The figures, source by source

Compiled 26 July 2026 and checked against the 152 sources. Purpose: to record what each figure was, who stated it, when, and on what stated basis, so the record remains usable as figures are revised. Nothing here is settled; everything is attributed.

1. How many people were aboard

Other figures in this record depend on this one.

Certified capacity. Three inconsistent versions appear in the corpus, all attributed to officials. No published certificate appears in the corpus:

Cargo actually carried: 268 tons manifested, with Edghill's own line-by-line audit returning 260 tons "early in the day" on 19 July.

Thomas cannot say when or on what basis the passenger rating rose from 150 to 394 — "I don't have that information."

What the sources record. Only 35 of the first 67 survivors appeared on the manifest, per Edghill on 19 July — so thirty-two people recorded as rescued were not recorded as having boarded. DPI releases of 19 July state that families were being updated using the vessel's manifest. The figure of 179 is attributed to a review of CCTV footage rather than to any paper record. A Demerara Waves column of 22 July noted that rescued plus recovered plus missing already exceeded 170 while the official complement stood at 133.

Contested The Jamaica Gleaner, 20 July, inverts the manifest finding — reporting that at least 35 of the 67 rescued were not on the manifest. Every Guyanese source has it the other way. Treat the Gleaner rendering as an error, but note it, because the inverted version circulated internationally.

AP supplies the mechanism the official account omits: ticket clerks had sold unmanifested seats for cash "for decades at state-run ferry terminals." This is not a one-night lapse but a standing informal practice.

2. Survivor figures and their revision

8 → 53 → 67 → 69 → 77 → 76, where it froze.

Two outlets reported a reason for the reduction. News Room attributed it to de-duplication across uncoordinated reporting channels; Kaieteur described the same verification and reconciliation process taking 77 down to 76. Demerara Waves reported that ten people recorded as found overnight Monday were already at home.

3. The dead — two parallel sequences

Official (bodies recovered), attributed almost entirely to PM Phillips rather than police or coroner: 2 (20 Jul, Edghill) → 10 → 27 (20 Jul, 18:30) → 41 (21 Jul, 14:00) → 53 (21 Jul) → 65 (22 Jul) → 72 (23 Jul) → 73 (24 Jul), unchanged through 26 July. Of the 73: 69 identified, 63 released to families.

Unofficial and inferential: - 103 — Kaieteur News, 22 July. Not a government figure and not a claim by the Opposition Leader: it is Kaieteur's own tally compiled from three days of search work by the Coast Guard, Azruddin Mohamed, private vessel owners and local fishermen. Mohamed himself said only that "it looks like over 100." The state reported 53 the same day — a fifty-body gap. - ~100 — AP, 22 July, arithmetic: 179 minus 76 survivors. Echoed by Al Jazeera on 23 July while that outlet still carried 53 recovered, so an inference and a body count coexisted in international copy for about forty-eight hours. - "over one hundred" — WIN MP Sweatnam. - "unofficially more than 100" — Demerara Waves.

Missing: 83 (20–21 Jul, Phillips, and cited by Bulkan) → 30 unaccounted for (24 Jul, Phillips). Guyana Times on 24 July says "at least 31." Nobody reconciles 83 against 30.

Figures not published. 179 aboard, less 76–77 survivors, leaves 102–103 dead. Seventy-three bodies have been recovered. The state reports recoveries and never the residual, so the roughly thirty people still in or near the wreck have never appeared in an official total. Al Jazeera's 21 July piece is a live-updated artefact that preserves two tolls at once — the slug says 41, the headline says 53.

4. Was she seaworthy?

The state's position, stated precisely. DPI's headline of 20 July is "MV Barima was seaworthy – Min Edghill," but Edghill's own words are narrower: "There was no report of any kind that suggested the vessel was unseaworthy." That is an absence-of-reports claim, not a survey finding, and the distinction should be preserved in any citation. On the machinery he was more categorical — "There was no engine problem or any mechanical problem" (19–20 July), and "there was no reported engine troubles from the crew when the boat came in and when the boat left" (22 July). MARAD's Thomas, 22 July: "not overloaded," citing the load line, and stating that age is not determinative because modern thin steel is stronger.

