United Kingdom Supreme Court

You are here: BAILII >> Databases >> United Kingdom Supreme Court >> Harpur Trust v Brazel [2022] UKSC 21 (20 July 2022)
URL: http://www.bailii.org/uk/cases/UKSC/2022/21.html
Cite as: [2023] 2 All ER 113, [2022] ICR 1380, [2022] IRLR 867, [2023] 1 CMLR 18, [2022] UKSC 21

[New search] [Printable PDF version] [Buy ICLR report: [2022] ICR 1380] [Help]

Press Summary Issues & Facts

Hilary Term

[2022] UKSC 21

On appeal from: [2019] EWCA Civ 1402

JUDGMENT

Harpur Trust (Appellant)

Brazel (Respondent)

before

Lord Hodge, Deputy President

Lord Briggs

Lady Arden

Lord Burrows

Lady Rose

JUDGMENT GIVEN ON

20 July 2022

Heard on 9 November 2021

Appellant

Caspar Glyn QC

Nathan Roberts

Catherine Meenan

(Instructed by VWV Solicitors LLP (Bristol))

Respondent

Mathew Gullick QC

Lachlan Wilson

Naomi Webber

(Instructed by Hopkins Solicitors LLP (Nottingham))

Intervener (UNISON)

Michael Ford QC

Mathew Purchase QC

(Instructed by UNISON Legal Services (London))

lady rose and lady arden (with whom Lord Hodge, Lord Briggs and Lord Burrows agree):

1. Overview of the issue of law arising on this appeal

This appeal raises an important issue about the statutory leave requirement for part-time workers who may also be described as part-year workers, namely workers who work for varying hours during only certain weeks of the year but have a continuing contract throughout that year. These workers neither work the full number of hours worked by full time workers nor the full number of weeks worked by part-time workers. Their work is irregular. The issue is whether their leave entitlement is calculated on the same principle, proportionally, as full-time employees (which would mean that the weeks that they do not work reduce their entitlement) or whether their leave must be calculated ignoring those weeks. The latter would leave them with an entitlement which proportionally exceeds that of other employees. Nonetheless, the Court of Appeal held that the proper construction of the domestic law led to that result and further that such a construction was consistent with the applicable EU law.

Prior to the UK’s withdrawal from the European Union, the statutory leave requirement was governed by the EU Council Directive 2003/88/EC generally referred to as the Working Time Directive (“the WTD”), and its predecessor, which were implemented in the United Kingdom by, in particular the Working Time Regulations 1998 (SI 1998/1833) (“the WTR”), which also conferred additional rights. The WTD was adopted pursuant to article 137 of the Treaty Establishing the European Community (“article 137 EC”), now article 153 of the Treaty on the Functioning of the European Union. Article 137(2) EC empowered the Council to adopt directives setting minimum requirements for working conditions for gradual implementation by the member states. The obligation to implement the WTD and its predecessor was satisfied in the United Kingdom by the WTR. The WTR, as amended (inter alia) in 2007, provide that as from 1 April 2009 workers are entitled to 5.6 weeks’ leave (or 28 days, whichever is less) rather than the four weeks provided for by the WTD. The additional 1.6 weeks is conferred only by UK law. The WTR were made under section 2(2) of the European Communities Act 1972 and they are preserved in UK law by section 2 of the European Union (Withdrawal) Act 2018 (“the Withdrawal Act”). They are therefore “retained EU Law” as defined by section 6(7) of that Act. By section 6(3) of the same Act, retained case law (that is, retained domestic and EU case law) continues to apply to any question as to the meaning or effect of retained EU law (see section 5(2), section 6(3) and section 6(7) of the Withdrawal Act).

The respondent, Mrs Brazel, is a visiting music teacher at a school run by the appellant, the Harpur Trust. The Harpur Trust accept that Mrs Brazel is a “worker” within the meaning of the WTR as they applied during the period covered by this appeal. She is therefore entitled to 5.6 “weeks” of paid annual leave in the leave year. The arrangement between the parties is that Mrs Brazel takes her annual leave during the school holidays, when she is not required to give lessons. That is in accordance with the decision of this Court in Russell v Transocean International Resources Ltd [2011] UKSC 57; [2012] ICR 188. There the employee was an offshore oil worker who worked two weeks on a rig and then had two weeks’ break onshore. It was held that an employer could require an employee to take leave in a break period.

The Harpur Trust contend that a part-year worker’s leave entitlement must be prorated further to take account of the weeks not worked. So, the issue is essentially one of statutory interpretation, and we set out the relevant provisions below. The Harpur Trust argue that the domestic provisions prescribing how to calculate holiday entitlement and holiday pay must be interpreted so as to comply with what it refers to as “the conformity principle”. That principle, they say, emerges from the case law of the Court of Justice of the European Union (“the CJEU”) interpreting the provisions of the WTD. The amount of annual leave should, in accordance with the conformity principle, reflect the amount of work that Mrs Brazel actually performs during the annual leave year.

2. The facts and the proceedings below

The Harpur Trust run Bedford Girls School. Mrs Brazel started working at the school in September 2002. Mrs Brazel teaches pupils who want to learn to play the saxophone or clarinet. During the school terms, Mrs Brazel works different hours each week, depending on how many pupils need lessons in her instruments. She usually teaches between ten and 15 hours per week during term time but some weeks much less. During the school holidays she does not teach any lessons to the school pupils and is not required to work. She is paid only for the hours that she actually teaches in term time.

Mrs Brazel is currently employed under an employment contract dated 11 April 2011, and her claims for unauthorised deductions from her pay relate to periods between 1 January 2011 and June 2016. The schedule to her contract of employment sets out the role of the visiting music teacher. Her role includes teaching pupils, from complete beginners to those of a high standard, preparing them for exams and for concerts, festivals and competitions where appropriate. The contract provides that requirements for her services will depend upon a varying level of demand for individual personal tuition in her instruments. There are no minimum hours of work guaranteed to Mrs Brazel and she has no normal hours of work. Mrs Brazel’s pay during the period with which we are concerned was £29.50 per hour and she was paid in arrears at the end of each month.

The contract provides at clause 27 that the annual leave year runs from 1 September to 31 August and that during the leave year Mrs Brazel is entitled to 5.6 weeks’ paid leave. That leave must be taken during the normal school holidays or at such other times as are convenient for the school. Mrs Brazel has always been treated as having taken her annual leave entitlement in three equal tranches in the winter, spring and summer school holidays, that is to say, 1.87 weeks of each school holiday was treated as annual leave for which Mrs Brazel was entitled to be paid. Unused leave entitlement may not be carried forward to a subsequent leave year and there is no pay in lieu of unused leave except on termination of her employment. The contract confirms that there are no collective agreements which directly affect her terms and conditions: clause 52.

Before September 2011, Mrs Brazel’s pay for the 1.87 weeks she was treated as taking during each school holiday was determined in accordance with section 224 of the Employment Rights Act 1996 (“the 1996 Act”), as is required by regulation 16 of the WTR which incorporates section 224 for this purpose. Section 224 defined “a week’s pay” for this and several other purposes as the amount of Mrs Brazel’s average weekly remuneration in the period of 12 weeks ending with the start of her leave period, ignoring any weeks in which she did not receive any remuneration. The Harpur Trust therefore worked out how much Mrs Brazel had been paid during the twelve term-time weeks prior to the school holiday, divided that total by 12 and paid her 1.87 times that weekly average.

