Siegwerk Australia Pty Ltd (In Liquidation) v Nuplex Industries

Federal Court of Australia (Full Court)|2013-11-20|Before: Rolfe J, Robertson JJ

(2012) 189 LGERA 359; [2012] WASCA 79 State Rail Authority of New South Wales v Earthline (1999) 160 ALR 588 Transfield Constructions Pty Ltd v GIO Australia Holdings Pty Ltd (Supreme Court of New South Wales, Rolfe J, 2 March 1993, unreported) Transfield Constructions Pty Ltd v GIO Australia Holdings Pty Ltd [1996] NSWCA 538; (1997) 9 ANZ Ins Cas 61-336 Visy Packaging Pty Ltd v Siegwerk Australia Pty Ltd (2013) 301 ALR 560; [2013] FCA 231 Water Board v Moustakas (1988) 180 CLR 491 Whisprun Pty Ltd v Dixon (2003) 200 ALR 447 Wilkie v Gordian Runoff Ltd (2005) 221 CLR 522 Windsurf Pty Ltd v HIH Casualty and General Insurance Ltd (1999) 10 ANZ Ins Cas 61-447; [1999] QCA 360 Zurich Australian Insurance Ltd v Regal Pearl Pty Ltd (2007) 14 ANZ Ins Cas 61-715; [2006] NSWCA 328

Date of hearing: 15, 16 August 2013

Division: GENERAL DIVISION

Category: Catchwords

Number of paragraphs: 196

Counsel for Siegwerk Australia Pty Ltd: Mr A Kelly SC, Mr A Schlicht

Solicitor for Siegwerk Australia Pty Ltd: Wotton & Kearney Insurance Lawyers

Counsel for QBE Insurance (Australia) Limited: Mr R Cavanagh SC, Mr A Bowen

Solicitor for QBE Insurance (Australia) Limited: Gillis Delaney Lawyers

Counsel for Nuplex Industries (Aust) Pty Ltd: Mr S Donaldson SC, Ms S Ross

Solicitor for Nuplex Industries (Aust) Pty Ltd: Sparke Helmore

IN THE FEDERAL COURT OF AUSTRALIA VICTORIA DISTRICT REGISTRY GENERAL DIVISION VID 265 of 2013

BETWEEN: SIEGWERK AUSTRALIA PTY LTD (ACN 007 114 338) (IN LIQUIDATION)

AND: NUPLEX INDUSTRIES (AUST) PTY LTD (ACN 000 045 572)

JUDGES: PERRAM, DODDS-STREETON & ROBERTSON JJ DATE OF ORDER: 20 NOVEMBER 2013 WHERE MADE: MELBOURNE

The parties are to bring in short minutes of order or competing written submissions on the scope of the retrial and costs within 14 days. Note: Entry of orders is dealt with in Rule 39.32 of the Federal Court Rules 2011. IN THE FEDERAL COURT OF AUSTRALIA VICTORIA DISTRICT REGISTRY GENERAL DIVISION VID 713 of 2013

BETWEEN: QBE INSURANCE (AUSTRALIA) LIMITED (ACN 003 191 035)

AND: NUPLEX INDUSTRIES (AUST) PTY LIMITED (ACN 000 045 572)

JUDGEs: PERRAM, DODDS-STREETON & ROBERTSON JJ DATE OF ORDER: 20 NOVEMBER 2013 WHERE MADE: MELBOURNE

The parties are to bring in short minutes of order giving effect to the conclusion that QBE's appeal against the declaration and orders on Nuplex's cross-claim should be allowed and dealing with the question of costs within 14 days. Note: Entry of orders is dealt with in Rule 39.32 of the Federal Court Rules 2011.

IN THE FEDERAL COURT OF AUSTRALIA VICTORIA DISTRICT REGISTRY GENERAL DIVISION VID 265 of 2013

BETWEEN: SIEGWERK AUSTRALIA PTY LTD (ACN 007 114 338) (IN LIQUIDATION)

AND: NUPLEX INDUSTRIES (AUST) PTY LTD (ACN 000 045 572)

IN THE FEDERAL COURT OF AUSTRALIA VICTORIA DISTRICT REGISTRY GENERAL DIVISION VID 713 of 2013

BETWEEN: QBE INSURANCE (AUSTRALIA) LIMITED (ACN 003 191 035)

AND: NUPLEX INDUSTRIES (AUST) PTY LIMITED (ACN 000 045 572)

