Hui Ling McCarthy QC
Julian Ghosh QC
(Instructed by HMRC Solicitors Office (Bush House))
(Instructed by Pinsent Masons LLP (London))
LORD BRIGGS AND LORD SALES: (with whom Lord Reed, Lord Leggatt and Lord Burrows agree)
1. The complicated facts of this appeal raise some issues of general importance about the power of Her Majesty’s Revenue and Customs (in this context “the Revenue”) to make what is generally known as a discovery assessment. Broadly speaking, if the Revenue are not content with the accuracy of the self-assessment to income tax (and capital gains tax) in a taxpayer’s tax return, the Revenue may open an enquiry into the return and amend the assessment. The enquiry must be opened within one year of the filing of a timely return (and slightly longer if the return is late), although it need not be completed within that period. But if the Revenue find out (“discover”) that an assessment is or has become insufficient, then subject to satisfying one or the other of two conditions, they may make a fresh assessment (a “discovery assessment”) in the amount which they think will be sufficient to make good to the Crown the loss of tax. The first condition (which is the one in issue on this appeal) is that the insufficiency of the earlier assessment has been brought about carelessly or deliberately by the taxpayer (or their agent). The standard time limit for the making of an assessment by the Revenue, including a discovery assessment, is four years from the end of the year of assessment to which it relates. But if the insufficiency is brought about by carelessness the time limit is extended to six years, and if it is brought about deliberately by the taxpayer it is extended to 20 years.
2. In the present case the taxpayer Mr Raymond Tooth filed a return in 2009 which contained his self-assessment of income tax for the 2007-8 year of assessment in an amount which, from the outset, the Revenue officers dealing with his case regarded as understating his true tax liability by almost half a million pounds, by his use of a tax avoidance scheme which they regarded as ineffective for that purpose (“the Romangate scheme”). Whether the Romangate scheme was ineffective or not was a matter of controversy between the Revenue, Mr Tooth and other taxpayers who had used the same scheme, which was only finally resolved in the Revenue’s favour (with the assistance of retrospective legislation) at the end of 2013. The Revenue did not protect their position vis-à-vis Mr Tooth in the meantime by opening an enquiry into his return. The legal position at that time was obscure and they wrongly thought that they had protected their position by opening a form of enquiry using a different statutory power, but found out from a ruling by this court in 2013 that this had been ineffective. So the Revenue then issued a discovery assessment, alleging that the insufficiency in Mr Tooth’s self-assessment had been brought about deliberately, relying on the 20-year period within which to make the assessment.
3. It has never been the Revenue’s case that, simply by self-assessing in an amount much less than his true income tax liability, Mr Tooth thereby brought about its insufficiency deliberately, although of course he (or his agents) self-assessed in the lower amount intentionally. The Romangate tax avoidance scheme upon which he relied had been vouched as effective (albeit in the end wrongly) by leading tax counsel, and it has never been suggested that Mr Tooth did not genuinely believe that the amount at which he self-assessed was correct (or at least one which it was legitimate for him to put forward). Rather, the Revenue’s case is that the online tax return completed on Mr Tooth’s behalf contained a deliberate inaccuracy, because a loss which (under the scheme) was designed to be an employment-related loss incurred in the following year of assessment and then carried back, was wrongly inserted into a box on the electronic form reserved for partnership losses, and thereby found its way by deduction into the electronic calculation of the self-assessment tax liability, so as to have caused the insufficiency.
4. Mr Tooth appealed the discovery assessment to the First-tier Tribunal (“FtT”),  UKFTT 723 (TC). He put the Revenue to proof that there had been the requisite discovery. In any event he denied that his return contained an inaccuracy, let alone one which was deliberate. The Revenue pleaded that the discovery had been made by its officer Mr Nigel Williams in October 2014. The FtT accepted the Revenue’s case on discovery, but agreed with Mr Tooth that there had been no deliberate inaccuracy in his return. His appeal was therefore allowed.
5. On the Revenue’s appeal the Upper Tribunal (“UT”) found that there had been no discovery in 2014, mainly because the Revenue had formed its own view about the insufficiency in Mr Tooth’s return since 2009, and that a discovery in 2009, even if it had been pleaded, would have become “stale” by 2014. The UT also held that there had been no inaccuracy in Mr Tooth’s return, read as a whole, and that even if that were wrong, the inaccuracy had not been deliberate.
6. The Court of Appeal agreed with the UT on the absence of a qualifying discovery but found, by a majority, that there had been a deliberate inaccuracy in Mr Tooth’s return. Therefore, by reason of the conclusion on the first point, the Revenue’s appeal was dismissed. The Revenue appeals to this court, seeking to restore the FtT’s decision that there was a discovery and to uphold the judgment of the majority in the Court of Appeal that there was a deliberate inaccuracy in Mr Tooth’s return, while Mr Tooth seeks to restore the conclusion arrived at by the FtT and the UT that there was no deliberate inaccuracy in his return.
7. Mr Tooth’s activities during the year of assessment 2007-8 generated income on which, leaving aside his later participation in the Romangate tax scheme, he in fact incurred an income tax liability in excess of £475,000.
8. On 23 January 2009 Mr Tooth signed the documents requisite for his participation in the Romangate scheme, which was promoted by NT Advisors (“NT”) and sanctioned by leading tax counsel. The scheme was designed to generate (for him) an employment-related loss incurred in the tax year 2008-9 which (so he was advised) he could utilise by setting it against income both for that year and (by way of carry-back) for the previous 2007-8 tax year. On advice Mr Tooth initially sought to utilise £1,185,987 of that scheme-generated loss against his 2007-8 income.
9. Mr Tooth’s tax advisors Grunberg & Co (“Grunberg”) prepared his 2007-8 tax return in January 2009. By then the time had passed to do so on paper and they used IRIS software which is approved by the Revenue. NT had advised that the utilisation of losses derived from the Romangate scheme should be achieved by entering the claimed amount of loss in box 3 on Additional Information (“AI”) page 3, and entering “2007-09” in box 4. But Grunberg found that it was impossible to access box 3 on AI page 3. On enquiry of the IRIS software engineers they were told that this was due to a technical software issue, and that they should enter the loss in some other box, and then explain what they had done in the “white space” included in the form to allow for written explanations to be provided by a taxpayer.
10. Following this advice, and with NT’s confirmation that the method chosen was appropriate, Grunberg entered the loss claimed from their client’s participation in the scheme in the following way. First, on page 6 of the main return (“TR6”) at box 19 they wrote:
“During the tax year ending 5 April 2009, I sustained an employment related loss for which relief is being claimed now in accordance with section 128 ITA 2007. Please refer to the partnership pages of my return. Full details of this loss will be reported on my 2008/09 tax return in due course.”
11. Then they entered the loss claimed from participation in the scheme (£1,185,987) in boxes 7, 19 and 20 on the Partnership (short) (“SP”) supplementary pages of the return so as to claim the loss as a deduction against income in 2007-8, and set out the following explanation in the Additional Information box 30 on SP2:
“During the year ending 5 April 2009, I sustained an employment related loss for which relief is being claimed now in accordance with the provisions of section 128 ITA 2007 (via section 11 ITEPA 2003). I have reported the details of the loss claimed against my other income using box 3 [sic] above, which relates to a claim for a partnership Loss from this tax year set-off against other income for 2007-08. However, there is no equivalent box to claim relief now for employment related losses despite the provisions of section 128 ITA 2007. Full details of this loss will be reported on my 2008-09 tax return in due course. The loss arose pursuant [sic] arrangements for which a scheme reference is required under DOTAS (from AAG at HMRC) at this time the scheme has not been granted a reference number. When such number is obtained I will report it on my 2008-09 tax return, as that is the year in which the loss arose. I acknowledge that my interpretation of the tax law applicable to the above transactions and the loss (and the manner in which I have reported them) may be at variance with that of HM Revenue and Customs. Further please note that although I have reported (and hereby claim the loss pursuant to section 128 ITA 2007) in box 3 [sic] above I wish to make it clear that the deduction I am claiming on my return is not what you regard as a Loss for this tax year set-off against other income for 2007-08 - for all these reasons I assume you will open an enquiry.”
12. Use of the partnership pages of the online form required first the entry of a Unique Tax Reference number for the supposed (but of course non-existent) partnership. So Grunberg just made one up, using the obviously artificial number 99999 99999 in box 1 on SP1.
