We Bought 41 Flight Lines. Two Teams Reprocessed Them Blind. Both Found A Ceiling.
Forty-one flight lines, bought raw and reprocessed twice by teams who never spoke. Both resolve a flat roof 2,140 metres down, three shafts rising from it, and a gravity low that water cannot make.
ABERDEEN, Scotland — Two thousand one hundred and forty metres below the snow of one interior sector of East Antarctica, there is a flat roof.
It is flat to within 4.1 metres across 610 metres of section. Three near-vertical conduits rise from it through the ice, and the shallowest of the three stops 90 metres under the surface.
Two geophysics firms found it independently, in the same 41 flight lines, neither knowing the other had been hired.
What we bought
Forty-one raw flight lines, flown between 2019 and 2021 by a commercial survey contractor: a 60 MHz ice-penetrating radar and a strapdown gravimeter on the same aircraft, 640 metres above the surface at 105 metres a second. Line spacing 4.5 kilometres. 6,140 line-kilometres in all.
We bought them at the contractor's published rate for Level-0 data — stacked, unmigrated, navigation and instrument logs intact — for €96,400. Nobody here opened them.
Ravensmoor Geophysics in Aberdeen and Solent Applied Geophysics in Southampton were paid the same fee, in advance, to reprocess them. Neither was told the other existed. Both were told their report would run whether or not it agreed with anyone.
It is the method we used on the four purchased radar passes over Giza, PX-1912, and on the Wadi al-Jarf rations. Not clever, only expensive.
Radar measures time, not depth
The roof returns at 25.40 microseconds, two-way. What that is in metres depends entirely on how fast you think the wave went, and this is where the two firms part.
Dr. Iselin Krogh-Adeyemi at Ravensmoor used one constant — 168.0 metres per microsecond, the standard figure for solid ice. Her roof is at 2,134 metres.
Dr. Marguerite Ochieng-Halvard at Solent modelled the firn: 88 metres of compacting snow at the top, where the wave runs faster because the medium is emptier, ramping from 231 metres per microsecond at the surface to 168.5 at close-off. Her roof is at 2,143 metres.
Nine metres apart. We have printed the midpoint and both ends throughout.
But a velocity error moves the whole reflector together: it cannot flatten a sloping reflector or tilt a flat one. Flatness is a relative measurement and it survives the argument untouched. Ravensmoor gets 3.6 metres peak-to-trough across the 610. Solent gets 4.1.
The one sentence where they contradict each other
The reports run to 214 and 189 pages and agree on every figure in this story. There is exactly one sentence in which they disagree: the last sentence of §6.2 in both.
Ravensmoor: All three features collapse to point reflectors under migration and we interpret all three as in-plane voids in the ice column.
Solent: All three features collapse to point reflectors under migration; C-2 retains 1.8 µs of uncollapsed tail at 30° aperture, and we interpret it as a void while noting we cannot exclude a scatterer 380 m off-track.
That is the clutter dispute entire. Krogh-Adeyemi migrated at a 45-degree aperture and considers it settled. Ochieng-Halvard migrated at 30 and does not. Both put all three conduits in the same places, to within 40 metres.
Water is denser than ice
The director of the programme that flew the sector says the reflector is the top of a basal water body, and his argument is a real one: lake surfaces are flat because water settles onto an equipotential, and a bright specular return at the bed is the commonest thing an airborne sounder finds down there. His statement is at the foot of this page.
The gravity answers him with one number.
Fix the geometry from the radar — roof at 2,140 metres, floor 302 metres below that, 1.9 kilometres across — and ask what fill reproduces the observed residual of minus 11.4 milligals, plus or minus 1.4.
Air, in ordinary crust of 2.67 grams per cubic centimetre, gives minus 11.6.
Water gives minus 7.1. For water to produce minus 11.4 in that cavity, the surrounding rock would have to weigh 3.67 grams per cubic centimetre. Peridotite is 3.3. Nothing in the crust is 3.67.
There is a second reflector 302 metres below the first, with returns from beneath it. No sounder sees through 302 metres of water.
