HM Revenue and Customs v The Rank Group Plc

This appeal is by The Commissioners for Her Majesty’s Revenue and Customs (‘HMRC’). The respondent is The Rank Group plc (‘Rank’). The appeal concerns certain types of slot machine. It raises the issue whether the takings from such machines were or were not entitled to the VAT exemption for gaming activities applicable at the material time, namely 1 October 2002 to 5 December 2005. The critical question, posed by reference to the VAT legislation then in force, is whether ‘the element of chance in the game [played on such a machine] is provided by means of the machine’ (my emphasis). If, as HMRC claims, it was, the machine’s takings were taxable. If, as Rank claims, it was not, they were exempt.

HMRC has described the disputed machines as ‘reconfigured machines’. Rank takes exception to that, preferring to call them ‘section 16/21 machines’, references respectively to section 16 of the Lotteries and Amusements Act 1976 and section 21 of the Gaming Act 1968 under which Rank says the machines were regulated. HMRC in turn objects to that description on the basis that it begs the question at issue: because if they were section 16/21 machines, they would have been entitled to the tax exemption that Rank asserts and HMRC deny. HMRC’s position is that the machines were not section 16/21 machines, but are ‘Part III machines’, that is machines regulated by Part III of the Gaming Act 1968, section 26(2) of which includes the like language I have quoted in paragraph 1. If so, it is said that they were not entitled to the tax exemption that Rank asserts. I shall call the equipment in dispute ‘the disputed machines’.

There is no dispute that an amendment to the claimed exemption that came into force on 6 December 2005 ended the distinction between the types of machine that fuelled the issue raised by the appeal, but the point remains important. If Rank is right that the takings from the disputed machines enjoyed the claimed exemption, it is common ground, following a preliminary ruling in these proceedings by the Court of Justice of the European Union (‘the CJEU’), that Rank is entitled to invoke the EU principle of fiscal neutrality to assert that the takings on different machines which did fall to be taxed under the domestic VAT provisions ought not to have been taxed, and that it is entitled to reclaim the VAT it paid in respect of such takings. The machines whose takings were so taxed were indisputably Part III machines.

The appeal is, in money terms, potentially significant. Rank’s tax reclaim for the period is of the order of £35m. The outcome of the appeal may also govern other outstanding claims by Rank and other operators. HMRC assert that a failure by it on this appeal may cost it more than £1 billion. Rank is sceptical about that but the rival views on that are anyway legally irrelevant. The appeal is against one aspect of an order dated 12 June 2009 made by Norris J.

By decisions released on 15 May and 19 August 2008, the VAT and Duties Tribunal (Theodore Wallace and A.J. Ring CTA, ‘the tribunal’) allowed appeals by Rank against decisions of HMRC rejecting ‘voluntary disclosures’ by Rank, being in effect claims for the repayment of tax said by Rank to have been overpaid. The first appeal was the ‘Bingo appeal’, the second ‘the Slots appeal’. Norris J’s judgment ([2009] EWHC 1244 (Ch); [2009] STC 2304) dismissed HMRC’s appeals against both decisions.

HMRC sought permission for a second appeal to this court against Norris J’s order, which Jacob LJ gave on 11 August 2009. HMRC advanced two grounds of appeal in the Bingo appeal and three in the Slots appeal. On 20 April 2010, with the parties’ consent, this court (the Chancellor of the High Court, Etherton and Elias LJJ) made a reference to the CJEU under Article 267 of the Treaty on the Functioning of the European Union (‘TFEU’). The CJEU gave its ruling on 10 November 2011.

Following that ruling, HMRC pursued neither the Bingo appeal, nor ground 3 of the Slots appeal. Rank accepted that the ruling also decided ground 2 of the Slots appeal in HMRC’s favour and so we are not concerned with that either. All that remains for decision is ground 1 of the Slots appeal. That is whether the takings on the disputed machines were or were not at the relevant time exempt from VAT under the United Kingdom’s domestic legislation. In fact, HMRC accepted at the time that they were exempt, but they have since changed their mind about that. No-one suggests they were not entitled to.

The tribunal’s 2008 decision on the slot machine issue was not its last word on the topic. There was a further hearing before the tribunal (by then the First-tier Tribunal) in October 2009, following which the tribunal released its decision on 11 December 2009: [2009] UKFTT 363 (TC). I shall refer to this decision as ‘Slots 2’ and to the earlier tribunal decision on slot machines as ‘Slots 1’. I shall explain later what was in issue at the Slots 2 hearing.

The question of law raised by the appeal falls to be considered against a backdrop of two streams of legislation, referred to in argument as the ‘social stream’, the main element of which is the Gaming Act 1968, and the ‘fiscal stream’, the VAT legislation.

The Gaming Act 1968 was, until its repeal by the Gambling Act 2005, the principal legislation regulating the gaming industry. It was heavily amended over the years, but we were referred to it as originally enacted. It introduced different regulatory regimes in relation to ‘gaming’, defined in section 52 as ‘… the playing of a game of chance for winnings in money or money’s worth, whether any person playing the game is at risk of losing any money or money’s worth or not’.

Part 1 is concerned with regulating ‘Gaming elsewhere than on premises licensed or registered under Part II’ of the Act. Section 1(2), however, excludes the application of Part 1 to (a) ‘gaming by means of any machine to which Part III … applies’, (b) gaming to which section 41 (in Part IV) applies, and (c) gaming which constitutes the provision of amusement with prizes in the circumstances specified in sections 48 and 49 of the Betting, Gaming and Lotteries Act 1963. Section 2 shows that no gaming to which Part I applies shall take place where one or more prescribed conditions are fulfilled, the essential effect of which is that commercial gaming is prohibited.