13. The IRIS online return form contained a Tax Calculation Summary (“TCS”) section setting out the amount of the self-assessment. Its figures are computer-generated from the information inputted on the earlier pages of the return. Taking into account the carry-back loss claimed in the partnership pages, the TCS generated a net claim by Mr Tooth to repayment of income tax for 2007-8 of £13,212. Although neither Mr Tooth or Grunberg appreciated this, if the software had permitted them to have inserted the employment-related loss in box 3 on AI page 3 (as originally planned) this would have generated a free-standing tax credit in the full amount of the loss, rather than a reduction of the amount in box 2 on TCS page 1 from tax due of £475,000 odd to a repayment claim of £13,212. The use of the partnership pages rather than the unavailable AI page 3 therefore obtained for Mr Tooth an unintended timing advantage, but no different amount in the overall position on his account with the Revenue.
14. In August 2009 the Revenue wrote to Mr Tooth to give notice that they were opening an enquiry under Schedule 1A to the Taxes Management Act 1970 (“the TMA”) (“the Schedule 1A enquiry”) into what they described in the letter as his claim for relief in 2007-8 for employment losses incurred during 2008-9. By this the Revenue intended to protect their position as against Mr Tooth’s use of the Romangate scheme, having by then promoted a provision in the 2009 Finance Bill to nullify the effect of the scheme retrospectively. By this time the Revenue had already formed the view that, even under the existing law, the Romangate scheme was ineffective, albeit they recognised there was doubt about this and in due course promoted retrospective legislation to ensure this result. In the letter the Revenue referred to an announcement made on 1 April 2009 that such legislation was to be introduced which would have the effect of refusing relief for losses under the Romangate scheme. By their letter they made it clear that, from their reading of Mr Tooth’s 2007-8 return, they well understood that his Romangate-based losses had nothing whatever to do with a partnership, but rather related to employment, and involved a carry-back from 2008-9 to 2007-8.
15. In December 2009 Mr Tooth amended his 2007-8 return to increase the amount of losses claimed to £1,210,229. At no time did the Revenue open an enquiry into Mr Tooth’s 2007-8 return under section 9A of the TMA, although they purported to amend the return by removing the claimed loss of £1,210,299 (sic) by letter in April 2010. At this stage, due to obscurity in the drafting of the TMA, it was not appreciated that a Schedule 1A enquiry was not available as a means to challenge a claim to tax relief set out in the calculation in a self-assessment return and that the appropriate route for this was by an enquiry under section 9A. The result of that omission was, as is now recognised but was not understood by the Revenue at the time, that subject only to being re-opened by a discovery assessment, Mr Tooth’s 2007-8 return and the self-assessment in it became final by, at the latest, the end of 2010: see section 9A(2)(c) of the TMA.
16. Mr Tooth and the Revenue eventually agreed to put their differences on hold pending the outcome of the litigation between the Revenue and another user of the Romangate scheme, a Mr Cotter. That was finally resolved in favour of the Revenue, by this court, in November 2013: see Revenue and Customs Comrs v Cotter  UKSC 69;  1 WLR 3514. In that case Mr Cotter had not, as Mr Tooth did, made his own self-assessment of the tax due, but rather left the assessment to the Revenue, while claiming separately (although within his tax return) a relief in respect of his Romangate-derived 2008-9 losses against his liability for tax for 2007-8. The Revenue contested the claim for relief by a Schedule 1A enquiry, and meanwhile assessed the tax due in respect of the 2007-8 year without allowing the loss. The effect of the enquiry was to suspend the claim for relief. They then sued Mr Cotter for the 2007-8 tax in the County Court, and eventually won in the Supreme Court. Critical to their success was the fact that Mr Cotter had not (unlike Mr Tooth) done his own self-assessment of the 2007-8 tax due. But in giving the leading judgment Lord Hodge observed, at para 27:
“Matters would have been different if the taxpayer had calculated his liability to income and capital gains tax by requesting and completing the tax calculation summary pages of the tax return. In such circumstances the revenue would have his assessment that, as a result of the claim, specific sums or no sums were due as the tax chargeable and payable for 2007/2008. Such information and self-assessment would in my view fall within a ‘return’ under section 9A of the 1970 Act as it would be the taxpayer’s assessment of his liability in respect of the relevant tax year. The revenue could not go behind the taxpayer’s self-assessment without either amending the tax return (section 9ZB of the 1970 Act (inserted by paragraph 2(2) of Schedule 29 to the Finance Act 2001) or instituting an inquiry under section 9A of the 1970 Act.”
17. The Cotter judgment did not in terms deal with the merits or otherwise of the Romangate scheme but, in December 2013, NT stated their view to the Revenue that the retrospective amendment which became section 128(5A) of the Income Tax Act 2007 rendered it ineffective.
18. In correspondence in March and May 2014 Grunberg and the Revenue agreed that Lord Hodge’s observation in para 27 of Cotter applied to Mr Tooth, because he had included a deduction for the Romangate-derived loss in his self-assessment return for 2007-8, so that the Revenue had not by its Schedule 1A enquiry protected its position. But in July 2014, while repeating that concession, Mrs Smith, an Inspector of Taxes, gave advance notice of her intention to make a discovery assessment against Mr Tooth, on the basis that he had deliberately brought about a loss of tax by claiming an employment-related loss from 2008-9 in the partnership pages of his 2007-8 return.
19. The precise method by which a decision was taken within the Revenue to make and give notice of the discovery assessment which followed on 24 October 2014 needs to be described, in the light of Mr Tooth’s contention, successful in the UT and the Court of Appeal, that the Revenue had not proved a qualifying discovery. The decision to issue the discovery assessment was made by Mr Williams, but he had retired by the time of the hearing in the FtT and did not give evidence. Instead, Mr March and Ms Thorley gave evidence for the Revenue at the hearing and were cross-examined. Ms Thorley’s evidence was that she worked in a team dealing with individuals who participated in the Romangate scheme and liaised closely with Mr Williams as the “technical lead”, whose role was to review papers on a case by case basis and decide whether an assessment should be issued. A caseworker in the team would send a submission to the technical lead who would check the circumstances of the case, decide whether to issue an assessment and provide a template letter for the caseworker to send to the taxpayer to give notice of the assessment. In the case of Mr Tooth, it was Mr Williams who had decided to make a discovery assessment and who gave instructions for the assessment to be issued to the taxpayer. Mr March, who had also worked for a period as the technical lead, being preceded by Mr Clarke in that role and succeeded by Mr Williams in August 2014, confirmed that Mr Williams reviewed Mr Tooth’s file in October 2014 and had come to the conclusion that a discovery assessment should be issued. Mr Williams delegated the formal making of the assessment to Ms Thorley, and she sub-delegated it to Mr Anders. Mr Williams had not been involved in considering Mr Tooth’s case before October 2014.
“In October 2014, Mr Williams reviewed Mr Tooth’s file and concluded, in line with mine and Mr Clarke’s thinking before him, that he had discovered a loss of tax as a result of the appellant deliberately filling out his return in the manner he did. Mr Williams therefore wrote to the operational team responsible for the administration of sending out assessments with that recommendation. This conclusion was reached following the advice given to the Technical Leads from policy specialists within [the Revenue].”
21. The FtT also had before it the discovery assessment issued by Mr Anders on 24 October 2014 and internal emails passing between Mr Williams, Ms Thorley and Mr Anders on 23 October 2014 in which Mr Williams first gave and then confirmed an instruction that there should be a discovery assessment in relation to Mr Tooth in the sum of £475,498.37. This replicated the figure which the Revenue had previously intended should be assessed by means of its Schedule 1A enquiry in 2009.
22. On the issue of whether Mr Williams had made a “discovery” as required by section 29 of the TMA, the FtT directed itself by reference to the judgment of the Upper Tribunal in Charlton v Revenue and Customs Comrs  STC 866 (“Charlton”), which is considered further below. The FtT found, at para 45 of its decision, that:
“Mr Williams did make a discovery of an insufficiency of tax in October 2014 as confirmed by the emails of 23 October 2014 ...”
The Statutory Framework
24. The resolution of the issues arising on this appeal requires the court to construe two sections of the TMA. The first, section 29, confers the power to make discovery assessments while the second, section 118(7), provides an interpretation of section 29(4). The material parts of each section (for present purposes) are as follows:
“29. Assessment where loss of tax discovered
(1) If an officer of the Board or the Board discover, as regards any person (the taxpayer) and a year of assessment -
(b) that an assessment to tax is or has become insufficient, or
the officer or, as the case may be, the Board may, subject to subsections (2) and (3) below, make an assessment in the amount, or the further amount, which ought in his or their opinion to be charged in order to make good to the Crown the loss of tax.
(3) Where the taxpayer has made and delivered a return under section 8 or 8A of this Act in respect of the relevant year of assessment, he shall not be assessed under subsection (1) above -
unless one of the two conditions mentioned below is fulfilled.
