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 7-page quantitative aroma study using stable-isotope dilution, odor-activity values, aroma recombination, and omission tests on an American bourbon.
Author argument
Poisson and Schieberle show that a bounded set of odorants can reconstruct a bourbon’s typical aroma and that omission tests are needed to test which compounds or groups materially affect the whole mixture.
Researcher synthesis
This is a cornerstone for explaining bourbon aroma as a mixture system: concentration and threshold identify candidates, but sensory recombination and omission reveal interaction, masking, redundancy, and emergent character. Blending guidance should therefore avoid one-compound/one-flavor determinism.
Evidence assessment
Strong primary flavor-chemistry study combining quantitative and sensory validation.
Limitations and open questions
Limitations: One bourbon and one model matrix cannot represent the entire category; odor-activity values depend on thresholds and matrix; recombination captures aroma rather than full retronasal, taste, texture, and finish.
Open questions: Which later bourbon studies replicate the key-odorant set across mash bills, proofs, and maturation regimes?
Connected records
- Contributors: 2
- Verified Excerpts: 3
- Citations: 1
- Zettels: 1
Completion record
Full-source pass completed: 7/7 local PDF sheets and 6,926 extracted words reviewed, including quantitative tables, recombination, omission experiments, and references. Evidence locators retained at local sheets 1, 5, and 6. No completion hold remains.
Full review audit — 27 September 2026
All seven PDF pages (printed 5820–5826) read in full, including synthesis procedures, every row and footnote of Tables 1–6, Figures 1–3 (both mass-spectrum panels), acknowledgments and references 1–30. This supplied article is complete. The separately cited companion identification/AEDA article (reference 6, pp. 5813–5819) has not been read in this audit and is not silently included in completion.
Study and contribution
Poisson and Schieberle studied a supermarket-purchased Kentucky straight bourbon described by its label as sour mash and aged at least three years in new charred oak. The supplied text does not identify a usable brand name; do not infer one. The main measurements use three aliquots from the same bottle, not three independent brands, barrels or production batches. A separate limited comparison measures 15 odorants in two label-dated batches (1996 and 1998).
This is particularly valuable because it goes beyond chemical detection: stable-isotope dilution quantification, thresholds, recombination and omission tests connect analytical candidates to an actual model aroma. It does not establish a universal recipe for all bourbon or reproduce taste, texture, retronasal perception, finish or liking.
Methods and complete coverage
- PDF 1 / p. 5820: abstract, prior literature and aims. The authors distinguish merely identifying volatiles from determining sensory importance. Prior work found 45 odor-active regions with 42 identities; those prior identification experiments are cited, not reproduced here.
- PDF 2–3 / pp. 5821–5822: all reagent sources and labeled-standard synthesis, calibration, extraction, SAFE distillation, two-dimensional GC/MS and selected-ion procedures read. Figure 1 is synthesis of a labeled analytical standard, not a biochemical pathway occurring in a barrel. Figure 2A/B (PDF 4) compares EI and CI spectra; the +2 mass-unit shift supports the deuterated standard's identity, not a whiskey age marker. Table 1 includes response factors rather than assuming every analyte has identical response.
- Ethanol is measured by density after steam distillation. Most odor thresholds use the intended water/ethanol 6:4 matrix; ethanol's own OAV uses a water threshold, and cis/trans whisky-lactone thresholds come from reference 25. Do not claim all thresholds were newly measured by this panel in the same matrix.
- Ten trained assessors evaluate orthonasal aroma, eight defined attributes, on a seven-step 0–3 intensity scale in three sessions at 21 ± 1 °C. Training anchors include malty, fatty, coconut-like, fruity, flowery, vanilla-like, smoky and phenolic odors. This is trained description, not consumer acceptance.
- The 26-component model uses measured concentrations, nominal 40% ethanol and pH 4.1. A second model adds deodorized nonvolatile whisky residue. Similarity scores are 2.7/3 and 2.8/3 respectively. Figure 3's eight-axis profile and three lines were inspected; no error bars or inferential comparison establish that the 0.1-point increase is significant.
- PDF 4 / p. 5823: all Table 2 concentration/deviation rows and Table 3 two-batch comparison read. Concentrations are micrograms per litre, including ethanol 316,000,000 µg/L = 316 g/L. Analyte concentration alone is not sensory rank.