JUDGES: PERRAM, DODDS-STREETON & ROBERTSON JJ DATE: 20 NOVEMBER 2013 PLACE: MELBOURNE

I. Introduction 1 This is an appeal from orders made by Gray J (Visy Packaging Pty Ltd v Siegwerk Australia Pty Ltd (2013) 301 ALR 560; [2013] FCA 231) and concerns cans of tuna in various, including 'zesty', vinaigrettes. In what remains of this complicated litigation the manufacturer of a lacquer (the present appellant, 'Siegwerk') sues the supplier of a resin used as a component in that lacquer (the present respondent, 'Nuplex'). The cans of tuna had ring pull ends which are also known as 'easy open ends'. The cans were manufactured by Visy Packaging Pty Ltd. As part of the process of manufacture it applied the lacquer to the inside surface of the cans. It did so to provide a protective coat between the tuna (and the vinaigrette) and the can surface with the aim of preventing degradation of the latter by the corrosive action of the former. Once manufactured, the cans were supplied to a canner who put tuna in them which, in turn, was supplied to it by a canned tuna purveyor. The filled cans were then delivered back to the canned tuna purveyor who sold them to distributors and retailers under the well-known brand 'John West'. 2 From July 2004 internal corrosion was observed on the score line of the easy open can ends. In some cases, the corrosion would make its way from the inside of the can to the exterior. When this occurred, the contents of the cans spilled out on to other cans then spreading the corrosion further. Most cans were not affected by the problem but a significant number were. There followed two product recalls by the canned tuna purveyor of its product from the retail market. Initially this was confined to tuna in the zesty vinaigrette but the second recall swept up tuna in brine as well as in lemon and cracked pepper. Claim and counter claim ensued. The canned tuna purveyor claimed on the canner and the canner claimed on the can manufacturer. The can manufacturer settled the claims against it paying $6,105,786 to the canned tuna purveyor and $1,630,000 to the canner. Having done so, the can manufacturer then sued the lacquer supplier, Siegwerk; Siegwerk sued the resin supplier, Nuplex, and it sued its insurers. 3 Before the matter was called on for trial, Siegwerk settled the claim made against it by the can manufacturer for $2,250,000. All that was left of the suit that was eventually heard by the learned trial judge was the Siegwerk claim on Nuplex and Nuplex's claim on the insurer. The trial judge determined that the winner in this imbroglio was Nuplex. It was held not to be liable to Siegwerk and, going in the opposite direction, its insurer was held to be liable to indemnify it for the cost of the litigation subject only to an excess. 4 Leaving the insurance issues to one side, the main issue at the trial was whether Nuplex was liable to Siegwerk for the amount for which it had settled with the can manufacturer. As a result of some interventions by the trial judge the parties had agreed that the sole question for determination was whether the resin supplied by Nuplex had caused the internal corrosion of the cans. Before this Court an attempt was made by Siegwerk to complain that this had resulted in the trial judge failing to deal with all of the issues raised by its claim. This submission is unsound. Once the parties had agreed, as they unquestionably did, that there was only one issue - causation - all of the other issues in the case ceased to exist as matters between them. Litigation could scarcely be conducted on a rational basis if trial judges were obliged to consider not only the matters said by the parties to be in issue between them but also those issues said by them not to be so: see Metwally v University of Wollongong (No 2) (1985) 60 ALR 68 at 71, 'It is elementary that a party is bound by the conduct of his case'. For the reasons given, therefore, by Robertson J I agreed at the hearing that Siegwerk's application to amend its notice of appeal to pursue this point ought be refused. 5 The forensic task faced by Siegwerk was then to link the behaviour of the resin supplier, Nuplex, to the corrosion of the can ends. As it happened, liability was an easy burden. Nuplex was obliged under its contract with Siegwerk to provide it with a resin known as Nuplex 3490 EP. The terms of the contract provided that Nuplex 3490 EP was to include an epoxy resin ingredient called DER 669-E and that this was to be sourced exclusively by the resin supplier from Dow Chemicals in Germany. 6 In the events which occurred, Nuplex did not use the mandated DER 669-E but instead another epoxy resin called Epikote 1009. This action by Nuplex in substituting DER 669-E with Epikote 1009 when making the Nuplex 3490 EP phenolic resin was, therefore, a direct breach of contract and Siegwerk was entitled, at least, to nominal damages for that breach of contract. 7 The debate which ensued at the trial was concerned, therefore, not with the issue of whether the use by Nuplex of Epikote 1009 was in some way actionable at law. No question of negligence arose and there was no inquiry into whether the Nuplex 3490 EP was of merchantable quality. There mere fact that Epikote 1009 was used in substitution for DER 669-E was sufficient to provide a basis for legal liability in contract. The issue instead was only whether that, ultimately admitted, breach of contract was the cause of the corrosion in the cans. 8 The remaining forensic task confronting Siegwerk was therefore clear: it had to show that the cans were corroded because Epikote 1009 had been substituted for DER 669-E. 9 Many ways of proving such a case might be imagined. Siegwerk decided to pursue only two which overlapped. The first was a circumstantial case by which it sought to show that there was a temporal link between the time when the Epikote 1009 substitution began and the time when the corrosion in the cans occurred. The second was a more theoretical case in which an attempt was made to show that certain chemical consequences of using Epikote 1009 in the lacquer could be seen as making it more likely that corrosion would occur. 10 The circumstantial case was hampered in a number of key respects. 11 The first was that the switch to Epikote 1009 from DER 669-E was not the only change which had occurred in the lacquer around the time in question. To understand the issues of chronology it is useful to know that the lacquer Siegwerk supplied to the can manufacturer under its specifications was known as '650814'. Siegwerk was only the most recent provider of 650814 and, indeed, it was not until October 2003 that the lacquer supplied by Siegwerk came to be used in commercial production by the can manufacturer on can bodies and ends. It was at that time that the cans first came to be exposed to lacquer 650814 made by Siegwerk using Nuplex's epoxy phenolic resin 3490 EP. It was not until three months later, in January 2004, that the canner began filling cans made using Siegwerk's lacquer with tuna. It was six months later still, on 25 July 2004, that corrosion in these cans was first noticed. The trial judge found, and it did not appear to be in dispute, that where corrosion occurred it typically started a minimum of four to six weeks after the tuna was put in the cans. 12 The nature of the record keeping which was available did not permit any direct link to be made between individual cans and whether the lacquer used had been made using Epikote 1009 or DER 669-E. The impermissible substitution of Epikote 1009 appears to have occurred only between October 2003 (when the can manufacturer first started using the Siegwerk lacquer) and March 2004. In that time, there were apparently 49 batches of Nuplex 3490 EP provided to Siegwerk of which it could be shown that 13 contained Epikote 1009. There was no way thereafter of tracing directly which cans were treated with lacquer using those batches. 13 Looked at this way, it is easy to see Siegwerk's enthusiasm for the proposition that there was circumstantial evidence linking Nuplex's use of Epikote 1009 in batches made from October 2003 to March 2004. There was a reasonably distinct temporal coincidence. 14 The difficulty was that this was not the only change which occurred to lacquer 650814 at the time that Siegwerk's lacquer began to be used. Another change which took place in the same month that Siegwerk's lacquer began to be used, October 2003, was the can manufacturer's decision to specify to Siegwerk that the lacquer should be less viscous. The trial judge found at [15] that it changed the viscosity specification for the lacquer from 80-100 FORD to 65-95 FORD. Thus it was not the case that the lacquer which began to be used on actual cans in October 2003 and which was detected as failing in July 2004 differed in only one way - the substitution of Epikote 1009 for DER 669-E - from its predecessor; it differed also in that it was being applied to the cans with decreased viscosity. There was a further difficulty which was that in the preceding year, April 2002, there had been another change in the recipe for lacquer 650814 with the substitution of one aluminium product for another. Unlike the reduced viscosity (which occurred in the same month the substitution of DER 669-E began) this occurred 18 months earlier. Nevertheless it was another change. 15 Thus, the timing of the can failure was consistent with it being caused by the substitution of Epikote 1009 or by the use of a lacquer with reduced viscosity or, perhaps, or although this seems less likely given the longer period, the change in the aluminium product. By itself such a circumstantial case could not succeed. Although it was consistent with the cause of the failure being the lower viscosity of the lacquer or the use of Epikote 1009 it simply provided no basis for choosing between them and hence no path to saying that one of those conclusions was more probable than not. 16 If Siegwerk was to succeed, therefore, it needed something additional and for this it turned to industrial chemistry. Expert evidence was adduced which, if accepted, could have tended to show that: (a) Epikote 1009 had a lower average molecular weight than DER 669-E; and (b) lacquer produced using epoxy phenolic resins with lower average molecular weights tended to be more brittle than those with higher average molecular weights. 17 The argument then was that given the more brittle nature of the ensuing lacquer and the timing of the can failures it was established on the balance of probabilities that the cause of the failure of the cans was the use of Epikote 1009 in the Nuplex 3490 EP. 18 The trial judge did not accept that the lacquer supplier had discharged its burden of showing, on the balance of probabilities, that the use of Epikote 1009 was the cause of the can failure. He did not accept that it was proved that Epikote 1009 did have a lower average molecular weight than DER 669-E. Further, although the trial judge was prepared to accept as plausible the idea that the use of an epoxy phenolic resin with a lower average molecular weight might result in a more brittle lacquer, his Honour was not willing to find that this had, in fact, been proved. The trial judge noted some other matters which might also be responsible for the failure, two of which were the decision by the can manufacturer to specify a lacquer with lower viscosity and the possible excessive wear by the scoring tool responsible for the creation of the easy open lid. His Honour's ultimate conclusion was that there was insufficient evidence to establish Siegwerk's theory of causation. 19 Siegwerk filed a notice of appeal containing four grounds. Its written submissions developed each of these. It is convenient to commence with the challenges to the trial judge's factual conclusions, first, that the lacquer supplier had failed to show that Epikote 1009 had a lower average molecular weight than DER 669-E and, secondly, that lacquers made with epoxy phenolic resins having a lower average molecular weight were more brittle and more likely to fail.

