During the day, most people sit outside when they are at home, so the enumerators simply walked up to each dwelling and announced to the assembled company that they were here to fill in the census. Most people already knew what they were doing, and very few raised any kind of objection, or asked for further information about the census. People who evinced shyness or reluctance (these were mainly youths or young adults of both sexes) were cajoled by their assembled relatives into cooperating. Here, the fact that the enumerators were well known to everyone worked to their advantage, and to the advantage of the enumeration process. People largely took it on trust that the census was a good thing, and saw their cooperation in personalised terms: they were helping their enumerator kin to do their job properly.

Everybody found the process of the interview a bit strange: they were Indigenous actors in a non-Indigenous scenario. Although everyone would normally speak to one another in the local language, the enumerators initially asked the questions in English, and the respondents replied likewise. However, if the meaning of a question was unclear, or the answer not straightforward, people usually switched to the local language for the ensuing discussion. The enumerators were asking kinds of questions that local people rarely ask one another, either because both parties already know the answer (name, sex, kinship relationships, marriage status, place of residence, origin and ancestry), or because they are about matters that are not of central concern (age, level of education, source of income, nature of paid work, hours worked). Humour was the main device used by both the enumerators and the interviewees to reduce the awkwardness of the occasion. There were many jocular false answers, and jokes at the expense of white people for wanting to know these things.

On the first day, E1 and E6 quickly settled into a standardised routine, that varied only slightly according to circumstance. Each person at the dwelling who was on the SIHF was asked to come and sit in turn with the enumerators. E1 asked the questions, while E6 filled in the form. In general, E1 also supplied the answers, verbally, to the questions for which he thought he knew the answers, and on occasion the interviewee would correct him, or a discussion would ensue before the question was finally answered. He directly questioned the interviewee only when he was uncertain of the answer.

Once everyone who was present had been interviewed, the household in general was asked about those who were absent. If a person was somewhere in the general area, but not- 41 - actually at home, they were kept as candidates for SIPFs and noted down (mentally) for a later visit. In some cases it was discovered that people who had been marked as 'no' on the SIHF, because the enumerators had thought they were away, were in fact in the community. Their status was changed to 'yes' on the SIHF, and if they were actually present their SIPF was done. If a person who had been marked as 'yes' was actually away (at community B for example), their status on the SIHF was changed to 'no'. Significantly, the enumerators did not ask if there was anyone else who should be on the SIHF, and no one at the dwellings volunteered extra names, at least not in the cases I observed. Thus, if the enumerators had missed off any 'usual residents' when they compiled the SIHFs, the chances were that those people would be missed, unless they were actually present at the time of the interviews.

If there was anyone present at a dwelling who was on the SIHF of another dwelling, the enumerators also took the opportunity to do a SIPF for them, if they had not already been interviewed, on the principle of 'a bird in the hand'. This was a potential source of error: SIPFs could have got into the wrong envelopes, and the possibility of double counting also arose. Such errors did occasionally occur, but, as we have seen, they were picked up and corrected during the checking sessions at the school.

The interviewers were clearly committed to completing every SIPF in the presence of the interviewee. But had they asked everyone to respond to each question, the sessions would have dragged on interminably. In most cases their solution to this problem speeded up the interviews while not seriously compromising the quality of the data. E1 had a clear idea about which questions required answers from the interviewees and which questions he knew the answers to. Because the forms were filled out in the presence of the interviewee, that person had the opportunity to correct E1 if he got something wrong. That certain categories of people were less likely to do so was a possible source of bias: men and senior women were more likely to interject than were children and young women. Every interview took place within earshot of other household members, and there were often interjections from them in such cases.^[4] It could be argued, then, that the SIPFs of young women and children reflected less of their own knowledge about themselves and more of the opinions of the enumerator and more senior family members.

