This manual provides guidelines, not guaranteed recipes. The outcomes described are grounded in published research applied to the coffee leaf — but Thumpassery's Chandragiri variety at 130m under rubber shade has its own chemistry. Observe your leaf. Record what happens. Adjust accordingly.
If you are a farmer, café owner, or first-time Citane processor — start here. These five sections explain what this manual does, how to use it, and what the most important concepts are before you pick up a single leaf.
Paper I — the Citane Reactive Landscape — explains the chemistry of the coffee leaf and why different processing choices produce different results. It is theory. It is a map of what is possible.
This manual, Paper II, is the practical companion. It tells you what to actually do — with your hands, with your leaf, at your estate. It assumes you do not need to understand every chemical reaction before you start. You need to know which leaf to pick, what to do with it, and how to tell if the result is what you were aiming for.
Paper I answers: why does the leaf behave this way?
Paper II answers: what do I actually do with it?
You can use this manual without reading Paper I. But when something unexpected happens — and it will — Paper I is where you find out why.
Cross-references to Paper I appear throughout as [Paper I, Section X]. They are signposts, not prerequisites.
↑ ContentsBefore any processing decision is made, answer these three questions. They appear at the beginning of every pathway in this manual for a reason — they anchor every choice that follows.
What leaf am I starting with?
Age, position on the tree, sun exposure, season, condition. The leaf you harvest is your raw material. Every processing decision works with what is already in that leaf — not with a theoretical ideal.
Where am I trying to go?
Green and fresh? Amber and tea-like? Roasted and deep? Something unknown? Choose your destination before you start processing, not after. The pathways in Part 5 are organised by destination, not by technique.
Am I recording this batch?
Citane is exploratory. Without records, every good batch is a lucky accident you cannot repeat. The batch sheet in Part 8 takes under three minutes to fill in. It is the difference between exploration and craft.
Every processing pathway in this manual carries a confidence rating. This tells you how well-established the pathway is, and how much variation you should expect in the outcome.
| Rating | What It Means | Expect |
|---|---|---|
High |
Well-established in research or practice | Predictable outcomes when parameters are followed. Suitable for production batches. |
Medium |
Grounded in research, less tested at scale | Good directional reliability. Some variation expected. Start with small batches. |
Experimental |
Theoretically grounded, limited precedent | Interesting territory. Outcomes may surprise. Always record these batches. |
Exploratory |
Open framework — you design the route | No predicted outcome. The outcome IS the finding. Rigorous recording essential. |
The confidence rating reflects the evidence base, not the quality of the outcome. An Experimental batch that produces something remarkable is not a lower-quality batch — it is a discovery. The rating simply tells you how surprised you should be prepared to be.
↑ ContentsCoffee leaf has been processed and consumed as a beverage for centuries across Ethiopia, Indonesia, and other regions. Four traditions appear as sidebars throughout this manual: Engere (Ethiopia), Kuti (Harar, Ethiopia), Kawa Daun (West Sumatra, Indonesia), and Chemo (Southwest Ethiopia).
These appear as inspiration, not instruction. Citane at Thumpassery uses a different variety, at different altitude, under different shade conditions, in a different climate, in a different cultural context. To claim fidelity to any of these traditions would be inaccurate and disrespectful.
Each sidebar tells you: what the tradition does, what its processing logic is, and what Citane can learn from it. The pathway that follows is labelled "inspired by" — not a replication. As your processing experience grows, these reference points become more useful, not less.
This manual teaches evaluation alongside processing. For every destination, there is a description of what a well-made cup looks, smells, and tastes like. There is also a troubleshooting section (Part 9) that maps cup symptoms back to processing causes.
The most important habit a Citane processor can develop is tasting with intention — not just "is this good or bad?" but "what does this tell me about what happened in processing?"
Harsh bitterness is not just unpleasant — it is information. It is probably telling you that thermal development went too far, or that steep time was too long. A flat, hay-like cup is probably telling you that the leaf sat in dry storage without proper sealing. A muddy, vinegary note is probably a fermentation that ran beyond its useful duration.
The cup is the last step in a long process. Read it as a record of every decision that preceded it.
Citane is a coffee leaf beverage produced by KoffyKraft at Thumpassery Estate, Karavaloor, Kollam, Kerala. The leaf is Arabica Chandragiri, grown under rubber shade at approximately 130 metres above sea level. The beverage category is new — there is no established playbook for Citane specifically. This manual is the beginning of one.
Coffee leaves have been consumed as beverages for centuries across Ethiopia (Engere, Kuti, Chemo), Indonesia (Kawa Daun), and other regions. The chemistry that makes those beverages possible is the same chemistry that makes Citane possible. But the variety, altitude, shade conditions, and cultural context at Thumpassery are different. Citane is not a replication of any tradition. It learns from all of them.
This manual covers ten areas: understanding the leaf, choosing a destination, the processing toolbox, seven documented pathways, fermentation, brewing and evaluation, recording, troubleshooting, and reference tables. It is designed to be used by a farmer who has never processed a leaf for beverage before, and by a processor who wants to move beyond intuition toward deliberate craft.
Citane processing is genuinely exploratory. Some of what is written here will be revised as more batches are made and recorded. Where guidelines are grounded in published research they are marked with a source note. Where they are extrapolated or adapted, that is stated. Treat this as a living document — the first edition of something that will grow.