II. The Average Molecular Weight of Epikote 1009 20 At trial Siegwerk called Dr Scheirs to make good this point and Nuplex called Mr Haig. There was no question about either man's qualifications. Mr Haig did not venture a view on the question of whether Epikote 1009 had a lower average molecular weight than DER 669-E, confining himself to criticism of Dr Scheirs' methods. 21 Dr Scheirs had been involved in the litigation for a longer period than Mr Haig. His first report was prepared in March 2006 at a time when the can manufacturer was accusing Siegwerk of providing a defective lacquer. In 2006, Siegwerk was seeking to resist that proposition. As noted above, before October 2003 the can manufacturer had sourced its lacquer from a different manufacturer to Siegwerk. Its contention was that the problems had started when Siegwerk took over the supply of the lacquer in that month. The can manufacturer's expert had suggested that the new lacquer provided by Siegwerk had significantly different molecular weights to the old lacquer. Dr Scheirs had rejected this proposition, however, because he felt that the process of calculating molecular weights used in the can manufacturer's expert's report could not be relied upon. 22 In 2010 it emerged for the first time that there had been a substitution of DER 669-E with Epikote 1009 by Nuplex. Dr Scheirs then changed his views. In his second report (which was issued after Siegwerk had settled with the can manufacturer) he expressed not only the view that the average molecular weight of Epikote 1009 was 4000 whilst DER 669-E was 5000-8000 but also that this was likely to make the ensuing lacquer more brittle. 23 In a third report Dr Scheirs arranged for samples of Epikote 1009 and DER 669-E to be tested to see what their average molecular weights were. The first test was conducted by a chemical analysis company called Leeder Consulting. As it happens, there are in fact three measures of average molecular weight Mw (weight average molecular weight), Mn (number average molecular weight) and Mz (sedimentation average molecular weight). Dr Scheirs thought Mw the most useful. The result of the Leeder Consulting analysis was as follows: Mn Mw Mz Epikote 1009 Epoxy resin 3901 8425 10914 DER669E Epoxy resin 4401 11215 17770

