^[1] CHINS provides an estimate of the service population rather than a census-style head count. CHINS estimates invariably, therefore, produce larger populations than does the census enumeration.

^[2] It may have been a deliberate strategy to give E4 this house, but unfortunately it did not occur to me to check this. At the time the dwellings were apportioned I was not aware of the potential problem.

^[3] This process sometimes presented a challenge to my position as a neutral 'observer': it was hard for the enumerators not to cast me as a 'helper'. From the time of the 'training' session onwards I was appealed to intermittently for clarification on particular questions. I resisted giving advice as best I could (by pleading ignorance, and saying they should ask the CFO or the CC), except in the area of the translation of kinship terms. Here I simply followed where they led: they were bent on devising English 'translations' for local kinship terms that have no English equivalents, rather than classifying as non-kin people whom they regard as kin.

^[4] Daly and Smith's discussion of the interview techniques devised for community case studies at Kuranda and Yuendumu is pertinent here: 'The social pool of people contributing as additional de facto "respondents" to each questionnaire were included in the interview process…This approach…is positively oriented to socially embedded and constructed Indigenous modes of communication where the individual cannot be effectively "quarantined" for the purposes of eliciting information' (2000: 20). The enumerators at community A were, in effect, applying the same technique.

^[5] The form of Q. 2, which is not a direct question (Sex:) betrays the presence of a topic that the compiler of the questionnaire (presumably a member of the Euro-Australian mainstream), found inappropriate as the subject of a direct question. None of us is free of cultural taboos.

^[6] From day two onwards, E1 took a calculator with him. When someone knew their year of birth, this was subtracted from 2001, and the answer was recorded as the person's age.

^[7] Most local marriages would be classified as de facto relationships in mainstream terms, but it would be quite inappropriate, in the view of local Indigenous people, to classify them thus. The category 'de facto' occurs neither on the SIPF nor on the mainstream SIHF, and perhaps the injunction '"Married" refers to registered marriages', which does not appear on the SIPF, should also be removed from the mainstream HF.

^[8] The word Dreaming is not used in this area, but the word translated here as 'sacred ancestral inheritance' corresponds more or less to what mainstream Australia understands by that concept.

^[9] In contrast to the situation in some other parts of Australia (see David Martin, this volume), local people in this area attach no cultural value to fluency in English. Being able to speak English is seen as a useful skill, but inability to do so is not valued negatively.

^[10] The actual amount of money that people get in their account each fortnight varies, because the HA deducts money from people's CDEP payments to cover their outstanding bills before crediting the remainder to their account.

^[11] The fourth box in Q. 29 refers to 'sorry business' as a possible reason for being off work. People in this area do not use this phrase to describe funerals, and I suspect that there are other places in Australia where it is also an unfamiliar usage. It would be better to use a standard English term such as 'funeral'.

^[12] I was told this by one of the absent 'usual residents'.

^[13] The 2001 Census Dictionary (ABS 2001) defines a person's usual residence as the place where the person 'has lived or intends to live for a total of six months or more'.

^[14] When I returned to community A in June 2002, at least one tent which had been erected just after the census enumeration in 2001 was still in place outside its associated dwelling. It had acquired a protective awning consisting of a large tarpaulin on a bush timber frame.

^[15] This is not to say that people do not distinguish between their actual mother (and other 'close' M such as their mother's sisters) and other M, in terms of sentiment and behaviour.

^[16] In the great-grandchild's generation, female children are merged under the terms M or FZ, depending on the patrilineage to which they belong.

^[17] As mentioned previously, E1 and I had several conversations about how to translate (Z)DCFZC. On one occasion he reasoned that the reciprocal term MBCMM means 'mother-in-law's mother', so in English its reciprocal term should be 'mother-in-laws' son's son-in-law'. This actually works for a MBCMM and a male DCFZC, but whereas E1 made the computation instantly in his head, I had to go away and puzzle over a kinship chart for some time to confirm it for myself. This is not necessarily because E1 is smarter than I am (although that may well be true). Rather, a 'first' kinship system is like a first language: the individual 'born into' the system carries its grammar unconsciously in their head. A 'second' kinship system is like a second language: the rules must be learned consciously and laboriously, and full fluency is extremely hard to achieve.

^[18] Many other features of the kinship terminology support the thesis that a major organising principle of the system is the relationship between lineages over time. These will be discussed in another paper.

^[19] Diane Smith (pers. comm.) suggests an alternative strategy: getting the enumerators to draw up genealogies for each dwelling. In this way, household structures would be recoverable in finer detail. However, such a strategy would require extra training for the enumerators, would add to the time that it takes to complete the SIHF, and would be easily implemented only in communities where the enumerators were well known to the interviewees, and where the enumerators themselves were familiar with the details of how everyone was related to each other. Moreover, a small proportion of local people are in 'wrong' marriages, or are the children of such unions, and such information has to be treated carefully. While everyone in a community is aware of such facts, it is not usual or acceptable to discuss them openly in public.

^[20] Roger Jones (pers. comm.) doubts whether this would produce data that was any more accurate. However, an 'age-range' as opposed to an 'exact age' question would be more in tune with how local people view the question of age, and the results would certainly be no less accurate.

^[21] The local language distinguishes six seasons in the year; 'dry season' is not one of them. For most locals the year of the last census would not be memorable for that particular reason.