The trial of Vusimuzi “Cat” Matlala and his co-accused is increasingly turning into a battle not only over what allegedly happened to Musa Kekana after his arrest, but over how reliably the evidence surrounding those events was recorded, preserved and reconstructed.
As the Gauteng Hig Court in Johannesburg hears a trial-within-a-trial to determine the admissibility of evidence obtained following Kekana’s arrest on 17 April 2024, testimony has exposed a troubling pattern, important events were not always documented contemporaneously, witnesses have struggled to recall key details, and questions have emerged about how physical and digital evidence was selected and preserved.
The issue is potentially far more consequential than any single disputed version of events.
It goes to the question of whether the court can reconstruct the chain of events from a record containing gaps, delayed statements, incomplete documentation and contested evidence handling.
One of the clearest examples came from Netcare 911 paramedic Mary Kgapele.
Kgapele testified that Kekana told her he had been assaulted by police officers.
But the allegation did not appear in the contemporaneous patient report she completed after treating him.
Under cross-examination, she accepted that she should have recorded the allegation.
The omission became significant because much of her evidence about the events surrounding Kekana’s treatment was also affected by the passage of time.
She repeatedly told the court that she could not remember particular details, including how many times she checked Kekana’s vital signs, details concerning the blood he was coughing up, the colour of a cloth used around him and aspects of the handover process at hospital.
At one point, her evidence placed Kekana inside a property on Fifth Road for approximately an hour while police dealt with him.
She said she periodically went inside to check his vital signs. But when pressed on the precise number of times she did so, she could not remember.
A witness saying an event happened is one thing. A witness being unable to reconstruct when, how often and under what circumstances it happened is another.
The defence repeatedly tested those gaps. Kgapele also conceded that there were periods when she was not necessarily with Kekana and could not dispute that events may have occurred outside her view.
That left the court with an obvious evidentiary problem, the paramedic could testify about what she personally observed, but could not account for everything that may have happened while she was elsewhere.
Another potentially important fault line concerns the medical records themselves. Kgapele testified that she used an electronic system on a tablet to capture patient information.
She explained that the electronic system contained more information than the paper patient report, potentially including repeated vital-sign readings and pain assessments.
Yet the fuller electronic record had not formed part of the information she initially provided to the State.
The court therefore heard evidence suggesting that the contemporaneous electronic record may contain information not reflected in the paper version placed before the court.
Kgapele also testified that Kekana’s blood pressure was elevated, with readings including 153/100 and 154/95.
She agreed that stress and fear can cause elevated blood pressure and heart rate. But she also told the court that Kekana had identified himself as a known hypertensive patient, a detail that the defence challenged because it did not apparently appear in the hospital records in the same way.
The evidence therefore raises a deceptively simple question, what does the original medical record actually show?
And, perhaps more importantly, what information was lost when the fuller electronic record was reduced to a paper summary?
The problems did not end with the medical records. Kgapele testified that police only contacted her months after the April 2024 incident around the end of July or beginning of August.
She told the officer that she needed her documents and notes to provide accurate information. According to her testimony, three officers later went to her home carrying the patient report.
A male officer typed her statement on a laptop while she provided information based on her recollection and the documents before her.
The statement was then printed and signed. That process is not, in itself, proof that the statement is unreliable but it introduces an important evidentiary distinction.
The court is not dealing solely with a contemporaneous account written immediately after the incident. It is also dealing with a later reconstruction of events, produced months after the incident and after the witness’s memory had had to carry the burden of time.
Then came the crime-scene photographs. The testimony of Sergeant Thabiso Lerato Mohale introduced another layer to the evidentiary puzzle.
Mohale, a provincial crime-scene investigator, testified that he arrived at 80 Bramley at about 09:50 on 17 April 2024 and processed the scene.
His evidence provided the court with a window into how physical evidence was selected.
Crime-scene investigators, he explained, may take hundreds of photographs at a scene and later select the most relevant images for the formal album.
For this case, however, he had retained the raw photographs and brought two CDs containing the original images to court.
Two photographs became particularly important. The digital printouts showed that they were created on 17 April 2024 at 11:03:42 and 11:03:46 respectively, just four seconds apart.
