Use of genetic algorithms with multivariate regression for determination of gelatine in historic papers based on FT-IR and NIR spectral data

Abstract

Quantitative non-destructive analysis of individual constituents of historic rag paper is crucial for its effective preservation. In this work, we examine the potentials of mid-and near-infrared spectroscopy, however, in order to fully utilise the selectivity inherent to spectroscopic multivariate measurements, genetic algorithms were used to select spectral data derived from information-rich FT-IR or UV-VIS-NIR measurements to build multivariate calibration models based on partial least squares regression, relating spectra to gelatine content in paper. A selective but laborious chromatographic method for the quantification of hydroxyproline (HYP) has been developed to provide the reference data on gelatine content. We used 9-fluorenylmethyl chloroformate (FMOC) to derivatise HYP, which was subsequently determined using reverse-phase liquid chromatographic separation and fluorimetric detection. In this process, the sample is consumed, which is why the method can only be used as a reference method.The sampling flexibility afforded by small-size field-portable spectroscopic instrumentation combined

Non-destructive dating of fiber-based gelatin silver prints using near-infrared spectroscopy and multivariate analysis

“…The loss of accuracy is most probably related to the sampling variance. The quality of the model is similar to the quality of other dating models for chromogenic color prints (RMSECV = 5.4 years / 1942-2008 period) [32], and paper produced post-1850 (RMSEP = 9 years / 1850-1990 period) [31], and better than obtained for parchment (RMSECV = 72 years / 1200-1800 period) [35]. While the error of prediction (RMSEP) is acceptable, it is worth examining the uncertainty in the reference dates for the film stills.…”

“…The second component accounts for the variation in the scattering contribution (10000 -7000cm -1 region). The PC1 vs. PC2 scores plot (figure 5) evidences that the scattering is more pronounced for the thinner papers as expected if related to the presence of the baryta layer (previous studies have shown that effective path length in the case of paper is~0.5 mm [31], which is more than the thickest photographic paper in the sample set). The loadings plot for the third component confirms that some of the variation in the spectra is related to the presence of kaolin as already evidenced above.…”

“…A small band is observed at 4620cm -1 (shoulder), not common to the 100% cotton paper. In this region, combination bands of the N-H stretching mode are usually observed [42], suggesting that this band, also observed from the recto side (gelatin image layer) might be related to the presence of gelatin [31,51]. …”

“…It has already been established, in the case of books, that NIR is sensitive to changes in the composition of paper as a result of ageing and different manufacturing processes, and that an indirect correlation between NIR spectra and a paper's age or manufacturing date can be derived [31]. A similar correlation is the main objective of this study.…”

“…In terms of other organic constituents besides cellulose, previous studies have established correlations between NIR spectra and the concentration of rosin [50] and gelatin sizing [31] in historical papers. A small band is observed at 4620cm -1 (shoulder), not common to the 100% cotton paper.…”