Wednesday, October 7, 2026
They Buried
We Dug It Up
⌕ Search
AnalysisCryptids

He Was Inside The Searched Area For Nineteen Hours. Two Teams Walked Past.

A hiker found alive on day four had lain 400 metres inside ground cleared twice. His watch logged altitude and steps throughout. Matched against the teams' GPS tracks, it puts searchers within 90 metres of him, twice.

Sal BrindisiField Correspondent, Cryptids
4 min read · 881 words 81,208 readers

Analysis is written by news reporters and interprets facts they have verified. It is not opinion, and it takes no position on what should be done.

Segment D-2 from the assignment map, with the hollow marked where the volunteer found it, four days later. Engraved for They Buried

PACKWOOD, Wash. — Wesley Tarn-Ochoa was found alive on the fourth day, 400 metres inside a drainage that three teams had cleared on day one and cleared again on day two.

He had not moved. His watch says so: a barometric altimeter and an accelerometer, logging continuously, recovered from an evidence bag and read out for this newspaper by a firm that was not told what it was.

Matched second by second against the incident commander's assignment log and eleven GPS tracks bought from Lewis County for $27, the watch puts a searcher 88 metres from him at hour 19, and 71 metres from him at hour 31.

Two numbers in this piece are both nineteen and they are not the same nineteen. Teams logged nineteen hours and twenty minutes of boots-on-the-ground time inside the segment that held him. Hour 19 is when the first of them came closest.

The hollow

He slipped at 09:12 on the first morning, took forty-one steps, and stopped. The watch records nothing at all for the next thirty-one hours.

He was in a rock hollow under a fallen hemlock, four metres below the line a walker takes on that bank, wet through, and hypothermic by the end of the first night.

At hour 31 the watch logs four steps. He says he heard voices and called out. Of the first pass, in the dark at four in the morning, he remembers nothing.

Figure One man's watch against eleven GPS tracks, on the same clock
Hour 0 — 09:12, day oneThe last stepsForty-one steps at 1,402 metres, a drop of nine metres in eleven seconds, and then nothing at all for thirty-one hours.Hour 4 — 13:40Reported overdueHis partner calls the sheriff's office four and a half hours after the agreed return time. The incident is opened at 14:02.Hour 11 — 20:11Segment D-2 assignedTeam 3, three searchers, hasty. The log gives the segment as 0.7 km of drainage: steep, wet, heavy blowdown.Hour 19 — 04:08, day twoFirst close passTeam 3's track runs 88 metres from him, uphill and across the creek, in the dark. The watch logs no steps. He does not remember it.Hour 24 — 09:30ClearedD-2 signed off. The probability-of-detection box on the form is filled in by hand, in the field, as 0.5.Hour 31 — 16:20Second close passTeam 7 passes 71 metres below him on the creek line. Four steps are logged at 31:14, and then none for the forty-nine hours before he was found.Hour 33 — 18:00Cleared a second time, and releasedThe drainage comes off the active board. The search area is expanded to the ridge, six kilometres out, where twenty-eight of the next thirty-one assignments go.Hour 52 — 13:00, day threeThe dog teamA single air-scent dog works the creek, assigned 900 metres downstream of the hollow. Wind from the north-east all afternoon.Hour 81 — 18:20, day fourFoundA volunteer working off his assignment, on his own initiative, walks the fallen timber on the north bank and sees a boot.Hour 83 — 20:0531.4 degreesCore temperature at the aid station. He was discharged from hospital on 6 November and walked out under his own power.
Every time below is the incident clock, hour zero at the first logged step of the walk. The watch is accurate to about fourteen metres vertically and the team tracks to about five metres horizontally, so the two closest approaches could each be out by a house-length. Neither error is large enough to matter. They Buried, from the watch read-out, the assignment log and eleven county GPS tracks, all published entire

The arithmetic is already published

None of this is remarkable. It is the number the discipline prints in its own manuals and teaches to volunteers on their first weekend.

A hasty team clearing steep timber achieves a probability of detection of roughly 0.4 for a subject who is still, silent and lying down. Three passes at 0.4 leave better than one in five such subjects uncounted. Segment D-2 was signed off twice, the first time with 0.5 written by hand in the box.

