30 
BSH Report No. 56 
lation (marked in green). Again, the relations based on similarity (Fig. 18) are 
reflected In the sample positions In Fig. 19: NFI and IAEA on the bottom left- 
hand corner, BSH and RWS upper right-hand corner. 
Fig. 19: Zoom into the cluster of the samples of the first set. Red circles indicate results of duplicates 
(sample 1 and 10), green triangle frames the pattern of sample 1 to3. Blue arrow indicates higher peak 
intensities in the higher boiling area. 
As this triangular pattern of the samples Is repeating itself among the results, 
the relative difference between the data is somehow constant and might be of 
significance. It is to be remembered that shifts in standard deviation and 
weight average are strongly dependent on the number of Individual homo 
logues of n-alkanes present and their Intensity. As Identical samples were 
used, the differences are mainly caused by the analysis itself. Consequently, in 
such a comparison, shifts in standard deviation and weighted averages indi 
cate Increased or decreased detection of higher n-alkanes (direction of the 
blue arrow In Fig. 19). 
Generally, for the assessment of the analysis of paraffin-wax samples within 
one laboratory (meaning analysis run by the same analytical Inventory), the 
above-mentioned method presents a solid way to spot similarities and dissimi 
larities. Once a Euclidean distance (in terms of standard deviation and 
weighted average) or threshold value for the certainty of discrepancy is estab 
lished (comparable to the 5% and 14% for RSD and CD, respectively), a 
match or no match could be concluded. 
The inter-laboratory comparison conducted in the presented study has the 
additional benefit of monitoring differences of analytical performance as well.