005 — Floral & honey starter¶

WB-ING-2026-005 · status: designed · Disclaimer

Related: Maillard preparations · Skill user guide

Date: 2026-09-17 · Skill: fruit-maillard-prep v5 · Calculator: prep_calc.py on design-004.json

1. Request summary¶

Same brief as design-004 — floral and honey flavours and aromas, with the base beer explicitly out of scope — but asked for as the simplified, starter design: the one to actually build first, against design-004's advanced two-jar build.

Entry mode: 2, target-first. No mode 3 intake; the base beer was ruled out of scope, so this design reports its beer-side load rather than fitting it to a known beer.

Archetype: accent. Stage B concentrate 20 g/L (anchor 10–30), total fruit 23 g/L (anchor ≤100). Both clauses fit.

What "simplified" means here, concretely: one jar, one cook, one tincture, four weighed ingredients, two shopping items, and no A1, A2, Stage C, Stage D, kernel, inversion, phosphoric acid, protease, collagen, enzyme hold or warm hold of any kind. The only hold in the whole process is a 30-minute soak that starts at boiling point.

This design is also the brewery's first pH-calibration entry. data/batch-log.md's calibration table is empty — the meter arrived with skill v4 — so until three batches of consistent delta are in, the native-pH-anchored base solve is theory. A small, cheap, single-jar batch at the standard 6.15 target is the right shape for that first measurement, and it is a second reason to run this before design-004.

The binding constraint¶

The cook cannot make floral, so the tincture is not an optional extra — it is half the design.

Floral is monoterpenes: linalool, geraniol, citronellol, rose oxide, nerol oxide, hotrienol. A 121 °C cook rearranges or destroys them, which is why every floral row in ingredients.md is marked Stage T or D and never B. Worse, v4 sharpens the point on the other side too: free terpenes added before or during active fermentation are transformed, not merely faded — Sacch and Brett reduce geraniol to citronellol, cyclise linalool to α-terpineol, and acetylate both (§9, Med-high). So they cannot be cooked and they cannot meet a live fermentation.

That leaves the design with two separate jobs:

1. The jar makes the honeyed bed — phenylacetaldehyde → 2-phenylethanol, β-ionone, β-damascenone precursors, furanones. All heat-stable, all survive 18 months of Brett.
2. A cold spirit tincture makes the floral, dosed at bottling, to taste.

Do only the tincture and you get a perfumed beer with nothing under it. Do only the jar and you get a honeyed beer that is not floral. They are different operations, not different intensities of one.

Second constraint, which sizes the nitrogen: added FAN. 1 g of glycine into 4.5 L is 41 mg N/L at full carry-through (§10) and the warn is 30 mg N/L added. This design lands at 27 mg N/L with glycine 1.5 g and L-Phe 0.8 g. Anything much larger trips it.

2. Formula by stage (accent, one 4.5 L carboy)¶

Every stage is listed. Rows marked n/a are not used, with the reason. ×4.4 for 20 L.

Shortfall rule: there is no fruit mass to reallocate — the design has two fruit ingredients and one jar. If the apricot comes up short, cut the date proportionally and recompute the base dose: the apricot carries 18.5 of the jar's 20.1 meq of TA.

3. Nitrogen¶

• Glycine 1.5 g (19.95 mmol), L-Phe 0.8 g (4.84 mmol), collagen 0 g. No protease step — there is nothing to protease.
• Reactive amine 24.79 mmol against 49.6 g of reducing sugar = 11.1:1. The jar is glycine-only, so the 10:1 rule does not apply (it exists to out-number lysine and arginine in a collagen-bearing jar). The ratio sits inside v4's preferred 10–14:1 band.
• Collagen is 0 g and is not a candidate. At 4 g it would supply roughly 3 % of this jar's reactive amine and 100 % of its methionol, Lys and Arg — a 2–5 kDa hydrolysate has one free α-amino group per peptide, not per residue (§14). Dropping it also removes the papain step, which is most of what made the old version of this design fiddly.
• Protein carried in: 2.6 g (apricot 1.5, date 1.1). No almond, kernel, cacao or shiitake. Lys+Arg 47 mg.
• Glycine was held at 1.5 g, not raised: at 2.0 g the ratio improves to 9.0:1 but added FAN goes to ~31 mg N/L and trips the warn.