Part II applies to ‘Gaming on premises licensed or registered under [Part II]’. It applies, by section 9, to all such gaming ‘which is not gaming by means of a machine to which [Part III] applies’. Section 10 established the Gaming Board for Great Britain, whose duty was to keep under review the extent, character and location of gaming facilities provided on premises licensed or registered under Parts II or III. Section 21 contains special provisions as to ‘gaming for prizes’ and is the provision under which gaming machines not falling with Part III were regulated. Such machines were known as ‘section 21 machines’.

Part III applies to ‘Gaming by Means of Machines’. Section 52 defines a ‘machine’ as including ‘any apparatus’. Section 26 provides, so far as material:

The machine had, therefore, to be a slot machine and the game played on it had to be a ‘game of chance’, which section 52 defined as not including ‘any athletic game or sport, but, with that exception, and subject to sub-section (6) of this section, [including] a game of chance and skill combined and a pretended game of chance and skill combined’. The crucial words for present purposes are those in section 26(2) that confine the application of subsection (1) to a machine in respect of which ‘the element of chance in the game is provided by means of the machine’. As I indicated in paragraph 1, like words were later incorporated into the relevant VAT legislation.

The main regulatory provisions in Part III relating to machines to which it applied are in sections 31 to 39. I shall refer to these provisions more fully when I come to the arguments. Such machines were known as ‘section 31/34 machines’ or, more generally, simply as ‘Part III machines’.

The domestic legislation relating to the VAT treatment of slot machines has changed over time. The original legislation was in the Finance Act 1972, which introduced VAT to the United Kingdom. Schedule 5 set out the exemptions from VAT. Group 4, headed ‘Betting, Gaming and Lotteries’, provided:

With effect from 1 November 1975, the Notes to Item 1 of Group 4 were amended by the Value Added Tax (Betting, Gaming and Lotteries) Order 1975 (‘the 1975 Order’, subsequently consolidated into the Value Added Tax (Consolidation) Order 1976, SI 1976/128). A new paragraph (d) was added, reading ‘the provision of a gaming machine’ and a new Note (4) defined that term as follows:

The VAT position remained unchanged in the Value Added Tax Act 1983, as it did in the Value Added Tax Act 1994, where it was contained in Item 1 of Group 4 to Schedule 9, although what had previously been Note (4) became Note (3), as I shall hereafter call it. A minor amendment to Note (3) effected by section 10(4)(b) of the Finance Act 2003 substituted for the words ‘coin or token’ in paragraph (b) the words ‘coin, token or other thing’, which we were told might have been directed at including the insertion of paper money but was an immaterial change for present purposes. The same position remained in force until 6 December 2005, the close of the period relevant to the appeal. Material changes to Note (3) were, with effect from that date, introduced by article 2 of the Value Added Tax (Betting, Game and Lotteries) Order 2005 (SI 2005/3328). They put beyond doubt that takings from the disputed machines were thenceforth taxable.

There is limited material now available as to the factual background against which the change by the 1975 Order was made. What at least is clear is that the effect of Note (3)(c) and of section 26(2) of the Gaming Act was respectively to exclude from the scope of VAT and from regulation under Part III games of prize bingo played on machines of the nature described in Lord Denning MR’s vivid exposition in Regina v. Herrod, ex parte Leeds City District Council [1976] 1 QB 540, at 558D to H:

HMRC assert that, for the purposes of the issue raised by the appeal, the meaning of ‘machine’, ‘gaming machine’ and ‘provided by means of the machine’ was the same for both VAT and Gaming Act purposes. Norris J said, in paragraphs 11 and 62 of his judgment, that it was common ground that the relevant words in Note (3)(c) must be given the same meaning as the like words in section 26(2) of the Gaming Act and that his judgment proceeded upon that concession, upon which he refrained from commenting. On this appeal, Rank departed from that position: it disputed that whatever meaning is to be attached to the critical words in section 26(2) of the Gaming Act must automatically also be regarded as borne by the like words of Note (3)(c).

To state the obvious, however, the 1975 VAT amendment that introduced Note (3)(c) was directed at drawing a distinction between different types of gaming machine: only certain machines were to be excluded from the scope of the VAT exemption. As Norris J noted in paragraph 10, ‘there were bound to be some machines that fell on one side of the definitional line, and some that fell on the other.’

The provision of amusements with prizes, where the amusement constituted a lottery or gaming (or both) but was not ‘gaming by means of a machine to which Part III [of the Gaming Act] applies’ was regulated by section 16 of the 1976 Act. Gaming machines regulated under that section were known as ‘section 16 machines’.

The question is, therefore, whether during the period 1 October 2002 to 5 December 2005 the takings on the disputed machines were subject to VAT, or enjoyed the exemption in Item No 1 in Group 4. The answer turns on whether the takings resulted from ‘the provision of a gaming machine’ as defined in Note (3). If they did, the takings were subject to VAT. If they did not, they were exempt. There is no dispute that the disputed machines satisfied the requirements of paragraphs (a) and (b) of Note (3). The dispute is whether they also satisfied paragraph (c), namely that ‘the element of chance in the game is provided by means of the machine.’

There has never been any dispute that during the relevant period there was a class of machine that satisfied all the conditions of Note (3), including paragraph (c). That was the type of machine in which the element of chance was provided by an electronic or mechanical component within, and forming an integral part of, the body of the machine. The takings from such machines were taxable. The equipment, to use a neutral term, in issue in this case raises a more difficult question: it is equipment in which the element of chance is provided by a component that is separate and remote from, but linked to, the terminal used by the player. Reduced to its simplest, the question is whether the terminal, or the terminal together with that component, is to be regarded as a ‘machine’ within the meaning of Note (3). The question is, however, in fact more complicated, since the disputed machines with which the appeal is concerned are systems under which a single separate component serves several terminals.

