Juma closed the shop on Sunday and spread every record across the counters. Bi. Nuru had given them a narrow assignment: reconstruct cash movement by date and source. Do not invent trades they could not trace.
Mariam made four columns: money into Hassan’s account, money out, description given at the time, independent record available now. Omari and Peter supplied redacted statements. Rehema supplied her messages. Musa supplied transfers.
On the date Juma sent his first amount, another participant deposited money the same day. Three days later Stella received what she had been told was a return. There was no record showing that Juma’s specific money paid Stella, and they refused to write that conclusion.
“We can say new money and old payouts were in the same account,” Mariam said.
“Not that this exact deposit funded that exact person.”
“Right.”
They repeated that discipline across the timeline. New deposits often sat close to withdrawals to earlier participants. There were periods of market loss while internal screenshots still showed positive participant “returns.” Hassan described those as long-term allocations, but no independent ledger had yet verified them.
Bi. Nuru reviewed a sample. “Don’t call a transfer a return just because that was the label. Write: *transfer described as return*. A label is not the nature of the record.”
Juma saw how language itself could create an account that did not exist. A personal bank transfer called a “profit withdrawal” felt like proof of an investment system. Written plainly as a transfer from Hassan, the source question remained open.
They preserved original files and worked on copies. If new information changed interpretation, they added a dated note instead of rewriting old evidence.
Omari brought an old receipt Hassan had sent him. It named a “custody partner” and carried a registration number. Rather than announce that it was fake, they checked the public business registry. The company had once existed, but it had been dissolved years before the receipt date.
“This is fake,” Omari said.
“Maybe,” Juma answered. “First we ask whether it was an old template or some successor name.”
They called Hassan. He admitted he had continued using the old partner’s name on an internal template after the relationship ended.
“Was that company holding Omari’s money on the date of this receipt?” Mariam asked.
“No.”
Omari went silent.
Juma did not need a dramatic adjective. The document had been presented as something it was not.
They also created a list of material they would not use: screenshots with unknown origin, third-party voice notes without consent, gossip about Hassan’s lifestyle. Omari suggested checking Hassan’s travel photos as proof of spending. Mariam rejected it. “A holiday is not an account reconciliation.”
Before they finished, Juma wrote a limitation beside the registry result. Even an alarming finding should leave room for a documented explanation.
When Omari wanted a copy of the full timeline, Juma refused to hand over other people’s names and amounts. He gave Omari only the lines tied to his own transactions and the public registry note. The shared reconstruction would be available later under agreed rules. Omari was irritated, then conceded that he would not want his own balance passed around casually either. The decision reinforced a principle that had become central: evidence could be pooled for a defined purpose without turning every participant’s private finances into common gossip. Their search for truth did not require unlimited visibility.
They checked spelling variants and successor-company names before finalizing the registry note. Nothing in the available record showed that the dissolved entity had lawfully continued under the same identity on the receipt date. The finding stayed qualified, but the qualification did not make the date discrepancy disappear.
The receipt named a custody company that had been dissolved long before its date.
#