4. Base dose (Stage B only)¶

Native-pH-anchored solve (chemistry.md §12). Predict → dose 80 % → measure cold → titrate → log predicted vs measured.

Measure the mixture's native pH before anything goes in. The estimate is Low and the whole solve is anchored on it.

Note that near pH 6.15 the anchored and unanchored solves agree to 1.5 % — which is exactly why the old flat factor never looked wrong on a Stage B jar. The two methods diverge sharply only at a furanone-jar target; design-004 shows a 64 % gap at pH 4.4. That is the reason to run this batch at 6.15 for the first calibration entry: the arithmetic is least likely to be the source of any delta you measure, so the delta you record is telling you about the meter, the probe and the paste rather than about the solve.

No inversion, no phosphoric acid. The jar carries 5.1 g of sucrose. Inverting would recover about 3.7 g of reducing sugar into a jar whose ratio is already where it should be, and it would add a phosphoric dose that then costs base to neutralise. That 5.1 g is counted as 100 % fermentable in the beer either way (§18). This also removes the second pH measurement (the ~2.9 inversion target) from the process.

Stage C is not used, so the unbased-jar question does not arise.

Order of operations: soak the apricot → blend → reduce on the hob to 60–68 °Brix → cool → base to 80 % → measure → titrate to 6.15 → stir in the glycine and L-Phe → fill → seal. Nothing is ever sealed based and unreduced. Note that the KHCO₃ and the amino acids themselves read as Brix and depress a_w efficiently (§5), adding roughly 4 points to the final reading.

5. Ingredients list¶

Not needed: phosphoric acid, collagen, papain, sweet almond, kernels, shiitake, oak, vanilla, hibiscus, marigold, cacao, aronia, sloe, blackcurrant, quince, rosehip, muscat raisin, maltodextrin, and any fresh fruit at all.

Vessel check: none. The whole prep is about 90 g of paste plus 20–60 mL of tincture. It goes into the existing carboy with no volume problem.

6. Cook and process¶

Measure first: calibrate the pH meter at 4.0 and 7.0, then take the native pH of the blended mixture before anything is added. Record it. The base dose is anchored on it and this is the brewery's first calibration entry.

Holds (§21). There are none. The apricot soak starts at ~95 °C and cools; it is not a warm hold and it needs no time limit beyond "use it the same session". Nothing is enzyme-treated raw, nothing is held at 40–50 °C, nothing warm-held enters the beer uncooked.

Method.

1. Pour 60 mL of just-boiled water over the 50 g of chopped dried apricot, cover, 30 min.
2. Add the chopped date; blend to a purée with a stick blender.
3. Take the native pH. Record it against the predicted 3.50.
4. Reduce uncovered on the hob until a thin smear or the pressed liquid reads 60–68 °Brix on the jam refractometer. Sizing: about 75 g of water comes off, taking ~165 g of mixture down to ~90 g. prep_calc.py's known-solute lower bound is 55.0 °Brix as mixed and 90.2 g at target, before the 60 mL of soak water is counted — so the real reduction is a little more than it says. The refractometer decides, not the arithmetic. Never above 72 °Brix.
5. Cool. Add 1.52 g of KHCO₃ (80 % of the full dose) in two or three portions — it foams. Stir.
6. Measure cold with the flat-surface or spear probe, then titrate the last of the KHCO₃ in to pH 6.15. Record predicted (1.90 g), first-pass measured pH, and the actual grams it took. That is the batch-log calibration line.
7. Stir in the glycine and L-phenylalanine at base-adjustment time.
8. Fill a sanitised Kilner to 85–90 % (headspace 10–15 %), lid finger-tight. ~90 g of paste is about 70 mL, so use the smallest jar that gives that fill.
9. 121 °C, 90 min, counted from full pressure, jar standing in 3–5 cm of water — never on the cooker base, and never below the cooker's stated minimum liquid over a 90-minute run.
10. Hot vent: open the jar and stir 5–10 min while still hot. Furan boils at 31 °C and leaves; the pyrazines, furanones, melanoidins and norisoprenoids stay. See §15.
11. Slow-cool, closed. The jar is at a_w ≤0.86, below every relevant growth minimum, so the slow cool is free — and hours in the 60–100 °C band continue Amadori decomposition and residual Strecker chemistry. Refrigerate afterwards; the labelled jar is the storage vessel.

No roast finish. It pushes pyrazines and bitterness — the wrong axis here — and it is untested (§1, paired-jar only).