What sits against it, from the state's own record: - DPI, 8 September 2020: a release recording the Barima's history of mechanical failures. ⚠️ Do not quote. That file is a tool-condensed capture whose own note states only the Edghill quotation inside it is verbatim — and that quotation is about engaging India, not the vessel. Treat the mechanical-failure characterisation as the substance of a government release, not as anyone's words. - DPI, 20 September 2024: repairs pending, "more than 80 years of service." - DPI, 10 March 2025: the T&HD has "longstanding issues of negligence, mismanagement, and accountability." (Verbatim, file unflagged — the strongest state admission in the corpus.) - March 2026: a tender putting a further G$124.5m into the hull, four months before the sinking. Per unbylined CMC copy in the Jamaica Observer (25 July), voiced by APNU MP Saiku Andrews, the bill of quantities required examination of about 10,000 m² of hull plating and frames "with deteriorated sections to be renewed where necessary." Read the conditional carefully — this is a specification anticipating deterioration, not a survey establishing it. It has been reported as though it were a finding. - Capt. R. E. W. Adams, Stabroek News, 2 October 2015, on this named vessel: deck, winch and generator area "a mass of junk and rust," and such a ship would be barred from sea or scrapped elsewhere in the Caribbean.

What the corpus does and does not contain. The seaworthiness statements in the corpus rest on a docking chronology — last docked 2024, next due latter half of 2026. No source in the corpus names a class certificate, a passenger safety certificate, or a load-line survey date. The "not overloaded" statement rests on a capacity figure that MARAD's Director-General said he could not trace back from the 1938 rating, and which he and Edghill stated differently. Neither the claim that she was overloaded nor the claim that she was not can be checked against a published document, because no certified capacity has been published.

On the machinery, the 2017 rehabilitation is described as including "installation of two engines"; whether the original Paxman engines were entirely replaced is not established in the corpus. The figure of 87 years is therefore supported for the hull and not for the machinery.

5. Proximate cause — three incompatible accounts

Official: rough seas at a known-hazardous location compounded by crew impairment. Thomas's stability argument — a vessel with positive stability rights itself even at twenty degrees of heel, so failure to right indicates human operational error.

Survivors: a progressive loss of stability over four hours. Leon Murray (AP): "packed, packed, packed all downstairs"; one engine lost at about 19:00 and repaired at sea rather than returning to port; a water-ingress warning at about 20:00 answered with "They told us this is normal." Elena Moonsammy is the sole source for cargo stacked at the stern. Wayne Kitson describes a list that failed to right. Thomas's stability explanation and the survivor accounts of loading were not reconciled in any source in the corpus.

Systemic: an 87-year-old hull kept on a route with no alternative while a US$12.7m replacement idled, and a second vessel, the Kalliopi N, sat uncommissioned at the Kingston Goods Wharf since May awaiting a presidential ceremony.

Sinking duration Contested: about four minutes (Murray) versus "less than a minute" (Guyana Times).

Grounding Single source: Janette Bulkan states the vessel became stuck on a mudbank offshore Iron Punt. No official account mentions it.

6. Rescue response

The Mayday itself is contested, and this is easy to miss. A 23:01 Mayday is the spine of every published timeline in the corpus. But on 25 July Prime Minister Phillips said of a Mayday from the captain: "We have no such record." Whether the vessel transmitted a Mayday, whether it was logged, and whether "received by aviation authorities" and "recorded by the state" are the same thing, are all open.

Gouveia was airborne at 01:36, two and a half hours after the reported Mayday, in a fixed-wing Islander. Bulkan asks why the rescue helicopters and trained swimmers at the Omni Helicopters base at Ogle never launched. Two GDF helicopters were reported unserviceable — raised by the opposition via News Americas Now (19 July) and addressed by GDF officials via Kaieteur (23 July).

Who found the wreck [CONTESTED, possibly sequential]: fisherman Haresh Singh located the hull and directed the Coast Guard to fire flares (Guyana Times); Roraima Airways — private — reported the coordinates (Demerara Waves); ExxonMobil seabed mapping and sonar located it (UPI). These may be stages of one process rather than rival claims, but no source sequences them. Single source The detail that Exxon sent three offshore supply vessels and that divers came from "Exxon or the French" is Bulkan's alone.

Pan Pan Unresolved: whether an urgency call preceded the Mayday by four hours is asked by Bulkan and answered by no one. Her letter is also the only source in the corpus for Guyana's SOLAS accession on 10 December 1997, effective 10 March 1998 — a checkable fact that has not yet been independently checked.

