23 October afternoon · 28 October morning · 2026
The rescue picture
becomes clearer.
A family’s account, two rescue operators, a shore worker and another passenger fill in more of the night’s aftermath. They also show why a careful account must keep each person’s observations and limits in view.
On this pageWhat we know now
The evidence in context
More of the response. Important limits on the collision account.
The journeys after impact connect.
Nemius Joseph, known as Nemo, describes moving people from a yacht to Ti Kaye, then collecting Tyson and Dylan. He says Tyson asked to return to the damaged boat with a light, while Dylan travelled with him to Soufrière.
Police presence enters the record more clearly.
Nemo reports two Marine Police vessels. Larry John reports seeing a police vessel searching. These accounts add evidence of presence; they do not establish dispatch times or the adequacy of the whole response.
Not seeing lights is not always observing an unlit boat.
John explains that he was inside and not facing the sea until the crash. The new passenger repeatedly says he does not know whether his own boat’s lights were on. Both limits belong beside the lighting evidence.
The family testimony also adds a reported injury concern and a further account of notification and burial costs. Together, the files deepen two questions: how the collision was not avoided, and how rescue, care and family support worked afterwards. They provide no final Commission finding of fault.
23 October · Afternoon family testimony
A brother’s life, and a family seeking news.
The sister of the deceased referred to in the record as Mali or Marley William describes a quiet, loving man devoted to his daughter and family. Her own name appears as Zoya Jamali in the automated self-identification, with other surname variants elsewhere. The spelling remains under review.
She recalls learning of the collision at about 3 a.m. through relatives, with information relayed by a friend overseas. She describes limited information reaching relatives in Vieux Fort, and says she was told on Wednesday, 22 July that her brother had been found. She then describes identifying him at hospital.
That adds a family notification and identification date to compare with records. It does not settle the earlier recovery-date discrepancy: being told, recovery, identification and postmortem are different events.
Support has practical costs.
She says relatives pooled money for the burial and describes the weight of several family bereavements. A commissioner proposes counselling or debriefing. The exchange records a proposed offer, not proof that a referral or service followed.
Survival also raises a medical question.
She reports serious head injuries that she attributes to autopsy findings. We have not examined that report. Her account gives a reason to seek medical evidence alongside evidence about swimming, flotation and rescue. It cannot establish the precise cause of death, what produced the injury, or whether a particular precaution would have changed the outcome.
This is family testimony about her brother and the aftermath. It adds no direct view of impact. Read alongside the three morning family accounts →
28 October · What each witness could know
Follow the observation, not just the answer.
He heard the crash. He did not watch the approach.
John recalls engine noise and a crash. From sound, he thought a boat was coming south and moving fast, but he could not reliably tell whether he heard one vessel or two. Those are sound-based impressions, not measured speed or visual identification.
Some questions invite the conclusion that he would have seen any vessel lights. Later questioning establishes that he was on break inside the restaurant, not facing the sea. The crash made him look outside. His later answer that he saw no lights “before” the collision sits uneasily with that sequence and needs recording review.
John describes staff contact by radio, help at a floating dock and care in the restaurant. His radio was used between staff; the account does not establish marine distress-channel capability. He says no ambulance reached the resort, and confirms a questioner’s account of a roughly 90-minute wait followed by onward boat transfers. Dispatch and resort records are needed to test that sequence.
A further account of the handover, with careful limits.
The passenger says Tyrone initially operated Ocean Angel and Dylan later asked to take over. He places this after Canaries, whereas the earlier account placed the handover near Anse Mamin. Both are recollections; we retain the difference rather than choose a single location. His estimate of about fifteen minutes concerns the period from handover to impact, not an advance sighting of the other boat.
Speed increased, in his account.
He says Dylan used a little more speed, but adds “not too fast.” He describes bouncing, later distinguishing it from slamming, and says waves were not entering the boat. Those qualifications should travel with any summary of increased speed.