(4) The first condition is that the situation mentioned in subsection (1) above was brought about carelessly or deliberately by the taxpayer or a person acting on his behalf.
(5) The second condition is that at the time when an officer of the Board -
(a) ceased to be entitled to give notice of his intention to enquire into the taxpayer’s return under section 8 or 8A of this Act in respect of the relevant year of assessment; or
(b) informed the taxpayer that he had completed his enquiries into that return,
the officer could not have been reasonably expected, on the basis of the information made available to him before that time, to be aware of the situation mentioned in subsection (1) above.
(6) For the purposes of subsection (5) above, information is made available to an officer of the Board if -
(a) it is contained in the taxpayer’s return under section 8 or 8A of this Act in respect of the relevant year of assessment (the return), or in any accounts, statements or documents accompanying the return;
(b) it is contained in any claim made as regards the relevant year of assessment by the taxpayer acting in the same capacity as that in which he made the return, or in any accounts, statements or documents accompanying any such claim;
(c) it is contained in any documents, accounts or particulars which, for the purposes of any enquiries into the return or any such claim by an officer of the Board, are produced or furnished by the taxpayer to the officer; or
(d) it is information the existence of which, and the relevance of which as regards the situation mentioned in subsection (1) above -
(i) could reasonably be expected to be inferred by an officer of the Board from information falling within paragraphs (a) to (c) above; or
(ii) are notified in writing by the taxpayer to an officer of the Board.
(8) An objection to the making of an assessment under this section on the ground that neither of the two conditions mentioned above is fulfilled shall not be made otherwise than on an appeal against the assessment.
(5) For the purposes of this Act a loss of tax or a situation is brought about carelessly by a person if the person fails to take reasonable care to avoid bringing about that loss or situation.
(7) In this Act references to a loss of tax or a situation brought about deliberately by a person include a loss of tax or a situation that arises as a result of a deliberate inaccuracy in a document given to Her Majesty’s Revenue and Customs by or on behalf of that person.”
25. The relevant parts of the statutory context against which section 29 and section 118(7) fall to be construed are not themselves subject to any dispute as to their interpretation. The principal points may be summarised as follows. First, the expression “the Board” in section 29 means the Commissioners of Inland Revenue: see section 118(1). Secondly, as already noted, the time limits for a discovery assessment depend upon which of the two enabling conditions (or which part of them) in section 29(4) and (5) is satisfied. The second condition attracts the general time limit for assessments of four years from the end of the relevant year of assessment: see section 34(1). If the “carelessness” part of the first condition is satisfied, the time limit is extended to six years: see section 36(1). If the “deliberate” limb of the first condition is satisfied, the time limit is further extended to 20 years: see section 36(1A)(a).
26. In each case, the time limit for an assessment is expressed in positive terms. Section 34(1) states, “[s]ubject to the following provisions of this Act, and to any other provisions of the Taxes Acts allowing a longer period in any particular class of case, an assessment to income tax … may be made at any time not more than four years after the end of the year of assessment to which it relates”. Section 36(1) states, “[a]n assessment on a person in a case involving a loss of income tax … brought about carelessly by the person may be made at any time not more than six years after the end of the year of assessment to which it relates …”. In a case involving deliberateness, section 36(1A) states that “[a]n assessment … may be made at any time not more than 20 years after the end of the year of assessment to which it relates …”. Section 34(2) provides: “An objection to the making of any assessment on the ground that the time limit for making it has expired shall only be made on an appeal against the assessment”.
27. Thirdly, a broadly similar differential treatment of careless and deliberate conduct by the taxpayer is reflected in different levels of penalty which may be imposed. The penalty scheme is contained in Schedule 24 to the Finance Act 2007. A necessary condition for the imposition of a penalty is that an inaccuracy in a specified type of document (including a return) is either careless or deliberate: see Part 1, paragraph 1(3). Culpability is then divided into three ascending classes of seriousness: (i) careless, (ii) deliberate but not concealed and (iii) deliberate and concealed, each attracting progressively higher levels of penalty: see Part 1, paragraph 3 and Part 2, paragraph 4.
28. Some enlightenment as to the meaning of section 29 may also be gleaned from setting it in its historical context. The version of section 29 in its current format was substituted by the Finance Act 1994, to have effect from 1996 as part of the new system of self-assessment introduced by that Act. The equivalent of what is now section 29(1) was then section 29(3). Under the original section 29 there were wider powers to make discovery assessments. The distinction between decision-making by an officer of the Board and the Board itself was drawn very clearly. In 1970 income tax was assessed by inspectors (who were officers of the Board) while surtax was assessed by the Board. The TMA was a consolidation Act, and this distinction reflected the position that income tax administration was locally based for the most part, but surtax was administered centrally (originally by the special commissioners but, from 1965, by the Board): Sir Alexander Johnston, The Inland Revenue (1965), chapters 11-13. Section 29(1) of the TMA stated that, except as otherwise provided, all assessments to tax should be made by an inspector. Section 29(2) stated that all assessments to surtax should be made by the Board (defined in section 1(1) and again in section 118(1) to be the Commissioners of Inland Revenue). Section 29(3) provided, “[i]f an inspector or the Board discover … (b) that an assessment to tax is or has become insufficient … the inspector or, as the case may be, the Board may make an assessment in the amount, or the further amount, which ought in his or their opinion to be charged.”
29. Section 1(1) of the TMA provided that income tax was under “the care and management of the [Board]” and section 1(2) stated, “[t]he Board shall appoint inspectors and collectors of taxes who shall act under the direction of the Board”. The Act included various provisions requiring notices to be given by or to “an inspector” and others requiring notices to be given by or to “the Board”. However, in practice, pursuant to section 1(2) the Board delegated the exercise of its powers in relation to making assessments to officers. Accordingly, there was not a significant difference in practice between the two cases so far as section 29(3) was concerned. The significance of this is discussed below.
30. The new system of self-assessment which came into effect in 1996 was summarised by Moses LJ in Revenue and Customs Comrs v Tower MCashback LLP 1  EWCA Civ 32;  STC 809, paras 1-3 and 12-27. The previous wide power to issue a discovery assessment was restricted by the introduction of the two limiting conditions referred to in the substituted version of section 29(2)-(5): see Tower MCashback LLP 1, para 24. But the basic definition of what constituted a discovery so as to bring section 29 into operation was left unchanged: Hankinson v Revenue and Customs Comrs  EWCA Civ 1566;  1 WLR 2322, paras 15-18 (Lewison LJ).
31. In the form relevant to this appeal revised versions of sections 29 and 36 were introduced with effect from 1 April 2010 by Schedule 39 to the Finance Act 2008. Prior to that, the first condition in section 29(4) was that:
“the situation mentioned in subsection (1) above is attributable to fraudulent or negligent conduct on the part of the taxpayer or a person acting on his behalf.”
32. Prior to 1 April 2010 section 36 provided for the same 20-year time limit for both fraudulent or negligent conduct. Thus one of the reforms introduced by the Finance Act 2008 was to give careless taxpayers much greater protection from discovery assessment than had been enjoyed by their negligent predecessors, by the reduction of the time limit from 20 years to six years. This leads to an irresistible inference that, precisely when Parliament changed the language of the former fraudulent limb of the first condition to one of deliberate conduct, it nonetheless regarded that as substantially more blameworthy than carelessness.
33. The change from “fraudulent” to “deliberate” in section 29(4) was accompanied by the introduction of what is now section 118(7), with effect from the same date. There is some indication in the Explanatory Notes to the 2008 Act that the changes in language from negligence and fraud to carelessness and deliberate conduct were designed to align section 29 with the language of the penalty regime in Schedule 24 of the 2007 Act, or that the new terminology was at least borrowed from it.
34. The administration of the Inland Revenue and Customs and Excise was amalgamated in 2005 pursuant to the Commissioners for Revenue and Customs Act 2005 (“the CRCA 2005”). The term “Commissioners” in that Act is a reference to the Commissioners of Revenue and Customs, which includes the Commissioners of Inland Revenue, ie the Board as that term is used in the TMA. By virtue of section 5 of the CRCA 2005 the functions of the Board were vested in the Commissioners of Revenue and Customs. Section 2(1) of that Act states that “[t]he Commissioners may appoint staff, to be known as officers of Revenue and Customs”; and section 2(3) states, “[a]n officer of Revenue and Customs shall comply with directions of the Commissioners (whether he is exercising a function conferred on officers of Revenue and Customs or exercising a function on behalf of the Commissioners)”. By virtue of section 7 of that Act, functions conferred on an officer of the Board are vested in an officer of Revenue and Customs. Section 13 of the CRCA 2005 is headed “Exercise of Commissioners’ functions by officers”; subject to certain reservations in relation to non-delegable functions which are not material, subsection (1) provides: “An officer of Revenue and Customs may exercise any function of the Commissioners”. Section 14 provides for other powers of delegation; but subsection 14(4)(b) states that, subject to certain reservations which are not material, such delegation “shall not … prevent the exercise of the function by an officer of Revenue and Customs”, ie in recognition of the general delegation of the Commissioners’ functions pursuant to section 13 to enable any of them to be exercised by “an officer of Revenue and Customs”. All these provisions contemplate that functions relevant to the assessment of taxes are to be exercised by individual officers of Revenue and Customs, either on the basis of powers vested directly in them as officers or on the basis of powers of the Commissioners delegated to them by virtue of section 13(1), when they are identified internally as the officer responsible for acting as such in relation to a particular matter.