Eleven cases, unlabelled

We took the eleven recoveries most often described as physically impossible, stripped every name, place, date and source from them, and gave them to two search statisticians who were not told whose list it was.

Both placed each case against the published distance-from-point-last-seen distributions, built from about 150,000 logged incidents and broken out by subject category: child, hiker, hunter, despondent, dementia.

Nine of the eleven fall inside the 95th percentile for their category. Several are dull. One much-cited hunter was recovered at 6.1 kilometres from his point last seen, where the 95th percentile for hunters is 9.6.

One is a transcription error. Case four turns on a recovery at 4,100 feet of elevation. The original incident report, which we obtained, gives 1,400. The digits were reversed somewhere between 1998 and the first book to carry it, and the figure has been reprinted eleven times since.

Roderick Cavanaugh-Pyle, who compiled the eleven, has told us he is withdrawing case four. His reply runs below in full and it is the best argument against this article that anybody made.

The discipline does this to itself in public, and better than its critics allow. Our vault holds PX-2173, in which a slab-avalanche model is calibrated against a set of 1959 autopsy findings and then tested by people who went back to the pass.

The one that does not fit

Case eleven is a man of seventy-four with dementia who walked out of a care home at about half past six in the morning and was found alive on the fourth day by a farmer, in a hay barn, 31 kilometres away, on the far side of a river with no crossing within 19 kilometres of either point, in indoor shoes.

Ninety-five per cent of subjects in his category are recovered within 3.2 kilometres. Both statisticians, working separately and blind, put him beyond the 99.9th percentile.

We could not get the recovery team's track. The county says the device was replaced in 2016 and the data never migrated. One of the two searchers who found him declined to speak to us; we could not locate the other. The incident report is internally consistent, which is all we can say for it.

He never said how he got there. He died in 2021.

We print it here, at the foot, unexplained, because leaving it out of a piece that has just explained ten other things would be a way of winning an argument rather than reporting one.

What this desk takes from it

The nineteen hours are the story and the 0.4 is the reason for them. A man lay four metres off a trail, inside ground that had been searched, signed for and released, and the searchers who walked past him were not careless. They were operating at the detection rate their own manual predicts, which is much lower than the word cleared sounds.

Mr Cavanaugh-Pyle's reply makes the point we did not: we have printed his cases as ordinary and filed the finding under a headline that makes the searchers the subject. He thinks the scandal is the word on the form. So, now, does this desk.

Disclosure. This newspaper paid $1,480 to the data-recovery firm, $27 in public-records fees, and $4,200 in total to the two statisticians, at the same rate, in advance, with no clause tying payment to any finding. Mr Tarn-Ochoa was paid nothing and asked for nothing; he was given the complete read-out of his own watch before anything was published from it, and could have stopped the piece at that point. Mr Cavanaugh-Pyle was paid nothing and sells a book that contains case four.

How Others Covered This

The same events, as reported elsewhere on the same day. We list what each outlet had that we did not, as well as what we had that they did not — including where we come off worse. Why we print this.

  1. The Meridian Telegraph
    FOUR DAYS IN THE DRAINAGE: THE GROUND THAT GAVE HIM BACK

    Took the double clearance as evidence that the drainage had been empty and then was not, and built the piece on the gap between the two.

    Had that we did not

    The assignment log, quoted accurately and at length, and a genuinely good account of the volunteer who broke his tasking and went back.

    Left out

    The detection-probability figure printed on the same form, three lines under the sentence they quoted, in a box the team had filled in by hand.

  2. Aperture
    The Arithmetic Of Not Being Seen

    An essay on hypothermia, silence and the failure of the will to shout. The best writing anybody has done on this incident.

    Had that we did not

    Tarn-Ochoa's own description of deciding not to call out, which is the most useful thing said by anyone about why searches miss people.

    Left out

    It never asks for the GPS tracks. They are public records in Washington, they cost this newspaper $27, and they contain the whole answer.

