Source A: The cabinet that sent an alarm

An original contemporary non-fiction-style practice text, written for this question.

At the demonstration, the glass cabinet appeared to defend itself. A technician released a puff of damp air near its open door; the door closed, a fan whispered and the line on the monitor returned neatly to the green band. Around me, trustees nodded. Our museum’s collection of maps, letters and photographs would, they were told, be protected by a system that reacted before a person could even notice a change.

I wanted to be convinced. As the museum’s registrar, I had spent too many wet afternoons moving boxes away from unreliable walls. Yet the smooth line on the monitor seemed to be doing more work than the evidence justified. We had watched one cabinet respond to one planned disturbance in a conference room. The proposal was to install similar cabinets throughout six galleries.

We agreed to a thirty-day trial in the map room. The funding decision had to be made soon afterwards, which meant that the trial was treated less like an investigation than an audition. Staff photographed the glowing display. A trustee asked whether we could announce the complete installation before the trial had finished.

During the second week, the room’s humidity rose from 47% to 61% overnight. The cabinet sent an alert, but the message reached my telephone three hours later. By breakfast, the numbers had returned to normal. The supplier described it as an isolated communications problem. One trustee called it proof that the entire system was unreliable. Both conclusions arrived before anyone had opened the records.

The records complicated the story. The cabinet had detected the change within four minutes and had kept its interior stable. The room sensor had risen first. A cleaner had wedged open a fire door while a washed floor dried in the corridor, allowing damp air to be drawn into the room. Separately, the museum’s router had restarted during a routine update, delaying the message. The trial had revealed two weaknesses, but they were not the same weakness. The cabinet had protected its contents; our building and our communication arrangements had not protected the cabinet.

After the door procedure and router settings were changed, the readings remained between 47% and 50% for the rest of the trial, including a week of heavy rain. Alerts from later tests arrived in under a minute. It was tempting to replace the first dramatic verdict with a second one and declare the system proven. I could not do that either. One room had been tested for one month, with staff behaving carefully because they knew they were being observed.

I recommended installing the cabinets in two further rooms while retaining manual checks. The trustees had wanted a confident yes or no. Instead, the trial gave them a smaller decision, followed by another opportunity to examine what happened. That seemed less impressive than the demonstration. It was also more useful. A trial should not be a rehearsal for applause; it should turn one large promise into questions small enough to answer.

Source B: The field that was allowed to flood

An original contemporary non-fiction-style practice text, written for this question.

In July, the proposed flood-storage field looked like an expensive way of doing nothing. The grass had been cut around several new channels, wooden posts marked future water levels and a low bank curved along the boundary. At the public meeting, an animated map had shown storm water spreading obediently across the field instead of entering the village. On the ground, there was no animation: only dry soil, grazing sheep and an engineer pointing at places where water might go.

The field was a twelve-hectare trial. If it worked, similar arrangements could be made across nine farms higher in the valley. Some residents welcomed any plan that might protect the houses beside the river. The farmer whose land formed the trial asked who would remove debris, repair damaged fences and replace the winter grazing he would lose. His questions sounded inconvenient beside the clean blue shapes on the engineer’s map. I initially suspected that the scheme’s modest appearance was being used to conceal modest benefits.

In October, rain filled the channels and then covered most of the field. Water reached the lowest fence wire. The temporary riverside road was closed, and part of a public footpath was damaged, but the water at the bakery wall stopped below the mark left by an earlier flood. The council issued a statement saying that the trial had proved the valley plan worked.

The figures encouraged me, but they did not produce the simple comparison promised by the statement. The trial storm brought less rain, and the river had started lower. Nobody could say exactly where the water would have reached without the storage field. The engineer acknowledged this. What the storm had demonstrated, she said, was that water entered the field by the intended route, remained there and drained away without the bank collapsing. It had not demonstrated how nine storage areas would interact during a more severe storm.

The farmer was less hostile than he had been in July. “The field did a job,” he said, before showing me silt pressed against a gate and a section of flattened fencing. His approval was conditional on those costs being included in the next plan. The shopkeeper near the river was more enthusiastic: for her, a dry storeroom was evidence enough.

I had arrived in summer ready to dismiss a field that seemed to contain nothing but promises. After the storm, that position was no longer reasonable. Water had gone where the plan said some of it would go. Yet calling the trial either a miracle or a fraud would make the same mistake: treating one event as a final verdict. The flooded field was evidence that the idea deserved a larger test. It was not evidence that questioning could stop.