This study was conducted to characterize the variability of mainstream PAH emissions data and to investigate the ability to detect differences amongst products. The mainstream yields of a reference cigarette (Kentucky Reference 3R4F) were obtained for a period of approximately one year in order to characterize variability in analysis results across 19 PAH compounds using estimates of repeatability and reproducibility.

For the compounds measured, eight had yields that were below the limit of quantification. Averaged quantitative yields ranged from 324 ng/cig for naphthalene, to less than 0.4 ng/cigarette for 5-methylchrysene and Dibenz(a,h)anthracene. Repeatability ranged from 7.6 % for benzo[a]pyrene to 21.5 % for benzo[c]phenanthrene. The corresponding reproducibility estimates ranged from 43 % to 121 %.

Precision is largely dependent on the value of the number of replicates (n), since the standard error decreases as the number of replicates increases. To illustrate this, B[a]P emissions from the 3R4F, with a relative standard deviation of 8 % (most precise) and Cyclopenta(c,d)pyrene with a relative standard deviation of 21 % (least precise) were chosen. Given 5 replicate observations, the minimum detectable difference between the sample mean and the true population mean for B[a]P would be 7 % and 19 % for Cyclopenta(c,d)pyrene. However, if the sample size was increased to 20, the minimum detectable difference for B[a]P would be 3 % and 9 % for Cyclopenta(c,d)pyrene.

Additionally, potential year-to-year variability of mainstream PAH yields from four Canadian flue-cured cigarette brands, manufactured over a 30-year period, were tested. The year-to-year consistency of the yields per unit ‘tar’ of the four commercial flue-cured brands ranged in yield from 0.3 to 0.6 ng per mg tar. The yields were quite stable over the 30-year time frame. Thus, year-to-year product variation may not be a significant source of variability in mainstream PAH yields.