Beyond the Millennium - The Legal Issues: Sale of Goods Issues and the Millennium Bug

Robert Bradgate
Faculty of Law
University of Sheffield
[email protected]

Abstract

The law of sale and supply of goods seems to provide an attractive means for a person who suffers loss or expense due to a Y2K problem to seek to recover compensation. However, the classification of 'software' and other computerised items as 'goods' is problematic and open to question, and the application of the concepts of the law of sale to intangible items such as software is equally problematic. This piece considers first the various approaches to the classification of software, in order to assess whether it should be treated as 'goods' and what the consequences of so-doing will be. It proposes an alternative mechanism by which strict liability for defective software could be imposed. It then considers whether the law of sale, and in particular the quality warranties in sections 13 - 14 of the Sale of Goods Act, could be applied to software affected by Y2K problems. In particular, can it be argued that software, or other products, which fail due to being non-Y2K-compliant, lack 'durability'? Finally it considers whether the law of sale is an appropriate mechanism for determining the policy questions involved in allocating losses caused by the Y2K problem.

This is a Refereed Article published on 30 June 1999.

Citation: Bradgate R, 'Beyond the Millennium - The Legal Issues: Sale of Goods Issues and the Millennium Bug', 1999 (2) The Journal of Information, Law and Technology (JILT). <http://elj.warwick.ac.uk/jilt/99-2/bradgate.html>. New citation as at 1/1/04: <http://www2.warwick.ac.uk/fac/soc/law/elj/jilt/1999_2/bradgate/>

1. Introduction

At midnight on 31 December 1999 the world will plunge into chaos as computers, unable to recognise dates containing the year 2000, fail. Banking, financial and insurance systems will collapse. Traffic will grind to a halt as rail and road traffic management systems fail. Aeroplanes will fall from the sky. Electricity, gas and water systems will shut down. Business records will be thrown into chaos and as computerised ordering, stock control and distribution systems fail, whole supply chains will break down. There will be no escape. The car, with a computer controlled engine management system, will refuse to start. Nor will the home be a haven. Even if electricity supplies remain intact, many domestic appliances such as VCRs, microwave ovens, washing machines, central heating programmers, burglar and fire alarms contain embedded chips and to that extent are computerised and they will fail.

Such, at least, is the pessimistic view. One does not have to subscribe to this 'Cassandra' scenario to anticipate that the Y2K problem - the so-called 'millennium bug' - will cause losses on an enormous scale. Indeed, it has already done so. Over the last two to three years enormous resources have been committed to attempts to manage the problem and minimise the effects of the 'bug', checking computer systems for Y2K compliance, replacing or repairing those which are not and preparing contingency plans to deal with problems which arise. Some systems will have to be replaced. The same is true of computer dependent consumer items. Some forecast a boom for producers and retailers of consumer electronics.

On any view, then, the Y2K problem will be expensive. If the Cassandras are right, it will cause damage on a massive scale - not only economic losses to business but property damage and even personal injury. But even if they prove wrong, vast sums will be spent auditing, reprogramming and replacing computer systems. Those affected are likely to seek ways to recover some at least of the costs they have incurred dealing with and/or losses they will incur as a result of the Y2K problem.

As is well known, the cause of the Y2K problem is that many computer systems record the date in a six, rather than eight, digit field so that 1 January 2000 will appear as '01.01.00', causing logic failures as computers read '00' as earlier than '99'. The decision to record the date in this way was taken deliberately in the early days of computing, to save memory space and thus to reduce the cost of computers. At that stage the Y2K problem was no doubt unforeseen. Even when smaller, more powerful, computers began to appear in the 1970's and 1980's few people were thinking ahead to the year 2000. It is clear, however, that the problem would have been apparent even then, to any one who bothered to think about it. Moreover, it seems that non-compliant systems, incapable of handling the millennium date change, continued to be produced and marketed after the problem should have been known and probably even after it was[ 1 ].

1.1 Pledge 2000

In an effort to discourage litigation in favour of co-operation to deal with, and thus minimise the effects of the Y2K problem, the government's 'Action 2000 Task Force' has produced the 'Pledge 2000'. Businesses which sign the Pledge commit themselves to take steps to minimise the impact of the 'bug', to 'work co-operatively and in good faith with out supply chain partners on Year 2000 Issues' and to work together 'to solve the Year 2000 problem rather than taking legal action'. It closes with the words 'We will, whenever appropriate and possible, use consensual means for resolving any disputes over year 2000 problems that may arise'. The Pledge 2000 is, however, expressly stated not to 'create any legal obligation or establish any legal relationship whatsoever' but to be a 'non legally binding statement of our future intentions'.[ 2 ] Clear words can, of course, deprive of legal effect an agreement which would otherwise be contractually binding ( Rose and Frank Co v JR Crompton & Bros. Ltd ( 1925 )) and it is therefore clear that the Pledge will not prevent any signatory taking legal action against any other. It is a fair bet that, notwithstanding the bad publicity which might result, many may be tempted to do so if the stakes (or losses) are sufficiently high. The Pledge certainly will not bind non-signatories (and it seems reasonable to assume that the businesses most likely to fail to deal with the 'bug' are also the ones least likely to sign the pledge). Nor will it affect claims by private individuals, for it is expressed in terms for signature by business organisations only.

2. The Application of Sale Law

The law of sale seems a natural avenue for anyone seeking redress for losses caused by items proving not to be Y2K compliant to consider. We all, lawyers, consumers and businesses alike, are familiar with the notion that where goods are sold, the seller assumes legal responsibility for them. Moreover, the statutory implied terms that goods supplied under a contract must be of satisfactory quality and reasonably fit for the buyer's purpose and must conform to the description by which they are supplied, offer a number of advantages to a potential plaintiff. They create a regime of strict liability with an easily identifiable target - the supplier is liable for the goods supplied, without the need to prove fault. They are relatively easy to prove - the core term, that the goods supplied must be of satisfactory quality, arises automatically wherever goods are sold in the course of a business (Sale of Goods Act 1979 s.14 (2)). They are long established: notwithstanding the replacement of the requirement that goods be of 'merchantable quality' with that that they be of 'satisfactory quality' in 1994, the essence of the implied terms has been established for almost 200 years. There are statutory controls on attempts to exclude the implied terms under the Unfair Contract Terms Act 1977 ss 6 and 7. Where goods are supplied to a person who deals as a consumer the terms can never be excluded. In other cases exclusion is permitted only where the exclusion satisfies the test of reasonableness. Since the implied terms are classified by statute (in England and Wales, at least[ 3 ]) as 'conditions', breach of them entitles the buyer to reject the goods and claim a full refund of the price, without any allowance for any use he may have had of the goods. The right to reject is, however, lost if the buyer has 'accepted' the goods or is taken to have done so, including by retaining them beyond a reasonable time without rejecting them. Since 1994 the assessment of what is a 'reasonable time' must take into account whether the buyer has had a reasonable time to examine the goods for conformity with the contract (Sale of Goods Act 1979 s.35(5)). However, it is clear that a buyer may still be taken to have accepted goods before a latent defect manifests itself, and it seems probable that although the time before a buyer is held to have accepted goods and thus lost the right to reject them may now be longer than some pre-1994 cases suggested (see Bernstein v Pamsons Motors (Golders Green) Ltd . ( 1987 )), the right is still likely to be lost after a relatively short time. In most cases, therefore, except where items are discovered to be non-Y2K compliant within a short time of purchase, the buyer not be entitled to reject them. But even a buyer who has lost the right to reject, or chooses not to exercise it, may nevertheless claim damages for the breach, and a claim for damages may be brought up to six years after the date of delivery of the goods, and even later in some circumstances. And, perhaps most important in the present context, since the claim is one in contract, there is no restriction on the recoverability of economic loss. The implied terms thus provide a basis for claiming not only for damage to property or personal injury, but also for the cost of repairing defective items, or for the difference in value between the item as is and as it should have been. Thus a claim could be made for the costs of making an item compliant. And if, say, a VCR proves not to be Y2K compliant and has to be replaced earlier than might reasonably have been expected, it may be possible to claim damages for its being worth less than would have been the case had it been compliant.

