In order to enable the characterisation of historical paper, indeed of whole collections, in a non-laboratory environment, the technique must be robust and should enable the user to determine a number of properties simultaneously and without any sample preparation. Spectroscopy is the method of choice and due to the information rich spectra, near infrared (NIR) spectroscopy turned out to be enough information rich and at the same time the recent advances in hardware led to the availability of portable lightweight spectrometers just waiting to be used in heritage research.
However, the spectra alone are of no help. In order to extract the necessary information, one has to compare the spectra with hard analytical data, obtained on a reference collection using the traditional analytical method. Based on this comparison, one can then model the analytical data for unknown papers in a non-destructive manner.
The task seemed daunting at the beginning: we had to analyse more than 1000 papers for 15 properties, which took us more than two years. Armed with all the data, the calculations began: chemometrics is a discipline which enabled the analysis of large multidimensional sets of data. On these subpages, you can find out how this was done.