As from September 2011 the Harpur Trust changed the calculation. Mrs Brazel was still treated as taking her annual leave entitlement in three equal tranches. But the Harpur Trust calculated Mrs Brazel’s hours worked at the end of each term, took 12.07% of that figure and paid her the hourly rate for that number of hours. We will call this “the Percentage Method”. The Harpur Trust say that in calculating her leave entitlement in that way, they were following the method recommended by Acas in its guidance booklet Holidays and Holiday Pay for calculating the pay of casual workers. The relevant passage in the booklet states that if a member of staff works on a casual basis or very irregular hours it is “often easiest” to calculate holiday entitlement that accrues as hours are worked. 12.07% is the proportion that 5.6 weeks of annual leave bears to the total working year. The working year is the whole year (52 weeks) minus the annual leave (5.6 weeks) and so 46.4 weeks. 5.6 weeks is 12.07% of 46.4 weeks. The Harpur Trust therefore treated Mrs Brazel as entitled to 12.07% of her total pay for the term.

The relevant part of the guidance issued by Acas has now been rewritten. Furthermore, in 2020, subsequent to the decision of the Court of Appeal in this case, the Department for Business, Energy and Industrial Strategy issued guidance (see Holiday Pay - Guidance on calculating holiday pay for workers without fixed hours or pay) (“the BEIS Guidance”). There is a separate section in this document dealing with term-time and part-year workers, including those who, like Mrs Brazel, only receive pay during the periods when they are working and not during the non-working periods. The BEIS Guidance reflects the decision of the Court of Appeal in this case. The employer should not: (1) include in the holiday reference period any whole week in which no pay was received, or (2) apply the Percentage Method. The example given in the BEIS Guidance helpfully sets out the effect of the decision of the Court of Appeal:

“a part-time music teacher has a zero-hours contract entitling them to 5.6 weeks’ annual leave. They have a term-time contract meaning they work 32 weeks per year but remain in employment for the full year. They must take their 5.6 weeks of annual leave during the school holidays. They should therefore be paid for 5.6 weeks of leave taken at some point during the school holidays. The school breaks up for summer holidays on Friday 25 July and the teacher decides to take a two-week paid holiday in mid-August before school returns on 10 September. The employer should therefore take an average of the teacher’s pay rate over the last 52 weeks in which they worked, starting with the last week at the end of the summer term and omitting any other periods of school holiday in which the teacher was not paid.”

This example in the BEIS Guidance refers to a 52 week reference period. That is because, as a result of amendments to the WTR introduced with effect from 6 April 2020, the reference period applicable for calculating the average week’s pay due for statutory leave was increased from the 12-week period that is set in section 224 of the 1996 Act to 52 weeks (see the Employment Rights (Employment Particulars and Paid Annual Leave) (Amendment) Regulations 2018 (SI No 1378) Pt 3, reg 10(3)(b)).

The difference in pay resulting from the two methods used before and after September 2011 can be shown by using the school year 2012-2013 and the annual leave that Mrs Brazel was treated as having taken during the Easter school holidays in April 2013 as an example. That spring term was made up of ten working weeks running from 7 January to 18 March 2013 (ignoring half term breaks during which she received no remuneration). During those weeks Mrs Brazel worked different hours, from a minimum of 10.5 hours in the first week of term to a maximum of 14 hours each in two of the term weeks. She was paid at a constant rate of £29.50 per hour.

The pay to which Mrs Brazel claims she is entitled would be worked out as follows. As we have said, the assumption made is that Mrs Brazel is entitled to be paid for 1.87 weeks in the Easter holidays. She worked 127 hours over the whole spring term. But that term was only ten weeks long and section 224 requires a 12-week reference period. One must therefore add in the hours she worked in the last two weeks of the Autumn term 2012 to make up the 12 weeks. She worked 22.5 hours in the last two weeks of the Autumn term so her total number of hours in the reference period was 149.50. Multiplied by the hourly rate of £29.50, that makes the total pay received in the preceding 12 weeks £4,410.25. Dividing that by 12, one arrives at the average week’s pay of £367.52. Her pay entitlement for the 1.87 weeks’ leave she took during the Easter holiday 2013 was therefore £687.26. We shall call this method the “Calendar Week Method”.

According to the method now adopted by the Harpur Trust, one takes the total number of hours worked by Mrs Brazel during the spring term, that was 127 hours. 12.07% of that is 15.33 hours. The Trust multiplied that number of hours by the hourly rate of pay of £29.50 to arrive at £452.20 for her pay for the annual leave she was treated as taking during the Easter holiday. Under this method, the Harpur Trust say, the leave requirement accrues in proportion to the time the worker works so that this method, unlike the Calendar Week Method, is compliant with the conformity principle.

Mrs Brazel brought a complaint before the Employment Tribunal under Part II of the 1996 Act for unlawful deductions from her wages by underpayment of her entitlement to holiday pay. On 15 January 2017, the Employment Tribunal at Bury St Edmunds dismissed her claims. Broadly, the tribunal accepted that Mrs Brazel’s holiday pay needed to be pro-rated to reflect the fact that she only worked during term time rather than the whole working year of 46.4 weeks. The Employment Appeal Tribunal (HHJ Barklem sitting alone) allowed Mrs Brazel’s appeal, holding that there was no justification for departing from the clear statutory wording that she should be paid a week’s pay for each of the 5.6 weeks’ leave to which she was entitled, calculated using the average of the 12 preceding weeks.

The Court of Appeal (Underhill, Hamblen and Moylan LJJ) dismissed the Harpur Trust’s appeal: see [2019] EWCA Civ 1402; [2020] ICR 584. They rejected reliance on the conformity principle which we explained earlier (para 4 above). They recognised that the Calendar Week Method put Mrs Brazel in a more favourable position than some full-time workers, in the sense that the amount of pay she received for her annual leave worked out to be a higher percentage of the total pay she received over the year than would apply for a worker who worked all 46.4 weeks of the year. They held that that was not sufficient reason to justify departing from the statutory scheme.

The Harpur Trust now appeal with the permission of this court granted on 19 June 2020.

3. The legislative framework

The WTD derives from EU Working Time Directive 1993/104, which was introduced under former article 118a of the Treaty Establishing the European Community (now article 153 of the Treaty on the Functioning of the EU).

The recitals to, and article 1 of, the WTD set out the scope of the WTD. Recital (5) of the WTD provides:

“All workers should have adequate rest periods. The concept of ‘rest’ must be expressed in units of time, ie in days, hours and/or fractions thereof. Community workers must be granted minimum daily, weekly and annual periods of rest and adequate breaks. It is also necessary in this context to place a maximum limit on weekly working hours.”

Article 1 defines the purpose and scope of the WTD, and makes it clear that the WTD only provides for a minimum leave requirement:

“1. This Directive lays down minimum safety and health requirements for the organisation of working time.

2. This Directive applies to:

(a) minimum periods of … annual leave …”

Article 7 of the WTD is conceptually very important because it provides for the right to four weeks’ paid annual leave (increased by the UK as explained below), and that the worker must take the leave as a period of rest away from work rather than working through his leave period and receiving more pay in lieu of leave:

“1. Member states shall take the measures necessary to ensure that every worker is entitled to paid annual leave of at least four weeks in accordance with the conditions for entitlement to, and granting of, such leave laid down by national legislation and/or practice.