The tincture, separately and first. Elderflower 8 g, chamomile 3 g, coriander seed 2 g (cracked), rose petal 4 g if used, in 100 mL of 40 % neutral spirit. 48 h, dark, room temperature, strained through muslin. Keeps for months sealed. Make this before the jar — it needs 48 hours and no equipment, and tasting it on its own is how you learn the dose rate before committing anything.

Dosing the tincture: bench it first. 1 mL into 100 mL of the finished beer is the reference ratio; scale from what that tastes like. Typical 20–60 mL into 4.5 L at bottling.

7. Predicted compound inventory¶

At cook: a dark amber paste, apricot and honey over a faint nutty edge, no cocoa or roast. In beer at 6 months: honey and dried apricot over the base beer, with the tincture's elderflower clearly on top. At 18 months: damascenone and ionone grown, the tincture's fresh top thinned, the melanoidin body and pyrazine edge unchanged.

8. Survives vs destroyed (18 months)¶

Survives intact: 2-phenylethanol (rose, honey); β-ionone (violet, tea) and β-damascenone (honey, stewed apple) — the latter still increasing; furanones and maltol (cotton candy, burnt sugar; furaneol partly reduced or methylated over a long secondary, Med); alkylpyrazines (roasted hazelnut); terpene oxides from the tincture (tea, lychee, floral); melanoidins (body, colour, mild bitterness).

Reduced or transformed: - Phenylacetaldehyde → 2-phenylethanol. Honey becomes rose. On-target either way. If some of the aldehyde is wanted in the finished beer — its threshold is 5–10 µg/L, so very little is needed — hold 20 % of the jar back to bottling and account for its ~9 g of sugar outside the gravity gate. - Free monoterpenes are transformed, not just faded (§9, Med-high): geraniol → citronellol, linalool → α-terpineol, both partly acetylated. This is the whole reason the tincture goes in at bottling and not before. - Furfural and HMF → furfuryl alcohol (neutral loss).

Partial protection: EC-1118's SO₂ forms reversible bisulphite adducts with aldehydes, releasing them slowly in bottle (Med).

9. Fusel, methionol, 2-PE, FAN and THP budget¶

10. Kernel calculation¶

n/a. No kernels. No A2, no benzaldehyde, no HCN, no % ARfD, no ethyl-carbamate route, and no dark-storage instruction needed on that account.

11. Colour and body¶

Melanoidin colour is permanent and non-fermentable. About 90 g of a dark Stage B paste into 4.5 L, from a jar run at pH 6.15 where browning is fastest. There is no anthocyanin in the design, so nothing is lost to thermolysis and nothing red arrives.

Estimate: +2 to +3 SRM, toward amber-gold. Confidence Med-low. Body rises slightly from the melanoidins. Tweak 2 roughly halves this if the base beer is pale and the floral needs to read before the first sip.

12. Fermentable load¶

Includes all 5.1 g of sucrose, uninverted. Plan on the worst case.

prep_calc.py warning, quoted in full — this is the design's only warning: "worst-case fermentable 12.2 g/L: add early in secondary and hold the gravity gate absolutely."

The gravity gate. Ferment to a stable gravity, read twice 4–6 weeks apart, before bottling. The jar goes in early in secondary and is covered by it.

Outside the gate: only the tincture, 20–60 mL at bottling. It carries no sugar; its cost is ABV only, +0.18 to +0.53 %. Total ΔABV with the tincture: about +1.0 to +1.3 % worst case.

13. Beer-side acid, potassium and FAN¶

14. Safety — warm holds (§21)¶

There are no warm holds in this design. The apricot soak starts at ~95 °C and cools through the growth range in the wrong direction; it is not a hold.

• Nothing held warm enters the beer uncooked — vacuously true, since nothing is held warm.
• No enzyme touches raw fruit. The Lallzyme goes into the beer post-primary at 50 mg/L, which is its designed use and not a §21 risk.
• No low-acid material is held warm. The jar mixture is low-acid after basing (pH 6.15) but it goes straight from basing to the cooker.
• Stage B Brix confirmed before seal. Jar B at pH 6.15 is a low-acid mixture and its only hurdle is water activity: a_w ≤0.86 at 60–68 °Brix, below the growth minimum for C. botulinum (~0.93), B. cereus (~0.92) and C. perfringens (~0.93). That hurdle does not exist until you have actually reduced. Read the refractometer before sealing.
• Slow-cool closed, then refrigerate. There is no high-moisture Stage C jar, so the ice-bath-and-5-days rule does not apply.

