Balmoral Group Ltd v Borealis [UK] Ltd. & Ors

This judgment is divided into the following sections:

Balmoral Group Limited (“Balmoral”) is a privately owned company, registered in Scotland. It was founded in the 1970s by Dr James Milne, its Chairman, Managing Director and controlling shareholder. Its head quarters are in Aberdeen. In the year to 31^st March 2005 it had a turnover of over £31,000,000. Much of its business involves supply to the oil, gas and marine industries. It has a wide range of products, including sectional tanks, modular buildings and pipes. The Group has a transport business. Balmoral Marine Ltd, originally an associate and now a subsidiary company, provides specialist services for the offshore oil industry through the hire and manufacture of items such as navigational and buoyancy aids and wire rope. The Group has subsidiary companies in Norway, Houston and Holland.

Balmoral manufactures, amongst other things, storage tanks by a process known as rotational moulding (“rotomoulding”). Balmoral entered the rotomoulding business in 1992, when it began to design tanks for oils and other liquids, such as water or wastewater, which it began to manufacture towards the end of that year (Footnote: ¹). If used as oil tanks, they usually contain kerosene; less often, diesel. Balmoral’s designs although similar, in some respects, to those of other tank manufacturers, are unique to it. Balmoral’s staff includes engineers, designers, and technical support staff. Most of

Rotational moulding works in this way. A quantity of polyethylene in powdered form (the charge) is placed into one of the two halves of a steel mould, which are then clamped together. The mould, which is mounted on the arm of a machine, is introduced into an oven and continuously rotated on two axes at a low speed (typically 4- 8 revs/min), so that all parts of the interior surface of the mould pass through the pool. The charge becomes tacky and then melts in the heat (of up to around 300° C), forming a molten pool. As the powder becomes tacky it starts to stick to the mould. More powder sticks to particles that have become tacky before. As a result a layer of viscous, largely immobile (Footnote: ²), liquid forms over the entirety of the mould. The wall thickness distribution of the melt is largely determined at this stage. The product has, however, to remain in the oven in order to eliminate the bubbles (trapped pockets of gas) in the melt. After about 20 minutes the mould is removed from the oven and cooled. At this stage crystallization of the melt will occur. The crystallizing melt will shrink away from the mould, the rate of crystallization being dependent, in part, on the rate of cooling. When the polyethylene is sufficiently cool, the tank, as it has now become, is removed from the mould. A very high proportion of rotomoulding production is of tanks of one form or another.

Natural polyethylene is white or yellow in colour. A very large percentage, of the oil tanks in the British Isles (Footnote: ³) are “oil tank green” (Footnote: ⁴). The standard pigment used in almost all cases, including by Balmoral, is a green phthalocyanine based pigment. There are several different brands of such pigment. The addition of a pigment has a deleterious effect on the properties of the natural product. Further, pigment particles affect the way in which the polymer crystallizes as it cools and hence the crystalline structure. Some pigments, of which phthalocyanine green is one, are nucleating agents i.e. they encourage early initiation of the process of crystallisation.

Polymer can be mixed with a pigment in a number of different ways. The two can be dry-blended. In this process polyethylene in powder form is mixed with the pigment either in a blender before being added to the mould or in the mould itself. An alternative process is grind blending whereby polymer pellets are mixed with pigment and the mixture is ground into a powder. Another option is to use a masterbatch. A masterbatch is created by taking a high concentration of pigment and grind blending or compounding it with a base polymer. The resulting masterbatch is then mixed either in a blender or in the mould with virgin polymer. A further process is compounding. In this case polyethylene powder is mixed with a pigment under heat in an extruder. The resulting pellet is then ground to powder. The blending process is sometimes carried out by the product manufacturer and sometimes by a third party.

In broad terms the evenness of the spread and distribution of the pigment and the homogeneity of the resulting structure are increased, according to which option on a scale from dry blending to compounding is selected. So is the expense. Dry blending involves no grinding. Grind blending a masterbatch involves grinding only a fraction of the material which will form the charge. Compounding is the most expensive of all because it involves an additional process. But it provides the best method of mixing and dispersion of the pigment in the product, and the least reduction in the properties of the base resin.

Balmoral ordered its first machine, a Ferry/Rotospeed 330, in July 1992. During 1993 Balmoral acquired three further machines – a 430, 280 and 220. Machines 430 and 330, which were used for the production of fuel tanks, work on a carousel basis with a fourfold cycle of (i) loading/demoulding; (ii) heating in the oven; (iii) pre-cooling and (iv) cooling (Footnote: ⁵), so that at any one time there is a mould at each of the four stages. In 2003 or 2004, when it entered the chemical tank market, Balmoral purchased a “rock-and-roll” machine in which, as its name implies, the product being moulded is rocked back and forward about a perpendicular axis, whilst being rotated continuously through 360°. Balmoral built up their production of tanks from 8 original moulds on the 330 to a complete range of tanks with a production of over 10,000 tanks a year, some 10% of the available market, within 2 or 3 years.

Until 1997 the polyethylene used by Balmoral had been manufactured by manufacturers (largely other than Borealis) using a type of catalyst known as Ziegler-Natta (“ZN”), the names of its inventors. For the most part (Footnote: ⁶) Balmoral used a blended grade of powder supplied by DSM United Kingdom Ltd (“DSM”) comprising two Stamylex materials one of which was a fully compounded green masterbatch and the other a natural grade polyethylene, with a melt flow rate (“MFR”) of 3. A melt flow rate is a measurement of the melt viscosity i.e. resistance to flow, of a polymer. It is a measure of the mass in grams extruded through a standard die in ten minutes under standard temperature conditions (usually 190º C) and with a standard applied load (2.16 kg) (Footnote: ⁷). Viscosity decreases in inverse proportion to melt flow: the higher the MFR number the lower the viscosity.

Between the spring of 1997 and July 2002 Balmoral purchased from one or more of the Borealis defendants large quantities of polyethylene which had been manufactured from polyethylene, the monomer, and hexene, the comonomer, using a metallocene catalyst. This polymer was known as borecene. A metallocene catalyst is one that contains organometallic components. Balmoral purchased borecene in three grades with different melt flow rates – MFR 3, MFR 6, and MFR 4. Metallocenes now represent about 20% of the European and American rotomoulding market. There are now over 100 rotational moulders using borecene in the UK.

Balmoral used borecene to manufacture many thousands of tanks, including many thousands of oil tanks for both industrial and domestic use, which it sold to customers in the British Isles and abroad. Oil tanks manufactured by Balmoral using borecene suffered a high failure rate, much higher than what was to be expected from rotomoulded tanks. Within a short period of service tanks produced from borecene were splitting, leading to customer dissatisfaction, oil spillage, and claims. Such failure was particularly prevalent in certain designs. Products manufactured from MFR 6 material showed a much higher failure rate than those manufactured from MFR 3 or MFR 4 material.

It is common ground that the mode of failure was environmental stress cracking (“ESC”). This type of failure occurs when the chemical environment of a material, in this case kerosene or diesel, causes premature brittle failure (Footnote: ⁸) of a plastic material that has been subject to stresses below its yield stress. All polyethylenes when subjected to long term stress below their yield stress will eventually fail in a brittle manner, although in the absence of a stress cracking agent the time that this takes may well be measured in decades. Water, which has a very low absorption in polyethylene compared to kerosene and diesel, is not an appreciable stress cracking agent but kerosene and diesel are. Other such agents are organic solvents or detergents and even greases, gases or vapour.

The effect on Balmoral’s oil tank business was disastrous. Balmoral claims damages of up to £50,000,000 as damages for breach of contract, misrepresentation or negligence.

There are three defendants:

The parties are in dispute as to:

Balmoral contends that, if the contracts were on anybody’s standard terms and conditions, they were on those of Balmoral. Borealis contend that all the contracts were on their terms and conditions. These exclude any warranty of fitness for purpose and limit recoverable damages (if the seller has not replaced the goods) to the price of the polyethylene in question. If that be so, Balmoral relies on the Unfair Contract Terms Act 1977 and contends that the exclusion does not satisfy the requirement of reasonableness.

Accordingly the following issues arise for decision

Balmoral manufactures a substantial range of tanks in over two dozen designs. For some of theses designs there was a “bunded” version, with a secondary skin to ensure containment in the event of failure of the inner skin. The designs fall into four broad categories:

Balmoral regards and marketed itself as at the Rolls Royce end of the storage tank market with a reputation for quality rather than cheapness.

Balmoral’s tanks, when designed, were submitted to two bodies for official certification. These were:

As its name implies, Oftec is a trade association, of which Borealis was a corporate member – the only polymer manufacturer who was. It confers a certificate of quality. The relevant Oftec standard is the Oftec Oil Equipment Standard OFS T100 “Polyethylene Oil Storage Tanks and Bunds for Distillate Fuels”. This standard, which came into force in 1994 and has been the subject of revision, has become the main standard for oil storage tanks. In part this is because many distributors are members of Oftec, have greater links with them than with BBA, and want tanks that pass the Oftec standard. A manufacturer has to submit three tanks of each design to an Oftec approved test house in order to gain Tank Type certification in an approved resin. Type Approval must then be secured if a new resin, not previously approved, is used. Balmoral went through this process. Most of Balmoral’s range of oil tanks had obtained OFTEC approval by 1996 and did so, thereafter, with each change of material.

For the storage of Class C1 and C2 fuels (which include kerosene and diesel) OFTEC specified a minimum wall thickness of 4.5 mm. That does not mean that a tank whose wall thickness is never less than 4.5mm will automatically be fit for purpose. What is the appropriate thickness must depend on the particular design of the tank. At the design stage Balmoral consulted an outside consultant called Tony Taylor, who had designed tanks for Titan (the brand name for tanks manufactured by Kingspan Group Plc [“Kingspan”], the market leader) and was said to have had “vast experience in product design, including that for oil tanks”. (Footnote: ⁹) He gave general guidance that there should be a material thickness of 8mm for smaller tanks, 10mm for medium tanks, and 12 mm for larger tanks, although in a memorandum of 28^th October (Footnote: ¹⁰) 1992 to Mr Hay Mr Gibson, one of the design group, observed that this view had “not been borne out by our calculations”. The memorandum gave estimates of wall thickness of 8mm in all the eight tanks concerned save for a 5.5 mm measurement for LP 275. This was apparently based on Finite Element Analysis (“FEA”); but no details of that have survived (Footnote: ¹¹). FEA is a computer based method of estimating stresses and deformation: see paragraph 261 below.

OFTEC had no requirements in relation to the environmental stress cracking resistance (“ESCR”) of the materials of which tanks are made. Prior to the events which have given rise to these claims and those of other manufacturers, ESCR does not appear to have been a topic of much concern. When borecene was being developed, Borealis carried out the then standard Bell telephone tests (to which I refer in paragraph 179 below); but did not until later carry out creep rupture testing (a method of determining long term creep resistance), since it was not the practice for raw materials manufacturers to do so. When Borealis piloted a borecene with an MFR of 1.9 with Tyrell Tanks (part of the business of Kingspan) in late 1995 or early 1996, no inquiry was made by them about creep rupture tests, and it did not occur to Borealis to carry out such tests until later. Nor did anyone raise that subject when borecene was launched at a conference in Vienna in October 1996. Rotomoulders assumed that borecene would have the same or better long term qualities as ZN material, which had proved satisfactory, and in respect of which creep rupture tests were not conventionally undertaken. Borealis also regarded borecene as having better physical characteristics than its ZN material and saw no obvious need at this stage to carry out creep rupture testing.

In March of 1993 Mr Bob Wood, who was then employed by Neste and who later became Borealis UK’s Sales Manager, countersigned an internal customer credit request for credit of £1,000 in relation to an estimated supply of polyethylene to Balmoral from Neste’s Stenungsund plant. By then Mr Wood had been to Balmoral and met Dr Milne.

In the second half of 1994 Mr David Cartwright, who was then a Borealis UK sales representative, visited Balmoral in Aberdeen where he spoke to two individuals, one of whom was probably Mr Alexander Hay, Balmoral’s Production and Development Manager (Rotomoulding), and explained Borealis’ then product range (which did not include borecene). On 14^th October 1994 a product designer at Balmoral sent a fax to Mr Michael Shorter, who had been Neste’s, and was now Borealis’, main technical advisor on rotomoulding products, in which he said that Balmoral were currently looking for a grade of rotationally moulded polyethylene to store polyester resin and asked him to recommend a resistant form of polyethylene. A manuscript notation on the fax shows that Mr Wood told Mr Shorter not to respond. (Footnote: ¹²)

Mr Cartwright was, however, interested in doing business with Balmoral and, having spoken with Mr Hay, he applied on 27^th October 2004 for credit approval, which was granted to the extent of £150,000 (Footnote: ¹³).

On 15th December 1994 Balmoral ordered 22 tonnes of a material called NCPE 8089, 2 tonnes to be delivered in December and the remaining 20 tonnes in January 1995. Balmoral’s purchase order contained no reference to any Balmoral conditions and had no conditions on the back. No copy of the invoice or invoices has survived but I accept Borealis’ evidence that Borealis would invoice all deliveries and that each of Borealis’ invoices set out Borealis’ standard terms and conditions either on the reverse or on a following page (Footnote: ¹⁴). This is consistent with the fact that all relevant Borealis invoices that have survived do this and with the inherent probabilities. Those terms and conditions exclude any warranty of fitness for purpose and contain a limitation of liability to the purchase price of the goods.

On 18^th January 1995, Ms Audrey Smith, a customer service co-ordinator of Borealis, sent a fax to Mr John Forbes, Balmoral’s Group purchasing manager, which read:

“Further to your meeting with David Cartwright, we wish to confirm that the price of NCPE 8089 to Balmoral Group will be fixed at 1160 per tonne effective for deliveries from January to June 1995 inclusive. This price is subject to off take of one load (19,8ts) max per month, full loads only.

Prices are quoted subject to normal terms and to current conditions of sale. Should you have any queries, please contact us.”

The meeting between Mr Cartwright and Mr Forbes had probably taken place on 12^th January.

In February 1995 Balmoral ordered 60 tonnes of NCPE 8089 on a call-off basis. This material was delivered over the next few months in three tranches, although only two order acknowledgments and no invoices have survived.

The terms and conditions on the back of Borealis UK’s invoices stated as follows:

On 28^th February 2005 Mr Cartwright visited Balmoral and discussed future business. No mention was made of Balmoral’s (or Borealis)’ terms.

In August and November 1995 Borealis quoted prices for three grades of material (ME 8150 – natural, ME 8153 – black, and ME 8125 – dark green) as from 1^st September 1995 and for the first quarter of 1996 respectively.

On 6^th December 1995 Balmoral faxed purchase order number R 745 for 500 kilos of ME 8152 - natural. The purchase order contains in typescript on the bottom left hand corner the words:

“BALMORAL GROUP LTD.

TERMS & CONDITIONS APPLY

AVAILABLE ON REQUEST”

This was the first time that any reference to terms and conditions had appeared on a Balmoral purchase order sent to Borealis. The purchase order was expressed to be a “Confirmation of telecom J. Ball/A. Smith”. There is no evidence, and it was not suggested, that there had been any reference to any Balmoral terms in that telephone conversation and I am sure that there was none. Nor was any reference to Balmoral’s terms raised in any discussion with Mr Cartwright (who was home based and not a recipient of purchase orders which went to Borealis’ Wilmslow Sales Office). Had it been raised he would have told Balmoral that Borealis would not agree to a change from its terms of supply.

Both then and thereafter Balmoral’s purchase orders did not have any terms and conditions on the back of, or accompanying, the order. Balmoral did not supply Borealis with a copy of those conditions. Nor did Borealis ask for a copy. The suggestion that they were relevant was first raised on the pleadings in December 2005.

If Borealis had asked for Balmoral’s terms and conditions of purchase they would have discovered that they included the following:

“2.0

BASIS OF PURCHASE

2.1.

The purchase order constitutes an offer by the Buyer to purchase Goods and/or acquire the services subject to these conditions.

2.2.

These conditions shall apply to the Contract to the exclusion of any other terms and conditions on which any quotation has been given to the Buyer or subject to which the order is accepted or purported to be accepted by the Seller.

2.3.

The purchase order will lapse unless unconditionally accepted by the Seller in Writing within 14 days of its date.

2.4.

No variation to the purchase order or these conditions shall be binding unless agreed in Writing between the authorised representative of the Buyer and the Seller.”

3.0

SPECIFICATIONS

3.1.

The quantity, quality and description of the Goods and the Service shall, subject as provided in these conditions, be as specified in the purchase order and/or in any applicable Specification supplied by the Buyer to the Seller or agreed in Writing by the Buyer

….

8.0

CONDITIONS AND LIABILTY

8.1.

It shall be a material condition of the contract and the Seller shall warrant that the Goods:

8.1.1.

will be of merchantable quality and fit for the purpose held out by the Seller or made known to the Seller in Writing at the time the purchase order is placed

8.1.2.

will be free from defects in design, material and workmanship;

8.1.3.

will correspond with any relevant Specification or sample ….

On 24^th January 1996 Mr Cartwright introduced Mr Mark Caygill, his successor, to Mr Len Belsher, Balmoral’s General Manager, Moulding. The Balmoral representatives explained the blend of ZN material that they were buying from their current principal supplier DSM. Mr Caygill’s note records that “It was apparent that Balmoral are looking for the cheapest possible [product] achieving “acceptable” mouldings”. Mr Joyce, who was Sales & Marketing Manager of Balmoral Mouldings from January 1994 and also acting General Manager from 1996, after the departure of Mr Belsher, confirmed in evidence that this was so. DSM’s current “obviously very competitive” price was £800/mt. Mr Belsher asked Borealis for prices for its product range.

On 19^th February 1996 Mr D.E. Boersch, a rotational moulding consultant retained by Balmoral in connection with a “plant optimisation project”, faxed a message to Dr Milne about a forthcoming meeting, in which he observed

“The next topic for that planed (sic) meeting is the cooling part of the overall cycle. You’ve got very fine state of the art equipment, but the usage of all the possibilities built in is very poor in Balmoral. Here a lot of improvement potential is available”.

Further purchase orders were given by Balmoral in February (250 kilos ME 8131 – black) and August (1000 kilos ME 8133/8089). They contained the same reference to Balmoral’s terms save that in the latter case there were added the words “**Invoices not quoting valid order no shall be returned**”. Borealis’ invoices continued to set out Borealis’ terms on the reverse.

In September 1996 Mr Robin Tanner, who had been the General Manager of a firm of rotomoulders called Tank & Drum Industries Ltd (“T & D”) moved to Balmoral where he became Managing Director of Balmoral Mouldings (Footnote: ¹⁵). At this time the division was in a bad financial state. It had lost over £1,000,000 in 1995 and was to lose about £800,000 in 1996. Mr Tanner regarded saving on raw material as “essential”. One method of doing so was for Balmoral to cease purchasing DSM’s masterbatch and powder mix and, instead, to colour the polyethylene powder itself by dry blending it with pure pigment. For that purpose Balmoral purchased a turboblender, which it first used in about April 1997.

In October 1996 DSM indicated that it was going to cease production of the ZN material that Balmoral had been using. By a letter of 25^th October 1996 it confirmed to Balmoral that it would maintain current price levels until 31^st March 1997 but that it would be “tight on supply” and:

“that you should with some degree of urgency develop an alternative source of supply so that you can pursue the dry blend route with a view to reducing cost”.

In October 1996, following various tests and trials, Borealis launched borecene, to much interest, at a conference in Vienna of the Association of Rotomoulders (“ARM”) as a new generation of polyethylene product for rotomoulding.

Mr Tanner, whose former firm T & D had been using Borealis ZN material ME 8154 and had trialled borecene, and who had good relationships with Mr Shorter and Mr Andre van Uffelt of Borealis, contacted Borealis and invited them to visit Balmoral. This led to a meeting between Mr Tanner and those two in October and, then, on 2^nd December 1996, Mr Van Uffelt and Mr Mark Caygill of Borealis visited Balmoral in order to introduce Balmoral to borecene. They met Dr Bob Oram, Balmoral’s Group Technical Director, and Mr Tanner.

Mr Tanner’s evidence was that borecene was presented as an upgraded version of Borealis’ ZN material or, at any rate, that that is what he assumed. I prefer in this respect Mr Caygill’s evidence that it was described as a wholly new material, a new generation of polyethylene, based on a new metallocene catalyst technology. That is what borecene was, and was being marketed as generally (see the brochure referred to in paragraph 47 below) and it seems to me inherently likely that Mr Caygill said so.

As Balmoral’s notes record, Balmoral was told that borecene had a number of potential advantages over Borealis’ existing grades of polyethylene:

The trial batch was delivered on 13^th December 1996. Balmoral started some trials using it without Borealis personnel taking part. By February 1997, Mr Tanner was very keen, because of DSM’s position about supply of material, on setting up a trial of borecene with Borealis. In the event the joint test did not take place until 10^th April because Mr Shorter was not available.

Mr Caygill of Borealis visited Balmoral again on 12^th March 1997 when he

At some stage in 1997 Borealis produced a glossy brochure entitled “Borecene polyethylene for rotational moulding” which specified the several advantages of borecene (reduced cycle time, warpage control, improved flow properties/faster sintering, broader processing window, improved mechanical properties and weight reduction potential). Under the heading “Borealis new catalyst – the key to success” the brochure stated that:

“when developing our catalyst for rotomoulding polymers we targeted a polyethylene with improved characteristics as wanted by the rotomoulding industry.”

It then referred to the fact that borecene polymers were:

“characterized by a very narrow molecular weight distribution, an even comonomer distribution and optimized rheological properties such as low zero shear viscosity”.

The brochure indicated that borecene enabled the rotomoulder to reduce drastically the cycle time or the energy cost of the process. Under the heading “Reduced weight” the brochure recorded that an advantage of borecene:

“because of the easy flow properties, is the possibility to reduce the weight of the moulded article, due to the even wall thickness control”.

Under the heading “Improved mechanical properties” the brochure recorded that:

“Although the polymer has a narrow molecular weight distribution it gives improved ESCR results when compared with conventional polymers”.

Beneath this was a bar chart in which borecene with an MFR of 3 and a density of 940 kg/m³ i.e. ME8166/7 (Footnote: ²¹) was compared with standard material with an MFR of 3.4 and shown to be markedly superior in terms of time to failure.

Borealis produced a data sheet for ME 8166/7 which showed

On 7^th April 1997 an internal purchase requisition form for the purchase of 20 tonnes of ME 8167 from Borealis was completed by a Miss Sheena Murray in the Production Department of Balmoral because stocks of DSM’s material were running low. Thereafter an order for 20 tonnes was placed, probably by her (Footnote: ²²), with Borealis over the telephone. On 9^th April 1997 Borealis delivered to Balmoral 7.7 tonnes of borecene ME 8167. Borealis’ invoice for that quantity, dated 9^th April 1997 was stamped as received by Balmoral on 14^th April.

On 10^th April 1997 the joint trial took place. Mark Caygill and Mick Shorter were there on behalf of Borealis. Dr Kevin Walls (Footnote: ²³), Balmoral’s technical manager, who had joined after Mr Hay left, and Mr Charles Ingram, the Quality Assurance manager, were the persons principally concerned on behalf of Balmoral. These representatives of Balmoral were not clear as to exactly what tanks had received Oftec approval. The first topic considered at the meeting was the Oftec oil standard. Mr Shorter’s notes recorded that although Balmoral was producing tanks which aimed to pass the Oftec standard it was very obvious:

“that they did not know what the test was, if specific tanks easily passed or failed, if they passed could they save weight, or if they failed how they failed.”

The trial started with production of specimens of a particular oil tank (BH 3410) using DSM material and a lot of wall thickness data was generated. Then, as the note records:

“On selecting several tanks for study, the wall thicknesses were totally different on all, but above all, they were thick at the top and thin at the bottom – the opposite to what is required”.

In relation to the second topic – borecene - samples of the BH 3410 tank were produced with different charge weights of borecene, and samples were made of a smaller tank – H 1590 - comparing 72 kgs of DSM material with 62 kgs of borecene. Mr Shorter’s note records that:

“The trial was not progressed further, because on studying the tanks and trying to optimise the distribution (Footnote: ²⁴), it was clear that the major effect/cause of a good/bad criteria, was tool mounting and axis of rotation, etc on the machine. It was left that Balmoral have to optimise their machine, before material can become relevant.”

Dr Walls told me the material thicknesses of the bottom of the tank could be increased if the tool mountings on the machine were changed, although this was not the only way of doing that.

As can be seen these trials related to the rotomoulding process and not to the chemical composition or characteristics of the material, and were incomplete. It was left that Dr Walls was to carry out some in house studies into process and revert to Mr Shorter when Balmoral had optimised their machine and he was ready to re-evaluate borecene.

On 15^th April 1997 Balmoral issued its purchase order R 2278 for 20 tonnes of Borecene. The delivery of the 7,700 kgs on 9^th April 1997 appears to have been treated by Balmoral as the first delivery under this order because one of Balmoral’s copies of the order has the notation “9.4.97 - 7700 kg” written in manuscript beneath the typed order and another has that notation and beneath it the notation “7.5.97 - 12,300 kg”. Similar notations appear on copies of the requisition form. By now the words at the bottom left hand corner of Balmoral’s orders read:

“BALMORAL GROUP LTD.

TERMS & CONDITIONS APPLY A.O.R. (Footnote: ²⁵)

TIME IS OF THE ESSENCE

**INVOICES NOT QUOTING VALID

ORDER NO SHALL BE RETURNED**

On 7^th May 1997 Borealis delivered 19,800 kgs of material, of which 12,300 was attributed to purchase order R 2278 of 15^th April which, when added to the 7,700 kgs delivered on 9^th April, produced a total of 20 tonnes: see paragraph 52 above. The invoice of 7^th May 1997 in respect of order R 2278 was, however, for 19, 800 kgs, perhaps because 19, 800 kgs was a full load and the compiler of the invoice did not realise that this order had been the subject of two deliveries - on 9^th April and 7^th May. Someone at Balmoral noted on the invoice that 20,000 kg had been ordered

On 8^th May 1997 Balmoral issued a purchase order for 200 tonnes to be taken on a call off basis. This order must have been preceded by a telephone order because it was in part satisfied by 7,500 kgs of what was delivered on 7^th May.

On 11^th May 1997 Balmoral began to use borecene in the production line and by 21^st May 1997 borecene had become the material used in the production of all of Balmoral’s tank range. From then onwards Balmoral used borecene in various different forms almost exclusively until the middle of 2002. Balmoral was the first oil tank manufacturer in the United Kingdom and Ireland to use borecene in any significant quantities.

Dr Walls did not get back to Mr Shorter to re-evaluate borecene. In the months after April 10^th Mr Walls, who had studied at Queen’s University, Belfast (“QUB”), spent time producing a set of sample tanks with lesser charge weights than had been used for the DSM material for testing by OFTEC. In the course of doing this he optimised processing parameters such as rotation speed and oven and cooling settings, and carried out localised lowering and raising of temperatures at different parts of the mould (by shielding it (Footnote: ²⁶) or applying heat to it). He took extensive measurements by ultrasound. The ultrasound cannot, however, produce reliable data at corners (Footnote: ²⁷) and is not used for that purpose. Sporadically he cut up tanks and measured the wall thicknesses (Footnote: ²⁸). He was “primarily trying to get a more even wall thickness and more material where we wanted it overall” and was looking to avoid both unacceptable thinness and unacceptable thick/thin changes. He managed to obtain acceptable wall thicknesses. The tanks were passed by the Polymer Development Centre, a testing agency in Ireland, which acts on Oftec’s behalf.

In the event Balmoral decided not to reduce its charge weights. When borecene went into production, with the charge weights unchanged, no changes in the processing parameters was made (save that adjustments were made to the cooking time where necessary to make sure that it cured well (Footnote: ²⁹)). Whatever optimised production parameters that had been developed by Dr Walls in order to get the best distribution on lower charge weights were not adopted. Balmoral’s thinking at the time, according to him, was that because they were using the same charge weights and had been told:

“that metallocene was superior to our existing material strength-wise, so it was a no-brainer that if you keep the same wall thickness we should be able to have the same mechanical properties in our products”.

Between March and May 1997 Balmoral and Borealis contemplated entering into an overarching Industrial Supply Agreement. In the event they never did so. The sequence of events was as follows. On 12^th March 1997 when Mr Caygill of Borealis was introduced to Mr Forbes, Balmoral’s Group Purchasing Manager, by Mr Tanner of Balmoral, Mr Forbes had expressed interest in coming to some sort of supply agreement for the group covering both the rotomoulding and pipe coating businesses. He suggested that both sides got together in April. On 22^nd April (as I infer from Balmoral’s letter of 15^th May 1997) Borealis sent Balmoral a set of documents constituting their standard Industrial Supply Agreement. These included (a) the basic supply agreement to which the other provisions were to be exhibits; and provisions on (b) confidentiality; (c) price and rebates; (d) consignment stock; (e) research and development; (f) licensing. Mr Forbes passed the papers to Mr R W Byth, Balmoral Group’s solicitor.

Clause 3.1 of the price and rebate provisions read:

“The parties agree to comply to Borealis payment terms and general terms and conditions of sale for each country where sale takes place.”

Clause 4 of the basic supply agreement provided that:

“4.1.

The Customer shall be obliged to examine the products and to test them properly immediately on receipt, and if defects are ascertained, Borealis shall be notified in writing immediately. Borealis shall make replacement delivery of goods ascertained to be defective and the Customer shall have no other remedy against Borealis.

4.2.

Borealis shall not be responsible for the applicability and suitability of the products and shall only be liable for damages due to product liability in accordance with the provisions in the General Terms and Conditions for Borealis attached as Exhibit 3 to this agreement”..

The Consignment Stock Agreement provided, by clause 11, for the incorporation of the standard terms and conditions of the Seller unless otherwise specifically stipulated in the contract.

On 14^th May Mr Byth, responding to a note from Mr Forbes, returned to him the Borealis drafts with a number of comments. In relation to the price and rebate provisions he said:

“I see your comments and agree them. Have we actually seen the general terms and conditions of sale referred to in 3.1? We certainly need to do so if they are to be picked up in this way.”

In relation to the supply agreement he said, inter alia:

“Article 4.1

Are we expected to test every batch or just how practical is the first sentence of 4.1? Delete the words “and the customer shall have no other remedy against Borealis”.

