Fourteen searchers, overlapping sweeps, every GPS track saved and plotted afterwards. Over eleven years not one track has crossed a half hectare clearing, and nobody has ever reported turning aside.
I am a search planner for a lowland rescue team. We are the volunteers who get called when somebody vulnerable walks out of a house at night, and most of what we do is find a confused person in a hedge at four in the morning and put a blanket on them.
The planning is the part nobody sees and it is all arithmetic.
You divide the ground into segments with real boundaries, a fence, a ride, a stream. You give each segment a priority from what we know about the missing person, because a man with dementia goes downhill and follows linear features and is usually found inside two kilometres. Then you put a team in with a spacing that depends on what you are looking for and how thick it is, and you record what you did.
Recording it is not optional. Every searcher carries a GPS that logs a track, and afterwards the tracks are plotted over the segment and we calculate coverage: how much of that ground was actually swept at the spacing we intended. Somebody walking into a segment does not make it searched. The tracks make it searched, and until they are plotted we do not claim it.
I have eleven years of those plots.
The wood is about three hundred hectares on the edge of our area, mixed conifer and broadleaf, cut up by rides into eleven segments. We have searched it thirty one times since 2015. It is close to a hospital and a care home, so it comes up a lot.
In segment D there is a clearing.
It is roughly half a hectare, an old stacking area by the look of it, flat, grassed over, with bramble round the edges and a rotten pile of brash in the middle. It is about sixty metres across. You can see the whole of it from any edge of it.
There is no track across it.
I did not find that. A new member found it, doing exactly what I tell every new member to do, which is go through the old plots and look at how coverage is calculated, because a planner who cannot read a coverage plot is a planner who sends people out twice.
He came to me with segment D from a search in 2019 and asked whether the white patch in the middle was a no go area, a pond or a dump.
It is not on the hazard map. It is not fenced. It is the easiest walking in the segment.
So I went back through all thirty one.
Twenty two of those searches included segment D. Across them, about two hundred and sixty individual tracklogs.
None of them crosses it.
That is not the same as nobody going near it. Tracks run along all four sides of it, at between two and eleven metres from the edge, which is what you would expect from people sweeping at a twenty metre spacing through a wood with a clearing in it.
On nine occasions a track comes straight at it, reaches the edge, and runs along the bramble line to the far corner before carrying on in the original direction.
That is a perfectly normal thing for a tracklog to do when there is an obstacle. It is what a tracklog looks like when somebody walks round a pond.
There is no pond.
Then I did the thing I should have done first, which was ask the people.
I have asked nineteen team members, over four years, casually, one at a time, never two together, and never telling them why. They are searchers and they notice being managed, so I have been careful and slow about it, and I am aware that is its own kind of odd.
Not one of them remembers a clearing in segment D.
Several of them know the wood extremely well. One has searched it eleven times. He drew me the rides from memory and got them right, and he described segment D as solid conifer with a brash line through the middle of it.
There is no brash line through the middle of it. There is a clearing you could land a helicopter in.
I walked it myself in the October of 2021, deliberately, alone, with the team GPS running, because I wanted one track across it in the file.
It is a clearing. Grass, an anthill, the brash pile with elder growing out of it, a view up through the gap at the sky. I stood in the middle of it and I ate a sandwich and I was there for about twenty minutes.
The tracklog from that day goes along the southern edge and out at the corner.
I have looked at the raw file, not the plot. The points are one a second, there are no gaps in the timestamps, the whole twenty minutes is there, and for those twenty minutes the points are in a loose cluster eight metres south of the bramble, all of them, with a spread of about four metres, which is what a handheld GPS does when you stand still.
I was carrying the unit in my hand, not clipped to the top of my pack where the antenna ends up under a hood. I have the photographs I took with my phone while I was in there, and the phone's own location data for those photographs puts me in the middle of the clearing.
So I have two devices that disagree by about forty metres, in the open, under a gap in the canopy, which is the best reception in that entire wood.
The part I have to write down.
Our coverage calculation is not a presentation tool. It drives decisions. A segment that comes back at ninety per cent coverage with nothing found gets a lower priority for the second sweep, and when you have fourteen people and four hours of usable light you go where the arithmetic sends you.
Segment D has come back at between eighty eight and ninety six per cent, twenty two times.
That half hectare is inside the twelve per cent. It has been inside the twelve per cent for eleven years, and the reason it reads as ordinary unsearched margin is that the unsearched margin in a wood is always the bramble and the wet and the steep, and nobody reads a coverage plot and wonders about the flat bit in the middle.
Three of the thirty one searches did not find the person. Two of those were resolved somewhere else entirely, days later, and the outcomes were what you would expect and I am not going to discuss them.
One is still open.
I have done one thing about it, and it is not the thing a competent planner would do.
The hazard overlay that goes out on every search map for that wood now has a hatched box in segment D marked OLD STACK, BRASH, AVOID, which is a thing I typed in myself in 2022 and which nobody has ever queried, because avoiding a brash pile is the most boring instruction in search management.
It means the clearing is now excluded ground, and excluded ground comes out of the denominator, so segment D reads ninety nine per cent and the twelve per cent hole is gone from the plot.
I know what that is. I have been doing this for nineteen years and I have sat on debriefs where somebody's judgement was taken apart for less, and if it is ever taken apart I will say what I have written here and they can make of it what they like.
I did it because the alternative was to put fourteen volunteers at twenty metre spacing through a clearing that eleven years of instruments say is not there, in the dark, looking for somebody, and then read their tracklogs in the morning.