Neil Chinn, of Astra Games Limited, gave evidence in support of Rank’s case. Astra designs and manufactures gaming machines and claims to be the market leader in its field. He referred to slot machines, fruit machines and one–armed bandits as being colloquial terms for a certain type of gaming machine, but he referred to all such machines as ‘slot machines’, which traditionally are coin-operated, with three or more mechanical or video reels which spin when a button is pressed or, in the case of older machines, when a handle is pulled. The machine typically pays out according to the patterns or symbols on the machine when it stops.

Mr Chinn described slot machines as typically including some 18 key parts. His evidence related to slot machines during the period 2002 to 2005. The hardware of a slot machine consists of a cabinet containing the electronic control board, power supply coin insert and payout mechanisms, reels and/or video screens and cash boxes. The electronic control board is an embedded microprocessor control system that generates the winning and losing games and displays the results to the player via the reels, lamp displays or video screens. The machine’s software is a list of instructions that the processor executes in order to generate the winning or losing games. Such software is controlled by a ‘random number generator’ (‘RNG’), which in his diagram of a typical slot machine Mr Chinn shows as included within the cabinet. He described the function of an RNG in this way:

Mr Chinn also explained the concept of a ‘remote RNG’. This is a physically separate microprocessor unit – separate, that is, from the cabinet and software comprising the terminal before which the player stands. It is, however, linked to it by a cable and it has as its sole function the production of random numbers. The program that generates such numbers is the same as that usually found in the microprocessor that controls the slot machine. The same principles that ensure randomness in the sequence of numbers are also used in the remote RNG. The remote RNG delivers a new number on demand from the slot machine control system. This is usually done via a serial connection to the slot machine control board. The essence of Mr Chinn’s evidence is that the remote RNG works in the same way as an RNG comprised within the terminal at which the player plays.

Mr Chinn was asked what the advantage was of removing the RNG from the machine and instead using a remote RNG. His answer was that there was no commercial advantage. The perception was simply that a terminal with a physically remote RNG would not be a Part III machine but would instead be a machine to which Part II of the Gaming Act applied, in particular section 21. The concept of the remote RNG was therefore driven by a consideration of the applicable regulations. The belief was that a gaming machine in which the element of chance was determined by an RNG physically separate – or remote - from the terminal itself, and only connected to it by a cable, was not a Part III machine but was instead a so-called section 16/21 machine.

These machines are the ones that HMRC describe as ‘reconfigured’. That may perhaps be a reasonable way of describing terminals that had originally incorporated an RNG, but from which the RNG was physically removed whilst continuing to be connected to it by a wire, as appears to have happened in some cases. It may, however, be a less apt way to describe systems that were originally designed to have a remote RNG, as also happened.

The tribunal summarised the evidence relating to the machines in paragraphs 14 to 23 of its judgment. It dealt first with ‘multi-terminal RNGs’, namely an arrangement in which a single remote RNG was linked by cable to a number of terminals and it is only with systems of that type that this appeal is concerned. It then dealt with ‘single-terminal RNGs’, being systems in which a single terminal was served by its own remote RNG. The systems to which the tribunal was referring covered a spectrum, and Mr Peretz, for HMRC, also took us across it, starting with the single-terminal systems. I shall summarise the evidence.

One category of machine was the subject of evidence from Brian O’Kane, a senior officer with HMRC, with excise duties responsibilities in relation to betting, gaming and lotteries. He exhibited a picture of a terminal produced by an Astra competitor and Mr Chinn was asked about it in cross-examination. The picture showed a cabinet (or terminal) with a separate RNG on the floor below it linked by a wire to the cabinet above. Mr Chinn expressed his view that the terminal could not in practice have been operated like that and that ordinarily the RNG would hang either at the back of the machine or else on the wall, but not actually on the machine. The system appears to have been created by simply removing from the terminal the RNG which had originally been installed as an integral part of it.

The tribunal said of this category:

Another single-terminal variant was described by the tribunal as follows:

Mr Chinn, whose company had created this product, explained that it entered the market in 2004. It was ‘a two piece cabinet with a separate bottom plinth that contained the cash boxes and no mechanical/electronic workings’. The bottom plinth contained the RNG, which was connected by wire to the ‘main workings’ of the upper cabinet. The RNG had its own power supply. The two parts of the system were bolted together for safety purposes.

The tribunal did not decide whether or not systems in this category were Part III machines; and Rank advanced no case to us that they were not. The appeal is also not directly concerned with them.

These are multi-terminal systems in which several terminals are served by a single remote RNG. The systems are of different types, although their essential characteristics appear to be the same. The tribunal referred to three types of system. The appeal is concerned with the characterisation of systems of this nature.

The tribunal, in paragraph 19, described machines manufactured by Bally Gaming Systems. They were installed in Grosvenor Casinos, Stanley Casinos and Gala. The system was that of several terminals served by a single remote RNG. Simon Beacham, head of electronic training at Rank, described ‘the element of chance … [as] located outside of the machine’. He said that Rank ensured that the RNG was hung on the wall and did not touch the cabinets, and that the RNG was independently powered. The RNG was connected to the terminals by a standard Cat 5 network cable.

The tribunal, in paragraph 18, described the Megaslot machine. The evidence includes a factory photograph of four player terminals and a separate box housing the RNG, with its own power supply, above which is a display structure bearing the word ‘Megaslot’. Mr Chinn said there could have been anything from two to six terminals. The Megaslot unit would be linked to the terminals by a daisychain cable, which he explained was like a standard telephone cable but stronger. He also explained that the four terminals, display and RNG would be sold as a package for £22,000.

The last multi-terminal system was described by the tribunal in paragraph 17. This is manufactured by IGT, and it includes a configuration in which a single remote RNG, with a separate power source, serves six terminals. Mr Beacham’s evidence, to which the tribunal did not refer, was that whilst the RNG did not have to be game-specific, it had to be manufacturer-specific. Were it to break down, it is unlikely, for example, that Astra could provide a compatible replacement: IGT would have to supply it.