2. The minimum period of paid annual leave may not be replaced by an allowance in lieu, except where the employment relationship is terminated.”

Article 15 of the WTD provides:

“This Directive shall not affect member states’ right to apply or introduce laws, regulations or administrative provisions more favourable to the protection of the safety and health of workers or to facilitate or permit the application of collective agreements or agreements concluded between the two sides of industry which are more favourable to the protection of the safety and health of workers.”

Turning to the WTR, these were designed to implement the WTD, but since the WTD permitted the member state to introduce more favourable options for employees, consideration of the WTD alone cannot resolve the issue of construction raised by this appeal. The court needs to search for indications as to whether the WTR were designed to implement the minimum requirements of the WTD or to confer more favourable benefits on employees. If the text points in the latter direction, then, contrary to the submission of the Harpur Trust, there can be no question of the WTR having breached the obligations imposed by the WTD under EU law.

There are many definitions in regulation 2 to the WTR, including the definition of the word “day” as a period of 24 hours beginning at midnight. Neither “year” nor “week” is defined, but, according to regulation 2(2), where there is no definition in the WTR, the meanings used in the corresponding provisions of the WTD apply.

Regulation 13 is headed “Entitlement to annual leave”. It provides for a worker to be entitled to four weeks’ annual leave in each leave year. As far as relevant, it provides:

“(1) Subject to paragraph (5), a worker is entitled to four weeks’ annual leave in each leave year …

(5) Where the date on which a worker’s employment begins is later than the date on which (by virtue of a relevant agreement) his first leave year begins, the leave to which he is entitled in that leave year is a proportion of the period applicable under paragraph (1) equal to the proportion of that leave year remaining on the date on which his employment begins.”

The WTR were amended, with effect from 1 October 2007, by the insertion of regulation 13A(1), which provides for workers to have an additional 1.6 weeks’ leave unless the worker already has an entitlement to such additional leave by contract (Working Time (Amendment) Regulations 2007 (SI 2007/2079)). As far as relevant, regulation 13A of the WTR provides:

“(1) Subject to … paragraphs (3) and (5), a worker is entitled in each leave year to a period of additional leave determined in accordance with paragraph (2).

(2) The period of additional leave to which a worker is entitled under paragraph (1) is –

…

(e) in any leave year beginning on or after 1 April 2009, 1.6 weeks.

(3) The aggregate entitlement provided for in paragraph (2) and regulation 13(1) is subject to a maximum of 28 days …

(5) Where the date on which a worker’s employment begins is later than the date on which (by virtue of a relevant agreement) his first leave year begins, the additional leave to which he is entitled in that leave year is a proportion of the period applicable under paragraph (2) equal to the proportion of that leave year remaining on the date on which his employment begins.”

Regulations 13 and 13A thus deal with the duration of leave. Regulations 13(5) and 13A(5) provide for the time-apportionment of the leave to which an employee is entitled where they start employment in the leave year.

Regulation 14 (excluding amendments made subsequent to the periods in issue in these proceedings) provides for the calculation of outstanding statutory leave when a worker is dismissed during a leave year, and it is important because it is clearly applicable to both full-time and part-time workers:

“14. Compensation related to entitlement to leave

(1) This regulation applies where –

(a) a worker’s employment is terminated during the course of his leave year, and

(b) on the date on which the termination takes effect (‘the termination date’), the proportion he has taken of the leave to which he is entitled in the leave year under regulation 13 and regulation 13A differs from the proportion of the leave year which has expired.

(2) Where the proportion of leave taken by the worker is less than the proportion of the leave year which has expired, his employer shall make him a payment in lieu of leave in accordance with paragraph (3).

(3) The payment due under paragraph (2) shall be –

(a) such sum as may be provided for for the purposes of this regulation in a relevant agreement, or

(b) where there are no provisions of a relevant agreement which apply, a sum equal to the amount that would be due to the worker under regulation 16 in respect of a period of leave determined according to the formula –

(A × B) - C

Where –

A is the period of leave to which the worker is entitled under regulation 13 and regulation 13A;

B is the proportion of the worker’s leave year which expired before the termination date, and

C is the period of leave taken by the worker between the start of the leave year and the termination date …”

Regulation 15A stipulates that, in the first year of employment, leave is limited to the amount that is “deemed to be accrued” but this is apportionment across the first year not on the basis of work done but on the basis of time. That method of accrual (ie apportionment) uses the month as the unit of time and takes no account of when the work is actually performed:

“15A. Leave during the first year of employment

(1) During the first year of his employment, the amount of leave a worker may take at any time in exercise of his entitlement under regulation 13 or regulation 13A is limited to the amount which is deemed to have accrued in his case at that time under paragraph (2) or (2A), as modified under paragraph (3) in a case where that paragraph applies, less the amount of leave (if any) that he has already taken during that year …

(2A) Except where paragraph (2) applies, for the purposes of paragraph (1), leave is deemed to accrue over the course of the worker’s first year of employment, at the rate of one-twelfth of the amount specified in regulation 13(1) and regulation 13A(2), subject to the limit contained in regulation 13A(3), on the first day of each month of that year.”

Regulation 16 of the WTR provides for the entitlement to pay during statutory leave, and the mechanism for determining the amount of a “week’s pay”. The version of regulation 16 as it applied during the period of Mrs Brazel’s claim provided, so far as material:

“16. Payment in respect of periods of leave

(1) A worker is entitled to be paid in respect of any period of annual leave to which he is entitled under regulation 13 and regulation 13A at the rate of a week’s pay in respect of each week’s leave.

(2) Sections 221 to 224 of the 1996 Act shall apply for the purpose of determining the amount of a week’s pay for the purposes of this regulation, subject to the modifications set out in paragraph (3).”

Section 224 of the 1996 Act, to which regulation 16 refers, is the relevant provision for calculating the week’s pay where there are, as in this case, no normal working hours. Section 224 provides:

“224. Employments with no normal working hours

(1) This section applies where there are no normal working hours for the employee when employed under the contract of employment in force on the calculation date.

(2) The amount of a week’s pay is the amount of the employee’s average weekly remuneration in the period of twelve weeks ending –

(a) where the calculation date is the last day of a week, with that week, and

(b) otherwise, with the last complete week before the calculation date.

(3) In arriving at the average weekly remuneration no account shall be taken of a week in which no remuneration was payable by the employer to the employee and remuneration in earlier weeks shall be brought in so as to bring up to twelve the number of weeks of which account is taken.”

The BEIS Guidance is based on this provision, although as we have mentioned, that Guidance incorporates the modification made to regulation 16 by the Employment Rights (Employment Particulars and Paid Annual Leave) (Amendment) Regulations 2018/1378, regulation 10. That changed regulation 16 with effect from 6 April 2020, ie subsequently to the period covered by this claim, so that from that date section 224 applied as if the reference to 12 weeks was a reference to 52 weeks.