  3. They Buriedthis newspaper
    He Was Inside The Searched Area For Nineteen Hours. Two Teams Walked Past.

    Had a survivor's watch read out, put it on the same clock as eleven team tracks, and then tested eleven famous recoveries blind against the published distributions.

    Had that we did not

    The watch log entire, all eleven tracks, both statisticians' unblinded worksheets, and the case we cannot explain, printed unexplained.

    Left out

    We reconstructed a hypothermic man's night second by second and put his four steps in a diagram. That he is certain he shouted, and that his certainty is not evidence of a sound, sits in a footnote and not in the diagram. — V. Ashcombe-Doyle, standards editor

Right of Reply

They Buried contacted Roderick Cavanaugh-Pyle, compiler of the eleven cases on 2 July 2026 with the full draft, both statisticians' worksheets and the original incident report for case four, and no deadline set. Replied 20 July 2026, asking that it run in full and unedited. Printed as sent.

Case four is wrong and I am withdrawing it. It was 1,400 feet. I have reprinted 4,100 in three editions and in a talk I have given perhaps sixty times, and I took it in 1998 from a secondary source I did not check because it agreed with me. I would rather have learned that from you than from nobody, and I am saying so first because everything after this is going to sound like a man defending himself.

Now. You have proved my case and printed it as a refutation, and I would like to know whether anybody on your desk noticed.

I have never once written that these distances were impossible. I have written that the searches were declared complete. Those are different sentences and your own arithmetic is the argument for mine. You tell your readers, correctly, that a hasty team in blowdown finds four subjects in ten. You tell them that three passes therefore leave better than one in five uncounted. And then you print, without comment, that segment D-2 was signed off, released from the active board, and that twenty-eight of the next thirty-one assignments went six kilometres away to a ridge.

That is the thing I have spent nineteen years complaining about, and you have documented it better than I ever did, and you have filed it under a headline about a man in a hollow so that the complaint reads as an explanation.

A number that describes a thousand searches does not explain one search. Your two statisticians placed nine of my eleven inside a distribution. Good. A distribution is a shape made of other people. It tells the county what to plan for. It does not tell a family what happened on their mountain, and when you write that a case falls at the ninety-first percentile you have not said anything about that case at all — you have said something about the other hundred and forty-nine thousand.

I object to your phrase most often described as physically impossible. Described by whom. Not by me, and not in print, and not once in nineteen years. That sentence lets your reader believe you are correcting my claim when you are correcting a claim assembled out of other people's summaries of it, and you are a newspaper that has printed a standards column on exactly this, so I will hold you to it.

I object to being the foil. Your headline is about Wesley Tarn-Ochoa. Your piece, from the fourth section onward, is about me, and I was given a draft on the second of July, which is fair, and eighteen days, which is generous, and no way at all to answer the framing except at the bottom of the page where you have put me. That is the standard arrangement in this trade and I am not asking for special treatment. I am pointing out that the arrangement flatters whoever writes the top.

On case eleven. You have printed it unexplained, at the foot, with the missing track and the searcher you could not find, and you have not reached for anything. That is the only paragraph in your piece I would have written the same way, and I expect it to be the least read thing in the whole newspaper this week. I have had that experience. It does not get easier.

And on Mr Tarn-Ochoa, since he is the reason any of us are here: he lay four metres under the line a walker takes, for four days, and two teams went past, and a man who broke his own tasking found him. Every serious person in search and rescue already knows that is ordinary. It is the people who read your newspaper who do not, and if your piece teaches them the 0.4 and nothing else, it will have done more good than my eleven cases ever did. I would still ask you to consider that I am the reason you looked.

Published unedited under our right-of-reply guarantee.

How was this story?

We publish the result, whatever it is. Reader verdicts appear on the front page and in our newsroom metrics.

162 verdicts · 98.8% loved it

Readers' Letters 0

Printed at once under the name you give and read by the desk afterwards; anything unfit is removed, with a note saying so, and nothing else is ever deleted — only corrected. Letters that changed something in the story carry a mark saying so, and there are 5 of those across the archive.

  1. No letters yet on this story. Yours would be the first.

Write to the desk