Content
Citation(s) & Footnote(s)
Book(s)Whiskey Knowledge Databases › Literature Notes
Complete analytical Literature Note produced from a full-source review. Claims below are bounded by the recorded evidence and limitations.
Scope and core summary
Complete synthesis of a 12-page study profiling twenty-five whiskies at multiple dilution levels and testing a six-whisky subset with a trained panel.
Author argument
The authors find nonlinear chemical and sensory changes with added water; beyond roughly 20% water in their design, aroma differentiation among whiskies diminished, and style- and sample-specific responses remained important.
Researcher synthesis
Water is an experimental and serving variable, not a universal enhancer. Native-proof and normalized-proof sessions should be separated, additions measured, and conclusions tied to the specific whiskey, dilution range, rest time, vessel, and sensory question.
Evidence assessment
Strong primary study integrating untargeted HS-SPME–GC-MS, descriptive analysis, and multivariate statistics across American and Scotch samples.
Limitations and open questions
Limitations: Headspace extraction approximates but does not duplicate human olfaction; the trained-panel subset was small; the study measures aroma more directly than taste, mouthfeel, or preference; the 20% point is not a universal threshold.
Open questions: How do smaller consumer-scale water additions affect target-fit judgments in high-proof bourbon blends?
Connected records
- Contributors: 5
- Verified Excerpts: 3
- Citations: 1
- Zettels: 2
Completion record
Full-source pass completed: 12/12 local PDF sheets and 6,091 extracted words reviewed, including all dilution treatments, sensory and chemical results, conclusions, and references. Evidence locators retained at local sheets 1, 2, and 10. No completion hold remains.
September 27 Full Publication-Package Audit
The complete 12-page main paper and revised 29-page supplement have now been read, including every table, all 32 references, methods, credits, and the eleven supplementary correlograms. Main Figures 1–5 were visually examined on pages 5–9; supplemental statistical tables were checked visually on pages 10–14 and every correlogram on pages 19–29. This extends the earlier main-PDF-only review. Raw chromatograms, participant data and R scripts are available by author request, not supplied with this publication package; no independent computational reproduction or author contact occurred.
What the design supports
Twenty-five purchased whiskies were chemically profiled at 100, 90, 80, 70, 60 and 50 percent whisky in the prepared mixture. The sensory subset was six 43% ABV whiskies: three Bourbons, two single malt Scotches and one blended Scotch. Those were examined at 100, 80, 60 and 40 percent whisky. These percentages describe the fraction of whisky, not alcohol strength. Nominal final ABVs in that subset are approximately 43, 34.4, 25.8 and 17.2 percent; do not confuse 80 percent whisky with 80% ABV.
Twenty-five consumers completed the preliminary CATA exercise; twenty trained participants completed descriptive analysis after two training sessions, with duplicate assessments across four sessions. Randomized presentation, coded black covered Glencairn glasses, reference standards, and scheduled breaks reduce several practical biases. This was aroma evaluation, not a liking test or a direct test of identifying brands correctly.
HS-SPME–GC-MS used triplicate samples, 30°C equilibration/extraction conditions, 40-minute fiber exposure and tentative NIST library assignments. Those triplicates are analytical replication, not independent production batches. The instrument's heated, agitated, sealed-vial exposure differs from a person's brief sniff over a tasting glass. Ethanol competition on the fiber may affect response; the authors' analogy to olfactory masking is suggestive, not proof that both processes are equivalent.
Findings and their limits
Main Figures 1–2 show dilution-dependent changes in the chemical profiles. Their first two axes represent 24.7% and 51.8% of total variation, respectively. Some dilution-group ellipses overlap at greater water additions. A two-dimensional ellipse comparison does not establish universal indistinguishability, sensory equivalence or a threshold applying to every individual whiskey.
Figures 3–5 connect measured volatile patterns with aroma ratings. Several acetate esters correlate with pome fruit, and phenol-related patterns correlate with smoke, rubber and bacon descriptors. Correlation can also reflect compounds that travel together within a whisky style; it does not establish that each correlated molecule causes the named aroma. No omission/recombination experiment or independently validated consumer-prediction model is reported.
The authors interpret dilution beyond approximately 80:20 whisky:water as reducing distinctiveness and call this deleterious. The Academy should present the observed study-specific convergence while separating it from enjoyment or quality. A drinker may prefer a less distinctive aroma. Neither an optimum dose nor a universal ban on more water follows.