35. The issues on this appeal divide cleanly into two groups. The first group of issues (in terms of historical time) relate to the question whether the “deliberate” limb of the first condition for a discovery assessment was fulfilled, because of the way in which Mr Tooth and his advisors completed his 2007-8 tax return in 2009. That depends in part upon what may loosely be described as the place where the requirement for a deliberate inaccuracy (in section 118(7)) appears on a scale of blameworthy conduct ranging from mere conscious advertence at the bottom to something tantamount to fraud or dishonesty at the top. It also depends upon the question whether the word “in” within the phrase “in a document” in section 118(7) calls for a consideration of the accuracy or otherwise of the relevant document, read as a whole, or whether a sufficient inaccuracy may be identified by reading one part of it on its own, and disregarding the rest. It also calls for a consideration of the relevance, if any, of the fact that, in 2009 (as now), online tax returns were read, at least initially, by computers rather than by human beings, on their arrival at the Revenue.
36. The second group of issues relate to the question whether the Revenue pleaded or proved that there had been a qualifying discovery under section 29(1) in October 2014 to engage the power to make a discovery assessment at all. That depends in part upon the meaning of discovery, by an officer of the Board, in section 29(1).
37. While there is no escaping a meticulous examination of Mr Tooth’s tax return in order to ascertain whether it was inaccurate, or contained an inaccuracy, with due regard to the fact that (as the online form pointed out) it would initially be read by a Revenue computer, it is best first to form a clear view about what is required by the “deliberate” limb in the first condition under section 29(4) as illuminated (if that is the right word) by section 118(7). The trouble is that, read separately, section 29(4) and section 118(7) tend to pull in different directions. As Mr Julian Ghosh QC correctly submitted for Mr Tooth, section 29(4) read on its own suggests that to fall foul of the first condition (otherwise than by carelessness) the taxpayer must deliberately have brought about the “situation” mentioned in section 29(1), which in this case is that his self-assessment was insufficient. In other words, says Mr Ghosh, the taxpayer must have deliberately under-declared his tax liability. To allow this central feature of the requisite misconduct to be watered down by section 118(7) would be to allow the section 118(7) tail to wag the section 29(4) dog. By contrast Ms Hui Ling McCarthy QC for the Revenue submits that section 118(7) is a sort of deeming provision, which may perfectly legitimately control the meaning of section 29(4) which it seeks to interpret, and may thereby greatly reduce the seriousness on the scale of misconduct required to fulfil the first condition for a discovery assessment. In particular it may remove the requirement for a deliberate under-declaration of tax which section 29(4) read on its own might otherwise have appeared to require.
38. On this issue we consider that the Revenue must be correct. A deeming provision in a definition section of a statute may and commonly does give rise to a different meaning of the operative provision than the one which might have been arrived at by reading it on its own. That is, in a sense, what the definition section is frequently there for, although in many cases it may only be for the avoidance of doubt. It may in that sense wag the dog. Section 118(7) is not strictly a deeming provision, but rather one which includes conduct which it specifically describes within a class of conduct (that generally described in section 29(4)) which would not otherwise accommodate it. Its effect is that where the conduct which (in the present context) brings about or “results” in a situation consisting of an insufficiency within section 29(1) consists of an inaccuracy in a document given to the Revenue by or on behalf of the taxpayer, then the section 29(4) condition is fulfilled even if the insufficiency itself was not deliberate, provided that the inaccuracy was. In short, it decouples the insufficiency from the requisite intention, provided that the deliberate inaccuracy causes it in fact.
39. It may be, at least in theory, that section 118(7) does not cover the whole of the ground delineated by the “deliberate” limb of section 29(4). A taxpayer might bring about a relevant situation (here an insufficiency) by something said at a meeting with the Revenue. In such a case section 118(7) would have no application and the Revenue would have to show that the taxpayer intended by his statement to bring about the insufficiency, as section 29(4) would otherwise require.
40. It must be acknowledged that section 118(7) does therefore open the way to a discovery assessment with a 20-year time limit by reason of conduct by the taxpayer that falls well short of a deliberate under-declaration of tax, so that it might be hard to describe such conduct as fraudulent, using the language of section 29(4) in force before 2010. But the language of section 118(7) is clear and unambiguous in having the effect described above. Suppose for example that Mr Tooth had entered the employment-related loss in the partnership box without providing any explanation of its true source and nature, intending the Revenue to believe that it derived from a partnership business and occurred during the 2007-8 tax year. It would still have generated a negative tax liability for 2007-8 in the amount which he genuinely believed to be true. But if that negative amount (and the consequent insufficiency) was the result of entering the loss in the partnership box, as a partnership loss, (rather than elsewhere in the return) then we consider that the deliberate limb of the first condition in section 29(4) would have been fulfilled. Such a presentation in his return would have misled the Revenue from having a full understanding of the information relevant to assessing his self-assessment.
41. But the Revenue seek to go further. Ms McCarthy submits that the requirement in section 118(7) that the documentary inaccuracy should be deliberate means only that the statement in the document constituting the inaccuracy was deliberately made: ie made intentionally rather than for example carelessly or by mistake. This would apparently extend to an intentional statement which was in fact inaccurate, even though genuinely believed by the maker to be true, and not intended to mislead. She sought support for this submission by reference to West v Revenue and Customs Comrs  UKUT 100 (TCC);  STC 1004 (“West”). In that case the relevant question was whether the taxpayer knew that there had been a wilful failure by a company by which he was employed to deduct PAYE, within the meaning of regulation 72C(2)(a) of the Income Tax (Pay as You Earn) Regulations 2003 (SI 2003/2682). At para 64 the UT (Sir Geoffrey Vos C and Judge Berner) said:
“For a person wilfully to effect a particular legal outcome, it is not necessary for that person to be cognisant of the legal consequences of his or her actions. It is necessary only for that person intentionally or deliberately to put in train the various actions (or knowingly to fail to do so in the case of omissions) that in the event have the material consequences in law.”
42. The very different and complex context in which this apparently general observation was made was not explored in sufficient detail for it to be appropriate for this court to pass useful comment on it. It is sufficient to say (not least because it was a dictum about the contextual meaning of wilful rather than deliberate) that it offers no assistance on the true construction in this very different context of the phrase “deliberate inaccuracy” in section 118(7). The question is whether it means (i) a deliberate statement which is (in fact) inaccurate or (ii) a statement which, when made, was deliberately inaccurate. If (ii) is correct, it would need to be shown that the maker of the statement knew it to be inaccurate or (perhaps) that he was reckless rather than merely careless or mistaken as to its accuracy.
43. We have no hesitation in concluding that the second of those interpretations is to be preferred, for the following reasons. First, it is the natural meaning of the phrase “deliberate inaccuracy”. Deliberate is an adjective which attaches a requirement of intentionality to the whole of that which it describes, namely “inaccuracy”. An inaccuracy in a document is a statement which is inaccurate. Thus the required intentionality is attached both to the making of the statement and to its being inaccurate.
44. Secondly, “deliberate inaccuracy” is the gateway to the taxpayer’s exposure to a 20-year period for the making of a discovery assessment, because of the importation of that phrase from section 118(7) into section 29(4). If the first interpretation were to be preferred the taxpayer could incur that exposure by making an honest but in fact inaccurate statement, even after taking reasonable care as to its truth or falsehood. The taxpayer would not even need to be careless, and yet would incur a much longer exposure than if he had been.
45. Thirdly, the penalty scheme in Schedule 24 to the Finance Act 2007 had, shortly before the relevant amendments were made to section 29 (including section 118(7)), used the same concept of deliberate inaccuracy for the purpose of triggering penalties more serious than those arising from carelessness, at altogether higher levels of blameworthy conduct (even though subdivided by reference to the presence or absence of concealment). It seems inconceivable that Parliament would have chosen the same language to serve as the gateway to the longest available period of exposure to a discovery assessment, if the phrase was to be interpreted as meaning only that the statement was intentionally made.
