Inside White Paper No. 5: 85.5% of the Kava Reaches the Shell — and 90% of It Is Riding on Particles
aluball extraction kava science research white paper
1 min read

Inside White Paper No. 5: 85.5% of the Kava Reaches the Shell — and 90% of It Is Riding on Particles

Kava has been prepared the same way for three thousand years, and until now data is scarce in terms of measuring what that preparation actually accomplishes. American Kava Association Technical White Paper No. 5 does exactly that: It follows the kavalactones out of the root, into the bowl and accounts for every milligram on a closed mass balance from the spent material.

Root of Happiness was glad to be part of the work, donating the kava material and funding the independent laboratory analysis. Tyler Blythe designed the study, prepared every beverage and performed the mass-balance analysis, with co-author Jeffrey Bowman of Nakamal at Home. What follows is a plain-language walkthrough of the paper, its numbers and what they mean for anyone who makes kava at home or sells it for a living.

American Kava Association · Technical White Paper No. 5

Standardized preparation and extraction efficiency of traditional aqueous kava

A four-variety mass-balance study measured exactly where the kavalactones go when kava is prepared in an Alu Bottle Shaker. A single 60-second cycle moved 77.8% of the root's kavalactones into the cup and after a second wash it went up to 85.5%. 90% of what arrived there was carried on suspended particles, not dissolved in the water. The first wash extracted an average of 100mg of kavalactones into a 125ml (4oz) shell which translates to 142mg in a 6oz Alu Shaker shell.

Tyler Blythe (Root of Happiness) designed the study, prepared every beverage and carried out the mass-balance analysis, with co-author Jeffrey Bowman (Nakamal at Home). Root of Happiness was glad to take part by donating the kava material and funding the independent laboratory work.

  • Independent lab · HPLC-UV
  • 4 varieties · 11 beverages
  • Closed mass balance
  • 15 analytical reports
Published by
American Kava Association · Las Vegas, NV
Paper
Technical White Paper No. 5, Version 1.0 — August 2026
Authors
Tyler Blythe (Root of Happiness) · Jeffrey Bowman (Nakamal at Home)
Analytical laboratory
Flora Research Laboratories, LLC — Grants Pass, Oregon (jobs J25-0930-H, J25-0930-I)
Study period
Harvest September 2025 · analysis October 2025
Preparation system
Alu Ball® 70 µm strainer in an Alu Shaker® Pro bottle
Varieties
Kali-ISA · Kali-Rogu · Kali-Hiwa (American-grown) · PNG Koniak
Material donated by
Root of Happiness, Kali-Kava and the American Kava Association
Shop the AluBall ProJump to the key findings
Alu Shaker Pro bottles loaded with 10 g of milled kava beside the donated Root of Happiness Kali-Rogu, Kali-ISA and Kali-Hiwa bags, the Papua New Guinea Koniak bag and a gram scale

The full study rig: two 70 µm Alu Ball® strainers, an Alu Shaker® Pro bottle pair, a calibrated gram scale and the four donated root lots. Every beverage in the paper was made this way.

The question

How much of the kava actually makes it into the bowl?

Every kava drinker performs an extraction. Almost nobody knows its yield. Published estimates of how much kavalactone a water preparation recovers have ranged from single digits to above eighty percent which is a spread wide enough that no producer could honestly put a dose on a label.

The reason for that spread turns out to be a measurement artifact rather than a disagreement about kava. Studies that filtered or centrifuged the beverage before analysis reported very low recoveries. Studies that analysed the beverage as served, cloudy, unfiltered, the way a person drinks it, reported very high ones. Both were measuring correctly. They were measuring different things.

This paper closes that gap by refusing to discard anything. Root powder, first beverage, second beverage and the spent root left behind in the strainer were all assayed, so the kavalactones could be accounted for on every side of the ledger. On top of that, each beverage was assayed twice — once whole and once after filtration — which separates what is dissolved in the water from what is riding on suspended plant particles.

Why the Alu Bottle Shaker was chosen

Hand-kneading is skilled, variable work: grip pressure, cloth weave and wrist rhythm all change the result, and none of them are recordable numbers. A sealed shaker bottle with a fixed-mesh strainer replaces all of that with parameters you can write down — 70 µm pore size, 10 g of root, 900 mL of water, 38 °C, 60 seconds. Fixing the method is what makes the yield measurable at all.

Traditional aqueous kava is a suspension, not a solution and with the Alu Shaker, 85% of the Kavalactones in the root make it into that suspension.

The paper's central conclusion

Headline results

Four numbers that change how kava should be measured

Eleven beverages were prepared across four varieties and three protocols — PNG Koniak was run at pH 8 only. These are the means; the per-variety detail follows below.

  • 77.8%of root kavalactones reach the cup in one 60-second cycle

    Range across the four varieties: 71.3% to 85.7%. One minute of shaking, once.

