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 17-page ICP-MS and mercury-analysis study comparing trace-element patterns in Scotch, Irish, and American whiskies.
Author argument
The authors show that elemental and chemometric data can separate some samples or clusters, while finding no simple strict relation between whisky type and trace-element abundance across the dataset.
Researcher synthesis
This is useful for explaining authentication as multi-signal inference. Minerals may reflect water, grain, equipment, processing, or contamination, but elemental profiles are not a flavor map and should not be used to romanticize limestone or terroir without controls.
Evidence assessment
Primary analytical study with multielement measurement and multivariate analysis.
Limitations and open questions
Limitations: Small category sample, uneven product representation, semi-quantitative data, and potential packaging or process confounds. The observed Irish-sample signals are not category universals.
Open questions: Which validated American-whiskey authentication datasets combine elemental, isotope, volatile, and documentary provenance evidence?
Connected records
- Contributors: 4
- Verified Excerpts: 3
- Citations: 1
- Zettels: 4
Completion record
Full-source pass completed: 17/17 local PDF sheets and 10,174 extracted words reviewed, including method validation, PCA, conclusions, and references. Evidence locators retained at local sheets 1 and 15. No completion hold remains.
September 27, 2026 — independent complete-source audit
All 17 supplied pages, all 22 reference entries, both tables and Figures 1–5 (every panel) were read. Rendered pages 6, 7, 10, 11 and 14 were inspected. No supplementary file is identified in the supplied article. The existing original attachment, canonical Source SRC-396, Literature Note LIT-320 and evidence identities are preserved. This is a complete review of the article, not a replication of its measurements.
What this paper contributes
Pawlaczyk, Gajek, Jozwik and Szynkowska demonstrate an exploratory elemental-screening workflow and its limited ability to distinguish a selected collection. Twenty retail samples comprise 18 Scotch whiskies, one Irish whiskey and one American whiskey. The Scottish set contains 13 blends and five single malts, with ten Speyside, six Highland and two Lowland assignments (Table 1, page 11). Four of the five malts are Speyside products. National origin, type, age and producer are therefore confounded. A one-bottle Irish outlier cannot establish a national fingerprint; three analytical repeats do not create three independent Irish producers.
Brands are deliberately anonymized. Do not infer Jim Beam or Jameson from the codes JB and J. The US sample is called bourbon in the prose but classified “Blended” in Table 1 and all plots. The category validity cannot be checked from the published identity information. Regional assignments of blends also do not identify the provenance of every constituent distillate. Ages run from three to 18 years; the proposed age analysis was inconclusive and dropped, rather than establishing no age effect.
Measurement and quality controls
Each sample underwent three separate mineralizations of 2 mL whisky, acid treatment and dilution to 25 mL. Three mass spectra were collected for each replicate; subsequent analysis used averages (pages 12–14). Twenty-one selected isotope channels and separately quantified total mercury formed 22 variables. The channels include two titanium isotopes, so this is not 22 independently measured elements. Repeated measurements estimate laboratory variability, not bottle-to-bottle or batch-to-batch reproducibility.
ICP-ToF-MS values are counts or semi-quantitative signals, not calibrated concentrations. The paper reports a reagent blank and multielement standard before and after a one-day cycle but no matrix reference material. Matrix recovery, detection limits, full uncertainty, interference correction and long-term stability are not established. Page 15 explicitly acknowledges Ca-48/Ti-48 and Cr-54/Fe-54 isobaric interference. The strong titanium-channel correlation is partly shared elemental information and should not be treated as independent confirmation of geography.
Mercury used direct thermal decomposition, gold amalgamation and atomic absorption with a separate calibration. At the stated 50 µL sample volume, the reported 1.82–4.56 µg/L averages correspond to roughly 0.091–0.228 ng mercury. These masses are far below the lowest nonzero low-range calibration point stated (2 ng), although above the instrument's quoted lower measurement capability. Low-level calibration performance needs scrutiny; instrument capability alone is not sample-matrix validation. The article says all samples were below a 20 µg/L limit attributed to a 2003 Polish regulation. Preserve this as a dated author-reported comparison, not a current compliance or general safety assurance. The rough Pb/Cd/As standard comparison is also explicitly not the study's validated toxic-metal-monitoring objective (pages 5, 13–16).