Have we thought about adding a warranty requirement? At the very least I think we should ask them to add a new article in the following terms: ‘Borealis warrants that the products shall be delivered in first-class condition and conform to specification. Borealis further warrant the products shall be of first-class quality and fit for the purpose for which they are purchased”. I would regard this as being a reasonable warranty to request and stop short of my normally more aggressive position on such matters, knowing that we need to try to get this agreement without too much hassle. If all else fails, I guess we would have them at common law if they provided us with defective material. I am surprised to see that there is no exclusion of consequential losses here and that can only work in our favour so we should not raise the point”.

On 15^th May Mr Forbes of Balmoral wrote to Mr Caygill at Borealis with comments and amendments for consideration. In respect of the price and rebate provisions he required to see a copy of Borealis’ payment terms and general terms. He also sought the deletion of the words “and the customer shall have no other remedy against Borealis” in clause 4.1. He asked for the warranty that Mr Byth had suggested.

As is apparent from the letter of 15^th May Mr Forbes had not seen Borealis’ terms. There is no record that they were ever sent to him by Borealis. They were, however, on the back of every invoice, which Mr Joyce, who became Balmoral Tanks’ Managing Director from August 1997 (replacing Mr Tanner), would initial. Borealis knew that Balmoral were in receipt of those invoices. They did not know that Mr Joyce initialled them, although they could legitimately make the correct assumption that someone looked at them and passed them for payment.

Borealis’ copy of the letter of 15^th May has the manuscript annotations of Mr Robert Wood of Borealis, to whom the letter had been passed by Mr Caygill. He amended the warranty so as to read “Borealis warrants that the products are according to given agreed spec”. Thereafter the negotiations ground to a halt, although, when on December 17^th 1997 Mr Caygill wrote to Mr Forbes to confirm an agreement that had been reached, subject to conditions, for Borealis to give a month’s notice of withdrawal of supply of raw materials, he headed the letter “Re: Industrial Supply Agreement – Notice of Withdrawal”,

No suggestion was made by Balmoral in the communications about an Industrial Supply Agreement that Balmoral’s terms already applied.

What happened thereafter was that the parties met regularly to negotiate prices and volumes to be supplied and over what period those prices would apply (whether six monthly or quarterly).

Polyethylene is produced in either pellet or powdered form. The relevant Borealis product codes are as follows:

Balmoral purchased a variety of types of borecene from Borealis which was mixed with pigment in different ways. In broad terms the borecene material (Footnote: ³⁰) purchased from Borealis and the accompanying pigment was as follows:

During 1997 and 1998 Balmoral experienced problems with warping, where the shape of the tank distorts during the cooling process, usually due to differential cooling across the wall thickness of the polymer (Footnote: ³⁴), and blooming, where white patches appear on the outer surface of the tank. Blooming is caused by microscopic surface tearing of the polymer on the outer section of the tank, when the polyethylene crystallises and tears away from the tool surface, leaving a residue of microscopic pieces of still soft polymer on the mould. It is extremely common in the case of dark materials, whether metallocene or ZN, and is usually dealt with by flaming, i.e. applying heat to the surface of the tank, so as to produce an even appearance. Warping is not uncommon. Neither of these phenomena form any part of Balmoral’s present claim against Borealis (Footnote: ³⁵). Balmoral relies on them because it raised a claim against Borealis for which Borealis paid it compensation without referring to, or seeking to rely on, its terms and conditions. The two problems appear in the end to have been resolved, even if not completely, although it is not entirely clear to me what it was that did so.

On 14^th August 1997 David Cartwright and Anders Sandaas of Borealis visited Balmoral to discuss Balmoral’s problems with “pre-release” (where the product comes away from the mould earlier than it should and, therefore, warps) and blooming. Prior to then Mr Shorter had advised Balmoral to change from a solvent to a water based release agent in order to reduce blooming. A release agent is a coating applied to the metal mould to assist release. A water based agent would lead to a slower release and, therefore, allow the plastic moulding to stay in contact with the mould wall for as long as possible. Parting of the moulding from the mould wall reduces the control of cooling. Balmoral had changed their pre-release agent, which had reduced the risk of blooming but increased the amount of pre-release and consequent warpage. Mr Sandaas, who questioned the need to change pre-release agent, suggested that the blooming effect might be eliminated and warpage reduced if the cooling cycle was slowed down and the rate of crystallization thus reduced. This did not produce an improvement in blooming although it helped somewhat with warpage. Mr Sandaas suggested the use of a non nucleating pigment (Footnote: ³⁶), dispersal of the pigment through the material in a different manner, and working with ICO to find the best combination of pigment percentage, pigment formulation, resin density, resin crystallization temperature and processing conditions. A nucleating pigment is one that tends to initiate crystallisation, so that crystallisation will take place at a higher temperature. This may cause stresses over the structure or lead to embrittlement of an otherwise ductile material.

On 27^th August Mr Joyce complained strongly to Borealis that Balmoral was being used as a guinea pig (Footnote: ³⁷). A further fax of complaint followed on 9^th September. On 15^th September 1997 Mr Sandaas of Borealis provided Balmoral with a long report. He again recommended the use of a non-nucleating pigment. He recommended demoulding earlier (Footnote: ³⁸) and reverting to a solvent based release agent. This led to the use of the tecroblend masterbatch referred to in paragraph 70 above.

Borealis, in the persons of Mr Bob Wood and Mr Shorter, visited Balmoral at the end of September. On 29^th September Mr Joyce supplied Borealis with the costings associated with obtaining approval for Balmoral’s range of products (£55,280), said that scrapped material associated with eliminating problems had cost in excess of £18,000, to which was to be added loss of business and market positioning, together with the cost of flaming all products to deal with blooming.

In September 1997 Borealis delivered a batch of material – R 00107 - which was faulty, having an MFR of either 2.16 or 2.6, and which, Balmoral claimed, led to a three day loss of output.

After various visits and correspondence Balmoral’s claims were eventually settled. On February 10^th 1998 Mr Wood wrote saying:

“As we have not received from Balmoral a sample of the green blended material ME 8167, batch R 000107, for technical analysis, we have decided to settle the claim on a commercial basis”

The letter proposed a payment of £85,000 to cover (i) Oftec/BBA testing costs - £25,000; (ii) 3 days lost production on account of batch R 00107 - £45,000; and (iii) credit for returned material £15,000 (Mr Joyce described this as relating to warpage and blooming). On February 20^th Balmoral’s offer was increased to £90,000 by increasing item (ii) to £50,000. The claim was settled at that sum, credited against future purchases. Neither Borealis nor Balmoral said anything in the course of the negotiations about anybody’s standard terms.

Further product claims were settled in a similar way. In August 1999 Borealis supplied ICO with an allegedly defective batch of material which caused variable wall thickness in the product. Balmoral blamed ICO. ICO blamed Borealis. Borealis could find nothing wrong with the retained sample of powder and complained that Balmoral had not notified them of the problem in sufficient time to allow them to investigate and had retained no samples for testing. The dispute was settled in April 2000 with a three way split in respect of the claim of £17,880.

In 2001 Balmoral received a defective batch of borecene due to the presence of the wrong form of stearate, as a result of which numerous tanks had to be pre-released from the moulds and scrapped. Borealis, through its insurers, agreed to pay £170,000 by way of damages together with a credit note for the value of the material supplied. The claim for loss of profits on lost sales was left open because of a potential overlap with this action. Again neither party relied on any standard terms.

In late 1998 Borealis introduced Balmoral to the new grade of borecene with an MFR of 6 (for which Balmoral later secured Oftec approval (Footnote: ³⁹)). On 16^th November 1998 Mr Halvorsen and Mr Wood of Borealis visited Balmoral. It was probably then that Mr Halvorsen handed Balmoral Borealis’ Processing Guide.

According to Mr Macleod (Footnote: ⁴⁰) of Balmoral’s note, the Borealis representatives told Balmoral that Borealis had developed a new grade (ME 8161) with an MFR of 6, to alleviate problems with powder flow and hopefully ‘blooming’, and that this higher melt flow material would improve material distribution within the mould, and hence the overall wall thickness of the products (Footnote: ⁴¹), and could improve cycle time by up to 30%. All other physical properties were the same as MFR 3. Mr Wood recalled this being said, as it probably was, by reference to provisional data sheets. The data sheets in bundle 32 show that, with the exception of the MFR, the physical properties were said to be the same. The ESCR figure in those data sheets was 4. In fact the ESCR of MFR 6 was less than the ESCR of MFR 3, albeit within the same No 4 category used by Borealis.

A series of tests were carried out trialling three different materials on the H 1590 tool: (i) Pegasus B 408, (ii) Borealis ME 8155 (both of which are ZN materials) and (iii) Borealis ME 8161 i.e. borecene with MFR 6. The same moulding parameters were applied to the three materials. The rotation ratio was 4:1. Blooming appeared on the surface of all three products, more prominently on ME 8161. Wall thickness measurements were taken from spots on the walls of the stiffening ribs. The ME 8161 material gave the lowest thickness (on average) for all sections measured. Borealis explained that material with a higher melt flow required higher rotational speeds than normal in order to distribute product more evenly throughout the product because at low speed more material would be left in pockets and flow away from peak areas because of the product’s low viscosity.

Further trials took place on the mould for the SL 1250 and LP 1250 tanks, because the H 1590 tank mould had sustained some damage, and products using ME 8167 (i.e. borecene MFR 3) were also made. Rotation speeds were increased from 4:1 to 7:1. The results showed an all round improvement in wall thickness of approximately 1 mm using the ME 8161 (MFR 6) compared with the ME 8167 (MFR 3). These results followed and may reasonably be attributed to a change in rotation speed, as Borealis had recommended.

The conclusions drawn from the test included the following:

Professor Crawford points out that, although the fault addressed was blooming and not wall thickness distribution, these results show (a) that different wall thickness distributions were obtained for each plastic when the same moulding conditions were used in the same mould and (b) that, if the manufacturing conditions were altered then the wall thickness distributions could be changed.

On this and subsequent visits in February 1999, February and March 2000, and November 2001 Mr Halvorsen observed what he regarded, and describes in his statement as, inappropriate processing practices by Borealis operatives such as (a) failing to make proper use of the appropriate release agent by following the instructions on the can; (b) removing tanks whose upper parts had stuck suspended in the mould with a “crowbar”; (c) reheating the outside of the tank mould in order to melt the plastic and so loosen it (Footnote: ⁴³); (d) want of familiarity with the Borealis processing guide; and (e) a lack of interest on the part of some Balmoral staff in solving problems. He also found that Balmoral operatives would, whilst he was there, try to follow what he was explaining and seek to optimise the machine settings so as to avoid warpage, but, when he returned, he would find that much of what he had explained had been sidelined and old practices restored. He explained to Balmoral operators that if borecene was processed in the same way as ZN it would not achieve the distribution that was necessary so that issues such as temperature, speed and ratio of rotation, and the cooling cycle were very important, detailed discussion of which was not particularly welcome to Balmoral machine operators.

In at least one of these criticisms Mr Halvorsen was mistaken. What he described as a “crowbar” was in fact a form of paddle often used by rotomoulders for removing tanks from moulds. Mr Halvorsen did not, as it turns out, intend to refer to a crowbar in the generally accepted sense of the term, but the instrument that he saw was not one with which he was familiar. The generality of this evidence renders detailed refutation or assessment difficult. But I found Mr Halvorsen, who left Borealis in 2005 and now works for a Borealis distributor, a reliable witness, and I regard his general criticism as probably well founded. Mr Wood’s evidence was of the same tenor.

On 3^rd December 1998 Mr Halvorsen sent a fax to Mr Taylor and Mr Macleod at Balmoral recording the result of the trials in November and expressing the following Conclusions:

“Conclusions

We see potential for time reduction, approx 30%.

The blooming we have more to look into. I think this is a combination between the pigment and the release agent. I have tried the same release agent here in Norway, and I cannot see any blooming. For this kind of release agent, it is very important to add it properly.

You have to follow the instructions on the can.”

Further tests were carried out on 16^th February 1999 at Balmoral with a Rotolog. These showed that if MFR 6 material was used there could be a 31% reduction in cycle times and that careful selection of a release agent with optimum cycle times and a reduced cooling rate could reduce the blooming effect or might eliminate it. Fast cooling causes the product to pluck itself away from the surface of the mould resulting in blooming. A reduction in cycle time of 21% was also experienced in relation to MFR 3 material. Mr Halvorsen’s record of the meeting recorded that the most important thing was that the customer had to be interested in solving the problems, and described the trials as badly arranged, and the Production Manager as seemingly not interested in what was going on

In May 1999 Borealis agreed to supply 4 tonnes of MFR 6 material free of charge to enable Balmoral to obtain Oftec and other approvals. In June 1999, Balmoral requested 2 tonnes of MFR6 for product trials. In July Borealis agreed to supply 10 tonnes of ME 8160 free of charge to cover the cost of OFTEC testing. In August Balmoral told Borealis that it needed 19 tonnes free supply in order to cover WRC, OFTEC and BBA testing costs, and Borealis agreed that it would meet the cost of re-approving Borealis MFR 6 grade by supplying material free of charge. By August 1999 Borealis had told Balmoral that it was going to cease production of MFR 3 and would only be supplying MFR 6 in future. In December 1999 Borealis ceased to supply Balmoral with MFR 3 and began supplying MFR 6.

The Borecene processing guide indicated that borecene would need an increase in oven temperature and rotation speed and an optimisation of rotation ratio in order to achieve the desired even wall thickness. It observed:

“Optimising of processing conditions for Borecene is very dependent on type of rotomoulding machine used, machine settings, mould, mould material etc. Therefore only these general guidelines are given. If more detailed advice are (sic) requested contact your Borealis Technical Service representative.”

It also contained the following statements in a section headed “Colouring of Borecene”:

“When changing from a conventional material to a Borecene, the same pigments can be used in the same quantity….

It is recommended that the quality of the Borecene product is initially monitored to ensure correct pigmentation. Details regarding pigments can be supplied by the pigment manufacturer.”

This general statement was based, according to Miss Fatnes, on some testing especially for impact properties, on some pigments for specific customers and on the experience that Borealis’ technical service people had in the market. Borealis had not done testing specific to the oil market. I do not regard it as established that Borealis had made the same claim in respect of pigmentation when originally introducing borecene to Balmoral. At the same time Borealis had not indicated that Balmoral would or might have to change its pigment.

In November 1998 Mr Wood of Borealis sent to Mr Joyce of Balmoral a letter on Borealis U.K. paper with an enclosed packet. The letter informed him that Borealis was “preparing to implement some important changes in our business practices in order to service you better”. The letter indicated that the packet contained information about the following:

The enclosed packet included a circular letter which read as follows:

“Dear Business Partner

Re: Shared Service Centre & Direct Invoicing

Continuously striving to be a top performer in our processes to serve customers in the most original way, Borealis has decided to establish a Shared Service Centre (SSC) for managing the transaction processing of a major part of our financial administration. The new concept will go into effect on January 1 1999.

The SSC will be located in Woluwe, Belgium under the legal umbrella of the existing Borealis Coordination Center (BCC).

We are convinced that by centralising financial tasks, Borealis’ personnel will be able to focus even more on core activities like production, logistics and sales and thus respond faster to your needs.

For the same reasons Borealis has decided to optimise its current invoicing process. This means that there will be a change from the re-invoicing principle through the local sales company to direct invoicing by the delivering production unit. Settlements, however, will be done via the local BCC account as described in the attachment. As before, all the specific payment instructions will continue to be featured on our invoices.

……..

Because of the integration of PCD (Footnote: ⁴⁴) and the direct invoicing from our production units in different countries, we recognise the need for harmonising our General Terms and Conditions of Sale, You will find the revised document attached. Irrespective of the source of delivery, you will deal with a commercial partner who applies unified conditions with respect to commercial, insurance and legal maters. We are confident that you will find the harmonised terms to be even more customer oriented and simplified.

If you have any questions or require further information, please do not hesitate to contact (name and contact information of your local credit controller (Footnote: ⁴⁵)). Thank you”

A further letter dated November 1998 contained the following paragraph:

“Borealis sales invoices dated 01.01.99 and onwards are factored to Borealis Coordination N.V., which will be the sole legal beneficiary of all collections.

We kindly request you to follow the payment instructions on our sales invoices and quote invoice numbers as reference in all payments and enquiries.

In order to provide you with continuous and further improved services we recommend and promote standard bank transfers as the primary payment method..”

One of the attachments read, so far as relevant, as follows

“Borealis Coordination Center N.V.

Bank Transfers

Country Currency Bank Address Account number

UK GBP Citibank N.A. London, U.K. 8543992

Cheques and remittance advices with full sales invoice numbers to be mailed to:

Borealis Coordination Center N.V.

c/o Citibank European Lockbox Service

P.O. Box 17663

London SE1 4ZT

U.K.”

Also attached was the following:

There was also attached (a) a form of acknowledgement of receipt, which Balmoral did not return, (b) Borealis’ General Terms and Conditions and (c) letters about Borealis and (i) the euro and (ii) Year 2000 compliance. The General Terms begin with the following preamble:

“These General Terms and Conditions of Sale … together with the order confirmation shall in so far as no other agreement regulating the issue has been concluded in writing apply to all products … sold by the Borealis entity given in the front page (in the following “Borealis”) ..”

No sample front page of an invoice was included.

Mr Joyce expressed some concern to Mr Wood (either on the telephone or face to face) as to whether the changes would mean increased cost to Balmoral in some way.

On 11^th December 1998 Mr Wood wrote to Mr Joyce, again on Borealis U.K. paper. The letter was in a standard form sent to many Borealis customers. It included the following:

“Dear Allan

Re: Shared Service Centre

Our recent mailing on the euro into our business practices (sic), the Year 2000 challenge and the establishment of a Shared Service Centre (SSC), have raised, especially for the latter, some concerns and remarks from your side.

With the creation of a SSC, it is definitely not our aim to transfer costs, currently paid by Borealis, to you. Our customer service is always our first priority and we would therefore like to clarify the points of concern raised.

*Central invoicing: Your local Borealis Sales Office will print and dispatch the invoices in the name of the supplying Borealis plant and continue to offer assistance on any issues relating to payments. Invoicing from different sites instead of invoicing through one local sales company necessitates your maintaining different supplier files. The creation of one bank account reference for all the Borealis sites, will, however, soften the additional administrative effort. The planned second stage of the project will enable central invoicing, independent of local Sales Offices, thereby reducing the temporary complexity introduced by the first phase, and establishing one invoicing and collecting entity.

…..

We hope that this letter answers your questions and concerns, if you require any additional information, please do not hesitate to contact us.”

Up until the end of 1998 Borealis’ invoices contained no reference to any company other than Borealis UK. They were headed “Borealis” and contained at the base Borealis UK’s name, address, telephone and fax numbers, company and VAT registration numbers, bank account details and other information.

In respect of deliveries from January 1999 onwards the format of the Borealis invoices was changed in the following respects:

From January 1999 onwards a number of changes took place in the way in which Balmoral ordered material from Borealis and paid for it. After the first few orders Balmoral ceased to address its orders to Borealis UK and, instead, addressed them to Borealis Coordination NV but continued to send them to Borealis UK (Footnote: ⁴⁷) . Balmoral ceased making payment to Borealis UK and made payment, by cheques in sterling, to Citibank in London into Borealis Coordination Center’s “lockbox” account. VAT was only charged on invoices in which the “Place of Dispatch” specified in the invoice was Great Britain.

Prior to January 1999 producing companies such as Borealis Norway and Borealis Sweden (Footnote: ⁴⁸) had charged local Borealis sales companies, such as Borealis UK, for all deliveries to the territory of the local company, the product manufactured being sold by the producer to those local companies. After that sale Borealis UK would “re-invoice” their customer, to whom without mark-up they on sold. This arrangement meant that the local sales companies had separate balance sheets and separate administrative and accounting departments. Borealis came to regard this as less efficient than it could be.

From the beginning of 1999 the Borealis companies ceased to sell to the local sales company and, so far as they were concerned, sold products to the customers themselves. Borealis Coordination Centre NV, a Belgian company (“Borealis Belgium”), purchased the receivables from the production company and collected them (Footnote: ⁴⁹), as well as performing various accounting functions for the production companies. The invoices were generated over night in the UK Office in batches by the producing unit (by computer link) and then sent out by Borealis UK. The numbers of staff in the local sales companies decreased so as to consist only of sales managers and customer service representatives, administration being transferred to Borealis Belgium or the production unit.

On 14^th February 2000 Mr Wood of Borealis, who had had discussions on the rebate to be extended to Balmoral if the volume of material ordered by it exceeded a certain level, sent to Mr Joyce a copy of a written agreement, signed by Mr Webster on behalf of Borealis Denmark representing all its subsidiaries, headed “Bonus Agreement 2000”, with a request that Mr Joyce sign and take a copy of it and return the original. The agreement covered deliveries for the year 2000. It provided that, if during 2000 Balmoral ordered 2,000 tonnes or more of rotational moulding resin in that year, Borealis would pay a rebate of £20 per tonne on the entirety of the quantity ordered. Clause 5 provided:

“The parties agree to comply to Borealis payment terms and general conditions of sale”

No copy of the letter signed by Mr Joyce has survived (Footnote: ⁵⁰). But Mr Joyce told me that it was his usual practice to return documents presented to him for signature and, although he did not recall signing it, did not really dispute that he probably did. In the light of that evidence and the fact (a) that it would be in the interests of both parties to have the agreement signed, and (b) that a similar agreement was signed on behalf of Balmoral in 2001, I think the likelihood is that Mr Joyce did sign and return it. Balmoral ordered more than 2,000 mt in 2000 and was paid the agreed rebate.

On 3^rd November 2000 Mr Wood of Borealis sent to Mr Joyce of Balmoral a letter describing a further change of practice. It included the following:

“Borealis is committed to continuously improving Borealis’ customer service. This is why we are establishing a single invoicing entity. Effective January 1 2001, all material delivered by Borealis, irrespective of production site, will be invoiced from Borealis A/S, the Danish parent company.

…… In the future you will deal with only one supplier.

Our commercial, day-to-day contact is not affected by this change in invoicing….”

This letter had attached to it a package of documents of which only a draft version has survived . One of those documents informed Balmoral that

In January 2001 the first three Borealis invoices contained underneath details of the total sum due the following:

“Exporter: Borealis AS, Ronningen

This document is solely payable to Borealis A/S, Lyngby Hovedgade 96, DK-2800 Kongens Lyngby, Denmark via Banktransfer to the account 40092550 at ABN AMRO (London) (swift code ABNAGB2L) for beneficiary Borealis A/S.

Cheque: Cheque to be sent to Borealis A/S, Hovedgade 96, DK-2800 Kongens Lyngby, Denmark.

Remittance advices are to be sent to our Borealis Coordination Center, Woluwedal 26, B-1932 St Stevens-Woluwe Belgium or faxed to + 32 2 715 0459.”

The goods were described in the bottom left hand corner as invoiced by Borealis Denmark and Borealis UK was described in the bottom right hand corner as Commercial Agent.

From invoice 9800006198 of 5^th January 2001 the words on the invoice after the price were

Every Borealis invoice consisted of three pages. The General Terms and Conditions of Sale, which were not changed, were set out in full on the face of the third page.

From 1^st January 2001 the invoices were no longer printed in England or posted by Borealis UK. They were despatched from Denmark by courier. Balmoral did not immediately start making payment to the Amro account; and had to be reminded to do so in October 2001.

On January 17^th 2001 Mr Joyce and Mr Forbes met with Mr Wood and agreed rebate terms for 2001 of £20 per m.t. on all tonnage purchased if 2000 m.t. or more were ordered, and £30 if 2500 m.t. or more were ordered. A typed rebate agreement for 2001 in the same form as the 2000 agreement, save for the change of year, and dated 17^th January 2001, had been prepared on or before that day. On that day Mr Wood amended the typed agreement in manuscript so as to provide, in addition, for a £30 rebate if 2500 m.t. or more was ordered. Mr Wood initialled and dated that amendment. Mr Forbes signed the amended document. He did not date his signature in manuscript but he signed immediately below the date of 17^th January. I infer that he did so on that day. On 14^th February Mr Wood faxed through to Mr Joyce and Mr Forbes a note, dated 12^th February 2001, confirming , inter alia, these rebate terms, together with a fully typed up rebate agreement for 2001, signed by Mr Webster on behalf of Borealis. In an email sent to Mr Joyce on 12^th February, to which the fax was an attachment, Mr Wood had indicated that he would be sending the fax, signed, when he was back in the office “together with the rebate forms for your signature”. There is no evidence that the fully typed up agreement was ever signed by Balmoral.

Towards the end of 2001 Balmoral came to realise that oil tanks manufactured with Borecene were failing (by cracking) at an alarming rate. In an internal memorandum of 2^nd November 2001 Mr Woolley of Balmoral highlighted “a worrying trend in the number of tanks cracking since the introduction of [borecene]”, and that the number of complaints in respect of two of the tanks (H 2500 and SL 1250) was “worryingly high, more so when considering the young age of many of those failing”. The latter included tanks made from MFR 6 borecene to which Balmoral had only recently changed. Borealis’ immediate reaction, as expressed in an e-mail from Mr Halvorsen of 14^th November, was that the reason for the cracks was that the material was:

“mixed with a nucleating pigment creating differences between the particles – so called transcrystallinity (Footnote: ⁵¹) – creating stress and weak points in the product”.

He recommended either changing to a non-nucleating pigment or, preferably, to producing tanks only from compounded green material.

Balmoral had had failures of tanks when it used ZN materials. The underlying failure rate over a ten year period was 2.96% of tanks manufactured. The rate of failure with borecene was much higher.

On 22^nd November 2001 a meeting took place between Balmoral personnel and Mr Halvorsen at which he suggested that the cause of failure might be the combination of a grind blend mix together with a nucleating pigment, causing stress points due to the formation of transcrystallinity, a concept that, according to Balmoral’s note of the meeting, he “could not fully explain”.

On 29^th November 2001 a meeting took place between representatives of Balmoral, Borealis and ICO, the pigment suppliers/incorporators, at which Borealis claimed that no other tank manufacturer using Borecene had experienced cracking problems except for one very small one. Whether or not that was true at the time, it has become apparent that Borealis has received claims from most British tank manufacturers of any size. At this meeting Borealis advised Balmoral that they should move to a further version of borecene, with an MFR of 4, and that it should be fully compounded with the pigment. Balmoral followed this advice and on 17^th December 2001 began production with fully compounded MFR 4 material, which, by arrangement with ICO, whose David Howarth was at the meeting, they obtained for the same overall price as a grind blended product. By January 2002 Balmoral was experiencing difficulties with MFR 4.

In November 2001 Balmoral approached Rapra Technology Ltd (“RAPRA”) for advice. It gave definite instructions to them in March 2002 and one of their representatives gave evidence at the hearing.

By February 2002, if not before, Balmoral had become thoroughly dissatisfied with what they regarded as Borealis’ lack of any sense of the urgency or of the importance of the problem that Balmoral faced; and regarded Borealis as bound to underwrite the cost of complaints caused by the failure of tanks made by Borecene in the 10 year warranty period. Balmoral decided to switch from borecene for all non-vertical tanks and to raise a claim against Borealis for tank failures. At a meeting between Balmoral and Borealis on 29^th February Mr Halvorsen suggested that the cracking might be attributable to problems of “sharp geometry”, the application of the release agent or aggressive demoulding.

On 12^th March 2002 at a meeting at which Messrs Joyce, Forbes, Wood and Halvorsen were present, the parties agreed to continue the 2001 rebate agreement into 2002.

Balmoral implemented a change back to grind blend in April 2002 (at Borealis’ suggestion) in order to deal with a tendency of the material to stick in the moulds. But problems continued and Balmoral ceased to use Borecene altogether on 29^th July 2002. Thereafter the polymer that they used was, largely, compounded BP Rigidex HD 4330 until October 2003 and then Dowlex NG 2432 (Footnote: ⁵²), both ZN materials with MFRs of 3 and 3.5.

On 23^rd April 2002 Mr Shorter, who, having been an employee of both Borealis and ICO (Footnote: ⁵³), had then set himself up as a consultant, attended Balmoral’s premises to express a view on Balmoral’s production methods and to draw up materials requirements specifications for future material purchases. This turned out to be a bad day to call because only the day before Balmoral’s production manager, Alan Taylor, had died in a house fire, and Mr Shorter looked round without his assistance. During the course of his visit he talked to a machine operator and had a meeting with the Group Technical Department.

On 30^th April Mr Shorter e-mailed to Balmoral a series of comments which he asked them not to take as points of criticism but on the basis that an independent pair of eyes might see things which Balmoral could not, recognising that Mr Taylor would have been on to some of his points. His report included the following:

“Process parameters. These are set by experience and some go back to Alex Haigh (sic) days. Rotation requirements is (sic) “a black art”, No set procedures for part manufacture. Some parts are generated with 2 oven temperatures, some with 1. Precooling and cooling operation is not optimised. It would appear also that some operators “modify” the process parameters to suit their requirements.

….

Mould design and installation. Nearly all moulds are mounted very close to the arm! This will have effect on temperature distribution around the mould, and hence wall thickness variation…

….

Moulding quality – get feeling that operators dont (sic) take ownership for the quality that they generate.

QA studies. No equipment to check incoming resin or powder, or moulding parameters. (Footnote: ⁵⁴)”

Balmoral was surprised by Mr Shorter's report which did not appear to it to be consistent with how Mr Shorter had portrayed the situation when at ICO. Mr Woolley’s evidence was that the products were mounted on the arms as per the machine design with the base of the tanks connected to the “plate”. The way of mounting was not changed and the only place where heat could have been minimised by virtue of the shielding effect of the arm mounting would be the bases, and there was no problem around the base (Footnote: ⁵⁵). After another visit Mr Shorter was not further engaged, not least because he appeared to be able to offer no useful advice on the cracking issue as he had no experience of cracked tanks.

I recognize that Mr Halvorsen and Mr Shorter will have had a different viewpoint to that of Balmoral. Nevertheless the fact that both of them were critical of Balmoral’s processing practices in the way in which they were gives some indication that Balmoral was not optimising its practices nor applying them uniformly.

On 14^th August 2002 Balmoral withdrew its SL 1250 (old style) (Footnote: ⁵⁶), H 1135, H 1590 and H 2500 designs, save for bunded versions of the 1135 and 1590. These tanks were superceded by the SL 1000, SL 1250 (new style), H 1500 and H 2000 (Footnote: ⁵⁷). The inner tanks of the horizontal range and of the new SL 1250 tank had been redesigned in order to accommodate pumps, gauges and ancillary equipment so that the bunded version could take fuel dispensed by a pump rather than by gravity. Balmoral also introduced a new single skin vertical tank - the V 1800. The H 5000 design had been withdrawn in 2001.

