We Paid Three Groups To Fit The Ringing. All Three Needed A Cavity.
Nine long-period events, three seismology groups, one criterion registered before any of them saw data. No solid interior reaches forty minutes of ringing. All three fits converge on a rigid shell and a void beneath it.
SILVER SPRING, Md. — No model of a solid Moon rings for forty minutes.
Nine events in the public lunar record ring for longer than that. One of them rings for three hours and fifty-one minutes.
This newspaper paid three seismology groups to close that gap. They were kept apart, paid the same fee in advance, and worked against a criterion we registered in writing on 14 April, before we had approached anybody: reproduce the measured attenuation and the shape of the coda envelope, first arrival to noise floor, to within ten per cent on at least seven of the nine events.
All three groups reported that they could.
None of them could do it with a solid interior.
What they were given
Nine events, drawn from the long-period catalogue this paper spot-checked entry by entry in July and holds as PX-1788.
1969-311-B 1970-127-A 1971-062-C 1971-284-A 1972-095-A
1972-201-B 1973-158-A 1975-330-B 1976-044-A
selection rule: the longest event in the catalogue, then eight drawn at
random from entries over ninety minutes. Rule written before the draw.
Each group received the raw traces, the station geometry, and nothing else. None was told how many other groups there were. Two assumed one. There were two.
The criterion, and the day we wrote it
The criterion matters more than the result, because it is the only thing standing between a commissioning newspaper and the answer it would like.
We also wrote down what we were not permitted to do. We could not exclude a model class. We could not see interim results. We could not withhold a fee.
Solid-throughout interiors were in every group's model space from the first hour, and all three spent the bulk of their compute there.
Where the solid Moon stops
Between them the groups ran 11,900 solid-interior models. Fractured, dry, cold, layered, brecciated to two hundred kilometres — every arrangement a rock can be in.
The best of the 11,900 rings for thirty-eight minutes.
Scattering in dry, broken rock is a real mechanism and it does real work. It is the explanation the catalogue's own compiler has argued for nine years, and it is the explanation this paper printed in July. It gets you to about forty minutes.
It does not get you to two hundred and thirty-one.
What the accepted fits require
All three accepted fits have the same shape: a rigid outer shell, and beneath it a large volume of very low density.
The shell comes out at 66 kilometres on one fit, 62 on another, 71 on the third.
The volume comes out at 1.2 billion cubic kilometres, give or take fifteen per cent across the three. That is about five and a half per cent of the Moon. Gathered into one sphere it would be thirteen hundred kilometres across.
Three groups who have never spoken arrived at the same geometry from the same nine traces. The Moon rings for three hours and fifty-one minutes because there is enough nothing inside it to ring for three hours and fifty-one minutes, and that sentence now has three separate derivations on the record.
The man who had to change his mind
We drove back to Maryland in July with the reports in a folder. Ellery Vaughan-Kimathi, who compiled the catalogue and who told this paper last month that the Moon "is a bell because it is broken and dry, not because it is empty," read all three at the same kitchen table over two days.
He asked hard questions about the envelope on the fourth event. He argued about where the noise floor sits. On the second afternoon he said the sentence above, and asked us to print it in the same place we printed the other one.
That is this paragraph. It is the correction he is owed and the one he asked for.
What is still shut
Dr. Marguerite Ansell-Duplessis, who runs the office that holds the archive, has replied at the foot of this page and her reply is worth more than our headline. She says the noun is ours. She is right that it is: the reports bound a density, and we wrote "void". She is also exact about what that bound rules out — ice, broken rock at any porosity she would defend, every mantle composition in the literature — and about what it does not settle, which is whether the space is empty or holds something at better than ninety per cent void.
She also says, in the same letter, that a working group inside her office reached a comparable shell figure in 1998 and never published it.
The nine events are all near-side. Four far-side records exist and have never been read by anyone now living. We applied on 2 June, were refused on 19 July, and appealed on 11 August.
The shell has been measured on one hemisphere. The next locked door is the other one.
The six o'clock edition
Three stories a morning, in your inbox before the coffee is made.
Sources & Method
We registered the acceptance criterion in writing on 14 April, before approaching anybody, and had a seismologist with no stake in the outcome try to break it first. We then paid three groups the same fee in advance, sent each the same nine raw traces and nothing else, and undertook in the commissioning letter to publish every report whether or not it met the criterion and whether or not the three agreed. All three reports, the failed runs and the criterion with its date are published with this story.
Who we spoke to
- Dr. Ingrid Kirkbride-Osei, Planetary seismologist; led the first of the three groups. Commissioned in writing on 21 April; report published entire, including her objections to our criterion April–July 2026 Accepted fit: shell 66 km, interior volume 1.31 billion km³. Thinks our ten per cent tolerance was too generous and says so in her covering letter.
- Dr. Rui Tanaka-Ferreira, Seismologist; second group, working blind. Given the identical nine traces with no sight of any other report and no contact with any other group April–July 2026 Accepted fit: shell 62 km, interior volume 1.09 billion km³. The thinnest shell and the smallest volume of the three.
- Prof. Halvard Ostrowski, Geophysicist; third group, working blind. Same traces, same terms, same fee; told only that at least one other group existed April–July 2026 Accepted fit: shell 71 km, interior volume 1.24 billion km³. Ran the largest solid-model sweep — 6,400 of the 11,900.
- Ellery Vaughan-Kimathi, Retired network engineer; compiler of the catalogue filed as PX-1788. Interviewed in person at his home over two days, with all three reports in front of him 28–29 July 2026 Has argued for nine years that the ringing is scattering in dry, broken rock. Changed his position on the second afternoon and asked us to print that he had.