The general atmosphere of cooperation was weakened or broke down with two kinds of respondents: visitors from other communities and women from community C families, whether they were visitors or 'usual residents'. Although all visitors were kin to someone in the community, they did not necessarily know the enumerators very well. The community D visitors were cooperative when 'caught', but elusive: hunting and fishing were much higher on their list of priorities than filling in the census. The community E visitors, as we have seen, showed no desire to cooperate: the one who was 'caught' at his brother's house at first tried to evade the process ('We're only passing through, we don't live here'), but eventually gave in to pressure from the enumerators and his brother's family.

For the community C women, especially the younger ones, the whole process was just too culturally inappropriate. They refused to answer out loud, retreated physically into the dwelling, or refused to come out to be interviewed. The enumerators adapted their interview strategy accordingly. Only female enumerators attempted to conduct these interviews, and they took place inside the house, or off to one side, in lowered voices. When all else failed the forms for the women were filled in later by the enumerators on the basis of their knowledge of the individuals concerned.

Other circumstances sometimes forced variations on the theme. One of E1's MMBD (the sister of his actual MMBD, or 'mother-in-law') lives at one dwelling. In local society a FZDC ('son-in-law') and the women he calls MMBD, whether actual or classificatory, must avoid physical proximity and eye contact, and may only converse indirectly through an intermediary. The fact of working in pairs, as the CFO had correctly predicted, rendered this unproblematic. The CFO was also correct in his estimation that it was best to leave the handling of such situations to the enumerators themselves: avoidance behaviour is an everyday part of local life, and there are strategies already in place to accommodate it. It is not uncommon for a MMBD and her male FZDC to be members of the same household, since elderly mothers often live with one of their daughters. E1's MMBD simply sat where she was. E1 had positioned himself at some distance from her and out of her direct line of sight, and E6 had sat down between the two of them. Interestingly, E1 and E6 still maintained their roles. E1 asked the questions in a quiet voice. E6 relayed the question to E1's MMBD, E1 then answered the question (or the MMBD provided the answer herself), the answer was relayed by E6, and then E6 wrote the answer down if there was no dissent from either of the other two parties. E1 was unable to ask Q. 2 and Q. 4–6 (the questions on sex, kinship, and marriage), because a FZDC may not speak about, or hear anyone else speak about such matters in regard to his MMBD. E6 asked these questions herself, in a whisper, or simply wrote the answer down without asking. When it came to Q. 39, E1 tried to get E6 to ask it, but she was hesitant. So he asked the question quietly and she then repeated it word for word to the MMBD.

Although the enumerators tried always to do one SIPF at a time, there were times when they departed (or were forced to depart) from this principle. In the case of small children, where it was a parent or another relative who was helping with the SIPF, the enumerators often took one form each and worked on them simultaneously. This was not because children were considered less important, but rather because the enumerators quickly realised that filling in a form for a person under 15 years of age was more straightforward since several questions did not apply to them.

At two of the dwellings I visited, the enumerators were compelled to fill in more than one adult form at a time, or even to hand over the SIPF to the interviewee and allow them to fill it in for themselves. This was because the interviewees were about to go off hunting or fishing, and were anxious to be on their way. In one case the interviewees were some of the community D visitors, and in the other they were a group of local young men. The enumerators judged (almost certainly correctly), that people would vote with their feet if the interview process took too long, and so they resorted to this procedure. These SIPFs- 43 - were subsequently reviewed during the checking sessions, and in some cases added to or amended in the light of the enumerators' personal knowledge of the interviewees.

I now consider briefly the responses to each of the questions on the SIPF, roughly in the order in which they appear on the form. The full versions of the questions on the form are found in Appendix C.

Because the SIHFs had been filled in previously, the interview always began at Q. 1 of the SIPF. The enumerators did not follow the instruction to copy the answers to Q. 1–4 from the SIHF, but instead used Q. 1 and Q. 2 as a joking introduction. It is noteworthy that Q. 1, Q. 3, and Q. 4 are posed as direct questions on the form, which seems to contradict the instruction at the top of the form, but another factor that might have been significant here is that the enumerators considered it no more peculiar to ask these questions than any of the others.^[5]

In Q. 1 (What is your name?), people's first names were sometimes spelled differently from how they appeared on the SIHF. One or two people appear under one name on the SIHF, and another on the SIPF. Everyone has several names, and the name by which they are referred to can change over time, for example on the death of someone who shares the name that is in current use. In one or two cases, interviewees said that their 'registered' name should be put on the SIPF, even if it was not the name they were currently addressed by.

