Is It Geisha? A Variety Claim With No Paperwork
A broker offers six bags of 'Panama Geisha, washed, Boquete, 91-point potential' at roughly 40% below the going price for that name. The offer sheet lists screen 14/15, moisture 10.9%, water activity 0.55. You receive a 300 g green sample and a 100 g roasted sample. The bag carries a farm name but no farm-gate documents, no lot code, no exporter certificate. In the cup: strong jasmine, bergamot, peach, silky body, tea-like finish, and a faint geranium note as it cools to 40 °C. The green is short and roundish with some silver skin still attached. The buyer asks you to 'just tell me if it is Geisha or not'.
1. What can green appearance and the cup actually establish about variety, and what can they not?
Read the worked answer
Green appearance establishes almost nothing about variety. Bean size and shape are driven by screen, altitude, position on the branch, cherry maturity, processing and drying as much as by genotype. Geisha is characteristically long and slender, but a small, roundish screen 14/15 sample does not exclude Geisha, and a long large bean does not prove it. Silver skin retention is a processing and milling signal, not a varietal one.
The cup establishes a profile, not an identity. Jasmine, bergamot, peach and tea-like body are characteristic of Geisha and consistent with the claim - but characteristic is not diagnostic. Several Ethiopian landraces and some Bourbon-adjacent lots can throw jasmine; the geranium note at 40 °C could be a genuine floral-rose expression or a fermentation artifact. A cup profile raises or lowers probability; it does not name a cultivar.
What actually identifies variety: genetic markers (SSR/SNP panels) run on a leaf or green sample, or a documented chain of custody from a verified mother-garden planting through lot codes to export. Everything else - appearance, cup, price, reputation, the broker's word - is inference.
So the honest answer to 'is it Geisha' is: the cup is consistent with the claim, the physical sample neither confirms nor refutes it, and nobody in this chain has produced evidence that identifies a cultivar.
2. What is a defensible way to answer the buyer's question, and what do the moisture and water activity figures add?
Read the worked answer
Reframe the question. You were not asked to certify a variety; you were asked to reduce commercial risk. Deliver three separable findings:
- Sensory: describe the cup in neutral language, note the geranium at cool, give a score for the liquid in front of you - not for the name on the bag.
- Physical: screen, moisture, water activity, defect count, colour - measured facts.
- Evidence gap: name exactly what is missing (lot code, farm-gate/harvest records, exporter documentation, any genetic test) and state that without it variety is unverified.
Then propose the test that would settle it: ask the seller for a genetic panel, or run a blind side-by-side against a verified reference lot of the same claimed variety, same roast, same rest, coded cups. That is a discrimination test, and it should be interpreted as one - a 'no difference' result is not proof of identity either.
The numbers. Moisture 10.9% sits inside the normal green range and water activity 0.55 is workable, if on the higher side; nothing here is a storage emergency. But neither number authenticates anything, and in a chain with no lot code they simply cannot be tied back to the bags in front of you. The 40% discount is the loudest datum in the file: names in strong demand do not trade far under market without a reason - old crop, mislabelled lot, re-export, or a name being used loosely. Ask what the reason is and write down the answer.
3. The broker offers you a commission if you 'confirm Geisha' in your written report. How does that bear on what you write?
Read the worked answer
The commission makes the requested conclusion inadmissible. Payment contingent on a specific finding converts an assessment into a purchased opinion. Even if your cup notes would have been identical, the report is now worth less to everyone who reads it - and worth nothing at all if the buyer later discovers the arrangement.
What to do:
- Decline the arrangement explicitly, in writing, before issuing anything. If the broker wants a paid assessment, the fee is fixed and independent of the result.
- Disclose the offer to the buyer who engaged you. Undisclosed contingency is the part that damages the credential, not the tasting itself.
- Keep the report to what the evidence supports: cup description, measured physical data, the documentation gap, and a clear statement of the limits of the assessment. No use of 'certified', no implication that a Q Grader credential certifies cultivar, no lending of the mark to a name you have not verified.
- Separate the commercial recommendation from the identity question. You can say 'this is a very good floral washed coffee and it is priced attractively for what it is in the cup' without ever answering 'is it Geisha'.
The general principle worth carrying into any grading job: a grader's product is a defensible record, not a favourable one. Any request that a conclusion be pre-decided - confirm this name, pass this lot, round this score up - is answered the same way: here is what the sample shows, here is what it does not show, and here is what would be needed to close the gap.