In ordertodeterminenewdiscriminatingcriteriaforthe
organic substances,multivariateanalyseswerecarriedouton
the referencematerialsasdescribedinparagraph2.3.Theresults
of PCAcalculationusingtheCHspectralregiondatapointsas
variableswitha2cm−1 steparepresented Fig. 4. Itillustratesthe
distribution ofthe37referencematerialsonthePC1andPC2score
plot. ThegluescanformaclusterthankstothePC1(negative
contribution); theoilsareallononesideofthePC1(between
10and20).However,theresinsaremuchdispersedandseemto
be hardlyseparatedwithPC1butquitesortedwithPC2.Mostof
the triterpenicresinshaveanegativecontributiononPC2andthe
diterpenicarefrom0to15onPC2.However,thisisinsufficient to
consider thatthedifferentmaterialsarewellsortedorclustered.
Indeed, somesamplesarequitedispersed:forexamplepineresins
(colophonyandVeniceturpentine)andothersareclusteredbut
among othermaterials;forinstancefrankincensesamplesare
“mixed” with thediterpenicresins.Moreoversomeoftherefer-
ence samplesareprocessed:forexample,twosamplesofcoloph-
onyareinarawformandoneisacolophonyvarnish,sameforthe