The defining feature of section 224(3) is that the employee is employed for the week in question but remuneration is not payable for that week. Such weeks are then ignored in calculating the amount of a week’s pay due for each week of the statutory leave requirement. In order to extend for 12 weeks, the reference period used to calculate pay must be based on earlier weeks when the employee was both employed and received remuneration. Section 229 of the 1996 Act is not expressly referred to in regulation 16 of the WTR but it contains the following supplementary provision:

“(2) Where under this Chapter account is to be taken of remuneration or other payments for a period which does not coincide with the periods for which the remuneration or other payments are calculated, the remuneration or other payments shall be apportioned in such manner as may be just.”

4. The pro-rating of the annual leave requirement under EU law

The Harpur Trust contend that the governing principle is that the leave requirement of full-time and part-year employees must be adjusted according to the amount of time spent working (the conformity principle described in para 4 above) so that leave depends on time spent working. Therefore, a full-time employee, working each week, will get the full 5.6 weeks’ paid leave and part-time employees who work every week but only for part of each week will have their leave entitlement adjusted pro rata to that of full-time employees. It follows, argue the Harpur Trust, that the leave requirement for part-year employees should be pro-rated to allow for the weeks in which they are not required to work.

The principal authority of the CJEU on which the Harpur Trust rely was decided after the decision of the Court of Appeal in this case, QH v Varhoven kasatsionen sad na Republika Bulgaria (C-762/18) [2020] EU:C:2020:504 (“QH”). It concerned the leave entitlement of workers who had been unlawfully dismissed and then reinstated in their employment following the annulment of the dismissal by the domestic court. The question referred to the CJEU was whether a national law which precluded any entitlement to annual leave accruing during the period between dismissal and reinstatement was contrary to article 7 of the WTD. The CJEU’s starting point in QH was the importance of article 7 of the WTD and the right to leave:

“53. … in the first place, it must be borne in mind that, as is clear from the very wording of Article 7(1) of Directive 2003/88, every worker is entitled to paid annual leave of at least four weeks. That right to paid annual leave must be regarded as a particularly important principle of EU social law …

55. In addition, as the court has previously held, the right to paid annual leave cannot be interpreted restrictively (judgment of 30 June 2016, Sobczyszyn (C‑178/15) EU:C:2016:502, para 21 and the case law cited).

57. In the second place, it should be borne in mind that, according to the court’s settled case law, the right to paid annual leave, as laid down in article 7 of Directive 2003/88, has the dual purpose of enabling the worker both to rest from carrying out the work he or she is required to do under his or her contract of employment and to enjoy a period of relaxation and leisure (judgment of 20 July 2016, Maschek (C‑341/15) EU:C:2016:576, para 34 and the case law cited).

The CJEU went on in QH to express what the Harpur Trust have termed “the conformity principle” in the following way, although that is not a phrase that the CJEU itself used:

“58. That purpose, which distinguishes paid annual leave from other types of leave having different purposes, is based on the premiss that the worker actually worked during the reference period. The objective of allowing the worker to rest presupposes that the worker has been engaged in activities which justify, for the protection of his or her safety and health, as provided for in Directive 2003/88, his or her being given a period of rest, relaxation and leisure. Accordingly, entitlement to paid annual leave must, in principle, be determined by reference to the periods of actual work completed under the employment contract (judgment of 4 October 2018, Dicu (C‑12/17) EU:C:2018:799, para 28 and the case law cited).”

The CJEU proceeded, however, to hold that the conformity principle was not applicable in certain situations, in particular if and to the extent that the worker had been prevented from working during the reference period because she had been unlawfully dismissed or, as the CJEU had decided in earlier cases, because she had been absent on sick leave. The CJEU held:

“59. Nonetheless, in certain specific situations in which the worker is incapable of carrying out his or her duties, the right to paid annual leave cannot be made subject by a member state to a condition that the worker has actually worked (see, to that effect, judgment of 24 January 2012, Dominguez (C‑282/10) EU:C:2012:33, para 20 and the case law cited).

60. The same applies, in particular, with regard to workers who are absent from work on sick leave during the reference period. As is clear from the court’s case law, with regard to entitlement to paid annual leave, workers who are absent from work on sick leave during the reference period are to be treated in the same way as those who have in fact worked during that period (judgment of 4 October 2018, Dicu (C‑12/17) EU:C:2018:799, para 29 and the case law cited).

61. Thus, according to article 7 of Directive 2003/88, any worker on sick leave during the reference period cannot have his or her entitlement to at least four weeks’ paid annual leave affected (see, to that effect, judgment of 24 January 2012, Dominguez (C-282/10) EU:C:2012:33, para 30).

62. In that context, the court has held that article 7(1) of Directive 2003/88 must be interpreted as precluding national legislation or practices under which the right to paid annual leave is extinguished at the end of the leave year and/or of a carry-over period laid down by national law where the worker has been on sick leave, for the whole or part of the leave year, and therefore has not actually had the opportunity to exercise that right (judgment of 30 June 2016, Sobczyszyn (C‑178/15) EU:C:2016:502, para 24 and the case law cited).

63. Under the case law set out above it cannot be accepted that a worker’s right to a minimum paid annual leave, guaranteed by European Union law, may be reduced where the worker could not fulfil his or her obligation to work during the reference period due to an illness (judgment of 19 September 2013, Review Commission v Strack (C‑579/12 RX‑II) EU:C:2013:570, para 34 and the case law cited).

64. Thus, Directive 2003/88 does not allow member states either to exclude the existence of the right to paid annual leave or to provide for the right to paid annual leave of a worker, who was prevented from exercising that right, to be lost at the end of the reference period and/or of a carry-over period fixed by national law (judgment of 29 November 2017, King (C‑214/16) EU:C:2017:914, para 51 and the case law cited).

65. It must therefore be ascertained whether the principles deriving from the case law on the right to paid annual leave of a worker who, because of sickness, has been unable to exercise his or her right to such leave during the reference period and/or the carry-over period fixed by national law may be transposed, mutatis mutandis, to a situation, such as that at issue in the main proceedings in the present cases, in which a worker who has been unlawfully dismissed then reinstated in his or her employment, in accordance with national law, following the annulment of his or her dismissal by a decision of a court, has not, during the period between the date of the unlawful dismissal and the date of reinstatement, actually carried out work for his or her employer.

66. In that regard, it should be observed that, in order to derogate, as regards workers absent from work owing to sickness, from the principle that the rights to annual leave must be determined by reference to periods of actual work, the court has relied on the fact that incapacity for work owing to sickness is, as a rule, not foreseeable and beyond the worker’s control (see in particular, to that effect, judgment of 4 October 2018, Dicu (C‑12/17) EU:C:2018:799, para 32 and the case law cited).

67. It must be stated that, like incapacity for work owing to sickness, the fact that a worker was deprived of the opportunity to work owing to dismissal that was subsequently held to be unlawful is, as a rule, not foreseeable and beyond the worker’s control.

68. As the Advocate General observed in point 48 of his Opinion, the fact that the worker concerned has not, during the period between the date of his or her unlawful dismissal and the date of her reinstatement, in accordance with national law, following the annulment of that dismissal by a decision of a court, actually carried out work for his or her employer is the consequence of the latter’s actions that led to the unlawful dismissal, without which the worker would have been in a position to work during that period and to exercise his or her right to annual leave.

