Date: 15 February 2026
Machine: Gene Café CBR-301
Batch: 250g
Set Temperature: 250°C
Preheat: No
Cooling Initiated: 12:25 (745 seconds)
First Crack (logged): 11:20 (680 seconds)
Development Time: 65 seconds (crack to cool)
Weight In: 250g
Weight Out: 211.25g
Weight Loss: 15.5%


Context

Roast 22 was conducted as a controlled compression test.

Previous roasts of this Guatemala Caturra washed (Roasts 18, 20 and 21) repeatedly clustered around ~16.1% weight loss under the same baseline configuration:

  • 250g batch
  • 250°C set temperature
  • Fan level change around ~4 minutes
  • No mid-roast parameter changes

The objective here was simple:

Shorten post–first crack development and observe the resulting structural shift.

No changes were made to temperature, airflow, or batch size.
Only cooling timing was adjusted.


Observed Roast Behaviour

Early Phase (0–5 minutes)

Fan behaviour remained consistent with prior roasts.
No deviations in airflow timing were observed.

Mid Phase (5–11 minutes)

Temperature progression appeared linear and stable.
No evidence of stall or crash around crack onset.

First Crack

First crack was logged at 11:20.
Cooling was initiated at 12:25.
Development time was therefore 65 seconds.

This represents a reduction of approximately 15–25 seconds relative to previous ~16.1% roasts.


Numeric Comparison to Prior Cluster

RoastWLTotal TimeDevelopment
18~16.1%~12:50~80–90s
20~16.1%~12:50~80–90s
21~16.1%~12:50~80s
2215.5%12:2565s

This roast reduced:

  • Total applied time by ~25 seconds
  • Development by ~15–25 seconds
  • Weight loss by ~0.6% relative to the 16.1% band

Observational Positioning

Roast 22 does not enter the previously observed 14–15% structural range seen in other coffees.

Instead, it sits between:

  • The ~16.1% Guatemala cluster
  • The ~14–14.7% range observed in certain naturals

This suggests (observationally, not causally) that:

  • The system retains sufficient thermal momentum at 250g / 250°C to resist large structural shifts unless cooling is triggered earlier still.
  • Weight loss in this configuration may exhibit a natural clustering zone around ~16%.

Further repetitions would be required to confirm this.


What This Roast Does Not Conclude

  • It does not establish that 15.5% is preferable.
  • It does not establish that 16.1% was excessive.
  • It does not establish process-dependent structural behaviour.
  • It does not establish perceptible cup difference.

Cupping is required before any interpretation is promoted.


Next Step

Roast 22 will be cupped against a 16.1% Guatemala example.

The question:

Does a 0.6% reduction in weight loss and ~20 second reduction in development produce a perceptible structural difference?

If yes, this becomes a meaningful structural threshold marker.
If no, this may indicate a tolerance band within this system.

No conclusions are drawn prior to cupping.


Reflection

This roast represents the first deliberate compression of this coffee’s development window.

The system responded predictably.
The outcome remained within a tight numeric band.

The dataset is becoming more controlled