It seems reasonable therefore to assume that the failure of items which are not Y2K compliant will give rise to claims for breach of the implied terms, either by consumers (and businesses) seeking to reject items on this basis, or by persons seeking compensation for the costs of repairing items to make them compliant or for losses caused by failure of non-compliant items. The prospect of consumers en masse seeking refunds for non-functioning VCRs, microwaves and so on on January 2 2000 must be a nightmare for retailers; that of claims for damages for consequential losses caused by malfunctioning items, or for the costs of repair of such items even more so. Even one such claim might be ruinous. Such damages claims may give rise to special problems of remoteness and causation. (see Macdonald ( 1999 )). The focus of this piece is on the special problems posed by attempting to base claims for non-Y2K compliance on the statutory implied terms in contracts for sale and supply of goods. Although the implied terms are long established, their application to computer software and related items may be problematic. How can rules devised in the context of disputes about parcels of waste silk ( Gardiner v Gray ( 1815 )), Manilla hemp ( Jones v Just ( 1868 )) and carriage poles ( Randall v Newson ( 1877 )), be applied to disputes about allegedly defective computer software? It was long ago recognised that the requirement that goods be of merchantable quality was difficult to apply to complex, manufactured, mechanical items as opposed, say, to commodities ( Bristol Tramways Carriage Co Ltd v Fiat Motors Ltd ( 1910 )). Indeed, that difficulty was one of the reasons for replacement in 1994 of the 'merchantable quality' requirement with that of 'satisfactory quality', but there is as yet virtually no case law on the meaning of 'satisfactory quality', and relatively little case law on the application of the implied terms to computer software before or after 1994. However, before considering the application of the implied terms to the problem at hand, we must first consider the questions 'is computer software 'goods'?' and 'is software supplied under a contract for the sale or supply of goods'. These questions have only recently, and as yet inconclusively, been considered in case law.

3. Is Software 'Goods'?

Terms of satisfactory quality, fitness for purpose and correspondence with description are implied by implied by statute into (i) contracts for the sale of goods (Sale of Goods Act 1979 ss.13, 14(2), 14(3)) (ii) contracts of hire purchase (Supply of Goods (Implied Terms) Act 1973) and (iii) other contracts of supply involving the transfer either of possession of or property in goods (Supply of Goods and Services Act 1982). Although the basis for implication and the remedies available to the customer may differ according to the classification of the contract, the content of the terms is the same in each case. However, the terms are only implied by statute into contracts for the supply of 'goods'.

3.1 Appliances with Computerised Components

The Y2K problem is generally seen as a software problem, but if software is seen through a layman's eyes as being programs supplied on floppy disk or CD-ROM this is a gross oversimplification. In some cases the problem may lie in the computer's real time clock chip, where the software is 'hard wired'. In others it may lie in a particular program, but that may be an operating or applications program. Applications software created using bespoke software 'shells' such as Excel or Access may be non-compliant even though the 'shell' is compliant. In some cases software which would run happily on one operating system or machine may fail on another. Finally, and perhaps most importantly, a wide range of products which are not obviously computers or computer related actually contain and often depend on computer technology. Examples would include the domestic VCR, motor cars with computer controlled engine management or other systems, and aircraft. It is submitted that classification of contracts for the supply of such items as contracts for the supply of goods is unproblematic. Although the courts have at different times suggested different tests for classifying contracts, classification is essentially a matter of identifying the intention of the parties. The proper test in each case is therefore to ask what is the item sold/bought, and the answer, it is submitted, will be 'a VCR', 'a motor car' or 'an aeroplane' as appropriate. The presence of a computerised component will make no difference. The essence of the item supplied is not the computer program, but the goods of which the program is part. Since the contract in each case is for the supply of goods, the supplier will be subject to the statutory implied terms, and therefore subject to potential liability if the goods fail to function due to a Y2K related problem.

The result should be similar where items containing computerised components are supplied under a contract for work and materials, as, for instance, where a security engineer installs a burglar alarm using components some of which contain computerised components, or a mechanic fits a new computerised engine management system to a car. The contract in each case is one for work and materials, not sale of goods, because the essence of the contract is the supplier's skill and labour. However, the components are properly categorised as goods notwithstanding that they may be dependent on a computer chip or software program. Under a contract for work and materials the supplier is subject to (i) an implied term that the work will be performed with reasonable skill and care and (ii) implied terms that the materials supplied will be of satisfactory quality and reasonably fit for purpose and will correspond with description. Thus if the system fails because of a Y2K related problem in a component, the supplier will be strictly liable.

3.2 Computer Hardware

It seems clear that computer hardware should be regarded as goods. Certainly it has been held to be so where the hardware proves to be mechanically defective ( Amstrad plc v Seagate Technology Inc . ( 1998 )), and it is submitted that it should make no difference that the defect is due to a failure of 'embedded' software, so that if, say, a computer fails due to an error its real time clock, the supplier will be potentially liable. Although it may at first seem different from the cases considered above - perhaps because the item affected is recognisably a computer - the case is directly analogous to that of the supply of (say) a VCR, and this has been accepted by the courts. Thus in Toby Constructions Products Pty Ltd v Computa Bar (Sales) Pty Ltd ( 1983 ) the Supreme Court of New South Wales, held that a sale of a computer system comprising hardware and software was a sale of goods. A sale of hardware alone should therefore, a priori , be a sale.

3.3 Computer Software

Is the position then different where the defect relates to an item recognisable as software and recognisably separate from the hardware on which it runs? We must here consider the ways in which software is supplied. First, it is common for computers to be supplied as a package, with a bundle of pre-loaded software, including both operating system such as DOS and/or Windows, and applications software. The customer who buys such a package may be aware that he is acquiring software as part of the bargain but may make no mental distinction between the various components. Strictly speaking, however, in such a case the customer buys (a) property in the computer - i.e.: the hardware - and (b) a licence to use the bundled, pre-loaded software. It would therefore be perfectly possible in law to draw a distinction between the hardware and the software elements. Nevertheless, in the Toby case the Supreme Court of New South Wales held that a sale of a computer system comprising hardware and software was a sale of goods.

Strictly speaking Toby is not binding on an English court. It was accepted as good law by Sir Iain Glidewell LJ in St Albans City Council v International Computers Ltd ( 1996 ) but his comments were obiter. One argument which a supplier might run against it is that software supplied pre-loaded on a computer is not actually supplied to the customer under a contract but as a 'free' gift. The leading case on the status of 'free gifts' is Esso Petroleum Co Ltd v Commissioners of Customs and Excise (1976). In 1970 Esso promoted its petrol by issuing a collection of 'World Cup Coins', each bearing the likeness of a member of the England 1970 World Cup Squad. Motorists who bought four gallons of Esso petrol were entitled to receive a 'free' coin. The question arose whether the coins were 'sold' for the purposes of assessing the supplier's liability to purchase tax. The company raised two arguments against liability, (a) that there was no contractual supply of the coins, which were a genuine gift and (b) that if there was a contract it was not one of sale, since the consideration for the supply of the coin was not money[ 4 ]. Four members of the House of Lords accepted that if there was a contract the consideration for the supply of the coin was not money but the motorist's entering into the contract to purchase petrol, with the result that the contract was not one of sale. Two of their Lordships however preferred the view that there was no contract at all.