Results and limits of inference
The Scottish-only comparison retained reported differences for iron and copper: more Fe signal in blends, more Cu signal in single malts. In the full 20-sample set, Cr, Fe, Cu, Zn and Ba differed by the authors' type grouping. Geographic comparisons changed with grouping: pooled Scotland versus two “other” samples gave Ba/Zn/Sn differences; within Scotland the reported differences were Pb and Bi between Highlands and Speyside (pages 6–9). This is an exploratory set of multiple tests with unequal, sometimes singleton groups. Exact p-values, effect-size uncertainty and a comprehensive multiplicity strategy are not supplied. Nonparametric tests do not cure design confounding or automatically isolate median differences when distribution shapes differ.
PCA is a visualization of covariance, not a tested classifier. The authors found no clear Scottish regional separation, even after variable reduction and removal of several outlying products. Single malts occupy a relatively compact area that overlaps blends; no out-of-sample accuracy or counterfeit decision threshold is reported. The interpretation that equipment may affect copper and other metals is plausible but not experimentally isolated here. Water, grain, equipment, casks and contamination were not independently measured. There was no sensory panel: the abstract's organoleptic claims and the “softer” spirit explanation are not demonstrated sensory results.
Internal errors and visual checks
Figure 1's four spectra use different vertical ranges; equal graphic peak height does not imply equal count intensity. Figures 2–3 show medians, 25–75% boxes and min–max whiskers. Page 15 mistakenly calls the lower quartile the second quartile and describes extrema as outliers. Teach the plotted definitions, not that paragraph.
Figure 4 axes explain 34.93% and 21.80% (56.73% total). Figure 5, after dropping Ru/Ag/Hg, explains 39.69% and 24.34% (64.03%). The prose says the percentage did not improve; the plotted percentages actually increase. This increase still does not establish better regional classification, and removing variables changes the denominator of explained variance.
Page 9 says principal components are correlated; standard PCA components are orthogonal/uncorrelated in the fitted data. Page 15's suggestion that a p-value measures how true or population-representative a result is is incorrect. Page 9's “107 Au” conflicts with the Ag channel throughout the methods; conclusion “88 Ti” conflicts with Sr-88 in the channel list. Merck IV versus VI naming also differs between pages 12 and 13. These transcription and statistical-explanation errors must not become Academy teaching content.
Historical and regulatory background is unreliable
Pages 1–3 contain a speculative early-origin narrative, “Andreas Coffey,” a misleading hierarchy of single malt/cask strength/pure malt, universal distillation-count claims and several legal errors. These introductory passages are not suitable authorities for Academy history or production definitions.
For a verified example, the current US whisky standard, §5.143 distinguishes bourbon from straight bourbon: two years is the straight designation's minimum, not a universal bourbon minimum; there is no maximum four-year age for bourbon. The paper also applies bourbon-specific grain requirements too broadly to US whisky. The cited 2009 Scotch Regulations, regulation 3 require production and maturation in Scotland but do not require locally grown cereals or exactly two distillations. They define blended Scotch as containing single malt and single grain, not simply a mixture of malts. Only these bounded legal checks were made; the remaining historical and foreign-regulatory assertions are quarantined for separate authoritative verification.
Proposed Academy applications and linked synthesis
Use this paper in an authentication research companion as a lesson in sampling: display “18 + 1 + 1” before the PCA plot. Ask learners what a single unusual Irish bottle establishes, and what a representative national study would require. Pair a count-intensity plot with an explanation of calibration, matrix controls, interference and unknown identity. Do not use the reported mercury figures as consumer-risk rankings or the elemental patterns as a flavor wheel.
Link the comparison to 
All three existing excerpts were checked. EXT-2035's section is Conclusions across pages 15–16; EXT-2036's clearest regional-separation support is pages 9–10, with single-malt clustering also on page 16. Native read-back is recorded separately from browser rendering and Proton cloud byte verification.
