Advanced Statistics
29
significantly higher, compared to the samples of the other sets. The GC-FID
method used (CEN2012) is not very suitable for these high boiling compounds,
the separation of components above C40 shows an irregular pattern (RT-intervals
are increasing, decreasing height through peak broadening, etc.).
The data from the second set (sample 4, 5 and 6) of both RWS and BSH are
assembled In the upper right-hand corner. The data from the same samples
acquired by IAEA and NFI, however, form a second group at around 4/29.7
(stdev/wa).
Comparison of sample 6, set 2
mean (n=2) of maximum-normalized intensities
r^00CTiQ*r-irMrrj'=l-mU>r'‘00<nO*r-ifMrrj'i3-mU>r'‘0Q<nO*-ir‘irrj*rm
HHrlolNrJrtNrJNrtnolrflrtfOrOrtfOrtrOffliO'i^-^-^^t^-
OOtJUOCJUOCJUOCJUUOtJUOtJUOtJUOtJtJUtJtJ
Fig. 18: Comparison of normalized peak intensities of the second set’s sample #6, from all participants.
Due to this way of compressing the curve’s properties into 2-D data, the range
and relative Intensity of the considered n-alkane species are of great influence.
Fig. 18 shows a comparison of the average means of the intensity of sample 6
from all participants. Especially for higher n-alkanes, larger discrepancies
occur. As mentioned above, these discrepancies might be related to chroma
tographic issues (resolution) or mass discrimination of the targeted sub
stances. A comparatively high similarity between RWS and BSH Is revealed as
both participants had a higher gain of the hba (C35-C45). Even though the NFI
provided data just up to C38, the resulting weighted averages and variances
are still corresponding very well with IAEA’s data. One reason for this is that the
relative Intensities of the data of both laboratories are quite alike from C32 to
C38, leading to the mentioned similarities.
As described above, the values of the duplicates of set 1 (Fig. 19, sample 1
and sample 10, marked with red circles) are in close vicinity to each other, Indi
cating a strong bell-shape similarity. Closer Investigation reveals a typical pat
tern between the samples of this set, which appears in similar form for every
participant’s data: the samples 1,2 and 3) are arranged in a triangular constel-