- PDF 5 / p. 5824: all Table 4 OAV/threshold values and notes, all ten Table 5 omission conditions, and Figure 3 examined.
- PDF 6 / p. 5825: all eleven Table 6 water-versus-ethanolic threshold rows, interpretation, conclusion and acknowledgments read. PDF 6–7 includes all thirty reference entries and publication history.
What the omission tests actually support
Table 5 reports detectable differences when removing the ester group (ethyl esters plus 3-methylbutyl acetate), vanillin (4-hydroxy-3-methoxybenzaldehyde), the cis-whiskylactone/nonalactone pair, cis-whiskylactone alone, or ethanol. Removing gamma-nonalactone alone, beta-damascenone, diacetyl, the two branched aldehydes, or those aldehydes together with 3-methylbutanol did not yield a significant difference under this design.
High OAV is therefore a useful selection criterion, not proof that a component is independently indispensable in the mixture. Conversely, a nonsignificant omission is not proof of zero contribution or universal absence of effect: the panel, concentrations, model, power and task matter. The authors propose masking by fruity esters for damascenone; this is an explanation consistent with the result, not an isolated factorial test proving that mechanism.
Ethanol omission changes both ethanol's sensory signal and the solvent matrix controlling other compounds' release. It cannot separate these mechanisms. Its very large water-based OAV is not directly comparable to ratios for odorants tested in 40% ethanol and is not a linear measure of perceived dominance.
Precision and reporting audit
- Table 2 is headed “32” odorants, but its printed rows contain 31 including ethanol; Table 4 also lists 31 and has 26 above one. Table 1 lists 30 nonethanol analytical entries. The abstract/prose assay counts are not fully reconciled. Retain the directly observable table counts rather than invent a missing analyte.
- The Table 2 narrative says standard deviations were at least 10%, while its rows show 1–10%. This is a visible reporting inconsistency, not an extraction error.
- Methods contain a reversed ethanol/water ratio once, while the abstract, recombination method and discussion identify the intended 40% ethanol matrix. Table 6's column wording also needs care. Record the inconsistency rather than use the reversed wording as a 60% protocol.
- Table 1 labels delta-decalactone while the synthesis, concentration and threshold material refer to gamma-decalactone. Its 4-allyl-2-methoxyphenol row and surrogate-isotope footnote also differ in isotope-label notation. These need clarification before laboratory replication.
- Printed OAVs cannot all be regenerated exactly from rounded published concentrations/thresholds: for example 30/0.2 = 150, while the ethyl (S)-2-methylbutanoate OAV is 138; 11/0.1 = 110, while damascenone is 79. Coarse threshold rounding can contribute; do not silently “correct” the authors' results or pretend exact reproduction.
- Table 3 ratios do not all match the compressed narrative: cis-lactone rises 3880/2490 ≈ 1.56, vanillin 3060/2130 ≈ 1.44, and 2-methylpropanal 417/233 ≈ 1.79. Two batch samples cannot establish general manufacturing consistency or population variability.
- The cis-lactone-only omission is Table 5 condition 4A, despite the narrative calling it 4B; damascenone's omission is in Table 5, not Table 4. Use the actual table identifiers.
- Table 5 provides significance categories, not raw response counts or a complete repeated-trial/randomization analysis. Exact probabilities, power and multiplicity handling cannot be independently reconstructed from this report. No claim of experimental replication is made.
- Excluding OAV-below-one components from the model is a selection strategy, not exhaustive proof that all such compounds have no mixture effects. “Typical American bourbon aroma” in the conclusion should remain bounded to this tested product/model.
Academy uses and linked synthesis
Proposed sensory lesson: move through detection → quantification → matrix-matched threshold → reconstruction → omission, asking what each stage can and cannot show. A table exercise can contrast high-OAV damascenone with a nonsignificant omission, then contrast lower-OAV cis-whiskylactone with a detectable omission. Keep published results, student predictions and new experimental observations distinct.
Use Figure 3 as a reading exercise: a close spider plot is descriptive similarity, not proof of identical complete flavor. Design an original diagram for public teaching and check rights before reproducing the journal's figures. Laboratory standard synthesis is reference material, not a learner mixing recipe.
Connect to 
EXT-1995 requires the ethanol-water-threshold exception; EXT-1996 is supported for this model; EXT-1997 requires masking to remain an interpretation rather than a proven mechanism. Original preserved; complete reading does not imply Proton hash identity, browser rendering or current-product verification.
