Marzipan Flavour against Cyanide Cost¶

Version 1 · Disclaimer

The cyanide dose page answers one question: is this charge safe. This one answers the other: how much marzipan does it actually buy, what does that flavour cost in cyanide, and which way of preparing the kernel gives the most of the first for the least of the second.

The two are locked together at the point of release. Amygdalin hydrolysis yields one benzaldehyde and one hydrogen cyanide per molecule, so a kernel cannot deliver aroma without delivering cyanide in a fixed ratio: 3.93 mg of benzaldehyde per mg of HCN by mass. Nothing in the jar or the carboy changes that ratio at the moment of hydrolysis. What the processing route changes is what happens afterwards, because the two products are lost at different rates. A sub-boiling vent removes most of the HCN from an aqueous slurry and only some of the benzaldehyde; a Stage C cook costs benzaldehyde and removes no cyanide at all. The ratio that reaches the bottle is therefore a property of the route, and worth choosing deliberately.

The unhydrolysed fraction is the other half of the argument. Amygdalin that survives the A2 hold is not safe amygdalin, it is deferred cyanide: Brettanomyces β-glucosidase finishes the job over months in a sealed bottle, where nothing can be vented, so the model retains it at 1.0 for both products. That is why a raw kernel and a hydrolysed-but-unvented one come out identical on cyanide. The vent is the only step in the whole sequence that removes any.

The bench¶

It also runs on its own at /assets/apps/kernel-marzipan-bench/index.html if the frame above is awkward on a phone.

Reading the flavour scale¶

Benzaldehyde is what marzipan smells of, so its concentration in the finished beer is the flavour gained. The scale runs on the same principle as the cyanide one: a log axis with the bands marked and the charge's position on it.

The ceiling is an input, because it is a preference rather than a fact: someone building a noyau-forward beer will set it higher than someone adding a marzipan edge to a kriek.

The routes and what they retain¶

Entry is amygdalin rather than HCN, so the two potentials stay stoichiometric: a kernel at \(A\) mg amygdalin per gram carries \(0.2320A\) mg of benzaldehyde and \(0.05909A\) mg of HCN. With \(c\) the A2 hydrolysis completeness and \(s\) the benzaldehyde surviving in the beer:

Vent residual 0.40 for HCN and survival 0.70 for benzaldehyde, Stage C survival 0.50, post-primary survival 0.8–1.0, default completeness 0.70: all Med-Low except the stoichiometry, and all from the cooked-fruit primer §8.

The cyanide cost is the headline number: milligrams of HCN per litre of beer for every mg/L of benzaldehyde delivered with it, so lower is more flavour per unit of hazard. At the planning defaults it is 0.21 for the recommended route (A2, vent, Stage D), 0.28 for a raw kernel, 0.30 for a Stage C jar and 0.62 pre-primary. The Stage C route being worse than doing nothing is the result worth staring at: the cook burns off aroma while the deferred cyanide rides through untouched. That conclusion rests on the Stage C benzaldehyde survival of 0.50, the weakest number on the page, and it is a good candidate for a paired jar test.

Flavour-limited or toxicity-limited¶

Setting the charge that reaches the flavour ceiling against the charge that puts one bottle at half a reference dose — the primer's warn line — gives the regime:

• flavour-limited, the comfortable case: the beer stops being pleasant before it stops being safe, so the sensory judgement is also the safety margin;
• toxicity-limited: full marzipan intensity is not available without crossing the warn line, and the charge has to be capped on the arithmetic rather than the nose.

The regime does not depend on the kernel's amygdalin content or on the batch size — both scale the flavour and the hazard identically — only on the route, the bottle size, the drinker and which reference dose is used. At 375 mL, 70 kg and the EFSA acute reference dose every route except pre-primary is flavour-limited, which is the reassuring answer; on the UK COT value of 5 µg/kg none of them is.