  • 89.6%of beverage kavalactone is carried on suspended particles

    Only 10.4% is truly dissolved. Filter the drink and you throw away roughly nine tenths of its activity.

  • 85.5%cumulative recovery after a second extraction

    A second pass adds 7.7 percentage points of yield — but doubles the volume, so per-serving strength falls from 100 mg to 55 mg.

  • 8.1%of the root's kavalactones end up dissolved in water

    This is the number a filtered assay would report. It is not what the drinker receives.

Extraction efficiency by variety

Table 1 — Kavalactone recovery at pH 8 (900 mL per extraction)

Percentage of the kavalactones present in the starting root that were recovered in the beverage, calculated on a closed mass balance against the spent root.

Variety1st extraction1st + 2ndGained by 2nd
Kali-ISA74.2%77.7%+3.5
PNG Koniak71.3%80.7%+9.4
Kali-Rogu85.7%94.4%+8.7
Kali-Hiwa80.1%89.1%+9.0
Mean77.8%85.5%+7.7

All values normalised to 3.0% moisture. Single preparation per variety per condition.

Kali-Rogu was the standout: 85.7% in a single minute, and 94.4% once re-extracted. Even the weakest performer, the Papua New Guinea Koniak, released 71.3% on the first pass. There is no variety in this set for which a single shake was a poor extraction.

Two Alu Shaker bottles holding opaque tan kava beverages labeled Kali-ISA and PNG Koniak immediately after a single 60-second extraction

Kali-ISA and PNG Koniak beverages straight out of a single 60-second cycle. The opacity is the point: that cloudiness is suspended root particulate, and it is carrying roughly 90% of the kavalactones in the bottle.

The central finding

Nine tenths of the strength is riding on particles

Each beverage was assayed whole and again after filtration. The difference between the two is the particle-associated fraction — kavalactone that is physically present in the drink but not dissolved in the water.

Table 3 — Partitioning after a single extraction (900 mL)

Particle-bound versus dissolved kavalactone (KL) as a share of total beverage kavalactone, with the delivered dose per 125 mL serving.

VarietyParticle-bound KLDissolved KLKL per 125 mL servingFlavokavain per serving
Kali-ISA90.3%9.7%147 mg2.6 mg
PNG Koniak89.5%10.5%79 mg2.5 mg
Kali-Rogu87.6%12.4%84 mg3.4 mg
Kali-Hiwa91.2%8.8%90 mg0.9 mg
Mean89.6%10.4%100 mg2.3 mg

The consistency is the striking part. The four roots differed twofold in kavalactone content, yet every one of them produced a beverage that was 87.6% to 91.2% particulate. That uniformity says the partition is governed by physical chemistry — kavalactone solubility in water — and not by variety, potency or handling.

Table 4 — Partitioning after two extractions (1800 mL combined)

VarietyParticle-bound KLDissolved KLKL per 125 mL servingFlavokavain per serving
Kali-ISA89.6%10.4%77 mg1.3 mg
PNG Koniak85.0%15.0%45 mg1.3 mg
Kali-Rogu83.7%16.3%46 mg1.8 mg
Kali-Hiwa85.3%14.7%50 mg0.5 mg
Mean85.9%14.1%55 mg1.2 mg

The dissolved share rises with the second pass simply because more water is present to dissolve into — but the drink is still overwhelmingly a suspension.

Table 5 — Dissolved kavalactone as a share of the starting root

This is the reconciliation. An assay that reads only filtered liquid lands in the 7–15% band — precisely the range of the older low-recovery literature.

Variety1st extraction1st + 2nd
Kali-ISA7.2%8.1%
PNG Koniak7.5%12.1%
Kali-Rogu10.7%15.4%
Kali-Hiwa7.1%13.1%
Mean8.1%12.2%

The reconciliation, in one sentence

Based on the kavalactone solubility numbers in this study, a method that assays only filtered liquid would report between 8.8% and 16.3% of what the consumer actually receives or roughly one tenth under a single extraction, and nearer one seventh after a second. The historic disagreement about kava extraction efficiency was never about kava; it was about whether the analyst kept the cloudiness or filtered the aqueous suspension.

A solubility ceiling, not a process limit

To test whether better technique could raise the dissolved fraction, the authors summed the published 21 °C solubilities of the six major kavalactones: 13.8 mg per 100 mL, or about 124 mg in a 900 mL preparation. Every variety in the study landed at 46% to 83% of that theoretical ceiling. The dissolved fraction is bounded by physics, not by effort. Shaking harder, longer or hotter cannot meaningfully change it.

