<abstract xmlns="http://www.w3.org/1999/xhtml"><p>The first part of the paper describes a new method of obtaining reproducible and meaningful headspace profiles of tobacco lamina by using a modified closed loop stripping apparatus. The complex chromatograms are obtained by high-resolution glass capillary gas chromatography. The second part summarizes the results of a chemometric approach to interpret the chromatograms obtained from a series of nine Virginia flue-cured tobaccos from different origins and belonging to different quality groups, each one analysed three times by the method described above. After the elimination of peaks containing redundant information, the resulting data set, consisting of 27 × 17 data points, was analysed to detect natural groupings by using an in-house program (in Basic) for principal component analysis. A subsequent discriminant analysis yielded two discriminant functions capable of separating the nine Virginia tobaccos into three quality groups as defined by a conventional organoleptic analysis carried out by a smoking panel. All the tobaccos could be classified correctly (100 %). A first attempt to classify, by the procedure described above, a group of six Virginia tobaccos whose organoleptic scores were not known, did not yield clearly interpretable results, possibly because the performance of the capillary column used for analysis had slightly deteriorated during the experiment with resultant changes in retention characteristics, which led to wrong identifications of certain peaks.</p></abstract>

The first part of the paper describes a new method of obtaining reproducible and meaningful headspace profiles of tobacco lamina by using a modified closed loop stripping apparatus. The complex chromatograms are obtained by high-resolution glass capillary gas chromatography. The second part summarizes the results of a chemometric approach to interpret the chromatograms obtained from a series of nine Virginia flue-cured tobaccos from different origins and belonging to different quality groups, each one analysed three times by the method described above. After the elimination of peaks containing redundant information, the resulting data set, consisting of 27 × 17 data points, was analysed to detect natural groupings by using an in-house program (in Basic) for principal component analysis. A subsequent discriminant analysis yielded two discriminant functions capable of separating the nine Virginia tobaccos into three quality groups as defined by a conventional organoleptic analysis carried out by a smoking panel. All the tobaccos could be classified correctly (100 %). A first attempt to classify, by the procedure described above, a group of six Virginia tobaccos whose organoleptic scores were not known, did not yield clearly interpretable results, possibly because the performance of the capillary column used for analysis had slightly deteriorated during the experiment with resultant changes in retention characteristics, which led to wrong identifications of certain peaks.

Pattern Recognition of Tobacco Headspace GC Profiles: A Potential New Analytical Tool for the Classification of Raw Tobaccos

Research Division, F. J. Burrus S.A., Boncourt, Switzerland

Contributions to Tobacco & Nicotine Research

This work is licensed under the Creative Commons Attribution-NonCommercial-NoDerivatives 3.0 License.

The first part of the paper describes a new method of obtaining reproducible and meaningful headspace profiles of tobacco lamina by using a modified closed...

Pattern Recognition of Tobacco Headspace GC Profiles: A Potential New Analytical Tool for the Classification of Raw Tobaccos

Published Online: Jan 06, 2015

Page range: 93 - 103

Received: Dec 10, 1986

Accepted: Jan 12, 1899

DOI: https://doi.org/10.2478/cttr-2013-0589

© 1988 Beiträge zur Tabakforschung International

This work is licensed under the Creative Commons Attribution-NonCommercial-NoDerivatives 3.0 License.