46. Fourthly, as already noted, the phrase was introduced at the same time as a substantial shortening of the exposure period for carelessness, which leads to the clear inference that Parliament must have regarded “deliberate inaccuracy” as conduct substantially more blameworthy. It is to be noted that, as Ms McCarthy submitted, there are other triggers for a 20-year time limit for an assessment which do not necessarily lie on a scale of blameworthiness between carelessness and fraud, but their existence does not displace the powerful inference as to Parliament’s intention already described.
47. It may be convenient to encapsulate this conclusion by stating that, for there to be a deliberate inaccuracy in a document within the meaning of section 118(7) there will have to be demonstrated an intention to mislead the Revenue on the part of the taxpayer as to the truth of the relevant statement or, perhaps, (although it need not be decided on this appeal) recklessness as to whether it would do so.
48. The final question of construction of section 118(7) is as to the meaning of the phrase “in a document” as the place where the relevant deliberate inaccuracy is to be found. The Revenue submitted, and persuaded a majority of the Court of Appeal (Males and Patten LJJ), that it was enough if a deliberate inaccuracy could be found somewhere in the document, interpreted on its own and without regard to the rest of the contents of the document, always provided that it had the requisite causative effect in bringing about the insufficiency of tax or other relevant situation under section 29(1). By contrast Mr Tooth submitted, and Floyd LJ held in the Court of Appeal, that the accuracy or otherwise of the statement relied upon could only properly be assessed in the light of the whole of the document in which it was contained.
49. It almost goes without saying that the meaning of particular words or phrases in a document of any kind is generally to be ascertained by a contextual approach, that is by appraising the critical passage in the light of its context as part of the document read as a whole. There is no reason in principle why the same should not be true of a tax return. But in this respect the sheet anchor of the Revenue’s case was the fact that, as the online tax return form clearly stated, the return would be read upon receipt at the Revenue by a computer rather than, initially at least, by a sentient, literate, human being. Computers, it was said, do not do contextual interpretation, but look at each part of, or box in, the return separately.
50. This is, with respect, a very unattractive argument. A document written in the English language (or any language other than computer language) does not have a different meaning depending upon whether it is read by a human being or by a computer. A choice by the recipient of such a document to have it machine-read cannot alter its meaning. Furthermore, the Revenue-approved online tax return form used by Mr Tooth and his advisors contained numerous “white spaces”, that is, sections of blank white-coloured space, usually headed “Additional Information”, within which the taxpayer is invited to add information using his own words and phrases, so as to ensure that the declaration required to be made and signed, namely that:
“The information I have given on this Tax Return is correct and complete to the best of my knowledge and belief”
is actually true.
51. The Revenue cannot in our view have it both ways. If they sensibly include ample white spaces in their approved form of online returns so as to ensure that the taxpayer is not constrained by the limitations of the boxes for figures from making a correct and complete return, then they cannot thereafter assert, for the purpose of advancing a non-contextual interpretation of one or more boxes, that their computer cannot read what is written on the white spaces. This must be a fortiori true where, as in the present case, the white space used by Mr Tooth to provide a full and frank explanation of the true meaning of the supposedly offending insertions in the partnership boxes was to be found immediately adjacent to those boxes.
52. It follows that the task of interpretation required by the need for the Revenue to prove that there was a deliberate inaccuracy in Mr Tooth’s tax return was to interpret the meaning of each relevant part of the document by reference to its place in the context of the document as a whole, just as is normally done when interpreting any other document. This is not to say that the word “in” as part of the phrase “deliberate inaccuracy in a document” is to be ignored. It simply means that, rather than concluding that there is an inaccuracy by applying tunnel-vision to a particular part of the document, and ignoring the rest, the true meaning of that part is ascertained from a reading of the document as a whole.
53. Applying that interpretation of section 29 and section 118 to Mr Tooth’s 2007-8 tax return, we are satisfied that both the FtT and the UT were correct to conclude that there was no deliberate inaccuracy to be found in it, sufficient to fulfil the first condition for a discovery assessment. The main reason is, as Floyd LJ also held, that there was simply no inaccuracy at all, deliberate or otherwise. Although the application of tunnel-vision to boxes 1, 7, 19 and 20 of the partnership pages might suggest at first blush that Mr Tooth was claiming to have incurred a partnership-related loss in the 2007-8 tax year (which was plainly inaccurate to his and his advisors’ own knowledge) perusal by a reasonably well-informed and careful reader of the detailed explanations provided in box 19 on TR6 and box 30 on SP2 revealed, to the contrary, that he was claiming to carry back an employment-related loss from the 2008-9 tax year derived from participation in a tax avoidance scheme, and doing so by entering the relevant figure in the partnership box because the form failed to provide any employment-related box within which to do so, contrary to his interpretation of a requirement that it should do so in order to reflect section 128 of the Income Tax Act 2007. And he even invited the Revenue to open an enquiry into his return on the express basis that he expected that the Revenue would disagree with his understanding as to his right to do so.
54. It is no coincidence that, in the present case, the officials in the Revenue dealing with his case were under no misapprehension at any time once they had read the return about what Mr Tooth was seeking to do. It is plain from the correspondence between the parties that the Revenue understood perfectly well from their first (human) reading of Mr Tooth’s return that he was seeking to carry back employment-related losses derived from his participation in the Romangate scheme in the 2008-9 tax year as a deductible against his 2007-8 income, and that what he was doing had nothing whatsoever to do with any partnership.
55. Ms McCarthy sought to support the Revenue’s case by referring to the statutory power of the Revenue to specify the form on which a self-assessment return should be presented. She submitted that it was not open to a taxpayer to enter figures incorrectly in the boxes shown on the return, which would be read in the first instance by computer. This was a reason for departing from the usual approach to interpreting a document as a whole.
56. The difficulty with that, however, is that the electronic form itself was defective for the purpose it was supposed to fulfil. The figure in box 1 for self-assessment to tax was automatically generated as a result of a calculation based on figures entered in other boxes on the form. According to the technical advice given upon enquiry, other than by inserting figures in the boxes for partnership information it was not possible for a taxpayer in Mr Tooth’s position to generate the proper self-assessment figure in box 1 which he, for legitimate reasons, wished to present. So either Mr Tooth had to present his return with a figure in box 1 showing a self-assessment to tax in an amount in excess of £475,000, which would not have represented the correct amount of his self-assessment, or he had to put figures in an unsuitable box which would generate the correct amount of his self-assessment in box 1 (a credit of about £13,000 due to him). Either way, he would have to make a return with an incorrect amount in one or other box and accompany that with a written explanation in a white space on the form.
57. The solution which Mr Tooth adopted had, as it seems to us, the considerable merit of showing the correct overall figure at which he self-assessed his liability to tax. Ms McCarthy said, however, that he should have taken the first route and shown an incorrect self-assessment figure in box 1, and then explained in the white space that this was not the correct figure and what his true self-assessment was. But, like the different solution which Mr Tooth adopted, this solution would also have depended upon showing an incorrect figure in a box on the form and accompanying that with text to explain the true position. In our view, the fact that the Revenue themselves suggest a solution to the impasse with which they confronted Mr Tooth which would involve presentation of an incorrect figure in a relevant box on the form to be explained away by accompanying text in a white space shows that there is nothing in the structure of the electronic form which can be relied on to convert honest presentation of his claim for relief in his self-assessment return into a deliberate inaccuracy. This suggested solution by the Revenue is just a variant of the solution Mr Tooth adopted. As Lord Carnwath observed in R (Derry) v Revenue and Customs Comrs  UKSC 19;  1 WLR 2754, para 65, the fact that a form filled in with an inappropriate number entered in one or other box would bypass the Revenue’s automated checks is “not the fault or the concern of the taxpayer”.
58. Even if we could have been persuaded, contrary to our conclusion above, that by some tunnel-vision approach to the interpretation of parts of Mr Tooth’s return, ignoring all context, it contained an inaccuracy, we would not have been satisfied that it was deliberate in the sense, explained above, that Mr Tooth or his advisors knew that the relevant statements were inaccurate and intended thereby to mislead the Revenue. Ms McCarthy submitted that all that was achieved by the extensive passages in the white spaces on Mr Tooth’s return, referred to above, was to explain why the statements in the relevant partnership boxes were deliberately inaccurate. That cannot be right. Reading the return as a whole, Mr Tooth and his advisors did their best, in the context of an intractable online form which did not appear to enable them to do it more directly, to explain the employment-related and scheme-derived basis of his ambitious claim to extinguish his 2007-8 tax liability by an admittedly contentious carry-back. It is unnecessary in this context to decide whether they had no alternative but to do it that way, a point briefly but inconclusively argued in this court, nor whether the requisite causative link between the alleged inaccuracy and the insufficiency of tax was established.
