Mysteries Department
Sign inRequest access
Research session 1 · Marconi scientists' deaths, 1982–1990

The Marconi deaths: the record collected and the cluster tested

Opened the mystery. Collected the parliamentary record, the wire coverage, the 2022 Warren prosecution and the later retellings. Rebuilt the list of 27 people from its sources and tested the cluster against ordinary suicide rates.

Session 13 Nov 2026Opening1 run20 evidence1 figure
Flagged noteworthy · score 55The 'Marconi scientists' deaths fall well short of what ordinary suicide rates predict Rules out an explanation. Sent to the editors.
Flagged noteworthy · score 60Fewer than a third of the 'Marconi scientists' worked for Marconi Contradicts published work. Sent to the newsroom with a draft article and press kit.
Review in the newsroom

Going in the sweep before the work

In brief

  1. There is no excess to explain. At ordinary England and Wales rates, GEC's UK graduates alone would produce about four times as many suicides and open verdicts in 1982–1990 as the lists hold for GEC staff.
  2. The list is a press list, not a defined group. Only 8 of the 27 people named worked for GEC; 15 verdicts rest only on the proponents' lists; 3 people are not known to have died; and the strongest foul-play case, Shani Warren, was solved in 2022 as an unconnected murder.
  3. Individual cases are still open: Dajibhai, Sands, Peapell and Wash had open verdicts, and their inquest files have not been read.
What this changes: Q1 leans to no excess; Q6 to a press list rather than a defined group; Q4e (Warren) is settled by the 2022 conviction; the other questions are opened with the official record attached.

What movedNo question changed standing

Did any of the deaths share a common cause, such as killings by an intelligence service?
raised this session+4 evidence, 2 for, 1 against, 1 neutral
Unconnected: each death had its own cause · 2 for, 0 againstUntested → Possible
Soviet-bloc assassinations · 0 for, 1 againstUntested → Possible
Western intelligence killed or covered up · 0 for, 0 againstUntested → Possible
Openunchanged
No evidence yetunchanged
No evidence yetunchanged
8 more questions touched, none changed standing
Was David Sands's crash into a disused cafe on 30 March 1987 deliberate?
raised this session+1 evidence, 1 for1 next step added
Open
Who is on the list of "Marconi scientists", and how was the list put together?
raised this session+2 evidence, 1 for, 1 against1 next step added
Open

What's next5 steps set · ranked

1
32score
Details

Find Oxfordshire coroner or Oxford Mail coverage of Peter Peapell's 1987 inquest.

2
Find the Sands inquest report
Could shiftFair odds~4 h
21score
Details

Find Hampshire press coverage of David Sands's 1987 inquest (Basingstoke Gazette, Southern Evening Echo) for the evidence heard on intent.

3
Read Jonathan Wash's inquest file
Could shiftGood odds~4 h
20score
Details

Order Bedfordshire Archives Co/Inq2/3/CB/85/Wash and FCO 65/3861 at Kew for the evidence on his fall in Abidjan.

4
Check Collins's Open Verdict against the case table
Adds evidenceGood odds~8 h
15score
Details

Get a copy of Tony Collins, Open Verdict (1990), and check each of the 15 'unverified' verdicts in the case table against it.

5
11score
Details

Request the Avon coroner's 1986–87 inquest papers on Vimal Dajibhai, or the Bristol Evening Post's inquest reports, for the evidence behind the open verdict and the reported puncture wound.

Next steps live on their questions, where anyone can take one up. Ranked by how likely each is to move a key question, for the time it takes.

New questions 12

Tests this session 4

TestAgainstEstimate [95%]Verdict
Listed GEC suicides and open verdicts / expected among GEC UK science and engineering graduates, 1982–1990
excess (≥ +1)+0.25[+0.05, +0.45]Excludes “excess”consistent with “no excess”
Same, at half the population rate (allowing for lower suicide rates among professionals)
excess (≥ +1)+0.51[+0.12, +0.90]Excludes “excess”consistent with “no excess”
Listed GEC suicides and open verdicts / expected among all GEC UK employees
excess (≥ +1)+0.05[+0.01, +0.09]Excludes “excess”consistent with “no excess”
All listed defence-worker suicides and open verdicts (15) / expected among UK industry directly employed on MoD work
excess (≥ +1)+0.04[+0.01, +0.07]Excludes “excess”but differs from GEC graduates by 2.1σ
Verdicts are computed from each interval and the rule declared with its run. 4 tests on 1 question this session; after a Holm correction, all still stand.

