Repeatable, not occasional — the chain behind a certificate of analysis

Any competent operation can produce one excellent lot of dried aromatic material. Producing the same lot again, ten runs later, in a different season, with different staff on the floor, is a completely different problem — and it is the only one that matters to a buyer, because a buyer is never purchasing the batch that was photographed. They are purchasing the next one.

The instinct is to reach for the certificate of analysis. It is the wrong instrument for the question. A certificate describes one sample, on one day, against the parameters someone chose to test. It is a necessary baseline and it is genuinely good at catching what it was designed to catch. What it cannot tell you is whether the process behind it will hold. Two lots with near-identical paperwork can arrive weeks apart in completely different condition, because almost everything that determines condition happens either before the sample was drawn or after.

This guide walks the chain end to end, in the order it actually runs: variety selection, growing environment, inputs, handling before drying, the drying rate itself, conditioning, the water activity gate, cold storage and release cadence, packaging, and what a lot code has to resolve to. The vocabulary is shared across every aromatic crop that is dried rather than sold fresh — hops, lavender, peppermint, chamomile, rosemary, green tea, saffron. The species differ. The physics and the discipline do not.

1. It starts with the variety, not the batch

The largest single source of variation between lots is not the process. It is the plant material going into it. Two plants of the same nominal variety, grown side by side, can differ substantially in essential oil content, in the ratio of compounds within that oil, and in how well the dried material holds its structure. Selection is how that variance gets narrowed, and it is slow work: candidates are trialled across multiple runs, and only the ones that reproduce their behaviour are kept.

The traits worth selecting for are unglamorous and mostly physical:

• Structure and density. How the material holds together determines how it dries, how far bound water has to travel, and how much of it survives handling intact.
• Yield of the aromatic fraction. In hops this is measured as alpha acids and total oil; in lavender and peppermint as essential oil percentage on a dry basis. It is heritable, it varies enormously between varieties, and it is the difference between a crop that smells like itself on arrival and one that does not.
• Reproducibility. The trait that matters most and is measured least. A variety that performs brilliantly once and unremarkably three times is worse than one that performs well every time, because a supply chain is built on the second kind.

In practice this is why varieties carry names at all. A house designation such as Toad Venom or Uncle Snoop is not decoration and it is not a marketing label. It is the handle a record hangs on. Hops have worked this way for a century: Citra, Mosaic and Simcoe are names attached to a specific, repeatedly propagated line, and a brewer buying one of them is buying the expectation that it behaves the way it behaved last year. A named variety that drifts is worse than an unnamed one, because it carries a promise the material no longer keeps.

A cross written out, as Animal Face x Sin Mintz is, tells a reader the parentage and almost nothing else. Parentage sets the range of what is possible. It does not say which of the two parents the material will favour in a given season, how it will behave on a drying rack, or whether it will reproduce itself at all. That is what trialling is for, and it is why a designation like Uncle Snoop is only worth something once it has been run enough times to be boring.

This is the part of the chain a certificate of analysis is structurally unable to see. It reports what a sample contained. It says nothing about whether that composition was a property of the variety or an accident of the season.

2. A controlled environment is a repeatability tool

Growing under controlled conditions is usually described as a quality decision. It is more accurately a variance decision. An outdoor season delivers whatever it delivers: a cool wet August, an early frost, three weeks of unusual light. The material still gets harvested, and the differences travel all the way down the chain into the finished lot. Controlled rooms remove that input, which is why the argument for them is not "better" but "the same, again".

The variables being held steady are the ordinary ones — light intensity and photoperiod, temperature and its day-to-night swing, relative humidity and vapour pressure deficit, air movement, and carbon dioxide. None of them is interesting on its own. What is interesting is the discipline of keeping each within a narrow band cycle after cycle, and of logging them so that a difference between two lots can actually be traced back to a cause rather than guessed at.

Dedicated rooms matter for the same reason. A space used for one crop, on one schedule, with one set of conditions, has fewer ways to drift than a shared one. It also makes the sanitation cycle between runs a routine rather than a negotiation.

3. Clean inputs set the ceiling

What goes in shapes what comes out, and the two inputs that matter most are also the two that get the least attention.