Real difficulties began at Q. 3 (How old are you?): almost no one knew how old they were. Where they did know their year of birth they calculated their age with reference only to the year, not their actual date of birth.^[6] Quite often, the age recorded on the SIPF was different from the guesstimate that had been entered by the enumerators on the SIHF.

The responses to Q. 4 (How are you related to Person 1 (Head of house)?) will be discussed in detail in a later section. The enumerators tried very hard to follow instructions, used English kinship terms themselves, and encouraged the interviewees to do so as well. The results will be uninterpretable. Q. 5 (Are you more closely related to anyone else in the house?) was quickly abandoned, at least in the interviews that I observed, for reasons that will become clear.

The question (Are you married?) posed no problems to the enumerators or the interviewees.^[7] However the data collected at community A may cause some puzzlement to the sharp-eyed analyst. Two dwellings contain a female person 1 who describes herself as married, but no evidence of a cohabiting spouse in either case. Why do these women describe themselves as married, rather than as widowed, separated or divorced? The reason is that local Indigenous marriages are often polygamous: it is not at all uncommon for a man to have more than one wife (and often those wives are sisters). It has become common- 44 - nowadays for a man to live with only one of his wives, and for the other wives to occupy one or more separate dwellings. In local terms this does not constitute either separation or divorce, if the relationship between the two parties remains amicable. In both the cases mentioned here, the husband lived in a nearby dwelling with another of his wives.

The 'place' questions also caused few problems for the enumerators and the interviewees, but the data almost certainly do not reflect what the census was after. 'Place', which is not further defined, was interpreted to mean 'community' rather than 'dwelling'. Moreover, people identify closely with the community as their home, even though they are at times highly mobile, and a 'yes' answer to Q. 7 (Do you live at this place most of the time?) reflects that identity rather than a computation about which community the person actually lives at 'most of the time'.

The instruction to give a street number, street name, suburb, and postcode if the answer to Q. 7 was 'no', as in the case of visitors, caused some hilarity. No one thinks in these terms. Although the community does have street names they are never used as a point of reference. Mail is not delivered house to house in this community. It does not have a street directory. People think in terms of 'X's house', or 'the new house', or 'the blue house' (see Musharbash 2001: 4 for similar observations about Yuendumu). This is generally true of all communities in the region, even the larger settlements.

The ambiguity of the word 'place' did surface with respect to visitors, because the dwelling at which they were enumerated tended to be the 'place' at which they usually stayed when visiting the community. The use of the verb 'live' (as opposed to 'stay') was not enough of a trigger to automatically generate the correct response. However, after discussion, the meaning of place as 'community' took precedence, and the home communities of visitors were put down as their usual place of residence. Little attempt was made to think seriously about where anyone was actually living one and five years ago (Q. 8 and Q. 9). The default assumption was that the answer was the same as Q. 7, that is the community with which the person identifies.

This group of questions was invariably answered hastily and in a formulaic way, for three reasons. One derives from local Indigenous culture, in which place has very different connotations from those it has in the mainstream, and in which years, as a unit of measurement, have little significance. The two other reasons concern the design of the SIPF questionnaire. Because the questionnaire took so long to administer, the enumerators tended to supply, or readily accept formulaic answers to, questions that appeared to have self-evident answers, or which appeared, in local terms, to be unimportant, or even absurd. For this particular group of questions this impulse was compounded by the design of the questions themselves. There is something very offputting about being asked to provide a complete address, three times in a row, when one's dwelling does not have that kind of address, or, even if it does, is never thought of in those terms.