As time went by Balmoral came to realise what a disastrous position it was in. Complaints of cracking tanks continued to rise. Distributors began to express worry and concern. The suitability of plastic tanks to hold oil was called into question. Legal proceedings were begun against Balmoral in Belgium and France. Because some of the cracked tanks produced oil spillages the Environment Agency became involved, Balmoral’s problems became public knowledge. The Irish Rotational Moulders Association formally asked for an explanation of the nature and extent of the problem and what was being done about it.

A number of meetings took place between Balmoral and Borealis in late 2001 and 2002. The tenor of Borealis’ response to Balmoral’s complaints was that the failure of the tanks was or might be attributable to a number of factors, e.g. poor processing or incorrect pigmentation, the unifying characteristic of which was that none of them was Borealis’ responsibility.

On 20^th January 2003 a detailed technical meeting took place between Balmoral and Borealis. Borealis refuted any suggestion that borecene was not fit for its purpose and suggested a number of key contributors to failure other than borecene (wall thickness distribution, use of grind blended pigmentation, defective demoulding procedures, and design that was a trigger for cracking).

In a letter of 4^th February 2003 Mr Joyce claimed that borecene was:

“intrinsically not “fit for purpose” when considering the design, manufacturing and application of the finished product. At every stage Borealis have had full knowledge of the intended use of the material purchased by Balmoral and have actively recommended its use. Borealis cannot therefore escape responsibility for the cracking problem and its consequences”...

On 13^th February 2003 Borealis responded to the letter saying that they were taking the issue very seriously indeed, and were working urgently to obtain a technical determination of the cause at the earliest possible moment. The letter expressed the view that until the cause of the problem was identified it was premature to discuss a commercial settlement. Further tests were proposed and it was asserted that Borealis could not be held responsible for any

“final moulded part defects arising from a coloured powder, the specification of which we have no control over. As such, Balmoral’s involvement of ICO in investigating the issue is again requested”.

No mention was made by either side of any applicable terms and conditions.

During 2003 and 2004 the number of failures increased. The figures for failures in 2005 are less than those for 2004 but there is an issue between the parties as to whether or not this is a temporary lull. In mid 2003 Balmoral set up a Customer Services Department to deal with calls from the public and to organise tank change outs (such change outs being described as “emergency”, in contrast to the “planned” change outs which took place from autumn 2003 for tanks made from borecene MFR 6). Prior to then complaints had been dealt with in the Sales Department, initially by one individual working on a part time basis. When a complaint is deemed to have been resolved the relevant file is closed and passed to the “Archive” department. This Department was set up in late 2003 to collate information on individual complaint files, beginning with files where the complaint was closed. The Accounts Department was trawled for evidence of the specific costs for individual complaints. Staff in the Archive Department compiled a summary of cost information in relation to complaints on a sheet, and key data was input into an “Emergency Spreadsheet”, which has formed the basis for assessing the cost of tank failures.

The first report from Capcis, Borealis’ experts, summarised data provided by Balmoral revealing that between October 1997 and July 2002 approximately 54,000 oil tanks were manufactured. The total numbers of tanks manufactured from borecene (and ZN) were approximately as follows:

During that period failure statistics in respect of the six most produced tanks, out of about 27 designs, constituting 82% of production, were as follows:

The number of failures for the SL 1250 and H 2500 tanks is significantly higher than for the other major tanks designs. The third report of Miss Sally Hassell, Balmoral’s forensic accountant, reveals (paragraph 2.3) that the SL 1250 and H 2500 account for almost two-thirds of tank failures to date, although they are only 30% of production. The failures were particularly high in respect of tanks manufactured in 2000 and 2001, when MFR 6 borecene was being used. Of the H 2500 tanks manufactured in ZN material 131 of those manufactured in 1995 failed and 117 of those manufactured in 1996 – on average 8.2% of production (appendix 1).

Updated figures were produced during the course of the trial by Miss Sally Hassell and Mr Flemming Jensen, the expert for Borealis. Miss Hassell’s figures are to 31^st March 2005 and Mr Jensen’s to 31^st December 2005. Both experts had to make assumptions in relation to certain tanks because certain information was either unavailable or wrong. Their approaches to estimation employed different methodologies. Each agrees that their respective methodologies are both acceptable and that the results produced are reasonably accurate.

The resulting figures for the production of tanks from borecene during the period 1997 to 2002 are:

During 2000 there had been what was described by Mr Joyce as an investigation into the horizontal range. That review appears from a subsequent document to have included the conclusion:

“that the higher incidence of failures of horizontal tanks was mainly attributable to high cyclic loading at the edges where there are stress concentrations that are caused by tight radii and disruption of the continuity of the ribs. Thinning at the internal edges was also more prevalent with tight radii. The thickness checks that we normally carry out with the ultrasonic do not identify thinning at the edges. The FEA that we have carried out have been on simplified models that do not properly represent the edges”. (Footnote: ⁵⁸)

The most up to date failure rates for borecene are Mr Jensen’s 31^st December 2005 figures. These reveal that, out of total production of 72,214 tanks in the years 1997 to 2002, 2,668 tanks made of MFR 3, 4,330 tanks made of MFR 6, and 408 tanks made of MFR 4 had failed. The percentage failures, and the percentage failure in respect of all materials, were as follows:

In their supplemental report Capcis presented a further analysis of failure data which showed, amongst other things, that in respect of tanks manufactured in the period from October 1997 to July 2002 in the six most produced tanks designs and the H 5000 tank :

Between June 1999 and December 2004 approximately 200,000 tonnes of borecene grades were supplied to some 400 companies world wide, of which RM 8402/3 – the MFR 6 material – accounted for about 49,000 tonnes. Borecene has been used for many different applications. Metallocenes occupy about 25% of the American and European markets and are used successfully by over 100 rotomoulders in UK.

The essence of Balmoral’s claim is that borecene was not fit for the purpose of making standard green oil tanks and that as a result of this unfitness Balmoral’s borecene-made tanks failed in alarming quantities. Balmoral’s claim extends, on the pleadings, to a claim that borecene was not of satisfactory quality but, as Balmoral’s final submissions implicitly recognize, if Balmoral have a valid claim it is likely to be one on the former basis.

Section 14 (2) of the Sale of Goods Act 1979 is primarily directed towards substandard goods. Although there is an overlap between sections 14 (2) and (3) the function of 14 (2) is to establish a general standard which the goods in question are required to reach, and not to ensure that they attain some higher standard of fitness for a particular purpose made known to the seller. In appropriate cases the question as to whether goods are of satisfactory quality may be determined by considering whether they are fit for all purposes for which goods of the kind in question are commonly supplied: section 14 (2B) (a); Jewson Ltd v Leanne Teresa Boyhan [2004] I Lloyd’s Rep 505 (Footnote: ⁶⁰). For a material that has a very wide range of possible uses, and which is to be used and transformed by a specialist manufacturer for his own particular purposes, that seems to me somewhat too wide a test, particularly when polyethylene although commonly supplied for oil tanks is not , in respect of some grades, suitable for that purpose.

In my judgment Balmoral have not established that borecene was of unsatisfactory quality. The material supplied was not defective or incorrectly manufactured. Borecene was suitable for rotomoulding generally. The question is whether it was suitable for the particular purpose of constructing above ground static tanks to be used for storing oil over long periods. That is something that falls within the reach of section 14 (3).

A buyer will, in the absence of any terms of exclusion or restriction, have a good claim for breach of an implied warranty of fitness for purpose if :

I turn, therefore to consider whether, on the assumption that no exclusionary terms are applicable, there was an implied warranty of fitness for purpose.

All the sales were made in the course of Borealis’ business.

Borealis was well aware from the start of its dealings with Balmoral that Balmoral was making green oil tanks with a ZN polymer and a phthalocyanine green pigment by rotomoulding. Borealis introduced borecene to Balmoral as a possible substitute polymer. Balmoral made known to Borealis, insofar as that was necessary, that they would be buying borecene to use to make green oil tanks in their rotomoulding machines.

Borealis indicated to Balmoral that Borecene could provide a number of significant advantages as a polymer, including reduced cycle time; warpage control; improved flow properties; a broader processing window (Footnote: ⁶²); improved mechanical properties and the potential to reduce charge weights due to even wall thickness control – all these are enumerated in the Borealis brochure. What was said by Borealis’ representatives is reflected by what is said in the brochure.

MRF 6 was put forward in the meeting of 16^th November 1998 on the basis that it would improve material distribution within the mould and hence the overall wall thickness of the product, that it could improve cycle time by up to 39%, and that all its other physical properties were the same as MFR 3.

Borealis were the makers of borecene, and therefore in the best position to know its intrinsic properties, both chemical and mechanical. They had, as their literature made clear, a sizeable research and technology department. The intrinsic properties of borecene were within their sphere of expertise. They were also well equipped to perform a range of standardised test on samples of their products, and customers would be entitled to assume that when Borealis’ literature set out the results of those tests it did so accurately.

Borealis had the facility to make and did make relatively simple rotomouldings, usually in the shape of small boxes, with conventional moulding equipment (Footnote: ⁶³). This enabled them to make several copies of a standard product for testing purposes. (QUB did the same). Mr Halvorsen carried out extensive rotomoulding trials of borecene, experimenting with changes of temperatures, ratios, machine settings etc. The boxes were then sent to the laboratory for testing of mechanical properties. But Borealis were not professional rotomoulders, let alone conversant with the particular way in which Balmoral designed, manufactured and tested oil tanks. That was within Balmoral’s sphere of expertise.

In those circumstances Balmoral, in ordering borecene, reasonably relied on Borealis to supply a polymer whose properties made it reasonably suitable for the purpose of making green oil tanks by rotomoulding, in the sense that it was capable of being used to make consistently satisfactory tanks. By “satisfactory” I mean tanks which would last as for a reasonable minimum period of time in ordinary use. The ten year period covered by Balmoral’s warranty (and that of many other tank manufacturers) is a suitable measure. (Footnote: ⁶⁴)

Borealis would not have fulfilled that obligation if borecene’s properties were such that, whatever reasonable (Footnote: ⁶⁵) adjustment was made to Balmoral’s processing parameters, including grade of pigment and method of pigment incorporation, satisfactory green oil tanks could not be made despite an adequate design of tank. Nor would Borealis have fulfilled its obligation if the possibility of a need for an adjustment of processing parameters would not have occurred to a competent rotomoulder, or if such a rotomoulder would not be able to discover by reasonable endeavours what change of parameter was needed.

But Balmoral could not reasonably rely on Borealis in respect of design or rotomoulding both of which were within its sphere of expertise. As to the former, it is the job of the rotomoulder to design his tanks. A polyethylene supplier is not responsible for failings in design. As to the latter, a competent rotomoulder would realise that all raw material products are different, that a new product may perform in ways that differ from the way in which products that he has previously used have performed, and may react differently to the same pigments and methods of pigment incorporation as he has used before, as well as to different ones. One of the reasons why raw material suppliers supply samples of their products free of charge before purchases begin is in order to enable manufacturers to gauge the suitability of the supplier’s material for the particular purpose for which they intend to use it, i.e. by using their own moulds, pigments, blending methods, and processing parameters. This happened in the case of MFR 3 and MFR 6.

Fuel oils, and, in particular, kerosene are environmental stress cracking agents. This means that tanks filled with such oils are prone, over time, to crack at stress points, even though subject to stress levels lower than those at which they would be expected to fail. In the end all polyethylenes subject to long term stress below their yield stress will fail in a brittle manner. In the presence of an ESC agent they will fail sooner. ESC comes about because the ESC agent plasticizes and, thereby, softens the material. Softening, in the presence of stress, causes crazes (small voids) to develop in the amorphous phase of the material. In time these develop into cracks on the outside of the tanks (Footnote: ⁶⁶). Once the cracks reach a critical length, catastrophic brittle failure occurs. Kerosene is a weak ESC agent so that failure, if it occurs, may not happen for 10 – 20 years or more.

Water and other bulk liquids are denser, and therefore heavier, than oil, and thus exert greater load stress than kerosene. But water, and other bulk liquids, are not ESC agents to any appreciable extent.

Balmoral sold borecene to other manufacturers who have had problems with cracking tanks. In several cases, including DESO Engineering Ltd (“DESO”) in Great Britain, and Carbery Plastics Ltd (“Carbery”), Kingspan and Clarehill Plastics Ltd (“Clarehill”), whose claim is small, in Eire, they seek to hold Borealis responsible (Footnote: ⁶⁷). Between them these companies constitute a very large proportion of the oil tank manufacturing industry in the British Isles, which takes a sizeable proportion of the annual European tonnage of polyethylene for making oil tanks (12, 429 out of 20,367 tons for 2001).

It is common ground that the relevant Balmoral oil tanks have failed because of ESC. Broadly speaking the tanks have failed, irrespective of material, at particular locations specific to each model where the stress (Footnote: ⁶⁸) is high compared with other areas of the tanks. For the horizontal tanks the tendency is for the cracking to occur on the top of the central bumper bar near where the vertical corrugations start. The great majority of the SL 1250 failures consist of vertical cracks at the lower part of the side face: see figure 3.1. of Capcis’ first report, where there is what was termed a “step” where one half of the mould meets the other. Other, less frequent, areas of failure are around the holes through the tank and at the central base of the long side of the tank. Vertical tanks, whose failure rates are a small percentage of production, tend to fail towards the base of the tanks around the “mouse hole” or on the top of the tank around the lifting eye.

In the case of the two tanks with the highest degree of failure (H 2500 and SL 1250) FEA analysis, carried out by a firm called Wilde FEA Ltd, shows that high stress regions coincide with the primary points of failure. The failures, which Capcis also observed in tanks that they examined, are usually associated with changes in section (such as an internal corner) producing stress concentration and, usually, a change in wall section thickness, or where localised thinning concentrates stress. Thus there are high stress concentrations in the areas where thinning is most likely. This does not mean that the point of failure is always at the thinnest point in the area. Failure is a function of stress. Stress at point A may be insufficient to lead to failure; whereas a greater stress at point B will do so, even though the material is thicker at point B than at point A.

Rapra’s first report expressed the opinion that:

“… the poor ESC performance , coupled with the fact that Borecenes are at higher stresses due to wall thinning generate performance failures.”

The tenor of the Rapra reports was that in all but the simplest structures products made with borecene would inevitably display wall thickness variations.

There was some somewhat unsatisfactory evidence to the effect that Balmoral tanks had also failed in thick sections where there was no localised thinning. Dr Oram of Balmoral gave evidence in re-examination to the effect that the “norm of wall thicknesses where cracks had occurred” was 5.5 - 6mm. He had not in fact taken, or witnessed the taking, of these measurements but was reporting on data from Rapra which has not been identified or produced. Those figures are, however, not inconsistent with some of the Capcis measurements of sections of the tanks they examined. The wall thickness at the point of cracking identified in figures 3.4 - 3.19 to their first report varies from 4.64 to 7.42 mm in respect of borecene. Those figures do not however address the question of thickness variations in the area surrounding the cracks.

Mr Clements of Rapra indicated in his oral evidence that there had been failure where there was no change in wall thickness and where the failure occurred in a thick section. When challenged he referred (a) to a DESO tank where the sample made had failed on the top in thick section, (b) to vertical tanks (unspecified), (c) to his understanding that Carbery tanks had, to Capcis’ knowledge (Footnote: ⁶⁹), failed on the roof section, (d) to slim line tanks of other manufacturers failing in thick section, and (e) that he had seen a vertical Balmoral tank fail in thick section at the bottom of the outlet port. But he accepted that he had never referred Capcis to Balmoral tanks failing in thick section, nor discussed it with them, and that the tanks examined by Capcis were treated as representative by all concerned.

Dr Wright of Capcis, when he came to gave evidence, accepted that two of the measurements of the tanks of which he took samples – namely the “a)” measurements for tanks V 3655 and V 2600 – were measurements recording wall thickness sections around cracks in the tank top. In the former case the site of the crack has a thickness of 5.64 (see figure 3.7) and the minimum/maximum thickness ratio of the sample was 7.78:5.36. In the latter case (see figure 3.9) the thickness at the site of the crack is not specified and the min/max range is 10.52:4.95.

In the light of that evidence I do not accept that Balmoral’s borecene made tanks failed to any material extent in areas where there was no significant wall thickness variation.

Variation of wall thickness is graphically illustrated by the photograph at page 37 of Mr Clements’ first report. This photograph shows sections of a H 1590 tank manufactured from:

It is practically impossible to rotomould a tank without any wall thickness variation. The skill of the rotomoulder is to reduce the variation to an acceptable level. Substantial wall thickness variations are prejudicial to the integrity of the product for two reasons. Firstly, thinning of one area will mean that that area is more stressed than it would be if it was thicker because of the reduced cross sectional area. The flexural stiffness of a structure is proportional to the cube of its thickness. So, doubling a wall thickness will increase flexural stiffness by a factor of 8 (2³). Secondly, a difference in wall thickness will concentrate stress in the thinner portion, particularly if there is a change from thick to thin to thick (Footnote: ⁷¹).

Balmoral contend that the reason for the massive increase in failure of oil tanks is that borecene is fundamentally unsuitable for the manufacture of green oil tanks in that:

Borealis contend that borecene does not have either of those defects and that the reason why so many tanks have failed is because Balmoral failed appropriately to design, manufacture and test tanks made with borecene, as a result of which there were, in some tanks, inadequate wall thicknesses and/or excessive wall thickness variation at points of high stress at which failure occurred. In relation to “design” Borealis do not suggest that the general shape of the tanks had to be different; rather that the tanks should have been redesigned so as to reduce stresses at critical points, e.g. by the elimination of sharp corners or rapid changes of section, and by optimising processing parameters. In particular designs, such as the SL 1250 and H 2500, which showed much greater failure rates than others, stress was concentrated in areas of the tank where the design meant that wall thickness variation was likely. Balmoral, they say, failed to ensure that their rotomoulding process would secure a sufficient wall thickness and a sufficiently even wall thickness distribution, particularly in areas of high stress. The combination of a design that concentrated stress and promoted wall thickness variation in certain areas and a manufacturing process that led to wall thickness variations in those areas caused the tanks to fail.

Borealis also draw attention to the fact that, when Balmoral and others were using ZN material there was an average failure rate of about 3% a figure described by Balmoral as “commercially acceptable”; but, they suggest, unacceptably high for a product with no redundancy. In this Borealis is, in my view, correct. Given the potentially catastrophic consequences of an oil leak, an accepted failure rate of 3 in a 100 is too much.

The expert witnesses called by the parties in respect of this dispute were the following:

The expert evidence in this case is voluminous and complex and I have not always found it easy to see the wood for the trees. This has, on occasion, been because of the confusing manner in which some of the material was presented.

Mr Clements’ first report was considerably less clear than it might have been. It, also, failed to refer to what he had written in another report in a different action which was highly significant. Borealis categorised this omission as misleading. I do not believe that Mr Clements was intending to mislead me but, as will become apparent, I think that there are justified criticisms to be made of the contents of his reports, which contain omissions and errors, usually to the detriment of borecene.

Professor Pethrick is a recognised authority on the chemistry of plastics. His reports were clear and well written. At the beginning of his evidence he introduced for the first time certain diagrams, avowedly for the purpose of exposition. These included a new point, namely the significance for ESCR purposes of a polymer having a value greater than “Mc”. This criterion had not featured in his earlier reports (and thus took Borealis by surprise). Its relevance in determining the issues before me appeared to me increasingly questionable. I found some of his oral evidence difficult to follow, a circumstance not assisted by his introduction of points (e.g. on Mc and the criticality of viscoelasticity) not foreshadowed in his reports.

Dr Adrian Wright wrote only a small part of the CAPCIS report, much of it a gathering together of material derived from others. He is not a polymer scientist, nor an expert in tank design, rotomoulding, or statistical analysis nor a practitioner in FEA. Balmoral, although not objecting to his giving evidence, invited me to reject his evidence completely on the basis that he had no relevant expertise. I regard his evidence as at the borderline of admissibility. But, since none of my conclusions, with one exception, rely on an acceptance of his evidence the question of rejecting it need not arise. The exception lies in the significance of the pattern of failure, where his view (held also by Dr Clemens and Professor Crawford) is one which I would have reached in any event. Were it necessary for me to determine the question of admissibility I would have ruled his evidence admissible on the ground that he had relevant expertise as an examiner of material failures and their causes.

Dr Mark Clemens is a polymer physicist/technologist. He is neither a chemist nor an expert on the chemical architecture of polymers. I found him a reliable and straightforward witness.

Professor Crawford is a world authority on rotomoulding. He was for 15 years Professor of Engineering Materials at QUB, where he established the Rotational Moulding Research Centre, and then Professor of Mechanical Engineering at the University of Auckland, New Zealand, where he established research activity in rotational moulding. He was then Pro-Vice Chancellor of QUB, and is presently Vice Chancellor of the University of Waikato. I found him a straightforward and impressive witness.

I was invited to treat Professor Crawford’s evidence with caution on account of the fact that, when he was at QUB, his department had a Ph D student whose 3 year research project into the rotational moulding of metallocene polyethylenes, under his supervision, was sponsored by Borealis under a contract which showed that Borealis had an interest in using the fruits of her research towards marketing borecene as a superior material for rotomoulding in comparison with ZN material. The work did not include testing for ESCR (Footnote: ⁷²). QUB also had research contracts from most of the major material suppliers, machinery manufacturers, mould manufacturers and moulders.

I do not regard Professor Crawford’s evidence as compromised by that link, especially in circumstances where he has not had any contact with Borealis for some six or seven years, has done far more research on ZN material than this sole project on borecene, and has carried out research projects for several rotomoulders. I do, however, take into account the fact that Professor Crawford has not seen Balmoral’s practices in action nor do his criticisms amount to any specific suggestion as to how any specific design or production method should have been altered in order to secure a satisfactory result.

Compared with ZN Borecene has a narrower molecular weight distribution (Footnote: ⁷³), and a more regular chain architecture. ZN materials are much more irregular. They have a wider distribution of short, medium and long chains, and those chains have a wider range and distribution of side branches. A small percentage of the chains are short and highly branched. The majority are of medium length and there is a small percentage of long chains. By contrast metallocenes have a more regular structure with a high percentage of chains of similar length but with a few short or long chains. They have a more even pattern of side branches on the chains, with a small number of chains having a low amount of side branches. The side chains tend to collect, in the case of ZN material more on the shorter chains; metallocenes have proportionately more side chain branches on the longer chains (Footnote: ⁷⁴).

During the course of developing borecene Borealis carried out tests for ESCR on borecene (of various MFRs) and their own conventional ZN material - ME 8154. This had been Borealis’ standard rotomoulding grade prior to the introduction of borecene and was used in the UK for the production of green oil tanks by at least two manufacturers – T & D (Footnote: ⁷⁵) and Clarehill.

In the Bell telephone test (“BTT”) (Footnote: ⁷⁶) 10 notched samples of a standard thickness are bent into a holder which is then immersed into a cracking agent, usually Igepal, a form of detergent, at 100% or 10% solution, at 50°C. The time to failure of 50% of the samples is plotted. Despite its widespread use, the test is not very satisfactory for a number of reasons including the following (i) the stress to which the samples are exposed is usually in excess of the sample’s yield stress and does not therefore reflect reality; (ii) the individual times to failure of the 10 samples often vary very greatly, by as much as a factor of between 10 and 100; (iii) results taken at different centres vary very widely. The Bell telephone test is very sensitive to minor changes e.g., in how you make the notch, how you bend the sample into its holder, and temperature control.

As a result of the deficiencies of the Bell Telephone test , the Association of Rotomoulders (“ARM”) introduced a Constant Tensile Load (“CTL”) test. This is a form of creep rupture test, in which a notched sample, in the shape of a dog bone is suspended vertically with a weight attached to the bottom. The time to failure of the specimen is then monitored. In the ARM CTL test the specimen is immersed in 10% Igepal at 30° C. Borealis used CTL tests for pipe applications, but it was not in 1996 an established test of theirs for rotomoulding. Accordingly all their ESC tests before the launch of borecene were BTT tests, the then market standard. They only became able to do the ARM CTL test on borecene for rotomoulding in about 1997.

Borealis’ tests, so far as presently relevant (Footnote: ⁷⁷), were carried out on unpigmented polymer. The following table gives the average of many results derived by Borealis from Bell telephone tests together with the results of ARM CTL tests (Footnote: ⁷⁸):

As can be seen all the borecenes produced results better than those for ME 8154, Borealis’ ZN polymer. ME 8154 had a butane comonomer, which, all other things being equal, might be expected to have results lower than for a ZN with a hexene comonomer. As I have said, when quoting a rating for ESCR in their data sheets Borealis used a scale of 1 to 5 (5 being at the top). The reason why Borealis used such a rating rather than a figure in hours was because their tests were often carried out at any one of three different sites, in Norway, Sweden and Belgium. The absolute values as between those sites varied very considerably, making comparisons between those results and the results obtained at other testing places on other materials inapposite.

There was some contradiction in the evidence as to whether ME 8154 had been rated as having an ESCR of 3 or 4. In her second statement, paragraphs 12 – 14, Miss Fatnes says that ME 8154 had been assigned a rating of 4 before she assumed responsibility for the development of borecene and thought it appropriate to put borecene with 940 density in the same category but did not think it was significantly better so as to increase the rating to 5 (Footnote: ⁷⁹). In her third statement she corrected that and said that ME 8154 had a rating of 3, as she confirmed, no doubt correctly, in cross examination. This is consistent with her first statement, paragraph 114, as amplified in her third statement, where she said that a material which survived between 50 and 200 hours, in the 100% Igepal test, would be a 4. In due course the 100% test was phased out in favour of a test using the 10% solution, which is a more aggressive ESC agent.

Borealis also undertook a direct comparison between ME 8161, i.e. borecene MFR 6, and Dowlex NG 2432, a ZN grade with an MFR of 3.8. which is used for rotomoulding green oil tanks. The samples were immersed in 10% Antarox (similar to Igepal) and produced similar BTT results, the figures being 21 and 25 hours respectively. Creep rupture testing data on borecene for up to 1000 hours in air was generated at various temperatures and stresses and the borecenes were shown to be more creep resistant than Borealis ZN rotomoulding grades.

Borealis regarded any further long term testing (Footnote: ⁸⁰) as the responsibility of the product manufacturer. So do Capcis. Borealis informed Balmoral (see the letter of Reidat Anderssen of 20^th July 1997) that they had not studied the effect of petroleum and fuel oil on long term creep data. Borealis did not carry out any ESCR testing on samples in kerosene (Footnote: ⁸¹).

Somewhat surprisingly neither the BBA nor the Oftec standards make any reference to ESCR, and it does not seem to have been something that Balmoral had in mind until a very late stage. Dr Oram's evidence was that the first time that he had heard of a creep rupture failure mechanism was in 2002 when Rapra mentioned it as a possibility describing it as “a mode of failure after long term service well recognised in the polymer manufacturing industry”. In the light of that evidence it seems to me unlikely that Balmoral had it in mind at the time when the tanks were designed.

Rapra carried out creep rupture testing (with the dog bone specimens held in a clamp and subjected to a constant load) of samples of borecene made tanks. Since, in practice, tanks will fail over a period of years, some means must be found to make an earlier assessment of their propensity to brittle failure. Creep rupture testing at elevated temperatures is a means of evaluating that propensity. In the case of MFR 3 borecene the samples were taken from a failed H 1590 tank. In the case of MFR 6 and MFR 4 tanks the samples were taken from unused H 1590 tanks.

Creep rupture tests can be carried out in different mediums such as (i) Air (ii) Igepal; (iii) kerosene and (iv) diesel, and at different temperatures. Testing in air at ambient temperature is of limited significance, unless carried out over a very extended period. Testing in Igepal at an elevated temperature was a recognised method of providing a rapid qualitative assessment of susceptibility to ESC. Testing in kerosene and diesel at elevated temperatures is another method that allows a quantitative prediction of service life in the actual environment in which the Balmoral tanks were to be used. Testing at elevated temperatures simulates the effect of time. The results at those temperatures have then to be converted to results at ambient temperatures by applying a “time-temperature superposition” i.e. a method of calculation which will convert results at one temperature to results at another. The results have a considerable degree of variability. Mr Clements indicated that at any particular stress level a +/- shift in time of 0.25 of a decade was typical: in other words if failure occurred at 1000 seconds in one sample (Footnote: ⁸²) another sample might fail at between 750 and 2000 seconds. Capcis agreed that this was valid as a general indication of the degree of scatter in the testing.

Paragraphs 73 - 75 of Mr Clements’ first report read as follows:

“73

Specimens were... tested at 60, 70 and 80° C after saturation in

kerosene, at 80 ° C in diesel and at 60° C in air at various stress levels in order to generate a time to failure vs stress applied curve. The point at which the specimen failed in a brittle manner was also recorded.

74

Additionally creep rupture curves in Igepal...were generated for MFR4 Borecene….MFR6 Borecene…and the Z-N grades of BP...and DOW at 60°C. This was undertaken to accelerate the cracking behaviour with a view to ranking the materials with respect to their ability to cope with ESC.

75

The technique of time- temperature superposition was then applied to generate a derived master curve at ambient (20°C) in accordance with EN 1788 and ISO 9080.”

The results of that testing are contained in a number of figures in which the time to failure in seconds (on a logarithmic scale) forms the x axis and the applied stress (in MPas) the y axis. The figures are as follows:

Paragraph 116 of the Report stated:

“Creep rupture curves for each material examined at 60, 70 and 80° C saturated in kerosene together with 60° C Igepal and master curves at 20° C are contained in Appendix 1 of this report.”

Appendix 1 contains nothing of the kind. It contains results of tank strain gauging. Appendix 3 does contain creep rupture curves for MFR 6, MFR 4 and BP at 60, 70 and 80° C in unsoaked kerosene, i.e. material which has not, prior to the commencement of the test, been soaked to equilibrium in kerosene. It contains no data for MFR 3.