The harvest decision is the first and largest processing decision. Before any tool in Part 4 is applied, the chemistry of the leaf is already determined by when and where it was picked. This section gives you the knowledge to make that first decision well.
Leaf age is the single most powerful determinant of the chemical starting point for any processing pathway. The research is unambiguous: the concentrations of chlorogenic acids, mangiferin, and catechins all decrease significantly as leaves age. A young leaf and an old leaf from the same tree are chemically different raw materials.
Highest chlorogenic acid concentration. Mangiferin at peak. Highest catechin levels. High moisture content (~73%). More reactive — responds dramatically to oxidative and enzymatic processing.
Moderate chlorogenic acid levels. Good catechin pool. Easier to handle mechanically than young leaves. The standard starting material for most processed Citane pathways. Moisture ~63%.
Chlorogenic acids declined (~74% drop from young). Mangiferin declined (~85% from young). Milder flavour potential. Lower bitterness precursors — can produce softer, rounder cups with less astringency. Moisture ~62%.
Lowest levels of all active compounds. Kuti tradition (Harar, Ethiopia) specifically uses fallen leaves for their mildness. Appropriate for very mild brews or specific traditional-style pathways. Not generally recommended for Citane's main pathways without specific intention. Moisture: ~56%.
Mangiferin decrease during leaf aging: approximately 85% reduction from young to mature in C. arabica var. Bourbon (Monteiro et al., 2019, Antioxidants). Chlorogenic acid decrease: approximately 74% in 5-CGA from young to mature. Catechin content: highest in young leaves (Campa et al., Frontiers in Plant Science). Moisture content data: Rigling et al., Foods 2022.
Thumpassery's Chandragiri is grown under rubber shade. This is a defining characteristic of your leaf's chemistry — and it matters for processing.
Research on C. arabica confirms that the phenolic pool — including chlorogenic acids and catechins — is generally greater in full-sun leaves than in shade-grown leaves. Full-sun plants produce more phenolics as a photoprotective response to high light intensity. Shade-grown leaves accumulate less of these compounds but tend to have a different balance — more chlorophyll-rich, with a different fatty acid profile in the membrane lipids.
Your rubber-shaded leaf may have a lower total phenolic concentration than a sun-grown equivalent, based on the general pattern in C. arabica — though this has not been measured for Chandragiri at Thumpassery specifically. If that pattern holds, a softer, less astringent starting point would be plausible: oxidative processing (Pathways 3 and 4) could produce a more delicate amber style than equivalent pathways from sun-grown leaf, and the thermal pathway (Pathway 5) could produce a milder roast character. This is not a disadvantage either way — it is a character to observe and record, not a deficit to correct.
This also means your leaf's mangiferin concentration — already the most distinctive compound in coffee leaf — may be somewhat different from the Cangeloni 2022 baseline (which used 1,700m Colombian leaf). Record your sensory outcomes carefully. Over time, the data from your own batches becomes more useful than any published figure.
Source basis: Campa et al. — phenolic pool greater in full-sun vs shade leaves. Frontiers in Plant Science, 2017. Consumer perception study — shade vs sun effects on sensory profile (Fibrianto et al., 2024, IOP Conf. Series EES 1302).
↑ ContentsWithin a single branch, leaf chemistry varies with position. Terminal flush leaves — the youngest at the growing tip — contain the highest concentrations of chlorogenic acids, mangiferin, and catechins. As you move inward along the branch toward older leaf pairs, concentrations decline progressively.
| Position | Leaf Pair | CGA Relative Level | Mangiferin | Best Use |
|---|---|---|---|---|
| Terminal tip | L1 (youngest) | Very high | Very high | Green Citane, enzymatic pathways |
| Second pair | L2 | High | High | Oolong-style, general processing |
| Third pair | L3 | Moderate-high | Moderate | All pathways, good balance |
| Inner leaves | L4–L6 | Declining | Declining | Roasted pathways, decoction |
Source: Phytochemical Profile and Antioxidant Capacity of Coffee Plant Organs (Antioxidants 2020) — L1 to L6 leaf pair sampling, C. arabica.
↑ ContentsA stressed leaf is not simply a lower-quality version of a healthy leaf. It is a chemically different raw material. Under biotic stress (pest damage, fungal infection) or abiotic stress (drought, waterlogging, nutrient deficiency), the leaf reallocates resources — sometimes increasing certain phenolics as a defence response, but doing so unevenly and unpredictably.
Insect-damaged: PPO (the oxidation enzyme) activates immediately at wound sites. The leaf has already begun transforming before processing begins — and not in a controlled way.
Fungal or mould-spotted: Introduces unwanted microbial populations. Their metabolites will appear in the brew. Avoid entirely.
Yellowing from disease (not natural senescence): Indicates cellular breakdown. Different from the normal ageing of a senesced leaf — contains degradation products rather than rebalanced compounds.
Physical damage at harvest: Torn, crushed, or bruised leaves begin oxidising immediately. Unless you intend to oxidise (Pathway 3 or 4), handle all leaves without disruption.
Firm, not limp. Consistent colour for its age. Clean surface — no spots, no webbing, no sticky residue. No unusual smell at picking. Stem cleanly removed without tearing. These are your working standards at harvest.