Water. Municipal supply carries dissolved minerals that vary by source and by season, and those minerals accumulate in the growing medium over a cycle. Reverse-osmosis purification removes that variable so the feeding programme starts from a known baseline instead of an unknown one. This is not a purity claim; it is the same repeatability argument as the controlled room, applied to the irrigation line.

Crop protection. Residues concentrate as material dries — remove three-quarters of the mass as water and whatever was on the plant is now present at roughly four times the concentration per gram. That arithmetic is why a residue policy has to be a policy about what is applied, not a policy about what is detected afterwards. Testing catches what it screens for. It cannot catch a compound it was not asked to look for, and no screen covers everything.

4. Handling: the aromatic fraction is on the outside

In most aromatic crops the compounds people are paying for are not distributed through the tissue. They are concentrated in surface glandular structures — the lupulin glands of hops, the peltate glands on lavender and mint leaves — and those structures are fragile, exposed, and detach easily. Every additional transfer, every drop into a bin, every rough pass through a machine knocks some of them off.

This is the argument for hand-sorting and for minimising the number of times material is moved. It is not sentimentality about craft. It is that the aromatic fraction is physically located in the part of the material most likely to be lost to mechanical handling, and there is no downstream step that restores it. Losses attributed to age are frequently losses from handling that happened on day one.

The same reasoning applies at the other end of the chain, which is why a receiving procedure that tips a delivery from one container into another before assessing it has already cost something.

5. Drying is a rate problem, not a target

Fresh plant material is roughly three-quarters water by weight. Drying takes that down to somewhere near a tenth. It is tempting to treat that as a destination — get it dry, then move on — but almost everything that determines the quality of the finished material is decided by how fast the water leaves, not by where the process stops.

Water moves out in two stages. First, surface and free water evaporates quickly. Then the bound water held inside cell structures has to migrate outward before it can evaporate at all, and that migration is slow. Push the drying environment too hard and the outside dries and case-hardens while the inside is still wet — a lot that feels dry in the hand, weighs as if it is dry, and then re-moistens itself from the inside within a day of being sealed.

The conventional environment for a slow dry sits at roughly 15–20 °C and 55–62 % relative humidity, with gentle indirect airflow, over seven to fourteen days depending on density. Every one of those three variables can substitute for another, which is where operations go wrong: raising temperature to compensate for high humidity dries the surface faster, not the core, and volatilises exactly the aromatic compounds the crop was grown for.

Density is also why one schedule cannot simply be copied across varieties. Run a single drying curve on a dense, tightly structured lot such as Toad Venom Flower, and the same curve on a looser, more open one such as Uncle Snoop Flower, and the second reads dry on the outside days before the first while still holding a comparable amount of bound water inside. Two lots, one schedule, two different outcomes, and only one of them has a problem that shows up later.

These are conventional operating ranges for aromatic botanicals generally, not a specification for any particular crop. The written instructions supplied with a consignment beat any general table, including this one.

Kiln drying at higher temperatures is standard practice for some crops and a deliberate trade: speed and throughput against essential oil retention. It is a legitimate choice when the material is destined for extraction. It is the wrong choice when the aroma of the material itself is the product.

6. Conditioning: the part that looks like nothing happening

Conditioning is what happens after drying, in a sealed container, over weeks. From the outside it looks like storage. Three things are going on inside it, and none of them can be rushed.

Moisture equalises. Even a well-dried lot is uneven — denser parts hold more water than thin ones. In a sealed container the wetter parts release moisture into the headspace and the drier parts take it back up, until the whole lot sits at one water activity instead of a range. This is the reason a container is opened briefly each day in the first week: the released moisture has to leave, or the equalisation simply stops at a level that is too high.

Residual plant compounds break down. Chlorophyll and simple sugars left in the tissue degrade slowly, and their breakdown is what removes the raw, grassy, hay-like note of freshly dried material. This is enzymatic and temperature-dependent, which is why conditioning warm goes faster and worse — the same warmth accelerates aromatic loss.

Volatile aromatics either stay or leave. The compounds that give aromatic botanicals their character are volatile by definition: they evaporate at ordinary temperatures, and the lighter ones evaporate first. Every hour spent warm, every extra opening of the container, and every transfer between containers costs some of them. A conditioning phase done cool, in the dark, with the lid opened deliberately rather than often, keeps more of them than one that is not.