69. Therefore, in a situation such as that at issue in the main proceedings in the present cases, the period between the date of the unlawful dismissal and the date of the worker’s reinstatement, in accordance with national law, following the annulment of that dismissal by a decision of a court, must be treated as a period of actual work for the purpose of determining the rights to paid annual leave.

70. Accordingly, the court’s case law relating to the right to paid annual leave of a worker who, owing to sickness, has not been in a position to exercise his or her right to such leave during the reference period and/or the carry-over period fixed by national law may be transposed, mutatis mutandis, to a situation, such as that at issue in the main proceedings in each of the present cases, in which a worker who has been unlawfully dismissed and subsequently reinstated, in accordance with national law, following the annulment of the dismissal by a decision of a court, has not, during the period between the date of that dismissal and the date of his or her reinstatement, actually carried out work for his or her employer.”

The Harpur Trust refer to the principle of pro rata temporis as a “parallel” principle. That principle is a settled principle of EU law and thus of the employment law of the UK. It is set out in Council Directive 97/81/EC of 15 December 1997 concerning the Framework Agreement on part-time work concluded between the European Trade Union Confederation (“ETUC”) and others (“the Part-time Working Directive”). This was implemented in the UK by The Part-time Workers (Prevention of Less Favourable Treatment) Regulations 2000 (SI 2000/1551), which came into force on 1 July 2000. The essential principle of the Framework Agreement, and thus of the Part-time Working Directive, is that employers may not discriminate against part-time workers. This is expressed in clause 4, which reads in material part:

“1. In respect of employment conditions, part-time workers shall not be treated in a less favourable manner than comparable full-time workers solely because they work part time unless different treatment is justified on objective grounds.

2. Where appropriate, the principle of pro rata temporis shall apply.”

But, again, the pro rata temporis principle thus only applies “where appropriate”. Clause 4(2) permits apportionment on some different basis or indeed that there should be no apportionment if that were appropriate. Clause 4 does not in any event preclude legislation which discriminates in favour of part-time employees.

QH concerning unlawful dismissal and cases such as Dominguez cited in Dicu concerning periods of sickness are exceptions to the conformity principle. An example where the CJEU has refused to depart from the conformity principle is Hein v Albert Holzkamm Gmbh & Co KG (Case C-185/17) [2019] 2 CMLR 19. The issue was whether German law complied with the WTD when it enabled a collective agreement to provide for periods of short-time working imposed by the employer to be taken into account in determining the employee’s normal remuneration for the purpose of the leave requirement. The CJEU held that it had first to determine the duration of leave. It held as regards that aspect of the leave entitlement that weeks in which no work was done had to be deducted when calculating the worker’s leave entitlement. But it did not affect the determination of the amount to be paid. That was to be “normal” pay. However, the CJEU went on to make a qualification to this. A member state could decide to make the leave requirement more generous to workers because of article 15 of the WTD.

A situation where the conformity principle applied to increase the amount of holiday entitlement is where a part-time worker takes leave based on her then work but then increases her hours: see, Greenfield v The Care Bureau Ltd (Case C-219/14) [2016] ICR 161. Mrs Greenfield took leave during a holiday year which exhausted her (then) statutory leave, but she then increased her hours of her work. The CJEU held that her leave entitlement had to be recalculated to take account of these further hours. The leave entitlement would then be prorated according to the periods when a different number of hours was worked.

Turning to the present appeal, Underhill LJ in the Court of Appeal was minded to accept that the conformity principle was part of EU law, and he termed it “the accrual approach” which we understand to mean the same as the conformity principle: see para 66. In the light of the decisions which we have cited, we consider that he was right to accept that the CJEU had interpreted the WTD as adopting the conformity principle, namely that subject to exceptions such as those recognised in QH and Dominguez, workers only have an entitlement to paid annual leave in proportion to the time they work. Underhill LJ did not need to express a final view because he went on to decide that the domestic implementation in the WTR, properly interpreted, gave the part-year worker four weeks’ leave without providing for any adjustment to reflect the number of weeks worked. His cautious approach was justified because the CJEU had not dealt with a case such as the present.

One of the strongest arguments against this interpretation of the WTR is that it is absurd for the legislature to allow part-year workers to obtain the right to an amount of leave which is disproportionate in relation to that to which other workers are entitled. We examine below the various ways in which the Harpur Trust contend that the WTR can be interpreted so as to avoid this result and answer the argument based on absurdity.

A necessary step in the argument of the Harpur Trust is the principle of EU law that domestic legislation which implements EU law should be interpreted by domestic courts so as to conform with the EU legislation it is designed to implement. This principle is often called the Marleasing principle, after the case of Marleasing SA v La Comercial Internacional de Alimentatión SA [1990] ECR I-4135, a decision of the European Court of Justice, which preceded the CJEU. But EU law only requires UK legislation to be so interpreted if that result is consistent with domestic rules of statutory interpretation: see the decision of the House of Lords in Fleming v Revenue and Customs Comrs [2008] UKHL 2; [2008] 1 WLR 195, and Hein at para 50.

The expression “week” is defined for the purposes of domestic law by section 235(1) of the 1996 Act as, so far as relevant, a period of seven days (section 235(1)). That meaning is not incorporated into the WTR. It is clear from regulation 2(2) of the WTR (see above para 24) that Parliament recognised that in general at least the meaning of “week” for the purposes of the WTR would be that given to the word by the CJEU in interpreting that word in the WTD.

Our task is, however, to construe the WTR and the meaning of “week” used there. In our judgment, the conformity principle enunciated by the CJEU cannot operate so as to allow a construction of the provisions so that they apply to part year workers in the way that the Harpur Trust contend.

It is not enough for the Harpur Trust to show that the Court of Appeal’s interpretation leads to the part-year worker receiving disproportionately more paid leave than other workers. Given that, as we have said, a more generous entitlement for part-year workers would not infringe either the WTD or the Part-time Working Directive so as potentially to engage the Marleasing principle, the issue is one of statutory interpretation. It is not sufficient simply to assert that “the conformity principle offers a simple, principled solution to all working patterns” (para 98 of the written case of the Harpur Trust) unless they can show that the provisions are drafted on the basis that that is the principle underlying the provisions and so should be applied.

The Harpur Trust point to the fact that regulation 13, which stipulates that there shall be four weeks’ leave must be subject to adjustment if the worker works part-time. As a matter of common sense, the expression “week” in regulations 13 and 13A must include fractions of a week where a person works part-time. We agree with that submission to the extent that, as Underhill LJ explained in para 63 of his judgment, a part-time worker who takes four calendar weeks’ leave is in fact only relieved of working on the particular days in those calendar weeks on which they would have worked otherwise. But fractions of a week are still fractions of a calendar week. So, that would not justify an approach which involves taking a fraction of a week which has been derived by making an adjustment for the fact that the worker has only worked for some of the year. Moreover, there is no mechanism for working out the corresponding amount of four weeks’ leave for a part-year worker as opposed to other part-time workers. Furthermore, as we explain below when examining the alternative methods put forward by the Harpur Trust, there may be difficulty in establishing the amount of leave to which a worker is entitled except at the end of the year.