He did not see the other vessel approach. At first he thought they might have struck a rock and says he learned at hospital that another boat was involved. He explicitly says he does not know whether Ocean Angel’s lights were on.
Briefing, jackets and visibility remain central.
He says he received no safety briefing and wore no life jacket, following the others who boarded without one. He recalls everyone remaining seated after the handover, with Tyrone moving forward. This adds an observation about people’s positions, not proof of a formally assigned lookout or a measured view from the helm.
He describes reaching floating wreckage, waiting an estimated half-hour, being rescued and taken to Canaries, then travelling by ambulance to Soufrière Hospital. His injury and temporary difficulty seeing also limit what he could observe after impact.
Different boats · Different stages · No single synchronized clock
The rescue is a set of journeys.
Nemo’s account supplies several missing links. The table keeps each journey separate. Approximate counts and times remain attached to their source; they are not added together to manufacture a precise response timeline.
| Journey or stage | What the new testimony says | What still needs checking |
|---|---|---|
| Yacht → Ti Kaye | Nemo says he took people including Lester from the yacht to the resort at their request, because an ambulance had reportedly been called. | Exact group, time and landing records. The resort witness’s observations need not cover the same people at the same moment. |
| Search → damaged Ocean Angel | Nemo says he looked for Tyson and Taylor, was directed to Tyson and Dylan, and took both aboard. | Who directed him, the location and timing. He did not witness the collision and does not know its precise position. |
| Tyson returns; Dylan → Soufrière | He says Tyson asked to return to Dylan’s boat with a light to warn other traffic. Nemo then took Dylan to Soufrière with his own lookout. | A reported safety action after impact. It does not establish either man’s responsibility for the collision. |
| Rescue boat → Canaries | Clear English recaps during Darren Joseph’s mostly Kwéyòl testimony describe collecting about six people and a total of about ten or eleven aboard going to Canaries. | The underlying answers are poorly captured. Counts, identities and the later journey require audio review; this is not a verified translated manifest. |
| Canaries → Soufrière Hospital | The new Ocean Angel passenger describes an ambulance transfer after landing at Canaries. | This is a separate pathway from the reported unfulfilled ambulance request at Ti Kaye. Compare both dispatch and handover records. |
Police presence is clearer. Timing is still estimated.
Nemo reports two Marine Police vessels and estimates their arrival at five to ten minutes after his. He also thinks he arrived about an hour after the collision, although he did not see it and supplies no clock anchor. Adding those estimates would not establish a reliable police response time.
John also describes a police vessel searching. These accounts concern the immediate response. Kaitha Paul’s earlier account concerned a particular later search trip without Marine Police or SLASPA personnel aboard. That narrower statement is not disproved by police being present on another journey.
Calls and statements need exact labels.
Nemo says he tried numbers belonging to people who work at Marine Police and got no answer. The transcript does not establish that he called an official emergency line. Call recipients, duty arrangements and dispatch logs matter.
He is initially introduced as having no police statement, but that is corrected during the hearing: the chair refers to a statement dated 21 July. The new passenger separately says a planned police follow-up did not happen. These are different record histories, not evidence for one blanket claim that rescuers or passengers were never interviewed.
Darren Joseph · 28 October · Approximately 01:07–01:22
Where the automated text loses Kwéyòl, we show the gap.
Darren Joseph’s evidence matters. The difficulty lies in the automated capture: long answers become scattered English words, uncertain names and missing phrases. There is not enough reliable Kwéyòl text here to produce a complete translation. A confident reconstruction would hide the source’s weakness.
Clear exchanges and labelled recaps.
The English exchange supports an approximate Canaries rescue load. The commissioner’s closing recap describes the strain of the rescue and says two people asked to remain. We identify this as the commissioner’s recap, not a verified English translation of Joseph’s own words.
Names, missing answers and sequence.