59. That conclusion, which departs from the reasoning of the majority in the Court of Appeal, and of all three of its members on the question of deliberateness, is sufficient to require the appeal to be dismissed. But the important question as to whether there was even a qualifying discovery was fully argued, in fact at greater length than the issue as to deliberate inaccuracy. To that we now turn.
60. The Revenue’s case in the FtT was that the relevant discovery for the purposes of section 29(1) of the TMA was made by Mr Williams as the assessing officer when he looked at Mr Tooth’s tax return in October 2014 for the first time and decided that Mr Tooth’s self-assessment to tax was understated. The FtT accepted the Revenue’s submission on this point. It rejected a submission by Mr Tooth that for the purposes of the application of section 29(1) it was appropriate to have regard to the collective knowledge of the Revenue, including knowledge other officers had acquired in 2009 when Mr Tooth’s return was received and first considered, so that whoever made the relevant discovery it was not Mr Williams.
61. The UT reversed the FtT on this point. The UT took the view that if two different officers of the Revenue independently make the same discovery at different times, it only qualifies as a discovery for the purposes of section 29(1) of the TMA in relation to the first discovery; although the second officer finds out something that is new to him, it is not a discovery for the purposes of the provision. In the UT’s view, such a construction is necessary to protect the taxpayer from a stale assessment, ie an assessment based on a discovery made some time previously but allowed to lie unactioned on the file for a significant period. Relying on and affirming the view of the UT (Norris J and Judge Berner) in Charlton, para 37, the UT held that a discovery would lose its quality as a discovery, and would cease to qualify as such for the purposes of section 29(1) of the TMA, if no assessment was issued while it was new, and it was allowed to become “stale”. This was what had happened in this case. The Revenue (acting by officers other than Mr Williams) discovered the insufficiency of tax in 2009 and first identified it in the letter to Mr Tooth in August 2009, so “this was no discovery by Mr Williams at all [in October 2014]” (para 85). The UT stated that the discovery in 2009 was stale by October 2014 and could not be relied on for issuing an assessment then (para 83). We would observe that the Revenue did not identify the officers who reviewed Mr Tooth’s return in 2009 and did not seek to rely on any discovery made then; nor did they seek to rely on the discovery made by Mrs Smith in July 2014.
62. The Court of Appeal unanimously upheld the decision of the UT on the discovery issue for reasons given by Floyd LJ. Like the UT, Floyd LJ endorsed the reasoning in Charlton, para 37. He held that Mr Williams did not make a relevant discovery when he reviewed Mr Tooth’s file in October 2014 because it was clear that other Revenue officers had previously considered that Mr Tooth’s return involved the same insufficiency of tax as Mr Williams identified in October 2014, quite possibly in 2009 and certainly by the time of Mrs Smith’s letter in July 2014.
63. The FtT, the UT and the Court of Appeal all based their consideration of this issue on what was said by the UT in Charlton. In that case the UT affirmed the judgment of the FtT that the officer who reviewed the case (Mr Cree) had made a discovery for the purposes of section 29(1) of the TMA that there was an under-assessment of tax in the taxpayer’s returns after the closure of the section 9A enquiry window, even though the Revenue had failed to open such an enquiry in time by an oversight, being aware at or about the time the returns were received that they wished to challenge the tax avoidance scheme relied upon by the taxpayers (and where they had in fact opened section 9A enquiries into a considerable number of returns by other taxpayers relying on the same scheme) and that they already had a ruling of the Special Commissioner in their favour in respect of the ineffectiveness of an equivalent scheme. In the event, despite this error by the Revenue, the UT held that this would not have prevented Mr Cree from being entitled to issue a discovery assessment under section 29(1). At para 28 the UT observed that the word “discovers” in section 29(1) of the TMA connotes a change in a state of mind: “At one point an officer is not of the view that there is an insufficiency such that an assessment ought to be raised, and at another he is of that view”. The UT said this at para 37:
“In our judgment, no new information, of fact or law, is required for there to be a discovery. All that is required is that it has newly appeared to an officer, acting honestly and reasonably, that there is an insufficiency in an assessment. That can be for any reason, including a change of view, change of opinion, or correction of an oversight. The requirement for newness does not relate to the reason for the conclusion reached by the officer, but to the conclusion itself. If an officer has concluded that a discovery assessment should be issued, but for some reason the assessment is not made within a reasonable period after that conclusion is reached, it might, depending on the circumstances, be the case that the conclusion would lose its essential newness by the time of the actual assessment. But that would not, in our view, include a case, such as this, where the delay was merely to accommodate the final determination of another appeal which was material to the liability question. Such a delay did not deprive Mr Cree’s conclusions of their essential newness for section 29(1) purposes.”
64. On the appeal to this court, Ms McCarthy submits that the first four sentences of this paragraph are correct, but disputes the correctness of the latter part of the paragraph and the idea that a discovery can lose its quality as such if there is a delay after it is made and, as it has been put in some cases, it becomes stale. She also submits that the question whether there is a discovery for the purposes of section 29(1) depends upon the state of mind of the individual officer of the Revenue who decides to make the assessment. For these purposes there is no concept of the Revenue having collective knowledge such that if one officer makes a discovery that is to be regarded as a discovery made once and for all by the Revenue as a whole. The result, according to Ms McCarthy’s submission, is that a second or third officer (and so on) can also make the same discovery for themselves, so that each of them becomes entitled serially to issue a discovery assessment. The legal controls against misuse of the discovery assessment regime lie in the conditions referred to in section 29(3)-(5) and the statutory time limits and in the field of public law and the possibility of obtaining relief in judicial review proceedings against any abuse of power. Ms McCarthy submits that the FtT found that the officer who issued the discovery assessment in this case, Mr Williams, made the relevant discovery for himself in October 2014, and was entitled so to find. Mr Ghosh, on the other hand, seeks to uphold the decision and reasoning of the UT and the Court of Appeal.
65. We accept the submissions of Ms McCarthy. The points she makes have wider significance than the present case, so we set out our reasoning below. We will deal with the issues of collective knowledge, what is required for a discovery within the meaning of section 29(1) and the legal protections where there is a delay between discovery and the issuing of an assessment.
66. First, however, we observe that it is a significant feature of the legislation that the time limits which apply are not triggered by a discovery event, but run from the end of the year to which the relevant assessment relates. Thus, in the context of the statutory regime, the concept of a discovery is employed not to determine the period beyond which the taxpayer is safe from further assessment but to trigger the powers of an officer or the Board to revisit an assessment already made.
67. Before self-assessment was introduced with effect from 1996, the first assessment of tax based on a taxpayer’s return was made by a tax inspector, an officer of the Board. In a case falling within what is now section 29(1)(b), a discovery assessment would typically be made when a second officer reviewed the file at a later date and formed the view that this assessment to tax was insufficient (a relevant discovery might also be made if the first officer came to think that they had made an under-assessment when they first dealt with the file). Under the self-assessment system, the first stage of assessment has been passed to the taxpayer, but the concept of discovery has not changed. An officer makes a discovery that an assessment to tax is insufficient upon reading the form containing a self-assessment and forming the opinion that the assessment it contains is incorrect and too low.
68. Section 29(1) of the TMA, concerning discovery assessments, operates in two cases. First, it confers a power on “an officer of the Board”, if he discovers a matter falling within sub-paragraphs (a) to (c) - we are concerned in this case with sub-paragraph (b) - and subject to subsections (2) and (3), to make an assessment in the amount which ought “in his opinion” to be charged to make good the loss of tax. Secondly, it confers a power on the Board themselves to make an assessment according to “their opinion”, if the Board discover the deficiency of tax. These cases are distinct, as the language of the provision makes clear. Discovery by “an officer of the Board” is treated as something separate from discovery by “the Board”, as the phrase “as the case may be” in the last part of section 29(1) also highlights. Moreover, the language and structure of the provision would make no sense if its operation turned on a concept of collective knowledge of the Board, derived from the knowledge of any and all of its officers. The reference to discovery by an officer of the Board would be otiose, since such discovery would on that hypothesis always constitute discovery by the Board. Further, the condition in section 29(5) operates by reference to the state of mind of a particular hypothetical officer of the Board dealing with the taxpayer’s case at a particular point in time (either when the time limit for commencing an enquiry into a return made under section 8 or section 8A expired or when he informed the taxpayer that he had completed his enquiries into the return), and does not involve any concept of collective knowledge on the part of the Revenue: see Langham (Inspector of Taxes) v Veltema  EWCA Civ 193;  STC 544, para 36 (Auld LJ); Sanderson v Revenue and Customs Comrs  EWCA Civ 19;  4 WLR 67, para 17(5) (Patten LJ); Charlton, paras 65-66. Section 29(1), which also in relevant part by use of similar language focuses on the state of mind of an officer of the Board, should not be interpreted differently.
