Eleven Days To Dig Out. The Requirement Was Twenty-Four Hours.
Four egress studies and nine trial runs at a quarry test bed. The planning figure was 27 metres an hour through shot rock. The machine managed 3.1, and broke twice.
A months-long reporting project. Documents cited below are held in The Vault and available to readers.
PEAK FOREST, United Kingdom — The requirement was twenty-four hours. It is stated once, in a 1978 staff paper, and it was never relaxed.
Nine trial runs at a quarry test bed in Derbyshire, in the autumn of 1983, produced a mean advance of 3.1 metres an hour through shot rock, and two broken machines.
At the rate the machine actually managed, counting the hours it stood still, the crews would have reached daylight in eleven days.
What the whole thing rested on
The proposal was to put launchers under 900 metres of rock. Everything else in it — the cost, the shift patterns, the argument for building it at all — followed from one assumption: that after an attack, the crews could cut their own way out through the collapsed access drift within a day.
The modelled debris column is 648 metres. Divide by twenty-four and you get 27 metres an hour.
The 1979 planning figure was 27 metres an hour.
We asked where that came from and the answer is in the study's own footnote: the logged advance of a roadheader in a bedded coal seam in Nottinghamshire in 1976. For four years the requirement was justified by the figure and the figure was justified by the requirement, and nobody had cut anything.
Four studies
There are four. The 1979 planning study, the 1981 study that first derived a rate from broken ground rather than coal and recommended trials, the 1983 trials report, and the 1985 review.
Three came to us on internal review after a two-year appeal. The fourth, the 1981 study, was bought for £40 at a house clearance in Matlock with the executor's consent, and it is the one that says the quiet part: that no figure then in use had been obtained in fragmented ground.
What the machine did
Three beds of graded shot rock. Nine runs, between 12 September and 25 November 1983.
The best single run was 5.2 metres an hour, in bed C, the most uniform of the three. The mean across all nine, counting cutting time only, was 3.1.
A cutter-head bearing went on run 3 and a hydraulic ram on run 8. Together they cost 104 hours. Include them, as the 1985 review did, and the rate is 2.45 metres an hour. Six hundred and forty-eight metres at 2.45 is 264 hours.
Dr. Innes Halvorsen-Achebe's group, working the same limestone today with modern plant and disc cutters, measures between 6 and 9 metres an hour and describes that as good.
The memorandum
The programme ends in two paragraphs.
It does not mention deterrence. It does not mention strategy, treaties, allies, or cost. It notes that the achievable advance rate in fragmented ground appears materially below the planning assumption, and asks the programme office for a revised figure by 30 April 1986.
There is no reply on the file. There is nothing further on the file at all.
What is still there
A 1,400-metre test adit, driven in 1980 and never used for anything else, leased since 2004 to a university rock-mechanics group for £18,000 a year. It is the only part of the programme that has ever paid for itself.
And the machine, under a canopy outside a town museum, on a bed of chippings, with an interpretation board that gives its weight as 42 tonnes. The contractor's own weight ticket, in the archive, says 61.4.
The thing we cannot settle
Two of the three engineers who ran the trials thought at the time that the number was a tool problem and would come right with a better head. One of them, Eirwen Trethowan-Baptiste, stopped thinking so at some point in the early nineties and cannot say when.
The other is Douglas Pethybridge-Rees, who is seventy-nine, who ran the machine, who told the programme in writing that the head was wrong for the material, and who still believes the rate would have doubled or trebled with the right one. His letter is below and it is the best thing on this page.
He may be right. We could not test it, because the machine is a museum exhibit and the drift it was built for was never cut.
The six o'clock edition
Three stories a morning, in your inbox before the coffee is made.
Sources & Method
We went after the rate rather than the programme. Four studies were identified from citations in one another's bibliographies, three obtained on appeal and the fourth bought from an engineer's estate; the contractor's trial sheets were read in a business archive and every one of the nine runs transcribed. We then asked a working rock-mechanics group what the same rock does today, and put the whole of it to the two surviving trial engineers.
Who we spoke to
- Egress studies of 1979, 1981, 1983 and 1985, Four studies on achievable advance rates in fragmented ground, commissioned by the programme office. Three released on internal review after a two-year appeal, annexes intact; the 1981 study bought at an estate sale for £40, with the executor's written consent Appeal decided March 2026; the estate purchase November 2025 The 1985 study is the only one of the four that states a rate including stoppages.
- Trial record, egress machine, beds A to C, runs 1 to 9, The prime contractor's logged trial sheets and two failure reports. Held in a deposited business archive; read over five days and photographed with the archive's written permission April 2026 Every run is timed to the minute. Both failures are written up by the men who were standing there.
- Douglas Pethybridge-Rees, Chartered mining engineer; one of three engineers who ran the 1983 trials. Interviewed at his house near Chesterfield three times, and sent the full draft with six weeks to answer February to July 2026 Still believes the rate was a cutter-head problem. His reply is printed at the foot of this page, entire.