The potential significance of this muddle is this. Firstly, paragraph 73 referred to specimens tested at 60, 70 and 80° C after saturation in kerosene. The data from those tests was not annexed. What was annexed was data at those temperatures from unsoaked specimens, although Appendix 3 did not state whether the kerosene was soaked or unsoaked. The ordinary reader would, however, assume that the data in the appendix supported the curve in the text (Footnote: ⁸⁴). In their supplemental report Capcis commented that they had been unsure whether the data used to construct the master curve was indeed that in Appendix 3. Secondly, figure 5 which shows master curves of time to failure at 20° C, i.e. at ambient temperature, is an extrapolation from the raw soaked kerosene data at elevated temperatures. That made it particularly important that there should be sufficient testing (at different temperatures and stress levels) in order to provide a sound basis for extrapolation. If the values taken before extrapolation are invalid or unrepresentative the error will be magnified in the extrapolation and produce an inaccurate master curve. Thirdly, there is a dispute as to whether or not soaked data is a reliable guide.

Mr Clements identified, by strain gauging, the stress levels at what he described as the primary failure sites on a set of H 1590 tanks made in each of the borecene grades and the BP and Dow ZN grades. Four gauges were placed two above and two below (so that there were two gauges on each side) the horizontal rib section traversing the tank. The tanks were filled at the industry standard fill rate of 400 litres per minute. Tensile specimens were cut from the tanks and tested at 50 mm/min to derive a stress/strain ratio. Strains derived from the gauge (Footnote: ⁸⁵) were used to calculate stresses and a graph of stress against capacity derived (Footnote: ⁸⁶). Figure 9 of his report gives the following stresses at full capacity at gauge location 1 (Footnote: ⁸⁷):

He regarded the 10% Igepal results (figure 6) as providing a qualitative confirmation of that result. In the light of what he understood (erroneously) to be data on the failure rates for all borecene made tanks and the results of the H 1590 testing he expressed the opinion (a) that all tanks made from MFR 6 would fail before the end of 10 years from production, (b) that it was highly likely, in the case of tanks made with MFR 4 that a high percentage would fail within that period, and (c) in the case of tanks made with MFR 3, it was likely that a further number of tanks produced would fail within that period. But he made clear in his evidence that he no longer held that view in relation to the MFR 6 tanks, although it would apply to some designs of tank, and only held the views expressed in relation to the MFR 4 and MFR 3 material in respect of some designs.

I view these extrapolations with great scepticism. On the footing of the table in paragraph 193 all the MFR 6, MFR 4, and MFR 3 tanks would now have failed and they have not. Part of the reason why the extrapolations are flawed may be because the strain gauging is unreliable. Or the assumption of 2/3^rd use may be unreliable, although figure 9 indicates that at 1000 litres (i.e. about 2/3^rd full for a tank of 1590 litres ) the stress is near to zero, so that the real life stress could not be less. As will become apparent hereafter, there are more deep seated objections to figure 5.

Mr Clements also expressed the view:

that Balmoral’s designs were similar to those of other manufacturers.

As stated in paragraph 191 above, the data contained in Appendix 3 to Rapra’s first report is derived from specimens of material that have not been soaked in anything prior to the commencement of the test (“unsoaked data”). The data which underpins figure 5 is derived from samples that had been soaked in kerosene to equilibrium prior to the commencement of the test (“soaked data”). In a letter dated 3^rd February 2006 Mr Clements invited Capcis to accept that “unsoaked samples give unrealistic and potentially prejudicial data (and if not, why not)?”

This view was not expressed in either of his reports and took Capcis by surprise. In the light of that letter, and other matters, Borealis sought to introduce, and on 20^th February 2006 I allowed to be introduced, two reports that Rapra (including Mr Clements) had filed on behalf of Deso in their claim against ICO Polymers (Footnote: ⁸⁸), to which Borealis had been a Part 20 defendant. The first report had already been filed. The supplementary report, which Balmoral invited me to allow to be introduced if the first one was to go in, was signed by Mr Clements on that day.

In his first report in the Deso action Mr Clements had relied on testing on unsoaked data (Footnote: ⁸⁹) which showed that in kerosene and diesel MFR 3 had outperformed Dowlex MFR 3.8 and borecene MFR 6. The latter two were relatively comparable with the Dow material being slightly better in kerosene. In diesel borecene MFR 6 was better at high stresses (above about 4.5 Mpa) and Dow better at lower stresses. Mr Clements had however pointed out that the rate of permeability (Footnote: ⁹⁰) for borecene was substantially lower than that of the Dow material and that this would influence the creep rupture performance of thick samples in favour of borecene. In his second report in that action he said that the creep rupture curves in his first report “may have been influenced by the fact that lower permeation rates in the Borecene material meant that the relative ranking of the materials was favoured towards the borecene”.

Mr Clements did not at any time prior to the production of the Deso report in this action reveal that he had relied on a large quantity of unsoaked data in the Deso action or state what that data indicated. After the Deso reports had been admitted Mr Clements produced a statement explaining why he had not carried out any pre-soaked testing for the purposes of the Deso action.

In that statement he explained that creep rupture testing first started for Balmoral in 2002 showed that the ZN material used by Balmoral after it stopped using borecene significantly outperformed MFR 6 and MFR 4 borecene , the tests being done at 20°C in air and Igepal. The results were inconclusive as against MFR 3. Subsequently samples of MFR 3 and 6 were tested against two further ZN grades in air and Igepal at 60°C; and borecene at MFR 3 performed substantially worse. These results were disclosed. In Summer 2004 a testing programme was undertaken for Deso using the BTT test and creep rupture testing with unsoaked samples. The BTT tests showed Dowlex 2432 substantially outperforming MFR 3 and MFR 6 borecene. But in the creep rupture tests Dowlex 2432 performed worse. These results were sent to Capcis. In 2005 further creep rupture tests were carried out for Balmoral at elevated temperatures. These revealed that MFR 6 performed much worse than the ZN comparator but MFR 4 only marginally worse. In summer 2005 Rapra were instructed by Kingspan and Clarehill, two manufacturers who had only used MFR 6 borecene. In the course of creep rupture tests on unsoaked samples of MFR 6 in kerosene at elevated temperatures Mr Clements noticed that the thinner samples provided by one manufacturer failed notably earlier despite being under the same stress. He suspected that the difference in performance was directly related to permeability. “The penny dropped” at the very end of summer 2005. Accordingly a testing programme using pre-soaked samples was started in late summer /early Autumn. Only a very few results were available by the time of his first report in this action on 27^th October 2005. Further data from the testing programme was contained in his supplemental report. Deso never instructed Rapra to carry out testing on soaked samples. As a result his report in that action relied on unsoaked data, but with a reference to the possible significance of permeability. He remained of the view that the only way in which finally and firmly to test the ESCR properties of borecene was to use samples pre-soaked to equilibrium in order to reflect the in service conditions.

Mr Clements was well aware that Capcis were relying on unsoaked data, which is a normal method of testing that he had used himself (Footnote: ⁹¹). It seems to me singularly unfortunate and surprising, therefore, that his view as to the unreliability of unsoaked data did not feature in either of his reports, the first of which appended curves based on unsoaked data (without revealing whether it was soaked or unsoaked) (Footnote: ⁹²), or the joint expert’s report, or in the reports in the Deso action, or in the pleadings or in Balmoral’s opening, both written and oral (Footnote: ⁹³), such that this controversy came to the fore only in February 2006.

Mr Clements’ evidence is that he said that unsoaked data was invalid at the expert’s meeting in November 2005. Dr Clemens of Capcis denied this, although he accepts that shortly before service of the original reports in October 2005 Mr Clements mentioned in passing that he was conducting his CRT testing on soaked samples and that, after what he believed was a DESO experts meeting in 2005, Mr Clements had indicated as they were leaving that he had found some strange results with materials from a different manufacturer.

I am prepared to accept that in late 2005 Mr Clements made some reference to the fact that he had found a difference in results from unsoaked data as between samples from different manufacturers depending on the wall thickness of the sample, which might make comparative permeability a relevant factor, and that he was going to undertake a limited series of tests in the time available on soaked samples (Footnote: ⁹⁴). Mr Clements’ first report did, of course, produce a curve avowedly based on soaked data. But I do not accept that, prior to his letter of 3^rd February 2006, he made plain that he was contending that unsoaked data was unrealistic. If he had done so, I am satisfied that this would have been referred to either in the joint expert’s report or one of the supplementary reports and that his letter of 3^rd February would not have caused the surprise that it did.

A test based on samples soaked in kerosene (whether soaked until equilibrium prior to the test or soaked during the test) does not replicate the in-service conditions of a tank. In real life there will be kerosene on the inside and air on the outside. The polyethylene is not saturated throughout its entire volume. There will be a gradient of absorption from the inner to the outer surface, from which vapour will escape. ESC starts on the outer surface, which will have a lower concentration of diesel or kerosene than the inner surface. The outer surface will be subject to tensile stress and thus be more permeable than the inside. In the conditions of the creep rupture test both sides are permeated by kerosene.

Mr Clements’ view is that, in order to obtain a valid comparison between two different materials, which in use will be in long term contact with kerosene, it is necessary to have both samples at equilibrium so that each starts off in the same state of permeation. Otherwise you are not comparing like with like. If both samples have been permeated to equilibrium, the only relevant difference between the two samples is their chemical composition, and it is that which is to be tested.

Dr Clemens’ view, which I accept, is (a) that using unsoaked data was common practice, a proposition with which Mr Clements agreed; and that using soaked data is not (Footnote: ⁹⁵); (b) that in using samples pre-soaked to equilibrium “you are altering the whole gross material, severely differently to what happens in practice (Footnote: ⁹⁶)”; and (c) that if the material was pre-soaked to equilibrium, it would not break in a truly brittle manner; the failure mechanism would be different to that which would occur in real conditions (Footnote: ⁹⁷). Moreover, he disagreed with the suggestion that the ESC process will only start after the point of equilibrium absorption. On the contrary crack initiation and propagation takes place at or very near to the exposed surface of the samples as the cracking agent penetrates the surface of the polymer and initiates craze formation by disentangling or pulling out tie molecules and exploiting existing micro-defects, long before equilibrium is reached. I accept that this is so, not least because it is consistent with the chemical evidence to which I refer below.

Lastly it is to be noted that Capcis’ tests, to which I shall refer below, were conditioned to the test temperatures in the test environment for about 18-24 hours during which, on Mr Clements’ uptake figures at 80° C, some 40-50% of the total liquid which could have been absorbed would have been absorbed before the test began; and the nature of the test is such that the surface of the sample experiences the maximum tensile stress (so that if ESC is to begin it will do so there) and would be soaked by the time the test began.

I am not persuaded that results from unsoaked data are unreliable, or even that results from soaked data are more reliable. In either case the test conditions will not replicate reality since tanks are never immersed on both sides, whether before they enter service or during it. Pre-soaking to equilibrium is, in my view, a more significant departure from reality than using unsoaked specimens. The fact, if it be such, that borecene takes up kerosene more slowly might create a bias in its favour (of uncertain extent and cause). But, for the reasons give by Dr Clemens, pre-soaking to equilibrium before the test starts would make so fundamental a change in the material and its eventual method of failure, compared with reality, as to cast doubt on the validity of the resulting data, or at any rate its reliability when compared with unsoaked data. Mr Clements has made use of extensive unsoaked data in this action and the DESO action without any more than a tentative qualification.

The evidence as to the slower rate of uptake of borecene as compared with ZN material is, in any event, not very firm. Mr Clements’ first report has figures of the percentage kerosene uptake of borecene and Dow and BP material at 80° C. That is a very high temperature. It is by no means clear that the same relationship would apply at lesser or ambient temperatures, particularly when borecene has, as it does, a fraction which melts at 75° C which ZN material does not. In addition there appears to be a greater variation in the speed of uptake between the two ZN materials than between the borecenes and the Dow material.

In his supplementary report Mr Clements did not deal with the question raised by Capcis as to whether the data in Appendix 3 of his first report supported the master curve in figure 5; nor did he reveal that that data was unsoaked. He produced at page 302 a new master curve, from soaked data, comparing borecene with MFRs 3, 4 and 6 with BP and Dow. Following Mr Clements’ letter of 3^rd February Capcis asked for the data supporting the master curves. This was produced on 17^th February and can now be found, not in any report of Mr Clements, but at Appendix 2 to Capcis’ 2^nd supplemental report.

Dr Clemens’ opinion was that:

“the absolute minimum amount of data required before one could begin considering attempting to derive a master curve is 3 different stress levels for each of three different temperatures. If one has less than 3 stress levels at a given temperature, the potential for error is so enormous that attempting to derive a master curve is a thoroughly unsound practice which is liable to lead to gross inaccuracies in the time-shifted master curve so obtained”.

As to this data he said:

“We were thoroughly taken aback at the lack of data from which the master curves presented in Mr. Clements’ reports were generated … In our opinion, it is inappropriate to construct a master curve from any of this data … If one considers the data which Mr. Clements had at the time he constructed the master curve in his first report, we find it extraordinary that he felt able to proceed in the way in which he did. Coupled with the misleading presentation of graphs at pages 80-82(i.e. Appendix 3) which did not correspond to the master curve, we are bound to say that we feel somewhat deceived. If the true data had been presented, it would have been immediately apparent how inadequate it was.” (Footnote: ⁹⁹)

I accept the validity of these criticisms of the data and Rapra’s reliance on it, without thereby suggesting that Mr Clements intended to deceive. Mr Clements referred to an academic paper (Footnote: ¹⁰⁰) which suggested that data from two temperatures at the same stress, and two stresses at the same temperature, i.e. a minimum of 3 data points, would be something that you could start with, provided that the stresses were low and the fractures brittle, and provided a particular formula was used. He did not use that formula.

Capcis plotted the soaked data from which Mr Clements had constructed the master curve in his second report (notwithstanding their view as to its inappropriateness for the purpose) and discovered that in the plots for materials at 70 and 80° C (Footnote: ¹⁰¹) the BP material performed worse, although in the master curve in Rapra’s second report it appears as second best. This result they regarded as “obviously purely a function of the application of dubious time shift factors”. They also plotted a master curve based on the unsoaked data in Appendix 3 of the first report using average values of time shift factor for each material. It showed that MFR 6 material was, on a general ranking, the best and MFR 4 the worst, although the curve was not one in which they placed any confidence.

In his supplementary report Mr Clements set out a table of ESCR figures which contained (a) the data set out in the table at paragraph 181 above derived from Miss Fatnes’ witness statement; and (b) “additional data from rotational moulding static oil tank grades from several manufacturers of LLDPE (datasheets provided in appendix 4)”. On its face the table, which is reproduced in the first two columns of the table below, showed borecene to be markedly inferior in respect of ESCR. As appears from the third column the significance of the table is not quite what it seems.

The considerable dangers involved in comparing BTT results in one laboratory with those obtained in another is apparent from the different figures obtained for MFR 3 and MFR 6 borecene and for the Dow material when tested by Borealis and when tested elsewhere: see columns 2 and 3. In the case of Dow there was an 18 fold difference (450/25). Mr Clements accepted in cross examination that Borealis’ test results were consistently lower than those of others regardless of the type of material used.

In footnote 2 on page 16 of his second report Mr Clements expressed the view that borecene MFR 3 would have been “nowhere near” having an improved ESCR compared to DSM material. In so doing he compared:

In the light of the matters to which I have referred I do not regard the data relied on by Mr Clements as establishing that borecene had a very poor ESCR.

I have not ignored the fact that Rapra did creep rupture testing in Igepal which showed a ranking of earliest times to failure at 60° C of (i) MFR 6; (ii) MFR 3; (iii) Dow, and (iv) BP. I am wary of the utility of these results derived from Igepal. At best (Footnote: ¹⁰⁵) they reveal a qualitative ranking of those materials, so far as ESCR is concerned, at that temperature in that material. But the tanks in question were to store kerosene or diesel, not soapy water, and at a lesser temperature. I note that in the DESO action no Igepal testing was put in evidence by RAPRA. There is no evidence which establishes that a reliable correlation can be made between Igepal results at 60° C and diesel/kerosene results at ambient temperature (Footnote: ¹⁰⁶); and it is not self evident that there must be such a correlation, especially since Igepal is not absorbed into polyethylene (Footnote: ¹⁰⁷).

Prior to their first report Capcis caused to be carried out a creep rupture test known as a “3-point bending test” on a number of different types of polyethylene (injection moulded unpigmented, rotomoulded unpigmented, and unused Balmoral green tank samples) in different materials. This test method ensured that the maximum stress was applied to the outer surface, that being the point at which failures started. The apparatus used establishes a bending moment similar to that experienced by the tank wall in the field. Tests were undertaken at 60, 70 and 80° C in air, diesel and kerosene. The specimens tested in diesel and kerosene were totally immersed (but not pre-soaked to equilibrium). At the time of their first report there was insufficient data available for time temperature superimposition. On the data available Capcis concluded that, at temperatures of 60, 70 and 80° C in diesel and kerosene, unused Balmoral green tank samples performed worse than unpigmented materials produced using optimised rotomoulding conditions (as was to be expected), and that there was no consistent evidence of any difference in the CRT curves of their borecene materials and BP Rigidex 4330. By the time of the second report (January 2006) the only brittle failures that had been recorded in respect of green Balmoral tank samples were in respect of a number of the ZN samples at 70 and 80° C.

In his supplementary report Professor Pethrick sought to investigate what might explain why Borecene had lesser ESCR than ZN polymers as apparently found by Mr Clements. What appears in the three following paragraphs, and in paragraphs 226 and 227, is derived from his evidence and I accept it as a general description of the relevant chemistry.

As the molten polyethylene cools the molecules become less mobile until at about 140° C they start to fold up into crystalline polymer chains called lamellae, similar in appearance to a snake whose body is coiled in descending horizontal rows. These chains grow outward radially from a central nucleation site, progressively extending the crystalline phase. The crystalline phase is to be distinguished from the amorphous phase which is the proportion of the melt that is not incorporated into the controlled crystalline arrangement of lamellae during solidification, and which ends up as a solid comprising irregular, unstructured or tangled chains. During this process, irregular chains and branches that cannot readily be incorporated into the regular structure of the lamellae become segregated to the surface of the growing crystal. Occasionally a chain may bridge the gap between growing neighbouring crystals, and these linkages between crystals (bridging or tie molecules), together with physically entangled amorphous chains and branches, provide a path for force to be transmitted between the crystals and for inter crystallite bond strength to be achieved. Adjacent crystalline regions connect either directly by branched chains or via entanglements.

In the case of ZN material the lamellae group into three-dimensional structures called “spherulites” in which individual lamellae are distributed spherically around the “seed” crystals, giving the appearance of a Maltese Cross. The molecules that are not incorporated into spherulites remain as amorphous material located between the lamellae and between spherulites. Branched chains cannot, however, all fold into the regular lamella structure and elements of the polymer are segregated to the amorphous region between the crystalline lamellae. Bridging/tie molecules arise from the natural tendency, as a result of entanglement and branching (particularly the former), for elements of the polymer to be segregated to the amorphous regions between crystalline lamellae. Entangled molecules will slow down the formation of crystalline regions. Entanglement is best envisaged by the analogy of spaghetti. Short lengths of spaghetti can remain relatively separate. Longer lengths entangle into a mass. The greater the number of entanglements the more force is need to break them. The wide distribution of branches in ZN materials assists entanglement.

Because metallocene polymers have a narrower molecular weight distribution, with chains more even in length and with a more even distribution of branches on the chains, they are more effective at folding (Footnote: ¹⁰⁸) during the crystal phase, as a result of which crystals will grow more readily and therefore faster. This means faster cycle times for rotomoulders. The lamellae will be longer. As a result there will be an increase in strength, but there will be less tie molecule density.

The tenor of Professor Pethrick’s reports was that, whereas in the case of ZN materials the lamellae group into spherulites, in the case of metallocene they formed into a wheat sheaf structure and that there was less of a tendency for the polymer chains to form an amorphous phase. The implication was that the difference in structure and the lesser amorphous phase would or might explain why the two materials have different ESCR. But, whilst maintaining the significance of differences in lamella structure as between ZN and metallocene, Professor Pethrick accepted in cross examination:

Professor Pethrick explained that low molecular weight hydrocarbons such as diesel and kerosene can interact strongly with the amorphous phase of linear polyethylene. At ambient temperatures diesel and kerosene are not solvents for crystalline polyethylene; but they will be able to interact strongly with the amorphous regions. These agents will endeavour to plasticize the amorphous regions between the crystallites. That plasticization is resisted by the tie molecules that bridge the lamellae. The ability of the polymer to resist cracking is dependent on the extent to which the tie molecules can retain structural integrity and resist crack propagation. If they cannot resist the applied stress, scission of the molecules occurs and crazing of the surface takes place. This provides a means for crack propagation, ultimately leading to failure in those regions of the tank the subject of higher stresses.

A high density of tie molecules bridging the lamellae will resist ESC; a low density of such molecules will assist it. ZN materials have a higher content of longer polymer chains than borecene and therefore a higher resistance to ESC. So, whilst borecene will, on account of any increased crystallisation be stronger, its ESCR will be less than ZN (Footnote: ¹¹⁰). Metallocenes have a high aspect ratio. This can inhibit the diffusion of ESC so that borecene is likely to take up an ESC agent more slowly than the lower aspect ratio ZN material. But once ingress has occurred borecene is less likely to accommodate the plasticization action of an ESC agent without rupture.

This conclusion is, Professor Pethrick claims, supported by an article by Wright – “Environmental Stress Cracking of Plastics”, Rapra, 1996 - which indicates that ESCR correlates with the fraction of higher molecular weight polymer present in a material; and with an article “The Critical Molecular Weight for Resisting Slow Crack Growth in a Polyethylene” (Footnote: ¹¹¹), which concluded that the resistance to slow crack growth in the particular commercial ethylene-hexene copolymer there examined (Footnote: ¹¹²) resided in those molecules whose molecular weight was greater than 1.5 x 10 to the power of 5.

Professor Pethrick’s view expressed in his supplemental report is that borecene’s inadequate ESCR is by far the most significant factor behind the massive increase in Balmoral’s fuel tank failures; and that Borecene did not have the ability to resist ESC together with appropriate rheological characteristics to make Borecene easily processable into the desired structures.

As the mould rotates, different parts of it will pass through the powder which becomes tacky and then molten. As the powder becomes tacky it sticks to the walls of the mould. The extent to which it does so is in part dependent on the viscosity (i.e. the resistance to flow) of the polymer. The lower the viscosity the greater the flow and the more the material will fall away from internal protrusions. For that reason a lower viscosity may make it more difficult to avoid wall thickness variations at those protrusions. But if the polymer is highly viscous the result may be that air is trapped in the melt and the wall structure is unduly foam like rather than dense and compacted. Reducing viscosity thus assists the removal of bubbles. Another potentially relevant characteristic is the visco-elasticity of the material i.e. its ability to recover its form once any force applied to it is removed or, more graphically for present purposes, its ability to drape around a protrusion as the mould rotates so as not to fall away.

Professor Pethrick’s evidence was that the melt viscosity of metallocene polymers had been the subject of study at the University of Strathclyde. That research is potentially relevant both to ESCR and rheological properties. The viscosity of a polymer melt is measured by placing the polymer between two oscillating plates. The energy required to rotate the plates is then measured. The speed of rotation can be varied so as to produce a variation in shear rate (the measure of the speed of movement of liquid against a surface). The sheer rates experienced in rotational moulding are very low. Since, at those rates, viscosity is independent of shear rate, the viscosity, in this context is referred to as zero shear viscosity [η◦]. Rotational moulding is undertaken at shear rates of between 0.001 and 0.01 sֿ¹.

This research indicated a number of things. Firstly, the zero shear viscosity of the material decreased, as you would expect, as between two ZN materials and Borecene MFR 3, as between MFR 3 and MRF 4, and as between MFR 4 and MFR 6. Viscosity is measured in Pascal seconds. The viscosity of water is of the order of 0.001 of a Pascal second. The viscosity of treacle is of the order of 1 Pascal second. The viscosity of MFR 6 at 200°C is over 800 Pascal seconds. Mr Clements’ figures (Figure 11) were as follows:

Secondly, at low shear rates polymer material with short chains all the same length (like borecene) exhibit a viscosity independent of shear rate i.e. the viscosity does not reduce as the shear rate increases. If, however, the viscosity of a polymer is dependent on shear rates so that viscosity reduces as shear rate increases (frequency dependence) at relatively low shear rates, that is an indication, in Professor Pethrick’s view, of a number of factors: a broad distribution of chain length or molecular weight distribution, entanglement of the chains in the melt and clustering of polymers. Borecene showed a very different dependence on shear rate when compared with BP and DSM’s ZN materials. In the case of borecene with MFRs of 3, 4, and 6, frequency dependence did not occur until higher shear frequencies than was the case with the ZN material. This indicated that in the case of the ZN materials the polymer chains in the melt were either entangling or were much longer in length. Polymer chains above a critical molecular weight will entangle and the viscosity of the melt will increase significantly (the spaghetti analogy). The behaviour of the ZN catalysts leads to the creation of a small number of longer chains which in turn leads to the higher viscosity of ZN material as compared with borecene.

Thirdly, if the melt was kept at a constant temperature of 180° or 200° C for a period of time and at a constant shear rate there was an increase in viscosity after a period. That increase was more marked in the case of DSM and BP material than in borecene. In the case of melt held at a constant temperature of 180°C this difference occurred in the region of 3000 seconds. In the case of melt held at 200° any marked increase occurred at about 1250 seconds (Footnote: ¹¹³) in the case of the BP material, but much later in the case of the DSM material. But the MFR 6 material showed marked increases in viscosity earlier than the DSM and BP material and the MFR 4 material increases similar to those of DSM.

The consequence of these rheological characteristics in terms of viscosity and viscoelasticity was that, in Professor Pethrick’s opinion, metallocene polymers are:

“probably satisfactory for moulding simple box or barrel type structures where no protrusions exist. However, where the design of the container requires strengthening ribs then “drape” becomes a highly significant issue. Borecene polymers lack viscoelasticity and appear to be unsuitable for rotational moulding of any more complex structure than a symmetrical box or barrel without internal protrusions where even wall thickness is required. Where they are used in the production of articles by rotational moulding process, metallocenes will inevitably give rise to significant variations in wall thickness distribution”.

When he came to give oral evidence Professor Pethrick produced, to illustrate his general thesis, a schematic figure (Footnote: ¹¹⁴) in which the shear rate formed the x axis and viscosity [η] the y axis. On this figure were plotted values notionally attributable to three materials with a broad molecular weight distribution (represented by three dotted lines) and three materials with a narrow molecular weight distribution (represented by 3 undotted lines). The materials in the former category had viscosities in excess of those in the latter at all shear rates. The materials in the latter category had viscosity values which remained broadly constant until a given shear rate was reached at which stage they began to decline. Viewed from left to right those values, when plotted, displayed three roughly parallel lines, which at a certain shear rate turned downwards as viscosity reduced. By contrast the materials in the former category had, from the very start, a reduction in viscosity as shear rate increased (“shear thinning”), so that the three lines were never parallel but curving downwards as from the lowest shear rate.

In this figure the material with broad molecular weight distribution was characterised as having a molecular weight higher, and the material with a narrow molecular weight distribution as having a molecular weight lower, than a critical value of “Mc”. Mc is the point at which (Footnote: ¹¹⁵), if you plot viscosity against molecular weight, there will be a step change in the relationship between viscosity and molecular weight. Before that point viscosity will increase in proportion to the molecular weight (here Mn: see paragraph 239). After that point viscosity will increase at a 3.5 power of the molecular weight. This is because the longer chains (whose existence increases the molecular weight figure) produce a disproportionate increase in viscosity. It is these longer chains that produce entanglements which enable the polymer to resist ESC. Below the Mc point there is no entanglement and generation of tie molecules is dependent on bridging. As a result material with a molecular weight below Mc is less suitable for ESCR purposes. The molecular weights of borecene were he said (Footnote: ¹¹⁶) around or just below the Mc value for these materials. That value differs for different materials. Further the fact that, until relatively high shear rates, borecene had a constant viscosity against shear rate was indicative of the fact that the material was not entangling. He regarded entanglement as connoting viscoelasticity.

I did not find the manner in which Professor Pethrick introduced evidence as to the supposed criticality of Mc – in a document, not previously circulated, introduced without prior notice at the commencement of his evidence and then explained orally – helpful. I was told that Mc was “an established reference point in any of the polymer literature on rheology” (Footnote: ¹¹⁷) but no such literature was produced or identified by reference. Professor Pethrick’s evidence did not indicate, prior to my asking him questions, what was likely to be the Mc point applicable to Borecene. But he told me that for a polyethylene it was, typically an Mn figure of about 18,000, an Mw figure of around 35,000, and an Mz figure of around 45,000, although that evidence was given with the caveat that Professor Pethrick did not know what was the Mc value for the particular copolymers that constituted the material at issue because different copolymers have different Mc points.

The experts are agreed that, in general the higher molecular weight species in a polyethylene material will improve ESCR. Polymers do not, however, have a single molecular weight because the polymer chains vary in length, and, therefore, weight. Molecular weight is expressed in three different ways:

Mr Clements calculated, by means of gel permeation chromatography, the Mn, Mw, and Mz figures for borecene MFR 3, 4, and 6 and two ZN materials. The data (Footnote: ¹¹⁸) is as follows:

These figures are significant for a number of reasons. First, they show the ZN material to have significantly higher Mz values than borecene. The literature indicates that Mw “appears to be the best parameter for describing the effect of molecular weight on slow crack growth and on the low temperature fracture stress” (Footnote: ¹²⁰) in preference to Mz, but that, with materials of the same Mw “it has been reported that the preponderance of the high molecular weight fraction had the greatest influence on EC resistance (Footnote: ¹²¹) ” i.e. that Mz can rank materials with the same Mw because it highlights the quantity of longer chain molecules. Secondly, none of the data places borecene close to the Mc point indicated by Professor Pethrick. The figures for borecene are considerably in excess of that point whichever measurement of molecular weight is employed. Thirdly, Mr Clements’ data shows that the ranking of the Mw value of borecene is effectively (i) MFR 3 and BP Rigidex; (ii) MFR 4; (iii) Dow, and (iv) MFR 6.

There are other indicators that metallocenes have an advantage over ZN material so far as ESCR is concerned. In the case of ZN material the short chain branches (derived from the ethylene comonomer) are concentrated on the shorter chains; with metallocene they are evenly distributed irrespective of the length of chain. The literature suggests that that is an advantage in ESCR terms:

“Ishikawa et al first showed the profound increase in resistance to SCG when most of the short-chain branches are placed on the long molecules. Scholten and Rijpkema … also showed the importance of placing the chains on the high molecular weight molecules, and that at the same time, SCB on the small molecules appear to have little or no effect on SCG”. (Footnote: ¹²²)

At the same time metallocenes have a significantly lesser quantity of the long molecules.