That bound holds for total kavalactone rather than for every compound taken one at a time. Measured against the same compiled figures, dissolved yangonin exceeds its tabulated 0.3 mg per 100 mL ceiling in all four varieties, and four of the six kavalactones exceed theirs in Kali-ISA. Either very fine particulate is being counted as dissolved, or the compiled single-compound solubilities — which trace back to an unpublished 1997 conference presentation — are too low. The paper reports the discrepancy rather than resolving it.

The exact device used in the study

Every beverage in White Paper No. 5 was made in an Alu Shaker® Pro bottle with a 70 µm Alu Ball® strainer — 10 g of milled root, 900 mL of 38 °C water, one 60-second cycle. That is the whole method, and it moved a mean 77.8% of the root's kavalactones into the cup. This is a viable, repeatable, measurable way to prepare traditional kava, and it is the one piece of equipment that makes those numbers reproducible in a home kitchen.

Method

How the study prepared every beverage

These are the fixed parameters from Section 3 of the paper. Following them at home reproduces the study's conditions closely enough to expect similar behaviour.

  1. Mill the root for 30 seconds

    Grind fineness sets how much surface area the water reaches and how much fine particulate can pass the mesh. In this study it was standardised at a 30-second mill — and it is one of the two biggest levers on final strength.

  2. Weigh 10 g of powder into the strainer

    A 70 µm Alu Ball® strainer. Pore size is the other big lever: a coarser mesh lets more particulate through and yields a stronger drink, a finer one holds it back.

  3. Add 900 mL of water at 38 °C

    Reverse-osmosis water was used to remove mineral variability. 38 °C is warm to the touch, not hot.

  4. Shake for 60 seconds

    One cycle. Not five minutes, not until your arms hurt — sixty seconds was enough to move a mean 77.8% of the available kavalactones.

  5. Stir, then pour

    Because the strength is on the particles, the drink stratifies as it stands. A bowl poured off the top of a settled batch is weaker than one poured from the bottom. Stirring before every pour is not fussiness; it is dose control.

  6. Decide whether to re-extract

    A second pass on the same root recovers another 7.7 points of the total, but it arrives in a second 900 mL. Choose it when you want more volume, not when you want a stronger bowl.

The two levers that actually matter

Strainer pore size and grind fineness govern beverage potency more than temperature, time or pH do. If a preparation feels weak, change the mesh or the grind before you change anything else.

A tested folk remedy

Acidifying the water did not help

A common claim holds that acidic water pulls more kavalactone from the root. It was tested directly. Extraction at pH 5 averaged 85.1% against 87.0% at pH 8 — no improvement, and the per-variety response was inconsistent (Kali-ISA +3.4 points, Kali-Hiwa unchanged, Kali-Rogu −9.1 points). Adjusting water pH is not a reliable way to strengthen a preparation.

Fingerprint fidelity

The chemotype survives the shake

Six of the eleven beverages reproduced their starting root's kavalactone rank order exactly. The five that differed represent only two distinct transpositions — one adjacent pair in Kali-Rogu, one in Kali-Hiwa — and in each case the two compounds are separated in the raw material by less than 2%, which is finer than the method can resolve. Two compounds that close cannot be reliably ranked against one another, so their order in the beverage is not established either way and no selective extraction should be read into it. Practically: a shaker-bottle preparation delivers the variety's characteristic profile, not a distorted version of it. The chemotype you buy is the chemotype you drink.

Two patterns in the data run in a consistent direction without being established by it, and the paper reports them on exactly that footing: desmethoxyyangonin was the most retained of the six kavalactones in the first-extraction marc of all four varieties, and both observed transpositions demoted the less water-soluble member of the affected pair. Both are directions for replication to test, not findings.

Table 6 — Chemotypes and starting-material flavokavain load

All four starting materials fell under the 0.4% w/w flavokavain limit. Kali-Rogu and Kali-Hiwa carry noble-range, kavain-led codes. Kali-ISA and PNG Koniak carry Isa-type 254 and 256 codes, which occur in both Isa and tudei material; but because both lots sit well below the flavokavain limit they resolve to beverage-grade Isa rather than tudei. Beverage grade is not the same as a Codex noble classification, and the paper makes no noble claim for them. You can read about the Beverage-Grade classification HERE

VarietyChemotype codeTotal flavokavains in root (% w/w)Under the 0.4% limit
Kali-ISA2546310.253%Yes
PNG Koniak2564310.260%Yes
Kali-Rogu4235160.293%Yes
Kali-Hiwa4632510.102%Yes

Variety selection, not preparation technique, is the effective lever on flavokavain load. More on the distinction in our guide to noble versus tudei kava.

Implications

What this changes, and for whom

For anyone who makes kava

Stir before you pour, every time. Choose grind and mesh deliberately. Treat a second extraction as a volume decision rather than a strength decision. And stop chasing marginal tricks — hotter water, acidic water, longer shaking — because the measured gains are small or absent next to the two levers that matter.