Seasonal variation in coffee leaf chemistry is real and documented — though specific data for Chandragiri at 130m under rubber shade at Thumpassery does not yet exist. Based on what is known from broader C. arabica research:
| Condition | Leaf State | Processing Implications |
|---|---|---|
| Dry season, healthy growth | Firm, lower moisture, concentrated compounds | Drying faster, more predictable oxidation. Recommended for first batches of each pathway. |
| Post-rain / wet season | Higher moisture, may be more turgid | Withering takes longer. Sun drying may be unreliable. Microbial risk higher if humidity stays elevated. |
| Harvest during rain | Wet surface, compromised epiphytic microbiome | Avoid if possible for fermentation pathways. Acceptable for immediate heat-fix pathways (Green Citane). |
| Drought stress | Wilted, reduced turgor | Not ideal raw material. If unavoidable, document carefully and compare against non-stressed batches. |
The rubber shade canopy at Thumpassery will moderate temperature and humidity swings compared to an open plantation. This may produce a more consistent leaf chemistry across seasons than unshaded plots. Record seasonal timing in every batch sheet — over time, patterns will emerge specific to your estate.
The published baseline for coffee leaf chemistry in this project (Cangeloni 2022) uses C. arabica Castillo variety at 1,700m in Colombia. Thumpassery's Chandragiri at 130m is a different variety at a dramatically different altitude. The numbers will not be the same.
| Factor | Cangeloni 2022 Baseline | Thumpassery Estimate |
|---|---|---|
| Variety | Castillo (Colombia) | Chandragiri (Kerala) |
| Altitude | ~1,700m ASL | ~130m ASL |
| Shade | Field cultivation | Rubber shade canopy |
| 5-CGA | 16.27 g/kg DW | Direction unclear — lower altitude and shade are each independently associated with lower phenolic content in C. arabica generally, but the combined effect on this variety has not been measured. May be lower; could also be offset by other factors. Requires measurement. |
| Mangiferin | 4.43 g/kg DW | Direction unclear by the same reasoning — shade is associated with lower phenolic production generally, but mangiferin specifically has not been studied across altitude/shade gradients. Could be lower, comparable, or higher. Requires measurement. |
| Caffeine | 7.94 g/kg DW | Higher caffeine in shade-grown C. arabica is documented in some studies — comparable or somewhat higher is plausible here, but this is not a Thumpassery-specific finding. Requires measurement. |
These are directional estimates, not measurements. The only way to know Thumpassery's actual leaf chemistry is to measure it. In the meantime, use the Cangeloni baseline as a framework — not as a specification — and let your sensory evaluation fill in the gaps.
↑ ContentsThe harvest is the first processing decision. Use this decision framework before picking.
| Destination | Best Leaf Age | Best Position | Season | Condition |
|---|---|---|---|---|
| Green Citane (α) | Young to mature (L1–L3) | Terminal flush preferred | Either; avoid heavy rain | Perfect — no damage |
| Oolong-style (β) | Mature (L2–L4) | Second and third pair | Dry season preferred | Clean, firm, uniform |
| Black-style (β deep) | Mature to old (L3–L5) | Inner and mid pairs | Either | Clean — mechanical disruption will be extensive |
| Roasted (δ) | Mature to old (L3–L6) | Any except fallen | Either | Clean — heat will manage minor variation |
| Engere-inspired | Mature (L2–L4) | Mid-branch | Either | Clean |
| Koji / Experimental | Mature (L2–L4) | Consistent batch preferred | Dry season preferred | Clean — microbial work requires good starting material |
Pick cleanly — no torn stems. Move picked leaves immediately into shade. Do not pile — heat builds quickly in a pile of freshly-picked leaves and begins unintended transformation within 30 minutes. Transport to processing area within two hours of picking. If you cannot process immediately, spread leaves in a single layer in a cool, shaded, well-ventilated space.
Before teaching tools or techniques, this section asks the single most important question: where are you going? Your destination determines your leaf choice, your toolbox selections, and your pathway. Start here, not at the toolbox.
Fresh, green, herbal. Possible slight grassiness. Clean brightness. Light body. The leaf at its most recognisable — minimal transformation.
Young to mature. Terminal flush (L1–L3). Freshly harvested, no damage. Dry season preferred.
Pathway 1 — Green Citane. Minimal processing: clean, heat-fix, dry gently. No disruption, no oxidation.
Rounded, honeyed, tea-like. Amber liquor. Moderate body. Smooth rather than sharp. The most immediately accessible Citane style.
Mature leaf (L2–L5). Mid-branch. Either season. Clean and uniform.
Pathway 3 (oolong-style, lighter oxidation) or Pathway 4 (black-style, deeper oxidation). Choose based on how rounded and full you want the cup.
Toasted, nutty, caramel, and roasted notes. Coffee-adjacent but distinct — the leaf's own character remains. Deeper, drier finish.
Mature to old (L3–L6). Interior leaves acceptable. Either season. Some variation in starting leaf is manageable.
Pathway 5 — Roasted Citane. Thermal domain. Temperature profile is the primary variable.
Unknown. Possibly floral, fruity, complex, or entirely unexpected. This is where genuinely new Citane identity may emerge.
Mature, clean, consistent. You need a stable starting point when the process itself is variable.
