Build the trial on the ground.
Draw the parcel, then lay the experiment onto it level by level: site, trial, blocks, plots. Every plot has its place on the map and its treatment, set out by a real experimental design.
- 4 blocks
- 5 treatments
- 20 plots
Run the season.
Plan each operation against the calendar - sowing, fertilising, spraying, irrigation - assign it to a role, and follow every plot from emergence to maturity.
Collect every reading.
Observations are taken at the edge of the plot, on a phone, even without a signal. Each value lands on its plot, its date and its unit: nothing typed twice, nothing lost.
- Plot
- 302
- Measurement
- Grain yield
- Value
- 7.90 t/ha
Get the proof.
The analysis says which treatment came out ahead, by how much, and whether the difference is larger than the trial's own noise. When the data cannot support an answer, it says so, and why.
Treatment means with 95% confidence intervals. Means that share a letter do not differ (Tukey, 5%).
