Content
Citation(s) & Footnote(s)
Book(s)One Literature Note synthesizes one Source. It should be understandable without reopening the source while remaining traceable to exact Excerpts.
Source argument
Summarize the author's central argument and relevant reasoning in your own words.
Evidence map
Connect the exact Excerpts that support this note. Do not place unlocated quotations only in the page body.
Researcher synthesis
Explain what the source contributes to the project, how it connects to other research, and where you agree, disagree, or remain uncertain.
Assessment
Record evidence quality, source limitations, and open questions. Separate the author's position from your interpretation.
Completion checklist
Chemical Diversity and Complexity of Scotch Whisky — Complete Article and Supplement Review
Kew, W., Goodall, I., Clarke, D. and Uhrín, D. Journal of the American Society for Mass Spectrometry 28 (2017), 200–213; published online 17 October 2016. DOI: 10.1007/s13361-016-1513-y. Canonical SRC-127 / LIT-118. Private Academy research review, 27 September 2026.
Coverage and Provenance
The complete 14-page article, all five tables (every row and note), all seven figures and their panels, acknowledgments, license and all 70 reference entries were read. All text and tables in ESM1 were read, including all 85 sample entries and 15 calibration entries; all 13 supplementary figures were inspected across the 15-page local rendering (the original contents pagination runs to 16). ESM2 and ESM3 were each read visually in full: one cover and 85 sample plots, 86 pages apiece. This covers all 170 sample plots, not selected examples. Axes, color keys, bubble-size patterns and sample labels were inspected; overlapping points were not individually identified as chemical structures.
The original local main PDF is preserved. All three original supplements were recovered through the Europe PMC PMC5174148 supplementary archive. The separate Edinburgh dataset landing record and both small CSVs (85 sample records and 15 calibrants) were read. Its SampleInformation.docx download timed out on the initial attempt. The repository also lists 6.53 GB mzML and 5.33 GB raw/processed archives; these instrument datasets have not been reprocessed or exhaustively examined and are a separately bounded supporting resource. Reading the article's bibliography does not mean reviewing all cited publications. Proton-to-mirror identity remains provisional until cloud bytes can be hashed.
What the Study Establishes
A 12 T FT-ICR instrument exposes a large and structured molecular-formula space in 85 commercial Scotch whiskies: 55 malt (including blended malt) and 30 blends, plus a separate new-make comparison. The sample selection represented 73.15% of 2014 sales volume according to the authors, not that proportion of distilleries, styles or modern production. It contains repeated products across years and is not 85 independent distilleries. No standalone grain-whisky group establishes grain effects. [pp. 200–202; ESM1 Table S1]
The combined set contains 4,271 assigned monoisotopic formulae: 82.2% CHO and 17.8% CHOS, with 407 found in every whisky and 1,201 in more than 75% of samples. These are formula counts, not a count of uniquely identified flavor compounds. In the example sample, 1,902 monoisotopic assignments plus 571 isotope assignments account for 74.4% of 3,325 peaks; the monoisotopic share alone is about 57.2%. The headline 72–88% assignment coverage includes isotopic species. Isotopic fine structure strengthens elemental-formula assignment but does not uniquely resolve structure. The authors' database exercise associates 1,735 of the 1,902 formulae with over 320,000 possible structures. [pp. 203–205; Tables 1–3; Figures 1–2]
The method interrogates negative electrospray ions with constrained C/H/O/S formula rules over approximately m/z 98–1000. Nitrogen-containing and other unmodeled chemistry, neutral species, volatile aroma components and weakly ionizing compounds are not an exhaustive compositional census. Signal-to-noise and intensity thresholds and a minimum class size further shape what is retained. Blanks and selected cross-day repeat measurements are useful controls, but do not constitute replicate independent bottles of every product. [pp. 201–203]
Reading the Visual Evidence
Van Krevelen plots map H/C against O/C; DBE plots map formula-derived unsaturation against carbon number. Colors encode mass or oxygen count and bubble size expresses ion intensity. Broad chemical regions overlap, and neither color nor size supplies aroma intensity. Regular lines can be consistent with transformations such as oxidation or condensation without directly observing a reaction path or proving barrel kinetics. Ion abundance also depends on ionization efficiency; normalization imposes a common total. [pp. 204–207; Figures 2–4]
All 85 paired supplementary maps share interpretable geometry but differ in dominant nodes. Examples include large C30/DBE8 and C14/DBE12 nodes in S14-1941, S14-1942 and S14-2373; fatty-acid-region dominance in S14-1947, S14-1948 and S14-2374; and unusually large carbohydrate-region nodes in S14-2082, S14-2083 and S14-2372. These observations provide selected teaching examples after complete atlas inspection, not an unsupervised brand-identification rule. The paired maps show two views of the same assigned formula set, so apparent corroboration between them is not an independent experiment. [ESM2/3 pp. 48–49, 52–53, 61–62, 76–78]
The unmatched new-make comparator has only 252 monoisotopic and 49 isotope assignments among 1,084 detected peaks. Its 27.8% assignment rate differs sharply from the mature-whisky coverage. Mature-versus-new-make differences are consistent with maturation contributing chemical diversity, but a single unmatched comparator does not isolate oak extraction, chemical transformation, distillery effects or time. [p. 206; Table 4 and Figure 3]
Classification Strength and Its Limits
PCA's first nine components explain 85% of variance; the first two account for roughly half. Blends cluster more tightly than malts in this dataset. That is not evidence of lesser quality, simpler perceived flavor or a universal malt advantage. The malt/blend OPLS model reports R2X 0.64, R2Y 0.83 and Q2 0.61. For cask categories, the 43-sample model reports Q2 0.61 and the narrower bourbon-versus-sherry subset of 27 reports Q2 0.80. Q2 is a predictive variance statistic, not 80% classification accuracy. [pp. 207–210; Figures 4–6]
The sherry group includes seven observations representing only three products sampled across years. The report does not establish an independent blinded external test set or explain product-grouped cross-validation. Repeated product identity may leak across validation partitions. Missing peaks replaced by random noise and unit-variance scaling also warrant sensitivity tests and reproducibility documentation. These constraints make the models promising exploratory evidence rather than ready-to-deploy authentication tests.
