Roast 25 was meant to be simple.

Brazil Topazio.
250g.
250°C set.
No mid-roast adjustments.

The only deliberate change was the stopping logic: a fixed cool at 13:03.

No chasing weight loss. No reacting to crack sound. Just a hard time stop.

On paper, it felt clean. Controlled. Almost disciplined.

The roast looked darker than expected. There was visible smoke. The beans came out a deep brown. If judged by appearance alone, it would have been easy to assume it had pushed well into development.

But the scale said otherwise.

14.3% weight loss.

Lower than anticipated.

That moment was quietly revealing.

The total roast time was similar to other roasts that had reached higher development. The peak chamber temperature was normal. Nothing obvious suggested “underdevelopment” during the roast itself.

Yet the outcome told a different story.

Time alone wasn’t a reliable proxy.

That doesn’t mean time is irrelevant. It clearly matters. But it isn’t sufficient on its own to define roast level on this machine. Effective energy, airflow behaviour, and the shape of heat application matter more than a fixed timestamp.

Roast 25 wasn’t a failure. It was a calibration.

It reinforced something that had been suspected but not fully internalised: weight loss remains the only stable anchor across roasts, especially on a convection heavy system.

In hindsight, the darker appearance and smoke were misleading. Brazilian naturals can look developed earlier than they actually are. The scale doesn’t lie.

This roast quietly shifted the mindset from:

“Set a time and trust it.”

to

“Time guides. Weight loss confirms.”

Roast 25 wasn’t dramatic.

It was clarifying.

And in many ways, it set the stage for the decision to consolidate rather than complicate.