Demerara Waves reports that, except for the GDF, the government had no crisis-communications response on the first night. The qualifier matters.

7. Why the replacement never sailed Contested

• Wharf works: PM Phillips, 25 July — delayed Port Kaituma wharf works meant the Ma Lisha could not run the route.
• Cargo capacity: residents and contemporaneous reporting — the vessel could not carry the route's cargo, so the Barima stayed and was re-tendered.

The two accounts are not mutually exclusive: one describes incomplete wharf infrastructure, the other a mismatch between the vessel's specification and the route's cargo. The Ma Lisha was commissioned in 2023, and Parliament was told in 2023 that it would replace the Barima and the Kimbia.

8. Compensation and funeral costs as reported

The reported figures are funeral-expense figures, not compensation. APNU's Saiku Andrews: "It starts with an assistance for the funeral expenses and I believe that there is no less than $1M…", and some families "have been offered a mere $300,000 for funeral expenses." Phillips's position on the wider question is sequencing rather than legal characterisation: "The focus now is not on compensation. The focus is on recovering bodies, financing the burial and funeral rites…" The corpus does not contain a statement by any official characterising payments as "not compensation"; that wording is a compression of the quotation above. Separately, Edghill confirmed on 22 July that the state assumed the risk on an uninsured vessel, which means any eventual compensation is a treasury matter rather than an insurance claim.

9. Smaller discrepancies worth tracking

10. Claims resting on a single source

Flagged because they are consequential and uncorroborated. Each is a priority for verification.

Bulkan's letters carry an unusual share of the load: the mudbank grounding, the unanswered Pan Pan question, the SOLAS accession dates, and the Exxon vessel and diver detail. Demerara Waves alone reports the ten already-home "survivors" and the absence of a crisis-communications response outside the GDF. Guyana Times alone has Haresh Singh instructing the Coast Guard to fire flares. Elena Moonsammy alone describes cargo stacked at the stern. The March 2026 hull-plating specification reaches the corpus only through unbylined CMC copy voiced by an opposition MP. The uninsured-vessel disclosure and the presence of machinery and ATVs aboard come through Kaieteur and HGPTV. Venezuelan families disputing earthen burials is Kaieteur alone. Patterson's claim that expert procurement advice was disregarded is post-disaster and unverified. Aporrea alone frames the sinking as having occurred in Venezuelan-claimed waters. The vigil detentions' resolution — charges dropped, apology given — rests on the activists' own statement; no police or government statement appears in the corpus, and the event was a government national prayer service rather than a candlelight vigil.

11. Reliability guidance for this corpus

Earliest or most directly sourced: the Paxman History Pages engine register (pre-disaster, apolitical, corroborates build year, yard, ordering chain and dimensions independently of all disaster coverage — every figure in it was verified); DPI's own pre-disaster releases, which are admissions against interest; the 2015 Adams letter; and AP's original on-the-ground reporting (Drake in Charity, Wilkinson in Georgetown), which is the origin of most survivor testimony everyone else recycles and the only source for the decades-old practice of clerks selling unmanifested seats.

Handle with care: 592Hub, an aggregator and the sole source of several specification figures. Any Guyanese outlet's account of an opposing party's motives. Figures attributed to "officials" without a named speaker. Note that Elena Moonsammy (Guyanese press; lost seven children she travelled with) and Helena Moonsammy (AP and Gleaner; a grandmother who lost four grandchildren and spent nine hours in the water) may be two different people — the archive does not establish that they are one person with variant spellings, and earlier drafts of this analysis wrongly assumed so.

Files carrying a capture_fidelity, fidelity or verbatim: false warning must never be quoted as verbatim; their direct quotations are reliable, their connective prose is the retrieval tool's compression. This applies to the 2020 DPI release, the Christena Wikipedia capture, and the vigil-detentions report, all of which are cited above with that caveat attached.

Absent from the corpus entirely: any published certificate of survey, load line or passenger capacity; any Auditor General finding specific to the T&HD; any trade union statement; any gazetted terms of reference for the Commission of Inquiry; any IMO or PAHO statement; any MARAD advisory or notice to mariners; and any reference in either direction to charges being laid or an inquest opened. No BBC byline surfaced. Reuters did contribute — Al Jazeera's 21 July piece is bylined "Al Jazeera Staff, AP and Reuters" — so treat the wire coverage as present but not directly captured.