24 From this Dr Scheirs concluded that the Mw of DER 669-E was 25% greater than that of Epikote 1009. Dr Scheirs also noted that the Mn of the Epikote 1009 was 3901 and this was close to the Mn quoted in the product data sheet for Epikote 1009 which was approximately 4000. The point of that last observation was to demonstrate that the sample which had been tested was typical. 25 Mr Haig was critical of this process in a number of ways. 26 First, he did not think that only one sample of each substance was enough given that they were substances which exhibited a large range of weights. Additionally, nothing was known of the samples' age which could affect the results. In fact, as discussed below, Dr Scheirs did arrange for two sets of tests. 27 Secondly, he noted that Dow's data sheet for DER 669-E had said that it had a large molecular weight range Mw of 5000-8000 (Mn). That is, for an arithmetic average of 6,500 Mn there was a possible variability of ± 25%. That variability mattered because the difference found between the Epikote 1009 (3901) and DER 669-E (4401) samples analysed was only 11% (Mn). Mr Haig criticised Dr Scheirs for regarding that 11% as critical when the product variability of Mn was ± 25%. Dr Scheirs did not, however, say that. His evidence was that what was significant was the 25% difference in the Mw values. He did say that the Mn was, in fact, higher for DER 669-E but he was quite clear that the most commonly used measure was Mw and that there the difference was 25%. 28 Mr Haig's third criticism had more substance. The product data for DER 669-E suggested a Mn in the range of 5000-8000, but the ChemicalAnalysis results suggested a Mn of 4401, i.e., well outside the quoted ranges. This was capable of suggesting that there was something irregular about the sample of the DER 669-E which was used for testing or the testing process itself. 29 A fourth criticism Mr Haig made was that the Mw value found by Leeder Consulting for Epikote 1009 was at variance with the reported values for it in the literature. So Kioke had reported its Mw as 9454 (as opposed to Leeder Consulting, 8425) and Mori had reported it as 15300. 30 For those reasons Mr Haig was sceptical of the process by which Dr Scheirs had gone about measuring the Mw for Epikote 1009. 31 The trial judge approach the matter this way (at [73]): The first assumption on which Dr Schier's [sic] theory is based is that there was in fact a difference in molecular weight between the Epikote 1009 used by Nuplex in substitution for the DER 669-E that Nuplex was bound by its contract with Siegwerk to use. The validity of this assumption is very difficult to establish. In the first place, the use by the respective manufacturers of the two epoxy resins of ranges in the technical data they publish about molecular weight has the result that the only comparison possible on the basis of that technical data is a comparison of the ranges, rather than a direct comparison of the molecular weights of the particular inputs into resin manufactured by Nuplex. Dr Scheirs said that this difficulty was overcome by the analysis he had carried out of a sample of each of the two epoxy resins, which revealed significant differences in molecular weight. I accept Mr Haig's criticism of this proposition. Given that molecular weights are expressed in ranges, analysis of a single sample of each epoxy resin is unlikely to be of much significance in determining whether there were actual differences in molecular weights. This is not to say that I agree with Mr Haig's statement that he would need to be "convinced" of the difference in molecular weight. It is clear from his evidence that Mr Haig was applying a standard well above that of the balance of probabilities in saying that he was "[n]ot convinced". It is true, however, that separate comparative analyses of a series of samples from different batches of the two epoxy resins could have provided a firmer basis on which it could have been determined with any confidence whether there was a difference in molecular weights between the two resins used by Nuplex. The evidence as it is falls short of validating the first assumption. 32 The lacquer supplier's complaints about this reasoning are fivefold: (a) the trial judge erred in concluding it was difficult to assess the molecular weights involved because Dr Scheirs had, in fact, done so; (b) the manufacturing data sheets were in evidence and these showed Epikote 1009 had a lower molecular weight; (c) Dr Scheirs gave evidence to the same effect; (d) Mr Haig agreed; and (e) the trial judge was wrong when he said that only one test had been conducted by Leeder Consulting; in fact, there had been two. 33 Proposition (a) is rhetorical only and assumes its own correctness. It is true that Dr Scheirs had assessed the molecular weights but the entire question at the moment is whether he did so correctly. The fact that Dr Scheirs had the molecular weights tested cannot, by itself, provide any more support for the correctness of those tests than the testing process itself does. 34 Proposition (b) is wrong. The Epikote 1009 data sheet was in evidence. It did not reveal an Mw for the product but it did say that the Mn was approximately 4000. The reference given in Siegwerk's written submissions to Tab 28 of Part C of the appeal books was not a reference to a data sheet but instead to an analytical report. The reference given in the same submissions to p 21 of Mr Haig's report is not a reference to a data sheet either. On that page Mr Haig reported that he had sent an email to the manufacturer of DER 669-E, Dow, which had replied that the Mw for DER 669-E was between 5000-8000. The evidence did not permit any comparison so far as data sheets were concerned between the Mn of Epikote 1009 and DER 669-E for the data sheet for Epikote 1009 quoted a Mn value and the response from Dow was an Mw value (even assuming it could be treated as a data sheet). It was not correct to say, therefore, that the manufacturing sheets showed Epikote 1009 had a lower molecular weight than DER 669-E without explaining the relationship between Mn and Mw. This was not attempted. 35 As to (c), it is true that Dr Scheirs gave evidence to this effect but, as the passage above plainly shows, the trial judge took this into account. 36 As to (d), it is not correct to say that Mr Haig agreed. The transcript reference given is to T54. At that part of the trial Dr Scheirs and Mr Haig were giving concurrent evidence and were engaged in a dialogue with each other. At the top of the page (and the bottom of the preceding page) Mr Haig said this: - - - in that the Epikote 1007 has a molecular weight range that's published that's only half the value published for the 1009 and to find a result that's comparable between those two is really quite disturbing as far as my acceptance of these results as being truly accurate. And in fact, one of the results that John produced for the D.E.R. 669, the ones done by GPC -and came back with a result of 4400 is in fact outside the published range for the D.E.R. anyway. So all in all I don't have a lot of confidence in these results and I'm quite willing to accept that there may well be a molecular weight difference on average between the two products. But we haven't been able to establish it with the testing that has been performed. 37 Unsurprisingly, Dr Scheirs did not agree. It is not necessary to set out his response. It suffices to say that he emphasised the orthodoxy of the testing procedure. Siegwerk then relies on Mr Haig's response to this: I'm quite happy to accept John's contention that, in fact, samples analysed in the one laboratory on the one machine at the same time and calibrated at the same time are going to produce a relative figure that's quite acceptable. I agree with that. I guess in this whole molecular weight argument and the differences between the two starting epoxies, at the end of the day I don't see that that is significant, and we will discuss that further through a number of avenues, one of which being the published literature for the performance, and also the fact that there are many other valid reasons for the failures without becoming involved in these subtle differences of molecular weight. 38 Siegwerk submits that this is evidence that Mr Haig accepted that the molecular weight of Epikote 1009 was lower. It does not say that. 39 As to (e), the parties' submissions were difficult to follow. Siegwerk submitted that the trial judge had made a mistake in thinking that Dr Scheirs had only commissioned a single test of Epikote 1009 and DER 669-E. My reading of paragraph [73] of the judgment is that it proceeds on the basis that there was only a single test arranged by Dr Scheirs. There were, in fact, two tests. The first was commissioned by Dr Scheirs from Leeder Consulting on 4 November 2010 and the results received on 18 November 2010. The results of that test are set out above at [23]. Dr Scheirs appears then to have obtained another analysis on around 13 July 2011. This time the analysis was conducted by an entity called 'ChemicalAnalysis'. This Court was not taken to any material as to how this came about. The result of the analysis was before the Court, however, and it showed a Mn for DER 669-E of 6834 and for Epikote 1009 of 4789. It was this analysis that Siegwerk elsewhere erroneously referred to in its written submissions as a data sheet. 40 Nuplex's submissions took issue with the idea that the trial judge had not been aware of both tests. Reference was made in its written submissions to paragraphs [39], [45] and [47] of the judgment. These contain, however, no reference to the second test at all. Indeed, [45] explicitly referred to Mr Haig's criticism of Dr Scheirs for only using a single sample. Nuplex sought, in Mr Donaldson SC's address to the Court, to establish that what the trial judge meant was that there had only been a single sample of the two resins tested, albeit this sample had been tested twice. That is not the natural reading of [73] or [45]. The reasons for judgment contain no reference to the second test at all. In that circumstance, and even apart from the ordinary language of [73], it is much more likely that the trial judge was referring to there having been only one test and not two tests of a single sample. 41 Accordingly, I would conclude that this was an error. As I have already said, the second analysis suggested a Mn for Epikote 1009 of 4789 and for DER 669-E of 6834. It did not include any analysis directly of Mw but it did include a measurement of a variable called 'EEW' (epoxide equivalent weight) which Mr Haig elsewhere gave evidence could, in a rough and ready way, be half the value of Mw. There was no discussion of this effect in the appeal, however, and I do not think it appropriate to proceed on the basis that Mw is roughly half of EEW. 42 Taken together the two tests commissioned by Dr Scheirs, therefore, suggested the following: Leeder Consultants ChemicalAnalysis Mw Mn Mw Mn Epikote 1009 8425 3901 - 4789 DER 669-E 11215 4401 - 6834

