Room seven has no camera

The patrol wand logs every checkpoint with the hour and the minute, and the report comes out printed on the supervisor's desk every Monday.

It works like this: there are eleven points around the building, each one a round fitting screwed to the wall, and you touch the wand to it and it beeps. The system has no idea whether you looked at anything. It knows that at 01:14 somebody touched the wand to point 5.

We do four patrols a shift, seven in the evening to seven in the morning. Between them we sit in the control room, which has twelve screens and one good chair and one bad one.

Adilson and I have worked together Tuesday to Saturday since March.

The museum is municipal, housed in a 1927 mansion, with nine exhibition rooms over two floors.

From outside it looks like one thing. Inside, at night, on the emergency lights, it is an old building with floorboards that crack and a climate system that starts itself at three in the morning and frightens every new member of staff.

We have twelve cameras for nine rooms, the hall, the staircase and the loading bay.

The one in room seven burnt out in March. The replacement is stuck in a tender that was challenged in May and that nobody expects to move this year, so there is a printed sheet taped to the corner of screen seven that says awaiting replacement, in the handwriting of the previous supervisor, who retired in July.

Room seven is textiles. It is the smallest, it is the darkest, and it is the only room in the entire museum where nobody on earth can see what is going on.

About Adilson, what there is to say.

Fifty-one, twenty-two years in security, two grown children. He did twelve years in bank branches and says a museum is better because in a museum nobody turns up armed wanting the safe.

He patrols clockwise. I learned to patrol his way round.

On a twelve-hour shift with two people, conversation runs out in the third week. After that what is left is something else: we know by heart how many paces it is from the stairs to point 6, we know which board in room four cracks, and we can tell from the sound of a door whether the other one is going up or coming down.

That takes about four months to happen and it has no name.

What we do for the other ten hours of the shift is watch screens. Twelve screens, nine rooms, and nothing happening on any of them for twelve hours at a stretch, three hundred-odd nights a year.

Night security is the best-paid job of doing nothing that exists, and people who have never done it assume that is a good thing. It is not. The body tires more standing still than walking, and a head that spends the night awake with no task will go and find itself a task somewhere.

In August the museum closed for eleven days for pest treatment and an electrical inspection.

Closed to the public does not mean closed. Security carries on, the collection is still there, and both shifts keep running, with the difference that there is no other member of staff in the building after six.

For eleven days we were the only two people inside a 1927 mansion from seven at night until seven in the morning.

On the night of the fourteenth of August, a Thursday, the two o'clock patrol ran late.

That is in the report, because everything is in the report. Point 6 was read at 02:09. Point 7 was read at 02:43.

Between the two there are fifteen metres of corridor and one door.

Room seven has four upright cases, a wooden bench in the middle for anyone who wants to sit and look, and a boarded floor that is the only one in the museum that does not crack, because it was replaced in 2011.

The lighting is fifty lux, which is dark enough that a person takes about thirty seconds adjusting their eyes on the way in.

The air handling holds the room at twenty degrees and fifty-five per cent humidity all year, and it makes a low constant noise that covers any other.

We had not arranged anything. There was no conversation before, none at the time and none afterwards, and I have thought about it a great deal and I still think it was the empty building and not us.

The wooden bench in the middle is a metre twenty long and bolted to the floor.

I put the wand down on it, and that is why point 7 was not read until 02:43.

The museum reopened on the twenty-fifth of August.

With the public back, with the intern, the woman on the desk and the education staff walking school groups through the rooms on Tuesday mornings, a night shift becomes a different thing: there are still two people, but the building stops being empty, because a building that had people in it during the day is never properly empty at night.

We still do four patrols. He still goes clockwise.

The August report was filed without comment. Thirty-four minutes between two points triggers nothing in the system, which only flags a patrol not done, and nobody reads the patrol reports of a municipal museum unless something has happened.

Nothing happened. Not one item moved.

In case three of room seven there is a patchwork quilt in cotton and linen, made inland around 1890, donated by Maria Ovídia Fontes in 1974, and the room is kept dark because dyed cotton loses its colour in light and does not get it back.