Two to four weeks is where most of the change happens. Beyond about eight weeks the returns are small and the losses continue, which makes "conditioned for six months" a claim about storage rather than a claim about quality.

7. Water activity, used as a release gate

Almost everyone quotes moisture content — water as a percentage of total weight. It is easy to measure and it is the wrong number, because it says how much water is present and nothing about how much of it is available. Two lots at an identical 11 % moisture can behave completely differently depending on how tightly that water is bound into the tissue.

Water activity (written a_w) measures availability directly. Seal a sample in a small chamber, let it come into equilibrium with the air, and measure the relative humidity of that air; a_w is that figure expressed from 0 to 1. Pure water is 1.00. It is the number that microbiology actually responds to, which is why food, pharmaceutical and botanical handling all converged on it.

Note the asymmetry: too wet is a safety problem that can be caught and corrected, too dry is a quality problem that cannot. Re-humidifying an over-dried lot restores the weight and the feel. It does not bring back what evaporated.

A release decision looks mundane written down, and that is the point. A lot of Toad Venom read at 0.58 a_w at 21 °C is inside the band and can be released. The same lot read at 0.68 goes back to conditioning and is read again. The number decides, not the schedule, and the figure that gets recorded is the one taken at pack-out rather than the one taken on a good day two weeks earlier.

A reading is only useful where it is placed

The difference between an operation that measures water activity and one that uses it is where in the process the reading sits. Measured after dispatch, it is a record — something to consult when a complaint arrives. Measured immediately before release, as a condition of release, it is a gate: a lot outside the band does not ship, and the failure is caught by the people who can still do something about it.

A figure is also only meaningful with its temperature, because a_w shifts with it — typically a few thousandths per degree. A reading taken on a cold morning in an unheated room is not comparable to one taken beside a warm window. A useful record states the value, the temperature it was taken at, the instrument, and when it was last calibrated against a saturated salt standard. Anything less is a number, not a measurement.

8. Cold storage, and the cadence nobody prices

Cold storage slows every degradation pathway that matters: oxidation of the essential oil fraction, enzymatic breakdown, and the loss of the lighter volatiles. For bulk aromatic crops it is standard — the hop trade has stored baled material cold for decades for exactly this reason, because alpha acids and oil measurably decline at ambient and measurably hold when cold.

But cold storage is a way of slowing loss, not preventing it, and this is where a second variable does more work than most buyers realise: elapsed time from harvest to arrival. A lot dried and conditioned in the last two months, held cold, and shipped within days of being packed will usually outperform an older lot that was stored impeccably. Freshness is not a marketing claim here; it is the compounding of a rate.

Which is why release cadence is a quality variable in its own right. An operation that harvests on a regular rhythm and ships close behind it has a structurally younger inventory than one that produces in large infrequent runs and draws down from them for a year — regardless of how good the cold room is. The relevant question to a supplier is therefore not "how do you store it?" but "how old is what you are about to send me?", and the answer should be a date rather than an adjective.

9. Packaging: four variables, all of them cheap to get right

Once a lot is stable, everything that degrades it further comes through the packaging. There are only four routes, and each has an unglamorous, well-understood answer.

• Light drives photodegradation, and ultraviolet does most of the damage. Amber glass blocks a useful part of the UV range; opaque and foil-lined packaging blocks all of it. Clear packaging on a lit shelf is the single most common self-inflicted storage error, and it is also the one that looks the most appealing.
• Oxygen drives slow oxidation of exactly the compounds worth keeping. Reducing headspace helps; an oxygen absorber helps more; nitrogen flushing at pack-out helps most, and it is why vacuum-sealed and gas-flushed formats hold aroma longer than a jar does.
• Moisture exchange works both ways. A two-way humidity sachet holds the pack at a stated relative humidity — 58 % and 62 % are the common grades — by absorbing or releasing water as needed. It is a buffer, not a repair: it will not rescue material packed above the mould threshold, and it does not replace a water activity reading.
• Headspace is the one people forget. The air in a half-empty container is a reservoir of oxygen and of moisture that the material will come into equilibrium with. Right-sized containers, or fewer and fuller ones, beat one large container opened repeatedly.

Sealing at the point of packing rather than at the point of dispatch matters for the same reason. Material that sits packed but unsealed in a warehouse for a week has been exchanging with room air for a week, whatever the pack date says.