Pathway 6 (Koji-assisted) or Pathway 7 (Route Builder). Rigorous recording is non-negotiable for these batches.
| Destination | Leaf | Pathway | Confidence |
|---|---|---|---|
| Green / Herbal (α) | Young-mature, L1–L3, flush | Pathway 1 | High |
| Decoction / Traditional (β) | Mature, L2–L4, mid-branch | Pathway 2 | Medium |
| Amber / Tea-like light (β) | Mature, L2–L4 | Pathway 3 | High |
| Amber / Tea-like full (β) | Mature-old, L3–L5 | Pathway 4 | High |
| Roasted / Deep (δ) | Mature-old, L3–L6 | Pathway 5 | Medium-High |
| Koji / Complex (ε) | Mature, clean, consistent | Pathway 6 | Experimental |
| Open / Unknown (ε) | Mature, clean, your choice | Pathway 7 | Exploratory |
These are your tools. Each is independent. You can use one, several, or all of them. The order matters enormously — but the order is determined by the pathway you have chosen (Part 5), not by any fixed sequence. Read this section to understand what each tool does. Refer to Part 5 to know when and in what order to use it.
Withering is not just drying. It is a period of controlled moisture reduction during which the leaf's own enzymes begin to work at low, ambient-temperature rates. Glycosidases slowly unlock aroma-locked compounds. The leaf softens. Cellular structure changes slightly, making subsequent mechanical disruption more effective and more even.
The key distinction is between withering (ambient temperature, controlled) and drying (elevated temperature, intended to arrest chemistry). Withering activates. Drying stops.
Withering activates the Enzymatic Domain (Paper I, Section 26). Primary reservoirs engaged: E (Glycoside unlocking), K (Volatile precursor release). Withering without mechanical disruption does NOT significantly activate Reservoir C (Catechins via PPO) — PPO requires cell damage.
Rolling, wringing, cutting, or crumbling the leaf breaks cell walls, bringing the polyphenol oxidase enzyme (PPO) into contact with its substrates (catechins, chlorogenic acids) for the first time. This triggers the oxidation cascade that is central to amber-style Citane production. The degree of disruption is the primary control variable.
This tool has no reverse. Once the leaf is disrupted and oxidation begins, it cannot be stopped without heat. Choose the degree of disruption deliberately.
Mechanical disruption is the key activation event for the Oxidative Domain (Paper I, Section 18 and 24). Primary reservoirs engaged: C (Catechins → theaflavin analogues), A (CGAs → quinones), J (Lipids → C6 aldehydes via lipoxygenase). Irreversible above mild bruising.
After mechanical disruption, the leaf rests in open air while PPO drives the oxidation cascade. This is not a passive period — it is the most chemically active phase of the amber-style pathways. Colour shifts from green to yellow to amber to brown as catechin oxidation products accumulate. Time and temperature are your two controls.
Heat fixing (also called kill-green) denatures PPO and other active enzymes, freezing the leaf's chemistry at its current state. It is used in Green Citane to preserve the fresh, unoxidised profile; and in amber-style pathways to stop oxidation at the target depth.
Drying reduces moisture content to a level where biological activity ceases and the leaf can be stored without degradation. Target: moisture content below 8% (German tea standard, DIN 10809). This is non-negotiable — inadequately dried leaf will mould in storage and cannot be safely brewed.
Roasting is the thermal tool that drives the deepest transformation. At temperatures above approximately 130°C, amino acids react with sugars (the Maillard reaction) to produce toasted, nutty, and caramel aromas. At temperatures above 160°C, trigonelline degrades to produce pyridine compounds — the distinctive aroma of roasted coffee. The leaf must be fully dried before roasting — any residual moisture will steam rather than roast.
These are the routes through the leaf. Each pathway answers the three questions: what leaf, what tools, in what order. Each carries a confidence rating. Each connects to the traditions that informed it and to the Paper I chemistry that explains it.
Each pathway below is a conceptual procedure showing how the framework's tools (Part 4) can be combined toward a destination. It demonstrates how the framework describes a sequence of transformations — it is not necessarily a documented batch. Parameters (times, temperatures, ratios) are starting points drawn from analogous processing traditions and general food chemistry. Where a pathway predicts how Chandragiri leaf may behave relative to a published study, that prediction is directional, not measured — treat "may," "could," and "is expected to" as markers of an open question, not a guarantee. Treat the first several batches of any pathway as calibration, and record outcomes per Part 8.
Destination: Fresh, herbaceous, green. The intact leaf profile — lowest transformation, highest compound retention relative to the fresh leaf.
Best leaf: Young to mature (L1–L3). Terminal flush preferred. Freshly harvested. No damage.
Tools used: Heat fixing → Drying only. No withering, no disruption, no oxidation.
Infusion: pale gold to jade green. Aroma: fresh, grassy, herbal. Taste: clean brightness, mild vegetal note, low astringency if leaf was young. Light body. The most "coffee leaf"-tasting Citane — it references nothing else.
Heat fixing without prior disruption preserves Reservoirs A, B, C, E, and K in their most intact form. The cup will contain the highest proportion of unreacted compounds of any pathway. Oxidative Domain is explicitly avoided. [Paper I, Sections 14, 26, 33]
Destination: Full-extraction, robust amber cup. More body than infusion-style pathways. Coffee-leaf flavour is prominent rather than subtle.
Best leaf: Mature (L2–L4). Mid-branch. Either season.
Tools used: Drying → Grinding → Brewing by decoction.
Deep amber to brown liquor. Robust body. Coffee-adjacent but distinctly leafy. Higher bitterness than infusion pathways — this is normal for decoction. Sweetener (honey) can balance this for café service.