43 Mr Haig's criticism was that an accurate picture could not be obtained from a single sample. The trial judge rendered this criticism at [73] as an observation that there needed to be a series of tests done. I do not accept that if the trial judge had adverted his mind to the second test, that the result would necessarily have been the same. It would have been possible, it is true, to discount the second test, by noting that it dealt with Mn and not Mw or by saying that two tests were not enough. But there may have been answers to this. One answer might be calculating Mw from EEW. I cannot say this error was immaterial.

III. Whether lower molecular weight epoxy resins lead to more brittle lacquers 44 Dr Scheirs explained this point in his final report as follows: The use of the unauthorized lower MW Epikote 1009 (molecular weight 4000) in place of the stipulated DER669 (molecular weight 5000 - 8000) reduced the resistance of the coating to cracking under scoring stresses and reduced the level of adhesion of the lacquer to metal under autoclave sterilization. The low MW epoxy gave a coating that was more brittle and with less adhesion to the metal substrate than the PPG resin. The Epikote epoxy was used by Nuplex until March 2004 based on the information that I have seen. Use of Epikote 1009 in place of DER 669E would explain the variability in performance both of the Sicpa product as well as between the Sicpa product and the PPG product. 45 Dr Scheirs gave evidence that his view was supported by two academic papers; one by Kojima, the other by Kojima and Watanabe together, ('the Kojima Papers'). 46 Mr Haig had two points in response. First, he noted that the lacquer had other components in it besides the Epikote 1009 and reasoned that it was not possible to draw conclusions about the effects of Epikote 1009 without first understanding those effects. Secondly, he argued that the Kojima Papers did not apply to grade 9 epoxy resins and that both Epikote 1009 and DER 669-E were resins of that kind. Dr Scheirs did not accept this interpretation of the Kojima Papers. 47 The first point was not debated before this Court. As to the second point, it will be seen that what is involved is a dispute about the meaning of the Kojima Papers. 48 The trial judge dealt with the matter this way at [74]: The second assumption made by Dr Scheirs is that a lower molecular weight of the epoxy resin used results in a less flexible coating on a can end, after the epoxy resin has been mixed with a phenolic resin, the resulting epoxy-phenolic resin has been mixed with other ingredients to make a lacquer, the lacquer has been applied to can ends in two coats and each of those coats has been separately cured by heating. Dr Scheirs says that this assumption is supported by the Kojima papers, which show the results of experimental work demonstrating the correlation between higher molecular weight of epoxy resin and greater flexibility of final coating. There can be no doubt that the contents of the Kojima papers do support the existence of such a correlation up to a point. Mr Haig's criticism is that Grade 9 epoxy resins (which include Epikote 1009 and DER669E) appear to be beyond that point. In other words, the higher the grade of epoxy resin, the more likely it is that there will be a plateau effect or (Mr Haig says) even a downturn in the flexibility of the finished product. Grade 9 epoxy resins are very high grade products. I do not accept that the assumption made by Dr Scheirs that any lowering of molecular weight involved in the substitution of Epikote 1009 for DER669E resulted in less flexible lacquer coating on can ends is borne out. The assumption might well be correct, but further evidence would be needed to validate it. 49 Both Mr Haig and Dr Scheirs claimed that the Kojima Papers supported their position. In brief, Dr Scheirs pointed to certain parts of the Kojima Papers to make good the proposition that when used as part of a coating, lower average molecular weight epoxy resins were more prone to brittleness and adhesion problems. On the other hand, Mr Haig pointed to other parts of the papers to make good the proposition that, whilst this was generally true, it was not true in the case of what he called group 9 epoxies (of which Epikote 1009 and DER 669-E were apparently members). To resolve this debate it was necessary to grasp the point being made by both gentlemen and to assess it. 50 The debate between them was technical. They were ad idem in accepting that there was a general rule of thumb that as the average molecular weights of epoxy resins increases the flexibility of the epoxy phenolic resins also improves. Mr Haig's reasoning was that this rule of thumb ceased to apply to grade 9 epoxy resins. His evidence was as follows: So I guess in taking issue with John's theory, we can start I suppose by saying that, yes, there is quite an acceptable rule of thumb that as the molecular weight of epoxies increases and the epoxy phenolic resins that are manufactured from those epoxies increases, that you start - you get superior performance because of better adhesion and flexibility, and all of the good things that you want in this particular application. It is though a rule of thumb and it has its limitations, and the main limitation I see is that it is really only applicable over a range of epoxy resins up to a point. At the point of about 5, 9 epoxies we start to get into a region where the molecular weight differences are less critical on performance and, in fact, beyond that the molecular weight increases can actually be detrimental to performance. Now, I rely in forming that opinion on the work of Kojima in that paper that I quote in my report. We can delve into Kojima at this point if you wish, your Honour. It is going to be probably hard going but - - - 51 Thus the essence of the debate foreshadowed by Mr Haig was that the Kojima Papers would demonstrate that for grade 9 epoxy resins (such as Epikote 1009 and DER 669-E) there were little discernible improvements in flexibility and adhesion as average molecular weight increased. This proposition was disputed by Dr Scheirs who read the Kojima Papers differently. Both readings were quite technical and by no means susceptible to simple explanation. 52 It is evident that the trial judge accepted Mr Haig's point about grade 9 epoxy resins but neither at [74] nor anywhere else did his Honour record any of the detail of what the debate was or why he accepted Mr Haig's view. The reasons simply accept that the conclusions in the Kojima Papers did not apply to grade 9 epoxy resins. Siegwerk submits that this reasoning was not adequate. In my opinion, this contention is correct. 53 To see why, one needs to understand the debate. At its heart, not mentioned in the trial judge's treatment of the subject matter, was the variable Mc. Mr Haig and Dr Scheirs agreed on this definition of Mc: When describing both polymeric networks and crosslinked polymers, the basic quantity used is the so-called network parameter, Mc, defined as the number average molecular weight between branch points (junction points). In crosslinked polymers, we can distinguish primary chains and crosslinks, and thus Mc relates to the segment of the primary chain between two crosslinks. 54 They also agreed that the Kojima Papers took Mc and related it to the performance characteristics of the epoxy resins in question. Put another way, Mc measured the molecular weight between the crosslinks in the resin and the point of the Kojima Papers was to examine the relationship between flexibility and adhesion, on the one hand, and the variable Mc, on the other. It might be observed, as both Dr Scheirs and Mr Haig accepted, that Mc and average molecular weight are not quite the same thing. Both accepted that Mc was the average molecular weight between crosslinks. No attempt was made to explain to this Court what this meant. Both parties in this Court appeared to proceed on the basis that average molecular weight was some kind of proxy for average molecular weight between the crosslinks. Neither party sought to say whether average molecular weight between the crosslinks stood as a proxy for Mw or Mn or for something else. 55 The testing carried out by Kojima involved creating coatings of various resins with different varying levels of Mc, folding these coatings over themselves twice (referred to as '2T') and then dropping a steel block of fixed weight on to them from 40 cms above. After this the disruptions to the surface coating ('enamel') of the specimens were measured electrically (or 'rated') as a result of which a value was derived. This was called the enamel-rater value or 'ERV'. The lower the ERV, the more flexible the coating was. 56 Mr Haig's point about the Kojima Papers turned on Figure 6 in the second paper which looks like this: 57 Kojima and Watanabe explained Figure 6 as follows: Solvent-borne coatings were formulated using the 4-times extracted epoxy resin and the original epoxy resin at various epoxy/phenolic ratios. These coatings were coated on TFS and baked in an oven at 210˚C for various times to obtain the coating films with various degrees of cure. Figure 6 shows the evaluated results of flexibility by an enamel-rater as a function of calculated Mc. The ERVs were well expressed as a function of Mc, which possess a minimum at around Mc = 4,000. In the lower Mc region, the network structure appears to be too tight and rigid; whereas in the higher Mc region, it appears to be loose and imperfect. Thus, it appears important to control Mc to obtain optimum flexibility of the cured coating films, but not Ws. 58 Some things Figure 6 shows are perhaps tolerably clear to an appellate court, some not. Into what might be thought to be the clear category can be put the observation that the figure does appear to show that as Mc increases the flexibility of Epikote 1009 improves right up until just above a Mc of 3500. However, there are aspects of Figure 6 which are confusing to lay people. Why, for example, is it indicated that the