15. Safety — thermal-process compounds (§22)¶

Acrylamide levers this design pulls:

1. Glycine is present at 1.5 g and competes for the dicarbonyl pool — a recognised acrylamide-reducing additive. The glycine-led nitrogen policy is doing real work here, not incidentally.
2. Backbone. Medjool date is the lowest-asparagine reducing-sugar donor on the list, and asparagine is the precursor. Fruit is never the nitrogen source anyway.
3. pH is the lever this design does not pull. It runs at 6.15, which is the higher-formation end. That is a deliberate trade: the Strecker route to phenylacetaldehyde is the whole point of the jar. Tweak 2 pulls this lever instead, at the cost of the honey aldehyde.
4. No roast finish; the 140 °C ceiling is never approached.

Scale, for the record: at 500–1000 µg/kg in paste, ~90 g of paste into 4.5 L is roughly 4–8 µg per 375 mL bottle, against a typical adult dietary intake of 30–130 µg/day. Low confidence, model-system extrapolation. Not alarming; not zero.

Furan. A sealed jar of sugar and amino acid at retort temperature is the textbook condition. Nothing in this design is notably ascorbate-rich — no rosehip, sea buckthorn or blackcurrant — so the exposure is at the lower end, but the condition still applies. Hot vent, 5–10 min open and stirring while hot, before blending or refrigerating. Confirmed at step 10 in §6. Effectively free.

HMF. Not inverting, so there is no over-acidified route to it. If the paste comes out of the reduction much darker than expected, check the reduction rather than the acid — there is no phosphoric to have overshot.

Confidence¶

What this batch actually settles, and why it is worth running before design-004:

1. The first pH calibration entry. Predicted 1.90 g against measured. One batch of delta is a measurement; three consistent ones become a standing brewery offset (chemistry.md §12).
2. Whether 0.8 g of L-Phe reads as honey and rose at all, which is the assumption design-004 spends its entire FAN budget on.
3. The tincture dose rate in mL per 4.5 L, which is a number with no prior data behind it and which every later design in this programme will reuse.
4. Whether a 60–68 °Brix, properly based, nitrogen-loaded jar tastes clean, against batch 001 at 30 °Brix and 40:1 — and whether pH 6.15 tastes acrid, which is what an overshoot above 7 shows as within minutes.

Tweak points¶

1. Push the honey harder. L-Phe 0.8 → 1.2 g (2-PE 52–103 mg/L, midpoint ~77) and glycine 1.5 → 1.2 g to hold added FAN at ~27 mg N/L. The ratio moves to 13.7:1, still inside the preferred band. This is the cheapest single upgrade in the design.
2. Run it as a furanone jar instead. Target pH 4.4 rather than 6.15. Less colour, substantially less acrylamide, a much smaller base dose — and recompute the factor from the measured native pH, do not reuse 0.5 or any flat number (design-004 shows a 64 % gap at this target). Costs the Strecker route, so 2-PE drops to the Ehrlich share alone, roughly 16–32 mg/L. Gains furanones and maltol. A good paired jar against the base design.
3. Tincture only, no jar. Make the tincture, bench it at 1 mL per 100 mL, dose it. One evening, no cook, no arithmetic — and it isolates the single variable with no prior data behind it. Legitimate as a first move on its own.
4. Add a Stage D. 200–300 g of raw, frozen-then-thawed quince into the beer early in secondary, with the beer-side Lallzyme already in the design to release it. This is the one step that adds the bound floral pool — the thing that makes a beer get more floral between 6 and 18 months — and it is the single largest step from here toward design-004. Costs ~12 g of fermentable, one extra addition, and it needs quince, which in the UK means waiting for October. It also repairs the design's one soft spot: the carotenoid the unsulphited apricot has lost.

Log¶

2026-09-17 · Designed¶

Intaken from design-004-floral-honey-starter.md, generated by fruit-maillard-prep v5. Designed only: nothing bought, cooked or dosed.

2026-09-22 · Renumbered¶

Design number changed from 004 to 005 so that it matches the page ID, WB-ING-2026-005. References to the maximum-intensity design now read design-004 (was design-003). The source file names (design-004-…, design-004.json) are unchanged.