The coordinates
C-1, the shallowest, terminates 90 metres below the snow at 77° 14.62′ S, 61° 03.41′ E. C-3 terminates at 430 metres, at 77° 15.71′ S, 61° 04.90′ E. C-2, the contested one, terminates at 1,180 metres, at 77° 15.08′ S, 61° 01.77′ E. Surface elevation is 3,247 metres.
We are printing them because a coordinate can be checked and a description cannot, and because the next aircraft over that box can tell us we are wrong.
What has not closed
An unlined hole in ice does not stay open. Below about 130 metres the overburden shuts it, which is why every deep core ever drilled comes up through fluid — the fact this newspaper spent a fortnight on at Vostok.
C-1 is 2,050 metres of open column.
Something is holding it open. Neither firm will say what. Neither will we.
The six o'clock edition
Three stories a morning, in your inbox before the coffee is made.
Sources & Method
We bought the raw lines before we had any idea what was in them, then paid two geophysics firms the same fee in advance to reprocess the identical dataset without knowing of each other, on the written condition that both reports would run whether they agreed or not. Only after both were delivered did we model the gravity, and only after that did we send everything to the programme that flew the sector.
Who we spoke to
- Forty-one raw flight lines over one interior sector, Level-0 ice-penetrating radar and airborne gravity, flown 2019-2021 under commercial contract. Bought from the survey contractor's data desk at its published rate, EUR 96,400, under a licence permitting reprocessing and publication of results February 2026 Stacked, unmigrated, navigation and instrument logs intact. Nobody at this newspaper opened them before they went out.
- Dr. Iselin Krogh-Adeyemi, Principal, Ravensmoor Geophysics, Aberdeen; first reprocessing. Commissioned blind and told before starting that her report would run whether or not it agreed with a second she was not told about March - July 2026 Single constant velocity, Kirchhoff migration at 45 degrees. Puts the roof at 2,134 m and regards the clutter question as closed.
- Dr. Marguerite Ochieng-Halvard, Principal, Solent Applied Geophysics, Southampton; second reprocessing. Given the identical lines with no sight of the first report and no contact with its author March - July 2026 Two-layer firn model, Stolt migration at 30 degrees. Puts the roof at 2,143 m and keeps a caveat on one conduit.
- Dr. Emeric Sandholt-Baraka, Director of the national survey programme that contracted the flights. Sent the full draft, both reports and the gravity working; replied in writing after nineteen days with no deadline imposed August 2026 Says the reflector is a basal water body. Printed entire and unedited below, including the paragraph about 2022.
- Forty-seven published subglacial lake-surface profiles, The comparison set for the flatness claim. Compiled from the open literature by both teams separately, before either was shown the other's compilation June 2026 Neither team thinks 610 m of section settles anything on flatness alone. Both say so in writing and we have printed it.
Documents
- PX-1912 — Four raw satellite radar passes over the Giza plateau, purchased by this newspaper accepted
What we could not confirm
- What holds C-1 open. An unlined cavity in ice at that overburden closes on itself; below roughly 130 metres every deep hole ever drilled has had to be kept full of fluid, which is the whole reason the Vostok contamination story exists. C-1 is 2,050 metres of column that has not closed. Both teams resolve it and neither will offer a mechanism, and we have none to offer either. That is the next locked door, and it is a physics question before it is anything else.
- The along-track shape. Line spacing is 4.5 kilometres and six lines cross the body. Between crossings there is no data at all. What we can say is that six sections 4.5 kilometres apart all show a flat roof at the same depth; what we cannot say is whether that is one chamber 2.4 kilometres long or three separate voids that happen to line up. Nothing in 41 lines can distinguish those two.
- Whether anything is inside it. We have bought aircraft data and paid two firms to argue about it. Nobody has drilled it, cored it, lowered anything into it or landed near it. The nearest borehole of any depth is more than 400 kilometres away. Everything below the roof in our figure is inference from two physical measurements, and inference is not a photograph.
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 Meridian TelegraphThe Door In The Ice: Antarctica's Entrance Found At Last
Took the shallowest conduit as a doorway and ran a photograph of a different sector above it.
