Evaluation of Analytical Errors in Cigarette Tar Determination: Sources, Statistical Methods, and Qu

Cigarette tar analysis error evaluation is the systematic assessment of precision, bias, and measurement uncertainty associated with the determination of nicotine-free dry particulate matter (tar) in cigarette smoke. Because tar is a calculated value derived from three separate measurements, its error budget is inherently complex and must be rigorously controlled. This technical article reviews the principal error sources across the ISO smoking and analytical workflow, the statistical frameworks used to quantify them, and the quality-control practices applied by accredited third-party laboratories to ensure reliable and comparable tar results.

Technical Overview

Tar, formally defined as nicotine-free dry particulate matter (NFDPM), is one of the most widely reported smoke yield metrics in the tobacco industry and in regulatory frameworks worldwide. It is not measured directly. Instead, total particulate matter (TPM) is collected on a Cambridge filter pad during machine smoking, after which water and nicotine are quantified in the extracted condensate. Tar is then obtained by calculation: Tar = TPM − Water − Nicotine.

This indirect approach means that the analytical error of tar is an aggregation of the errors of three independent determinations, each with its own sampling, gravimetric, chromatographic, and calibration uncertainties. Error evaluation therefore addresses the entire measurement chain: cigarette sampling, conditioning, machine smoking parameters, condensate collection and extraction, instrumental analysis, calculation, and reporting.

The evaluation of analytical errors in cigarette tar determination is typically structured around internationally recognized frameworks, including ISO 5725 for precision (repeatability and reproducibility), the ISO/IEC Guide 98-3 (GUM) approach for measurement uncertainty, ISO 16055 for monitor test pieces used in routine quality control, and ISO/IEC 17025 for laboratory competence. Together, these frameworks allow laboratories to demonstrate that reported tar values are fit for purpose, traceable, and comparable across sites and over time.

Testing Samples

Commercial Cigarettes

Finished commercial products are sampled according to ISO 8243 or the corresponding national procedure such as GB/T 18771. Sampling plans are designed to capture lot-to-lot variability, and cartons are selected randomly from defined batches. Sample integrity during transport and storage is documented, since physical damage or moisture changes can bias subsequent smoke yields.

Prototype and Research Products

Development-stage products, blend variants, filter or ventilation modifications, and novel formats are analyzed to support design decisions. These samples often exhibit higher intrinsic physical variability, which must be statistically separated from analytical error during data interpretation.

Monitor Test Pieces and Reference Cigarettes

Certified monitor test pieces conforming to ISO 16055, together with reference cigarettes with assigned tar values, are used as quality-control materials. They provide a stable benchmark against which day-to-day system performance, port-to-port consistency, and long-term drift can be assessed.

Sample Conditioning

Prior to smoking, all samples are conditioned at 22 °C ± 2 °C and 60 % ± 5 % relative humidity for a minimum of 48 hours in accordance with ISO 3402. Cigarettes are marked for butt length and those with individual masses deviating markedly from the sample mean may be excluded, as specified by the analytical standard, to reduce variability unrelated to the measurement system itself.

Testing Items

The error evaluation program covers both the primary smoke yield measurands and the derived statistical quantities used to characterize analytical performance:

• Total particulate matter (TPM), expressed in mg per cigarette
• Water content of the smoke condensate, expressed in mg per cigarette
• Nicotine in the smoke condensate, expressed in mg per cigarette
• Tar (nicotine-free dry particulate matter, NFDPM), expressed in mg per cigarette
• Carbon monoxide in the vapour phase by non-dispersive infrared analysis, where full machine-yield panels are required
• Puff count, puff volume, puff duration, and butt length as process verification data
• Repeatability standard deviation, repeatability limit (r), reproducibility limit (R), estimated bias, and expanded measurement uncertainty (U, coverage factor k = 2)

Testing Methods

Machine Smoking and Condensate Collection

Smoking is performed on linear or rotary smoking machines conforming to ISO 3308, using the standard ISO smoking regimen of 35 mL puff volume, 2.0 seconds puff duration, and one puff every 60 seconds, with the smoking specified butt length and a defined number of cigarettes per Cambridge filter pad. Test cigarettes are allocated to smoking ports in a randomized layout to prevent systematic port effects from contaminating the estimate of analytical error. Ambient cross-flow around the smoking ports, leak integrity, and puff profile verification are checked before each analytical run.

Gravimetric Determination of TPM

Cambridge filter pads are equilibrated in the weighing room, weighed before and after smoking on a balance with 0.1 mg readability, and the mass gain is divided by the number of cigarettes smoked on the channel. Static electricity, buoyancy effects, balance drift, and pad handling are controlled through anti-static measures, draft shielding, and certified reference weights.

Determination of Water and Nicotine

The collected particulate matter is extracted with propan-2-ol containing suitable internal standards. Water is determined by gas chromatography in accordance with ISO 10362-1, and nicotine is determined by gas chromatography with flame ionization detection in accordance with ISO 10315. Multi-point calibrations, internal standard recovery checks, and control chromatograms are used to monitor retention-time stability, response linearity, and detector drift.