The reference period for entitlement to leave is different from the period which is used for determining the amount of weekly holiday pay, as Hein shows. To determine the worker’s holiday pay, regulation 16(2) of the WTR expressly applies section 224 of the 1996 Act for the purpose of calculating the amount of a week’s pay for the purposes of that regulation. This ensures that part-year workers are treated in the same way as other workers, full or part-time, that is, by ignoring any weeks during the preceding 12-week period when they do not work at all, when calculating their average week’s pay. But there is no provision in regulations 13 and 13A of the WTR for the entitlement to annual leave to be calculated by excluding the weeks where no work is performed.

If the conformity principle were the basis for the entitlement to leave governed by regulations 13 and 13A, there would have to be a provision like section 224. On conventional principles of interpretation, however, the fact that section 224 applies for one purpose, namely for calculating the average week’s pay and not for another, namely for calculating the number of weeks leave, is an indication that there was no similar provision for determining the length of the leave. That conclusion is reinforced by the fact that regulations 13 and 13A contain provisions for apportionment when employment begins part way through the leave year showing that the UK legislature had considered the question of the apportionment of leave and decided that, where it was necessary to do so, it would adopt a time-apportionment basis, not an apportionment on the basis of work actually done.

Moreover, the regulations dealing with leave entitlement and the amount of holiday pay provide for entitlements which are clearly inconsistent with the conformity principle. Regulations 13(5) and 13A(5) prorate the leave entitlement of the new employee according to the time that has yet to elapse in the leave year, not to the amount of work done in the leave year. The same is true of regulation 14, which calculates the outstanding holiday pay due to a worker whose employment is terminated before the end of the leave year. This is done by reference to the time which has elapsed since the start of the leave year, not by reference to the work actually done during that time. No distinction is drawn between full-time, part-time and part-year employees. The court should proceed on the basis that the regulations treat the calculations of the length of leave and holiday pay on a consistent basis, and thus that the correct conclusion is that the WTR do not enable the employer to reduce the leave entitlement of part-year employees still further to ensure that such workers receive an amount of leave which is proportional to that received by other workers.

We therefore conclude that the conformity principle is not the principle underlying the regime established by the WTR and that, on their true interpretation, the WTR do not base entitlement on the conformity principle as regards part-year workers so as to exclude the weeks in which they do not work from the reference period for determining the length of their leave entitlement. The calculation of the entitlement to annual leave is different from the calculation of average pay in this regard. In short, the amount of leave to which a part-year worker under a permanent contract is entitled is not required by EU law to be, and under domestic law is not, prorated to that of a full-time worker.

5. The Harpur Trust’s submissions as to how to apply the WTR

We now address the Harpur Trust’s submissions as to how and why the domestic provisions should be construed to reduce Mrs Brazel’s leave entitlement to reflect the fact that she does not work during the whole working year of 46.4 weeks. In summary, their case is that although the Calendar Week Method results in Mrs Brazel receiving a greater leave entitlement than she would under the Harpur Trust’s current calculation, other workers working other irregular hours patterns will be worse off under the Calendar Week Method than under the Harpur Trust’s method. The Harpur Trust argue that the Court should adopt one of those alternative methods for all workers including Mrs Brazel as a way of ensuring that every worker receives at least their entitlement under the WTD.

(a) The Harpur Trust’s alternative methods

The first of the two alternative methods proposed by the Harpur Trust is the way that they have calculated Mrs Brazel’s annual leave and pay since September 2011.

The Percentage Method works as we have already described in para 9 above: 12.07% is the proportion that annual leave of 5.6 weeks bears to the full working year of 46.4 weeks. The worker is entitled to paid leave for 12.07% of the work that they have actually performed. One therefore simply adds up all the hours or days worked and applies the hourly or daily rate of pay to 12.07% of that number of hours or days. The same amount can be arrived by an arithmetically different route, by applying the 12.07% to the pay received over their annual leave year. The Percentage Method does not therefore require one to work out the week’s pay in accordance with section 224 of the 1996 Act.

The second method is more complicated. We shall refer to this as “the Worked Year Method”. It takes account of the fact that the worker does not work during the school holidays by calculating the proportion of the working year of 46.4 weeks which comprises weeks in which the worker does actually work. Taking a worker with the same work pattern as Mrs Brazel, there are usually about 34 working weeks in the school year ignoring all holidays. That represents 0.73 of the full working year of 46.4 weeks. One then applies that percentage to the 5.6 weeks since that is the annual leave entitlement for someone who works all 46.4 weeks. The annual leave entitlement for this worker is 4.09 weeks of paid leave, that being 0.73 x 5.6. One then works out the average week’s pay according to section 224 of the 1996 Act but the worker’s entitlement over the year is only 4.09 weeks at that rate of pay, not the full 5.6 weeks.

(b) The effect of the alternative methods on part-year workers

The Harpur Trust say that the domestic provisions must be interpreted in a way that chooses one of these methods because one can posit workers who work a different pattern of irregular hours who either get an amount clearly in excess of any reasonable entitlement or who are worse off under the Calendar Week Method than under either of the two alternative methods. Case 1 is the infrequent worker, such as an exam invigilator, who works only three weeks a year but who works 40 hours in each of those three weeks. The Calendar Week Method would appear to entitle such a worker to 5.6 weeks’ paid leave at the average week’s pay which would be her weekly pay during the three weeks she works. It cannot be right, the Harpur Trust argue, that such a worker is entitled to holiday pay almost twice the amount of her actual annual earnings. The two different methods it proposes arrive at a more sensible result. Using the Percentage Method, the invigilator has worked 120 hours and, applying the 12.07% multiplier, is entitled to 14.48 hours of paid annual leave at the hourly rate actually paid. Using the Worked Year Method, one reduces the 5.6 weeks’ leave entitlement to reflect the proportion of the working year that the invigilator actually worked, that is three weeks out of a potential 46.4 weeks which is 6.47%. Since she has worked only 6.47% of the working year, she is entitled to 6.47% of the annual leave entitlement of 5.6 weeks, namely 0.36 of a week. One then calculates the average weekly pay in accordance with section 224 and arrives at the pay for annual leave by applying a multiplier of 0.36 to the average week’s pay so calculated.

Case 2 is the “Transitioning Down” worker (“TD”). TD is paid £100 per day. For the first half of the leave year (26 weeks) she works five days per week and for the second half of the leave year (20.4 weeks) she works one day per week. She takes her leave at the end of the leave year after 46.4 weeks of work. According to the Calendar Week Method, TD’s average weekly pay is based on the preceding 12 weeks during which she worked only one day per week in each of those weeks. Her week’s pay is therefore £100 and her entitlement to paid leave is 5.6 times that, namely £560. This must be wrong, the Harpur Trust argue, because it ignores the fact that she worked full time for the majority of the year. Under the Percentage Method, the employer calculates how many days TD in fact worked over the leave year. This is 150.4 days (26 weeks at five days plus 20.4 weeks at one day). The employer calculates her total pay over the working year, namely 150.4 x £100 = £15,040. She is then entitled to pay for her paid leave of 12.07% of that, namely £1,815.