69. This view is supported by consideration of section 29 as it was originally enacted. In section 29(3) (the equivalent of what is now section 29(1)) the distinction between decision-making by an officer of the Board and by the Board itself was drawn very clearly, as set out above. It is clear that section 29(3) (as originally enacted) focused on the state of mind of the relevant decision-maker, being an inspector or the Board depending on the distribution of functions between them. The question whether there was a discovery by an inspector depended on the particular inspector’s state of mind, not on what knowledge might be located elsewhere in the Revenue organisation. The only relevant change in the current version of section 29(1) is that the phrase “an officer of the Board” has been used in place of “an inspector”, to cover a wider range of officeholders with statutory functions in relation to the assessment of tax. The position remains that, for the “officer” limb of section 29(1) in its current form, the provision is concerned with the state of mind and knowledge of the particular officer who claims to have made a relevant discovery and then purports to exercise the power to make an assessment which arises under that provision when that condition is fulfilled. This is the ordinary meaning of the words used in the provision, and there is good reason to construe them in this way. The officer in question needs to know if a discovery has been made in order to know if they have power under section 29(1) to issue an assessment and reference to their own state of mind enables them to know with confidence whether they have that power. This interpretation also appears to match the way in which the Revenue works in practice, as illustrated by the evidence in this case, where a taxpayer’s file is allocated to a particular officer to review and take relevant decisions and actions, drawing as necessary on advice or submissions presented to the officer by others. The provision contemplates that a particular officer will personally have full decision-making responsibility in relation to a taxpayer’s file.
70. The same position applied when an officer of the Board exercised functions of the Board under section 29 in its original form on the basis of the delegation of those functions to that officer. Where the Board delegated its function of assessment under the original versions of section 29(2) and (3) to an officer, the effect was that the relevant state of mind and knowledge was that of the officer. As the delegate of the Board, the officer stood in their place. The officer would act on their behalf on the basis of the officer’s own state of mind and knowledge, not by reference to the state of mind or knowledge of anyone else in the organisation. The same would be true if, exceptionally, the Board decided to exercise the function of assessment under section 29 for themselves. The Commissioners would then act on the basis of their own state of mind and knowledge as a panel, not by reference to the state of mind or knowledge of anyone else in the organisation. This is all a matter of ordinary principle in relation to the exercise of powers in public law. See, for example, the analysis by the Court of Appeal in R (National Association of Health Stores) v Department of Health  EWCA Civ 154 in relation to the Carltona doctrine (Carltona Ltd v Comrs of Works  2 All ER 560) of implied delegation of decision-making functions to civil servants within a government department, at paras 23-38 (Sedley LJ) and 71-74 (Keene LJ) and also in R (Bancoult) v Secretary of State for Foreign and Commonwealth Affairs (No 3)  UKSC 3;  1 WLR 973, para 47 (Lord Mance). There is no principle of collective knowledge within a department. If a civil servant acts on behalf of a Minister, it is the civil servant’s knowledge and state of mind which are relevant; if the Minister decides to take the relevant decision himself or herself, it is the Minister’s knowledge and state of mind which are relevant. The Carltona doctrine of implied delegation has the same effect as the express delegation of powers effected by what is now section 13(1) of the CRCA 2005 and previously under section 1(2) of the TMA, and the same point about the absence of a principle of collective knowledge holds good.
71. The position set out above continues to apply in relation to the operation of section 29(1) in its current form both as a matter of principle and because there has been no material change from its predecessor provision in the original version of section 29(3). Although it did not need to decide the point, the UT in Charlton at paras 40-43 rejected the contention that the question whether a discovery had been made was to be tested by reference to the collective knowledge of the Revenue, rather than the knowledge of an individual officer. We consider it was right to do so.
72. This view of the operation of section 29(1) is supported by other authority as well. In Sanderson v Revenue and Customs Comrs  EWCA Civ 19;  4 WLR 67, para 25, Patten LJ explained that “[t]he exercise of the section 29(1) power is made by a real officer who is required to come to a conclusion about a possible insufficiency based on all the available information at the time when the discovery assessment is made.” From this and other authorities the UT (Morgan J and Judge Berner) in Anderson v Revenue and Customs Comrs  UKUT 159 (TCC);  STC 1210 derived a series of propositions (para 24), including that in section 29(1) the concept of an actual officer discovering something involves an actual officer having a particular state of mind in relation to the relevant matter, which requires the application of a subjective test (explained further at paras 25-28). There is also an objective test, in that mere suspicion of an under-assessment of tax is not sufficient and the belief which the officer forms regarding the under-assessment has to be one which a reasonable officer could form (paras 24 and 29-30). The UT in Anderson rightly acknowledged that section 29(1) set out public law powers and its interpretation was informed by principles of public law.
73. On the question of what qualifies as a discovery for the purposes of what is now section 29(1) of the TMA, Cenlon Finance Co Ltd v Ellwood (Inspector of Taxes)  AC 782 is the leading authority. It was concerned with section 41(1) of the Income Tax Act 1952, the predecessor of section 29. A first inspector agreed the taxpayer’s computation of trading profits and issued an assessment to tax on the basis of it. Subsequently, the file was passed to a new inspector who, on the basis of the same facts as had been before the first inspector, came to the conclusion that an additional sum should have been included in the profit figure and he issued an additional assessment for tax in relation to the revised figure. The taxpayer argued (p 788) that it could not be said that the second inspector had discovered anything new and that it would be unjust to put the taxpayer in peril for an extended period of time (six years under the regime then applicable) when he had made all the facts known to the Revenue at the outset. However, the House of Lords held that there had been a discovery by the second inspector within the meaning of the provision, so that he was entitled to issue the new assessment. Viscount Simonds (p 794), with the agreement of the other members of the appellate committee, approved the judgment of Lord Normand in the decision of the Court of Session in Inland Revenue Comrs v Mackinlay’s Trustees 1938 SC 765; 22 TC 305 and added, “I can see no reason for saying that a discovery of undercharge can only arise where a new fact has been discovered. The words are apt to include any case in which for any reason it newly appears that the taxpayer has been undercharged and the context supports rather than detracts from this interpretation.” Lord Denning observed (p 799), “[e]very lawyer who, in his researches in the books, finds out that he was mistaken about the law, makes a discovery. So also does an inspector of taxes”.
74. In the Mackinlay’s Trustees case, an assessment to surtax was made in 1935 on a view that the taxpayer was not entitled to certain profits under a partnership deed. In 1937 a different view was taken about that and a discovery assessment was issued to claim the tax in relation to those profits. The change in view about the legal effect of the partnership deed was held to constitute a discovery for the purposes of the provision which became section 29(1) of the TMA. Lord Normand said (22 TC 305, 312), “I think the word ‘discover’ in itself, according to the ordinary use of language, may be taken simply to mean ‘find out’. What has to be found or found out is that any properties or profits chargeable to tax have been omitted from the first assessment.” Referring to the condition in what is now section 29(1)(b), Lord Normand said (22 TC 305, 312):
“I think that, since these words must apply where the person chargeable has delivered a full and proper statement, they are apt to cover the case of a discovery of a mistake in the assessment caused by a mistake in the construction of the partnership deed or, it may be, caused by a mistake in the law applicable to such a deed, even where there has been a complete disclosure of all relevant facts upon which a correct assessment might have been based. I do not think it is stretching the word ‘discovers’ to hold that it covers the finding out that an error in law has been committed in the first assessment, when it is desired to correct that by an additional assessment.”
Lord Normand considered that this wide interpretation of the word “discover” in the provision was also supported by its application in relation to the other conditions in the provision.
75. There was no suggestion in the Mackinlay’s Trustees case or in the Cenlon Finance case that any notion of staleness applied or that a discovery might lose its quality as such simply by the effluxion of time. That would be contrary to the ordinary use of language. Viscount Simonds’ reference in the latter case to discovery covering “any case in which … it newly appears that the taxpayer has been undercharged” was a reference to the state of mind of the person said to have made the discovery, to whom it “newly appears” that an assessment to tax is insufficient. A discovery is a particular event in time, and does not cease to be such with the passage of time. As is made clear in both cases, a discovery within the meaning of what is now section 29(1) of the TMA may consist simply in a new appreciation of the legal significance of a set of circumstances.