Medium confidence because the specific parameter optimisation for Chandragiri leaf at Thumpassery via decoction has not been done. The Engere tradition provides directional guidance. Expect to adjust ratio and duration across the first 3–5 batches before settling on your preferred parameters.
Destination: Rounded, tea-like, possibly honeyed. Amber liquor. Moderate body. The most accessible Citane style for a café context.
Best leaf: Mature (L2–L4). Clean and uniform batch.
Tools used: Wither → Roll lightly → Oxidise (partial) → Heat fix → Dry.
Gold to amber liquor. Aroma: floral, possibly honeyed, light vegetal note. Taste: rounded, smooth, tea-like. Mild astringency. Medium body. This is the closest Citane analog to a high-quality oolong tea — but with the distinctive mangiferin character of the coffee leaf underneath.
Destination: Deep amber to red-brown liquor. Full body. Robust, tea-like character with depth. The most structured of the infusion-style pathways.
Best leaf: Mature to old (L3–L5). Interior leaves acceptable. Clean.
Tools used: Wither → Roll fully → Oxidise (complete) → Heat fix → Dry.
Deep amber to red-brown liquor. Aroma: warm, woody, slightly malty. Taste: full body, structured astringency, possible honey or dried fruit note. Finish: long and dry. Similar depth to a quality black tea — but the xanthone pool (mangiferin) may create a characteristic softness not found in Camellia sinensis black tea (see Paper I, Reservoir B buffering hypothesis — unconfirmed).
Destination: Toasted, nutty, caramel, coffee-adjacent. For customers who want a roasted character without coffee's intensity.
Best leaf: Mature to old (L3–L6). Fully dried base essential.
Tools used: Dry → Roast → Cool → Store.
Brown to deep brown liquor. Aroma: nutty, toasted grain, caramel, slightly roasted. Taste: smooth, roasted character without coffee's sharpness. Body: medium to full. The trigonelline-to-caffeine ratio in the leaf differs from the bean — expect a roast character that is adjacent to coffee but not a replica (see Paper I, Reservoir F, "The Trigonelline Opportunity" — the ratio itself is measured; the resulting flavour is not yet confirmed). Bitterness should be structured, not harsh — if harsh, the roast went too far.
Destination: Unknown, and that is the point. The Koji treatment enriches the amino acid and sugar pools before any thermal processing occurs, potentially producing a richer, more complex cup than a direct-roast pathway on the same leaf. But the specific outcome for Chandragiri leaf at Thumpassery is not yet documented anywhere.
Best leaf: Mature, clean, consistent batch. When the process is variable, the raw material should be as stable as possible.
Tools used: Wither → Koji contact → Dry → Light roast.
Koji (Aspergillus oryzae) must be food-safe. Source it from a reputable supplier — do not improvise with unknown moulds. All equipment in contact with Koji must be scrupulously clean. This pathway requires hygiene attention that the simpler pathways do not.
Unknown — and intentionally so. The working hypothesis is: deeper Maillard character from the same base roast temperature, possibly with an additional aromatic dimension from terpene glycoside unlocking during Koji contact. Observe with an open evaluation rather than a predetermined expectation.
This is not a pathway with predetermined steps. It is a structured framework for designing your own route. The moment a route is prescribed, it stops being experimental. Paper I's philosophy — map the route after you walk it, not before — applies here entirely.
Use the six steps below as a design scaffold. Fill in each step with your own choices before you begin processing. The batch sheet in Part 8 then records what you decided, what happened, and what you found.
Which leaf are you starting with? Be specific. Age, position, condition, season.
Record your choice on the batch sheet before processing begins.
Where are you aiming? Or are you explicitly aiming nowhere — entering to observe?
Which tools from Part 4 will you use, and in what order? You do not have to use all of them.
Write out your intended sequence in order before starting. This becomes the processing sequence on your batch sheet.
For each step in your sequence: temperature, duration, humidity, weight before and after, any observations during the step. Be precise. "Overnight" is not a parameter. "14 hours at 23°C" is.
Brew using a consistent method (infusion at 90°C, 5 min, 1:80 ratio, as a starting standard). Evaluate: colour of liquor, aroma dry and wet, taste, body, finish, duration of finish. Use the sensory vocabulary from Part 7. Compare to the closest established Emergent State and note how it differs.
Assign the batch to an Emergent State or describe it in your own terms if it does not fit. Note: what was unexpected? What would you change? Is this worth repeating? If yes, could you repeat it from the record you just made?
That last question is the test of a good route map. If you cannot repeat it from your record, the record is incomplete.
The Route Builder is the practical implementation of Paper I's framework for observation (Section 39). Every completed Route Builder batch is a potential new Citane identity waiting to be found — and a route map waiting to guide the next practitioner who wants to go back there, or to compare an isolated intervention against a combinatory one (see Paper I, Section 39, "Observation as the Central Activity" and "The Engere Comparison"). [Paper I, Sections 38, 39]
Roasting is well understood. Oxidation is reasonably understood. Fermentation is where the unexplored territory lies — and it deserves its own treatment, separate from the seven pathways above.
Fermentation differs from every other tool in this manual in one important way: it has its own agenda. Heat does what you tell it. Enzymes act on what they're given. Microbes reproduce, compete, and change the environment they're in — sometimes in directions you did not choose. This makes fermentation the most powerful and the least predictable tool available.
