights the central importance of kinship in local Indigenous life. To give it prime place on the census forms is to give people the impression that it is more important than anything else on the form, whereas, in fact, it is simply being used as a means to model certain kinds of demographic data.