The scope and concept of this mini Round Robin (miniRR) 
13 
The scope and concept of this mini Round Robin 
(miniRR) 
In a first approach on evaluating different analytical methods for paraffin Identi 
fication, an intercalibration study (mini Round Robin) with a small group of 
environmental laboratories was set up which Investigated the possibilities of 
environmental forensic analysis for source Identification of paraffin spills using 
and enhancing methods regularly applied In oil spill environmental forensics. 
The overall question of the miniRR was to what extend paraffin wax samples 
can be distinguished from each other or stated as a positive match, using the 
techniques at hand (GC-FID, GC-MS and GC-IRMS). 
For this purpose a collection of samples was sent to the participants. 
As the samples from the R&D-project are all from the German coast, having 
samples from a very distant origin might decrease the probability to be exactly 
the same product. Thanks to David Francois (Ministère de la Défense, France), 
we were able to use additional samples from France, where paraffin wax Is 
beached on the coast lines as well. After GC-FID analysis, the samples were 
compared with the pool of project samples. The collection of samples used for 
the miniRR finally consisted of samples with a) a high probability to be the 
same product, b) a low probability and c) a sample In duplicate. The assess 
ment of similarities was done through cross correlation based on the vector-llke 
representation. In addition to the method presented before (normalised on 
maximum, see above), a second approach was used. This second approach Is 
based on “next-a-nelghbour” normalisation, whereat each peak’s Intensity 
(peak height) was normalised to the following n-alkane In the homologous 
series. This method, suggested by Paul Klenhuls (Rljkswaterstaat-Laboratory, 
RWS), Is used to better compensate for Instrumental differences. 
As a result, there were two different types of data generated from the same 
compilation of samples. Cross-correlation using these two numerical rep 
resentations led to different results, meaning different best matches. 
Finally, three sets of samples were put together, grouping each of the best 
matches from the two different approaches In one set. A fourth set was added, 
consisting just of one sample, the duplicate of sample 1 set 1. This setup 
resulted In a total of ten samples. 
Approach 1 Approach 2 
set origin/date origin/date r origin/date r 
1 Fr, 05-2013 Amrum, Ger/05-2013 0.9977 Ockholm, Ger/12-2013 0.9980 
2 Fr, 2010 Trischen, Ger/05-2013 0.9949 Hooge, Ger/08-2014 0.9609 
3 Buesum,Ger, 08-2013 Buesum, Ger/07-2013 0.9991 HaHaKoog Ger/07-2013 0.9910 
4 duplicate