For our purposes, once the gift analysis is rejected, it matters little whether the contract is categorised as a sale, where the consideration is money, or a barter, for a non-money consideration. In either case the supplier will be liable for the goods supplied[ 5 ]. It is submitted that the 'gift' analysis is difficult to accept in the context of the Esso case itself. In the present context it should be rejected. The commercial reality is that pre-loaded software supplied bundled with a new computer is not 'free' in any real sense. The customer pays a global price for the whole package and has a contractual right to receive the software specified. It is submitted that, subject to the point to be made below, the contract should be analysed as being for the supply of software for a money consideration, in the sense that a part of the contract price should be attributed to the software although in reality there is one entire contract under which there is no separate allocation of part of the price to software. Even if, however, the software is held to be a gift, the supplier may be held liable for it. It has been held that where there is a contract for the sale of goods, the statutory implied terms apply to all goods supplied in pursuance of that contract, even if they are not sold. Thus, for instance, where a customer bought a soft drink in a returnable bottle and the bottle exploded because of a defect, the supplier was liable. Even though the bottle was not sold to the customer, it was supplied under a contract of sale ( Geddling v Marsh ( 1920 )).

This assumes, however, that software is 'goods', and this assumption is open to question. We must therefore now consider the status of a contract to supply software alone. This was not considered in the Toby case. It has now been considered in several English and one Scots case, but cannot be regarded as conclusively settled.

3.3.1 'Bespoke' and 'Off the Peg' Software: Goods and Services

One's first instinct might be to regard software as goods. After all, software is normally supplied on floppy disk or CD-ROM (or, increasingly, DVD) and therefore appears to be goods in the same way that a record or music CD is 'goods'. This analogy is, however, misleading. The software is not the disk or CD-ROM, but the program contained on it. Software does not have to be supplied in any physical medium. A program may be supplied on line, via the Internet. Or, where a software house produces a program for a client it may install it on the client's machine without supplying any physical medium. We must therefore consider the various ways in which software can be supplied. First is the situation where a software house is retained to produce a 'bespoke' program for a particular client. Second, the supplier may produce a 'bespoke' product using an existing proprietary program, as, for instance, where it produces a database by modifying a standard database package. Third, a software house may produce a more or less standard package for a number of customers, supplying it to each with appropriate modifications. Finally, there is the straightforward sale of a proprietary, or 'off the peg', package, sold over the counter in standard packaging through retail outlets in much the same way as books, music compact discs or any other product.

Some commentators propose that a distinction should be drawn between the production of a 'bespoke' program and the sale of a standard proprietary package. According to this approach the contract for production of bespoke software should be treated as one for work and materials, rather as a contract by a solicitor to draw a will or draft a contract, on the grounds that the essence of the contract is the supplier's skill as a programmer. In contrast, it is argued, the contract to supply a proprietary package can be regarded as one for the sale of the package. The consequence of this analysis is that under a contract for production of an 'bespoke' package the contract would include statutory implied terms that the supplier should perform his work with reasonable skill and care. He would therefore be liable only if shown to be negligent. In contrast, in the case of a sale of a proprietary product the contract would be one of sale, subject to statutory implied terms of quality, fitness for purpose and compliance with description which would impose strict liability on the retailer.

Some support for such a distinction between 'bespoke' and proprietary software can be found in the cases ( Saphena Computing Ltd v Allied Collecting Agencies ( 1995 )). The classification of contracts according to the 'substance of the contract' test is derived from the case of Robinson v Graves ( 1935 )[ 6 ] where it was held that a contract for a painter to paint a portrait was one for work and materials. Greer LJ observed in that case (at pp 587-8) that:-

However, Robinson v Graves has been the subject of strong criticism. In the earlier case of Lee v Griffin ( 1861 ) the court, adopting a different approach, ruled that wherever the contract results in the production of a finished product which is transferred under the contract, the contract is a sale and therefore held that a contract for a dentist to make and supply false teeth was a sale of the finished product. This test will often be preferable. In Australia the Supreme Court of Victoria has described Robinson v Graves as a 'hard case' and expressly rejected its test as 'illogical and unsatisfactory', 'wrong in principle' and 'too erratic' to be useful ( Deta Nominees v Viscount Plastic Products Ltd ( 1979 )[ 7 ]).

Distinguishing between mass produced and 'one off' products in this way at first seems attractive. However, whilst the distinction between the first and the fourth types of supply described above is readily apparent, the distinctions between those and the intermediate arrangements is less readily apparent. Moreover, in the present context application of the Robinson v Graves test may be misleading. First, it ignores the crucial question, whether software is 'goods' at all for the purposes of the Sale of Goods, or Supply of Goods and Services Act. Even if we accept, for a moment, that a contract for the production of a bespoke package is a contract for work and materials, it by no means follows that a contract for supply of a proprietary package is a contract for the sale of goods. The logic of such reasoning would lead to the conclusion that where a solicitor drafts a will or contract for a client there is a contract for professional services but that a sale of a pre-printed will form would be a sale of goods, as arguably would be the situation where a solicitor supplies a client with a contract in a pre-written office standard form. Conversely, if software is goods, application of the Lee v Griffin test would lead to the conclusion that a contract for the production and supply of a bespoke package may be one for the sale of goods. But even if we accept that a contract to produce a bespoke package is one for professional services, it by no means follows that the supplier's only liability is for failure to exercise reasonable skill and care. It is clear that under a contract for professional services, such as to design and build a structure, a term may be implied at common law that the finished structure will be reasonably fit for the client's purposes (see IBA v EMI Electronics and BICC Construction Ltd ( 1980 )). Similarly although electricity is not 'goods' for the purposes of the Sale of Goods Act, it has been held that a contract for the supply of electricity would include an implied term that the electricity should be reasonably fit for the buyer's purpose ( Bentley Bros v Metcalfe & Co ( 1906 ), and in Saphena Computing Ltd v Allied Collecting Agencies ( 1995 ) the Court of Appeal found a contract for the supply of software to be subject to an implied term of fitness for purpose without needing to classify it as a sale or otherwise.

3.3.2 The Medium is the Message

A different approach was adopted in the St Albans case. The defendants in that case, ICL, had supplied to the plaintiffs a software package it had produced for use by local authorities in administering the council tax. The package was therefore a more or less standard product, but supplied to several clients. Scott Baker J at first instance had no difficulty in treating the software as goods and the contract as one for the sale of goods ( St Albans City Council v International Computers Ltd ( 1995 )). In the Court of Appeal, however, Sir Iain Glidewell favoured a more complex analysis ( St Albans City Council v International Computers Ltd ( 1996 , p. 492)). He recognised that the word 'software' may be used to describe both the program algorithm itself and the floppy disk on which the program is supplied, but that 'goods' as defined in the Sale of Goods Act could only refer to a physical item. He concluded that the classification of the contract would depend on the manner in which the program is supplied. Where the program is supplied on a physical medium such as floppy disk or CD-ROM there is a supply of goods. Where, on the other hand, the program is supplied without the supply of any physical item, as where the program is installed on the customer's computer by the supplier's engineer, or is supplied on-line, there is no sale of goods. Nevertheless, he concluded, even if the contract is not one for the supply of goods, that does not preclude the implication into the contract, at common law, of terms that the program will be reasonably fit for the customer's purpose.