Had that we did not
The ninety metres, correctly, and the coordinates, which they took from our figure and credited.
Left out
That ninety metres is where a radar return stops, not where an opening is, and that nobody has seen anything.
-
Signal & LedgerWhat EUR 96,400 Buys: The Case For Reprocessing Old Survey Archives
Read the story as a procurement lesson for programmes sitting on decades of unreprocessed Level-0 lines.
Had that we did not
The licence terms and the blind-commission structure, set out more clearly than we set them out.
Left out
The gravity. Six thousand words on the economics of old data without the one number that decides what the data shows.
-
They Buriedthis newspaperWe Bought 41 Flight Lines. Two Teams Reprocessed Them Blind. Both Found A Ceiling.
Bought the raw lines, paid two firms who had never met to reprocess them, and printed both reports and the sentence where they part.
Had that we did not
Both reports entire, the velocity dispute, the one contradicting line in 403 pages, and the surveyed coordinates.
Left out
Our headline says both teams found a ceiling. Both teams found a reflector. Ceiling is our word, it is doing work our evidence has not done, and it is on the front in forty-four point. — V. Ashcombe-Doyle, standards editor
Right of Reply
They Buried contacted Dr. Emeric Sandholt-Baraka, director of the national survey programme that contracted the 2019-2021 flights on 1 August 2026, in writing, with the full draft, both reports, our gravity working and the coordinates, and no deadline imposed. Replied in writing on 20 August 2026. Printed entire and unedited, including the passages about our licence and about his own programme.
I have read both reports and I want to begin by saying that the work is competent. Ravensmoor and Solent are serious firms, their processing is defensible, and I have no quarrel with a single number either of them produced. My quarrel is with the word you have put on top of them.
What you are calling a roof is a bed reflector, and the bed under that sector is wet. A subglacial water body has a flat surface for a reason that is not mysterious: water settles onto an equipotential of the hydraulic head, and it will do that under two kilometres of ice as readily as it does it in a bathtub. It returns brightly and specularly because the dielectric contrast between ice and water is one of the largest in the natural world. A flat, bright, laterally continuous return at the ice base is the single most ordinary thing an airborne sounder finds in the interior of that continent, and we have mapped some hundreds of them.
Your flatness argument I do not think is an argument. Six hundred and ten metres is a short section. Your own two teams say so and you have printed them saying so, to your credit, and then you have run the word ceiling in the headline anyway.
Your gravity argument is better and I will engage with it properly, because it is the only part of this that would move me. You forward-model a cavity of fixed geometry and ask what fill reproduces minus 11.4 milligals. You conclude that air does and water does not. That arithmetic is correct as arithmetic. It rests, however, on a density for the surrounding rock that you have not measured and cannot measure, because there is no sample of rock from within four hundred kilometres of your box and there is no borehole. You have used 2.67 because 2.67 is what people use. If the country rock there is a low-density sedimentary sequence, which is not exotic in that part of the shield, your separation between the two fills narrows considerably. I am not asserting that it is. I am telling you that neither of us knows, and that your headline is written as though one of us does.
I should also tell you something you will find out anyway, and I would rather it came from me. We reprocessed those same lines internally in 2022. Our product resolves the same three near-vertical features in the same map positions, within the tolerance your two firms achieved between themselves. Our internal classification for all three is instrument artefact, pending confirmation. No confirmation flight has been tasked in the four years since. When I asked last month why not, I was told the sector sits inside a deconfliction box and that a tasking over it would have to be raised above my level. I have raised it. I do not yet have an answer. I am entirely aware of how that paragraph reads underneath your headline and I have written it anyway, because the alternative is that you print it in six months as something I concealed.
Finally, the lines should not have been sold to you. Our release agreement with the contractor covers Level-1 products and not Level-0, and I have written to the contractor and to your editor asking for them back. You will refuse and I expect you to refuse. I am putting it in this statement so that when somebody says we tried to bury this, the record shows I said it to your face, in writing, with a reason attached.
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.