It is also, on the evidence of Paper I's network model, a theoretically promising unexplored territory for new Citane character — though, as Paper I now frames it, one candidate among several rather than the unique answer. The intersection of glycosides, amino acids, microbial enzymes, and mild heat — discussed in Paper I, Section 39 as a frontier prioritised for practical accessibility — sits primarily in this domain.
↑ ContentsEvery leaf carries a population of bacteria and fungi on its surface — the epiphytic microbiome. This is your starting point for fermentation, even if you do nothing intentional. A withered leaf left at warm ambient temperature for an extended period will begin to ferment using this native population.
Koji (Aspergillus oryzae) occupies a special position: it is technically a fermentation organism, but its primary value to Citane is the enzyme cocktail it produces — proteases, amylases, and glycosidases — rather than its own metabolic byproducts. Full procedure is described in Pathway 6. This section covers the principles independent of that specific pathway.
Intentional inoculation with specific LAB or yeast strains offers more control than relying on native microbiota — at the cost of additional sourcing and hygiene requirements. This is genuinely exploratory territory for Citane.
Fermentation is the highest-risk tool in this manual. Knowing what failure looks like — and acting on it immediately — protects both the batch and your confidence to continue experimenting.
Ammonia smell: Protein putrefaction — undesirable bacterial activity has dominated. Discard.
Any mould colour other than expected: White/green is normal for Koji. Black, pink, or orange mould indicates contamination. Discard the entire batch — do not attempt to salvage by removing affected portions.
Sliminess: Indicates bacterial overgrowth, typically undesirable strains. Discard.
Off, chemical, or solvent-like smell: Possible undesirable fermentation byproducts. Discard.
Discarding a batch is not a failure of the process — it is the process working correctly. Every fermentation experiment that ends in a discard still produces information: record what conditions led to it, so the next attempt can be adjusted. The cost of a small discarded batch is far lower than the cost of serving an unsafe or unpleasant product.
Brewing is the last step of processing and the first step of evaluation. The same processed leaf can taste meaningfully different depending on brew parameters — and the cup is your record of everything that came before it.
Research on coffee leaf tea brewing gives a useful starting framework, while confirming that optimal parameters vary by processing style and leaf origin.
| Pathway | Method | Ratio | Temperature | Time |
|---|---|---|---|---|
| Green Citane | Infusion (steep) | 1:80–1:100 | 75–85°C | 3–5 min |
| Engere-inspired | Decoction (boil) | 1:100 | 85–100°C | 7–12 min |
| Oolong-style | Infusion (steep) | 1:80–1:100 | 85–90°C | 4–6 min |
| Black-style | Infusion (steep) | 1:80–1:100 | 90–95°C | 5–7 min |
| Roasted Citane | Infusion or decoction | 1:80–1:100 | 90–95°C | 5–8 min |
Lower brewing temperature and shorter infusion duration tend to reduce bitterness and improve consumer acceptance (Cao 2021, Ma 2024, cited in Yohannis 2026). For Robusta coffee leaf, optimal brewing was found at 93.4°C for 4.8 minutes (Hariyadi 2020). The Just-About-Right (JAR) studies (Fibrianto) found 95°C for 5 minutes optimal for decoction-style coffee leaf tea using a 1:100 ratio.
Do not agitate during steeping — stirring or pressing releases additional tannins and increases bitterness (functional compendium findings on withering-processed leaf).
Begin with 1:80 ratio, water just off the boil (90°C), 5 minutes, no agitation, filter promptly. Taste. If too bitter, reduce time to 3–4 minutes or temperature to 80°C on the next cup. If too weak, increase ratio to 1:60 or extend time slightly. Record what you land on for each pathway — your preferred parameters become part of your house standard.
This is the most practical section in the manual for day-to-day quality control. Each row connects something you taste to something that happened during processing — and what to try differently next time.
| Symptom in Cup | Likely Processing Cause | What to Try Next Time |
|---|---|---|
| Sharp, grassy, "raw" taste | Insufficient heat-fixing — PPO not fully denatured, or chlorophyll-dominant compounds dominant in an unintended way | Increase heat-fix duration or temperature. Check leaf snaps cleanly before brewing. |
| Flat, hay-like, dusty | Excessive drying time, or storage moisture re-absorption followed by re-drying — compounds have degraded through repeated stress | Reduce drying time/temperature. Check storage — airtight, dry, dark. Don't re-dry rehydrated leaf repeatedly. |
| Harsh, mouth-drying bitterness | Over-oxidation (too long in oxidation rest) OR over-roasting (temperature/time exceeded target) OR brew temperature/time too high | Shorten oxidation window and check colour more frequently. For roasted pathway, reduce roast time by 2 min increments. For brewing, reduce steep time or temperature. |
| Muddy, indistinct, "off" cup | Uncontrolled or excessive fermentation — possible undesirable microbial activity not caught in time | Reduce fermentation/oxidation duration. Check more frequently for off-smells. Ensure cleaner starting leaf next time. |
| Sour or vinegary note | LAB-dominant fermentation ran longer than intended, or oxidation rest extended into microbial territory | Heat-fix or dry earlier in the sequence. If intentional, this may indicate a Gamma-state direction — record and decide if it's desirable. |
| Weak, watery, underwhelming | Brew ratio too dilute, steep time too short, OR leaf was over-dried/over-aged and compounds have degraded | Increase ratio to 1:60. Extend steep time by 1–2 min. If leaf is old stock, consider it for the Roasted pathway instead. |
| Uneven colour — patches of green in an amber batch | Mechanical disruption was uneven — some leaf areas were not bruised/rolled sufficiently | Increase rolling time or pressure. Ensure leaves are turned/mixed during oxidation rest for even air exposure. |
| Burnt, acrid, ashy | Roast exceeded target temperature or time — likely scorching on pan surface | Reduce heat. Stir more frequently. Use a thermometer rather than judging by time alone. |
| Unexpected floral or citrus note | Possible Reservoir K (volatile precursor) release — terpene glycosides unlocked during extended wither or Koji contact | Not a fault — this may be a discovery. Record the exact processing sequence in detail. Attempt to repeat it. |
If Citane is genuinely exploratory, then record-keeping is not an appendix — it is part of the process itself. This section is as central to the manual as any processing pathway.
