Present ≠ Perceptible: How Scientists Find the Molecules That Actually Matter
Updated: Oct 20
A fragrance sample can contain dozens—or hundreds—of volatile compounds. An instrument can show that a molecule is present, perhaps even at a large measured signal. That still does not answer the question most people actually care about: does this molecule meaningfully shape what a person smells?
That gap is why odor research rarely stops at a single chromatogram. Scientists can move through several kinds of evidence—chemical profiling, headspace sampling, gas chromatography-olfactometry, detection thresholds, odor-activity values, and recombination or omission experiments. Each method asks a different question. None is a universal score for “importance.”
This is the methods companion to the abundance question. The point is not that analytical chemistry is misleading. The point is that chemical presence, gas-phase availability, human detectability, and contribution to an intact odor are different endpoints.
1. Chemical profiling: what can the instrument see?
Gas chromatography coupled with mass spectrometry, or GC-MS, can separate volatile compounds and support their identification. With an appropriate quantitative method, it can also measure selected compounds. That makes it indispensable for learning what is chemically present under the conditions of the analysis. [1][8]
But a chromatographic peak is not a sensory vote.
Peak size can reflect concentration, detector response, sampling efficiency, calibration choices, and the analytical method itself. Even when abundance is measured well, the result still says how much of a compound the method detected—not how strongly that compound will shape a person’s perception of a complex odor. [1]
The useful question at this rung is: what compounds are analytically detectable, and in what measured amounts under this method?
2. Headspace sampling: what reaches the sampled gas phase?
Before many volatile compounds can be analyzed, researchers have to decide what part of the sample to collect. Headspace methods focus on the gas above a material rather than the bulk material itself.
That matters because the composition of a product and the composition of its sampled headspace are not automatically the same. Volatility, polarity, partitioning, matrix interactions, temperature, equilibration conditions, and the sampling technique can all affect what reaches the gas phase and what the analytical system recovers. [1][2]
This is especially important for fragranced materials. Bothe, Dettmer and Engewald used headspace solid-phase microextraction in work that directly included scented candle wax, while also showing why the sampling system and matrix cannot be treated as neutral. [2]
Headspace analysis therefore answers a narrower question than “what does the product smell like?” It tells us what entered a sampled gas phase under defined analytical conditions. Heated headspace, SPME recovery, and instrument response are not substitutes for a person smelling a candle in a room.
3. GC-olfactometry: which separated fractions can a person smell?
Gas chromatography-olfactometry, or GC-O, adds a human observer to chromatographic separation. As compounds or fractions leave the chromatograph, an assessor smells them at a sniff port and records odor events. [3][4]
That can reveal something chemical detection alone cannot: a small analytical signal may still correspond to a fraction that a person can smell. In a study of scented consumer products that included various scented candles, Bartsch, Uhde and Salthammer combined GC-MS with GC-O and found odor-active trace compounds that chemical abundance alone would not have prioritized. [8]
But the sniff port changes the problem.
GC-O presents separated eluates one after another. The assessor is not smelling the original intact mixture in the same form, and certainly not experiencing whole-room candle throw. A sniff-port event establishes that an odor-active fraction was detected under that GC-O setup. It does not, by itself, prove that the compound dominates the finished fragrance, dominates across people, or drives room-level performance. [3][4][8]
GC-O results also depend on the method, sample preparation, assessor, repeatability, and analytical conditions. Different GC-O approaches answer somewhat different questions, so “GC-O detected it” should be read as method-specific evidence, not a universal verdict. [3][4]
4. Thresholds and OAV: which candidates deserve attention?
A detection threshold is the concentration or exposure level at which an odor becomes detectable under a defined sensory procedure. Thresholds are useful because molecules can differ dramatically in detectability.
They are not universal constants detached from context.
Perry and Hayes showed that threshold estimates for the same odorants changed substantially across different matrices. Ferreira and colleagues also showed assessor variation and method-dependent uncertainty in threshold-related dilution analysis. [5][6]
Odor-activity value, or OAV, tries to combine concentration with detectability:
OAV = measured concentration ÷ threshold estimate
A high OAV can help prioritize candidates whose measured concentration is large relative to a threshold estimate. That is useful screening logic. It is not proof that the highest OAV compound is the dominant perceptual cause of a complex odor. [7]
The threshold may have been measured in a different matrix, with a different delivery method, under different exposure conditions, or with a different panel. And once compounds are mixed, suppression, enhancement, masking, and other mixture interactions can change what the intact odor becomes.
So OAV is best read as “worth investigating,” not “case closed.”
5. Recombination and omission: does the candidate change a defined odor model?
Recombination and omission experiments move closer to contribution.
In a recombination experiment, researchers build an aroma model from selected candidate odorants. They can then compare that model with the target odor. In an omission experiment, one candidate—or sometimes a group—is removed to see whether the model changes in a meaningful sensory way. [7]
That gives stronger evidence than simply observing a large peak or a high OAV because the candidate is being tested inside a defined odor model.
Even here, the conclusion has boundaries.
The model contains selected compounds at selected concentrations under selected conditions. A successful recombination does not prove that every real product, every matrix, every person, or every room will behave the same way. An omission effect supports contribution within that defined system; it is not a universal product-performance certificate. [7]
What no single rung can prove
Chemical profiling, headspace analysis, GC-O, thresholds, OAV, and omission testing can build a much stronger explanation together than any one method can alone. They still should not be collapsed into claims they were not designed to support.
A GC-MS peak does not by itself rank sensory importance.
A headspace measurement does not by itself measure room throw.
A sniff-port detection does not by itself prove dominance in the intact mixture.
A threshold from one matrix does not automatically transfer to another.
A high OAV does not automatically prove causal importance.
A recombination or omission result is strongest inside the model that was actually tested.
A better way to read an odor study
When a study says a molecule is “important,” ask five questions before accepting the word at face value:
What sample or matrix was tested?
What phase was measured: bulk material, sampled headspace, separated GC effluent, or an intact mixture?
Was the endpoint chemical abundance, human detection, perceived intensity, odor quality, or model contribution?
Were compounds evaluated separately or in the intact mixture?
Was contribution directly tested through recombination or omission, or inferred from concentration, threshold, or sniff-port detection?
Those questions turn a methods section into a map of what the evidence can actually support.
For candles, that distinction matters. Analytical release data can tell us a great deal about chemistry, but room-level experience also depends on how a fragranced product releases material, how the mixture behaves, the environment, and the observer. For practical context on that use-level distinction, see Cold Throw vs Hot Throw (Why They Don’t Match) and What “Hot Throw” Really Means (And What It Doesn’t).
The most defensible conclusion is also the least dramatic: scientists do not find “the molecule that matters” with one master instrument or one magic number. They narrow the question, combine methods, and test contribution at progressively more relevant levels—while keeping each method’s limits attached to the result.
Comments