There was put before us, without objection, some evidence that was not before the tribunal or Norris J in Slots 1, but was before the First-tier Tribunal in Slots 2, namely a witness statement of James Thomas made on 5 May 2009. He and his son own United Leisure Gaming Limited, the parent company of Hippodrome Casino Limited. In the early 1960s, he founded Thomas Automatics Company Limited (‘TACL’), which became a major designer, manufacturer and supplier of leisure equipment, including gaming machines. At about the same time, he also founded Showboat Holdings Limited which, through various subsidiaries, owned amusement arcades and bingo clubs. Some of its interests were sold to Rank in 1987, but Mr Thomas’s family retained control of, inter alia, Thomas Holdings Limited, the parent of a group that over the following 20 years became leading operators of amusement arcades and bingo clubs. In 2006, the group was sold to Mayfair Acquisitions Co Limited and they became operated by Riva Gaming Limited. Mr Thomas has been involved with industry bodies representing the interests of the betting and gaming industry, including in the early 1980s a spell as the Chairman of the British Amusement and Catering Trade Association, and he has continued to act on its top committee as a trustee and life member.

Mr Thomas explained the development of bingo machines. In the 1960s, between the respective enactments of the Betting, Gaming and Lotteries Act 1963 and the Gaming Act 1968, bingo was not played for cash (the limits on cash prizes were too great) but for prizes, and was generally known as ‘prize bingo’. TACL developed types of prize bingo machines. The first such machine had up to eight playing positions. The player would give his money to a collector (the machines were not yet ‘slot’ machines). The random numbers were provided by a separate‘bingo blower’, a mechanical device that continually mixed the bingo balls and had a chute that randomly pulled a ball out for the caller. The bingo blower was not connected to the terminal in any way and was simply the mechanical equivalent of the gaily dressed lady referred to by Lord Denning in Herrod’s case (see paragraph 19 above). It follows that such a prize bingo machine would not have been a Part III machine under the Gaming Act. The next development was to introduce a slot to the machines, so that there was no need for a separate collector. This did not, however, result in the element of chance being provided by the machine.

Mr Thomas explained the development of gaming machines in the late 1960s and 1970s. In 1968, he (I presume he meant one of his companies) bought from Isca Electronics an electronic number generator that it had developed – what is now called an RNG. Mr Thomas regarded this as more advanced than his bingo blower. He developed from this his Bingo Electronic Number Generating Equipment (‘BENGE’), of which the main component was a computer chip. BENGE was about one foot square and was first shown at the Amusement Trade Exhibition at Alexandra Palace in 1969. Mr Thomas started using BENGE in the early 1970s by way of a replacement for the bingo blower. Mr Thomas did not, however, say that BENGE was linked to the terminal in any way. On the face of his evidence, it was simply a more sophisticated bingo blower, the use of which would not have brought the bingo machines within Part III of the Gaming Act.

Mr Thomas devoted several paragraphs to the development of machines that paid out prizes of Green Shield stamps. The machines were officially launched in January 1976. He described them as working with an external RNG that fed results to the machine which displayed them. The RNG was in a stand-alone cabinet, which was much larger than the RNG and was of such a size that it could not be placed inside a Green Shield stamp machine. He said that this was intentional, it being his way of ensuring that the machines met the requirements of section 16. The RNG had its own separate power source and was connected to the machine by way of a standard power cord. The Green Shield machines had both single and multi-player versions, with a single RNG in the latter case serving several terminals. The takings on these machines were not taxed because HMRC regarded them as exempt. HMRC’s revised position is that they were wrong in this respect and the machines should have been taxed as Part III machines.

Finally, Mr Thomas referred to the development of gaming machines in the 1980s, and he did so by reference to newspaper articles and other exhibits to his statement. At least by 1983, the bingo blower had been replaced by a separate electronic RNG linked to the terminal and generating the numbers on its screen.

It is worth summarising HMRC’s changing views about the taxability of the disputed machines. There are two classes of machine which were always regarded as exempt from VAT, being classes of machine in which the relevant element of chance was not provided by the machine, but by an external agent. These include (i) prize bingo machines, being slot machines for which the element of chance was provided by a mechanical or electronic RNG that was external, and unconnected, to the slot machine but was located in the same premises as the slot machine; and (ii) Fixed Odds Betting Terminals (‘FOBTs’), being slot machines in respect of which the element of chance was provided by a separate electronic RNG connected to the terminal but situated, for regulatory reasons, on premises other than those on which the playing terminal was situated.

The disputed machines were formerly known as ‘section 16/21’ machines, a reference to the provisions of the gambling legislation under which they were (or were believed to be) regulated, namely section 16 of the 1976 Act and section 21 of the Gaming Act (both sections applying to gaming other than on machines to which Part III of the Gaming Act applied). As with prize bingo machines and FOBTs, a game of chance was played on a terminal into which a coin was inserted, and the ‘element of chance’ was provided by an electronic RNG that was separately located from, but connected to, the terminal. As with FOBTs, the RNG for the disputed machines served a number of terminals and communicated randomly generated numbers to each terminal, each of which operated independently from the other terminals. Unlike FOBTs, however, the RNG would be on the same premises as the terminals it served.

