This paper reports on the assessment of lignin and holocellulose by means of ATR-FTIR analysis and multivariate PLS regression. The analysis was conducted on 59 samples coming from different excavations where wood had been preserved in waterlogged conditions. A range of results from different wood species (Alnus sp.p., Cupressus sempervirens, Larix decidua, Picea abies, Pinus sp.p., Quercus sp.p., Ulmus sp.p.), states of preservation, waterlogged environments, and burial times are presented. A calibration model was selected after comparing different reference data (samples extracted and not-extracted, and ash-rich and ash-free bases of calculation for the calibration values), and two different post-acquisition spectroscopic manipulations (both in terms of normalisation procedures and of spectral ranges used for the calibration). Results showed that the best models were different depending on which considered component (lignin or holocellulose) was measured and to which data set (softwood or hardwood) the samples belonged. It is shown that the predictive ability of the models is affected by high ash content (too contaminated samples had to be excluded in order to attain good results, because of excessive overlapping of bands related to the inorganic fraction) but not by the preliminary extraction of sample. Furthermore, the stability of best models is also demonstrated and a procedure of external validation carried out on an external set of samples confirmed the general validity of the identified models.
Keywords: Archaeological wood; Holocellulose; Infrared spectroscopy; Lignin; Multivariate analysis; Partial least square (PLS) regression; Wet wood chemistry.
Copyright © 2014 Elsevier B.V. All rights reserved.