For measured preparation, 80:20 means four parts whisky to one part water in the nominal mixture. Adding 20% of the starting whisky volume as water instead produces about 83.3% whisky. The volume-percentage denominator matters; precise laboratory preparation should account for final volume rather than assume perfect volume additivity.
What the full supplement reveals
Supplement p. 2 describes aroma references using bacon, cedar chips, polenta, malt, oak extract, peated barley, fruit purées, rubber mulch, solvent mixtures and vanilla extract. These are olfactory reference recipes, not drink ingredients. Preserve the source's procedure as research history; do not copy industrial mulch or isopropyl references into unsupervised learner tasting kits.
Tables S1–S2 (pp. 3–9) list tentative compounds and grouping choices. These categories are analytical groupings rather than a strict chemical taxonomy: for example, the saturated-ethyl-ester group contains some non-ethyl esters. Compound assignments and attribution to production stages should remain provisional without standards or additional confirmation.
Table S3 (p. 10) gives 25 named products, but lacks a full lot/strength record and does not separately identify the six sensory bottles in a clearly labeled table. Its abbreviations overlap across sample/style roles. Preserve the published identity list; do not silently fill in missing bottle details.
Table S4 (p. 11) shows that cedar (p=.0631), cornmeal/polenta (p=.0605), and malt (p=.168) were selected for descriptive analysis despite not reaching .05 in CATA. Thus the ten selected attributes should not all be described as statistically significant CATA discriminators.
Tables S5–S7 (pp. 12–14) distinguish panel interaction effects from final retained sample effects. Vanilla, solvent/chemical, pome fruit, bacon, rubber and peat smoke retain significance in the reported mixed-effects assessment. Cornmeal/polenta loses significance there (F=1.00 versus critical 1.554), despite its ordinary ANOVA sample p=.00283. Oak has ordinary sample p=.0919; cedar .114 and malt .081591. The main prose says seven retained descriptors but lists six. The PLS discussion includes oak as an association, which is a different claim from a significant DA sample effect.
Table S8 (pp. 15–18) maps 131 compound numbers to the PLS plot. Figures S1–S11 (pp. 19–29) show patterns by class. Acetate esters often track pome fruit, while PAHs and many phenol-related compounds track smoke/rubber/bacon; other ester, alcohol and terpene patterns are heterogeneous. These are associations across this sample set, not universal flavor dictionaries.
Reporting discrepancies to retain
The supplied main PDF names band-aid among the ten DA terms on p. 4, while pp. 5–6 and the revised supplement use bacon. Use bacon for the reported analysis and flag the discrepancy.
Methods and Figure 1 state 91 compounds for the 25-whisky analysis; discussion says 87. Counting the entries in revised Table S1 gives 87 (6 + 10 + 13 + 18 + 18 + 3 + 8 + 11). Report this inconsistency rather than silently treating 91 as verified.
The main paper's claim of seven retained sensory descriptors lists only six; revised Table S7 supports those six. Cedar's directional arrow is not evidence of a significant treatment effect. Pome-fruit attribution also varies between discussion wording and the blended-Scotch association in Figure 5. These reporting defects qualify precision; they do not erase the overall evidence that dilution changes aroma profiles.
Connection to the Molecular Model
The completed Karlsson–Friedman review, including its six-page supplement and 2018 correction, is here: 
Both studies support composition-dependent behavior. The earlier model concerns one molecule's location in a simplified solvent; Ashmore and colleagues add real-whisky instrumental and human aroma measurements. They neither directly validate the modeled guaiacol evaporation step nor establish that dilution must improve flavor. Treat molecular plausibility, observed aroma change, discrimination, and preference as distinct questions.
Proposed Academy Use
Build a sensory exercise around measured dilution, coded samples, consistent glass/temperature/rest, and separate descriptive ratings from liking. Show both starting and final ABV alongside whisky:water ratio. Teach the contrast between a model, an analytical profile and a human response. A methods exercise can compare the ordinary ANOVA and mixed-effects cornmeal results, and distinguish a compelling multivariate plot from an independently validated prediction.
This work is especially useful for teaching why water is a controlled treatment. It is less suitable as a universal serving prescription or a bottle-quality ranking. No public lesson has been changed.
Provenance and Verification
Main file: owner-authorized Movies mirror of the Proton-listed filename; Proton byte identity remains pending. Revised supplement recovered through the Europe PMC open-access supplementary-files service for PMC10048241; publisher version notes record a supplementary upload on March 28, 2023. Original files preserved. Content review covers the published package, not a reanalysis of unavailable raw data. Internal attachments, native Notion read-back and browser presentation are tracked separately.