Calculation of Tar and Propagation of Effects

Tar per cigarette is calculated by subtracting water and nicotine from TPM. Because the three quantities are measured on the same extract, partial correlation of the nicotine and water results is considered when constructing the uncertainty model. The subtraction structure explains why low-tar products show the largest relative errors: the water and nicotine corrections represent a larger fraction of TPM, so absolute uncertainties of similar magnitude translate into noticeably higher relative uncertainty.

Statistical Evaluation of Errors

Repeatability and reproducibility are estimated through interlaboratory collaborative studies designed according to ISO 5725-2, yielding the repeatability limit r and reproducibility limit R. In routine practice, the absolute difference between duplicate single results obtained under repeatability conditions is expected not to exceed r. Measurement uncertainty is evaluated following the GUM approach: Type A components are derived from replicate smoking and analytical data, while Type B components are taken from calibration certificates, equipment tolerances, and environmental specifications. These are combined into a standard uncertainty and reported as an expanded uncertainty, which for tar typically falls in the range of several percent relative, depending on yield level and machine configuration.

Ongoing Quality Control

• Daily verification of puff volume, puff duration, puff frequency, and leak integrity before analytical runs
• Control charts of monitor test piece tar values to detect drift and out-of-control conditions
• Replicate analysis of test samples at a defined minimum rate, commonly at least ten percent of throughput
• Scheduled calibration and intermediate checks of balances, gas chromatographs, and environmental monitoring systems
• Regular participation in proficiency testing schemes evaluated with z-scores to demonstrate external comparability
• Periodic port-to-port and machine-to-machine comparison studies to quantify and correct systematic differences

Principal Error Sources

• Deviations in puff volume, duration, and frequency, which translate approximately proportionally into TPM changes
• Cigarette physical variability, including mass, circumference, pressure drop, and ventilation, which is controlled through randomization rather than counted as analytical error
• Gravimetric effects such as balance calibration error, static charge on filter pads, and water uptake or loss by pads during handling
• Chromatographic effects, including calibration bias, extraction incompleteness, and solvent volume inaccuracy
• Environmental deviations during conditioning and smoking, notably temperature and relative humidity excursions
• Operator and data-handling errors, mitigated through validated software, second-person review, and traceable records

Applications

Error evaluation of cigarette tar data underpins a wide range of technical and regulatory activities. Manufacturers use documented precision and uncertainty to support tar declarations, to compare candidate designs during product development, and to harmonize results across multiple laboratories and production sites. Regulators and conformity-assessment bodies rely on characterized measurement quality when interpreting tar ceilings, yield classes, and mandatory reporting. Accredited third-party laboratories apply the same statistical framework to demonstrate compliance with ISO/IEC 17025 requirements, to validate methods against the precision criteria stated in ISO 4387 and its national equivalents such as GB/T 19609, and to support participation in interlaboratory comparisons.

Additional applications include long-term trend monitoring of analytical systems, justification of specifications for new smoking and chromatographic equipment, root-cause investigation of anomalous results, and the design of reduced variability in routine testing programs. In research contexts, quantified uncertainty is essential for distinguishing genuine product differences from measurement noise in comparative smoke chemistry studies.

FAQ

What exactly is tar in cigarette analysis?

Tar, or nicotine-free dry particulate matter, is the residue of cigarette smoke particulate collected on a Cambridge filter pad after subtracting the water and nicotine contained in that residue. It is a machine-derived analytical quantity, not a direct measure of human exposure.

Why is the relative error of tar higher than that of TPM?

Because tar is obtained by subtracting water and nicotine from TPM, the absolute errors of three determinations combine into the tar result. For low-yield products, the subtracted fractions are proportionally larger, so the same absolute uncertainty produces a substantially higher relative uncertainty.

What precision level is typically achievable for tar determination?

With well-controlled systems conforming to ISO 3308 and ISO 4387, within-laboratory repeatability standard deviations on the order of a few tenths of a milligram per cigarette are commonly achieved, corresponding to a few percent relative for typical commercial yields. Exact limits depend on tar level, machine type, and the number of cigarettes per pad.

How is analytical error separated from cigarette variability?

Randomized port allocation, replicate machine runs, and nested statistical designs allow the variance components attributable to the measurement system and to the product itself to be estimated separately, typically using analysis of variance techniques consistent with ISO 5725.

Do alternative smoking regimens change the error evaluation approach?

The evaluation principles remain the same, but intensive regimens with larger puff volumes, shorter intervals, and blocked ventilation generate higher absolute TPM and moisture levels, which alter the absolute uncertainty budget and sometimes the dominant error components.

How frequently should the analytical system be verified?

Puff parameter and leak checks are performed before each run, quality-control cigarettes are smoked at a defined minimum frequency such as ten percent of throughput, monitor test pieces are charted continuously, and full interlaboratory comparisons are undertaken at least annually to confirm ongoing reproducibility.