I found it difficult to discern what precise significance Professor Pethrick attached to the distinction between viscosity and viscoelasticity in the context of rotational moulding. Viscoelasticity is the ability of a material to recover after the application of a force has ceased. In rotomoulding, as the mould rotates and the melting powder tumbles, the force of gravity acting on the polymer continues although it operates in different directions. The upshot of Professor Pethrick’s evidence was, as it seemed to me, that viscosity and viscoelasticity are components of complex viscosity [η] (Footnote: ¹²³) – and that if the material has a degree of elasticity it is likely to be more viscous. This accords with common sense. A material that has a certain degree of elasticity is inherently likely to have a greater resistance to flow. It also explains why borecene has a lower viscosity than ZN material. It is common ground that a polymer with a narrower molecular distribution, such as borecene, will behave more like a Newtonian liquid (e.g. water whose viscosity does not change with an increase in shear rate) than a polymer with a wider distribution, such as ZN material, which will behave in a more pseudoplastic way.

Whether or not Professor Pethrick regarded viscoelasticity as necessary if even wall thicknesses were to be obtained was not entirely clear (Footnote: ¹²⁴). Nor was it clear to me whether it was being said that there was any given value of viscoelasticity which was (a) critical and (b) which borecene did not achieve. Professor Pethrick did not measure the viscoelasticity of borecene and comparator materials at the shear rates relevant to rotomoulding. In the end his evidence was that the importance of viscoelasticity in rotomoulding, so far as wall thickness was concerned, was unproven, “because I don’t think it’s been considered” (Footnote: ¹²⁵), and so I find it to be. (Footnote: ¹²⁶) I accept Professor Crawford’s evidence, which has support in the academic literature (Footnote: ¹²⁷) that so far as wall thickness distribution is concerned there is nothing to indicate that viscoelasticity is an important advantage.

I also note Dr Clemens’ evidence that figure 11 (see paragraph 232 above) of the first Rapra report sets out data relating to seven materials of which only one - Sclair/Nova - exhibits a significant pseudo plasticity. Capcis were unaware of that material ever being used by Balmoral in oil tanks. The other ZN material that shows a slight pseudoplastic tendency at the low shear rates applicable to rotomoulding is DSM Stamylex, which was the material used by Balmoral immediately before borecene. The other two ZN materials (Dowlex NG 2432 and BP HD 4330) are, at rotomoulding shear rates, effectively as Newtonian (i.e. not shear rate dependent) as the three borecene grades. The former was used by Balmoral immediately after the use of borecene ceased (and has a zero shear viscosity only very slightly higher than that of Borecene ME 8166/7). The latter is Balmoral’s current material of choice.

Professor Pethrick placed store by the shear rate dependence of ZN material at lower shear rates than Borecene - see paragraph 233 above – as an indication of the viscoelasticity of ZN material. It is, however, apparent from figure 2 of Professor Pethrick’s first report (paragraph 28) that the shear rates that he was taking to produce the data in that figure were way above those actually found in rotomoulding. The x axis of figure 2 starts at approximately 0.06 sֿ¹ (Footnote: ¹²⁸) and the lines of data stop some way short of the start of the graph. Further as the lines reach the start of the graph they flatten (Footnote: ¹²⁹). This indicates that, at the shear rates relevant to rotomoulding (put by Capcis at between 0.001 and 0.01 sֿ¹), there is no significant difference between the shear rate dependence of ZN and Borecene. (Footnote: ¹³⁰)

But Professor Pethrick’s point was that, if at shear rates higher than those used in rotomoulding, you see in relation to ZN material a decrease in viscosity as shear rates increase, as compared with none or a markedly lesser change with borecene, that is indicative of the fact that ZN material contains a degree of elasticity/entanglement which metallocenes lack, and that it is that entanglement which is significant for ESCR purposes. He disputed the proposition that the data showed that at shear rates found in rotomoulding significant entanglement is not happening in any of the materials, including the ZN materials. He was of the view that:

“if you have any shear rate dependence below angular frequency of around about 10 (i.e. 10 radian seconds), this is an additional contribution due to entangled polymers” (Footnote: ¹³¹)

So far as changes in viscosity over time are concerned the data as to the effect of holding a melt at a constant temperature of 180° C (figure 3) revealed no significant increase in viscosity until about 3000 seconds – a period greater than that over which the melt would be heated in rotomoulding.

So far as the data at 200° C is concerned (figure 4), there are a number of points. Firstly, figure 4 does not appear to me to display any noticeable increase in viscosity until about 1000 seconds which is close to the limit of time in which the mould would be in the oven. Secondly, any increase in viscosity could only contribute to satisfactory wall formation after a period of time during which the distribution of polymer will have been unaffected by this phenomenon because the temperature of the polymer is too low. This period will end substantially after the melt point, by which time the wall thickness distribution is likely to have been substantially determined. Thirdly, whilst the highest temperatures will be experienced where the melt meets the mould, lower temperatures will be experienced towards the interior surface of the mould. So any change in viscosity is likely to take place, if at all, at the mould wall. But the polymer is likely to be adherent or even bonded to the mould wall so that the benefit of any increase in viscosity is limited. Fourthly, I note that MFR 6 showed the steepest increase in viscosity of all, albeit not starting until about 2000 seconds, and DSM the second lowest, starting at about 3000 seconds. In these circumstances I am not persuaded that the data set out at figure 4 casts any light on the suitability or otherwise of borecene for rotomoulding. Professor Pethrick did not suggest that I should set much store by this data and I do not.

Borecene was marketed as having comparatively low viscosity. Mr Clements described borecene as having inherently low viscosity. For ease of reference I set out Figure 11 of Rapra’s first report again:

As is apparent the viscosity of borecene MFR 3 is about the same as Dow and that of MFR 4 and MFR 6 somewhat less. But the evidence does not establish that there is a figure for viscosity below which polyethylene is not suitable for rotomoulding oil tanks and below which borecene lies. In relation to figure 11 Mr Clements was asked:

“Q. But you don’t say in the context of this graph that you can only properly have a rotomoulding polyethylene if you achieve a certain figure on the vertical axis?

A. No”

In the light of the considerations that I have set out in the preceding paragraphs I was far from persuaded that Professor Pethrick’s evidence showed borecene either to have very poor ESCR or such rheological characteristics as made it impracticable to use for making oil tanks.

The Capcis reports make a number of points which I summarise in the following paragraphs.

Only the rotomoulder has control over the parameters of the moulding operation. These parameters will include (i) oven temperature; (ii) heating times; (iii) cooling methods (forced air, water); (iii) cooling times; (v) speed of rotation of the arm and the plate; (vi) speed ratios i.e. the ratio between the speed of the arm and the plate (Footnote: ¹³²); and (vi) release agents (Footnote: ¹³³). Each change in material requires a change in processing conditions and careful trialling of each model type in order to ensure the production of a good quality finished product. Each material has unique optimum processing conditions which can affect the quality of the finished product. Balmoral would have to determine how a new product reacts to its particular manufacturing processes. This trialling should involve destructive testing of trial products, including measurement of wall thicknesses, e.g. by callipers, particularly at areas where non destructive measurement techniques cannot operate e.g. at corners.

In relation to OFTEC Capcis say this:

“4.76.

Whilst OFTEC OFS T 100 states that for a tank designed to store either diesel or kerosene, the minimum wall thickness should not be less than 4.5mm, this does not mean that an oil tank with a wall thickness of 4.5mm is fit for the purpose of storing large volumes of kerosene or diesel. Indeed, an oil storage tank would fail the deformation requirement of this standard should the tank be made to this uniform minimum wall thickness. What it actually means is that, at non-critical points (i.e. regions of low loading) a minimum 4.5 mm wall thickness is required. Due to the complexity of oil tank designs, it is the responsibility of the manufacturer to assess the actual wall thickness required at any point in his tank design.

4.77

Unless the tank design is simple (i.e. cylindrical) only a design engineer with access to finite element analysis (FEA) would be able to determine the required wall thickness at any given point in a tank with any confidence. Prior to the use of FEA, a design engineer would calculate a “best guess” and build in a safety factor in the final recommendation of wall thickness. If a safety factor of 2 was applied then the recommended wall thickness would be twice the “best guess” estimate.”

Balmoral’s tanks suffered failures of the same type as later occurred with borecene when it was using ZN material. Balmoral’s figures for failure rates by 31^st December 2005 of the following models made in 1995 and 1996 i.e. not from borecene are as follows:

Borealis submit that these figures indicate a defect in Balmoral’s horizontal tank designs (increasing in seriousness with the size of the tank) for which the only explanation is that the stresses at the failure sites were too high. They point out that Balmoral have produced no evidence that would permit an analysis of their engineering design for the tanks and no relevant engineering calculations. In the light of those figures Balmoral should either have changed the design of the mould to reduce stress concentrations at failure sites, or increased the thickness of the tank walls either generally or at failure sites, and continuously monitored the result by a form of testing which would accurately reveal the wall thickness distribution in critical areas. In fact they did none of those things.

In October 2004 Dr Clemens and Dr Wright of Rapra visited Balmoral in Aberdeen and selected from a large number of failed tanks stored in a field 14 sample tanks: 1 made from MFR 3 ZN, 5 made from MFR 3 Borecene, 7 from MFR 6 borecene, and 1 from MFR 4 borecene. They cut out sections around crack sites, and took wall thickness measurements of those sections. A table of the maximum and minimum thicknesses and the max/min ratio of these samples is at paragraph 3.8 of their first report. In only one case – an SL 1250 tank - was the minimum wall thickness less than the OFTEC 4.5 mm and that was not at the point of failure. Cracking was found to coincide with minimum measured wall thickness in four of the failure sites examined (Footnote: ¹³⁴). The results showed significant wall thickness variations within individual tanks and as between one tank and another. The extremes of wall thickness were always located at a change in section area, with the thickest part usually being at an external, and the thinnest at internal, corners, i.e. the phenomenon shown in the page 37 photograph. All but two of the samples exhibited thick to thin ratios in excess of 2:1.

Figures 3.1 to 3.3 to Capcis’ first report contain photographs of typical locations of cracks. In the case of the slim line tanks a typical location is at the interface between the two halves of the mould where there is a “step” between the material on different sides of the join. An example of a section showing this was put before me at the hearing. It appears to me clear that there is, indeed, what could be called a step which is probably due to a misalignment of the two halves of the mould. The Capcis figures show that in the case of most of their samples the minimum thickness of the sample (Footnote: ¹³⁵) was below or very close to 5.5 (Footnote: ¹³⁶) and that there were considerable variations in wall thickness in the areas of the samples. Mr Richard Mawrey, Q.C., criticised Capcis’ reliance on these figures on the basis that they had taken failed to take measurements immediately adjacent to the crack site, and that the appropriate measurement would have been to take a section along the line of the crack. I do not accept the validity of this criticism. Such a measurement (which might well have been relatively uniform) would not measure variations in thickness around the area of the crack which affect the stress concentration at that point.

By the time of their second report Capcis had updated figures for their creep rupture testing. I have referred to this at paragraph 220 above.

CAPCIS arranged for finite element analysis to be varied out by Wilde FEA Ltd on two Balmoral tanks designs – SL 1250 and H 2500 - these being Balmoral’s two biggest sellers and the tanks with the biggest failure rates. Finite element analysis is a computer based method, originally designed for the aviation industry, of estimating the stresses and deformation likely to be experienced in a particular product design. Balmoral made use of FEA when designing the tanks. The process was, however, considerably less sophisticated, and slower, than it is now.

In the FEA exercise carried out by Wilde the tanks were assumed to have, in one exercise, 5mm and in another, 9.5 mm wall thicknesses at full load conditions in both cases, and, also, at half load in the case of the 5 mm wall thickness. The exercise therefore assumes rotomoulding perfection so far as wall thickness distribution is concerned. In reality tanks will not have uniform wall thicknesses and thickness variations will concentrate stresses and increase the potential for failure. An attempt was also made to model a variation in wall thickness at corners by taking a 5mm thickness with 9.5.mm at “rounds” and vice versa. Capcis regard the results as to be treated with caution and I do not propose to rely on them.

The simulations were run under linear-elastic conditions (Footnote: ¹³⁷) with certain assumed properties of polyethylene one of which was a modulus of 700Mpa (Footnote: ¹³⁸). These conditions do not reflect reality since polyethylene does not behave in a linear manner i.e. stress and strain are not directly proportionate. In consequence the resulting Mpa figures are greater than they would have been in reality at the upper end, i.e. from about 7 Mpa upwards, and conversely too low at the lower end. Further the simulation in the FEA exercise assumes that the base of the tank is fixed to the ground and cannot move in any direction when in reality the base of the tank may be capable of a little movement, although this will be limited because of friction if it is on a concrete base (more if the tank is supported on piers), and there can be bowing of the tank. As a result some of the stress readings close to the base would not be accurate, although I accept that the differences would not be significant. Stress contour plots (Footnote: ¹³⁹) were made with a cut off stress of 15 Mpa. This point was taken because polyethylenes typically yield at between 15 and 20 Mpa (Footnote: ¹⁴⁰) at 20° C. In reality stresses of the order of 15 Mpa or less would have led to immediate ductile failure.

In the case of the H 2500 at 5 mm wall thickness and full load, the plots indicated that maximum stresses were produced along edges (changes in section) between the stiffening rib sections and the wall, and at the ‘corner’ between sides and ends. Maximum stresses exceeded the minimum yield value of 15 Mpa in a number of locations, including sites where cracking had typically been observed i.e. around rib sections, and corners.

In the case of the SL 1250 tank at 5mm wall thickness full load maximum stresses were indicated along the edges between the sidewalls and the stiffening sections together with stress concentrations along the floor perimeter at the sides of the tank. Stresses exceeding the minimum yield value were indicated at a number of locations, including the edge along the lower horizontal rib where failures had occurred in service.

Capcis further record the existence, in the case of many of the SL 1250 tanks, of the “step” to which I have already referred, where there is a difference in thickness of around 2mm running vertically along one of the end walls. This feature is not reflected in the FEA analysis.

Capcis draw two conclusions from this data. Firstly the H2500 and SL 1250 designs, even assuming uniform thicknesses, would perform very close to the yield stress (Footnote: ¹⁴¹) of a typical PE material under static load. Wall thickness variations, and the step would increase the potential for failure. Capcis would have recommended reconsideration of the designs because of the high stress areas shown. (Footnote: ¹⁴²). Secondly the fact that several of the points of high stress are also points of failure where wall thickness variations would increase stress concentration supports the conclusion that tanks have failed because of the stresses imposed in consequence of those variations. Professor Crawford agrees with this analysis. The exercise tells you that the hot spots are placed where special care is needed to build up wall thickness so as to reduce the stress level, or that you should increase the wall thickness generally. If for some reason the stress cannot be so reduced a change in design (e.g. by changing the shape or the corner radii) would be necessary. There are several different ways of altering the manufacturing conditions, or the design, or both, to achieve the required objective. Capcis accept that the FEA analysis is not a perfect simulation of the tank material but say that it is sufficient to locate the areas of high stress and the fact that they coincided to a large degree with the observed position of actual tank cracking failures.

Balmoral suggests that this exercise tells us nothing of interest. Insofar as it reveals that the two most popular tanks failed at points of high stress that is no more than was to be expected. The mechanism of failure was environmental stress cracking; so failure was likely to occur at areas of higher stress. Those stress points were there when the tanks were made in ZN, and the tanks did not fail to anything like the same extent.

I am satisfied that the FEA results provide an instructive qualitative assessment of the location of the stress “hot spots” in the tanks, i.e. an indication of the relative stresses across the tank. They are not, however, a guide to absolute values. The fact that some of the readings exceed the minimum yield value of 15 Mpa is not, therefore an indication that, when filled the stresses will reach that figure; it is an indication of where the highest points of stress are to be found.

Capcis noted that when the method of pigment incorporation was dry mix or grind blended there was a very inhomogeneous structure. The pigment appeared in the form of tram lines within which lay the base resin particles so that there was a form of “schnitzel”: see figure 3.22c. The lines themselves were immediately surrounded by a transcrystalline structure (lighter in colour on the photograph – the same phenomenon can be seen on figure 2.35c surrounding a portion of masterbatch). A transcrystalline structure is known to be detrimental to physical properties. Where a tecroblend was used and the masterbatch was concentrated at the outer surface of the tank wall the transcrystalline structure was concentrated at the walls which would be detrimental to tank properties. In addition masterbatch particles were very angular leading to points of stress concentration. The BP Rigidex material, which Balmoral was using at the time of the report, was fully compounded and the structure was very homogeneous with no transcrystalline morphology.

But by the time of their second report Capcis had come to the conclusion that pigmentation was not a significant causal factor. I agree.

Capcis’ overall conclusion was that the two primary factors which had directly influenced the failures were:

Capcis rejected borecene as a causative factor because of the fact that it had been used successfully for oil tanks around the world (either in neutral or black) or, in the case of most of Balmoral’s tanks, green. If the resin was fundamentally at fault all Balmoral tanks manufactured would have experienced near total or very high failure rates. Their conclusion was that Borecene was fit for the purpose, if processed correctly, and if good design principles were employed.

Professor Crawford’s written evidence was to the effect set out in the following paragraphs.

The quality of a rotomoulded product is affected by a number of factors including:

Rotomoulding is an apparently simple, but in fact complex, manufacturing method involving a rotating mould and transient heat transfer phenomena that have a major effect on both the geometry and the properties of the end product. Few rotomoulders have accurate process control equipment on their moulding machines to monitor the temperature of the plastic or the mould, on account of the difficulty and expense of having such instrumentation on the rotating mould. As a result it is difficult for them to maintain the consistency of the product. Although operators will claim to use constant moulding conditions there are inevitable variations due to (i) start up of the machine, or shift changes; (ii) variations in ambient conditions and (iii) machine operator error (Footnote: ¹⁴³).

Moulders need to use their experience to adjust the cooling rate, which has an important effect on the properties of the moulded part, so as to obtain a part with acceptable tolerances. If process control equipment is not being used to maintain consistent cooling rates, tight quality control checks are needed to monitor the dimensions and properties of the moulded parts. Rotational speeds and ratios can also be adjusted to provide the desired wall thickness distribution. The role of the material supplier is to provide general advice about the likely behaviour of their material. It can be no more than general because the supplier will be ignorant of the details of the user’s moulds (which are likely to be unique to him), machines (of which there is a wide range of types), practices and settings. The responsibility for achieving the desired performance of the moulded part must lie with the moulder.

Normally, different moulding machine settings will need to be used when a material or grade change is made. Many manufacturers design tanks based on the current designs of other manufacturers’ tanks (thus potentially spreading bad design practice) with modifications; and then alter the designs to make improvements from time to time, without any real technical or design analysis. As a result they may have no real appreciation of the stress concentrations in their products. Tank designers should use FEA on new designs to ensure that the tank can withstand the stress concentrations inherent in its design.

Certain basic design rules are well established, one of which is that sharp corners or changes in section should be avoided because they cause stress concentration and can make it difficult to lay up the desired thickness of plastic, thus producing the phenomenon shown in Mr Clements’ page 37 photograph. With lower viscosity resins, such as borecene, speed ratios and moulding conditions such as oven time and temperature, are important if good wall thickness distribution is to be achieved at sharp internal corners. When switching from ZN material to borecene the moulder should know this.

Balmoral should have ensured:

Professor Crawford regarded his conclusions as confirmed by the fact that Balmoral had manufactured borecene tanks in designs other than H 2500 and SL 1250 which have had a very low failure rate; and that, even in those designs a large percentage are still performing satisfactorily. This shows that under appropriate moulding conditions borecene can produce good quality tanks.

As to the latter contention, Mr Jensen’s figures down to 31^st December 2005 show the following:

Professor Crawford fundamentally disagreed with the view expressed by Mr Clements (a) that once a speed ratio is optimised for a design of tank it then remains fixed irrespective of the grade of material subsequently processed;

In his supplemental report Professor Crawford identified the principal methods of controlling thickness distribution as:

Attention has, also, to be paid to charge weights. If, for instance, the manufacturer is working on minimum tolerances (i.e. has a charge weight as low as possible), the inevitable consequence is that there will on occasion be insufficient material at critical places to prevent ESC failure. If measurements are only taken in non critical areas, such as flat surfaces, and the weight is reduced to the minimum tolerance at those locations, an unacceptable level of failures will be inevitable. Self-evidently an increase in charge weight will, all other things being equal, increase the thickness of the product.

After Professor Crawford’s first report Balmoral served evidence about their processing parameters. The principal evidence was that of Mr Hay and Mr Ross. Mr Hay was the Production Manager until the end of 1996, but would not, therefore, have been involved in the production of tanks with borecene. Mr Ross was Production Supervisor over the period 1997-9. Thereafter, although remaining on the shop floor, he relinquished man management responsibilities and assumed a technical support role.

Mr Hay’s evidence, which I accept, was that when a mould was first introduced there was “a massive process of thickness checking, including destructive testing”. Tanks were cut up in order to examine the distribution of the polymer with the given charge weight. Part of the aim of the trials was to ensure that there was as little thickening and thinning at internal/external corners as possible. So far as rotation ratio was concerned, he would start with a 4:1 ratio as a starting point (which he usually found satisfactory), look at the distribution of the material and see whether or not it needed to be changed. If there was a thin area he would adjust the ratio so as to track that area through the powder pool. If there were areas with thick and thin sections, he would either use shields to reduce, or funnels or venturis to increase, heat and, therefore, thickness. If there was a change of material, Mr Hay would carry out the same sort of checking as he did with a new mould. In the early days he would even go through the full optimisation cycle for every material batch change.

As to speeds Mr Hay regarded rotation speed as a much less regularly used process control than ratios or shielding or funnelling. He found that the moulding operation ran better at lower speeds whilst higher speeds would give problems of vibration or excessive movement and throw the material around. He would never employ a speed above 5 rpm for the arm or 2.5 rpm for the plate, although he had experimented with arm speeds of up to 6, at which speed, he said, there was a risk of contact between the mould and fixed structures e.g. the oven doors, especially with big moulds and large charge weights.

Mr Ross indicated that speeds were not adjusted. When experimentation with speeds had been done it was found that a more even material distribution could be generated on the limited flat sections that could be measured by ultrasound, but the effect was not consistent, and he could not say what the effect was around internal protrusions which could not be measured with ultrasound. He also indicated that there were major concerns amongst engineering staff that at higher speeds, above 6 rpm, there could be mechanical damage to the mould or its components because of the greater vibration at those speeds.

Mr Ross’ evidence was that Balmoral used what he understood to be the industry norm of 4:1 for non-cylindrical tanks, and that that ratio was not something which Balmoral would consider changing without excellent reason. There were occasional trials during the last 8 years when other ratios were investigated but the traditional settings proved the best. However, the production specification sheets that have survived show something somewhat different. Some of these are undated; some are dated 2000 and 2001. Many of these have ratios of 5:1.

Mr Hay’s evidence was that, after initial optimisation, routine thickness testing was done by ultrasound. But there was regular use of a “lamp test”, where a powerful lamp is lowered into the product and the outer surface is scrutinised for “shine-through”. On some occasions, if so required, lamp testing was used in conjunction with thickness checking by cutting up the tanks into pieces. Mr Ross’ evidence indicates that cutting up tanks happened on relatively rare occasions during trials, and normally only if Balmoral was alerted to the need for it e.g. by lamp testing.

Professor Crawford expressed the opinion in his supplemental report that there was a fundamental failure by Balmoral to appreciate the importance of optimising processing parameters such as rotation ratios, speeds and cook time/oven temperature (Footnote: ¹⁴⁸) for the new borecene materials. In his view Balmoral certainly had scope for using higher speeds than 5 rev/min which is relatively slow. (Footnote: ¹⁴⁹) He found it surprising that Balmoral staff never found any justification for using speed ratios other than 4:1 (as he then understood to be the position) because it was “inconceivable” that that ratio could work for all tank shapes, and all positions and orientations of tank, and for all plastics, and because non-integer speed ratios give a much better distribution of plastic over the surface of the mould. He also said that it was useful to reverse the direction of mould rotation particularly when the powder is being deposited on the inside surface of the mould – otherwise a particular flow pattern of the powder over the surface of the mould gets locked into the system.

In Professor Crawford’s view the evidence as to Balmoral’s approach to processing reinforced his view that Balmoral did not use correct moulding conditions, and failed properly to monitor wall thicknesses by cutting up tanks and measuring the wall thickness in critical internal corner areas. Destructive testing should in any event have formed part of a competent moulder’s routine quality control checks.

Both sides make claims as to the deficiencies of the other side’s evidence. Balmoral submits that there is no scientific evidence to meet, or at any rate to match, that of Professor Pethrick; and that, whilst Borealis’ allegations of poor design and inadequate rotomoulding practices are long on generalities, they are short on specifics. Borealis retort that Balmoral have called no evidence to refute Professor Crawford’s evidence on rotomoulding practice, have produced little relevant evidence on design (Footnote: ¹⁵⁰), and have not attempted to show, by practical experiment, that borecene cannot make tanks with satisfactory wall thicknesses. Balmoral’s rejoinder is that Borealis have not produced any evidence of a manufacturer (or anyone else) producing a tank made from borecene with wall thickness variations within acceptable tolerances for tanks containing an ESC agent. (Footnote: ¹⁵¹). Borealis riposte that Balmoral have in fact made tanks with satisfactory wall thickness distribution (see the work of Walls and Macleod) and that a very large proportion of their tanks are still in service.

These submissions admit of more detailed discussion. I intend, however, to confine myself to two observations. First, the absence of a contradictor to Professor Crawford seems to me more significant than the absence of one to Professor Pethrick whose evidence does not, in my view take Balmoral far enough. Second, it would have been open to Balmoral to show, by practical experiment, that, using their machines and moulds, it was not possible or practicable to obtain an acceptable wall thickness distribution, however you change the parameters. I appreciate (a) the difficulties of proving a negative (albeit that is what Balmoral seek to prove) (Footnote: ¹⁵²); (b) that, had Balmoral conducted such an experiment it might have been said that, if only some change had been made which they did not in fact make, all would have been well, and (c) that there is more than one way of establishing Balmoral’s case. Nevertheless the absence of any empirical testing does seem to me of some significance, and of more significance than the fact that Professor Crawford was not able to specify exactly what should have been done in relation to each design by way of change of parameter, an exercise that cannot be done by reference to the design drawings alone.

Balmoral have, in my judgment, failed to establish that borecene was not reasonably suitable for the purpose of making green oil tanks by rotomoulding in that it was incapable of being used to make satisfactory oil tanks. On the contrary I accept the evidence of Professor Crawford that borecene is fit for that purpose. My reasoning is set out below.

I remind myself that the onus of proof in this respect lies on Balmoral.

I am not persuaded that borecene has been shown to be so defective in ESCR (or so low in viscosity) as not to be reasonably fit for the purpose of making oil tanks. Borecene has characteristics that differ from those of ZN materials, not least in the fact that it has a narrower molecular weight distribution than ZN material, two results of which are (a) that some grades of it will have a lesser ESCR than some ZN material; and (b) that it is less viscous than ZN material. But that does not mean that it is defective or unsuitable

Any conclusion on ESCR ought, if possible, to be based on satisfactory empirical testing of borecene in the relevant environment i.e. diesel/kerosene, from which quantitative conclusions can de drawn. That would be the best evidence as to what is likely to happen in practice. It is, no doubt, for this reason that Rapra carried out such extensive testing in these mediums in this and the DESO action. Their results do not, however, support the conclusion that borecene has very poor ESCR. As to them:

If the evidence as to the influence of the molecular structure on ESCR was sufficiently cogent or compelling, it might, either itself or in combination with empirical evidence, lead to the conclusion for which Balmoral contends. It is not, however of that character. I approach the evidence with a degree of caution since, as Borealis submits and as I accept, the ESCR of a polymer is largely dependent on the density of tie molecules, which in turn is dependent on a number of aspects of the polythene’s molecular and crystalline structure, such as molecular weight distribution, density of short-chain branches, whether those branches are located on small or long chains and the length of those branches. This list of factors I have taken from the article of Lu and Ishikawa cited at paragraph 177, footnote 74. That and other literature causes me to apprehend that it may be dangerous to draw conclusions from research that is not directed to the particular polyethylene in question and its peculiar characteristics.

Where the evidence has embraced measurable criteria in relation to borecene, it does not show that borecene has defective ESCR. Thus Borecene is shown to have figures for Mn, Mw and Mz which comfortably exceed Professor Pethrick’s figure for Mc; and an Mw figure which, depending on the grade of borecene, either exceeds, equates to or is not greatly inferior to BP Rigidex and Dow.

That borecene, had lesser viscosity than ZN material at the temperatures used in rotomoulding was well known to Balmoral. It gives borecene certain advantages, put forward by Borealis in its marketing statements in the form of ease of application (Footnote: ¹⁵³) and distribution of the polymer over the surface of the mould and the readier disappearance of bubbles in the melt – and thus reduced sintering/heating time. As Professor Crawford points out in his supplemental report (paragraph 50):

“it is accepted in the rotational moulding industry and in the rotational moulding scientific community that the desired characteristics of a rotational moulding resin are low zero shear viscosity and low elasticity” (Footnote: ¹⁵⁴).

Reduced viscosity is not, however, an unqualified benefit. The fact that the material is more likely to flow needs to be taken into account in determining processing parameters in order to ensure that this characteristic does not prejudice the achievement of satisfactory wall thicknesses and satisfactory wall thickness distribution in areas known to be difficult such as internal protrusions. Viscosity can be significantly altered by changes in manufacturing temperature (according to Professor Crawford by a factor of two within the normal moulding range).

Nothing in the evidence persuades me that the viscosity of borecene is so low that it cannot be used satisfactorily in the rotational moulding of oil tanks. In particular I do not regard it as established that, because of its rheological properties, it is not practicable to use borecene to rotomould tanks so as to achieve an even wall thickness. The page 37 photograph shows that using borecene can result in unacceptable wall thinness and unacceptable wall thickness variations. It does not, however, establish that that was a consequence inherent in any use of borecene. I accept Professor Crawford’s evidence that the viscosity of Borecene is more than adequate to provide the necessary wall thickness if correct rotational speeds and ratios are used (Footnote: ¹⁵⁵).