For producers and vendors

A kavalactone figure on a label is a claim about particle transfer, and it is only meaningful alongside the preparation parameters that produced it. Root-to-water ratio, grind, strainer pore size, water temperature and number of extractions all belong next to the number. Without them, the figure is not reproducible by the customer.

For laboratories and regulators

An assay of filtered liquid captures only 8.8% to 16.3% of the delivered kavalactone — roughly one tenth under a single extraction, nearer one seventh after a second — and should not be used to characterise a traditional beverage. A specification that governs only raw material is incomplete, because two preparations from identical root can differ severalfold in delivered dose depending on mesh and grind. Consumer exposure is determined at the preparation step, and the analytical method has to look at the drink as served.

For the shaker bottle itself

This is the practical vindication. A sealed bottle with a fixed-mesh strainer produced closed mass balances, high and repeatable recoveries, and faithful chemotype transfer across four varieties. It is a legitimate extraction method with parameters you can record and hand to someone else — which is exactly what traditional hand preparation cannot offer.

Stated plainly, and with limits

The paper establishes that the Alu Bottle Shaker is a viable and quantifiable extraction method. It makes no claim of superiority over hand-kneading or over any other device — no head-to-head comparison was run, and none is implied. What it does establish is that this method's output can be measured, described and repeated.

Taste the chemotype difference for yourself

The study's most useful practical finding for a drinker is that the shaker bottle transfers a variety's chemotype faithfully — so which root you choose is what actually shapes the experience. A sample pack is the cheapest way to feel that difference across kavain-led, dihydrokavain-led and methysticin-led profiles, prepared side by side under identical conditions. Pair it with the AluBall above and you can run your own comparison the way the paper did.

Honest limits

What this study does not establish

Stating the boundaries is part of the work. These constraints are disclosed in the paper and are repeated here in full.

Single replicate. One preparation per variety per condition, with marc dry weights determined once. The consistency across four independent varieties supports the central partitioning conclusion, but individual per-variety percentages carry more uncertainty than replicated work would.

Everything rests on one weighing. The dissolved mass is the marc dry weight multiplied by the difference between the raw and marc kavalactone concentrations, so extraction efficiency, particulate share and per-serving potency all scale directly with that single gravimetric measurement. No HPLC result can check it independently.

No comparison against hand preparation. No hand-kneaded control was prepared. The paper explicitly states that no claim of superiority over manual preparation, or over any other device, is made or supported by these data. Get ready for that next!

Not a health claim

This is analytical chemistry. Nothing on this page evaluates safety, efficacy or physiological effect, and nothing here is intended to diagnose, treat, cure or prevent any condition. Kava products are not evaluated by the FDA for such purposes.

Shop the AluBall ProBrowse kava sample packs

Citation and disclosures

Attribution, funding and conflicts of interest

Root of Happiness was glad to be part of this study. We donated the Papua New Guinea “Superior Koniak” material it was built on, batch VHM-020324, and provided financial support for the independent laboratory analysis. Tyler Blythe designed the study, prepared every beverage, ran the extractions and performed the mass-balance analysis.

Suggested citation. Blythe, T.; Bowman, J. Standardized Preparation and Extraction Efficiency of Traditional Aqueous Kava Beverages: A comparative mass-balance analysis of three American and one Papua New Guinea kava variety prepared with the Alu Bottle Shaker® system. American Kava Association Technical White Paper No. 5; American Kava Association: Las Vegas, NV, USA, 2026.

Analytical work. Flora Research Laboratories, LLC, Grants Pass, Oregon (jobs J25-0930-H and J25-0930-I). Kavalactones and flavokavains quantified by HPLC-UV; all values normalised to 3.0% moisture.

Disclosures. Both authors are commercially engaged in the kava trade. Kavafied manufactures and sells the Alu Bottle Shaker® system evaluated here and donated that equipment for the study, so it stands to benefit commercially from favourable findings. T.B. is affiliated with Root of Happiness, which donated the Papua New Guinea “Superior Koniak” material (batch VHM-020324), was the named client on the analytical reports and provided financial support, and in which T.B. holds an ownership interest; T.B. is also affiliated with Kali-Kava, which donated the Kali-Hiwa, Kali-ISA and Kali-Rogu material. J.B. is affiliated with Nakamal at Home, a kava retailer. The American kava materials were donated by the American Kava Association and were grown by T.B. The corresponding author is additionally a contributing author of AOAC SMPR 2018.005, cited in the paper's methods. The paper is published by the American Kava Association, which maintains the research greenhouse the American material was grown in, and both authors are members of that organisation. None of these parties participated in the analysis or generated the data, which came from an independent accredited laboratory working from samples submitted blind to the interpretations offered here.

Trademarks. Alu Bottle Shaker®, Alu Ball® and Alu Shaker® are registered trademarks of Kavafied. Root of Happiness is not affiliated with Kavafied beyond retailing its products.