There appears to be some support for this approach in the decision of Steyn J in Eurodynamics Ltd v General Automation Ltd ( 1988 ) (although he expressly left the status of computer software as 'goods' undecided). It is submitted, however, that this approach places too much emphasis on the manner in which the program is supplied, which may be a matter of pure chance. In the Scots case of Beta Computers (Europe) Ltd v Adobe Systems Ltd ( 1996 ) Lord Penrose, the Lord Ordinary, took a different view, and concluded that a contract for the supply of proprietary software was sui generis , comprising some elements of a sale and some of a licence. Although the contract may involve the supply of a physical item, such as a diskette, the essence of the contract is the grant of a licence to use the software contained on the diskette, without which the disk itself is useless. A distinction is therefore drawn between the physical medium on which the software is supplied and the intellectual content of the software itself. Lord Penrose accepted that if the physical medium is defective so as to prevent access to the intellectual content, the goods would be unsatisfactory, or unfit for purpose, and seems to have accepted (at p. 609) that the same would be true if the information content of the disk had been 'corrupted' prior to supply.

There is much to be said for this approach. It is a mistake to assume that all contracts must fit into one of the existing categories of nominate contracts, and it is clear from decisions in other areas that a contract may be a hybrid, containing some elements of a sale and some of other types of contract[ 8 ]. Moreover, as Lord Penrose recognised, the essence of a supply of software is the grant of a licence. In most cases, and certainly in the case of proprietary software, the producer retains intellectual property rights in the program. Although the buyer may become owner of the physical medium on which the program is supplied, there is no question of his acquiring property in the program itself. There is thus a distinction between a contract for the supply of software and one for the sale of goods, which involves the transfer of property in the goods. It may be objected, of course, that a normal sale does not effect a transfer of intellectual property rights to the buyer. Thus when I buy a book I acquire property in the physical item but copyright in the content remains in the author. Similarly when I buyer a patented product I acquire property in the product but not the patent. However, these situations differ from that of software in two crucial respects. First, the reader of a book or the user of a patented product does not require a licence to read the book or use the product. In contrast a licence is necessary to use software, since every use of it involves making a copy of it. Second, in the case of the product, the essence of the contract is the supply of the physical thing, although the attributes which make it desirable may derive from the intellectual skill of the inventor protected by the patent.

3.3.3 Does Classification Matter?

Classification of the software contract is not an abstract exercise. It is only necessary when consequences follow from classification. Does classification of the software contract as one of sale or otherwise matter? The point has already been made that even if a contract is not for the sale or supply of goods, either because it is a contract for production of bespoke software and therefore one of work and materials, or because there is no physical item supplied, terms analogous to those implied by statute into contracts for the supply of goods may be implied at common law. The same is true, it is submitted, if one accepts Lord Penrose's analysis of the software contract as one sui generis . Nevertheless classification of the contract may in fact have legal consequences. First, it is not clear that the same range of terms would be implied at common law as under statute. Dicta in the cases suggest that even in a contract other than for the supply of goods there may be an implied term that the item supplied should be reasonably fit for the buyer's purpose. Under statute there will also be terms that goods should be of satisfactory quality and correspond with their description, but the statutory terms implied into contracts of sale have their origin in the common law and it is therefore submitted, there is no reason why the range of terms implied at common law should not be the same as that implied by statute. However, the common law required that goods should be of 'merchantable' - meaning 'saleable' - quality. That term is now replaced in statute by a requirement that goods be of satisfactory quality, and reference is made to a number of factors which are said to be 'aspects of quality'. The satisfactory quality term is a purely statutory construct. It is not clear that the term implied at common law would be the same.

Classification of the contract may have more practical significance. Sections 6 and 7 of the Unfair Contract Terms Act restrict any attempt by a supplier to contract out of the statutory implied terms in a contract for the supply of goods. As against a consumer the terms can never be excluded. As against a non-consumer any attempt to exclude the implied terms is subject to a test of reasonableness. If however the software contract is not one for the supply of goods but contains similar terms implied at common law, attempts to exclude or restrict those implied terms will be subject not to section 6 of UCTA but to section 3, which provides a lower level of protection for consumers and non-consumers alike. Under section 3 exclusions are not wholly prohibited but are subject to a test of reasonableness if they appear either in a contract with a consumer or in a contract on the supplier's written standard terms of business. As a result terms which would be wholly ineffective under section 6 would be subject to a test of reasonableness, and might be valid, under section 3, whilst some terms which would be subject to the reasonableness test under section 6 might wholly escape scrutiny under section 3.

Another consequence of classifying software as goods, at least in some cases, is that strict liability is imposed on the supplier of goods. Thus in the case of proprietary software sold by retail, strict liability will be imposed on the retailer. Now one may question whether, in a modern context, it should be the retailer who bears primary liability for defects in mass produced and mass marketed products, sold pre-packaged so that the retailer has no opportunity to check the contents. This is true of most items but is especially true of software where unforeseen 'bugs' may be undetectable even to an expert. As has been noted, imposition of liability for Y2K related problems may be ruinous. We may doubt if liability for such losses should be imposed at all, but is it right that strict liability should fall on the retailer? The justification for retailer liability is that it provides a conduit by which liability may be passed back, via a chain of contractual claims, to the manufacturer actually responsible for defects in products ( Adams, 1997 ). We may note however that the contractual route has been by passed with the creation of a regime of strict manufacturer liability for injury caused by defective products (Consumer Protection Act 1987 see Howells ( 1999 )). We may note, too, that the liability chain may break down if, for instance, the manufacturer is insolvent, or successfully excludes his liability to his customer[ 9 ]. In those cases liability will fall and remain on the retailer.

3.3.4 An Alternative Analysis

A case may nevertheless be made for retailer liability. The system is familiar, especially to consumers. Moreover, if the manufacturer is insolvent, retailer liability provides the customer with a viable target for a claim. And the retailer may be as well, if not better, placed as the manufacturer to re-distribute losses through pricing adjustments. Nevertheless, there is an alternative, which is particularly apposite in the present context. The user of software needs a licence to use the program. The software producer normally seeks to impose such a licence to control the use of the software. Under a contract to develop a 'bespoke' program the licence will normally be an express term at the core of the development contract. In the case of proprietary software the user is sometimes required to register the software, in return for the provision of information about upgrades etc. In others he must obtain a pass word. Often the licence takes the form of a 'shrink-wrap' licence, under which it is claimed that the user agrees the terms of the licence simply by opening the packaging and loading the program. The contractual status of such licences is a subject of some debate. It is submitted however that there is in principle no reason why such a licence should not be construed as a contract between software producer and end user. There would be no problem finding consideration to support such a contract, in the manufacturer's agreement to grant a licence, and the customer's agreement, in return, to abide by the terms of the licence[ 10 ]. Once we accept the contractual analysis of the software user licence, there is contractual privity between software manufacturer and user, and that contract can be used as a vehicle for the imposition of implied terms of quality and fitness for purpose and so on at common law. Strict liability for the quality of the software product can thus be imposed, via the licence, on the manufacturer without the need to involve the retailer.

This analysis is very attractive. Above all it achieves the objective of giving the customer the protection of a strict liability quality regime without the need artificially to classify software as 'goods', and locates liability in the right place, on the manufacturer. However, ultimately there are several objections to it, at least in the context of consumer sales which account for an increasingly large share of sales of proprietary software. First, it is unfamiliar. Second, for reasons outlined previously, it might be possible for the manufacturer to exclude terms implied into a licence in this way where exclusion of the statutory terms would not be allowed a seller. And third, in many cases the manufacturer of proprietary software will be outside the jurisdiction of the English courts. It would be unacceptable to conclude that buyers of software deserve the protection of an implied terms regime and then deprive them of a suitable defendant to sue. We may conclude, therefore that there are good practical and policy reasons to impose primary liability for defective proprietary software, at least in a consumer context, on the retailer, even if the analysis of software as 'goods' is intellectually unsatisfying.

