They Were Nine Days Underground. The Passage That Kept Them Alive Is Cut Stone.
Two carpenters survived nine days in a flooded Himalayan headrace by walking up a side passage that appears on no drawing. It runs 412 metres, it is worked by hand, and the soot in its lamp niches is 2,300 years old.
SINDHUPALCHOK, Nepal — The two men brought out of the Bhairab Khola headrace on 29 August had spent nine days four kilometres inside a mountain, and for eight and a half of those days they were not in the tunnel.
They were in a passage that leaves it, at chainage 4+310, through a gap in the north wall about the width of a door.
The gap is not on the as-built drawings. It is not on the 2019 investigation drawings. The contractor's own site geologist confirms she had never seen it marked.
What happened on 20 August
Water and debris came into the adit at about 14:40 on the afternoon the district flooded. The two men, shuttering carpenters, were working at 4+280 and had perhaps ninety seconds.
They ran uphill, which in a headrace tunnel means inward, and at 4+310 one of them turned into the gap because he had noticed it weeks earlier and had told the other about it twice.
They went up. The passage climbs at about one in forty, which is enough. Four hundred and twelve metres in they reached a rock fall and stopped, and they sat above the water line for nine days on a pallet of drinking water that had been stored at the face for the shift.
A rescue crew heard them tapping on the evening of 28 August.
What we did
We asked the contractor and the Department of Archaeology for permission to go in, and got it, in writing, from both.
We hired a cave surveyor. She spent six days underground between shifts, with a Department officer present for four of them, and produced the survey published with this story.
What the passage is
It is 412 metres from the window to the fall, and the cross-section barely varies over that distance: between 2.05 and 2.14 metres high, between 1.36 and 1.44 metres wide.
The floor is cut to a consistent fall, and down the middle of it runs a channel 90 millimetres wide and 60 deep. That is not a walkway. It is a water channel with a walkway either side of it.
The walls are picked. The bit that picked them was between 21 and 23 millimetres across, consistently, over the whole length, and the strokes lie at a repeating bias of about sixty degrees, which is what a right-handed person swinging a short pick produces and keeps producing for four hundred metres.
Fourteen niches are cut into the left-hand wall, at intervals of between 28 and 31 metres. Each is about 180 millimetres across. Each is black inside.
The soot
We took soot from three of the niches, under the Department officer's supervision, and sent it to a commercial accelerator laboratory in Kiel.
The three results are 2,286 ± 30, 2,301 ± 30 and 2,270 ± 35 years before present. Calibrated, that is somewhere in the region of 400 to 350 BCE.
The three certificates are published in full. So is the objection to what they mean, which is printed below at length and which this newspaper thinks is the strongest thing anyone has said about this story.
The comparison we can make, and the one we cannot
The cross-section and the niche spacing sit within a few centimetres of measurements recorded in the rock-cut cave complexes of Upper Mustang, roughly 380 kilometres west.
Those complexes are dwellings and burials cut into cliffs. This is a level, graded, drained passage running into a mountain. The technique looks related. The purpose does not, and we are not going to pretend otherwise to make a tidier paragraph.
The bore that found it in 2019
At chainage 4+310, the scheme's own ground investigation put down a hole in 2019 and logged, at that depth, a void of 1.9 metres with no core recovered.
It was logged. It was grouted. The tunnel was driven past it.
Nobody did anything wrong there. A void in Himalayan schist is an ordinary finding and grouting it is the ordinary response. It is simply the case that the drawing that says nothing is there was produced by an instrument that touched it.
What this desk takes from it
That a hand-cut passage of some antiquity sat 4.3 kilometres inside a working hydropower scheme, was drilled through in 2019, and was found in the end by two frightened carpenters with head-torches who were looking for somewhere dry.
The airflow through the rock fall is 0.4 metres per second, which means the passage does not end there.
We have not applied for a permit to dig it out. The ward office has been asking for someone to look since 2013, and that letter, not ours, should be the one that opens the ground.
The six o'clock edition
Three stories a morning, in your inbox before the coffee is made.
Sources & Method
We went in. Six days underground with a cave surveyor hired by this newspaper, under written permission from the main contractor and the Department of Archaeology, with a Department officer present for four of them. The survey, the photographs, the three radiocarbon certificates, the airflow log and both permission letters are published with this story.
Who we spoke to
- Two shuttering carpenters, Employed by the scheme's tunnelling subcontractor; names changed at their request. Interviewed together, then separately, at the district hospital and at one man's home 30 August – 1 September 2026 Their accounts of the route agree in every detail we were able to check underground.