Under the Worked Year Method, one calculates the average number of days TD has worked per week over the working year. She worked a total of 150.4 days during the course of the whole working year of 46.4 weeks, so an average of 3.24 days in each of those 46.4 working weeks. That 3.24 days is then treated as the “week” in her case. One treats her as entitled to 5.6 weeks of those average weeks amounting to 18.14 days (5.6 x 3.24). The average weekly pay is calculated by applying section 224 to arrive at her weekly pay. That is £100 per week because that is her average pay over the 12 weeks immediately preceding her annual leave. But now one multiplies that weekly rate by 3.24 to arrive at £324 per week, giving a total of £1,814 (being £324 x 5.6). In TD’s case, therefore, the two proposed methods arrive at the same result.

The third hypothetical, Case 3, is the Transitioning Up worker (TU). TU also works each of the 46.4 weeks of the year and is paid £100 per day but she works the first 26 weeks of the leave year at one day a week and then takes her 5.6 weeks of leave. She then returns to work and works the last 20.4 weeks of the leave year at five days per week. On the Calendar Week Method, the Harpur Trust say, TU is only entitled to a total of £560 for her 5.6 weeks leave and the fact that she works full time for the second part of the year does not entitle her either to more leave or to more paid leave.

Under the Percentage Method, again, one looks at the number of days TU has worked over the whole leave year. This is 128 days (that is 26 weeks at one day per week and 20.4 weeks at five days per week). Her total pay for the year is £12,800. She is then entitled to 12.07% of that, namely £1,545. Under the Worked Year Method, one works out the average number of days per week TU has worked across the leave year which is 128 days over 46.4 weeks making an average of 2.76 days worked per week. Her entitlement is therefore 5.6 weeks of paid leave at her weekly pay (calculated in accordance with section 224 at £100) but then multiplied by 2.76 resulting again in an entitlement of £1,545.

The final example relied on by the Harpur Trust is the Restricted Leave Worker (RL). RL works irregular hours over the year and requests ten days of paid leave part way through the year, after working 30 weeks. She has worked a total of 370 hours during the course of those 30 weeks. She is paid £10 per hour. According to the Percentage Method, one works out how many hours she has worked across the year, say, 1,012; applies her hourly rate to that to arrive at £10,120 and applies the 12.07% multiplier to that to arrive at an annual entitlement to holiday pay of £1,221.48. But of course, those annual totals are not known by the time she wants to take her leave at the week 30 mark. The employer can limit her, the Harpur Trust say, to taking only the paid leave that she has already accrued. She has worked an average of 12.33 hours per week over those 30 weeks (that is 370 total hours divided by 30 weeks) and has worked at least some hours on 90 days over those 30 weeks, making an average of 4.11 hours per day (that is 370 hours divided by 90 days). RL’s working week is thus 12.33 hours per week so she is entitled to 5.6 weeks of annual leave, but treating a week as comprising 12.33 hours. RL is therefore entitled to 69.1 hours annual paid leave (that is 5.6 x 12.33).

The 30 weeks already worked by RL represent 65% of the total working year of 46.4 weeks so her entitlement to paid leave at the week 30 mark is 65% of 69.1 hours which is 44.65 hours. In other words, by the week 30 mark, RL has accrued 44.65 hours of paid leave. By requesting ten days’ leave at the week 30 mark, RL is asking for ten times her average working day of 4.11 hours which equals 41.1 hours of leave. That is within her entitlement so far earned of 44.65 so she can properly take that leave.

RL’s employer must then work out how much to pay RL for that leave. Applying the calculation in section 224, one looks back at the previous 12 weeks before she wants to take her leave. During those 12 weeks, let us say she worked a total of 224 hours so that the weekly average hours worked is 18.67 hours per week. RL earned £10 per hour so that her average weekly pay is £186.67. This then needs to be applied, the Harpur Trust say, at a rate of the working week. Since the average number of hours worked per day is 4.11 hours, a request for ten days’ leave amounts to a request for 41.1 hours. Since the average number of hours worked per working week is 12.33, 41.1 hours is equivalent to 3.33 average weeks (that is 41.1 divided by 12.33). The average weekly pay according to section 224 is £186.67. One multiplies that by 3.33 to arrive at £616.01.

However, to pay RL £616.01 would be to overpay her because it happens that she worked more hours on average during the reference period of 12 weeks (224 hours so an average of 18.67 hours per week) than she worked on average over the whole 30-week period since the start of the leave year (370 hours so an average of 12.33 hours per week). To adjust for the fact that the 12 week reference period was not representative of her working pattern over the 30 weeks of the leave year so far elapsed, one must reduce the pay by the percentage that her average working hours over the course of the 30 weeks (12.33 hours per week) bears to the average she in fact worked over the 12-week reference period (18.67 hours per week), that is 67%. Thus, the figure of £616.01 must be reduced so that the actual pay is 67% of that figure, namely £412.70.

The Harpur Trust go on to posit that at the end of the working year RL wishes to take some more paid leave. They then show how the employer can work out how much additional leave RL has earned in the second part of the year and how much she should be paid.

From the description above as to how the Harpur Trust submit an employer would be expected to apply their Percentage Method or the Worked Year Method, we can draw a few conclusions. The first is that the methods proposed are very different from the statutory method set out in the WTR. We agree with Mrs Brazel’s description of these methods as an entirely new scheme for calculating holiday pay entitlement. We would go further and say that in a number of ways the methods are directly contrary to what is set out in the WTR. The Percentage Method does not involve calculating a week’s pay despite the fact that regulation 16 of the WTR mandates that the pay entitlement should be calculated in accordance with section 224, as applied and adapted by regulation 16. The Worked Year Method involves treating the term “week” as meaning something very different from a calendar week of seven days.

Regulation 16 itself reflects a policy choice by Parliament to adopt the definitions of a week’s pay in sections 221 to 224 of the 1996 Act for the purpose of determining the pay due for annual leave. Section 224 is designed to deal with the difficult problem of defining, for a number of different statutory purposes, what should be taken to be a “week’s pay” for a worker whose pay depends on the number of hours worked in a week and who works different hours in each week. There were several ways in which this could be done, but Parliament decided to take a 12-week reference period and simply divide the total pay over those 12 weeks by 12 to arrive at the week’s pay. There is some rough justice in this which is mitigated by the provision that requires the employer to ignore weeks in which no pay is received. There may, however, be weeks in that 12-week period when pay is low. For example, in the final two weeks of the summer term of 2013, Mrs Brazel worked only 4.5 hours and seven hours whereas her hours of work were over ten hours in all but three of the other working weeks in the 2012/2013 academic year. That dip in pay in the last fortnight of the school year brought down her week’s pay used to calculate her pay for her annual leave taken in the summer holidays (£312.21) compared with the week’s pay used for the leave taken in the other two holidays (£367.52 for Easter holiday and £360.15 for Christmas holiday). That is the choice that the legislature has made and to that extent - but only to that extent - the hours worked do affect the pay received.

