One Sour Bean, Two Categories: Sorting a Lot Offered as Specialty
A 30-bag lot of washed Caturra from Huila arrives with a mill QC sheet reading 'Screen 15+, 0 full defects, Specialty grade'. You draw one 350 g green sample, spread it on a black tray under 5000 K light and sort it bean by bean. The tray holds: 1 bean containing a live insect and a four-hole tunnel; 3 beans with a single pinhole each; 6 broken and chipped pieces traceable to at least 4 beans; 2 husk and parchment fragments; 1 full sour bean that smells of vinegar when crushed; 4 immature beans with a green cast. The exporter's QC manager calls and argues: 'Put the pinholes and the broken pieces into your Category 2 total; if that total is five or under you are still Specialty. And just lift out the sour bean, it is one bean in 350 grams.'
1. Classify every finding on the tray and state the lot's grade outcome. Address the QC manager's proposal directly.
Read the worked answer
Category 1 (primary) defects
- 1 bean with a live insect and a four-hole tunnel: severe insect damage. One bean, but it is a Category 1.
- 1 full sour bean: Category 1.
Category 2 (secondary) defects
- 3 single-pinhole beans: slight insect damage, Category 2.
- 6 broken / chipped pieces: Category 2.
- 2 husk / parchment fragments: Category 2.
- 4 immature, green-cast beans: Category 2.
Outcome. Specialty requires zero Category 1 defects and no more than 5 Category 2 defects in the 350 g sample. Both conditions are tested; both must pass.
- Category 1 count = 2, so the lot cannot be Specialty or Premium, whatever the Category 2 arithmetic says.
- Category 2 count is already at least 9 (3 + 2 + 4) before the broken pieces are counted at all, so the 5-defect allowance fails independently.
Why the manager's proposal fails. Category 1 and Category 2 are not one pooled budget that trades off. Five Category 2 beans can never be 'spent' to absorb a Category 1. And a defect is recorded where it was found in the sample: lifting the sour bean off the tray after the sort does not un-find it, and the remaining beans are still the same sample, not a new one. Defects are counted as observed, not as curated.
2. The exporter asks you to re-sort a second 350 g sample 'to confirm', and offers to send one hand-picked 350 g taken from the top of a bag. How do you respond, and what would a clean second sample actually prove?
Read the worked answer
Respond with a sampling plan, not a favour. A second sample is worth something only if it comes from a defined sampling scheme: incremental samples or a probe drawn from multiple positions and depths across the 30 bags, combined and quartered down to 350 g, with the number of bags represented recorded. A hand-picked sample from the top of one bag is a grab sample; in a heterogeneous lot it is more likely to be clean than the lot is, so it biases the result toward the seller.
What a clean second sample would prove: that the lot is heterogeneous, and that defect distribution is uneven across bags. It would not erase the Category 1 finding. Both results belong in the record, each with its provenance.
Practical answer to the exporter: 'I will re-sample per a written plan and grade the aggregate; if the composite still contains Category 1 defects the grade stands. A clean sample from one bag tells us the lot is variable, which is its own quality problem.'
Document while you do it: sample ID, mass, date, light source, who sorted, the retain sample, and the exact counts per category. Without that, the second sort is just an opinion.
3. What should your written report say, and what does the 'five' in the specialty standard actually control?
Read the worked answer
The five is a Category 2 allowance, not a defect budget. The relevant criteria are two separate gates:
- Specialty: 0 Category 1 defects and 0–5 Category 2 defects in 350 g.
- Premium: 0 Category 1 defects and 6–8 Category 2 defects.
- Below that the lot drops into the Exchange bands, which tolerate limited Category 1 defects and a larger Category 2 band - read the current SCA green classification table for the exact numbers rather than quoting them from memory.
Report structure:
- Sample identity, mass, sampling method and provenance.
- Counts per defect, grouped by Category 1 and Category 2, listed individually rather than as a single total.
- Grade conclusion against both gates, stated as 'does not meet Specialty or Premium from this sample because of 2 Category 1 defects and a Category 2 count above the allowance'.
- A recommendation to re-sample under a defined plan before final rejection - the Cat 1 finding stands as recorded, but a proper composite is what supports a lot-level decision.
Avoid: a single headline number, and any phrasing that implies the Category 2 count was offset against Category 1. That is the misunderstanding that let this lot reach you labelled as Specialty in the first place.