These questions appeared at first sight to cause no problems at this community, where everyone is unambiguously of Aboriginal 'origin' (Q. 10) and 'ancestry' (Q. 13). A distinction between the two was also discerned—but it is not the distinction that the questions are aiming for. The details of the small print in Q. 13 (What is your ancestry?) draw on an implicit distinction between 'Indigenous', 'ethnic' and 'unmarked' (that is, Anglo-Celtic). These are not 'natural' categorisations, but ones embedded in the culture of mainstream Australia. Local Indigenous people do not categorise people in the same way. Moreover the content of these two questions is very problematic: neither says explicitly what information is really being sought.

I take Q. 10 to be a question concerning cultural identity (although it is not expressed overtly as such), and Q. 13 to be a question about biological (or 'ethnic' origins). The presence of the word 'ancestry' in Q. 13 led local people to a different interpretation. Ancestry was glossed by the enumerators as [sacred ancestral inheritance].^[8] Thus Q. 10 was interpreted more as an enquiry about ethnic origins (or biology pure and simple) and Q. 13 as a more complex enquiry about spiritual (or cultural plus biological) identity and origins. The 'right' responses were obtained, but for the wrong reasons.

The small print in Q. 13, which provided a clue as to the intent of the question, was noticed and often read out by the enumerators, and jokes were made about people's ancestry. The tendency to treat the small print as a joke in this question was encouraged by the examples given for possible ancestries (Vietnamese, Hmong, Dutch, Kurdish, Australian South Sea Islander, Maori, Lebanese). Most are improbable ancestries for local people of mixed descent. Because of this the examples given were not sufficient to override the interpretation of 'ancestry' as [sacred ancestral inheritance].

Q. 11 and Q. 12 (Was your father/mother born in Australia?) were generally treated humorously, because the answer to both was so self-evidently 'yes' in all cases.

The responses to Q. 14 (Do you speak an Aboriginal or Torres Strait Islander language at home?) were an accurate reflection of reality, in that everyone said they spoke an Aboriginal language at home. This is certainly true. For people who were from other linguistic regions, the 'language' name given tended to be a general one, and to correspond to what a linguist would classify as a language.

For local language speakers (the majority in this community), a much more complex picture emerged. The local Indigenous language is a language in the linguistic sense: it is a group of mutually intelligible dialects. But local people (and this includes the enumerators) rarely use the term that has been coined to designate the language. They commonly distinguish 'language' at two levels. They distinguish groups of dialects on the basis of their forms for the word 'this', and/or distinguish 'language' at the level of landowning group. Sometimes the labels are used separately, sometimes together. To the uninitiated, the census data would thus indicate that many different 'languages' are spoken in this community,- 46 - whereas in reality most of the 'languages' identified are variants of one single language. Moreover, the same variant may be recorded under three different labels: 'this', 'landowning group' or 'this + landowning group'.

Q. 15 (How well do you speak English?) was answered with a high degree of accuracy, although there was some debate about whether the phrase [a little bit, somewhat] should be translated as 'well' or 'not well'. People are in general able to assess objectively their level of competence in English.^[9] Adults tended to assess their own level of competence, and the enumerators accepted their self-assessment. For children, however, the enumerators adopted a formulaic approach: infants and pre-school children were said to speak English 'not at all', and schoolchildren were said to speak it 'not well', 'because they're just learning'. In one case a man wanted to put that his small grandson spoke English well, but he was overruled by the enumerators.

Q. 16 (What is your religion?) generated much debate; people wanted to mark more than one box. Many if not most local people are 'bi-religious'. As one interviewee put it: 'My beliefs are traditional, but my religion is [Christian denomination]'. The reasons for this lie partly in the mission history of the region, for by and large the founders of the missions in this area were respectful of the local religious belief system and did not attempt its wholesale repression, and partly in the nature of the local Indigenous metaphysic, which is incorporating and syncretic rather than exclusive. One might say that for local people their traditional religion is their Old Testament, and Christianity is their New Testament. Each has its place and function in the contemporary worldview.