Mc, of the Epikote 1009 increases apparently from 2000 to just under 4000? The answer to that may appear on the previous page where the authors explained that they had removed, in varying degrees, the low molecular weight elements in the resin to obtain different levels of Mc. They said: To examine the adverse effects of lower molecular weight components, a model calculation was carried out. We assumed that the molecular weight distribution obtained by GPC is segmented into fractions and that each fraction in the sol is formed by the fraction of the original epoxy resin corresponding to the same molecular weight. Then, we eliminated a certain fraction both from the original epoxy resin and the sol. Calculated Mc for such distributions corresponding to a certain sol content, Ws, gives information on how Mc changes by eliminating a certain molecular weight fraction. Figure 4 shows the calculation results for Ws = 0.10, 0.15, and 0.20. The contribution of the fractions with a molecular weight range between 1,000 and 10,000 appear to be most significant. In other words, a network structure of higher Mc can be obtained by eliminating such fractions. The influence of fractions with a molecular weight < 1000 appears not to be significant. However, this is because of their lower content and the influence of each molecule may be even more significant. Thus, it will be concluded that the epoxy resin without any lower molecular weight components is preferable to obtain a network structure with higher Mc. 59 If that be right (and I am by no means sure that it is) then the various levels for the Epikote 1009 displayed in Figure 6 reflect the fact that it has had varying bands of low molecular weight material removed from it to achieve varying Mc levels (on the basis that removal of lighter material will increase the average weight). On the other hand, Kojima and Watanabe also refer to varying curing times at the standard temperature of 210˚C and it may be that this is the reason for the fluctuations in Mc. Having no expertise in industrial chemistry, it is very difficult for me to say. 60 Whilst the trend of the Epikote 1009 (that is, the open circle dots) indicated that flexibility increased as Mc increased towards 3700, the black dots (which were the liquid-liquid extracts with even higher average molecular weights) revealed decreasing flexibility as Mc increased beyond 3700. Kojima and Watanabe explained this as follows: An epoxy resin, E1009, was used to prepare epoxy resins with sharper molecular weight distribution. Solid epoxy resin (300 g) was charged in a flask with 3000 g of ethanol and agitated at the boiling temperature for 3 h. After cooling, the mixture separated into two phases. The upper phase, which contained lower molecular weight components, was disposed. One third of the residue was dried and used as an extracted resin. By repeating this procedure, epoxy resins with a different molecular weight distribution were obtained. Figure 2 shows the molecular weight distribution patterns of these liquid-liquid extracted resins. 61 Mr Haig put it this way: … And having done that, he looks at the flexibility of commercial epoxy resins. So having performed experiments where he, if you like, purifies the resin, he actually produces an unrealistic series of lacquers that he tests. He then goes back to the flexibility of commercial epoxy resins and he tests Epikote 1009 and produces a range of final molecular weight products in the cured baked film by changing the stoichiometry or the baking conditions, and if we go to page 227, figure 6, you will see that the filled circles or the solid dot points represent the liquid-liquid extracted epoxy resins - that is, those that he has modified to produce sufficiently high molecular weight material, and also, the E-1009, which he has produced varying molecular weight final materials by changing the curing conductions. … 62 The significance of this then for Mr Haig was that it allowed Kojima and Watanabe to create a class of epoxy resins which had a very high average molecular weight between the crosslinks, higher indeed than that of Epikote 1009. When the trend line was extended what it showed was that the ERV started to go up when the Mc exceeded 4000, that is to say, that increasing molecular weight past 4000 with Epikote 1009 led to decreased flexibility and not, as the rule of thumb would have it, increased flexibility. Mr Haig explained this as follows: Now, a - a Mc value of about 3800, 3700 for the E-1009, if you look at figure 7, that is down in the - what is called the optimum range of performance for the enamel rater value for those epoxy resins. Now, in that area of about 3800, once can see that if you move lower or higher, you're not going to get any improvement in enamel rater value. So it's my contention that in fact molecular weight of the starting material as it is reflected in the final Mc value is in fact very insensitive in the range where we expect these type 9 epoxies to perform. 63 Thus was Mr Haig's argument. On any view, it was reasonably complex. As might be expected, Dr Scheirs' response was equally complex and had, in effect, two elements. The first was that the thrust of the Kojima Papers contradicted what Mr Haig was suggesting. He quoted the abstract from the second article: The coating films with a higher molecular weight of network chain, Mc, showed more flexibility. This was achieved by the use of an epoxy resin with higher molecular weight or an epoxy resin with a lower content of lower molecular weight components and by control of cure. 64 This left unchallenged, however, Mr Haig's treatment of Figure 6. Dr Scheirs dealt with it by focusing on the liquid-liquid extracted epoxy. His point was that this was not Epikote 1009 and that the increasing ERVs were to be seen as having been caused by increased curing times. Dr Scheirs put it this way: So it's quite clear what he is saying there, but now we need to revisit figure 6. So what he has got here in figure 6, by my estimation, is that he has reproduced those open circles for the Epikote. So we can take that data as being valid. He has then got some dark circles, and that refers to the liquid-liquid extractions. The problem I have with those dark circles is, if we go to his experimental section - or in fact, sorry, page 226, the previous page, the last paragraph: Solvent borne coatings were formulated at various epoxy-phenolic ratios. These coatings were coated and baked under various conditions to obtain coating films with varying degree of cures. So what he has done with those black circles, is he has taken firstly liquid-liquid extracted epoxy. It's no longer - yes? No? The black circles. … Okay, just the black. He has taken liquid-liquid extracted epoxy, which is no longer Epikote 109. Molecular weight has shifted. Then he has baked them under varying degrees of cure. Now, varying degrees would mean from both low to high. So - and once you overcure these systems, they become too densely crosslinked, and they then no longer have inherent flexibility. They become brittle. And that's what you see. That's why you see that curve moving up, because he has taken the higher molecular weight and then cured them to varying degrees. 65 Thus the debate between the experts was about what Figure 6 meant and how it was to be interpreted in light of the article as a whole. It was a complex issue. The only treatment of this question is at paragraph [74] of the trial judge's reasons (above). Of that paragraph the entire topic is dealt with in 6 sentences which I will repeat: Dr Scheirs says that this assumption is supported by the Kojima Papers, which show the results of experimental work demonstrating the correlation between higher molecular weight of epoxy resin and greater flexibility of final coating. There can be no doubt that the contents of the Kojima Papers do support the existence of such a correlation up to a point. Mr Haig's criticism is that Grade 9 epoxy resins (which include Epikote 1009 and DER 669-E) appear to be beyond that point. In other words, the higher the grade of epoxy resin, the more likely it is that there will be a plateau effect or (Mr Haig says) even a downturn in the flexibility of the finished product. Grade 9 epoxy resins are very high grade products. I do not accept that the assumption made by Dr Scheirs that any lowering of molecular weight involved in the substitution of Epikote 1009 for DER 669-E resulted in less flexible lacquer coating on can ends is borne out. The assumption might well be correct, but further evidence would be needed to validate it. (emphasis added) 66 With great respect to the trial judge, the italicised portion is inconsistent with a full grasp of the evidence which had been led. There was no plateauing effect. There was a trend down in ERV as Mc increased towards 4000 and then a trend up past the 4000 point using liquid-liquid extract resins. Mc was not a measure of how 'high grade' an epoxy resin is (whatever that meant), it was a measure of the molecular weight between the crosslinks. On my reading of the Kojima Papers there was nothing about 'high grade' epoxy resins in it at all. 67 All of that might well be material if one were trying to work out the answer. However, much more important is that there is no reference in the trial judge's reasons to Dr Scheirs' responsive argument that the data points for the liquid-liquid extract could be disregarded because they were not Epikote 1009. To engage with that argument it would have been necessary to have had a working theory of what Figure 6 meant. 68 In my opinion, the reasons of the trial judge on this issue are absent: cf. State Rail Authority of New South Wales v Earthline Constructions Pty Ltd (1999) 160 ALR 588 at 607 [64] and 622 [94] - [95]. Nor do I think this is a case where one could speculate about what the correct outcome should be so that this Court could attempt to answer the issue for itself. I do not have any confidence that an attempt on my part to say what the correct answer is would be correct.