10. Traceability: what a lot code has to resolve to

A lot identifier is worth nothing on its own; it is worth a great deal when it resolves to a record. A workable record ties the code to the harvest date, the pack date, the water activity at pack-out and the temperature it was read at, the packaging format, the storage condition since, and the test result for that lot rather than for a representative one.

A worked example makes the difference concrete. A code such as #190 Toad Venom should return, on request and without a delay: the harvest week it came off, the date it was packed, the water activity at pack-out and the temperature it was read at, the format it went into, the storage condition it has been held in since, and the certificate issued against that code. Six fields and a document. #09 Uncle Snoop should return its own six, not the same six with a different label on top. If any of them has to be looked up, reconstructed or approximated, the record was not being kept. It was being assembled after the question was asked. The designation is part of what the code resolves to, not a label applied afterwards. #190 Toad Venom Flower and #09 Uncle Snoop Flower are two records, two harvest weeks and two release readings, and the only thing keeping them apart six months later is that both were written down at pack-out.

That last distinction does most of the work. A certificate that is presented for a product line, with no lot reference, is a statement about something that was once tested. A certificate carrying the same lot code as the pack in front of you is a statement about the pack in front of you. They look almost identical on a screen and they are not remotely the same claim.

When a supplier can produce that record on request without a delay, it usually means it was being kept all along rather than assembled afterwards — which is, in the end, the only thing a buyer can actually verify from the outside.

The certificate is the baseline, not the difference

Read back through the chain and the pattern is hard to miss. Variety selection, environmental control, input quality, gentle handling, drying rate, conditioning, the release gate, cold storage, cadence, packaging and traceability — a certificate of analysis touches exactly one of them, at one moment, and only for the parameters someone thought to order.

This is not an argument against testing. Testing is the floor, it is non-negotiable, and material without it should not be bought. It is an argument against treating the floor as the ceiling. A lot can pass every screen it was given and still arrive dry, flat, brittle, or six months older than the buyer assumed, because none of those outcomes is what the screens were measuring.

The useful reframing is that quality is a property of a process, not of a batch. A good batch can happen by luck. A good tenth batch cannot. Everything above is simply the list of places where luck gets designed out — which is why the questions worth asking a supplier are about the chain rather than about the paperwork, and why the answers should be dates, numbers and procedures rather than adjectives.

Eight things that should stop you

1. A certificate with no lot reference. It describes something that was tested once, not the pack in your hand.
2. No water activity figure anywhere. Moisture content alone does not answer the question that matters.
3. A reading with no temperature attached. The value is not comparable to anything without it.
4. A pack date that is missing or vague. "This season" is not a date, and elapsed time is one of the two largest variables in the chain.
5. Clear packaging that has been sitting in light. Whatever the paperwork says, photodegradation has already happened.
6. Visible condensation inside the container. A local water activity spike, and a mould risk regardless of the lot average.
7. Dust or fines at the bottom of the pack. Over-drying, rough handling, or both — and a preview of what the rest will do.
8. A flat or absent aroma on opening. The volatile fraction is the first thing lost and the hardest thing to fake.

None of these require instruments to catch. They are all visible in the first ten minutes after a delivery is opened, which is what makes them worth building a habit around.

A ten-point receiving checklist

1. Record the arrival date and the condition of the outer carton before opening anything.
2. Check the packaging format against what was ordered — light barrier, seal type, pack size.
3. Confirm the lot code on the pack matches the delivery paperwork and the certificate.
4. Ask for the pack date and the water activity at pack-out, with the temperature it was read at.
5. Open one pack in a cool, dry, low-light space rather than wherever the box landed.
6. Assess aroma immediately on opening, before the headspace exchanges with the room.
7. Inspect for condensation, clumping, discolouration and fines at the base.
8. Check structure by feel: pliable is the target, brittle is over-dried, spongy is under-dried.
9. Reseal with the headspace minimised, and add a humidity buffer only if the material is within the safe band already.
10. File the record — pack date, lot code, readings, photographs — against the lot, not the delivery date.

Ten steps, most of them a few seconds each. The pattern behind them is consistent: suppliers who can produce a complete record tend to have the systems that produce complete records, and suppliers who produce a reassuring summary tend to have the summary.