The assumption in the computation for the hypothetical Restricted Leave worker proposed by the Harpur Trust, namely that she is only allowed to take and be paid for as many days leave as she has so far “accrued” in the first 30 weeks of the year, is not how the regime is supposed to work. It is clear from the case of R (Broadcasting, Entertainment, Cinematographic and Theatre Union) v Secretary of State for Trade and Industry (Case C-173/99) [2001] ICR 1152, (“BECTU”) that a worker is entitled to take all their leave and be paid for it at the start of the leave year if that is their choice and if it does not inconvenience the employer. The idea of leave accruing over the year as and when hours are worked is also inconsistent with regulation 15A that was inserted into the WTR in response to the BECTU ruling. That addresses the potential problem arising if an employee starts in a new job, takes 5.6 weeks’ paid leave after working a few months and then resigns before the end of the first year of employment. Regulation 15A limits the amount of leave to which that worker is entitled to the amount deemed to have accrued and, further, provides that the accrual occurs at a rate of one twelfth of the 5.6 weeks for each month she works. Thus, as Mrs Brazel points out, in this specific situation where there needs to be some reduction in the annual entitlement, there is an express deemed accrual at a specified rate which is not connected to the number of days or hours actually worked. The same point can be made as regards regulation 14 of the WTR which also deals with the situation where a worker’s employment is terminated in the middle of the leave year and she has taken a higher or lower proportion of days leave than the proportion of the leave year that has elapsed as at the date of termination.

The second problem with the two methods proposed by the Harpur Trust is that for a worker working an irregular number of hours per week over the course of the year, the calculations are extremely complicated. The proposed scheme would require all employers and workers to keep detailed records of every hour worked, even if they were not paid at an hourly rate. We note that recital (2) of the WTD refers back to article 137 EC which requires that directives adopted under that power “are to avoid imposing administrative, financial and legal constraints in a way which would hold back the creation and development of small and medium-sized undertakings.”

The Harpur Trust rely on their hypothetical examples as showing that there are workers for whom the WTR computation might mean that they receive less than their entitlement under the WTD. They say, for example, that the Transitioning Up worker described in para 60 above would be worse off under the Calendar Week Method than is permitted and worse off than she would be under their alternative methods. That is why, they argue, the Marleasing principle comes into play so that one of their alternative methods must be adopted to ensure that all workers receive at least their EU law entitlement. We do not accept that submission. Even if the application of the WTR which we consider is correct might result in a non-compliant transposition of the rights in the WTD in respect of some hypothetical workers, that does not mean that the Harpur Trust can rely on that non-compliance to justify a departure from the ordinary meaning of provisions in this case where it is clear that the Calendar Week Method results in Mrs Brazel receiving at least her entitlement under the WTD so that there would be no breach of the WTD in her case: see R (Z) v Hackney London Borough Council [2020] UKSC 40; [2020] 1 WLR 4327, para 114.

The Harpur Trust also argue that the construction upheld by the Court of Appeal leads to an absurd result, that absurdity being that a worker in Mrs Brazel’s position (and in the position of some of the other hypothetical workers posited by the Harpur Trust) receives holiday pay which represents a higher proportion of her annual pay than full time or part time workers who work regular hours. We recognise, of course, that a construction which leads to an absurd result is, in general, unlikely to be what Parliament intended. However, we do not regard any slight favouring of workers with a highly atypical work pattern as being so absurd as to justify the wholesale revision of the statutory scheme which the Harpur Trust’s alternative methods require. We agree with Underhill LJ’s observation as to the odd results produced in extreme cases, such as the exam invigilator in the Harpur Trust’s Case 1. General rules sometimes provide anomalies when applied in untypical cases and it would be unusual for a worker whose services are only required for a few weeks a year to be engaged on a permanent contract unless there was some other good reason to do so: see para 71 of his judgment.

The Harpur Trust argue that any difficulties arising from the uncertainty over the number of hours that are going to be worked and hence the amount of leave that is going to accrue by the year’s end should not be exaggerated. That risk is inherent in an annualised right and the regime must allow for changes in work patterns during the course of the year, particularly for irregular hours workers. They argue that even irregular hours may be irregular in a predictable manner and the employer can limit the leave taken to the number of hours accrued as the year progresses. As Underhill LJ said at para 77 of his judgment, commenting on Russell v Transocean referred to at the start of this judgment (see para 3 above), it was not contended by the employer in that case that the part-year workers were entitled to less than the full statutory holiday entitlement even though they only worked half the year.

As to the second point, the idea that leave must be treated as accruing with days worked and that it can only be taken once it has accrued makes it much less convenient for employers and workers to arrange for all the leave to which the worker is entitled to be taken in the leave year, particularly where the leave year coincides with the calendar year. Many casual workers on zero hours contracts are required to work many more hours in the weeks before Christmas than during the rest of the year. According to the Harpur Trust, they would then accrue more leave days. But it will probably be inconvenient for their employer to allow them to take those days during these busy weeks and according to regulation 13(9) of the WTR leave may only be taken in the leave year in respect of which it is due. This method may therefore make it more difficult and less convenient for employers to ensure that workers are able to take what turns out to be their full leave entitlement in the leave year.

(d) Other criticisms

The Harpur Trust criticise the Calendar Week Method for reasons other than its alleged failure properly to afford workers the rights conferred by the WTD. Mr Glyn submits that the changes in the Acas guidance have led to uncertainty. We do not accept that this incorrect guidance previously given by Acas can affect the proper construction of the statutory wording.

We reject the submission that the conformity principle is for the purposes of the WTR the answer to all these problems. There is no difficulty for the Harpur Trust in calculating Mrs Brazel’s annual paid leave in accordance with the Calendar Week Method, as they did prior to September 2011. We hold that that is the correct application of the statutory provisions in this case.

Finally, the Harpur Trust rely on section 229(2) which we have set out at para 33, above. This provision was, they said, at the court’s disposal for arriving at a “just” reference period to be used when calculating holiday pay. They accepted that that subsection is not incorporated into regulation 16 of the WTR but contended nevertheless that it could be deployed to give proper effect to the conformity principle. We do not consider that section 229(2) is relevant here. That provision is intended primarily to deal with, for example, a lump sum annual bonus related to the whole year’s work but paid at a particular date which may or may not be in one of the weeks falling within the reference period for the purposes of section 224. That bonus may be apportioned as seems just. We agree with Mrs Brazel that section 229(2) is not a general dispensing provision allowing the court to recalculate the amount of a week’s pay in any way it considers appropriate.

Conclusion

We therefore conclude that the Court of Appeal was correct to hold that the Calendar Week Method represents the correct implementation of the WTR and that this is fully compliant with EU law. We summarise our principal reasons as follows:

(i) Although the CJEU’s case law suggests that in general, the minimum entitlements prescribed by the WTD are calculated by reference to work actually carried out by the worker (subject to exceptions explained in, for example, QH and Dicu), the WTD does not prevent a more generous provision being made by domestic law.

(ii) Even if, therefore, the proper construction of the WTR results in Mrs Brazel being entitled to a greater amount of leave than she might be strictly entitled to under the WTD and to a proportionately greater leave requirement than full-time workers, such a construction is compliant with the WTD.

(iii) The incorporation into the WTR of the means of calculating an average week’s pay set out in section 224 of the 1996 Act for workers, including those who work very irregular hours, was a policy choice made by Parliament according to which the number of hours worked affects the amount of a week’s pay in some circumstances but not in others.

(iv) There is nothing in the WTR which indicates that the regulations should be construed so as to permit the alternative methods of calculating pay that have been adopted or proposed by the Harpur Trust and aspects of their proposed methods are directly contrary to what is required by the statutory wording and the WTR.

In short, the amount of leave to which a part-year worker under a permanent contract is entitled is not required by EU law to be, and under domestic law is not, prorated to that of a full-time worker.

Accordingly, we dismiss this appeal.