4. Does Non-compliance Involve a Breach of Implied Terms?

We may conclude (1) that many of the items likely to be affected by the Y2K problem are 'goods', and that whilst the status of contracts for the supply of software remains unclear, (2) it is at least arguable that some contracts for the supply of proprietary software are contracts for the sale of goods, and (3) that even if software is not analysed as goods, the contracts under which they are supplied may contain implied terms equivalent to (some of) those implied by statute into contracts of sale. The consequence is that the supplier of the software or goods in each case will be subject to liability for breach of the implied terms. We must therefore now consider the application of those terms to software, or any other item, affected by a Y2K related problem.

As noted earlier, all contracts for the supply of goods include implied terms that the goods supplied should (a) where sold by description, correspond with that description (Sale of Goods Act 1979 s.13); (b) be of satisfactory quality (Sale of Goods Act 1979 s.14(2)); and (c) where the buyer makes known the purpose for which the goods are required, be reasonably fit for that purpose (Sale of Goods Act 1979 s.14(3)). The terms are implied into all contracts for the supply of goods, and apply to all goods supplied in purported performance of the contract. They therefore apply to sales of hardware alone, sales of bundled hardware and software, sales of items which are 'computerised' in the sense that they contain an embedded chip, such as VCRs and so on. They also apply to a sale of software alone, if in the particular case the software is properly classified as 'goods'.

4.1 Goods Must Comply with Description

The requirement that goods must correspond with description applies where goods are supplied by description, and it has been held that in this context 'description' is limited to those words which identify 'essential commercial attributes' of the item sold ( Ashington Piggeries Ltd v Christopher Hill Ltd ( 1972 )). In most cases it will not be relevant where goods are found to be affected by a Y2K related problem. The exception will be where goods are expressly sold as being 'Y2K compliant'. Amongst the factors which the court will consider when deciding whether descriptive words form part of the description by which goods are sold is whether they were influential in the sale and whether the buyer could reasonably be expected to rely on the words in question when buying the goods ( Harlingdon & Leinster v Christopher Hull Fine Art Ltd ( 1991 )). The test is flexible, but essentially it seems that a statement will be regarded as part of the contractual description if it is such that it would be categorised as a term of the contract. It is submitted that in the present context it would be open to a court to hold that a specific claim that goods are 'Y2K compliant' is contractual and part of the description. The result would be that if such goods prove not to be compliant, the seller will be strictly liable, even if the claim of compliance was made by the manufacturer, say in promotional literature and the supplier was ignorant of its falsity. The one problem, however, may be in deciding what 'Y2K compliant' means. For instance, in 1998 The Guardian reported that certain Compaq computers were being marketed as 'Y2K compliant' on the basis of a test by the US National Software testing Laboratory, but that according to a different test they might not be regarded as compliant ( Guardian 1998 ). Compaq responded that their machines would only be non-compliant when used with software which accessed the computer's real time clock chip directly, and that such software would violate basic programming principles. In some cases under section 13 the courts have accepted that where descriptive words have a technical meaning, they should be interpreted in accordance with that meaning. Thus glass sold as 'safety glass' is not necessarily warranted as 'safe' ( Grenfell v Meyrowitz Ltd ( 1936 )). In the present context it might therefore be possible to establish that 'Y2K compliant' has a particular meaning, and does not necessarily guarantee that an item will be wholly free from Y2K related problems - so that, for instance, there would not necessarily be a breach if a computer sold as 'Y2K' compliant malfunctions only when used with a particular piece of non-standard software. It is submitted, however, that in interpreting descriptive words in this context the courts should adopt an objective test and interpret words in accordance with the reasonable understanding of the persons to whom they are addressed. Thus a claim of 'Y2K compliance' addressed to computing specialists might have a different meaning from the same claim addressed to consumers, who might reasonably interpret it as an absolute guarantee[ 11 ].

4.2 Goods Must be of Satisfactory Quality

The core requirement as to quality is that that the goods supplied under the contract should be of 'satisfactory quality'. Goods are of satisfactory quality 'if they meet the standard that a reasonable person would regard as satisfactory, taking account of the description of the goods, the price (if relevant) and all the other relevant circumstances' (Sale of Goods Act s14(2)). The standard is thus one of reasonable expectations. This is amplified by section 14(2B) which lists a number of factors to be taken into account as 'aspects of quality' in an appropriate case. They are:-

Factors (b) and (c) are unlikely to be relevant in the present context. Safety - or lack thereof - is always likely to be an important factor so that even a defect which is 'minor' in the sense that it may be easily or cheaply repaired may make goods unsatisfactory if it makes them unsafe, especially if it is hidden. The Y2K problem does not make items inherently unsafe in the way that (say) a wiring defect causing a computer to short circuit might do. On the other hand where software or computer dependent items are supplied for use in a safety critical context - such as fire alarms, or medical equipment - a Y2K problem which causes a malfunction may make the item unsafe and possibly therefore unsatisfactory.

4.2.1 The Problem of Durability

However, the most important of the factors, and the most difficult to apply, in the present context is likely to be 'durability'. An item which breaks down or wears out sooner than might reasonably be expected may be said to lack durability. But how is this to be applied to software?

As noted earlier, in many cases lack of Y2K compliance will necessitate the replacement of software, systems or consumer 'durables'. Can such items be said to lack durability on the basis that they cease to function (properly) on 1 January 2000? The list of factors in section 14(2B) was introduced into the Sale of Goods Act 1994 and there is as yet only one reported case in which they were considered ( Thain v Anniesland Trade Centre ( 1997 )). In that case the court held that a second hand Renault 19 which was 5- 6 years old and had done some 80 thousand miles, for GBP 2995, and which broke down some two weeks after sale when its gear box failed, was not unsatisfactory as durability was not a quality which a reasonable buyer would expect. However, the case provides very little guidance for the problem under consideration. The court was clearly influenced by the fact that the plaintiff had had the option to purchase an extended warranty which would have covered the cost of repairs to the gear box. More importantly, the court found as fact that the defect was not present in the car at the time of sale and developed later, and was a fault which could have developed at any time in a car of that age and mileage.[ 12 ] The situation is very different where non-Y2K goods or software are sold. In that case the 'defect' - the lack of compliance - is present at the time of sale[ 13 ]. To that extent the goods may be said to be 'doomed from the start'.

So, can it be said that software, or a computer dependent item, which is non-Y2K compliant and will therefore fail so as to require replacement or expensive repair on 1 January 2000 lacks durability? The answer must depend on two factors: the date of purchase and the reasonable expectations of the buyer. For how long can the buyer reasonably expect to have use of the goods before having to replace them? It is well known that software and consumer durables are regularly 'upgraded'. Is the reasonable life of a product the period between upgrades? It is submitted that in relation to consumer durables it is unlikely to be accepted that the consumer should expect to replace an item such as a VCR or car each time it is upgraded. Some consumers may do so, but that is a matter of choice. The reasonable expectation of durability may also vary according to the nature of the item - for instance, VCRs may now be expected to have greater longevity than washing machines. In relation to a motor car some repairs may reasonably be expected to be necessary after three or four years. But can a software supplier argue that because its products are upgraded each year, the consumer can expect no more than one year's use? It is submitted that such an argument has little chance of success. It is the buyer's not the seller's expectations which should be crucial. Most buyers buy computers and software in the same way that they buy other durables. They do not expect to have to replace items every year. Conversely, can a purchaser argue that a program is not expected to 'wear out' at all, and that provided that he is willing to forego the benefits of upgrades, improvements and new applications, he should be entitled to expect a program to last forever? Even if successful, this argument is unlikely to assist a plaintiff. Even if it is held that an early release of software lacks 'durability' because it is non Y2K compliant and fails in 2000, the buyer will almost certainly be held to have lost the right to reject, and damages for any breach of contract will be the difference between the value of the goods as supplied and their value had they conformed to the contract. Where the purchaser has had extended use from a program it is likely to be held that such difference in value is minimal or even nil.