76. In our judgment, contrary to the latter part of para 37 in the decision in Charlton, there is no place for the idea that a discovery which qualifies as such should cease to do so by the passage of time. That is unsustainable as a matter of ordinary language and, further, to import such a notion of staleness would conflict with the statutory scheme. That sets out a series of limitation periods for the making of assessments to tax, each of them expressed in positive terms that an assessment “may be made at any time” up to the stated time limit.
77. There is no basis for implication of an additional and stricter time restriction as suggested by the UT in Charlton. The UT in that case appears to have arrived at its view in the latter part of para 37 by misreading Viscount Simonds’ reference to something “newly appear[ing]” in the Cenlon Finance case. In our view, in using that expression, Viscount Simonds was only referring to the initial change in the officer’s state of mind required to satisfy what was described in the Anderson case as the subjective element of the test for a discovery in section 29(1). Further, the UT’s statement in Charlton was an obiter dictum not supported by detailed reasoning; on the facts in the case, it found that the relevant officer (Mr Cree) had acted in time.
78. Mr Ghosh submitted that once a discovery is made by one person it cannot be made again by another. We do not accept this. It is perfectly possible, as a matter of ordinary language, to speak of someone making a discovery for himself or herself even if it is something already known to others. The approach in the authorities supports this view. As we have explained above, since section 29(1) is concerned, so far as is relevant, with a discovery made by an officer of the Board, the question is whether the officer of the Board who is deciding whether to make a discovery assessment under that provision has subjectively made a discovery that there has been an under-assessment of tax.
79. This is the position which applies in the standard case contemplated by section 29(1), where a taxpayer’s file has been allocated to an officer of the Board who has responsibility for making all relevant decisions as regards assessment to tax in relation to it. The position is capable of being qualified by the operation of what is now section 2(4) of the CRCA 2005, which provides that “[a]nything … begun by … one officer of Revenue and Customs may be continued by … another”. This would allow for one officer to begin consideration of a file under section 29(1) of the TMA and make a discovery and then pass it on to another to complete the exercise of assessment without the second having to revisit the opinion of the first officer that there was an insufficient assessment to tax in the return. But in the present case there was no such division of responsibility between officers. The Revenue’s entire case before the FtT was based on the ordinary application of section 29(1), referring to the subjective state of mind of Mr Williams and a discovery said to be made by him.
80. This interpretation of section 29(1) and the way in which it operates alongside the relevant limitation periods seems to us to be clear. The subjective nature of the test in section 29(1) by reference to the state of mind of a particular officer means that there is the possibility that a series of officers might each make the relevant discovery if a taxpayer’s file is passed from one to another. However, this is not a reason for seeking to introduce further restrictions as a matter of statutory interpretation. The statutory regime makes clear the time limits within which the taxpayer may be exposed to a discovery assessment, and they do not run from the date of the relevant discovery.
81. As Ms McCarthy submitted, in interpreting section 29(1) there is no good reason in principle to distinguish between cases where a first officer who is allocated a self-assessment return to review puts it in a drawer unread and so does not see that it contains an under-assessment of tax, or fails to read it carefully and so does not spot that it contains an under-assessment of tax, or reads it carefully but makes an error of appreciation of the facts or the law and so again does not realise it contains an under-assessment of tax, or reads it carefully and finds the error but then goes on holiday, moves job or retires without issuing an assessment. In the first three cases it is clear that when the file is picked up by a second officer who conducts his or her own review of the file and realises or believes that the assessment in the return is wrong, that officer makes a discovery of that and has power to issue an assessment under section 29(1). The fourth case is not materially different. When the file is reviewed by a second officer, that officer makes his or her own discovery of the error of assessment in the return and accordingly also has power to issue an assessment under section 29(1). In each case, it might be said that there was an element of administrative failure which caused delay in the issuing of a discovery assessment. From the point of view of the taxpayer the four cases are the same. There is no good reason, under the statutory regime, to distinguish the fourth case from the first three. Also, there is no good reason to suppose that a purpose of the legislation was that a taxpayer who committed a blatant fraud, of a kind likely to be spotted by the first inspector to look at the file, should be better protected against being exposed to a discovery assessment issued by a subsequent inspector if the first one for whatever reason fails to issue an assessment, than a taxpayer whose fraud is better concealed and therefore more likely only to be discovered by another officer at a later stage.
82. Section 29(1) has to be capable of operating in a practical way in other situations as well. Varying the circumstances of the fourth type of case above, the first officer might leave a note on the file to say that he or she has made a discovery of an insufficiency of assessment to tax but the note might be difficult to interpret, so the second officer has to repeat the exercise of examining the assessment to form his or her own view. Or the first officer’s note might be thorough and clear, but the second officer may not have confidence in what the first has done and so may wish to make his or her own evaluation. In both these cases, it will be the second officer who makes the discovery which is relevant to clothe the officer with the power to issue an assessment under section 29(1). This makes the point, again, that section 29(1) contemplates that a particular officer has the responsibility for carrying out the exercise of re-evaluation and then issuing an assessment in light of it, and no one else can do that for the officer (subject only to the possibility of a division of responsibility by virtue of the operation of section 2(4) of the CRCA 2005, as discussed above).
83. There are a number of protections for the taxpayer in relation to exposure to a discovery assessment. As explained above, the new version of section 29 of the TMA which came into effect in 1996 contained important new conditions in relation to the previously much wider power to issue a discovery assessment. The taxpayer also has the protection of the statutory time limits, which as we have explained are linked to different levels of culpability on their part. It is typically only in relation to what amounts to fraud or is akin to fraud that the time limit becomes as long as 20 years. These are matters which fall within the scope of an appeal to the FtT against an assessment: sections 29(8), 31(1)(d) and 50(6) of the TMA.
84. In addition, as Ms McCarthy points out, where the statutory discovery condition in section 29(1) is satisfied, an officer or the Board has a discretion whether to issue an assessment (“may … make an assessment”) and they must act in accordance with ordinary principles of public law in deciding whether to do so. So, for example, they must act rationally, must not abuse their powers and may be required to respect any legitimate expectation which they have created. If they fail to do so, the taxpayer may seek relief in judicial review proceedings. This judgment is not the occasion to explore in detail the practical implications of these principles. Ms McCarthy accepts that, depending on the circumstances, a deliberate decision not to assess promptly might amount to irrationality which might in turn provide the basis for a remedy. In our view, the interpretation of the statutory regime should not be distorted to try to deal with issues which ought properly to be addressed by principles of public law enforceable in judicial review proceedings in the usual way. On the other hand, the fact that the application of these principles would proceed by reference to the state of mind of the officer or the Board in line with the approach in the National Association of Health Stores and Bancoult cases lends further support to the view that section 29(1) is intended to operate generally with that focus.
85. Turning back to the facts of this case, Mr Ghosh emphasised that the Revenue appreciated from the time when Mr Tooth’s self-assessment was first reviewed in 2009 that they would wish to challenge it and submitted that Mr Williams was in practice just the recipient of this previously formed evaluation and could not be said to have made any discovery himself. If this were a correct view of the facts, it seems to us that an officer of the Board other than Mr Williams would have made the relevant discovery in 2009 and that by operation of section 2(4) of the CRCA 2005 (see para 79 above) it would have been open to Mr Williams to make an assessment under section 29(1) on the basis of it, provided the conditions referred to in section 29(3) were satisfied. But then it might well have been the position, as the Court of Appeal thought, that the Revenue was precluded from relying on such an analysis because of the way it had pleaded its case.
86. However, the FtT found as a fact that Mr Williams did himself make a discovery of the under-assessment to tax. The FtT correctly directed itself that it was his state of mind which mattered for the purposes of section 29(1), and not the collective knowledge which might reside elsewhere within the Revenue’s organisation. There was sufficient evidence before the FtT on which it was entitled to make its finding. The evidence was capable of supporting the view that Mr Williams had taken responsibility for dealing with Mr Tooth’s file and that, albeit by relying on advice and submissions from others, he had formed his own opinion that the assessment of tax in his return was insufficient. Mr Tooth did not bring any Edwards v Bairstow  AC 14 type challenge to the FtT’s decision on appeal, based on insufficiency of evidence; rather, his challenge was based on legal points which we have found to be unsustainable. Therefore, if the Revenue’s appeal to this court had depended solely on the discovery issue, we would have been disposed to allow it. However, it does not.
87. For the reasons we have given, we would dismiss the Revenue’s appeal seeking to uphold the validity of the discovery assessment made in respect of Mr Tooth. Mr Tooth does not fall within the scope of the condition set out in section 29(4) of the TMA. The situation mentioned in section 29(1) was not brought about deliberately by him.