I do not reach that conclusion simply on a want of evidence, although there is, in my view, force in the submission that it is significant that Balmoral have not adduced evidence from an expert rotomoulder to the effect that borecene cannot practicably be used to rotomould oil tanks. Nor have they adduced any experimental evidence to the effect that, however you change the parameters, an uneven wall thickness will always result.

I accept Professor Crawford’s evidence (and that of Dr Clemens) that a competent rotomoulder would realise:

I further accept Professor Crawford’s evidence that appropriate quality control checks (post initial trialling) should have picked up the fact that tanks were being processed with localised thinning such as is shown in the page 37 photograph. Professor Crawford observed that photographs such as these would come as no surprise to a competent rotomoulder who would know that internal protrusions are a classic region where thinning will occur and that appropriate moulding conditions must be used to achieve the necessary thicknesses.

When, in 1997 Balmoral changed both material (from ZN to metallocene) and method of pigment incorporation (to dry blending) it did not carry out a re-optimisation process with respect to wall distribution. Joint trials took place on 10^th April 1997 at which, as Mr Shorter noted, the tanks made from DSM material were thick at the top and thin on the bottom. The trials were not progressed further because of problems of tool mounting and axis of rotation. In house studies were to take place before re-evaluation. In the event no further trials were conducted with Borealis.

After borecene had gone into production, Dr Walls spent several months in the absence of Borealis, optimising the wall thickness distribution of tanks made from borecene with reduced charge weights, in order that such tanks could be sent to Oftec for testing. In the course of this inspections took place of the wall thickness distribution of the tanks produced. On 3^rd July 1997 Dr Walls wrote a memorandum to Mr Tanner, the then Managing Director, Rotomoulding, which contained the following:

“It is likely that OFTEC will be concerned about the thickness of some of our tanks. This has been an issue in the past. In fact OFTEC have not actually tested the thickness of our tanks before, and have expressed concern about the thickness of our tanks already. In general our tanks our [sic] within spec, but the horizontal tanks have localised thinning around the inner rib area. Also, the smaller tanks are thin around the pockets at the bottom of the tanks. To resolve this problem, the rotation ratios were changed so that the tanks that were sent to OFTEC have better thickness distributions than our previous standard tanks”. (Footnote: ¹⁵⁸)

Dr Walls had clearly found that there was localised thinning in at least two areas and used a change of rotation ratio to improve the distribution in those areas. But, whatever changes he made when using lower charge weights, no change was made to the production settings. Balmoral decided not to go ahead with lower charge weights and the production settings remained as they had been with DSM material.

Borealis had a quality control system. This involved, amongst other things, a lamp test for each product. It also involved (according to the July 2001 update) testing for wall thickness of the first product made after each tool change and of the first product of the Sunday and Tuesday night shifts during each week that the tool was being used. This testing involved checking by ultrasound that a minimum thickness had been obtained at certain predetermined points on each particular design of tank. But it did not involve destructive testing or measurement at corner areas where the ultrasound would not function. Nor was the test concerned with changes in thickness, provided that the requisite minimum was achieved at the designated points. Nor does it appear to have focused on areas in particular designs where failures had been experienced. I accept Professor Crawford’s evidence that this is a valid criticism of the procedures. Failed tanks were coming back in ever increasing numbers, despite the quality control system. This was an indication that that system was defective and needed improvement e.g. by destructive testing. I note that Balmoral had been receiving complaints in 1998 and 1999 about cracked tanks not made in borecene (Footnote: ¹⁵⁹).

MFR 6 borecene was the subject of trials in November 1998 and Balmoral was supplied with the Borecene processing guide, which explains that:

“in order to achieve the desired even wall thickness in the finished article the rapid melting of borecene compared to that of conventional materials should be taken into account”.

It recommends an increase in rotation speed and optimisation of the rotation ratio, without both of which “uneven material distribution can occur”. At the trial in November Mr Halvorsen of Borealis told Mr Macleod that MFR 6 borecene required higher than normal speeds in order to distribute the material more evenly because:

“High Melt Flow materials at low speeds tend to leave more material in pockets and flow away from peak areas because of the very low viscosity”.

Mr Macleod ran a trial increasing speeds (and altering the rotation ratio) from 4:1 to 7:1. and the results showed an all round improvement in wall thickness, compared with MFR 3, of approximately 1mm. Mr Macleod’s conclusions included the following:

“3.4.

The higher melt flow material does result in an improved material distribution making charge weight reductions possible but not to the levels stated.

3.5.

Higher rotational speeds are required when moulding with ME 8161”

But, when borecene MFR 6 went into production in January 2000, the rotational speeds were not increased, nor does there appear to have been any change in the processing parameters or any check (other than the quality control system, with its limitations, referred to in paragraph 309 above) to see that appropriate wall thickness distribution was being achieved. I accept Professor Crawford’s evidence, which corresponded with what happened in November 1998, that Balmoral could safely have increased speeds to around 7 rpm (Footnote: ¹⁶⁰) and possibly more and could have benefited from doing so. If for some reason it could not have done so, it would, in the light of Borealis’ specific advice, have had closely to monitor the quality of the product manufactured using lower rotation speeds in order to determine whether, by changing other parameters, it could achieve a satisfactory wall thickness distribution particularly in the area of internal protrusions. If it could not do so, it would have to cease using the material.

Balmoral submit that the expression “optimisation of processing parameters”, apart from its inelegance, embraces only a limited number of types of change, such that it is unrealistic to attribute the failure of the tanks to a failure to optimise and that, in any event, appropriate optimisation was carried out, as is apparent from their moulding specification sheets.

As to the former, the permutations of charge weight, temperature, oven time, speed, speed ratio, rotation reversal, shielding and heating, the position of the mould on the arm, dwell time etc are, I accept, such as to enable rotomoulders, by what may, in some cases, include an element of fine tuning to secure a satisfactory product.

As to the latter the specification sheets show a change in parameters on occasion, because on some of them figures in respect of some entries have been crossed out and replaced by others. What the changed entries do not explain is the reason for or the effect of the change. There is no basis for inferring that checks made by Balmoral picked up wall thickness variations in corners or at protrusions or that changes were made in an attempt to remedy those defects. The standard testing by ultrasound did not reach these areas.

In January 2002 Balmoral switched to MFR 4 borecene and to fully compounding the pigment. There is no evidence that any process of re-optimisation in the light of the change in material and method of pigment incorporation took place, and I conclude that it did not.

In those circumstances I decline to conclude that the rheological properties of borecene are such that it is not practicable to use it to rotomould tanks because it is incapable of delivering an adequate thickness of wall or an adequate wall thickness distribution. On the contrary Balmoral, as it seems to me, failed to take the necessary steps to optimise their processing parameters and monitor the quality of their production, particularly in what it should have recognised as problem areas, such as corners, edges and protrusions, where there were stress hotspots and a likelihood of unsatisfactory wall thickness distribution.

I regard my conclusions on ESCR and rheological properties as reinforced by, and at any rate consistent with, the fact that certain of Balmoral’s designs have failed to a markedly greater extent than others (Footnote: ¹⁶¹), that, in general failure has occurred at specific locations in respect of each design, regardless of material, and that stress hotspots were often coincident with areas of failure. I do not regard these latter considerations as conclusive. It is possible to postulate a material so defective that it cannot, practically speaking, be used to make oil tanks, whatever design or process is used, so that tanks made from it, despite perfectly satisfactory processing, fail, not wholly surprisingly, at high stress points. But in the light of all the evidence I regard this as an unlikely explanation.

I prefer the simpler explanation that the tanks failed because, partly as a result of their design, they were subject to stresses for which the method of processing did not adequately cater (Footnote: ¹⁶²). Whether that is to be characterised as a design or a processing fault, or both, does not matter much. Design and processing are interacting factors. Stress hotspots thrown up by a design can be eliminated or reduced by changes in that design e.g. by providing a more generous curvature at corners or a less severe change in sections. Alternatively they can be addressed by a change of processing parameters to produce greater wall thickness and better wall thickness distribution. If, whatever the processing parameters, the stresses are so high that failure will ensue, the fault is one of design. If they can be addressed either by a change in design or an alteration in manufacturing conditions the fault can be put in either category, or both. If the change needed is a change in wall thickness, either generally or in specific places, that could be characterised as a change of design or one of process. Professor Crawford and Dr Clemens thought, and I accept, that, even if there was no design change, in the sense of a change in the geometry of the tank, tanks with satisfactory wall thickness distribution could have been made in borecene with appropriate rotomoulding processes (Footnote: ¹⁶³).

The absence of detailed design evidence from Balmoral means that we do not know on what stress calculations Balmoral’s designs were originally assessed as adequate, and with what safety margin. In the early stages wall thicknesses were reduced from a proposed 8 – 12 mm level. Reductions in wall thickness have a “dramatic” effect on the lifetime of a tank. The fact that there was, with ZN material, a 2.96% failure rate on average is an indication that with some of the designs there may have been no, or no adequate, safety factor for a product that had no redundancy. A change to borecene without optimisation may well have crossed the line leading to failures on a logarithmic scale, since crack growth rates accumulate on such a scale. Small changes in stress can produce large increases in failure numbers.

If, as Balmoral submit, borecene was intrinsically defective, I would expect to see a more widespread failure pattern by now, in terms of the proportion of tanks manufactured (as at 31^st December 2005 over 85% remain in service), of tank types affected, and in the locations of failure. The high incidence of failures in the SL 1250 and H 5000 designs strongly suggests a design weakness; and the fact that only a proportion of those tanks have failed points to inconsistencies in the manufacturing process.

In short I accept Professor Crawford’s evidence that a competent rotomoulder would be able to produce tanks from borecene with acceptable wall thicknesses. Others managed to do so, including Dr Walls, using lower charge weights, and Clarehill.

I have considered whether Balmoral are entitled to a robuster approach. They say, not without force, that they were given to understand that borecene was an improvement on previous ZN materials, with which they had, successfully (in their view), been making oil tanks. They were not told that they would have to make radical or significant changes to their processing parameters (or designs) or a change of pigment. Nor were they told that it would be more difficult to achieve even wall thickness with borecene or of the risk that wall thicknesses would reduce and variations increase to a point where ESC would occur. So they were entitled to assume, and did assume, that they could proceed as before, apart from any necessary tweaking of curing time. If borecene was only suitable for making oil tanks after an elaborate exercise to change either the design of the tanks or the processing parameters or both, then it was not reasonably fit for their purpose of making oil tanks without such changes.

I reject this. In deciding whether a product is reasonably suitable it is necessary to consider what a reasonable purchaser of the product could be expected to do. In the light of (i) Professor Crawford’s evidence (and that of Dr Clemens) as to what a rotomoulder should do, (ii) what Balmoral did in fact do (Footnote: ¹⁶⁴), I do not accept that Balmoral were entitled to proceed on the basis that, on switching to what they knew to be a new generation of polyethylene with different characteristics intended to produce better performance (the efficacy of which for them would, as they knew depend on trials (Footnote: ¹⁶⁵)), they could assume that borecene required no material change of processing parameters (or designs). Nor is it established that radical changes were needed, if by that is meant anything outside what Professor Crawford described as “the normal exploration of changes that would occur if you are switching from different grades of polyethylene”. The tenor of Professor Crawford’s evidence was that the alterations needed would not have been radical. It may be that borecene called for a tighter control of the processing operation. But that did not make it unfit for purpose.

Mr Mawrey relied on the fact that prior to changing to borecene Balmoral had produced satisfactory tanks with ZN material, but, when it changed to borecene, tanks produced by substantially the same methods as before, and with the same designs and moulds, suffered catastrophic failure. Other companies had had similar problems. In those circumstances the prima facie inference was that it was the material that was at fault. Was it realistically to be supposed that several rotomoulders had all independently got their processes (or designs) wrong? And what exactly should they have done?

I do not regard these considerations as persuasive. The fact that Balmoral made no change in their production methods does not seem to me to count in Balmoral’s favour. As to other companies I indicated, at the pre trial review, that I did not propose to admit some evidence then proposed to be tendered from other companies upon the grounds that, without a full investigation of the relevant circumstances, it would lack sufficient probative value, and that such an investigation would at worst involve litigating their cases for the purpose of this one and, at best would unduly prolong the trial.

Absent such investigation I do not derive much assistance from the fact that other rotomoulders have suffered similar problems; and such necessarily limited information as has come out during the trial does not suggest that further inquiry would have made any difference. My experience of this case indicates to me that each user is likely to have particular distinguishing characteristics. The only other claimant of substance in Great Britain is DESO, which is not large, and had made very few (about 50 -60) tanks before it began to use borecene, which it began using in 1997. They used both MFR 3 and MFR 6. Some of its wall thicknesses were, Rapra discovered, less than 3 mm at some locations. Rapra’s expert evidence in that case does not support Balmoral’s claims about the ESCR of borecene. The two Irish companies, Carbery and Kingspan, are large companies. In the case of Carbery, and, it may be, Kingspan the tanks have failed in the upper regions and the crack opens up in a form of fish mouth, suggesting a considerable residual stress. Carbery started buying borecene in May 2001 and Kingspan in early or mid 2002. They complain of increased failures from late 2002 in the former case and not later than 2005 in the latter. Clarehill had trialled tanks with borecene MFR 6 in February 1999 using a speed ratio of 8:1.4 instead of the 6:1 normally used with Borealis’ ME 8154 (ZN). In respect of a 1200 litre 40 kg tank the result of the change of ratio and adjustments to the temperature and oven cooling time was a 40% reduction in cycle time (approx) and improved material distribution. (Footnote: ¹⁶⁶) In May 2005 a representative of Clarehill confirmed to Borealis that borecene-made tanks had been in the field for 3-4 years and that Clarehill had suffered only background failure rates and that all failures had been identified as not being due to material issues (Footnote: ¹⁶⁷). Clarehill had a failure rate significantly lower than Balmoral’s ZN failure rate – about 0.5% as at 30^th November 2005 (Footnote: ¹⁶⁸).

It may be that, when the full facts are known and analysed, some or all of these manufacturers will prove to have good claims or the reverse. I do not find that the fact that they have unresolved claims of any substantial assistance in resolving Balmoral’s claim. Lastly I do not regard as fanciful the possibility that other rotomoulders have failed to optimise their processing parameters on a change of material.

Mr Mawrey invited me in his final oral submissions to consider the position in respect of the tanks that had not yet failed. In respect of those tanks, all made by the same process, there were two possibilities:

I do not derive much assistance from this conundrum. It does not, and is not designed to, provide any assistance to Balmoral’s case on ESCR. The selective ESC failure of Balmoral’s borecene-made tanks predicates a cause or causes which apply to some tanks but not others. If the cause is an inherent characteristic of the material, such as very poor ESCR, the proposition meets the difficulty that the mechanical properties of a material are a constant. If all the material had this defect, borecenes could be expected to have a very widespread failure.

I have no difficulty in accepting that all tanks are likely to have wall thickness variations and that the extent to which any particular tank does have such variations and fails on that account depends on (a) the extent and location of those variations; (b) the pattern of stresses actually imposed on the tank in question over its service life; and (c) the age of the tank.. Items (a) and (b) will differ for every tank and item (c) for most of them. As to (a), even though the processing parameters may be or purport to be the same, the precise thickness and composition of the polymer at different points will differ not only from batch to batch but from tank to tank. Further there may be minor, but important changes or adjustments in those parameters, e.g. in shot weight actually put in, temperature, and timing as between one tank and another. As to (b) the extent to which a particular tank is kept filled, or exposed to sunlight, (and therefore the extent and location of the stresses imposed), will differ from user to user. As to (c) Mr Clements has himself pointed out that “nearly identical products under nearly identical service conditions will exhibit a range of durability that is rarely less than a factor of 5” (Footnote: ¹⁶⁹). In those circumstances I see no reason to infer that the cause of the failure must be a rheological defect of random effect.

Lastly, Balmoral place reliance upon the fact that failure rates increased when borecene MFR6 was used – a circumstance which they submit points towards the material being the cause of (increased) failure. Miss Hassell’s statistics of borecene failures down to 31^st March 2005 (see JS page 42) are as follows:

In the light of the evidence I do not regard the fact that there was a greater degree of failure with MFR 6 as establishing the deficiency of borecene. It seems to me equally consistent with the fact that failure to optimise production had more serious consequences with the less viscous MFR 6 than it did with MFR 3 and MFR 4.

In the light of my conclusion on fitness for purpose, the question of misrepresentation does not arise, since, in that event, Balmoral does not advance a separate case in misrepresentation. The dispute between the parties as to the contractual position is immaterial. But, since this case may go further, and since the contractual issue has been the subject of extensive evidence and submission, I set out my conclusions on it below.

In the period between 1994 and mid 2002 Balmoral made nearly 400 purchases of polyethylene from one Borealis entity or another, in respect of which Borealis rendered invoices with their standard terms either on the back, or as one of the pages, of the invoice. From December 1995 Balmoral’s purchase orders referred to Balmoral’s terms.

Balmoral’s terms contained the following condition:

“8.1.1.

It shall be a material condition of the contract and the Seller shall warrant that the Goods …will be of satisfactory quality and fit for the purpose held out by the Seller or made known to the Seller in Writing at the time the purchase order is placed.”

There are no customary terms in the polymer trade in the United Kingdom in the sense of terms which are so “certain, notorious and reasonable” that anyone purchasing polymer must be taken to have contracted on those terms, unless expressly excluded or otherwise expressly agreed. But suppliers of polymer in the UK habitually seek to sell on their standard terms and conditions. I use the expression “seek to sell” because the fact that suppliers have standard terms does not mean that they are always successful in incorporating them. In many cases those terms exclude one or other, or both, of the warranties of satisfactory quality and fitness for purpose, particularly the latter, and limit liability to replacement of defective goods or repayment of the purchase price. But suppliers’ terms differ. Borealis UK’s terms excluded both warranties. Borealis’ General Terms only excluded the latter, but allowed for a warranty of fitness if Borealis had approved such purpose in writing (which Borealis UK’s terms did not). Both had the limit of liability to which I have referred.

Borealis have put in evidence the standard terms of five polymer suppliers, in addition to their own, which, whilst they exhibit similarities, also display differences only some of which appear in the following table:

Mr Joyce was aware that raw material suppliers did not give fitness for purpose warranties in respect of polymer (Footnote: ¹⁷¹).

In the present case Mr Mawrey submits that there were, in effect two parallel universes: the “real world” in which the parties moved and had their being, and an artificial world created for them by their lawyers when, but only when, a dispute arose. In the real world, as he submits, none of the individuals who were doing business with each other on behalf of Balmoral and Borealis paid any attention to the terms and conditions that the lawyers had drafted for them. Balmoral’s terms were adverted to in the bottom left hand corner of its purchase orders, but were never otherwise referred to nor provided to Borealis. Borealis’ terms appeared on the back of their invoices but were never discussed, either when complaints were made about blooming and warping, or when compensation was agreed, or when the dispute about the suitability of borecene arose. It was only when the lawyers came on the scene that the parties were transposed to an artificial world where reliance was placed on standard terms and a supposed change in contractual party (on two occasions) that the Claimant is said to be taken to have accepted, but never did.

The Court, in his submission, should prefer the real to the artificial universe and hold that neither party agreed to contract on the other’s standard terms. Alternatively, if the question has to be determined by a conventional application of the concepts of offer and acceptance, Borealis’ terms were never incorporated. Although it is true that Borealis’ terms were set out on the back of its invoices, all of these invoices post-dated the making of the contract and did not form part of the documentation constituting the subsequent one.

Mr Wood’s evidence (Footnote: ¹⁷²) as to the course of business, which I accept, was that, when an order was received, someone at Borealis UK in Wilmslow, after checking that the price on the order was the same as appeared on a price list that Mr Wood had given them, would check with the supplying plant that delivery could be made. If so, someone at Borealis would confirm the order (probably by telephone) to someone at Balmoral. Delivery would take place and within a couple of days Balmoral would be invoiced.

Accordingly, so Mr Mawrey submits, the offer was constituted by the purchase order, which was accepted either by the verbal confirmation of the order or by delivery of the goods. The invoice with the terms on the back came after each contract was made. That was not a counter offer. It was a claim for payment under the contract that had just been performed. Its dispatch at that stage did not alter the terms of contract already made, or those of the next contract, either on the first or any of the subsequent occasions, however many times the sequence was repeated.

There was, he further submits, no relevant course of dealing prior to 1997 when borecene was first purchased. Balmoral’s purchases pre 1997 were too few in number to constitute a course of dealing applicable to the dealing in borecene of a much greater order of magnitude that began in 1997.The inchoate negotiations for the Industrial Supply Agreement, although leading to no concluded agreement, signified to Borealis that its standard terms were not acceptable to Balmoral and precluded any inference that Balmoral accepted those terms.

Nothing changed in 1999 or 2001. The commercial contacts continued as before. Balmoral did not accept Borealis’ General Terms and Conditions any more than it had accepted those of Borealis UK. If Borealis’ conditions were not incorporated by being put on the back of invoices in 1997 and 1998, no more were they incorporated by this means in 1999, 2000, or 2001. As the letter of 3^rd November 2000 put it “Our commercial, day-to-day contact is not affected by this change in invoicing”.

Nor would the letters of November and December 1998 or November 2000 have signified to someone in Balmoral’s position that there was a change of contracting party, as opposed to a change of the invoicing and mode of payment arrangements relevant only for Borealis’ internal purposes.

Borealis contend that Borealis’ standard terms (first Borealis UK’s and then Borealis’ General Terms and Conditions) were incorporated into all its contracts with Balmoral. In Mr David Allen’s submission that incorporation was achieved by the following means:

Until January 1999, he submits, the contracting party was Borealis UK. In 1999 and 2000 it was the producing company, being Borealis Norway in the case of borecene and Borealis Sweden in the case of ZN material.

Whether or not one party’s standard terms are incorporated depends on whether that which each party says and does is such as to lead a reasonable person in their position to believe that those terms were to govern their legal relations. The Court has to determine what each party was reasonably entitled to conclude from the acts and words of the other: McCutcheon v David Macbrayne Ltd [1964] 1 W.L.R. 125,128; Kendall & Sons v Lillico & Sons [1969] 2 A.C. 31; Hollier v Rambler Motors (A.M.C. Ltd) [1972] 2 Q.B 71. The question is one of fact to which prior authority may form an uncertain guide.

Borealis relied on the Kendall case, supra. There the seller issued the buyer with a “sold note” shortly after each contract had been concluded. Nevertheless by the time of the transaction in question the terms in the sold notes were held to be incorporated in the contract. The trial judge had found that the conditions of sale were not incorporated. The Court of Appeal reversed the judge on this point, and the House of Lords agreed. In the Court of Appeal Diplock, L.J., as he then was, said this:

“Where, as in the case of the contract between SAPPA and Grimsdale, the parties have not agreed to embody their contract in a written document but have entered into an oral contract with the intention of thereby creating legal rights and liabilities and it is sought to rely upon a term contained in some written document as modifying their respective rights and liabilities which would arise by implication of law from the nature of the contract, the only question is whether each party has led the other reasonably to believe that he intended that the rights and liabilities towards one another which would otherwise arise by implication of law from the nature of the contract, namely a contract for the sale of goods, should be modified in the manner specified in the written document.

Upon the facts found by the judge the issue by Grimsdale of sold-notes containing “Conditions of sale” in respect of all previous sales to SAPPA should have led SAPPA reasonably to believe that Grimsdale were only willing to agree to sell goods to SAPPA upon the terms set out in the “Conditions of sale”. The conduct of SAPPA in accepting such sold notes in previous transactions as recording the sale without making any comment, query or objection about the “Conditions of sale” was conduct which would lead Grimsdale reasonably to believe that SAPPA intended to enter into a contract upon these terms. Whether Golden had ever read the conditions or what he thought SAPPA’s rights or liabilities under the contract would be is immaterial”.

In S.I.A.T. v Tradax [1978] 2 Lloyd’s Rep 470 Donaldson J., as he then was, observed that Diplock L.J. meant:

“that the words and conduct of Party A had to be considered with a view to what Party B, as a reasonable man, could and would, if he had applied his mind to the point, have been led to believe”.

In Circle Freight International Ltd v Medeast Gulf Exports Ltd [1988] 2 Lloyd’s Rep 427 exporters of dresses to the Middle East sued their forwarding agents, whose employee has negligently left the plaintiff’s van unattended in Fleet Street as a result of which the dresses had been stolen. The forwarders relied on their conditions, which excluded liability save in the case of wilful default and contained a limitation of liability. The conditions were printed on the back of the invoices, eleven of which had been sent for consignments that pre-dated the loss. The judge held that the conditions were not incorporated.

The Court of Appeal held that they were. Two relevant tests were identified, the first of which may, perhaps, be regarded as part of the second: (a) whether reasonable notice had been given of the forwarder’s conditions and (b) what had each party by his words and conduct led the other party reasonably to believe him to be accepting. It was not necessary that the terms relied on should be specifically set out, provided they were conditions in common form or usual terms in the relevant business, nor was it necessary that they should be contained in a contractual document. The parties were commercial companies. There had been 11 invoices which gave notice that business was conducted on IFF terms. The exporter’s managing director knew that forwarding agents might impose terms which would frequently be standard terms and would sometimes deal with risk. Accordingly reasonable notice was given by the forwarders and the exporters’ conduct would have led the forwarders reasonably to believe that their terms were accepted.

In Albright & Wilson UK Ltd v Biachem Ltd [20001] EWCA Civ 301; [2001] 2 All ER 537 the claimants ordered two different chemicals from two different suppliers who instructed the same company to deliver them. The driver of one consignment, by some mistake, presented the delivery note relating to the other. As a result his cargo was discharged into tanks containing the other product and a large explosion occurred. One of the issues was whether, in relation to one of the contracts, the standard terms of either party was incorporated. The order in question was the fifth in a series. The first order had been preceded by a quotation which was subject to the supplier’s terms and conditions of sale. After that the purchaser faxed a “blanket” order which stipulated that supply should be subject to the purchaser’s conditions. The order was confirmed by telephone. Delivery took place. The supplier then sent an invoice with his terms on the back. There were then four telephone orders which were followed by a standard form purchase order which called for supply on the purchaser’s terms and conditions. In each case an invoice with the supplier’s terms was sent after delivery.

Eady J., held (a) that the supplier’s quotation had become largely historical by September 1996 and could be regarded as an invitation to treat or, if an offer, matched by a counter offer in the subsequent order; (b) that a reasonable onlooker would determine that the contracts were governed by the purchaser’s conditions; (c) that an invoice subsequent to the contracts could not affect its terms and (d) that the fact that there had been several such invoices did not change the nature of the transactions. The invoice was too late to make a difference in every case. The Court of Appeal agreed. Circle Freight was not cited.

Similarly in Continental Tyre & Rubber Co Ltd v Trunk Trailer Ltd [1987] S.L.T. 58 manufacturers of trailers ordered a quantity of tyres from a tyre supplier on their printed purchase order which purported to incorporate their conditions of purchase. The sellers claimed that their terms applied because when the first batch of tyres was delivered one of the purchasers’ employees signed a delivery note with the words “All offers and sales are subject to the [seller’s] current terms and conditions, a copy of which will be supplied on request”. The Inner House of the Court of Session rejected this contention on the basis that the contract was complete as soon as the first batch of tyres was delivered. The signature of the purchaser’s employee on the delivery note had no contractual effect; it was not a contractual document and was only required for the purpose of confirming the quantity and description of the goods delivered.

These cases show two things. Firstly, at any rate where parties have dealt with each other more than once or twice, it may not be critical to the incorporation of standard terms that those terms be set out in a contractual document, i.e. one that itself constitutes an offer or its acceptance, or even in a purported record of the contract, nor that the document containing the terms relied on has preceded the making of every contract. Secondly, the sequence of events is important. An invoice following a concluded contract effected by a clear offer on standard terms which are accepted, even if only by delivery, will or may be too late. But, if there has been no reference to rival terms, the appearance of terms on the back of every invoice and the acceptance of delivery of goods without objection may indicate acceptance of the terms.

I am not persuaded that when Balmoral first contracted to purchase polyethylene from Borealis in December 1994 the common understanding between them was that the purchase was on Borealis’ terms and condition. In British Crane Hire both parties were crane hirers. They each hired out cranes on similar conditions. The defendants had signed and sent back the plaintiff’s conditions twice before; and would have done so again if the accident had not happened. In December 2004 Balmoral had not purchased material from Borealis before; nor had they ever signed or sent back Borealis’ conditions; nor, so far as the evidence shows, had they received a copy of those conditions when they sent their purchase order or took delivery. In those circumstances I do not think that Borealis was entitled to conclude that Balmoral had agreed to their terms.

But by the fax of 18^th January 1995 Borealis made it plain that its prices were quoted “subject to normal terms and to current conditions of sale” and Balmoral had received an invoice with Borealis’ terms on the back. By putting its terms on the back of their invoices Borealis indicated to Balmoral that, so far as Borealis were concerned, they intended the contract to be on those terms: see clause 2 cited at paragraph 30 above. Those invoices were seen and initialled by Mr Joyce, the Managing Director of Balmoral’s rotomoulding division. He realised that there were terms on the back of the invoices. He reviewed them but did not study them. (Footnote: ¹⁷³) Balmoral thereafter purchased material at the quoted prices with knowledge of Borealis’ conditions without ever suggesting that they were not applicable. Borealis did not know that each invoice in respect of borecene was initialled by Mr Joyce before payment. But they did know that they were received and paid without demur. A reasonable person in their position would be entitled to assume, as was the fact, that someone at Balmoral had considered whether or not to pay the invoice and had seen the conditions on the back.

In those circumstances Borealis was, in my judgment, reasonably entitled to assume that Balmoral accepted the applicability of the conditions subject to which the price had been quoted. If Balmoral wanted to pay the price quoted without accepting the terms, it was incumbent on it to say so.

Mr Mawrey submitted that the prices quoted were determined by market and trade conditions (the cost of polymer, prices charged by competitors, the size of the order, etc) and were not dependent on acceptance of Borealis’ terms. Prices were agreed in negotiation between Balmoral and Borealis and, absent an express acceptance of Borealis’ terms, the fax of 18^th January 1995 did not make those terms applicable.

Even if the first half of the submission is correct, the second is not. Firstly, Borealis would probably not have been prepared to supply its product at the price quoted or any price, if its terms were not, as it believed, accepted. Secondly, the effect of the fax was to make clear that the quoted price was contractually dependent on acceptance of the terms. In any event it seems to me unrealistic to suppose that the prices quoted would not differ at all if Borealis could not, as it believed, rely on its own terms. Borealis has insurance in respect of the claims the subject of these proceedings. If Borealis had no terms and conditions the cost of that insurance is likely to have been greater than it was and to be reflected in the price. Even if that were not so, exposure to liability for fitness for purpose would be likely to impel Borealis to take measures to reduce their exposure (e.g. by examining in detail the uses, designs, processing parameters, terms of sale and the like applicable to each particular customer) which would themselves impact on the price.