There is no explicit indication that it is permissible to mark two boxes for this question. E1's solution was to mark only 'Traditional Beliefs', often declaring as he did so, '[Indigenous] comes before [white man], so we'll put Traditional Beliefs'. Most interviewees agreed to this. The other enumerators sometimes marked both [Christian denomination] and 'Traditional Beliefs', and sometimes only one or the other, depending presumably on what the interviewee's response was.

There are no computers in the community, so not surprisingly everyone responded 'no' to these questions. This is not to say that no one knows how to use a computer. At least one person put a computer course down at Q. 23, and the teachers and HA employees in the community all use computers to some degree in their work. The base school at community B makes use of the Internet, and the community teachers may well have some experience of using it.

Q. 19 (Do you go to school, TAFE or university?) and Q. 20 (What type of school or place of education do you go to?) caused no problems, and were answered accurately. One person was in the middle of a teacher-training course of some kind, but it was difficult to decide- 47 - which box to mark in Q. 20 because it was an in-house course at the homelands school headquarters in community B.

Q. 21 (Are you 15 years of age or more?) was found to be odd, because people had already said their age earlier. The enumerators finally recognised it for what it is—a cue to go to the end of the questionnaire if the interviewee is under 15 years of age.

Nearly everyone had difficulty with Q. 22 (What is the highest level of primary or secondary school you have completed?), including those enumerators who are teachers. The prompt 'Year 8 or below' had no salience for them: had the term 'primary' been used there would have been no problem. Very few local people have more than primary education, and the outstation school has no post-primary section. Even most of the teachers went no further within the school system as such. For those people who had been educated in 'mission time' this question was even more difficult to answer, and it was sometimes left blank.

The vast majority of interviewees answered 'no' to Q. 23 (Have you finished a trade certificate/apprenticeship, TAFE course or university course since leaving school?). For those few who answered yes, the box that was always marked was 'Yes, other course'. This was a faut de mieux response, because people did not know whether the courses they had attended were or were not a 'trade certificate/apprenticeship'. Q. 24 (What is the name of that course?) caused some problems because what is really being asked for is the name of the qualification rather than of the course, but the wording is misleading. Most of the courses people had done were not certificate, degree or diploma courses. Under 'Full name of course' people invariably put a descriptive title rather the name of the qualification, so that the information supplied there tended to overlap with or be the same as that supplied for Q. 25 (What did you study?). Q. 26 (What was the name of the place you studied at?) caused difficulty for some because their courses had not taken place at a particular institution. For Q. 27 (In which year did you finish that course?), most people were able to produce the certificate that showed when the course finished.

HA policy is that people without other regular sources of income (e.g. pensions, or 'real' jobs) who live at outstations are paid for 20 hours of CDEP work per week. This was the case for the majority of adults in community A. For these people, the amount recorded at Q. 28 (How much money do you get each fortnight before tax?) was always the same, that is $320–$399 per fortnight. The enumerator rather than the interviewee supplied the answer. If people were on a pension of some kind, the enumerator also filled in the amount. I did not ask the enumerators how they knew the correct answers to this question: clearly they had either been instructed on the right figures to enter, or had thought to find them out prior to the enumeration.^[10]

No attempt was made to compute other sources of income, for example from art production or royalties, although several members of the community derive income from one or both.- 48 - I suspect that this is because, in comparison to the known income from CDEP and pensions, the enumerators felt it would be too hard to elicit this information. Income from these other sources does not come in a steady stream, and it would have been very difficult to work out a fortnightly amount. The subject was never raised, either by the enumerators or by the interviewees.

For people on CDEP, the answers given to this question were correct—but for the wrong reason. The question was interpreted as 'Did you have a pay last week?' Fortunately, the previous week had in fact been a CDEP pay week. This question needs to be rephrased if it is to be asked next time. The word 'worked' or 'work' needs to find its way into the large print in the main question.^[11] The local interpretation of the question suggests that for local people, as far as CDEP is concerned, the word 'pay' is more salient than the word 'job'.