The photographs showed a remote and a bunch of keys in the centre-console area of a vehicle. Mohale testified that he left the items where they were and that, during the period he remained at the scene, nobody removed them.
He said he only left the first scene at about 15:55 before proceeding to Fifth Road. That creates an unusually precise photographic timeline. The remote and keys were photographed at 11:03.
Mohale said they remained where they were when he left the scene nearly five hours later. But the more significant question was not necessarily where the items were. It was why they were treated differently from other objects at the scene.
Under cross-examination, Mohale explained his normal procedure. Items considered to have evidential value are photographed individually and then packaged.
The defence pointed out that numerous seemingly ordinary objects had been individually photographed and packaged, including water bottles, Vaseline and shoes.
Yet the remote and apparent key in photographs 44 and 45 were not individually photographed or packaged.
Mohale did not say that the objects had no evidential value. Instead, he explained that he did not regard them as sufficiently evidential to warrant individual packaging.
That answer opens another important line of inquiry. Who decides what becomes evidence? And what happens to an item that an investigator decides is not sufficiently important to collect?
Mohale’s evidence was that investigators photograph items for evidential purposes and then individually package those they consider evidential.
The defence therefore tested the apparent inconsistency between the treatment of numerous other objects and the remote and key.
It is not evidence that anything improper occurred. But it exposes a critical element of forensic investigation, evidence is not simply found, investigators make decisions about what to preserve as evidence.
The defence also scrutinised the digital metadata attached to the crime-scene photographs. Mohale accepted that the date and time settings on a camera can be manually changed.
He said, however, that he checks and double-checks the camera’s date and time before going on standby to ensure that it is correct.
He maintained that photographs 7805 and 7806 had not been modified. But another metadata entry showed an ‘accessed’ date of 28 July 2026. Mohale could not definitively explain the significance of that date, saying he was not sufficiently knowledgeable about the metadata to provide a technical explanation.
He suggested that it could relate to the transfer of photographs to a disk. Again, that does not establish manipulation. But it leaves a question hanging over the digital evidence.
What exactly does the metadata record show, and why does an access date appear more than two years after the photographs were created?
That is ultimately a technical question that may require evidence from someone with specialist digital-forensics knowledge.
The defence also probed who was actually in control of the crime scene. Mohale testified that there was already a Warrant Officer at the scene when he arrived. He explained that the Warrant officer explained what had happened and pointed out areas that needed to be photographed.
Mohale said that, ordinarily, the person already at the scene who points out the scene is the person in control, although he could not definitively identify who was in control in this particular instance.
That warrant officer subsequently left. Mohale continued working with his superior, a Colonel. He also saw two suspects who were handcuffed when he arrived. But because his focus was the processing of the scene rather than the suspects, he could not say what happened to them by the time he left.
That creates yet another gap in the reconstruction of events. The court is being asked to assess evidence generated across different locations, different officers, different witnesses and different points in time.
And not every witness can account for everything that happened outside their immediate field of observation.
Taken individually, none of these issues necessarily proves that evidence was fabricated, manipulated or improperly obtained.
The defence still has to persuade the court that the evidence should be excluded, while the State will have to defend the admissibility and reliability of the material it seeks to place before the court.
But the accumulating testimony has nevertheless exposed something potentially significant. The evidentiary record is not seamless.
It contains memory gaps, documentation gaps, questions about electronic records, delayed statements, disputed periods outside witnesses’ direct observation, questions about scene control and decisions about which physical objects warranted preservation.
And that makes the trial-within-a-trial about much more than the alleged treatment of Kekana.
It is becoming a test of how much confidence a court can place in an evidentiary chain when parts of that chain have to be reconstructed from imperfect memories and incomplete records.
For the defence, those gaps could become ammunition in its argument that the evidence obtained after Kekana’s arrest was compromised.
For the State, the challenge will be to demonstrate that despite the imperfections, the evidentiary trail remains sufficiently reliable and lawfully obtained to survive scrutiny.
Either way, the testimony is putting the evidence itself on trial. And that could ultimately prove just as important as the competing versions of what happened on 17 April 2024.