HMRC also regarded the disputed machines as section 16/21 machines, and so VAT exempt, just as the then Gaming Board for Great Britain regarded them as exempt from the regulatory provisions concerning gambling. HMRC’s view to this effect was reflected in guidance it issued in January 2005, which reads:

HMRC’s then view was, therefore, that the disputed machines were section 16/21 machines and so exempt from VAT. They disclaimed that they were Part III machines. Their revised position is that that view was wrong. They maintain their former position with regard to the exempt status of FOBTs, but now on the ground (inter alia) that they are not used for games of chance at all, but for betting. As for the disputed machines, they adopted their new position in defence of Rank’s claims for the recovery of VAT paid in respect of Part III taxable machines. Rank’s repayment case was advanced in reliance on the assertion that the levying of such tax infringed the principle of fiscal neutrality since the provision of similar supplies by section 16/21 machines was exempt from VAT. The basis of the case was that the Group 4 exemption for ‘Betting, Gaming and Lotteries’ derived from Article 13B(f) of the Sixth Directive, and in Finanzamt Gladbeck v. Linneweber (Case C-450/2) [2005] ECR 1-1131, [2008] STC 1069 the CJEU held that the limitations on the exemption under Article 13B(f) must comply with the principle of fiscal neutrality in that similar supplies must not be subject to different VAT treatment.

To meet that case, HMRC asserted that the disputed machines were not exempt at all and that to describe them as ‘16/21 machines’ is wrong. Their conversion to what they now assert to be the true view is so clearly held that they do not hesitate to describe as ‘absurd’ the unenlightened view they formerly espoused. The Gaming Board had adopted that same interpretation and so have the tribunal and Norris J. Everyone to date involved in determining the true characterisation of the disputed machines has therefore, at least at one stage, been of the like mind that their takings were exempt from VAT. The question for this court is whether HMRC are now the only soldiers marching in step.

Having introduced the subject of FOBTs, I should now identify what Slots 2 was about. It was concerned in part with the question, left unresolved by Slots 1, as to the extent to which Rank’s disputed machines were actually in use over the period in issue, namely from 2002 to 2005. The tribunal (by then the First-tier Tribunal) found that they were in operation during the whole of the relevant period and that is no longer in dispute.

Rank also ran a new point at that hearing in support of its case that the taxing of its disputed machines infringed the principle of fiscal neutrality. This was that the FOBTs were also exempt from VAT (as was accepted) and were similar to Rank’s taxed Part III machines (which was disputed). The tribunal found that the machines were similar. That holding was the subject of an appeal by HMRC to the Upper Tribunal, which made references in relation to that and other issues to the CJEU on 13 May 2010.

The CJEU’s ruling on the reference earlier made by this court dealt also with that reference made by the Upper Tribunal, and it gave guidance as to the approach by the national court to questions of similarity such as had been raised. The Slots 2 appeal then returned to the Upper Tribunal, when Norris J accepted that the First-tier tribunal had made two errors of law in its assessment of similarity as between the FOBTs and the Part III machines and held that that question had to be remitted to the tribunal. The remitted proceedings have been stayed pending the determination of this appeal. The Upper Tribunal’s decision in relation to this is reported at [2012] UKUT 347 (TCC); [2013] STC 420.

We are concerned only with the tribunal’s decision as to the multi-terminal systems. The tribunal’s conclusion as to these was as follows (I have set out Note (3), to which the tribunal refers, at paragraph 17 above, although I there gave it its earlier numbering of (4)):

Norris J provided careful reasoning for his ultimate conclusion that the tribunal had adopted the correct approach in relation to the characterisation of multi-terminal systems to each of which a single, remote RNG was linked. He discussed and considered the rival arguments in paragraphs 57 to 66, expressing his conclusion as follows:

Mr Peretz provided the court with a written summary of his submission in support of HMRC’s appeal, which reads as follows:

There is perhaps a logical difficulty in that summary of HMRC’s case, since it interprets the meaning of ‘gaming machine’ in Note (3) by reference to the sense of the word ‘apparatus’, a word that does not feature in the VAT legislation, although it does feature in the definition of a ‘machine’ in section 52 of the Gaming Act. Mr Peretz’s oral development of his argument made it plain, however, that the essence of his submission was that the disputed machines were ‘machines’ within section 26 of the Gaming Act and were, therefore, Part III machines; and that since the crucial language in Note (3)(c) was apparently drawn from that of section 26(2), it follows that if the disputed machines were Part III machines, so must they be ‘gaming machines’ within Note (3).

As to the first part of that submission, Mr Peretz emphasised that Part III of the Gaming Act 1968 created a specific regulatory regime for the machines to which it applied, and that the other Parts of the Gaming Act made it expressly clear that they did not apply to gaming by means of any such machines. Save that Part II (which includes section 21) included the like exclusion in relation to gaming by means of any machine to which Part III applies, it made no reference to machines. Section 21 was not ‘machine specific’ but simply imposed limits on stakes and prizes in Part II gaming generally.

Part III, however, was ‘machine specific’. Section 26 described the type of gaming machine to which it applied. Section 27 provided that, subject to exceptions, only those certified by the Gaming Board may sell and supply such machines. Section 28 empowered the Secretary of State to impose restrictions as to the terms and conditions of their sale, supply and maintenance. Subject to any direction given under section 32, section 31(2) limited to two the number of machines that might be made available for gaming on any premises licensed under the Act, a limit that was increased over time. Section 31(3) imposed a limit on the charge for playing a single game, subject, as happened, to changes as to the amount being made by the Secretary of State. Section 33 regulated the use of machines at non-commercial entertainments. Section 35 imposed a general restriction on the use of machines otherwise than under sections 31 to 34. Section 36 regulated who could take money out of the machines. Section 37 conferred a general regulation making power on the Secretary of State in relation to the sale, supply, maintenance and use of machines. Mr Peretz referred us to The Law of Betting, Gaming and Lotteries, Smith & Monkcom, 2nd Edition (the relevant one for present purposes), paragraph B10.7 of which summarised the primary objective of the Gaming Act 1968 as being to stamp out the abuses which had sometimes occurred in relation to the supply of gaming machines to site-owners, and continued:

Against those background considerations, Mr Peretz submitted that Parliament could not rationally have intended in Part III to draw a distinction between, on the one hand, machines where the playing equipment and the RNG are in the same box and, on the other hand, arrangements such as those I have summarised in the three categories referred to above, ranging from (a) those in which the RNG is simply removed from its original box and hung separately from it, to (b) those purpose-designed multi-terminal systems where a separate RNG serves all the terminals. Having regard to the plain objectives of Part III, there is no reason to attribute to Parliament an intention to treat systems falling on the latter side of the line drawn by Mr Peretz differently from those falling on the former side.

