Coffee | One Roast at a Time

Gene Café CBR-301 | Active Exhaust

Date: 2026-02-??
Batch: 250 g
Out: 217 g
Weight loss: 13.2%
Set temp: 250 °C
Preheat: None
End time: ~12:4x
Coffee: Honduras, Chinacla (La Paz) — Finca El Puente
Producers: Marysabel Caballero & Moises Herrera
Variety: Catuai
Process: Natural
Crop: 2025

Why this roast matters

The key insight behind this roast did not originate here.

That turning point occurred with Roast 13 (Ethiopia natural), where first crack became clearly audible again without any change to the system. That roast established an important baseline:

Crack audibility is coffee-dependent, not system-dependent.

Roast 14 exists to test the boundaries of that conclusion, not to replace it.

What happened

This roast behaved distinctly differently from the Brazilian naturals, and importantly, differently from Roast 13 in a useful way.

  • First crack was clearly audible and recognisable
  • Crack occurred earlier relative to overall roast progression
  • There was no need to push the roast to “find” crack
  • The roast ended calmly after a short post-crack development

Despite the clear crack, the roast finished at the lowest weight loss of any natural roasted so far (13.2%).

What this confirms (relative to Roast 13)

Roast 13 showed that crack could return.
Roast 14 shows it can return without increased development.

This removes a lingering ambiguity from earlier roasts:

  • Audible crack does not imply a darker roast
  • Quiet or absent crack does not imply underdevelopment
  • Crack intensity and clarity vary within naturals, not just between natural and washed coffees

Roast 14 therefore confirms and sharpens, rather than introduces, the earlier conclusion.

Comparison to other naturals

Across the natural coffees roasted so far:

  • Brazilian naturals
    Crack quiet or smeared; development best judged by time and weight loss
  • Ethiopian natural (Roast 13)
    Clear crack, but at a higher development level
  • Honduras Catuai natural (this roast)
    Clean, obvious crack at the lowest development level recorded

All were roasted on the same system, with the same constraints.

The variable is the coffee.

Key learning (restated, not reinvented)

First crack is descriptive, not decisive, even within natural coffees.

Roast 13 demonstrated that crack audibility could return.
Roast 14 demonstrates that it can return early and lightly developed.

Together, they remove crack as a universal decision trigger.

How this changes future roasts

This roast becomes a low-development natural reference for the system.

Future Honduras Catuai roasts can:

  • Be developed slightly further (+0.5–1.0% WL) with confidence
  • Use crack as confirmation, not a target
  • Be compared meaningfully against Brazilian naturals at similar WL

Closing reflection

Roast 14 does not add a new rule.
It closes an open loop left by earlier naturals.

The system is now stable enough that differences can be trusted to come from the coffee, not the process, and that is the real progression in this project.