IV. Conclusions on factual challenges 69 Error is shown in the trial judge's treatment of both of the critical assumptions. On the first assumption, his Honour failed to appreciate that Dr Scheirs had arranged for two tests, one with Leeder Consulting and one with ChemicalAnalysis. On the second assumption, the trial judge failed to come to grips with the central debate between the experts and in consequence gave no reasons at all for preferring Mr Haig's view. 70 It is not possible to say that either of these errors is not material. As to the first, whilst one can conceive of a chain of reasoning by which a tribunal of fact might determine that the second test made no difference, this is not, by any means, inevitable. It is possible that the second test might have persuaded the trial judge that Mr Haig's concerns about the sampling process were unfounded. Of course, his Honour could also have concluded that two samples were not enough either. The important point is that this Court cannot be definitive about what the outcome would, or should, be. 71 On the second point, the principal scientific question has simply not been determined. Again, this Court is in no position to determine the question for itself. Grounds 3 and 4.2 of appellant's notice of appeal are made out. Accordingly, there will need to be a retrial. 72 This makes it unnecessary strictly to deal with the remaining grounds. I would nevertheless make the following brief remarks. The trial judge reasoned that although his Honour was confronted with several different possible reasons why the corrosion occurred the question was whether Siegwerk had proved it was as a result of the use of Epikote 1009. His Honour concluded that whilst Dr Scheirs' theory about average molecular weight was plausible it had not been proved on the balance of probabilities. As a syllogism, I detect no error in this. Having found that it had not been shown that the average molecular weight of Epikote 1009 was less than that of DER 669-E the analytical theory could not succeed for the premise upon which it rested was not made out. I would therefore have rejected ground 4.1 which sought to criticise this style of reasoning. I would also have rejected ground 4.3 which sought to suggest that the trial judge had erred in not being more critical of Mr Haig. No such obligation existed. 73 Grounds 1 and 2 effectively sought to argue that the trial judge's conclusions were against the weight of the evidence. Had the trial judge noted the second test commissioned by Dr Scheirs and had he dealt with the debate about the Kojima papers and determined it adversely to Siegwerk, I would not have accepted this argument. The evidence before the trial judge was capable of supporting a favourable conclusion to either party. The difficulties in this appeal lie not in the trial judge having arrived at a conclusion which the evidence cannot sustain but, rather, in failing adequately to deal with the evidence. 74 My initial view is that the retrial should be confined to the single issue determined by the trial judge but I would permit the parties to debate this if they wish in brief written submissions filed within 14 days. 75 On the insurance issues, I have had the advantage of reading in draft the reasons of Robertson J. I agree that the insurer is entitled to succeed on its appeal for the reasons given by his Honour and join in his Honour's proposed orders. 76 Subject to the matter in [74] above, in Siegwerk's appeal I would propose the following orders: 1. Set aside the orders made on 19 March 2013. 2. There be a new trial on the single issue of whether the substitution of Epikote 1009 for DER 669-E caused the corrosion in the cans. 77 Pending determination of the appropriate orders the only formal order which should be made now is that the parties bring in short minutes of order or competing written submissions on the scope of the retrial and costs within 14 days. I certify that the preceding seventy-seven (77) numbered paragraphs are a true copy of the Reasons for Judgment herein of the Honourable Justice Perram.