Mr Peretz developed this by submitting that there was no evidence that, at the time of the passing of the Gaming Act, there were any machines or systems in use with separate RNGs: Mr Thomas’s evidence about BENGE was to the effect that it did not emerge until 1969. There was at the time of the passing of the Act evidence of the workings of prize bingo machines, involving the use of a human caller, and Parliament plainly intended to exclude systems such as that from the controls of Part III. That, Mr Peretz suggested, was because the required presence of a human agent rendered that particular type of gaming more transparent. Part III, however, was directed at regulating an industry in which the machines enabled gaming to take place without any such human participation. Part III was, it is to be inferred, directed at the control of the use of what Mr Peretz called ‘skewed’ machines. That is why Part III regulated the manufacturers and suppliers of the machines, and the numbers of machines that could be used in particular premises. The perceived need for regulation in respect of such machines applied as much to machines with integral RNGs as to those with remote RNGs. There is therefore no good reason not to interpret section 26(2) as also extending to systems with separate or remote RNGs.

Mr Peretz cited in support of such an interpretation the decision of the House of Lords in Regina (Quintavalle) v. Secretary of State for Health [2003] UKHL 13; [2003] 2 AC 687. Lord Bingham of Cornhill there referred to what he called the authoritative statement of Lord Wilberforce in his dissenting opinion in Royal College of Nursing of the United Kingdom v. Department of Health and Social Security [1981] AC 800, at 822, which I shall set out:

Mr Peretz submitted that the application of this interpretational approach justified appropriate liberality in the approach to section 26 of the Gaming Act so as to include the disputed machines. He recognised that Lord Steyn, at paragraph 21 of Quintavalle, observed that whilst the shift towards purposive interpretation was not in doubt, the degree of liberality permitted was ‘influenced by the context, eg social welfare legislation and tax statutes may have to be approached somewhat differently’. In this case, what is ultimately in question is the meaning of Note (3)(c) in the VAT legislation, which is tax legislation. Mr Peretz’s submission was that since this was plainly modelled on the language of Part III of the Gaming Act, the critical question was what section 26(2) in Part III meant: and if ‘machine’ in that context included the disputed machines, so does the language in Note (3).

Mr Peretz relied also on Victor Chandler International Ltd v. Customs and Excise Commissioners and Another [2000] 1 WLR 1296 as exemplifying an ‘always speaking’ interpretational approach to legislation even in the case of a penal statute, although Sir Richard Scott V-C, at paragraph 32, said that:

Mr Peretz said that the mischief at which Part III was directed applied as much to machines or systems of the nature in dispute as to those which indisputably fell within section 26. The machines were, he said, identical in their practical effect, it was only their configurations that varied. The court could, therefore, be clear that the disputed machines were within the mischief at which Part III was aimed.

If, therefore, the disputed machines could be said to fall within the purposive grasp of Part III, was there anything in the language of Part III legislation that nevertheless ruled out its application to the disputed machines? Mr Peretz accepted that if such language simply did not permit its application to the disputed machines, they were not within it. But he said that there was nothing in the language to exclude them. The key lay in the definition of machine set out in section 52 of the Gaming Act 1968, although all that it says is that a ‘machine’ includes ‘any apparatus’. He said that the inclusion of the concept of ‘apparatus’ in relation to the definition of a ‘machine’ assists in identifying what the boundary of the machine is. Whereas but for the inclusion of ‘apparatus’, one might regard the boundary of a ‘machine’ as the outer limits of the box encasing it, the inclusion of a reference to its ‘apparatus’ enables a wider interpretation. The Shorter Oxford English Dictionary gives, as its primary definition of ‘apparatus’, ‘the things collectively necessary for the performance of some activity or function’, and that enables the component parts of the systems in dispute, including the remote RNGs, to be regarded as a ‘machine’ to which Part III applied.

As I have indicated, if the reference to ‘apparatus’ in section 52 is the key to the interpretation of the word ‘machine’ in section 26(2), Mr Peretz’s submission that ‘gaming machine’ in Note (3) is to be read in the like way runs into the difficulty that Note (3) includes no like definitional reference to ‘apparatus’. Mr Peretz’s submission was nevertheless that ‘gaming machine’ in Note (3) must be interpreted in the same way as ‘machine’ in section 26(2).

Dealing specifically with the types of system directly in issue on this appeal, that is multi-terminal systems served by a single remote RNG, Mr Peretz submitted that one way of analysing the arrangement is that if, for example, there are seven terminals, each terminal is a ‘machine’ of which the common RNG forms part. He described such a set-up as involving seven ‘overlapping’ machines. Alternatively, all seven terminals and the single RNG could be regarded as a single machine. Mr Peretz recognised that the multi-terminal system arrangements raised questions as to what the ‘machine’ is for the purpose of the limits as to the numbers of machines on premises for the purposes of section 31(2) of the Gaming Act 1968, which originally limited the number to two. He also acknowledged that there were difficulties in the application of that subsection to such systems. I understood his primary suggested solution to that problem was that the Secretary of State could meet it by way of the exercise of his wide and general regulatory powers under section 37.