The boat name, aliases, the identity of people on particular legs and a fragment mentioning twenty to twenty-five minutes need recording review. We do not assign that fragment to a journey or turn an unclear answer into a quotation.
Return to the recording with a fluent reviewer.
Keep the original text and time anchors. Review the rescue sequence first, then names, numbers and the statement-history exchange. Log corrections in the reading layer so readers can distinguish the supplied text from later editing.
A commissioner’s question may contain an assumption. Agreement, a missing response or a short recap cannot always recover the witness’s full meaning. This limitation concerns the transcription, not the witness’s credibility. The 23 October afternoon file is largely intelligible English and does not require the same translation warning.
SLPA assessment · Inference, not a Commission finding
The aftermath is better explained. The final approach remains unresolved.
What is better supported
Rescue involved several private boats, a yacht, resort staff and Marine Police. People moved through different landing and medical routes. The new passenger supports a helm transfer, increased speed in his experience, absent briefing and no jacket worn by him. Nemo adds a reported reason for Tyson’s later return to the damaged boat.
What we cannot yet settle
The precise impact time, location, light operation, speeds and final turns remain open. So do the full passenger lists, exact lookout arrangements and official response timeline. The supplied material contains no measured reconstruction or complete medical evidence.
Our working inference remains that the vessels reached a dangerous encounter without enough time or distance for effective detection and avoidance. These new sessions do not identify one decisive cause. They strengthen the case for investigating the full sequence: preparation, approach, impact, injury, flotation, rescue, medical care and support afterwards.
They also require restraint in the narrative. John is not a pre-impact visual witness. The new passenger does not confirm unlit navigation lamps or extreme speed. A family’s report of an autopsy is not the autopsy itself. Keeping those distinctions visible makes the investigation more useful and fairer to everyone involved.
From testimony to action · SLPA proposals
Make the next questions answerable.
These are our proposed practices and records checks. They are not findings of a legal breach or final recommendations of the Commission.
Build a timeline from independent clocks.
Seek emergency call-taking, dispatch, Marine Police vessel and ambulance records. Separate call received, resource sent, arrival, pickup, landing and hospital handover. Identify personal-number calls separately.
Apply it: One incident record with named handovers can help responders know which people still need help and where.
Connect each transfer to a person and place.
Check Ti Kaye duty records, staff accounts, available call logs and any retained footage. Compare rescue-boat accounts and hospital arrivals. First establish which records actually exist.
Apply it: Give coastal staff a clear emergency contact procedure and a practical casualty handover record.
Test lights and sightlines physically.
Compare fittings, wiring, damage photographs, loaded trim and observer positions. Check command and lookout arrangements against the evidence. Passenger impressions cannot supply measured speeds or lamp performance.
Apply it: Use an explicit departure check, passenger briefing, named lookout acceptance and a clear helm handover.
Separate injury, notification and continuing care.
Seek appropriate medical evidence before drawing survival conclusions. Reconcile recovery and identification dates. Ask what support was offered, accepted, declined or still awaited, while protecting unnecessary personal detail.
Apply it: Assign a family liaison, record updates and offer follow-up support to relatives, survivors and civilian responders.
Public value depends on an inspectable record
Sources, uncertainty and editorial responsibility.
This coverage helps readers follow a long inquiry, compare accounts and see which records could answer the next question. Advanced AI agents and automated tools support organization, comparison and review. The site editor reviews published work and remains responsible for editorial decisions.
That publication-review process is different from checking every transcript word against the recording. Detailed human transcript editing is ongoing. We do not label these new files fully human-corrected or audio-verified. Multiple automated reviews do not create independent evidence.
The original supplied text remains available. Context-supported reading corrections are logged; uncertain names, headcounts, times and translations remain visible. Links above open the relevant hearing passages. Recording offsets locate testimony and are not times of day or precise video edit points.
This addition brings the source collection to ten files and 1,317 indexed passages. The new files add 190 passages and five examined witnesses. The 28 October morning file does not include the later sitting anticipated at its close.