The source itself provides counterexamples to overinterpretation: the regional model has Q2 about 0.20; peat discrimination is weak; the 10-versus-12-year comparison in supplementary Figure S11 has Q2 −0.099 despite visible separation in its plot. A visually separated chart is not proof of prediction. Failure in this design does not disprove regional influences or peat chemistry. [pp. 209–210; ESM1 Figures S9–S11]
Three years of one product show relatively consistent formula/intensity patterns (2,208 of 2,826 formulae shared), compared with greater between-product variability. The authors' suggestion that many reproducible compounds could matter sensorially remains a hypothesis: the study conducts no sensory panel, odor activity assessment, recombination or omission experiment. [pp. 208–210]
Audit Findings and Unresolved Contradictions
- The main text describes 24 samples from 2010, nine from 2012 and 52 from 2014. Table S1 and the dataset CSV instead identify 24 S10, seven S12, two S13 and 52 S14 codes. Preserve the discrepancy rather than silently inventing sample dates.
- Peat analysis is reported as n=31 and Q2 0.14 in the main text, while supplementary Figure S10 shows n=36 and Q2 0.094. Both indicate weak prediction, but the exact analysis cannot be reconciled from the supplied report.
- S14-2373 is described as both ten and twelve years old in the main text. Avoid using its age as a verified example.
- Figure 5's c/d caption reverses the bourbon/sherry colors relative to the plotted legend. Read B/S and symbols; do not copy its color key uncritically. Supplementary figures also contain numbering typos (S91/S21).
- Table S1 contains a BP wood code not defined by its supplied legend, and blank peat fields must not automatically become confirmed unpeated labels. The CSV repeats these gaps and supplies no age column.
- Table 5 gives different formulas for the same approximately 455.31608 fragment: the MS3 C29H41O3 entry is inconsistent with the MS2 C29H43O4 assignment and the mass. Some stated water-loss steps also do not differ by 18 Da (437 to 407 and 393 to 379). The supplement repeats the table, so duplication does not independently confirm it. Do not teach the detailed proposed pathway without reconciliation.
- A prose count of eleven assignments among twelve peaks conflicts with formulas provided for all twelve in Table 1. The central van Krevelen convergence visually lies near O/C 0.5; a textual value of 5 is a typographical issue.
The C30H46O7 marker is proposed as an oleanane-type triterpenoid such as bartogenic acid on fragmentation evidence, not conclusively identified with an authentic standard or NMR. Isomers remain possible. CHOS formulae here likewise do not establish the presence or sensory impact of volatile sulfur off-notes. [pp. 210–211; Table 5; Figure 7 and ESM1 S12–S13]
Critical Assessment
This is valuable primary comparative analytical research with exceptional mass resolution, explicit formula-assignment constraints, extensive visual disclosure and accessible supplementary material. Industry collaboration through the Scotch Whisky Research Institute provides sample expertise; BBSRC/SWRI support should remain visible. The article is under CC BY 4.0. Its strongest contribution is demonstrating a rich and organized formula space and plausible cask-related discriminators. Its weakest use would be turning exploratory formula patterns into direct assertions about quality, age, exact molecular identity or flavor causation. American-whiskey applications require separate validation.
Linked Ideas and Proposed Academy Uses
- A chemical fingerprint is not a flavor map. Pair this work with Poisson and Schieberle's bourbon recombination/omission study: one maps thousands of ion formulae; the other experimentally tests selected aroma contributions in one bourbon. Neither substitutes for the other. Proposed sensory-analysis exercise: identify what additional evidence would connect a large plotted bubble to a flavor claim.
- Every analytical method sees a selected portion of the spirit. Connect Kew's constrained negative-ion chemistry with Barnes's extraction-method comparison and Zhao's six-spirit profiling. Detection counts depend on ionization, extraction, thresholds and calibration. Proposed instrument-method worksheet: distinguish a peak, formula, identified structure, quantified concentration and demonstrated sensory contribution.
- Cask classification is a hypothesis to validate. Link the cask patterns to the Academy's wood-maturation reviews. Prior liquid, oak species, fill history, age and distillery may covary. Proposed advanced exercise: design product-grouped validation and an independent test set before making an authenticity claim.
- Chart separation and useful prediction are different. Use ESM1 Figure S11 as a compact case study: ask learners to interpret its negative Q2 before seeing a persuasive chart headline. Preserve aspiration and accessible explanation while keeping the overall impression accurate.
- Replicated figures are not replicated evidence. Teach why the two formula maps and repeated fragmentation table are alternate representations of the same observations.
These are proposed internal Academy applications; no public lesson has been changed. Existing Source, Literature Note, excerpt and Zettel identities are preserved. API read-back and browser rendering are separate checks; browser appearance is not claimed verified.
Existing Evidence and Connected Reviews
![[SRC-127] Abstract (PDF sheet 1)](https://images.spr.so/cdn-cgi/imagedelivery/j42No7y-dcokJuNgXeA0ig/95dc2f1d-5993-42f1-a7c0-c08fe42504f7/whiskey_research/w=1920,quality=90,fit=scale-down)