- Eirwen Trethowan-Baptiste, Second of the three trial engineers; later a tunnelling contractor. Interviewed by telephone twice, and shown the trial sheets she had not seen since 1984 May and June 2026 Thought at the time that a better head would fix it. Says she stopped thinking so in about 1991 and cannot date it more precisely.
- Dr. Innes Halvorsen-Achebe, Reader in rock mechanics; leads the group that leases the test adit. Walked this correspondent 400 metres in; supplied the lease, the rent history and the group's own instrumented advance measurements June 2026 Her group's measured rates in the same rock, with modern plant, are in the body of this story.
What we could not confirm
- The 648 metres. Every hour in this story is divided into that number, and it comes from one annex to the 1979 study which states it without working. We asked for the collapse modelling behind it four times and were told twice that it was not held and twice that it might be. A different collapse geometry moves the eleven days in either direction and we cannot bound it.
- Whether a quarry can stand in for a bombed drift. The three test beds were shot rock of known and recorded size distribution, laid to a specification. The case the programme was designed for is rubble of unknown grading with steel, cable and concrete in it. The 1985 study says this could make the rate worse or better and then declines to say which, and neither can we.
- Mr Pethybridge-Rees's tenth run. He told us, and showed us a pocket notebook to support it, that the machine was rebuilt with a second head and run again at a different site in the spring of 1984 at better than nine metres an hour. There is no tenth run in the contractor's sheets, no 1984 trial in any of the four studies, and no site we could identify. We print his account because he gave it to us on the record, and we have not been able to stand up a word of it.
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 BUNKER THAT WOULD HAVE ENTOMBED ITS OWN CREWS
Took the eleven days as a fate that was concealed from the men who would have been sealed in.
Had that we did not
The 3.1 figure, correctly, and from the same trial sheets we read.
Left out
That the body which commissioned the trials is the body that stopped the programme on the result, in a memorandum it wrote to itself.
-
Signal & LedgerFragmented Ground: What Nine Runs In 1983 Still Tell Anyone Buying Tunnelling Plant
Read the trials as a procurement lesson about pick patterns and bearing loads, for readers who specify machines.
Had that we did not
The cutter-head failure analysis, in more detail and with better drawings than we managed.
Left out
The requirement. Their piece never states the twenty-four hours, which is the only reason any of these rates matter to anybody.
-
They Buriedthis newspaperEleven Days To Dig Out. The Requirement Was Twenty-Four Hours.
Set four egress studies and nine logged runs against a requirement that was written once and never relaxed.
Had that we did not
All nine runs, both failure reports, the two-paragraph memorandum entire, and the engineer who still thinks the number was fixable.
Left out
Our headline is a number a machine produced in a quarry, and we have let it stand for a place that was never built and men who were never in it. That belonged above the fold, not in paragraph fifteen. — V. Ashcombe-Doyle, standards editor
Right of Reply
They Buried contacted Douglas Pethybridge-Rees, chartered mining engineer, one of the three engineers on the 1983 trials on 2 June 2026 in person with the trial sheets, and again on 19 June by post with the full draft, the chart and six weeks to answer. Replied by letter on 24 July 2026, asking that it run whole. Printed unedited, including the parts about this newspaper.
You have written a very careful story about a number and no story at all about a machine, and I would like the readers to understand the difference, because it is the whole of my objection.
The machine we took to that quarry in September 1983 had a drum head with a pick pattern laid out for coal measures. I said so at the time, in writing, twice, and the second time I was told the trial was to be run with the head that had been bought. That is not a discovery you have made. It is in the file you obtained. It is on the third page of the 1983 study, and you have quoted the sentence after it.
So here is what your chart shows. It shows what one wrong head did in three beds of graded rock over eleven weeks. It does not show what the ground would allow. Those are different quantities and your headline treats them as one.
On your two failures. Run 3 was a bearing and run 8 was a ram, and I will not defend either, but you have taken 104 hours of standing about and divided it into the cutting time to produce a rate of 2.45 metres an hour, and then you have multiplied that rate by 648 metres as though a crew digging for its life would sit in the dark for 63 hours waiting for a fitter. They would not. They would have a spare ram, because anybody who has ever been underground carries a spare ram. Your 2.45 is an arithmetic that could only be produced by somebody who has not done the work.
I am seventy-nine. I have driven 41 kilometres of tunnel in my working life, a great deal of it in rock that nobody would have described as favourable, and I am telling you as plainly as I know how that the difference between three metres an hour and nine is the head, the pick spacing and the water, and that in 1983 we had one of those three right.
Now. Your reporter came to my house three times, ate my wife's cake, and wrote down that I could not produce the 1984 sheets, which is fair, because I cannot. They existed. I saw them. I have the notebook and you have photographed the notebook and I understand that a notebook is not a record. I do not ask you to believe me. I ask you to say, somewhere a reader will reach it, that the man who ran the trials says the trials were run with the wrong tool and was told so at the time by nobody, because nobody was listening.
The last thing. You end on the interpretation board at the museum getting the tonnage wrong, and it is a nice ending and I laughed at it. But the board is wrong by nineteen tonnes and your story is wrong by ten days, and only one of those two errors has a newspaper behind it.
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.