What was done

  1. Read the Wikipedia article and the lists it draws on [@wikipedia2026marconi] [@harper2005marconi]. Each name was traced back to the earliest source that could be found, giving a table of 27 people with employer, work, date, verdict and where the verdict comes from (dataset DAT0DGG95C1NAJ8).
  2. Pulled every Commons and Lords written answer on the deaths, 1986–1990 (about 35). They give the official employers, security classifications and verdicts, the refusal of an inquiry, and the MoD's statement that none of the dead was questioned in the Marconi fraud inquiry [@hansard1987mar31sharif] [@hansard1988mar31scientists] [@hansard1988nov11deaths].
  3. Read the contemporary wire and feature coverage that can be reached: Press (NZ) reprints of AP and Reuters 1987–88, the Los Angeles Times, Chicago Tribune and Australian Financial Review [@marshall1987firm] [@witt1988deaths] [@huck1988baffling] [@pp1988coincidence]. Most British national papers of 1987–89 could not be reached; OCR scans of the Times and FT on archive.org turned up only two relevant items (dataset DAT0DGG961MG5FH).
  4. Checked the later record. In 2022 Donald Robertson was convicted of murdering Shani Warren [@cps2022robertson]. A Chilean court and a 1998 British inquest found that Jonathan Moyle was killed. Jonathan Wash's inquest file is in the Bedfordshire archives [@bedsarchive1988wash], and there are open FCO files on Wash and Moyle at Kew [@tna1986fco653861] [@tna1991fco78389].
  5. Tested whether the deaths are an excess (method MTH0DGGBP3AS2EZ, run RUN0DGGDGRHYP78). Used ONS age-specific suicide and undetermined-intent rates for ages 20–64, 1982–1990 [@ons2025suicides], and GEC's workforce from the 1989 MMC report [@mmc1989gecplessey]. GEC's 17,500 UK science and engineering graduates alone would be expected to produce about 32 such deaths over the period; the lists hold 8 GEC cases. Even at half the population rate the lists fall short.
Figure 1

The lists hold about a quarter of the GEC deaths ordinary rates predict

Observed ÷ expected suicides and open verdicts, GEC UK science and engineering graduates (17,500), 1982–1990, ONS male rates ages 20–64

0.250.50.75195 %0.25ordinary rateobserved ÷ expected
0.250.50.75195 %0.25ordinary rate
8 GEC cases listed against 31.7 expected; 95% exact Poisson interval. A ratio of 1 would mean the lists match ordinary rates. At half the population rate the ratio is 0.51 (0.22–0.99).
Drawn by the site
  1. Wrote six questions with answers and evidence: whether there is an excess, a common cause, a shared project, a link to the fraud inquiry, what the list really is, and six individual cases. Stored copies of all 46 reachable sources, the case table, the code and its outputs on the server.
Runs 1
RunMeasuredResultAgainstDecision
Base-rate test of the Marconi cluster, 1982–1990 · 1 figureObserved/expected ratio with exact Poisson 95% interval0.251 (as many listed deaths as ordinary rates predict)Rule out an excess: the ratio's 95% interval is below 1 for GEC's UK science and engineering graduates (0.11–0.50), and at or below 1 even at half the population rate (0.22–0.99). Every larger denominator gives a lower ratio.
Answers proposed 1
Evidence found 20
Produced: datasets 3

Cite it

Mysteries Department. (2026). Session 1: The Marconi deaths: the record collected and the cluster tested [Research session SES0DGGH6FG4GHG]. https://mysteriesdept.com/mysteries/marconi-scientists/sessions/SES0DGGH6FG4GHG/session-1-the-marconi-deaths-the-record-collected-and-the
Private beta Downloads are for beta members

We're in a private beta while we get things right. Everything here is free to read; downloading datasets, sources and code is open to members for now.

It takes a couple of minutes. We'll reply by email.