16 Two SPC employees went on board the Frixos on 30 April 2002 and 1 May 2002 when samples were taken from the engine room bilge, No.7 cargo tank and the HFO transfer pump in their presence by members of the ship's crew.
Expert Evidence
17 Dr John Francis Leeder, a qualified Analytical Chemist, received two samples for testing from SPC on 22 and 28 May 2002. Additional samples were received on 21 August 2002 and 1 April 2003. According to Dr Leeder the samples were analysed using approved methods of the American Society for Testing and Materials (ASTM) employing analytical techniques of Gas Chromatography for whole oil analysis. Three samples were selected for biomarker analysis by High Resolution Capillary Gas Chromatography/Mass Spectrometry and elemental analysis.
18 Dr Leeder found that the oil on the water from Gore Bay was a degraded bunker fuel oil. The sample from the bilge of the Frixos was found to contain lubricating oil and did not match the oil spill sample. The samples of HFO from the HFO transfer pump were found to have a perfect match with the spill sample. He said that the perfect match was indicated by the triterpane profile of the samples.
19 He summed up his analysis as follows:-
I have reviewed all of the analytical data including the chromatograms from the whole oil analysis and the data from the biomarker analysis. Based on the match between the triterpane biomarkers and the unresolved complex mixture (UCM) I have identified the 'Frixos' as the most probable source of the Gore Bay spill on 30/04/02. The triterpane profiles for the oil spill samples were found to be a perfect match with the Frixos HFO transfer pump sample. This match can be clearly seen in the chromatograms displayed in Figure 3. The probability of this match is 90%. I am therefore of the opinion that the Frixos is the source of the spill due to these similarities observed, to a 90% certainty using the ASTM method but the point of discharge on the vessel has not been sampled from this vessel.
20 Furthermore he found that the samples contained isomers of the unique compound dichlorotoluene ("DCT") and that the presence of these unique isomers in the spill samples and the two different samples from the Frixos, in his opinion, provide a clear link between the spill and the vessel.
21 He expressed his level of certainty in this respect:-
Based on the analysis I performed and the presence of these unique isomers of dichlorotoluene in the Gore Bay spill samples and in two different samples from the ship "Frixos" it is my opinion that there is a perfect match between the Gore Bay spill samples and the samples obtained from the "Frixos" with a 100% certainty.
22 Dr Leeder could not be shaken from the level of 100 per cent certainty in cross examination notwithstanding a conclusion inconsistent with his own expressed by Dr Scott A Stout, a Senior Consulting Geo-Chemist from Massachusetts in the United States of America engaged by the prosecutor and Dr Peter James Christopher Tibbetts a Chartered Chemist retained by the defendants.
23 Dr Tibbetts currently acts as a consultant to the company M-Scan Limited, a world leader in mass spectrometry consulting and analysis for chemistry, life science and environmental studies with offices in the United Kingdom, Switzerland, Germany, United States of America and an agency office in South Korea. In September 2004 the samples analysed by Dr Leeder were divided and sub samples were forwarded to M-Scan Limited for the purpose of conducting tests upon the samples to establish the veracity of the assertions made by Dr Leeder.
24 The results obtained by Dr Tibbetts suggested to him that the oil in the Gore Bay water samples had biodegraded since its analysis by Dr Leeder. He formed the opinion that the period of time between the docking operations of the Frixos and the collection of the water samples would not be sufficient to degrade the HFO. Dr Tibbetts described the samples taken from the HFO transfer pump as resembling a fresh heavy fuel oil that has been blended with a kerosene-based "cutter stock", whereas the engine bilge samples could represent a heavily weathered and biodegraded crude or heavy fuel oil or alternatively could represent a lubricating oil.
25 Dr Tibbetts obtained biomarker fingerprints of the spill oil, the Frixos bilge sample and the Frixos HFO transfer pump sample. The results he obtained suggested that the fingerprints are similar. He concluded as follows:-
In oil-oil correlation studies a positive match is not necessarily 'proof' that samples are related (Peters and Moldowan, 1993). In contrast, any differences in reliable 'biomarker fingerprint' parameters that are statistically significant, results in the two oils being considered a 'Non-match'. In the present case, there are statistically significant differences in a number of triterpane parameters. These differences, therefore, classify both the Frixos bilge and the Frixos HFO samples a 'Non-match' when compared with the oil from the Gore Bay water. On this basis the statistical differences in both of these oils precluded them as the source of the Gore Bay spill.
26 His analysis of whole oil for DCTs showed an increase in the relative concentration of the DCTs in the spill sample in the time since Dr Leeder undertook his tests.
27 Dr Tibbetts summarised his conclusions as follows:-
· The GC traces of the DCM extracts (whole oils) of the oil in the Gore Bay water samples are consistent with those expected for a weathered and biodegraded crude oil or heavy fuel oil.