Between the extreme interpretations of durability it is difficult to predict how the requirement will be applied. One factor which the court may take into account is the period for which the manufacturer supports a particular program. It is submitted that in the ordinary way of things a buyer may reasonably expect to use a software program for five years or so before replacing it entirely. He may reasonably expect to purchase upgrades during that time to obtain improved performance, but not to have to replace the program entirely. Nor can it make a difference that a Y2K problem can be fixed by purchasing additional software rather than replacing a program in its entirety. The purchase of upgrades is a matter of choice, not necessity.

4.2.2 Other Factors

Other factors may also be relevant. In Eurodynamics ( 1988 ), Steyn J indicated that 'bugged' software would not necessarily be considered to be unmerchantable, in accordance with the law at that time, since a buyer of new software must be taken to know that it may contain bugs, which will be ironed out over time. However, Steyn J appeared to have in mind the sort of unpredictable error which can only be discovered by running a program. The Y2K problem is not a true 'bug' in this sense. It was never unpredictable, merely unpredicted. Even if applicable at all in the present context, this reasoning would depend on the time at which the software or product was supplied. The buyer reasonably does not expect that software will continue to be supplied with a bug after the bug has been discovered, so that software and products supplied after the Y2K problem became known should reasonably be expected to be Y2K compliant. Steyn J also took into account the fact that the defendants in Eurodynamics offered an on site warranty under which they undertook to support the system and fix bugs which appeared. To this extent software may be different from other items. It has been held in the Court of Appeal that a manufacturer's warranty did not prevent a defective car from being unmerchantable ( Rogers v Parish (Scarborough) Ltd ( 1987 )). The availability of such a warranty, or a willingness to provide free upgrades may therefore prevent non-compliant software being regarded as 'unsatisfactory'. Alternatively, if software is upgraded, or 'repaired', free of charge, the buyer may suffer no loss even if there is a breach of contract[ 14 ].

4.2.3 When is Quality Assessed?

Although quality is judged at the time when risk passes from seller to buyer, which in most cases will be the moment of delivery of the goods, it is well settled that in making that assessment account must be taken of 'defects' not known at that time but which come to light later ( Kendall v Lillico ( 1969 )). Thus where a non-Y2K compliant item was supplied before the problem was understood, the lack of compliance may still be regarded as a 'defect' present at the time of supply, and, subject to the limitations on right to reject and the rules on limitation, give rise to a claim. It is important to note that to that extent liability under the implied terms is strict and there is no 'state of the art' defence, unlike, say, in product liability law (see Howells ( 1999 )). In general the limitation period for actions in contract is six years from the time of breach (Limitation Act 1980 s.5), which occurs at the time of delivery. Since the Y2K problem was widely known by 1994 at the latest it might be thought that limitation will therefore preclude actions in cases where the seller could not reasonably have known of a possible problem at the time of supply. However, where the contract is contained in a deed - as is often the case in large scale building or engineering contracts, and may also be the case for some software development contracts, the limitation period is twelve years (Limitation Act 1980 s 8), and so there may be situations where a claim could be brought in relation to a supply some time before 1994. In that case the fact that the 'defect' was at that time unknowable will provide the seller with no defence.

4.2.4 'Merchantable' or 'Satisfactory' Quality?

One other point must be made here. In relation to goods sold prior to 3 January 1995 the relevant requirement was that they should be of 'merchantable' rather than 'satisfactory' quality. It is likely, therefore, that some damages claims for lack of Y2K compliance could be brought under the old law, under which there was no equivalent to the list of factors now in Sale of Goods Act s.14(2B). However, it is clear that many of those factors, including 'durability', were considered by the courts when making the assessment of merchantability under the old law. Moreover, most commentators are agreed that the change from 'merchantable' to 'satisfactory' quality involved a change of terminology but not of substance. It is therefore submitted that the outcome of a claim relating to a particular item should be the same whether the item is judged by the standard of 'merchantable' or 'satisfactory' quality.

4.3 Fitness for Purpose

The third implied term requires that where goods are sold in the course of a business and the buyer expressly or impliedly makes known to the seller any particular purpose for which the goods are bought, the goods must be reasonably fit for that purpose except where the circumstances show that the buyer does not rely on the seller, or that it is unreasonable for him to do so (Sale of Goods Act 1979 s.14(3)). In many cases this requirement will add little or nothing to the 'satisfactory quality' requirement. Fitness for all common purposes is now an aspect of quality for the purposes of section 14(2). Thus where goods which have only one common purpose are bought for that purpose but prove unfit, they are likely to be considered unsatisfactory. In addition since the goods have only one common purpose, the buyer will be taken to have impliedly indicated his purpose when buying them, and there will also be a breach of the fitness for purpose requirement. Where goods which have a range of common purposes are bought for one of those common purposes, they may be considered unsatisfactory if they prove unfit for that purpose (or for any other of their common purposes). However, in that case the buyer will only be able to rely on the fitness for purpose condition if he indicated to the seller that he was buying the goods for that particular purpose.

The fitness for purpose term will come into its own where the buyer acquires goods for a purpose outside their range of common purposes and the goods, although reasonably fit for their common purposes, prove unfit for that particular purpose. The goods are likely to be considered 'satisfactory'. The buyer will then be forced to rely on the fitness for purpose term, and will have to establish that he indicated to the seller the particular purpose for which the goods were required. If he can do so, the seller will be liable unless he can establish either that the buyer did not rely, or that it was unreasonable for him to have relied, on the seller's skill or judgment. Where the buyer relied partly on his own and partly on the seller's skill or judgment the seller is liable for any lack of fitness arising from matters within his area of expertise ( Cammell Laird & Co v Manganese Bronze and Brass Ltd ( 1934 )). Thus if B asks S to supply a computer system and specifies the use of certain hardware components but leaves S to select other components and software, S will not be liable if the system proves unfit for B's purpose due to the unfitness of B's specified components.

Case law shows that the buyer must specify his purpose with sufficient particularity to enable the seller to exercise his skill or judgment[ 15 ]. It should be noted however that the seller's liability is strict. Thus provided that the buyer has indicated his purpose the seller will be liable if the goods prove unfit for the specified purpose, even if the seller has exercised all reasonable skill or judgment, unless the seller can prove either that (a) the buyer did not rely on the seller's skill or judgment or (b) it was unreasonable for him to do so. The seller may therefore be able to avoid liability in an appropriate case if he disclaims expertise in relation to the buyer's particular purpose (see below).