Mr Lasok QC, in his responsive arguments for Rank, made six opening points. First, Note (3) of the VAT legislation drew a line directed at taxing a defined group of slot machines and exempting other gaming machines. Second, the taxed group was identified by reference to those machines that satisfied three cumulative criteria, which can be taken to have been the fruit of careful deliberation. Third, the taxed machines indisputably included machines sharing a feature that does not apply to the disputed machines, namely that the element of chance was by a mechanism internal to the machine. Fourth, the exempt machines include prize bingo machines, which were already in existence by 1975, and FOBTs, which were in existence from 1998, whose common feature was that the element of chance was not provided by the machine but by a separate and distinct device. In the former case, Mr Thomas explained that it was originally a bingo blower but later an electronic RNG. As for FOBTs, whilst Mr Peretz sought to say that they were exempt because they provided betting games, HMRC’s previous view was that they were exempt because they had an external RNG, as they had said in the 2005 guidance. Fifth, the disputed machines are the multi-terminal machines served by a separate, remote RNG. Whether rightly or wrongly, they have in the past been regarded by reason of that feature of their configuration to fall outside Part III of the Gaming Act 1968, and within Part II, and they have in practice been regulated as ‘section 16/21’ machines. Sixth, such machines are like the prize bingo machines and the FOBTs, in that the element of chance is provided by a separate, remote RNG. There was, however, this difference, namely that in the case of FOBTs, which were in licensed betting offices, the RNG had, for regulatory reasons, to be outside the operator’s premises; whereas in the case of section 16/21 machines, the perception was that both the machine and the RNG had to be on the same licensed premises.

Mr Lasok moved to the question of whether the disputed machines are ‘gaming machines’ as defined in Note (3). The Gaming Act does not use the same phrase but defines in section 26 the type of ‘machine’ to which Part III applies. Whether a particular machine does or does not fall within either definition turns on whether or not it satisfies its respective statutory conditions. The draftsman of Note (3) did not adopt the definition of ‘machine’ in the Gaming Act: in particular, he did not provide, as did section 52 of that Act, that a ‘machine’ includes ‘any apparatus’. The ordinary meaning of Note (3) is that the ‘machine’ providing the element of chance must be the ‘machine’ on which the player is playing the game of chance. The critical question for the tribunal was whether the disputed machines, being those in a multi-terminal configuration, were ‘machines’ within the meaning of Note (3), and the tribunal found as a fact that they were not. This court, like Norris J, should respect that finding by the tribunal, which was a specialist tribunal.

Whilst Mr Lasok was, I understood, disposed to accept that configurations at what he called the ‘dubious end’ of the spectrum (see paragraphs 31 to 35 above) might arguably be regarded as ‘gaming machines’ within the meaning of Note (3), he said there is no justification for also so regarding any of the components of the purpose-built multi-terminal systems, in which the separate electronic RNG was one machine, and the terminals were separate machines. Such a system, under which two machines are connected by a detachable cable, or perhaps by a satellite link, cannot fairly be regarded as constituting one machine. A box comprising a CD player, amplifier and speakers will be one machine. A system of separates under which the CD player is linked to a separate amplifier and separate speakers will not: no one would regard the CD player and the amplifier as constituting one machine.

Whether in any particular case two entities linked by a detachable cable are fairly to be regarded as one or two machines is, however, a matter of fact and degree. Thus, when the tribunal regarded the RNG that was permanently connected to a single terminal as, together with the terminal, constituting one machine, it may well have been entitled to do so. The multi-terminal systems now in issue present, however, a quite different question. In such systems a single, separate RNG serves a number of terminals, say six. It is impossible to regard any such configuration as constituting one machine. Is the putative single machine the RNG and all six terminals? Or is it the RNG and any one of the six terminals? If the latter, does the single RNG form part of six separate machines? The difficulty in giving these questions affirmative answers explains why the HMRC’s original view as to the exempt status of the disputed machines was correct.

Mr Lasok also submitted that even if the language of Part III of the Gaming Act, construed against the legislative purpose of that Act, permitted the multi-terminal systems to be regarded as constituting one machine, that provides no assistance when it comes to the interpretation of the ordinary language used in Note (3), which did not simply incorporate the definition of a ‘machine’ in the Gaming Act. Part III of the Gaming Act defined a ‘machine’ by reference to criteria that were narrower than those of Note (3) It required the payment to play the machine to be by inserting a coin or token in a slot or aperture, whereas Note (3) did not so limit the method of payment, yet Mr Thomas’s evidence had shown that prior to the Gaming Act machines existed which did not have slots or apertures. Part III was only directed at regulating a ‘machine’ meeting the criteria of its own definition of a machine. Machines which fell outside its definition were regulated by other provisions of the Act. Part III imposes regulatory provisions, the breach of which constitutes a criminal offence. The provisions should therefore be interpreted strictly and literally, and their definition of a ‘machine’ cannot be transposed to the VAT legislation.

There can be no doubt that the definition of a ‘gaming machine’ in Note (3) of the VAT legislation was in material respects drawn from the definition of a ‘machine’ in section 26 of the Gaming Act. It is inconceivable that the draftsman of Note (3) did not have section 26 in front of him. Equally, however, he did not follow it slavishly. First, he did not confine a ‘gaming machine’ to a slot machine, although I do not regard this departure as of present significant materiality. Second, perhaps of more significance, he did not incorporate the definition of a ‘machine’ in section 52 of the Gaming Act. If, therefore, section 52’s amplification of the section 26 definition so as to include ‘any apparatus’ is critical to a conclusion that the disputed machines fell within Part III, whereas otherwise they would not, there may be a logical difficulty in regarding such machines as ‘gaming machines’ within Note (3).

Ultimately, this appeal is concerned only with the correct interpretation of the words ‘gaming machine’ in Note (3). Given, however, that it is apparent that the meaning of those words was in material respects drawn from section 26(2) of the Gaming Act, I regard it logical as to consider first whether the disputed machines were or were not Part III machines; and much of the argument before us was directed to that question.

