64 
5.3.3.3 PCA result comparability, own techniques 
Zencak el al. (2005) could show that deviations between PCA results obtained by 
different detection methods do not exceed the precision of a single methodology. Long 
term precision for spiked samples (five parallels over 4 months) was between 6-13 % 
depending on the tested method. The long term method precision for naturally exposed 
fish livers (five parallels over 4 months) includes additionally homogeneity variations of 
the sampling material. It was for the different techniques at maximum: 
ECNI-LRMS: <12 % (sPCAs), <18 % (mPCAs) 
CH4/CH2CI2-NICI-LRMS: <28 % (sPCAs), <10 % (mPCAs) 
EI-MS/MS: <20 % for s+mPCAs. 
• Main sources of uncertainty and systematic errors are: 
• Homogeneity variations and analyte losses during clean-up. 
• Integration of a signal hump over a retention time range of several minutes is 
highly dependent on base-line definition and not well handled by any integration 
programme. 
• Concentrations of single congener groups are often not too much above the 
quantification limit (defined at a signal-to-noise ratio of 10:1). Noise fluctuations 
increase the measuring uncertainty. This is particularly valid for formula and 
congener group-specific methods. 
Table 12 shows that deviations between different quantification methods are well within 
20 % for spiked samples and around this value for real samples with lower levels. 
CH2CI2/CH4-NICI gives often somewhat lower concentrations due to a complete 
suppression of remaining interferences. This method was mainly applied for the 
determination of formula and congener group pattern in sediments.