A batch without a record is an experience. A batch with a record is data. The difference matters enormously over time: ten batches without records are ten isolated experiences, each fading from memory. Ten batches with records are the beginning of a body of knowledge specific to your leaf, your estate, your equipment, and your hands.
The batch sheet below takes under three minutes to complete for a simple pathway, and perhaps five for an experimental one. This is a small cost for what it returns: the ability to repeat success, diagnose failure, and notice patterns that would otherwise be invisible.
Processing decisions rarely operate in isolation. Withering affects drying. Drying affects oxidation. Oxidation affects extraction. Fermentation affects multiple systems simultaneously. A batch sheet that records only the intended outcome misses most of what is actually happening.
When recording a batch, it is worth observing — and writing down — more than just "did it work":
Single interventions: what did this one step do, on its own, this time? Combined interventions: did this step behave differently because of what came before it? Sequence effects: would the same step in a different order have done something different? Unexpected outcomes: anything that didn't fit the plan — especially if it doesn't seem to matter at the time. It often does, later.
The framework in Paper I exists to improve observation, not to replace it. A detailed, honest record of something that didn't go as planned is worth more than a tidy record of something that confirmed what was already expected. (See also Paper I, Section 39, "Observation as the Central Activity.")
Photocopy this template, or recreate it in a notebook. One sheet per batch.
These show different levels of completeness — from a quick note to a fully detailed experimental record. All three are useful. More detail is always better, but a quick note is far better than nothing.
The three records below are illustrative formats showing how a batch sheet can be filled in — they demonstrate the level of detail this manual recommends. Where they describe specific outcomes (e.g. "gold-amber, floral/honey aroma"), treat these as example entries in the style records should take, not as a catalogue of confirmed Thumpassery results. As your own batch sheets accumulate, replace or supplement these examples with your own.
Quick-reference problems and solutions that don't fit neatly into the sensory diagnostic table — process-level issues encountered before the cup.
Most of the troubleshooting items above describe a single tool behaving unexpectedly. But many of the most useful things to notice happen at the junctions between tools — where the output of one step becomes the input to the next, and the combination behaves differently than either step would on its own.
A few junctions worth watching deliberately: withering × oxidation — a longer or more humid wither changes how fast oxidation proceeds afterward, not just how the leaf looks going in. oxidation × roasting — the degree of oxidation reached before heat-fixing changes what the leaf brings into the Maillard reaction during a later roast. fermentation × drying — how quickly and at what temperature a fermented batch is dried may shape which fermentation byproducts remain perceptible in the cup. Koji × thermal development — this is the pairing Pathway 6 is built around, and is itself an example of the principle.
None of these interactions are fully mapped. The troubleshooting table above will help you fix a single step that misbehaved. But the more valuable long-term records may be the ones that note: "I changed step 2, and step 4 behaved differently than it usually does" — even when step 4 itself was done exactly as before. The interaction may be as informative as either step alone.
| Pathway | Wither | Disruption | Oxidation | Heat Fix | Roast |
|---|---|---|---|---|---|
| 1 — Green | — | — | — | Steam 2-3min | — |
| 2 — Engere-inspired | — | Grind (dry) | — | — | — |
| 3 — Oolong-style | 12-18h, 20-26°C | Roll light, 3-5min | 2-4h, 20-26°C | Pan 5-8min / oven 90°C 15min | — |
| 4 — Black-style | 18-24h, 20-26°C | Roll full, 8-12min | 4-8h, 22-28°C | Pan 8-12min | — |
| 5 — Roasted | — | — | — | — | 130-160°C, 10-18min |
| 6 — Koji-assisted | 12h, 20-26°C | — | Koji 48-72h, 28-32°C | — | 130-140°C, 10-12min |
| Drying Method | Temperature | Duration | Best For |
|---|---|---|---|
| Sun (direct) | Ambient, ~30-40°C surface | 2-5 days | All pathways; risk of weather dependency |
| Sun (shaded) | Ambient | 3-7 days | Green pathway — gentler, preserves volatiles |
| Oven | 70°C | 4 hours | Most consistent; preferred for Green and Koji pathways |
| Pan (low heat) | 50-70°C | Variable, with stirring | Adds slight toast character; pre-roast pathways |
| If you want… | Pick this leaf | Avoid |
|---|---|---|
| Maximum freshness, brightest character | Young flush, L1-L2, dry season | Old, senesced, or stressed leaf |
| Balanced, versatile base for most pathways | Mature, L2-L4, either season | Mixed-age batches without recording |
| Mild, low-bitterness profile | Old or senesced leaf, clean | Young flush (too intense for mild target) |
| Consistent results for comparison/experiments | Single source tree, single harvest session, uniform age | Leaves collected over multiple days/trees |
| Roasted/thermal pathways | Mature to old, L3-L6 — variation matters less | Damaged or diseased leaf (still avoid) |
This manual is structured around branching: a destination is chosen, tools are selected, and a sequence is assembled from Part 4's toolbox. Paper I, Section 39 raises a related question from the chemistry side — whether some territories are reached through increasing process complexity (more steps, more domains, more sequence) or through increasing extraction depth within a comparatively simple pathway.