In that regard, whilst the appeal is concerned only with the characterisation of those multi-terminal systems which incorporate a single, separate RNG serving all the terminals, I consider it helpful to start by considering the status of systems at the other end of the spectrum. Taking first the configuration referred to in paragraphs 31 and 32 above (in which the RNG is removed from the terminal and hung on the wall), it appears to me clear, as the tribunal held, that the RNG and remainder of the terminal would together constitute a ‘machine’ within both section 26 and Note (3). It makes no sense that it can be possible to segregate the component parts of a single terminal into two separate, but connected, units which together continue to perform the identical function as they did before and then to be able to claim (i) that the ‘machine’ played by the player no longer provides the ‘element of chance’ since that is now provided by the separate RNG, and that (ii) therefore neither of the two units is a Part III machine. In my view, it is clear that such a configuration is as much a Part III ‘machine’ as was the single terminal previously. That is not because the new configuration ought to be notionally regarded as continuing to represent its former undivided construction. It is simply because it is obvious that a configuration of separate, but connected, items of equipment that together enable the playing of a game of chance at the terminal, being a game in which the element of chance is provided by means of such equipment, is a ‘machine’ within the meaning section 26(2).

I recognise that that involves attaching an element of plurality to the apparently singular concept of the ‘machine’ referred to in section 26, an interpretation to which the language may not be regarded as obviously susceptible. For myself, however, I have no difficulty in so interpreting the word ‘machine’. I readily accept that, for example, a stereo system of separates comprising a CD player, amplifier and speakers would ordinarily be regarded as comprising more than one machine. Analogies such as that are, however, of little help in interpreting the sense of the word ‘machine’ used in a statute that is plainly directed at providing protection to the public from perceived abuse, or the risk of abuse. The control that Part III sought to impose was on the availability for use by the public of mechanical equipment providing the opportunity for playing of games of chance when the element of chance was provided by the equipment so used. It cannot have been the purpose of Part III to confine its control to equipment comprised in a self-contained single unit or terminal and to exclude from such control two separate, but linked, items of equipment that together perform an identical function. If that were the purpose, regulation under Part III would become entirely optional. Casinos which chose to remain within such control could leave the RNGs in the terminal; and those which preferred to be controlled instead under Part II of the Gaming Act could remove them and hang them on the wall.

In my view, to interpret Part III in such a narrow, literal way would be absurd. It would virtually reduce it to a dead letter. That cannot be the correct construction of the word ‘machine’. The word must, if the language of Part III is to be given a sensible and practical effect that will enable it to achieve its obvious purpose, be interpreted as including equipment ancillary, and connected, to the playing terminal that automatically provides the element of chance that determines the outcome of the game played on the terminal. I would arrive at that interpretation without recourse to the definition of a ‘machine’ in section 52 as including ‘any apparatus’, a definition I anyway find less than illuminating.

I add that the fact that Part III is regulatory legislation, a breach of which can amount to a criminal offence, does not in my view compel the attaching of a narrow and unrealistic interpretation to the concept of a ‘machine’. Anyone seeking to evade the shackles of Part III by segregating the RNG from the playing terminal like this will be doing so with his eyes wide open; and, unless he has obtained a prior clearance, he will take the risk that he has misread the legislation.

If this is right, it follows in my view, and for like reasons, that a purpose built system comprising a terminal with a separate, but connected, RNG is also properly characterised as a ‘machine’. The terminal cannot be used for gaming purposes except by being linked to the RNG; and the RNG is designed to be linked to the terminal in order to enable the game to be played. Again, no doubt they constitute two separate items of equipment; but to treat the terminal as a separate ‘machine’ in considering the impact or otherwise of Part III is unrealistic. They are being used together for the purpose of playing a game on the terminal and the RNG forms an essential element of the system.

If right so far, I also do not understand why the multi-terminal systems should be treated any differently. The fact that there is only one RNG serving several terminals cannot make a material difference. In substance, the systems are exactly the same as in both previous configurations. By like reasoning, I cannot see why each terminal and the single RNG do not together constitute a machine within section 26. That is the substance of any such multi-terminal system; and it is the substance of the matter that counts.

The question, however, is not whether multi-terminal systems are Part III machines, but whether they are Note (3) ‘gaming machines’. It is true that Note (3) does not seek to serve the same, or any like, regulatory purpose as Part III. Mr Lasok was, in his submissions, apparently anxious that we should not forget that this appeal turns on the language of Note (3), not on that of Part III, from which I detected a distinct, albeit unspoken, concern that the court might well be unsympathetic to an interpretation of Part III that reduced it to the dead letter that is the logical consequence of Rank’s case.

Mr Lasok was, however, of course entirely right that we are concerned only with whether the multi-terminal systems constitute ‘gaming machines’ within the meaning of Note (3). In my judgment, however, they do. As I have said, there can be no doubt that the Note (3) definition of a ‘gaming machine’ was in all its material respects drawn directly from the definition of a ‘machine’ in section 26 of the Gaming Act. I can therefore see no reason for not interpreting a ‘gaming machine’ in Note (3) in the like way. On the contrary, I can see good reason for doing so. Once again, if it is instead to be construed in the narrow, literal way urged by Mr Lasok, it would reduce VAT on ‘gaming machines’ to a voluntary tax, in the sense that the tax could be avoided by a simple re-design of the playing equipment, whilst leaving its essential function unchanged. The tax, however, was plainly not intended to be a tax on takings from gaming machines of a particular, singular configuration. It was intended to be a tax on takings from equipment meeting the Note (3) criteria. The multi-terminal systems were such equipment; and each terminal and linked RNG were, in my view, ‘gaming machines’ within the meaning of Note (3).

I therefore respectfully disagree with the different conclusions arrived at by Norris J and the tribunal. In my view, on the facts found by the tribunal, the correct legal characterisation of the multi-terminal systems is that they were ‘gaming machines’. I would allow HMRC’s appeal.

I agree.

I also agree.