If the requirements for implication of the fitness for purpose term are satisfied, the seller will be in breach of contract unless the goods supplied are reasonably fit for the buyer's indicated purpose. Obviously, therefore, if the buyer specifically requests that a product be Y2K compliant, it may be inferred that that is the specified purpose and the product must therefore be compliant even if lack of compliance would not make it unsatisfactory. However, even where the buyer has not specified Y2K compliance, lack of compliance may make a product unfit. Correspondence with the fitness term is judged at the moment when risk passes from seller to buyer but reasonable fitness for purpose must require, inter alia , that goods are, at the time of supply, reasonably durable[ 16 ]. Thus an item supplied for a particular purpose which breaks down sooner than might reasonably be expected due to a defect present at the time of supply may be judged to have been unfit at the moment of supply. Subject therefore to the comments about 'durability' made earlier, section 14(3) may provide a buyer with a remedy where an item proves to be non-Y2K compliant having been purchased for a particular purpose which was notified to the seller at the time of the sale.

One particular situation must be considered. Where goods are bought for use in a particular system which is specified by the buyer it will be an implied term that the goods should be reasonably fit for use in that system. Thus, for instance, if a person buys software specifying that it is to be used on a particular model of p.c. it will be an implied term that the software will be reasonably fit for use on that p.c. Now, suppose that the p.c. functions satisfactorily with most software, and the software package in question functions satisfactorily on most hardware, but due to the way in which the software works and the particular characteristics of the p.c. the software is non-Y2K compliant when used on that particular make of p.c.[ 17 ]. Provided that the characteristic of the p.c. is known to the software supplier, or the p.c. is a common model and the problem is caused by a characteristic of the particular model of p.c. (so that it may be said that the characteristic ought to be known) it may then be argued that the software is of satisfactory quality but is not reasonably fit for the buyer's purpose. Where however the non-compatibility or error is due to a unique characteristic of the buyer's particular system, the supplier will not be liable. Suppose, for example, that a person buys software for his p.c., specifying that it is for use on, and providing details of, the machine, and the software proves unsuitable, perhaps because of an idiosyncratic manufacturing error in that particular p.c., or because the purchaser has previously modified its specification. It can be argued that the purpose indicated by the buyer is that the software is for use on that particular machine, but it cannot be inferred that the supplier undertook that the software would be suitable for that particular idiosyncrasy[ 18 ]. The explanation for this approach is that section 14(3) requires the buyer of goods to disclose his purpose in sufficient detail to enable the seller to exercise his skill and judgement.

4.4 Warranties and Disclaimers

It is clear therefore that the implied term that goods should be of satisfactory quality may provide a means by which a person who has bought software, or any other product, which is non-Y2K compliant, but its application may be unpredictable and difficult. However, the buyer may be able to obtain a greater measure of protection under the other implied terms. Where a person buys software he may therefore be advised to (a) request a specific warranty that it is Y2K compliant and (b) specify the purpose for which it is required including, where appropriate, details of the system on which it is to be run. If he does so and the software proves not to be compliant he may well be able to hold the seller liable, either for breach of an express warranty of compliance, or for breach of the implied terms in section 13 (description) or section 14(3) (fitness for purpose). A supplier who gives such a warranty, or supplies goods knowing the buyer's intended purpose, does so at his peril for by so doing he assumes liability for the goods and any defect in them which renders them non-compliant or unfit for the buyer's particular purpose, subject only to the restrictions outlined above. The supplier may therefore be advised to qualify any response he gives. Although attempts to exclude the implied terms are controlled by the Unfair Contract Terms Act, it has been held that a disclaimer of expertise in a particular case can prevent a sale being by description as where such a disclaimer is given the buyer cannot reasonably be expected to rely on the description ( Harlingdon & Leinster v Christopher Hull Fine Art Ltd ( 1991 ). The same reasoning must apply to the implied undertaking of fitness for purpose, since that is expressly stated not to apply if either the buyer does not rely on the seller's skill or judgement, or it is unreasonable for him to do so, and it must be unreasonable to rely where the seller has expressly disclaimed expertise. It is submitted, however, that neither of the terms in section 13 or section 14(3) should be capable of being excluded by a general disclaimer of expertise in a standard term contract, which should be construed as an attempt to exclude the implied terms and be subject to the Unfair Contract Terms Act.

5. Conclusion

This piece has considered whether the law of sale and supply of goods may provide a means by which a plaintiff may seek to recover compensation for losses caused by the 'millennium bug'. It has been argued, first, that the extensive discussion of the question whether software should be classified as 'goods' and whether it is supplied under a contract of sale may in fact have focused on the wrong issues. There is no good reason for liability to differ according to the manner in which software is supplied and therefore a classification which concentrates on the medium in which a program is supplied is open to objection and although there may be a more justifiable distinction between 'bespoke' and 'proprietary' software, in the final analysis it may be that the classification of the contract makes little difference to the legal obligations of the parties. The best argument for treating software as 'goods' is that it is commercially exploited as any other product. In the U.S.A. it has been held that software is goods ( Advent Systems Ltd v Unisys Corp. ( 1991 )[ 19 ]), and thus within the Uniform Commercial Code, for this very reason. It is clear, however, that the court was keen to apply the UCC to software contracts. The UCC is intended to be a complete code for commercial contracts, and has far wider ambit than the UK Sale of Goods Act. It is clear that many of the UK Act's provisions (such as those on the transfer of property) could not be applied to software as such (although they would apply to the physical medium in which the software is supplied). Moreover, some of the Act's provisions are less well suited to modern contracts than those of the general law[ 20 ]. The only provisions which are likely to be applicable to software in the normal run of things are the implied terms in sections 12- 15, and, as has been demonstrated, there is no need to classify software as 'goods' in order to apply those terms. Subject to the possibility that the supplier's power to exclude or modify the terms may be more restricted where the item supplied is categorised as 'goods', it has been suggested that implied terms equivalent to those implied by statute into contracts for the supply of goods could be implied at common law into a contract to produce bespoke software, or even into a software licence. The implied terms themselves, and, in particular, the requirement that goods be of satisfactory quality are however likely to prove difficult to apply to software and related items which are non-Y2K compliant.

However, these questions mask more difficult issues of policy. The crucial question in the present context, it is submitted, is whether there should be strict liability for losses caused by goods and software proving to be non-Y2K compliant, and, if so, where that loss should lie. Application of sale law imposes strict liability on the contractual supplier for what is essentially an intellectual product, and without the benefit of a development risks defence. It may seem strange that when the law of tort has shied from imposing extensive liability for inaccurate information widely disseminated, even where the information was given negligently, the law of contract should impose liability without proof of negligence. In response it may be argued that software is not 'pure' information and that to draw a distinction between information products and physical products is unreal in an information age. After all, software products are marketed and exploited on a commercial basis. They are tools or entertainments. If a tool fails to do the job for which it is bought, or if an item bought for pleasure fails to provide the expected entertainment, we would expect the buyer to get redress. Why should information products be different? Those who exploit software commercially should bear the risks associated with that commercial activity. The law should not create special categories for a particular industry. And if it is illogical to hold the retailer strictly liable for errors in the proprietary software he sells, it is no more illogical than it is to hold him strictly liable for defects in any other mass produced product.

But the present problem is different. The Y2K problem is essentially a design fault affecting the products of an entire industry. To permit contract actions to be brought for Y2K problems on the contract of supply exposes the seller to an indeterminate liability for economic losses to an indeterminate number of potential plaintiffs. To impose liability may cripple an entire industry of strategic importance. One is tempted to react 'Serves 'em right', but this is unreal. There are serious questions of economic and social policy to be considered in deciding whether or not to impose liability. If those questions are addressed in sale law at all - and it is submitted that they are considered in deciding whether to impose liability - they are not overtly referred to. For this reason it is submitted that sale law is a less appropriate vehicle for dealing with the liability issues which the Y2K problem may raise than the law of tort or product liability where policy issues are openly considered. Given its practical advantages, however, it seems unlikely that potential plaintiffs who can bring a case on a contract of sale or supply will shy from so doing.