Pathway 2 (Engere-Inspired Decoction) is, by the standards of this manual, the simplest pathway — dry, grind, boil. Yet the Engere tradition it draws on is reported to achieve substantial extraction, sweetness, body, and sensory persistence through that simplicity, sustained over time and concentration, rather than through branching into multiple tools.
This raises an open question for Citane: can some territories of the coffee leaf be reached through depth — longer extraction, higher concentration, sustained time in a single domain — rather than through branching complexity? This is not a conclusion, and it is not intended to privilege Pathway 2 over the others. It is raised here as a comparison worth making: when a batch from a complex, multi-step pathway and a batch from an extended, simple decoction both land in similar territory, that convergence itself would be informative. When they don't, that's informative too.
By this point in the manual, the seven pathways, the toolbox, the batch sheet, and the troubleshooting tables might suggest that the goal is to follow a recipe precisely and obtain a predictable result — the way a baking recipe works.
That is not quite what this manual is for. The purpose of the framework is not to predict every outcome. It exists to improve the quality of observation, and to make comparisons between batches possible. A pathway followed exactly will still produce a slightly different cup each time — because the leaf itself varies (Part 2), because ambient conditions vary (Part 4, Part 9), and because some of the chemistry at work is still only partially understood (Paper I). This variation is not noise to be eliminated. It is the record from which understanding of this leaf, at this estate, will actually be built.
A pathway, in this sense, is less like a recipe and more like a question put to the leaf, phrased in a particular way. The batch sheet records both the question and the answer. Over many batches, patterns in those answers — not any single one — are what the framework is for.
The reservoirs, events, domains, and states in Paper I, and the seven pathways in this manual, are the categories available today. They are not a ceiling. A batch that doesn't fit comfortably into any existing category — an unexpected aroma, a sequence that behaves differently than its parts would predict, a traditional technique not yet considered — is not an error in the framework. It is exactly the kind of observation the framework exists to make room for.
Where a new observation doesn't fit, the right response is to record it precisely (Part 8) rather than to force it into the nearest existing category. Future revisions of both papers may add new reservoirs, new tools, new pathways, or new states — or may revise existing ones — based on exactly this kind of accumulated observation. The framework is intended to expand, and to be corrected, as Thumpassery's own body of batch records grows.
Chen, X.-M. & Kitts, D.D. (2018). Effects of processing method and age of leaves on phytochemical profiles and bioactivity of coffee leaves. Food Chemistry, 249, 143-153.
Rigling, M. et al. (2022). Influences of Processing on Chemical Composition. Foods, 11, 2553.
Monteiro, A. et al. (2019/2020). Dietary Antioxidants in Coffee Leaves. Antioxidants, 9, 6.
Campa, C. et al. Phenolic pool and light response in coffee leaves. Frontiers in Plant Science.
Fibrianto, K. et al. (2024). Consumer perception of decocted coffee leaf tea originated from different altitude. IOP Conf. Series EES, 1302, 012097.
Hariyadi, D.M. et al. (2020). Optimization of brewing time and temperature for caffeine and tannin levels. Potravinarstvo Slovak Journal of Food Sciences, 14, 58-68.
Yohannis et al. (2026). Ethiopian Engere traditional practices. Discover Food. DOI: 10.1007/s44187-026-00927-8.
Defri et al. (2021). Kawa Daun smoking and processing parameters, cited in Annazhifah et al. (2024), Food Research, 8(5), 282-288.
Kuti and Chemo traditional practice documentation — Citane/KoffyKraft project compendia.
This is a working manual, version 2.0. Guidelines are grounded in published research applied to coffee leaf generally, adapted for Chandragiri variety at Thumpassery Estate. Where parameters are estimated or extrapolated, this is noted in context — preference throughout is given to "may," "could," and "requires observation" over directional predictions stated as expected outcomes. As batch records accumulate (Part 8), this manual should be revised to reflect Thumpassery-specific findings.
Traditional practices (Engere, Kuti, Kawa Daun, Chemo) are referenced as inspiration and reference points, not as claims of authenticity or replication. Source research belongs to its original authors. Traditional knowledge belongs to the communities of origin.
This manual should be read alongside Citane Reactive Landscape (Paper I) — the theoretical framework explaining the chemistry behind every pathway in this manual. Paper I answers why; this manual answers how.
CITANE PROCESSING MANUAL · PAPER II · VERSION 2.0
KoffyKraft / Thumpassery Estate · Karavaloor · Kollam · Kerala · India
Companion to: Citane Reactive Landscape (Paper I)
Guidelines, not protocols · A working manual for an exploratory craft