- Dr. Josefina Marek-Anand, Planetary seismologist; reviewed our criterion before it was registered. Sent the draft criterion on 2 April and asked to try to break it; four of her seven objections were adopted April 2026 Took no part in the fitting and was paid nothing. Her three rejected objections are printed with the criterion.
Documents
- PX-1788 — Lunar seismometer network, long-period event catalogue, unofficial compilation accepted
What we could not confirm
- Whether the volume is one cavity or many. All three fits model a single connected void because a single void is cheap to compute. Two of the three groups told us, without being asked, that a set of smaller volumes summing to the same total would satisfy the coda equally well, and neither had tested it. We do not know whether the Moon has one hollow inside it or forty.
- What is in it. Every fit here constrains a density — below roughly 0.3 grams per cubic centimetre — and a density is not a contents list. It is a severe bound and it does not admit the ordinary answers: water ice is three times too heavy for it, and so is broken rock at any porosity that has ever been measured in a laboratory. Vacuum satisfies it. So would a foam or a lattice at better than ninety per cent void fraction, and nobody has proposed a material that holds one of those open across thirteen hundred kilometres. This newspaper has written 'void' throughout because the three reports do, and the reports mean the bound.
- The far side. All nine events were recorded by near-side stations. Four far-side station records exist in the archive index and have never been released; we applied on 2 June, were refused on 19 July on grounds of media condition, and appealed on 11 August. Until they are read, the shell has been measured on one hemisphere and inferred on the other.
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 Moon Is Hollow And Somebody Has Known Since 1971
Took the shell figure straight from our chart and filled the interior with vacuum, which no report in the commission does.
Had that we did not
The 62-to-71-kilometre band, correctly quoted, with our groups named.
Left out
That all three fits constrain a density and none of them constrains a contents. Vacuum satisfies the bound. So would a structure at better than ninety per cent void fraction, which is not a solid Moon either.
-
Signal & LedgerBlind Commissioning Arrives In Planetary Geophysics: What It Costs
Read the exercise as a procurement story and priced it out for readers who buy research.
Had that we did not
Our registered criterion, reproduced in full and laid out more clearly than we laid it out.
Left out
The result. Nine hundred words on how we bought it and forty on what we bought.
-
They Buriedthis newspaperWe Paid Three Groups To Fit The Ringing. All Three Needed A Cavity.
Registered a criterion, paid three groups who had never met, and printed every fit including the ones that failed.
Had that we did not
All three reports entire, the 11,900 failed solid runs, and the criterion with its date on it.
Left out
Our headline says we paid. It does not say who paid us. Lunar seismology is one of three subjects named in the letters accompanying an endowment we cannot trace, and that sentence sits in a disclosure at the foot of this page instead of at the top of it. — V. Ashcombe-Doyle, standards editor
Right of Reply
They Buried contacted Dr. Marguerite Ansell-Duplessis, Director, Lunar Geophysics Coordination Office on 3 August 2026, in writing, with all three reports, the criterion and the full draft attached, and no deadline imposed. Replied 17 August 2026. Printed in full and unedited, including the parts about us.
I have read your three reports and I will say at the outset that they are competent work. That is not the same as saying they are right, and it is a considerable distance from saying that a newspaper ought to be commissioning them.
Begin with what you actually did. You bought nine fits to nine events against a criterion you registered in advance, which I grant is more discipline than most of what reaches this office. But a pre-registered criterion is only as good as the model space it is applied to, and you did not set the model space. Your three groups set it, separately, out of the same textbooks, the same public codes and the same twenty years of published parameterisation. Three people trained in one discipline reaching for the same shelf is not three independent measurements. It is one measurement made three times, and your convergence is a fact about their training before it is a fact about the Moon.
Second, on attenuation. You write as though Q were an observable. It is not. It is derived, and the derivation leans on an assumed source time function that a single station cannot constrain. Move that assumption within its own published range and the Q your groups were fitting moves by rather more than your ten per cent. Your tolerance is doing more work than your result.
Third, and I am aware you will print this entire, so I will say it plainly rather than have you find it later. The shell figure is not new. A working group inside this office produced a range for the rigid outer layer in 1998, by a different method, and it is not far from the range your groups report. It has never been published because that working group could not agree what to say about the interior beneath it, and in my judgement declining to publish a disagreement is not the same as concealing a finding. You will report it as a concealed finding regardless. I have told you because I would rather you had it from me than from somebody who wanted it to sound worse.
On the far-side records: they are not withheld. They are on a media format that has not been readable since the drives were scrapped, the recovery has been costed twice and funded on neither occasion, and I have wanted to read them for eleven years. If your appeal succeeds you will receive a refusal that says exactly this, at greater length, with a reference number on it.
What I object to is the noun in your headline. 'Cavity' is a decision, and not one of your three groups made it. Each of them bounded a density, and I am not going to soften what that bound does: below 0.3 grams per cubic centimetre excludes ice, it excludes fractured megaregolith at any porosity I could defend in front of colleagues, and it excludes every mantle composition in the literature. What it does not do is choose between vacuum and a structure — a foam, a lattice, a cellular fabric at better than ninety per cent void — and that choice is the whole of what is left to find out. You have taken a bound and made it a hole. The hole may in the end be the right reading of it. It is still your word and not theirs, and it is the only word your readers will keep.
Print the density. You will print the hole. I have written this knowing that.
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.