Associate: Dated: 20 November 2013 IN THE FEDERAL COURT OF AUSTRALIA VICTORIA DISTRICT REGISTRY GENERAL DIVISION VID 265 of 2013

BETWEEN: SIEGWERK AUSTRALIA PTY LTD (ACN 007 114 338) IN LIQUIDATION

AND: NUPLEX INDUSTRIES (AUSTRALIA) PTY LTD (ACN 000 045 572)

JUDGES: PERRAM, ROBERTSON AND DODDS-STREETON JJ DATE: 20 November 2013 PLACE: MELBOURNE

IN THE FEDERAL COURT OF AUSTRALIA VICTORIA DISTRICT REGISTRY GENERAL DIVISION VID 713 of 2013

BETWEEN: QBE INSURANCE (AUSTRALIA) LIMITED (ACN 003 191 035)

AND: NUPLEX INDUSTRIES (AUSTRALIA) PTY LTD (ACN 000 045 572)

JUDGES: PERRAM, DODDS-STREETON AND ROBERTSON JJ DATE: 20 November 2013 PLACE: MELBOURNE

Dodds-Streeton J 78 I have had the advantage of reading in draft the separate reasons of Perram J and of Robertson J. 79 For the reasons given by their Honours I agree that the appeals in VID 265 of 2013 and VID 713 of 2013 should be allowed and I agree with the orders each proposes. I also agreed, at the hearing of the appeal, that leave to amend the notice of appeal should be refused, for the reasons given by Robertson J. I certify that the preceding two (2) numbered paragraph is a true copy of the Reasons for Judgment herein of the Honourable Justice Dodds-Streeton.