The CFO’s difficulties in making contact with the CCs had necessitated the introduction of backup forces. On 20 September, I flew in to A8 in the company of the CI who had conducted the count at A2 and a non-local Indigenous Assistant who had previously been helping to finish off the count at F. The funeral there had finished, but there were still several ‘visitors’ in tents who had stayed on for a while.^[8] Meanwhile, the CFO was attempting to deal with all the other outstanding lacunae in the A/C area.

We counted nearly 90 people at A8. One household consisted of people who had been counted previously at A10. I had suspected that we would bump into them again and so had brought the A10 form with me, reasoning that if they were at home it would be better for them to be counted there and removed from the form at A10. I was also interested to see whether they would say that they had already been counted, and also whether the details on the two forms would differ. The Indigenous Assistant was assigned to their household. They did not say that they had already been counted, for the simple reason that they did not know they had been. At A10, the form that listed them had been filled in by a local CC, and only one person from the household helped with the form. The rest of the household had been out hunting at the time the form was filled in. Now, at A8, most of the household was present, but the person who had been interviewed at A10 was out hunting! Since I was officially a CC this time, I crossed all those present at A8 off the form that had been completed at A10. Four people were still at A10, so they were left as ‘visitors’ there.

The details supplied for certain individuals on the two forms did indeed differ. Because the Indigenous Assistant was not a Yolngu speaker, people gave him their English names, while at A10 their Yolngu names had been given. Estimates of people’s ages differed wildly. One child was put down as a three-year-old at A10 and as a seven-year-old at A8. The ages of several of the adults differed by a decade. It seems highly probable that had I not been the CC in this instance, the members of this large household would have been double-counted. Because of the differences between the two forms, this double-counting would have been undetectable once the forms were back at the CMU in Darwin.

Since I was a CC this time, I took the opportunity to act as the CI for one household at A10. I wanted to experience for myself the CI’s task. I chose a large household with several visitors who had not yet returned to A. There were 21 people in all, necessitating the use of two forms. Bearing in mind that I am a highly literate English speaker with some competence in the Yolngu languages and a fairly detailed knowledge of the community, I still found the task pretty gruelling. It took two full hours and keeping the interviewees focused on the task was difficult, when so much of the time I was simply ticking boxes rather than engaging with them.

I also had to make some tricky judgments about who to put on the form. There was a woman from A and some of her family who said they had not been counted at A, so I added them as visitors. This woman’s brother and his family had also been at the funeral and had left only that morning. The sister said she thought her brother’s family had not yet been counted. On balance, this was probable, because the man had a medical condition that necessitated him staying for prolonged periods in Darwin, and he and his family were often in transit between Darwin and A. It was possible that they had been at A on census night, and had missed being counted in Darwin. It was equally possible that they had been in Darwin when the count took place at A. I decided therefore to put them on the form as visitors and crosscheck the household form from A when I got back to the CMU. It transpired that they had, in fact, been counted at A. I do not know whether they were also counted in Darwin.

This was yet another instance of the complicating effect of the rolling count combined with mobility. I was in a unique position, with access to knowledge at the local level where the CIs were operating and at the more regional level, because of my observation of the entire process in the A/C region and my access to the CMU in Darwin. I found myself in the same position as everyone else at the moment of the count: forced to make a more or less educated guess about whether or not to include someone on a form.

On our return to A, I found that the CFO was about to board a plane to Darwin. He had spent a nearly fruitless day looking for people. He was exhausted and feeling very unwell and was clearly unable to continue. He had contacted the CMU and a replacement team was on its way: a member of the CMU staff and one of the other CFOs. They had been at F, tying up the loose ends there. They arrived just before he left and he handed over his paperwork to them. I was leaving for Darwin the next day, so I spent the evening with the new team and the Indigenous Assistant, who had stayed on to help. We reviewed what still needed to be done or checked. On 20 September, therefore, more than 10 weeks after the count had started, it was still not fully completed.