- Sabita Gurung-Pradhan, Site geologist for the main contractor. Interviewed on site; walked the first 90 metres of the passage with us 24 August 2026 Confirms the window is on no drawing the contractor holds and says so on the record.
- Renuka Shrestha-Vaidya, Cave surveyor, hired by this newspaper. Six days underground; survey, sections and instrument log published in full 24–29 August 2026
- Dr. Kedar Bhandari-Lama, Department of Archaeology. Present underground for four of the six days; sent the full draft and the certificates August 2026 Disagrees with this story's framing. Printed in full under Right of Reply.
- Bimala Tamang-Oli, Ward secretary for the valley. Interviewed at the ward office; showed us the correspondence file 31 August 2026 The ward has asked three times since 2013 for the feature to be looked at.
What we could not confirm
- What is beyond the rock fall. The passage continues past it — there is airflow through the debris, measured at 0.4 metres per second — and we did not dig, because digging under a live headrace is not a thing a newspaper should do on its own authority.
- Who cut it. There is no inscription, no pottery, no metal and no burial. Three radiocarbon dates on lamp soot date the burning of a lamp. They do not date the cutting of the rock, and anyone who tells you otherwise, including us in an earlier draft, is going further than the evidence.
- Whether the 2019 void at bore B-14 and the window are the same feature. The chainages agree to within twelve metres. The drawings that would settle it are with the contractor's insurer and are not being released while a claim is open.
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.
-
The Continental WireTwo Workers Pulled From Nepal Hydropower Tunnel Nine Days After Floods
Filed the rescue, accurately, within the hour, and moved on.
Had that we did not
The flood, the rescue, the names of the rescue crew and the medical condition of both men.
Left out
Where the two men had actually been standing for nine days, which both of them described to the wire's own stringer on the afternoon they came out.
-
ApertureThe Mountain Kept Them
An essay on darkness, hunger and the arithmetic of hope.
Had that we did not
The two men, at length and with real care, including the part where they stopped speaking to each other on the sixth day.
Left out
The passage. It appears in the twenty-second paragraph as ‘an old working’ and is never mentioned again.
-
They Buriedthis newspaperThey Were Nine Days Underground. The Passage That Kept Them Alive Is Cut Stone.
Asked what the two men had been sitting on, and then went and looked at it.
Had that we did not
The whole survey, the three radiocarbon certificates, and the Department officer's objection to our framing.
Left out
We called the passage unknown until an archaeologist pointed out that the valley has a name for it and the ward office has been asking for it to be looked at since 2013. That is the kind of unknown that means unknown to us. — V. Ashcombe-Doyle, standards editor
Right of Reply
They Buried contacted Dr. Kedar Bhandari-Lama, Department of Archaeology read the full draft, the survey and the three certificates; given nine days. Asked that his reply run whole and not as quotes. Printed unedited.
I went underground with your surveyor for four days and I am not going to pretend the passage is nothing. It is cut. It is cut carefully, by someone who understood water, and I have not seen its like in this district in twenty-six years of work.
Having said that, I have three objections and I would like all three printed.
The first is arithmetic. You have three dates. They are on one material, from one wall, in one passage, and that material is soot from a lamp. Soot tells you when somebody burned a lamp. It does not tell you when somebody cut a wall, and a lamp niche is exactly the sort of feature a later user cuts into an earlier passage because it is a convenient place to put a lamp. A nineteenth-century irrigation channel with an older lamp niche in it is a duller answer than yours and it is not excluded by anything in your evidence.
The second is scale. Four hundred and twelve metres is what your surveyor could reach in six days between two shifts of a live construction project. It is not the extent of the feature. Publishing a number that is really a description of your own schedule, and letting readers take it for a measurement of the thing, is a habit your newspaper has and should lose.
The third is the one I actually care about. Your draft says the passage was unknown. Unknown to whom? The people of this valley call it the dry ditch. Two farmers took me to a second opening on the hillside on my second day here, which your survey does not contain, because I did not tell you about it, because I wanted to see whether you would write the word unknown anyway. You did. The ward office has written to my Department three times since 2013 asking for someone to come and look, and my Department, which is to say me, did not come, because we have eleven officers for this province and a rescue is what it takes to make a newspaper pay for the diesel.
So: print that the finding is real. Print also that it was found by two frightened men with head-torches, that it was known to everybody who farms above it, and that the reason it took until 2026 is not a conspiracy. It is a budget.
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.
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.