Q. 30 (What job did you do last week?) and Q. 31 (What things did you do in that job last week?) caused difficulties initially until the enumerators worked out what the difference was. A standardised response emerged: Q. 30 was answered with 'community service' and Q. 31 with what the actual 'work' consisted of. 'Jobs' included 'community leader' and 'home management', and there were a lot of 'rubbish collectors'. The enumerators and interviewees were clearly responding to the prompt in the question, which lists only 'jobs' that would be considered as such in mainstream terms. No-one put down 'making art works', or 'hunting' or 'fishing', although many people on CDEP spend more time in those kinds of activities than in those contained in the prompt to the question. For example, after discussion with the enumerators, one person answered Q. 30 with 'community work', and Q. 31 with 'community leader'. His brother from community E remarked: 'you should put artist'—an entirely reasonable suggestion since person 1 is one of the best known artists from the region, whose works are found in national and international collections. The suggestion was not taken up.

The wording of Q. 32 (Who did you work for last week?) focused people on the local community, rather than on the HA which administers the CDEP program. The word 'for' in this question has more than one possible interpretation. It can mean for whose benefit did you work? Or alternatively it can mean 'who did you work for in your capacity as an employee? The use of 'who' forces the first interpretation: the community, which consists of a collection of known individuals, is a more salient candidate for 'who' than is an organisation based at community B. The small print did not help in this case, because the word 'community' cued the name of the community rather than the over-arching organisation.

Q. 33 (What is your workplace address?) caused the same difficulty as other questions that asked for addresses. There are only three official 'workplaces' in the community—the school, the community shop, and the clinic (which did not function on a regular basis- 49 - at the time). Like dwellings, these workplaces are not thought of in terms of their address. These questions were clearly not formulated with remote communities in mind.

Q. 34 (What work does your employer do?) focused people's attention on the HA, because of the presence of the distancing word 'employer'. People do not regard their own community in those terms. The counter-intuitive instruction in small print to write 'community council' was overlooked, except by one community D visitor. Instead, the enumerators devised a formula to describe the activities of the HA (variations on 'provides community programs').

The answers provided to this question (How many hours did you work last week?) were formulaic. If a person was on 20 hours of CDEP the answer was '20'; if they were in full-time CDEP the answer was '40'. The enumerators supplied the answer, and no one disagreed. No attempt was made to compute the actual number of hours worked, and it would not have been possible to do so: nobody wears a watch or keeps track of time by the hour in the outstation context.

This question provided another opportunity for humorous responses, because everyone gets to work on foot. Levity was not unwelcome at this stage in the proceedings!

Q. 37 (Did you look for work at any time in the last four weeks?) and Q. 38 (If you had found a job, could you have started work last week?) were both answered in the negative by everybody, including those on CDEP. A common comment by the latter in reply to Q. 38 was 'no, because already got a job'. By this stage the interviewees had clearly realised that CDEP was being treated on the SIPF as payment for 'work', and were happy to fall in with this definition.

The enumerators had been carefully schooled to ask Q. 39 (see Appendix C) in the local language; they had evolved a formula for it during their training session and had practiced asking each other. The CFO's concern, transmitted to them, was about people being uneasy about hearing the names of dead people. The question as asked in the local language focused on the form being put away in a safe place in Canberra for a long time, and being brought out [when you are a old person, so that you or your grandchildren can see it, and see your name there]. The period of 99 years was interpreted as [a long time], but the precise period of time was not really understood. Put in this way, consent caused no problems for local people, who recycle names from grandparent to grandchild: 'it's OK, because your [(Z)DC] will be using that name by then'. Everyone answered 'yes', but it is arguable that they were responding to a different question from that actually being asked in Q. 39.

Q. 40 asks for the interviewer to declare that they have explained the requirements of Q. 39 to the person interviewed, and that they have correctly recorded the person's views at Q. 39. However, the interviewers did not at first read the small print, and the early forms were signed by the interviewee, at the request of the interviewer. E1 noticed the small print at some point during the first day, and thereafter the interviewers signed the forms.