Cases

Amstrad plc v Seagate Technology Inc . (1998) 86 BLR 34.

Advent Systems Ltd v Unisys Corp. (1991) 925 F. 2d. 670.

The Aramis . [1989] 1 Lloyd's Rep 213.

Ashington Piggeries Ltd v Christopher Hill Ltd [1972] AC 441.

Bentley Bros v Metcalfe & Co [1906] 2 KB 548.

Bernstein v Pamsons Motors (Golders Green) Ltd . [1987] 2 All ER 220.

Beta Computers (Europe) Ltd v Adobe Systems Ltd [1996] SLT 604.

Bristol Tramways Carriage Co Ltd v Fiat Motors Ltd . [1910] 2 KB 831.

Cammell Laird & Co v Manganese Bronze and Brass Co Ltd [1934] AC 402.

Clay v Yates (1856) 1 H&N 73.

Crowther v Shannon Motor Co [1975] 1 All ER 139.

Deta Nominees v Viscount Plastic Products Ltd (1979) VR 167.

Dick Bentley Productions Ltd v Harold Smith Motors Ltd [1965] 1 WLR 623.

Eurodynamics Ltd v General Automation Ltd (6 September 1988, unreported), QBD.

Gardiner v Gray (1815) 4 Camp 144.

Geddling v Marsh [1920] 1 KB 668.

Grenfell v Meyrowitz Ltd [1936] 2 All ER 1313.

Griffiths v Peter Conway Ltd [1939] 1 All ER 685.

Harlingdon & Leinster v Christopher Hull Fine Art Ltd [1991] 1 QB 564.

Hyundai Heavy Industries Ltd v Papadopoulos [1980] 2 All ER 29, [1980] 1 WLR 1129, HL.

IBA v EMI Electronics and BICC Construction Ltd (1980) 14 BLR 9.

Jones v Just (1868) LR 3 QB 197.

Kendall v Lillico [1969] 2 AC 31.

Lee v Griffin (1861) 1 B & S 272.

Marcel Furriers v Tapper [1953] 1 WLR 49.

Oscar Chess Ltd. v Williams [1957] 1 WLR 370.

R v Wood Green Profiteering Committee ex p Boots Cash Chemists (Southern) Ltd (1920) 89 LJKB 55.

Randall v Newson (1877) 2 QBD 102.

Robinson v Graves [1935] 1 KB 579.

Rogers v Parish (Scarborough) Ltd [1987] QB 933.

Rose and Frank Co v JR Crompton & Bros. Ltd [1925] AC 445.

Saphena Computing Ltd v Allied Collecting Agencies [1995] FSR 616.

Slater v Finning [1996] 3 All ER 398.

St Albans City Council v International Computers Ltd [1995] FSR 686.

St Albans City Council v International Computers Ltd [1996] 4 All ER 481 (CA).

Stocznia Gdanska SA v Latvian Shipping Co [1998] 1 All ER 883.

Thain v Anniesland Trade Centre [1997] SLT Sh Ct 102.

Toby Constructions Products Pty Ltd v Computa Bar (Sales) Pty Ltd [1983] 2 NSWLR 48.

References

Journal Articles:

Adams J N (1997) 'The snark was boojum you see' Edinburgh Law Review p386.

Howells G (1999) 'The Millennium Bug and Product Liability' Journal of Information, Law and Technology (JILT) 1999 (2). < http://elj.warwick.ac.uk/jilt/99-2/howells.html >

Macdonald E (1999) 'Y2K and Contractual Exemption Clauses' Journal of Information, Law and Technology (JILT) 1999 (2). < http://elj.warwick.ac.uk/jilt/99-2/macdonald.html >

The Guardian ' Online ' 4 June 1998, p6 'At the Third stroke it will be 1900 precisely'.

Footnotes

1 . For instance, The Guardian reported on 4.6.98 that certain Compaq computers marketed as Y2K compliant were, strictly speaking, not, at least when used with software which directly accesses the computer's real-time clock.

2 . These words appear as an appended 'note' immediately after the space for signature of the Pledge. The preamble to the pledge contains a similar statement but couched in rather less firm language. It states that 'A signatory to Pledge 2000 is not constrained from taking appropriate action, legal or otherwise where this is due'. This could raise a number of nice problems. It is not clear if this statement is meant in any way to qualify the much more concrete statement in the notes, referred to in the text above. If so, there is enormous scope for debate, and ultimately litigation, about what is 'appropriate action' and whether legal action is 'appropriate' in any case, and when such action is 'due'.

3 . Scots law does not recognise the distinction between 'conditions' and 'warranties'.

4 . A contract is only one for the sale of goods if it involves transfer of property in the goods for a money consideration.

5 . Although classification of the contract as one of sale or otherwise may affect the customer's right to reject goods.

6 . See also Clay v Yates ( 1856 ) where a contract for a printer to print a book was interpreted as one for work and materials.

7 . See also Marcel Furriers v Tapper ( 1953 ); R v Wood Green Profiteering Committee ex p Boots Cash Chemists (Southern) Ltd ( 1920 ).

8 . See e.g. Hyundai Heavy Industries Ltd v Papadopoulos ( 1980 ), where a contract to build and supply a ship was held to be a contract of sale but with some of the characteristics of a building contract. See also, Stocznia Gdanska SA v Latvian Shipping Co ( 1998 ).

9 . This is a particular problem where the retailer's customers are consumers. UCTA s.6 will then prevent the retailer from excluding liability to his customers, but may permit his supplier to exclude liability to the retailer.

10 . See the similar reasoning, in a different context, in The Aramis ( 1989 ). The key to finding a contract is identifying a mutual intention to contract which can be reduced to an offer and an acceptance.

11 . Compare the approach to statements about motor cars in cases such as Oscar Chess Ltd. v Williams ( 1957 ) and Dick Bentley Productions Ltd v Harold Smith Motors Ltd ( 1965 ).

12 . The court may also have been influenced by the fact that the defendant had offered to replace the car, but the plaintiff refused those offers and demanded replacement of the gearbox.

13 . It may be noted in passing that the finding in Thain ( 1997 ) that the defect was not present in the gearbox at the time of sale may be questioned. The court seems to have ignored the fact that the reason the gear box failed was that it was seriously worn at the time of sale and was therefore, in all probability, then doomed to fail in a fairly short time.

14 . If the buyer unreasonably refuses the manufacturer's offer to upgrade or 'repair' he may be held to have unreasonably failed to mitigate his loss, thus reducing or even cancelling his right to damages.

15 . See Slater v Finning ( 1996 ); Griffiths v Peter Conway Ltd ( 1939 ); cf Ashington Piggeries Ltd v Christopher Hill Ltd ( 1972 ).

16 . See for instance Crowther v Shannon Motor Co ( 1975 ).

17 . As in the case of the Compaq computers mentioned earlier which use a 2 digit field to record the date in the computer's real time clock. Such machines would not function correctly with software which takes the time direct from the r-t-c, but such software would function adequately on machines whose r-t-c uses a 4 digit field.

18 . See Slater v Finning ( 1996 ); Griffiths v Peter Conway Ltd ( 1939 ); cf Ashington Piggeries Ltd v Christopher Hill Ltd ( 1972 ).

19 . The case is interesting because it did not concern a standard proprietary package.

20 . For instance, the rigid classification of terms as 'conditions' and 'warranties' and the rules on mistake and frustration in sections 6 and 7.