Mr Mawrey also submits that the number of orders prior to the deliveries of borecene was too small to amount to a course of dealing, and that the supplies that constituted any putative course of dealing were “as different as can be imagined from the borecene supplies of 1997 onwards”. I accept that, if a contract has been made for the supply of a relatively small quantity of product on one set of terms on only a few occasions, it may, depending on the precise circumstances, be difficult to establish that later contracts for large quantities of a different product are also made on those terms. But, here, at the beginning of the trading relationship Borealis made plain that its prices were quoted, and it was only willing to supply, on the basis of Borealis UK’s terms.

The quote of 18^th January 1995 was swiftly followed by an order for 60 tonnes and the payment of Borealis’ invoices without comment, query or objection. That signified to someone in Borealis’ position acceptance by Balmoral of the applicability of those conditions. Thereafter Borealis gave further quotes for material.

Thus, when on 6^th December 1995 Balmoral faxed another order which contained a reference to its terms and conditions, it did so in circumstances where Borealis had supplied material and given quotes for future supply, to which, as Borealis was entitled to assume, Borealis UK’s terms and conditions applied. The order of December 1995 was, in my judgment insufficient to reverse the previous position and to signify (much less to secure agreement) that Balmoral’s terms were now to apply instead. Balmoral said nothing about any change in terms. It did not say that Balmoral was going to treat Borealis’ quotes as no longer linked to Borealis’ conditions. Balmoral did not provide a copy of its conditions then or thereafter, nor, apart from the stamp on the order form, make any reference thereto.

In those circumstances I do not regard the reference in poor typescript at the bottom of Balmoral’s purchase orders as reasonable notice that Balmoral proposed to contract only on Balmoral’s terms in place of those of Borealis. A supplier in Borealis’ position would be entitled to expect that, if Balmoral was seeking to reverse the previous position, it would make explicit reference to the fact that it was proposing a change. A buyer in Balmoral’s position would not be entitled reasonably to assume that Balmoral had agreed to its conditions because of the stamp that now appeared on its purchase order or because Borealis did not comment on it.

Borealis UK made further supplies of ZN material and then in April 1997 started supplying borecene. For each order it supplied an invoice setting out its terms and conditions, and these were paid, after checking, without demur. This sequence of events served to confirm the acceptance of Borealis UK’s conditions.

I turn to consider the position if, contrary to my view, Balmoral’s conditions are applicable. The terms, it will be recalled, provided that:

“2.0.

BASIS OF PURCHASE

2.1.

The purchase order constitutes an offer by the Buyer to purchase Goods and/or acquire the services subject to these conditions

2.2.

……..

2.3.

The purchase order will lapse unless unconditionally accepted by the Seller in Writing within 14 days of its date.”

The question therefore arises as to whether clause 2.3 was intended to prescribe a mode of acceptance so that the only means by which the offer contained in the purchase order could be accepted was by the seller unconditionally accepting it in writing. I do not think that the clause should be so construed. The effect of clause 2.3. is that the purchase order ceases to be available for acceptance if it has not been accepted in writing within 14 days. It does not, however, provide that the only way in which an acceptance is possible is in writing. The contracts between Balmoral and Borealis were, as it seems to me, made when Borealis, having discovered that the producing company could deliver, orally confirmed Balmoral’s order. Although the evidence is unspecific, I infer that this would probably have occurred within the 14 day period. Accordingly, on this hypothesis, there would have been an acceptance of Balmoral’s order, on its terms, when the confirmation was given.

I do not regard the brief and inconsequential negotiations for the Industrial Supply Agreement as leading to a different conclusion. Those negotiations did not address the question of the terms upon which the parties were currently trading (Footnote: ¹⁷⁴). Balmoral sought the inclusion of a warranty and the exclusion of a limitation, which would be consistent with their standard terms (never shown to Borealis) and inconsistent with those of Borealis. Negotiations for an overarching agreement petered out in circumstances where Borealis did not accept changes which would have been inconsistent with their standard terms. I do not accept that, in those circumstances, the parties are to be taken to have agreed that neither of their terms should be applicable, nor that those of Balmoral should apply.

The case of Britvic Soft Drinks Ltd v Messer UK Ltd [2002] 1 Lloyd’s Rep 20 is distinguishable. In that the case the parties had been contracting on the basis of terms contained in a written agreement which was due to expire. Before that expiry negotiations took place at management level as to the terms upon which the parties’ commercial relationship should continue after the expiry of the agreement. Although agreement was reached as to prices, no agreement was reached as to the other terms prior to the expiry of the agreement. Tomlinson J held that after the expiry of the agreement neither party’s terms and conditions applied. In circumstances where there were management discussions as to the terms that were to apply he declined to hold that different contractual terms had been agreed either as a result of delivery notes incorporating standard terms which were handed by Messer to Britvic employees, and signed by them, or as a result of financial staff observing (if they did) those terms on the reverse side of the invoices they were required to process. Accordingly the contractual vacuum was filled by the terms implied by section 14 of the Sale of Goods Act 1979.

In the present case there was no agreement whose expiry created a vacuum that the parties were negotiating to fulfil. The parties were contemplating, making an overarching agreement. Balmoral hoped for, but failed to secure, agreement to terms which would represent a change from the terms on which they were currently doing business.

Nor do I derive assistance from the fact that Borealis settled claims by Balmoral without reference to its terms. Settlement of a claim on a commercial basis, without reference to applicable terms and conditions, is a not infrequent occurrence, although the fact that exclusionary terms are not relied on when they could be may bear upon their reasonableness for the purposes of the Unfair Contract Terms Act 1977.

The letter of November 1998 explained that there was to be:

“a change from the reinvoicing principle through the local sales company to direct invoicing by the delivering production unit”.

Borealis’ terms and conditions were to be harmonised so that:

“irrespective of the source of delivery you will deal with a commercial partner who applies unified conditions with respect to commercial, insurance and legal matters”.

The letter of December 1998 explained that:

“Your local Borealis Sales Office will print and dispatch the invoices in the name of the supplying Borealis plant and continue to offer assistance on any issues relating to payment. Invoicing from different sites instead of invoicing through one local sales company necessitates your maintaining different supplier files. The creation of one bank account reference for all the Borealis sites will, however, soften the additional administrative effort. The planned second stage of the project will enable central invoicing, independent of local Sales Offices….”

These letters were drafted by Mr Jan Lindh, the General Manager of BCC, who gave evidence, and members of a committee of which he was the Chairman. Although some of their language betrays the fact that they were written by persons whose first language was not English (Footnote: ¹⁷⁵), their content appears to me to be clear enough, even though they are couched in wording that presents an important change of contractual partner and terms as a service improvement. Whereas previously the delivering production unit would invoice the local sales company which would re-invoice the customer, from now on that unit was to invoice direct. That would require customers to maintain different files, because they would be dealing with different suppliers. But that additional administrative burden would be lessened by the fact that, whoever was the supplier, the payment would go to the same bank account.

These letters cannot, in my view, be treated as nothing more than an indication that, for purposes internal to Borealis, the mechanics of invoicing would change, but the contractual position would not. Their natural meaning is that there was to be a change from the previous position, where the supplier contracted with the sales office, which contracted with and therefore re-invoiced the customer, to one where the supplier to the customer was to be the producing company – hence the need for different supplier files. It was equally apparent from the letter of November 1998 that there was to be a change of terms (“You will find the revised document attached”). As Mr Joyce said in evidence “that is what it says”. (Footnote: ¹⁷⁶)

Balmoral’s acceptance of these new arrangements as to contracting party and terms is apparent from the fact (a) that it raised no objection to the proposals set out in the two letters; (b) that it ceased to address its orders to Borealis UK but addressed them to Borealis Coordination Center NV; (c) that it made payment to Citibank conformable to the attachment referred to in paragraph 94 against invoices which, by their contents, indicated a change of supplier and a new set of terms and which described Borealis UK as a commercial agent.

It is noticeable that Balmoral did not at this stage, or in 2000/1, suggest that its terms were in any way relevant, a circumstance which confirms me in the view I have reached as to the applicability of Borealis UK’s terms prior to 1999.

As appears from paragraph 104 above I conclude that Mr Joyce probably did sign and return the February 2000 rebate agreement, which included an agreement to Borealis’ terms. Even if I am wrong on that it seems to me that, in circumstances where Borealis’ conditions had applied throughout 1999, the failure of Balmoral to object to the term in the rebate letter to the effect that they should continue to apply, and their claim to the 2000 rebate without objection, would have further conveyed to someone in Borealis’ position that Balmoral agreed to the application of Borealis UK’s terms.

I do not accept that there is any difficulty in concluding that from 1999 onwards the contracting party was Borealis Norway in respect of borecene and Borealis Sweden in respect of ZN either on account of (a) the wording of the preamble to Borealis’ General Conditions; (b) the reference on the second page to the invoice to endorsement to Borealis Coordination Center; or (c) because of the difficulty of knowing who would have been liable if Borealis UK had accepted an order, and then lost it without having contacted the Borealis supplier.

As to (a) the preamble referred to the products as being sold “by the Borealis entity given in the front page”. There are at least two (Footnote: ¹⁷⁷) entities on the front page: the entity which is invoicing (“Invoiced by”) and the Sales Office or Commercial Agent. In the light of the explanation given in the letters of November and December 1998 and the use of the words “INVOICED BY” it seems to me clear that it is the invoicing party which is the entity that is to be the seller. After Borealis UK was expressly described as Commercial Agent the position was even clearer. The terms themselves make clear that there is only one entity which is selling.

As to (b) the reference to the invoice having been endorsed to Borealis Coordination Center and as solely payable to that entity may constitute (as I think it does) a notice of assignment of the debt (Footnote: ¹⁷⁸), or simply as a direction to pay the Center at its bankers, (a question properly to be determined under the law of Norway or Sweden) (Footnote: ¹⁷⁹). Either way it is consistent with the invoicer being the contracting party.

As to (c), the question does not arise on the facts of this case but, if it did, I should have thought that the contract was apparently made with Borealis Sweden if the order was for ZN, and Borealis Norway if it was for borecene, since the former did not supply Borecene and the latter did not supply ZN. If Borealis UK had no actual authority to confirm an order for borecene without contact with the producing company it would have had apparent authority so to do. If not, Balmoral UK would be liable for breach of an implied warranty of such authority.

The language of the letters of November and December 1998 and the changes brought about thereby showed that the producing company was to be the principal, which would invoice Balmoral the price, and that Borealis UK was to act as agent. Borealis UK was, by the letter of November 1998, to be the contact for commercial issues, but noticeably not, as previously, the recipient of payment. The local Borealis sales office was, according to the letter of 11^th December 1998, to “print and dispatch the invoices in the name of the supplying Borealis plant” i.e. on its behalf. From 29^th April 1999 Borealis UK was described on the invoice as “Commercial Agent”.

The letter of 3^rd November 2000 informed Mr Joyce of the proposed change for 2001 onwards, namely that all material would be invoiced by Borealis Denmark and that in the future Balmoral would deal with only one supplier viz Borealis Denmark. Balmoral did not demur to that. Balmoral also made payment to the AMRO account (after a period of continuing to pay Citibank). Balmoral thereby accepted the new contracting party and the continued applicability of Balmoral’s general conditions.

On 17^th January 2001 Mr Forbes signed the 2001 rebate agreement, incorporating Borealis’ terms, containing a manuscript amendment that Mr Wood had initialled. That constituted, in my judgment, an agreement that those terms applied – as was already the case. Even if I am wrong on that Borealis was entitled to conclude from the fact (a) that Mr Forbes had signed the original version of the agreement; and (b) that Balmoral had not demurred when Borealis sent a typed up copy signed by Mr Webster but had continued to order goods, and ultimately to claim the rebate, that Balmoral assented to the applicability of those terms.

Applying an objective test I conclude that Borealis’ standard terms (first those of Borealis UK and then the Borealis General Conditions) applied throughout. The application of such a test may produce a result inconsistent with one party’s subjective belief (or lack of one) but that does not mean, and I do not accept, that the result is thereby artificial.

The Borealis UK terms provide, so far as relevant as follows:

The effect of these terms, subject to the operation of the Unfair Contract Terms Act, is clear. If the goods do not comply with Borealis’ standard specification Borealis is bound to take one of the steps specified in 13 (A) (i) – (iii) (Footnote: ¹⁸⁰).Clause 13 (C) (i) excludes the statutory warranties of quality and fitness for purpose on which Balmoral rely. Clause 13 (C) (ii) absolves Borealis from any liability for any loss or damage arising from any defect in the goods. Clause 13 (E) restricts any liability of Borealis to the price of the Goods in any event.

The Borealis General Terms provide, so far as relevant:

“These General Terms and Conditions of Sale (in the following “General Terms”) together with the order confirmation shall in so far no (sic) other agreement regulating the issue has been concluded in writing apply to all products (in the following “Products”) sold by the Borealis entity given in the front page (in the following “Borealis”) to its customers (in the following “the Purchasers”).

….

Clause 5 is concerned with liability which is imposed on Borealis by mandatory rules in respect of product liability i.e. rules in relation to damage to person or property that cannot be excluded by agreement. It is not relevant to the present case.

Clause 6.1 does not exclude the implied term (in English law) as to satisfactory quality. But it does exclude any implied warranty of fitness for purpose, unless Borealis has approved such suitability in writing. It was not suggested that Borealis had given such an approval in writing.

Clause 6.3. limits Borealis’ total liability under the contract to either an exchange of defective for non defective product or a refund of the invoiced value. Mr Mawrey contends that this clause cannot have effect if the defective product cannot be exchanged because, in that event, there is no option that can be exercised in one of two different ways. I do not agree. The clause gives Borealis two options. If only one option is open to it, it is that option that it will have to choose. The fact that it cannot chose the other does not mean that the limit on liability cannot operate at all.

The relevant statutory provisions are as follows. Section 6 provides:

“Sale and hire-purchase

Section 11 provides:

Schedule 2 provides:

Sections 26 and 27 provide:

“26.

International supply contracts.

(1)

The limits imposed by this Act on the extent to which a person may exclude or restrict liability by reference to a contract term do not apply to liability arising under such a contract as is described in subsection (3) below.

(2)

The terms of such a contract are not subject to any requirement of reasonableness under section 3 or 4: and nothing in Part II of this Act shall require the incorporation of the terms of such a contract to be fair and reasonable for them to have effect.

(3)

Subject to subsection (4), that description of contract is one whose characteristics are the following-

(a)

either it is a contract of sale of goods or it is one under or in pursuance of which the possession or ownership of goods passes; and

(b)

it is made by parties whose places of business (or, if they have none, habitual residences) are in the territories of different States (the Channel Islands and the Isle of Man being treated for this purpose as different States from the United Kingdom).

(4)

A contract falls within subsection (3) above only if either-

(a)

the goods in question are, at the time of the conclusion of the contract, in the course of carriage, or will be carried, from the territory of one State to the territory of another; or

(b)

the acts constituting the offer and acceptance have been done in the territories of different States; or

(c)

the contract provides for the goods to be delivered to the territory of a State other than that within whose territory those acts were done.

27.

Choice of law clauses.

(1)

Where the law applicable to a contract is the law of any part of the United Kingdom only by choice of the parties (and apart from that choice would be the law of some country outside the United Kingdom) sections 2 to 7 and 16 to 21 of this Act do not operate as part of the law applicable to the contract.

(2)

This Act has effect notwithstanding any contract term which applies or purports to apply the law of some country outside the United Kingdom, where (either or both)-

(a)

the term appears to the court, or arbitrator or arbiter to have been imposed wholly or mainly for the purpose of enabling the party imposing it to evade the operation of this Act; or

(b)

in the making of the contract one of the parties dealt as consumer, and he was then habitually resident in the United Kingdom, and the essential steps necessary for the making of the contract were taken there, whether by him or by others on his behalf”. 

These statutory provisions only become relevant if, in the absence of Borealis’ terms and conditions, there would have been a breach of Borealis’ obligations in respect of the quality or fitness for purpose of the borecene supplied in the respects claimed by Balmoral. On that assumption it is necessary to decide:

If so, when were the contracts in

The effect of the Borealis UK conditions is that, if Borealis deliver Borecene which is either of unsatisfactory quality or not reasonably fit for the purpose for which it was required. Balmoral has no redress, beyond, at best, the possibility of a replacement of the defective product or the return of its price. Is that a term which it is fair and reasonable to have included in the contracts?

Balmoral submits that it is not. It is, they suggest, neither fair nor reasonable for Borealis to include in the contracts terms which exclude liability for breach of the core obligations of satisfactory quality and fitness for purpose with the result that Borealis escapes any real responsibility for the unfitness of borecene for its purpose, which has caused Balmoral foreseeable loss on a massive scale.

Borealis, which has the onus of proof of reasonableness, points out that this case concerns major commercial companies, not consumers. Although Borealis is a much bigger concern than Balmoral, Balmoral is a company with a turnover of over £30 million a year.

They also rely on the difference of position between themselves and Balmoral. Borealis produces raw materials with a wide range of uses and applications across the world. Borealis can and does perform a number of standard tests on samples of its product; but Borealis cannot be expected to test it for all uses of any prospective purchaser. Purchasers of borecene will, in their manufacturing processes, change its form, and may change its properties, by applying heat to it and mixing it with pigment by differing methods of incorporation. Those purchasers are specialist manufacturers. Borealis is not. It is the manufacturers who will know how they intend to use borecene, what will be the design of the product made from it and of the moulds used to produce it, how that product will be engineered, to what stresses it will be subject, and what method of processing will be used, as well as the uses for which and the terms on which it will be sold. It is they who are the experts in working out whether the material is going to work for their purposes and then making things out of that material. The way in which the product is made (e.g. the amount of wall thickness and the quality of wall thickness distribution) will affect the quality of the product. Borealis habitually provides, and in the present case did provide, free samples to enable testing to take place. If asked to carry out further tests in order to obtain data relevant to an intended use of the material Borealis is likely to do so. In the nature of things Borealis will have little or no control over the manner in which borecene is used, and little or no knowledge, much less control, as to the extent to which, and the terms upon which, products manufactured from it are marketed and sold. Every use or application may give rise to a large, even enormous, claim against Borealis. The success or failure of such claims may depend, to a significant extent, on matters that are internal to the claimant (such as the precise method of manufacture) in relation to which Borealis may have limited knowledge or access to evidence. In those circumstances, it was legitimate for Borealis to protect themselves against that exposure.

Balmoral’s standard terms contain provisions whose effect is similar to that of the Borealis UK terms. Balmoral’s terms provide:

“4 (a) the only liability accepted by the Company for the repair or replacement of any defective goods is a liability to the original purchaser for the repair or replacement, free of cost to the customer, of goods of its manufacture which are established to its reasonable satisfaction to have been defective at the time of despatch and where such defect is reported in writing before the expiration of the defects liability period (defined below)

…

(c)

The expression “defects liability period” means a period, running from the date of despatch, of six months…

(d)

The Company shall be under no liability whatsoever for any form of consequential or other loss howsoever caused, and is to be indemnified by the customer against all liability in respect thereof.”

The rival contentions of the parties raise in stark form the question whether a major supplier of raw material can, where UCTA 1977 applies, successfully avoid liability (other than for the price of the goods or their replacement) for breach of the conditions ordinarily implied under the Sale of Goods Act 1979.

In Granville Oil v Davis Turner [2003] 2 Lloyd’s Rep 356. 362, Tuckey L.J. said:

“The 1977 Act obviously plays a very important role in protecting vulnerable consumers from the effects of draconian contract terms. But I am less enthusiastic about its intrusion into contracts between parties of equal bargaining strength, who would generally be considered capable of being able to make contracts of their choosing and expect to be bound by their terms. Here the transaction includes carriage of goods by sea and insurance. These spheres of commercial activity standing on their own are excluded from the Act (see Schedule 1 para 1a (insurance) and 2c and 3 (carriage of goods by ship). In this case the element of road transport was sufficient to render the transaction subject to the Act, but the mixed nature of the contract of carriage emphasises the interest of the freight forwarder in having a time limitation which is applicable across the spectrum of his obligations”.

The freight forwarder in that case was held entitled to rely upon a term requiring suit to be commenced within 9 months of the occurrence of the cause of action.

I share Tuckey L.J.’s reduction in enthusiasm, not least because of the difficulty of applying so protean a concept as that of fairness and reasonableness in a context where the application of these criteria may produce different results according to whether they are looked at from the point of view of the buyer or the seller.

I turn then to consider the guidelines, other than (d) and (e) which have no application. I consider them in relation to both the Borealis UK terms and the Borealis General Terms and Conditions.

As to (a) the strength of the bargaining position of the parties, the position is as follows. So far as the terms of the contract are concerned, Borealis would not knowingly have supplied borecene otherwise than on its terms, or terms to equivalent effect, unless they had been persuaded to do so for some very good reason. If Balmoral had raised with Borealis the question of supplying Borecene on Balmoral’s terms, they would, in all probability, have been told that borecene was not available on those terms. As Mr Cartwright put it he would have “walked away” the moment the proposal was suggested. This is, also, demonstrated by what happened, in relation to the proposed Industrial Supply Agreement where the proposed Balmoral warranty was crossed out by Mr Wood.

But Balmoral was a large volume purchaser, which was able to (and did) negotiate actively, and successfully, on price. The possibility of Balmoral switching supplier, as in the end it did, was always a factor tending to keep the price down, and Balmoral succeeded in driving good bargains in that respect. Further Balmoral did not have to take Borecene. They could have continued with ZN material, even in 1997 when DSM had given notice of withdrawal, since Matrix Polymers was waiting in the wings.

So, in relation to price, the parties were on at least an equal footing (indeed Balmoral may have had slightly the upper hand), but on terms they were not. Borealis was only prepared to supply borecene, which became, as Borealis no doubt wished, Balmoral’s primary raw material, on its terms. Other suppliers, whether of borecene or ZN material were also likely to be willing to supply only on their similar standard terms, save that Matrix’s terms provided for liability of up to £1 million if negligence could be proved.

As to (b) Balmoral received no specific inducement to agree the exclusionary provisions of either the Borealis UK or the Borealis General Terms and Conditions.

As to (c) Balmoral knew that Borealis’ terms were set out on the back, or on a page, of the invoices, all of which Mr Joyce signed off. Mr Joyce knew that there were terms on the back of the invoices. He had reviewed those terms but did not study them. Balmoral personnel knew, or, at the least, ought to have known, that the price quoted was on the basis that Borealis’ terms applied. If they wanted to know what Borealis’ terms said and meant, they had only to read and consider them, if necessary with Mr Byth’s advice.

A material consideration, albeit not one reflected in the guidelines, is the reality of the customers’ consent to the terms. As to that, Balmoral did not pay much attention to the terms upon which they were ordering; Borealis’ terms were not up for negotiation anyway; and, save for the short-lived and inconclusive discussions about an Industrial Supply Agreement, there was no negotiation in which the applicability of the standard terms, with or without variation, was hammered out. I do not however accept that Balmoral’s assent to the terms should be regarded as more apparent than real, having regard to the number of different occasions when they either signed documents expressly incorporating those terms or had the opportunity to demur when Borealis explained that they were going to apply.

Another material consideration is the damage that could be suffered if the borecene provided should prove unfit for the purpose of manufacturing oil tanks. As the fact of this case show, such unfitness could foreseeably lead to very substantial loss to Balmoral, far exceeding the cost of the borecene itself.

Section 11 (4) in terms applies only to the restriction in clause 13 (E) of Borealis UK’s terms and to the option to limit liability to the invoiced value in clause 6.3 of the General Terms and Conditions. But it is well established that it is a relevant consideration in relation to the requirement of reasonableness as a whole.

Borealis has very substantial resources and also extensive insurance cover (sufficient as I understand it to cover this claim) although the terms of the insurance are not before me. That said, the potential size of the totality of claims to which Borealis could be subject, if the terms relied on are inapplicable, is unquantifiably large.

Balmoral had insurance cover in respect of product liability i.e. insurance against losses arising from damage done by Balmoral’s products, but not product recall and guarantee cover. Mr Joyce’s witness statement said that, as he understood it, it would have been difficult and prohibitively expensive for Balmoral to insure and that he was seeking further information. None has been provided. Mr Webster’s evidence is to the effect that the availability of such insurance and its price was linked to the quality assurance procedures at the insured’s premises. I infer from that that product recall and guarantee insurance would have been obtainable, if Balmoral’s quality assurance programmes were acceptable to insurers, but that it would probably have been expensive. I have, however, no idea of the sort of figures involved.

In Watford Electronics Ltd v Sanderson CFL Ltd [2001] EWCA Civ 317 Lord Justice Chadwick attached significance to the fact that a purchaser, who relied on the 1977 Act to prevent reliance on a supplier’s standard terms, had similar terms in his own conditions of sale. This showed “that Watford was well aware of the commercial considerations which lead a supplier to include a provision restricting liability for indirect or consequential loss; and, in particular , was well aware that a supplier would be likely to determine the price at which it was prepared to sell its products by reference (amongst other things) to its exposure to the risk of unquantifiable claims to indirect or consequential losses which might be suffered by the customer if things went wrong”. The fact, however, that a purchaser has similar conditions of sale to those which he impugns cannot be conclusive, not least because the purchaser may be hard pressed to defend his own conditions of sale against a 1977 Act challenge. Further, in the present case Balmoral supplied oil tanks with a 10 year warranty. Thus, although Balmoral’s standard sale terms indicate the type of awareness to which Chadwick L.J. referred, it cannot be said that it was reasonable for Borealis to include its exclusionary terms because Balmoral did much the same in relation to oil tanks manufactured from borecene.

I regard the question as to whether or not Borealis UK’s terms satisfy the requirement of reasonableness as quite finely balanced. In the end, however, Borealis have not satisfied me that their conditions satisfy that requirement.

The relevant provisions are (a) clause 13 (C) (i) which excludes all conditions and warranties as to quality or fitness for purpose or correspondence with description “to the fullest extent permitted by law”; (b) clause 13 (C) (ii) which relieves the seller from any liability for any loss that the buyer suffers by reason of any defect in the goods; and (c) Clause 13 (E) which limits liability in respect of any defect to the price in any event. These clauses exclude the core terms implied by the Sale of Goods Act of satisfactory quality and fitness for purpose and deny the buyer any redress if the goods are defective, save that, in the event of a failure of the goods to comply with the seller’s standard specification, and subject to compliance with the conditions in clause 13 (B), the buyer may recover their price.

The requirement is that the terms shall have been fair and reasonable ones to be included “having regard to the circumstances which were or ought reasonably to have been known or contemplated when the contract was made”. It is thus directed, in part, to potential rather than actual circumstances. The question is not, as it was under section 55 of the Sale of Goods Act 1893, whether it would be fair and reasonable to allow reliance on the term in the events which have happened.

When the contracts were made Borealis knew that Balmoral was buying borecene for the purpose of making oil tanks and that it was relying on Borealis to supply a polymer capable of being used to make consistently satisfactory tanks. It was the assumption of both sides that it was so capable. The supply of a product which, because of a latent defect (whether in respect of ESCR or rheological properties or both) made the manufacture of consistently satisfactory tanks impossible would confound those assumptions. In those circumstances a blanket exclusion of any liability whatever is prima facie unreasonable (as was the exclusion of any liability for the supply of carbon dioxide with an excessive benzene content in Britvic Soft Drinks Ltd v Messer UK [2002] EWCA Civ 548; and Bacardi-Martini Beverages v Thomas Hardy Packaging Ltd [2002] EWCA Civ 549). I do not regard the supply of product with a latent defect as so remote a contingency that it ought to play no weight in determining the reasonableness of the exclusion. Nor were these contracts in respect of which there were notorious difficulties in successful performance or a high risk of failure.

A determination of the reasonableness of a contractual exclusion requires consideration of whether the allocation of risk effected by the exclusion is appropriate. I have not been persuaded that requiring Balmoral to bear the entire risk of a latent defect in Borealis’ product is an appropriate allocation of risk. The Sale of Goods Act itself recognises that, all other things being equal, it should be the seller who bears the responsibility. Borealis has extensive insurance against just such a risk. Whilst product recall insurance would probably have been available to Balmoral, albeit expensively, Balmoral did not have such insurance. The evidence does not establish that product recall insurance would have been normal for someone in Balmoral’s position.

But commercial parties habitually make agreements amongst themselves that allocate risk; and the Court should not lightly treat such agreements as unreasonable. The present case is not, however, one in which the contracts made were the result of a serious negotiation as to the incidence of risk: cp the Watford case where that was exactly what took place. Borealis’ terms were presented on a take-it-or leave it basis and Balmoral’s scope for going elsewhere on any better terms was very limited (on the evidence before me to Matrix and, even with them, obtaining any substantial sum would depend on proving negligence). Whilst Borealis UK’s terms were standard in the trade they are not the product of any agreed process of negotiation between representatives of sellers and buyers.

I take into account the submissions made by Borealis which I have set out in paragraph 401 above as to the extent of their potential liability and the cost implications of their being potentially liable for claims in respect of unfitness for any of the many purposes for which borecene might be bought. In respect of these submissions two matters strike me as of particular relevance. Firstly, whilst liability for fitness for purpose can be onerous it is not unqualified. It only arises when the buyer reasonably relies on the seller and the goods are not reasonably fit for what the seller knows is their intended purpose. The defect must be one that lies within the seller’s sphere of expertise. The seller can, of course, still be liable for a latent defect of which he is himself unaware. But, as between seller and buyer, it is more appropriate that loss from such a defect should be borne by the seller. As Scott Baker, J, as he then was, said in St Alban’s City and District Council v International Computers Ltd [1995] F.S.R 686 is not unreasonable that he who stands to make the profit should carry the loss. Borealis is a very large organisation, much larger than Balmoral both in term of assets and technical expertise, and has insurance against liabilities such as these. It can spread the cost of that insurance, and of any exposure not covered by insurance, over a very wide range of purchasers. Balmoral might be able to pass on the cost of product recall insurance but whether that is so is, on the present evidence, an open question.

It is material also to look at the consequences. If borecene was, contrary to my findings, not suitable for making oil tanks because of a latent defect, Balmoral will have suffered a huge loss (a foreseeable consequence of such a defect) which they had no real opportunity to avoid at the manufacturing stage. It is not reasonable that they should be without any redress from the manufacturer at all.