· GC analysis of the whole oils indicate that the spill samples have been subjected to weathering and biodegradation in the intervening time between the analysis by Leeder Consulting in 2002 and the present analysis by M-Scan. This suggests that the storage conditions were not suitable for the prevention of deterioration of the samples.
· GC traces of the Gore Bay spill samples are dominated by peaks, which have been confirmed by GC-MS analysis, to be dichlorotoluenes (DCTs). These appear to have increased in relative concentration in the spill samples since the analysis by Leeder Consulting in 2002. It is difficult to envisage a mechanism that would result in an increase in the relative concentration of the DCTs in the sample over time, without addition of these components from outside.
· The relative concentration of the DCTs appears to be higher in the spill sample than in the "Frixos" HFO. Due to the higher water-solubility of the DCTs relative to the bulk oil, it is difficult to envisage a mechanism that would result in the increase in the relative concentration of the DCTs in a spilled oil sample.
· From a comparison of the GC traces of the oil in the Gore Bay water and those from the HFO transfer pump on the vessel MV "Frixos", the "Frixos" HFO samples cannot be precluded as a potential source of the contamination in the Gore Bay water. However, the degree of weathering and biodegradation in the two Gore Bay water samples would suggest that the oil had been released into the environment several days, if not weeks prior to sampling. It is understood that the vessel MV "Frixos" had not completed docking operations when the oil spill was observed. This period of time would not be sufficient to degrade fresh HFO to the extent seen in the Gore Bay waters.
· GC-MS analysis of the PAHs in the Gore Bay water sample and the Frixos bilge sample has shown that these parameters have been affected by weathering to such an extent that the normal diagnostic ratios have been rendered unusable.
· GC-MS 'fingerprinting' of the samples indicates that there are statistically significant differences in a number of triterpane diagnostic ratios, which classify both the "Frixos" bilge and the "Frixos" HFO samples as 'Non-matches' when compared with the oil from the Gore Bay water. On this basis, therefore, both of these oils can be precluded as the source of the Gore Bay spill.
28 Dr Stout was retained by the prosecutor to review the chromatographic and tabulated data contained within the reports of Dr Leeder and Dr Tibbetts.
29 He collectively considered data from both the Leeder and Tibbetts reports to develop his interpretation as to the chemical nature of the Gore Bay spill samples and the Frixos candidate source samples. It his opinion that the existing data contained within the Leeder and Tibbetts reports are of sufficient quality and extent to reach a defensible conclusion as to the relation between the samples.
30 Contrary to the opinion of Dr Tibbetts, Dr Stout opined that the degree of weathering observed in the Gore Bay oil sample is not inconsistent with what would be expected in a recently released heavy fuel oil.
31 Dr Stout offered a series of alternative explanations for the higher relative concentrations of DCTs apparent in the M-Scan data than in the Leeder data but in the end emphasised that the most important point is that the DCTs truly were present in the Gore Bay spill samples and the Frixos candidate source oils. He concluded that the variation in DCT abundance in the spill and bilge oils was due to sample preparation method and not to any external source of DCTs to the sample.
32 Dr Stout observed that on average many of the biomarker based diagnostic ratios for the Gore Bay spill oil have values that are in between those of the Frixos HFO and bilge oil. In his opinion this clearly reveals that mixing of the two Frixos oils might have produced the Gore Bay spill oil.
33 Dr Stout agrees with Dr Tibbetts that the two Frixos source candidates were "non-matches" for the Gore Bay spill oil and therefore could be "precluded as the source of the Gore Bay spill". However he has explained that the Nordtest methodology relied upon by Dr Tibbetts acknowledges that when mixtures are suspected "a simple linear mixing algorithm could be developed to explain the oil spill". Therefore according to Dr Stout, as Dr Tibbetts did not consider a mixture, he did not complete the Nordtest analysis. This he says is a severe shortcoming in circumstances where the Gore Bay spill oil had exhibited values in between the two Frixos candidate source oils. In his opinion the recommendation in the Nordtest method is appropriate in this instance.
34 Dr Stout chose to assess the potential for mixing by using two qualitative methods that each made an assumption with respect to biomarker concentrations in the samples. It was not possible to develop a quantitative linear mixing algorithm in the absence of biomarker concentration data or sufficient information to calculate these concentrations. Dr Stout described the reliability of the outcome of the qualitative methods as follows:-
Because each method required that an assumption be made, neither method should be considered as being quantitative in the sense that the "answer" reflects the absolute volumes of each oil in the mixture. However, it is very important to understand that an absolute "answer" (e.g., " X-percent HFO and Y-percent bilge oil ") is not necessary to defensibly establish if a mixture of the Frixos candidate source oils can provide a "positive match" (sic) the Gore Bay spill oil. Both methods simply are different mathematical approaches to testing this possibility - any "answer" that provides a match to the Gore Bay spill oil is proof that a mixture was possible (regardless of the absolute percentages).
35 Moreover both methods by which the M-Scan data were evaluated demonstrated to Dr Stout that mixing of the two Frixos candidate source oils could achieve a "positive match" with the Gore Bay spill oil. According to him both methods achieved a result of approximately 70% HFO and 30% bilge oil. He says that if different assumptions had been made the only impact would be that a different "answer" would be achieved - but a "positive match" by the Nordtest Method would still result. Dr Stout concluded that the linear mixing analyses conducted by him are proof that a mixture of the two Frixos candidate source oils would yield a "positive match" to the Gore Bay spill oil.
36 Despite the lack of direct evidence Dr Stout expressed the view that "it is more likely than not that the presence of DCTs in the three oils samples are the results of a particular solvent used onboard the MV Frixos". His opinion is that the DCTs provide additional support to the Nordtest method's results that the Frixos is the source of the Gore Bay spill.
37 The following extracts from the summary of his opinions reflect the essential findings by Dr Stout:-
· Qualitative inspection of the biomarker distributions reveals features in all three of the oils studied typically associated with oils derived from carbonate source rocks. More importantly, multiple biomarker features in the Gore Bay spill oil on average are in-between the two candidate source oils from the MV Frixos. The apparently intermediate character of the Gore Bay spill oil is evident in the biomarker-based diagnostic ratios reported by Dr. Tibbetts. As per Nordtest methodology, this observation warranted development of a linear mixing algorithm.
· The correlation analysis component within the revised Nordtest method is unable to deal with mixed sources because it only allows comparison of one source sample at a time against the spilled oil. However, the method clearly states that when a mixture of source oils is suspected to be the source of an oil spill "a simple linear mixing algorithm could be developed to explain the oil spill". The mixed oil can then be compared to the spill oil using the method's correlation analysis component.
· Dr Tibbetts' application of the revised Nordtest methodology to the M-Scan data was appropriate, but incomplete. Using the linear regression-correlation analysis component of the method he correctly concluded that individually the Frixos HFO or bilge oil were each "non-matches" to the Gore Bay spill oil. However, Dr Tibbetts failed to consider the potential for a mixture of these two source oils, which given the many intermediate features exhibited by the Gore Bay spill oil, was fairly obvious. Therefore, consideration of a mixture of the two source oils was necessary to complete the Nordtest method. Dr Tibbetts did not do so.
· As recommended in the revised Nordtest method when mixtures are suspected, two different qualitative linear mixing algorithms were developed using the same biomarker-based diagnostic ratios and data utilized by Dr Tibbetts. Both methods demonstrated that a mixture of the Frixos HFO and the bilge oil candidate source oils would produce an oil that, also according to the revised Nordtest method, was a "positive match" with the Gore Bay spill oil.
· Mixed isomers of DCT are commercially used in solvents marketed to the maritime industry. The specific DCT isomer pattern can vary between solvents. The DCT isomer pattern in the Gore Bay spill oil is comparable to the DCT isomer pattern in the Frixos source oils. Thus, it is far more likely than not that the DCTs in the Gore Bay spill and Frixos source samples studied are derived from a particular type of solvent used onboard the MV Frixos. The DCTs' presence in the Gore Bay spill oil and both of the Frixos source oils provide further support to the revised Nordtest method's "positive match" results and further indicates that the Frixos was the source of the Gore Bay spill.
· Collectively, the chemical data available provide overwhelming evidence that the MV Frixos was the source of the Gore Bay spill oil observed April 30, 2002.
38 In cross examination Dr Tibbetts said that his instructions were to compare the oil from the water to see if it matched either the oil from the HFO or the bilge, rather than to answer the question whether the oil in the spill sample was likely to have originated from the Frixos. Mr Sheahan SC, counsel for the prosecutor, submits that this would explain why Dr Tibbetts did not progress to a consideration of a mixture of the two source oils.
39 Following a challenging discussion regarding the proportion of water in the samples analysed at M-Scan, as well as the impact of water on biodegradation and after referring to the development of a blending logarithm by Dr Stout, the following exchange took place between Mr Sheahan and Dr Tibbetts:-
Q Looking at that data, I suggest, and having regard to what was described in the article that we've just been to, what you should have done, if you were attempting to the best of your ability to assist the court in this matter, was to follow through your investigation to see if a blending logarithm could be developed to explain the spill; do you agree?
A . Solely on this data, that's the case. In my mind, I had already excluded the bilge sample as a source of the spill due to the difference in the GC analysis.