Lastly, it is not without significance that Borealis settled some of Balmoral’s claims, without reference to Borealis’ conditions. That cannot automatically mean that reliance on the impugned terms would have been unreasonable, or that the terms themselves were unreasonable. But, especially when the settlement is as high as £170,000, it is an indication to that effect.

Balmoral rely on the general principle that, in the absence of evidence that foreign law differs from English law, the English Court will apply English law to the issue: Dicey Chapter 9-001.

Borealis rely on two authorities that show that the principle is not absolute. In Damberg v Damberg [2001] 52 NSWLR the New South Wales Court of Appeal refused, despite the assumption of the parties that it would be bound to do so, to accept that German law in relation to the taxation of capital gains, and the evasion or avoidance thereof, was the same as that of Australia. The assumption was, on the facts, particularly inapposite since there were indications that German and Australian law were completely different. At the relevant time for most of the transactions in issue Germany had a capital gains tax but Australia had none, save for short term capital gains. Heydon J.A., referred to a wealth of authority, English and Commonwealth, as to the presumption that the lex fori applies as well as to cases where the presumption had not been applied. The cases cited in the latter category reveal a number of possible criteria as to when the presumption will not be applied e.g. (a) where it is not in the interests of justice; (b) where the foreign law is not based on the common law; (c) where the English law relied on alters the common law; (d) where it is inherently improbable that the foreign law is the same; (e) where fairness requires it. The Court did not find it necessary to specify criteria that would determine when the presumptions would not be applied. It simply decided that it could not be assumed that German law and Australian law in relation to avoidance or evasion of capital gains tax were the same. Taxation law could not be assumed to rest “on great and broad principles likely to be part of any legal system”.

In Shaker v Al-Bedrawi [2003] Ch 350 the trial judge had applied the provisions of Part VIII of the Companies Act 1985 requiring companies to have “relevant accounts” before making a distribution to a Pennsylvanian company, upon the footing that such a requirement was a “generally applicable rule”. The Court of Appeal referred to cases where the rule was inapplicable e.g. (a) where English law creates some special institution; (b) in prosecutions in respect of acts committed abroad; (c) where summary judgment is sought. They reached the view that if Part VIII were to apply to the company the statutory provisions would have to be significantly adapted, and thought (obiter) that that might of itself be a sufficient indication that the case was one where some special institution was created in English law such that those provisions could not apply to a Pennsylvanian company. The basis of the Court’s decision was that the Second Directive could not, at that stage, be accepted as showing that the requirement for relevant accounts was one which was generally applicable to private, as opposed to public companies.

Borealis submit that there is no reason to assume, and that it is unlikely, that Norway or Denmark have enacted legislation that allows purchasers who are not consumers to circumvent contractually agreed restrictions in terms identical, or even substantially similar, to the 1977 Act. They further submit that section 27 (1) of the Act (which provides that, if the law of any part of the United Kingdom is the law of the contract only by choice of the parties, most of the Act is not to apply if apart from that choice the law of the contract would be a non UK law), shows that the draftsman of the Act did not intend that contracts which were, or, but for an express choice of foreign law, would be subject to foreign law should attract its provisions. Further section 27 (2) (a) provides that the Act has effect notwithstanding any contract term applying some foreign law if that term was imposed in order to evade the operation of the Act. That provision suggests that, absent an imposition for that purpose, the Act should not have effect. It would then be very odd, Mr Allen submits, if a contract expressly subject to a non United Kingdom law should, in the absence of evidence that that law does not have the same provisions as the Act, be subject to it.

In my judgment the Court should apply English law, including the Unfair Contract Terms Act. I have reached that conclusion for two reasons.

Firstly, I regard it as artificial to apply the English law of contract without reference to the 1977 Act. If, as I must, I have to have regard to the Sale of Goods Act 1979, why should I ignore the 1977 Act? A possible ground is that the 1979 Act, as successor to the Act of 1893, is, in relation to the matters which it addresses, largely declaratory of the common law. But the 1893 Act has itself been the subject of amendment. The provisions of the Unfair Contract Terms Act 1977 Act might well have been put into effect by way of amendment to the Act of 1893, as happened with the Supply of Goods (Implied Terms) Act 1973. The latter Act implemented the English and Scottish Law Commissions’ First Report on Exemption Clauses. The 1977 Act was passed in consequence of the Commissions’ Second Report. I respectfully concur with the decision of the Federal Court of Appeal in Canada in The Ship “Mercury Bell” v Amosin [1986] 27 DLR (4^th) 641 where Marceau J distinguished “substantive provisions of a general character” from “others of a localized or regulatory character” and treated the rule as rationally acceptable “only when limited to provisions of the law potentially having some degree of universality”. In the same case Hugessen J observed that “the proper expression of the rule … is that the court will apply only those parts of the lex fori which form part of the general law of the country”. The 1977 Act may properly be regarded as falling within that category.

Secondly, I think that it would be unfair to do so. The prime purpose of the rule is to establish on whom lies the onus of producing evidence of the content of foreign law. In the absence of any pleading that the 1977 Act had no application Balmoral was by the time of the trial reasonably entitled to assume that the case would be conducted by reference to English law including the 1977 Act. Balmoral pleaded the Act in its Reply. Borealis pleaded, and then abandoned, time bar defences in Danish and Norwegian Law in its Defence. In a Further Reply on foreign law Balmoral relied on the presumption. There was a rejoinder to that Reply. Borealis did not plead that the presumption was inapplicable. In those circumstances I do not think that I should decline to apply the 1977 Act.

I am confirmed in that view by two further considerations. Firstly it would have been very easy to adduce evidence as to whether Norway or Denmark has provisions identical or similar to those contained in the 1977 Act. Borealis, who, in effect, contend that neither Norway nor Denmark has any provision equivalent to the relevant provisions of the 1977 Act has failed to adduce evidence to that effect, notwithstanding that Norway and Denmark are the places of incorporation of the Borealis entities relying on those laws. If Borealis was going to say that these laws had no provision that would restrict the applicability of the exclusion clauses, it should, and could easily, have produced the evidence to establish that. Secondly, the laws of those countries may have provisions restricting the entitlement of suppliers to exclude or restrict liability which are (a) the same as English law; or (b) more; or (c) less favourable to the purchaser; or (d) may have no such provision at all. The effect of applying English law, but without reference to the 1977, Act is that I shall have assumed that the answer is (d). Whilst this is a possible answer it is by no means the only one. To make that assumption might be a great injustice.

I do not derive any assistance in reaching this conclusion from section 27 of the 1977 Act. The Act plainly contemplates that its provisions (a) will not apply to a foreign law contract unless the foreign law has only been chosen in order to evade the operation of the Act and (b) that its most significant provisions will still not apply even if the parties have expressly chosen the law of England, Scotland or Northern Ireland if, but for that choice, the contract would be governed by a foreign law. The purpose was, no doubt, to ensure (a) that people did not evade the operation of the Act by an artificial choice of law and (b) that parties who chose English law did not, by litigating in England, find themselves forced to have the 1977 Act apply in circumstances where that was not the natural law of the contract. But neither of those facts has any bearing on the incidence of responsibility for adducing evidence of foreign law. The section says nothing about the rule or presumption, which was in force well before the Act was passed.

The first question is: which parties are to be considered? Is it Balmoral, on the one hand, and Borealis Norway or Borealis Denmark, on the other? Or does it include Borealis UK as agent? If so, does 26(3) not apply because of the inclusion of Borealis UK as a party?

In Ocean Chemical Transport Inc v Exnor Craggs Ltd [2000] 1 Lloyd’s Rep 446 the Court of Appeal held that the reference in section 26 (3) (b) to a contract “made by the parties” referred to the principals in the contract in question and not the agents through whom the contract was made. Mr Mawrey submitted that that case applied in the ordinary case of an agent who incurred no personal liability on the contract in question; but not to a case where the agent is, as he submits to be the case here, an English agent who is personally liable under the contracts.

I do not accept that there is any relevant distinction between the Ocean Chemical case and this one. The basis of that decision was that it was “as clear as could be” that the reference to parties was to the principals. I do not regard that clarity as diminished if either the buyer or the seller’s agent assumes personal liability. If that is so, there are up to four “parties” - the seller, the seller’s agent , the buyer, and the buyer’s agent. The statute gives no indication that, if seller and buyer are in different States the fact that at least two of such “parties”, presumably one on either side of the contract, are in the same State, then the Act must apply. Nor does such a conclusion appear to me consistent with the statutory purpose. The policy behind the Act is that it should not strike down exclusions from liability contained in international supply contracts. It would be very surprising if a contact for the supply of a cargo of crude oil f.o.b. Bonny in Nigeria made between a Nigerian seller and an English buyer, both acting through English brokers, should cease to be an international supply contract, if either broker undertook personal liability on the contract. The purpose of the section is to deal with contracts where the supplier and the purchaser are in different States and the goods are being, or are to be, carried from one State to another, or to a State other than the one where the contract was made. It is not concerned with whether the agent of the supplier is in the same State as the purchaser or his agent.

But in any event I do not think that Borealis UK should be regarded as having undertaken a personal liability under the contracts with Borealis Norway or Borealis Denmark made after 1^st January 1999.

Mr Mawrey submitted that, when the draftsman referred to the conclusion of the contract, he did not have the making of the contract in mind - otherwise he would have used the language of offer and acceptance that he had used in section 26 (4) (b) and (c). The conclusion of a contract, he submitted, “logically follows its formation and must be either co-terminous with or shortly before its performance” and was, in the present case, “at or before the time when the obligation to pay arose: that is on the raising of the invoice”. At that stage the goods would have been delivered to Balmoral and would no longer be in the course of carriage.

I do not accept this submission. A contract is normally regarded as having been concluded when it has been made. It is at that stage that the negotiations have reached their conclusion. I see no reason to give the words “conclusion of the contract” anything other than their ordinary meaning. Further, if what is meant is the performance of the contract, there is no reason why that should mean anything other than full performance. If they are to refer to some point prior to complete performance, it is impossible to say what that earlier point is. Mr Mawrey’s submissions would have the curious consequence that a contract for the supply of goods to be manufactured in Germany and delivered in England would not be an international supply contract if payment was due after delivery.

If the relevant time is the time when the contract was made, that was either when Borealis UK confirmed the order or, at the latest, when Balmoral took delivery of the goods. If the latter, the goods were not then in the course of carriage. If the former, they may have been. In my judgment the contracts were made when the order was confirmed. By that stage there had been an offer to buy and an acceptance of the order. From then on the parties could reasonably regard themselves as having come under an obligation to deliver and accept delivery. There is no need to rely on delivery as constituting both the making of the contract and its performance by Borealis.

The borecene in question was manufactured in Norway and sold by either Borealis Norway or Borealis Denmark. It was delivered in the United Kingdom, either to Balmoral in Aberdeen or to ICO in Gainsborough, Lincs. That would suggest that when the contracts were made, the borecene was either in the course of carriage or was to be carried from one State to another.

But in some cases during 1999 and 2000, although the country of origin specified on the invoice for the Borecene was Norway the place of despatch was described as Great Britain. In those cases Borealis was the importer, the goods came from one of Borealis’ transhipment centres (in the form of rented warehouses) in Great Britain, and VAT was charged on the Invoice. Where the place of despatch was described as somewhere other than Great Britain Balmoral was the importer and no VAT was charged on the invoice. Balmoral had to account to the Customs for the VAT. In the case of goods sold by Borealis Denmark the place of despatch was always Great Britain. Borealis Denmark was always the importer of the goods, charged VAT on the sales and accounted for the VAT to the Customs. It was within Borealis’ power to determine whether it was the importer or not.

The invoices where VAT was charged, with Great Britain named as the place of dispatch, indicate that at some stage prior to delivery to Balmoral the borecene had arrived at a place in Great Britain from which it had been despatched. But the invoice does not indicate how long the goods in question stayed at the place of despatch, so that, looking at the invoice alone, it is possible that the goods were at the place of despatch to which the invoice refers when the contract was concluded.

Mr Allen submitted that it was inappropriate to adopt too technical an approach to this question. The reality was that Borealis Norway or Denmark were selling borecene to a purchaser in the United Kingdom. Even if the order was supplied from goods that had been stored in England for some time prior to the date of the order the goods were still in the course of carriage. They were only at a stopping place in the course of carriage between the country of manufacturer and the country of the purchaser.

Section 26 (4) (a) requires the court to determine whether, at the moment of contract, the goods supplied under it were still in the course of carriage from the territory of one State to the territory of another. In Amiri v BAE Systems [2004] 1 All ER 385, Mance L.J., as he then was, observed that section 26 (4) was “open to the comment that it has not been fully worked out” and that it may have been thought that the classical types of international trading contract (c.i.f., f.o.b, etc) should be excluded without more. Even contracts such as those may not come within section 26 (4) (a), if it is given a literal construction. Take the case where there is a string of CIF sales and the last one is made when the vessel is within the territory of the State where the cargo is to be discharged. On one view the goods are not then in the course of carriage from one State to another. Such a view is inappropriate. In the example given the goods can be properly said to be in the course of carriage from one State to another, just as a passenger may be said to be in the course of carriage from London to Paris when the Eurostar stops at Lille.

I am satisfied that I should regard the borecene the subject of the invoices naming Great Britain as the “place of despatch” as goods which, when the contracts were concluded, were to be carried or were in the course of carriage from the territory of one State (i.e. Norway) to the territory of another. The course of business involved Borealis UK checking with Borealis Norway that the goods could be produced and then confirming the order to Balmoral. It would follow that the goods would not, when the contract was concluded, have been in the United Kingdom and were to be carried thither once they had been manufactured. The evidence before me does not indicate that Borealis’ warehouses in the UK were anything other than transhipment centres or that borecene ordered by Balmoral was supplied to them from stock which, at the time of the contract, was in store in the United Kingdom waiting for a purchaser.

It follows that, in my judgment, whilst the 1977 Act applies to all the contracts with Borealis UK, it does not apply to any of the contracts with Borealis Norway or Borealis Denmark.

The General Terms and Conditions are in fact somewhat more favourable than the Borealis UK terms, since the implied condition as to satisfactory quality is not excluded and there may be an implied warranty of fitness for purpose if the seller has approved the suitability of the product in writing. It is not suggested that that occurred and the term of fitness for purpose is excluded. For the same reasons as I have set out in relation to the Borealis UK terms I am not satisfied that the clauses 6.1. and 6.3 of the General Terms satisfy the requirement of reasonableness.

In summary my conclusions on the contractual questions are as follows:

In the light of these conclusions no question of damages arises. Nevertheless, in case this matter goes further, I propose to set out my conclusions.

The parties advanced submissions on damages upon the hypothesis that Borealis was liable under all of the contracts, recognising that, if Borealis was only liable on some of them, their approach to quantum would have to be reconsidered. I shall proceed on the same basis. The parties also appreciated that any conclusion that I might reach might well require further working out in the light of whatever approach or combination of approaches I accepted as valid.

Before considering the issue it is necessary to record how Balmoral dealt with increasing tank failure. By the end of 2002 Balmoral was beginning to face a flood of complaints. The problem with borecene made tanks became public knowledge. The Environment Agency became involved. Balmoral had to increase the number of staff members devoted to dealing with complaints. By November 2003 9 staff were involved; by February 2004 13. Detailed records were kept of each complaint and entered on a spreadsheet (the “Emergency Spreadsheet”) which forms the basis of the quantum claim. Additional files were created for “Environmental” claims i.e. claims in respect of damage caused by oil spillages.

In the autumn of 2003, by which time the Environment Agency were expecting positive action, Balmoral adopted what became called a “planned change-out” policy whereby it would change tanks made of borecene MFR 6 before cracking developed, starting with tanks at the most vulnerable sites, and with the most vulnerable models (horizontal and slim line) and leaving vertical models until later. Balmoral contacted its distributors and advertised in the national media and on its website in order to identify as many of those tanks as possible. Distributors were invited to make contact with the end-user, sending a form to be returned to Balmoral if the user owned a tank with a serial number beginning with 0001 up to 0201. Advice was given to customers, distributors and fuel suppliers that tanks within that range should not be filled to over 50% capacity. A separate department of two people implemented the programme. In Spring 2005 some fuel suppliers decided to advise their customers to adopt this policy with all Balmoral tanks.

Matters were complicated by the coming into force of new legislation viz:

If the end user did not want a replacement he could get a refund of the current average sale price of the tank. If the replacement of an old single skin tank by a new one was inconsistent with the Regulations Balmoral would either give the end user an option to upgrade or a cash refund.

In April 2004 Balmoral replaced its 10 year guarantee with a 1 year one.

By early 2005 the cost of tank replacement was of the order of up to £350,000 per month and Balmoral was in severe difficulties. It decided that, in order to slow down the rate of its losses, it would have to take a tougher line with customers. Claimants were required to produce a proof of purchase invoice, and to pay for a site survey, including photographs of the tank and installation, by an Oftec registered company. If the survey showed that the problem was not the result of misuse or defective installation and was within the warranty period, then, unless the tank was made from MFR 6 borecene, the customer would be offered:

In the case of a tank made with MFR 6 borecene, owners were to be offered:

This change of policy produced bad feeling and bad publicity. Forcing end-users to deal with their insurers also meant that those insurers could later appear with substantial claims against Balmoral.

At present most planned replacements have been put on hold because of cash flow implications.

Balmoral’s claims can be expressed as follows:

Balmoral’s expert was Miss Sally Hassell, an experienced forensic accountant. I found her a highly competent and reliable witness in relation to matters which she had personally researched and calculated. She had, for instance, extensively reviewed the “Emergency Spreadsheet” and made adjustments which, I am satisfied, make it a reliable document from which to assess the cost of tank failures which can be attributed to borecene (Footnote: ¹⁸¹). Her checking, collation and amendment of data, some of it extremely voluminous, seemed to me thorough and conscientious. She was, inevitably, as she recognised, on less sure ground when dealing with projections as to what would have happened if the cracking problems had never arisen. Her reports were well constructed and, particularly considering the complexity of the material, clear.

Borealis’ expert was Mr Flemming Jensen. He is an insurance loss accountant. It was suggested that such a background would naturally incline him to take whatever point could properly be made in the reduction or resistance of the claim. I did not, however, regard him as having strained to take every point, although, understandably, he was not prepared to assume what had not been established. He seemed to me both competent and conscientious, although he could, on occasion, have painted with a broader brush.

The figure of £2,779,420 down to 31^st March 2005 is an agreed figure. It derives from a calculation, to be found at paragraphs 5.1 to 5.3. of Miss Hassell’s supplemental report (“Hassell 2”), of the cost of dealing with tanks made from borecene which failed in excess of the agreed, underlying, non-borecene rate of failure of 2.96 %. As at 31^st March 2005 a number of complaints had been dealt with (“completes”), a number had been deferred, and a number were open. The figures for dealing with completed (£526) and deferred (£51.48) complaints are agreed. A “deferred” complaint is one where the complaint has been rejected e.g. because the end user has failed to provide proof of purchase, or to organise and pay for his own site survey, or because installation was not in accordance with the guidelines. The sum for dealing with a deferred complaint covers the cost of initial processing and rejection of the complaint. The complaint is described as “deferred” (a) because of Balmoral’s doubts as to whether the policy will hold; and (b) because the end user may challenge this rejection. Miss Hassell’s calculation assumes that the open complaints will become completes or deferred in the same proportion as the actual figures in those categories.

In relation to the period down to 31^st December 2005 Borealis agrees the figure of £411,130 which appears in the table at paragraph 6.6.2 of Hassell 2. That figure is based on the same methodology as produces the £2,779,420 total but with more up to date figures for the average cost of dealing with complaints (Footnote: ¹⁸²). That cost has reduced, partly because there has been a higher proportion of refunds and refunds are now of residual value. The cost of deferred items has also reduced to £4 because Balmoral now shifts responsibility for obtaining site surveys to the end user. Miss Hassell also puts forward an alternative calculation of £578,131. This alternative calculation (paragraph 6.7.2) assumes that claims that have been completed or deferred will not be reopened, but that all open claims will be dealt with in accordance with the policy in force before the change in 2005, on the basis that that harder line policy cannot last. On that assumption £578,131 is agreed as a figure.

In my judgment it is the higher figure that is appropriate. The hypothesis upon which this question arise is that (contrary to the fact) I have decided that borecene is not fit for the purpose of making oil tanks and that, particularly in relation to MFR 6, borecene, tanks have failed in large quantities for that reason. On that hypothesis it does not seem to me that Balmoral would be able to resist having to maintain at least their former policy. It would not be possible to hold the line that end users must bear the costs of surveys, that only the residual value should be refunded, and that there should be very limited, if any, compensation.

In a document handed in on Day 22 Miss Hassell floated the idea that for this calculation I should use the agreed historic costs (£526 and £51.48) in relation to all change outs, including those completed and deferred, upon the basis that the tough line policy might backfire on Balmoral, complaints would be reopened, or new complaints would come in which would have been avoided if Balmoral had not declined to change out the tank involved, so that Balmoral would, in the end, become in at least as bad a state, in respect of complaints to date, as if they had maintained the previous policy. I regard this as too speculative to rely on.

The dispute relates to the costs alleged to have been incurred bringing defective tanks back to Aberdeen for disposal (Footnote: ¹⁸³) when the distributor was unwilling to deal with them. Until June 2005 when Balmoral ceased to have a Transport Division this transport was done by that Division. The claim is for £66,858.75 (£50,478.75 transport and £16,380 for handling). The transport cost is Balmoral’s standard delivery charge which varies according to the size of the tank. The handling charge is a “notional” charge to cover the labour in dealing with the tanks.

Mr Jensen would disallow the handling cost on the ground that it is a normal overhead expense, has not been proved, and would be easily absorbed. He would only allow £10,095.75 of the transport costs. The basis upon which he would only allow the latter sum is that he does not accept that none of these tanks could have been brought back as a backload after delivery of tanks to a distributor. The suggestion is that 81% of the tanks could have been backhauled: see paragraph 5.5 of Mr Jensen’s second report (“Jensen 2”) and paragraphs 363 – 364 of Borealis’ final submissions. The 19.2% figure, which reduces £50,478,75 to £10,095.75, is the proportion that the third party turnover of the Transport Division for the year ending March 2004 (£331,000) bears to the inter company turnover (£1,726,000).

I do not regard the £ 15 charge as having been proved as a loss. Some labour was no doubt involved in unloading the tanks and moving the so called “graveyard” stock to the graveyard i.e. the place at Balmoral’s site where returned tanks were stored. But no one at Balmoral has been able to identify any additional cost involved and the work, which was probably done by those whose pay appears in the group service part of the accounts, does not seem to me to be such that the fact that it had to be done, without identifiable additional cost, should be regarded as a loss to Balmoral.

As to the transport costs, there appears to me to be an inconsistency between the way in which the claim is in truth put forward and the manner in which it was, at any rate ultimately, resisted, namely that Balmoral could and should have used backhaul shipments to bring tanks back to Aberdeen. In giving evidence Mr Main, Balmoral’s Group Financial Director, said:

“A. No, the reason we put it (i.e. the breakdown of transport turnover) forward was I was asked a question by Ms. Hassell: Is there a cost in bringing tanks back? I said Yes, there is because if we bring a load of tanks back it means we're not bringing a paying load from a third party back so there is a real cost to us. She accepted that but in discussions with the experts from the other side this wasn't accepted, and they said they needed evidence to show that we did generate third party income and therefore we produced this”.

And later, in answer to questions from me:

“Do I understand from that that matters were so structured that if you had to collect a faulty tank, you usually, or often, arranged to collect it when you were delivering a load in the vicinity, so that you could bring back the defective tank on the transport with which you had delivered some goods to somebody?

A. Yes, the tanks would be changed out by an installer team, taken back to distributors based at their own base and then we would collect probably a number of tanks at one time, rather than picking individual tanks up.

MR. JUSTICE CLARKE: But you would endeavour to do this exercise of collecting tanks as a backload?

A. We wouldn't send a lorry down specifically to pick tanks up.”

In effect, therefore, Balmoral’s claim is that, because all the tanks were back hauled, it lost the opportunity to take back loads from paying third parties. Mr Main gave evidence that the transport division, which provided services to the other divisions at commercial rates, endeavoured to find backloads from third parties and did so with a high percentage of success, such that there were very few miles driven without loads. However in the marine part of the Transport Division the work, which often involved taking cranes and low loaders to the docks, was mainly local, and did not generate back loads.

This general evidence does not seem to be supported by the only figures available. The income from back loads is the third party element of turnover. Some back loads, however, involved the carriage of raw material for Balmoral’s Divisions, which generated an inter company charge. Inter company turnover also included the recharge of cranage and low loaders, and of third party transport sometimes used to deliver products. So, even if every delivery had had a backload of equal value the inter company and third party turnover figures would not be the same.

I approach the matter in this way. There are in the papers no documents that indicate which lorries returned empty, and which full. The evidence does, however, show that Balmoral tried, as one would expect, to obtain back loads from third parties in cases where it did not have a back load of its own. The fact that some 11% of the total turnover for 2003/4 (i.e. the ratio of £331,000 to £331,000 + £1,726,000) derives from third parties shows the extent of its success. But the fact that the third party proportion of inter company turnover is nearly 20% does not establish that in 80% of the cases there was no back load, not least because inter company turnover is not limited to receipts from outward journeys but includes the matters referred to in the previous paragraph, and because loads and backloads cannot be regarded as always of equal value. There was some confusion in the evidence as to whether the 80% figure included a charge to Balmoral for returning empty to Aberdeen. Mr Main’s evidence was that where raw material for the division was carried in a backload, there was an inter company charge. I infer from that evidence that empty back hauls were not charged. There is no positive evidence that they were. That would not be the way in which transport costs are usually charged.

In those circumstances it seems to me that Balmoral have established that they

The number of planned change outs to 31^st March 2005 is agreed. The number to 31^st December 2005 is one that Mr Jensen has no reason to doubt. A figure of £458 for the cost of the change out is agreed. The issue that divides the parties is whether the change out policy was a reasonable one and whether Balmoral is seeking to recover for tanks that did not need to have been changed out at all. Balmoral changed tanks whose production fell within the serial range “0001….” to 0201….” The first two digits - “00” and “02” - denote the years 2000 and 2002. The second two digits denote the month. Most of the tanks within that range were MFR 6 tanks. But some were not. For instance, some MFR 3 tanks manufactured in very early January 2000 immediately before the introduction of MFR 6 came into the 0001 range; and some MFR 4 tanks manufactured in January 2002 immediately after the change to MFR 4 came into the 0201 range. In the period up to 31^st March 2005 930 tanks were called in of which, as is now known or estimated (Footnote: ¹⁸⁴), 778 were made of MFR 6 borecene. If the claim for all 930 is valid its value is £425,940. If the claim should be limited to 778 tanks the claim figure up to 31^st March 2005 is £356,324. In relation to planned change outs down to 31^st December 2005 the relevant figures are £31,383 or £26,228.

Borealis contend (i) that they should not be responsible for the planned recall of tanks not made with MFR 6 borecene; and (ii) that in 2003 Balmoral, which knew that some tanks had low and some high failure rates, could and should have restructured the planned change out so that it was limited to MFR 6 tanks with failure rates higher than 10%. On Mr Jensen’s calculations that would produce a figure, for the period down to 31^st March 2005 of £174,498.

I agree with Borealis’ first, but not its second, point. As to (i), it was no doubt sensible to give the public a range of numbers between which the MFR 6 population would lie. But it was not impractical for Balmoral to check, from the serial number of any responding customer’s tank, whether it was made of borecene MFR 6. The production records contain details of the serial numbers, which are in sequential order, and the material (Footnote: ¹⁸⁵). Change out of a tank is quite expensive; and the cost of checking whether the material of the tank was borecene MFR 6 would have been small in comparison. Even if the records were archived they could be accessed. As to (ii), Borealis was faced with substantial failings of tanks made with MFR6 borecene. It was under pressure from the Environment Agency. By 2004 and 2005, the years when the policy was being put into effect, some tank designs were failing much more than others, as Balmoral would have known. As it has turned out so far, there have continued to be markedly different failure percentages for different designs. But, when the policy was made, Balmoral was not to know that this pattern would continue. Their policy to change out MFR 6 was, in my judgment, a reasonable response at the time to the dilemma in which they then found themselves, particularly having regard to the fact that a failure to call in a tank because, say, there had only been a 7.5% failure rate to date, could have catastrophic consequences. Further the matching of production data to failure data, which would allow an accurate calculation of those designs in which there was a failure rate higher than 10% has proved a mammoth task, only completed in the course of preparation of the experts’ reports in this action.

Accordingly the appropriate sum under this head is £356,324 plus £26,228 i.e. £382,552.

I ruled at an early stage that I would not deal at the hearing with the issue of environmental claims, i.e. those that have or may lead to environmental pollution. These claims can appear unexpectedly, or in unexpected amounts, either in addition to a claim for replacement of the tank or where no such claim has been made e.g. because the end user finds it easier to deal with his own insurer, or is forced to do so because Borealis has rejected his claim. Balmoral will often have no idea of the costs involved for a long time. The incidence of further unexpected claims is also very difficult to predict, both as to number and amount. It seemed to me that the appropriate course would be to make an interim award of damages in relation to the matters that could be dealt with now, leaving these claims over until later.

For similar reasons I decided not to deal with the issue of claims made against Balmoral outside the U.K.

Miss Hassell’s reports identify the following staff and staff related items:

The staff costs item consists of staff costs incurred in dealing with the cracking problem with the tanks and the planned change out. The staff costs of £821,051 are the product of the following:

The £342,531 figure is the cost of administrative personnel in the Customer Services Department from July 2003. The £321,086 is, as to £138,640, an allocation of 95% of the salary and other emoluments of Philip Pritchard, Balmoral’s site services director and the manager of the team dealing with customer complaints, and, as to £182,446, a proportion of the salary and emoluments of other staff, including Mr Joyce. The £157,434 figure is an allocation of a proportion of the salary and other emoluments of indirect staff (e.g. the finance director and staff in the accounts section). The proportions taken are percentage proportions of time spent in the years 2002 to 2005. The details are at Exhibits 27 - 30 of Miss Hassell’s first report (“Hassell 1”).

The issues between the parties on this aspect of the case are these. First, in respect of the costs of the Customer Service Department and Philip Pritchard, there is a relatively small difference of £17, 750. The relevant figures are as follows: