BAILII [Home] [Databases] [World Law] [Multidatabase Search] [Help] [Feedback]

United Kingdom Employment Appeal Tribunal


You are here: BAILII >> Databases >> United Kingdom Employment Appeal Tribunal >> British Airways v Pinaud (Part Time Workers) [2017] UKEAT 0291_16_0108 (01 August 2017)
URL: http://www.bailii.org/uk/cases/UKEAT/2017/0291_16_0108.html
Cite as: [2017] UKEAT 291_16_108, [2017] UKEAT 0291_16_0108

[New search] [View without highlighting] [Printable RTF version] [Help]


Appeal No. UKEAT/0291/16/DA

 

 

EMPLOYMENT APPEAL TRIBUNAL

FLEETBANK HOUSE, 2-6 SALISBURY SQUARE, LONDON EC4Y 8AE

 

 

                                                                                                             At the Tribunal

                                                                                                             On 1 August 2017

 

 

 

Before

HIS HONOUR JUDGE DAVID RICHARDSON

(SITTING ALONE)

 

 

 

 

 

 

 

BRITISH AIRWAYS                                                                                     APPELLANT

 

 

 

 

 

MRS F PINAUD                                                                                         RESPONDENT

 

 

 

Transcript of Proceedings

 

JUDGMENT

 

 

 


 

 

 

 

 

 

 

APPEARANCES

 

 

 

 

 

For the Appellant

MR ANDREW BURNS

(One of Her Majesty’s Counsel)

Instructed by:

Harrison Clark Rickerbys Limited

First Floor Suite

Thorpe House

29 Broad Street

Hereford

HR4 9AR

 

 

For the Respondent

MR MICHAEL SMITH

(of Counsel)

Instructed by:

OH Parsons LLP

Churchill House

Chalvey Road East

Slough

Berkshire

SL1 2LS

 

 

 

 


SUMMARY

PART TIME WORKERS

 

The Employment Tribunal correctly approached the question whether the Claimant was treated by the Respondent less favourably than it treated a full-time comparator as regards a term of the contract concerned with pattern of availability for work (regulation 5(1)(a) of the Part-time Workers (Prevention of Less Favourable Treatment) Regulations 2000).

 

The Employment Tribunal erred in law by holding, in effect, that statistical evidence produced by the parties was irrelevant to the question of justification (regulation 5(2)(b)).

 


HIS HONOUR JUDGE DAVID RICHARDSON

 

1.               This is an appeal by British Airways plc (“the Respondent”) against a Judgment of an Employment Tribunal (“ET”) consisting of Employment Judge Vowles, Mrs Brown and Mr Gregory sitting in Reading.  By its Judgment the ET upheld a claim by Mrs Florence Pinaud (“the Claimant”) that she had been treated less favourably than a full-time worker, contrary to the Part-time Workers (Prevention of Less Favourable Treatment) Regulations 2000 (“PTWR”).  The Respondent appeals against that Decision.  

 

The Part-Time Workers Regulations

2.               The appeal concerns the way in which the ET applied regulation 5 of the PTWR.  It is convenient to set out the key provisions of the legislation at the beginning of this Judgment.

 

3.               Regulation 5 identifies the right of a part-time worker not to be treated less favourably than a full-time worker.  It provides:

“(1) A part-time worker has the right not to be treated by his employer less favourably than the employer treats a comparable full-time worker -

(a) as regards the terms of his contract; or

(b) by being subjected to any other detriment by any act, or deliberate failure to act, of his employer.

(2) The right conferred by paragraph (1) applies only if -

(a) the treatment is on the ground that the worker is a part-time worker, and

(b) the treatment is not justified on objective grounds.

(3) In determining whether a part-time worker has been treated less favourably than a comparable full-time worker the pro rata principle shall be applied unless it is inappropriate.”

 

4.               The “pro rata” principle is defined in regulation 1(2) as follows:

“(2) In these Regulations -

“pro rata principle” means that where a comparable full-time worker receives or is entitled to receive pay or any other benefit, a part-time worker is to receive or be entitled to receive not less than the proportion of that pay or other benefit that the number of his weekly hours bears to the number of weekly hours of the comparable full-time worker;

…”

 

5.               There is also a definition of “weekly hours” which applies to the pro rata principle.  This is found in regulation 1(3):

“(3) In the definition of the pro rata principle and in regulations 3 and 4 “weekly hours” means the number of hours a worker is required to work under his contract of employment in a week in which he has no absences from work and does not work any overtime or, where the number of such hours varies according to a cycle, the average number of such hours.”

 

6.               The PTWR were intended to give effect to the Part-Time Workers Directive 97/81/EC, which was extended to the United Kingdom by Directive 98/23/EC.  The Directive implemented a Framework Agreement on Part-Time work concluded between social partners on 6 June 1997.

 

7.               The Framework Agreement provided as follows:

Clause 4: Principle of non-discrimination

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.

3. The arrangements for the application of this clause shall be defined by the Member States and/or social partners, having regard to European legislation, national law, collective agreements and practice.”

 

8.               The expression “pro rata temporis” found in the Framework Agreement means, in broad terms, “in proportion to the time involved”.  There is no equivalent in the Framework Agreement to the tightly drawn definition in regulation 1(2) and (3) of the PTWR by reference to weekly hours.

 

The Background Facts

9.               The Claimant joined the Respondent on 24 June 1985.  She was promoted to Purser in 1993.  She worked full-time at first.  On her return from maternity leave in 2005 she started to work part-time.  She remained on part-time work until she took voluntary redundancy on 30 April 2015.

 

10.            Some long-serving crew in the part of the Respondent’s business where the Claimant worked operated on a contractual working pattern.  The full-time pattern was called the “6/3 pattern”; six days off and three days on.  Over the year, this meant that the full-time crew member was available for work 243 days and off for 122 days.

 

11.            The part-time pattern, to which the Claimant moved, was called the “14/14 pattern”.  It was expressly described as a 50% contract, and the annual basic salary was 50% of that which the full-time crew member received.  The pattern was 14 days on and 14 days off, and within the 14 days on, the part-time crew member had to be available for 10 days.  Over the year, this meant that the part-time crew member had to be available for 130 days.

 

12.            Whether the crew member was full-time or part-time, she was required, on an available day, to bid for work.  Mr Andrew Burns QC for the Respondent briefly explained the system to me; the crew member could, depending on preference, bid for longer flights which might result in additional elements of pay, or shorter flights which may be less well remunerated, but better for her personal circumstances, or some combination of the two.  Bids for flights would generally be met on the basis of seniority, but there was also the possibility of being required for ground duties, such as training, or being on standby to cover eventualities or even being stood down altogether.  

13.            As the ET recognised in paragraph 43 of its Reasons, actual hours worked could fluctuate.  There was no clear pathway from days of availability to hours worked, or even precise days worked.  The annual basic salary, however, did not vary with the number of duty hours.

 

The ET Proceedings

14.            At the ET hearing, the Claimant was represented by Mr Michael Smith and the Respondent by Mr Andrew Burns QC.  The representation of the parties remains the same today.

 

15.            The Claimant’s case was put in two ways.  As we shall see, the ET reached a decision only on the first way.  Mr Smith’s skeleton argument for the ET hearing contained the following passage which the ET took as a summary of his case:

“19. In any event, from an examination of the figures provided, the following is clear:

(i) The Claimant, as a part-time worker, was required to perform 3.5% more work days proportionately than a full-time worker from 2005-2015; and

(ii) The Claimant was regularly required to work more than 50% of the Duty Hours of a full-time employee, despite receiving 50% of a full-time employee’s salary.

The above constitutes less favourable treatment under Regulation 5.”

 

16.            This paragraph was probably not intended by Mr Smith to be a legal analysis of his position and it requires a degree of unpacking.  The first way concerns available days; it engages regulation 5(1)(a) of the PTWR.  It was said that the Claimant was treated less favourably as regards the terms of her contract because she was required to be available for proportionately more days.  The full-time worker was required to be available for 243 days per year, 50% of which was 121.5 days, but the part-time worker was required to be available for 130 days.  This was proportionately 3.5% more days of availability for the same pay.  

 

17.            The second way concerns duty hours.  It is rather less clear how the second way relates to regulation 5(1).  It was said that the Claimant was treated less favourably because she was “regularly required” to work more duty hours proportionately than her full-time equivalent.  Mr Burns suggested that this was still about regulation 5(1)(a), and concerned pay as opposed to availability for work.  Mr Smith suggested that this part of his case could be put either under regulation 5(1)(a) or (b) and that, in any event, it flowed from the first way he had put his case; for if the Claimant was available for additional days, it was likely that she would work additional hours.

 

18.            There was a good deal of common ground at the ET hearing.  It was accepted that the Claimant had a full-time comparator, a Purser, Miss Fiona Evans who worked on the “6/3 pattern” for the 10 years when the Claimant worked on the “14/14 pattern”.  Statistics were available for that period for the Claimant, the comparator, and, I understand, some other potential comparators (who were, in the end, not relied on).

 

19.            It was also common ground that the pro rata principle as defined by regulation 1(2) could not be applied in assessing the first way in which the Claimant put her case.  Exactly why this should be so has not been entirely agreed today.  I will return to this matter later.

 

20.            For reasons that will become apparent, the ET made no findings about the statistics.  I will summarise, briefly, each party’s case about them, recognising that this is just a summary, and the ET heard witness evidence as well as receiving the statistics.

 

21.            The Claimant’s case about the statistics was that they showed over the 10-year period that the Claimant worked 53.17% of the duty hours of Miss Evans, and 53.83% of the flying hours of Miss Evans.  In other words, the contractual disparity was reflected in a disparity in hours worked.

 

22.            The Respondent’s case was that the statistics varied greatly.  They showed that over some periods, the Claimant worked less hours that Miss Evans, for example markedly less hours towards the end of her employment.  The actual hours worked depended on the choices and bids made by crew members.  Broadly speaking, she worked 50% of the weekly hours of her comparator.  A broad approach to comparability was appropriate and any differences could be regarded as trivial or minimal.

 

23.            This brings me to the Respondent’s case about justification.  The Respondent argued that if (which it denied) there was less favourable treatment, that treatment was justified on objective grounds.  The legitimate aim was to provide a workable 50% contract working pattern.  This could not be achieved exactly because the year was not divisible precisely by 6/3 or 14/14.  The slight difference was a proportionate means of achieving the legitimate aim even though it was not mathematically exact.  The adverse effect was trivial, especially bearing in mind the “swings and roundabouts” involved in the system of bidding for flights.

 

The ET’s Reasons

24.            The ET defined the Claimant’s case in accordance with paragraph 19 of the skeleton argument which I have already set out.  In quoting from paragraph 19, the ET omitted some wording from within paragraph 19(ii).  I have no doubt that this was simply a “slip”.

 

25.            The ET first addressed the way in which the case was put in paragraph 19(i).  It said:

“24. It was not in dispute that the Claimant was a part-time worker and the comparator was a full-time worker.

25. It was also not in dispute that the Claimant had to be available for work on 130 days per year and the comparator had to be available for work on 243 days per year.

26. The Respondent accepted that the requirement to be available for 243 days and 130 days respectively amounted to terms of the respective contracts of employment of the comparator and the Claimant.

27. 50% of 243 days would be 121.5 days.  The Claimant had to be available for 130 days.  That was 8.5 days more than 50%.  Put another way, she had to be available for 53.5% of the days on which the comparator had to be available, but was only paid 50% of the comparator’s salary.

28. The Tribunal found that the Claimant had therefore been treated less favourably than the comparator as regards the terms of her contract within the meaning of regulation 5(1)(a).

29. Under regulation 8(6) where less favourable treatment is found, it is for the employer to identify the ground for that less favourable treatment.”

 

26.            The ET went on to find that the claim in this respect had been presented in time.  It rejected an argument that the additional availability days were trivial.  It said, “The requirement to be available for work for an additional 8.5 days over the course of a year was a significant period” (paragraph 33).  It found that the less favourable treatment was on the grounds that the Claimant was a part-time worker.

 

27.            The ET then turned to the question of justification.  It quoted the Respondent’s submissions at some length.  It held that less favourable treatment would only be justified on objective grounds if it was to achieve a legitimate objective, and was necessary to achieve that objective and was an appropriate way to do so.  It concluded:

“37. The Tribunal found that there was a legitimate objective.  That was to provide a part-time shift pattern which was workable, practical, predictable, flexible and popular with cabin crew to enable BA to run its business effectively and the cabin crew to organise their home lives efficiently with more predictable working time.

38. The Tribunal also found, however, that the less favourable treatment was not a necessary or appropriate means to achieve the objective.  A non-discriminatory means of achieving the same legitimate aim would be to simply re-name the part-time 14/14 contract as a part-time 53.5% contract and pay an annual salary of 53.5% of the full-time 6/3 salary.  That simple expedient, although it would cost the Respondent more in salary, would remove the less favourable treatment and retain all the current advantages of the part-time 14/14 contract.

39. Ms Elliott accepted in her evidence that although other part-time shift patterns were reviewed when the Part Time Worker Regulations were implemented in July 2000, the part-time 14/14 shift pattern was not reviewed to assess whether it may be discriminatory.

40. It follows that the Tribunal found that the less favourable treatment was not justified on objective grounds.

41. The complaint of part time worker discrimination was well founded.”

28.            The ET did not go on to consider the Claimant’s alternative case set out in paragraph 19(ii) of the skeleton argument.  Under a heading “Statistics”, the ET said the following:

“42. The Tribunal was presented with a range of statistics showing the actual hours worked by the Claimant and the comparator and others who were also on a 6/3 full time contract.

43. The Tribunal found that so far as the complaint based upon paragraph 19(i) was concerned, those statistics were not only unnecessary but irrelevant.  The Tribunal’s finding is based upon the requirement to be available for work on a specific number of days and not on the actual work which was carried out on those days.  There was no dispute that on any particular duty day, cabin crew could be employed on different tasks.  For example, flying duties or ground support duties.  The actual period of work in hours could, and did, fluctuate and, on occasions, the worker would not be required to attend for work at all.  The crucial point, however, was that she would still have had to be available for work on that day.  The Tribunal’s finding was based upon the Claimant having to be available for work on 53.5% of the days which the full-time 6.3 comparator had to be available.  It was that which constituted the less favourable treatment.

44. The Tribunal therefore considered it unnecessary to consider the complaint under paragraph 19(ii), though that matter, and the statistics referred to above, may be relevant to the issue of remedy.” (Original emphasis)

 

Submissions

29.            On behalf of the Respondent, Mr Burns has made submissions to me on support of three grounds of appeal.  The three grounds of appeal, in summary, are that the ET erred in law in: (1) applying the pro rata principle; (2) making a comparison of available hours rather than weekly hours worked; (3) on the question of justification, holding that statistical evidence was irrelevant to its decision and deciding the case not put to the Respondent’s witnesses.

 

30.            Developing these arguments, he submits that the first way in which the Claimant put her case was to do with available hours, not pay.  The comparison of available hours could not be done, taking into account the pro rata principle, because this was only to do with pay or other benefits, not available hours.  Yet, he submits, this is what the ET effectively did, applying the pro rata principle, and incorrectly importing the salary the Claimant was paid in order to do so.

 

31.            The ET’s true task, he submits, was to compare the hours the Claimant and her comparator actually worked, and apply the pro rata principle to those to see if there was less favourable treatment.  If it had performed this exercise, it would have found there was no less favourable treatment in the last six months of her employment, and a complaint relating to any earlier period would have been out of time.

 

32.            He further submits that the ET was wrong in law to regard the statistical evidences irrelevant to the question of justification.  The ET decided that the “14/14 pattern”, with its differential in available days, pursued a legitimate aim, and in deciding if the less favourable treatment was a proportionate means of achieving that aim, it was relevant to take into account the practical impact of the less favourable treatment.  A suggestion that the 3.5% increase in pay was the answer was never put to any witness.  If it had been, the ET would have received evidence as to the practical difficulty and implication of such an increase.

 

33.            In answer to these submissions, Mr Smith argues that the ET was required, by regulation 5(1)(a), to engage in a comparative exercise regarding the terms of a part-time and a full-time worker.  The pro rata principle was not applicable because the relevant term of the contract related to days available, rather than hours worked, but a comparison still had to be made and there was no error of law in the way in which the ET carried out that exercise.  The fact that the Claimant was receiving 50% of the basic salary of the full-time comparator was essential background against which the comparison had to be made.

 

34.            On the question of justification, Mr Smith submits that the ET was correct to regard the statistics as irrelevant to the question whether the terms were less favourable.  Irrespective of the hours the Claimant actually worked, she had to be available for work for a greater period proportionately than the comparable full-time worker.  Mr Smith says the statistics would actually have shown that the hours worked over the 10-year period reflected that underlying difference in the contract.  Further, he argues that the ET’s suggestion of an increase in salary of 3.5% to reflect the disparity in terms was not essential to its reasons.

 

Less Favourable Treatment

35.            It is important to keep in mind that the ET reached its decision by application of regulation 5(1)(a).  It found that the Claimant had been treated less favourably, as regards the terms of her contact; see paragraph 28 of its Reasons.  Under a full-time contract, the comparator was required to work 243 days to earn the basic salary.  Under her part-time contract, she was required to work 130 days (not 121.5 days) to earn half the basic salary.

 

36.            I will begin with Mr Burns’ complaint that the ET imported the question of salary when it was irrelevant to do so.  I have no hesitation in rejecting that submission.  It was plainly relevant to the way in which the Claimant was putting her case.  The 53.5% to which she referred, reflected the fact that she was earning 50% of the salary of her comparator, but had to work 53.5% of the days.  In principle, this was a matter about which she was entitled to complain.  It would have been irrational and incoherent to have assessed whether the contractual term, relating to days of availability, was less favourable treatment without taking into account relevant pay of the full-time and part-time comparator.

 

37.            I turn next to Mr Burns’ complaint that the ET erred in law in applying the pro rata principle, when it had been agreed that this principle did not apply to the question raised by regulation 5(1)(a).  I agree that it was not appropriate to apply the pro rata principle, as defined by regulation 1(2) and (3), to the question whether the Claimant had been treated less favourably as regards to the term of her contract relating to available days.  This term of the contract did not dictate any particular number of weekly hours the Claimant would work, for reasons I explained earlier in this Judgment.  It dictated only the days she had to be available and left the hours worked to be allocated by subsequent steps in the process.  It would have been irrational to compare the position of the Claimant and her full-time comparator in relation to this contractual term by reference to the number of weekly hours.

 

38.            However, it is to my mind plain that the ET did not fall into the error of applying the pro rata principle, as defined by regulation 1.  It made no comparison of weekly hours.  It correctly compared days of availability.  Given the terms of the contract, this was plainly the correct comparison to make.

 

39.            It is true, of course, that the ET’s comparison involved an assessment of time and proportion.  I do not think there was any other rational basis on which to proceed.  Regulation 5(1)(a) requires a comparison to be made and it is not possible sensibly to make a comparison of terms relating to days of availability without assessing the days the Claimant was required to be available against the days the full-time comparator was required to be available and the respective pay of each of them.

 

40.            I reject Mr Burns’ submission that the ET should have restricted itself simply to the available days.  It is important to keep in mind that in the Framework Agreement, the term “pro rata temporis” does not have a narrow definition.  Even if the narrow definition in the PTWR was inapplicable, the ET was still permitted to make a comparative assessment by reference to time and proportion if it was appropriate.  Here it plainly was.

 

41.            It is also part of Mr Burns’ complaint that the ET undertook the exercise of comparing days of availability at all.  He argues that the Claimant’s real complaint is that she was not paid more.  This depends on a comparison of duty hours worked.  I do not agree with this submission.  There was a term of the Claimant’s contract which required for her to be available for work, in accordance with a particular pattern.  Once she raised the question whether it offended against the PTWR, the ET was bound to decide it.  It did not depend on a comparison of duty hours worked.

 

42.            Accordingly, I consider the approach of the ET to the question of less favourable treatment to have been correct.

 

Justification

43.            I then turn to the question of justification.  The leading authority is the decision of the Supreme Court in Ministry of Justice v O’Brien [2013] ICR 499.  The Supreme Court adopted guidance given by the European Court and the Advocate General in that case which had been the subject of a reference.  It is sufficient to quote the opinion of the Advocate General:

“The unequal treatment at issue must therefore be justified by the existence of precise, concrete factors, characterising the employment condition concerned in its specific context and on the basis of objective and transparent criteria for examining the question whether that unequal treatment responds to a genuine need and whether it is appropriate and necessary for achieving the objective pursued: see Del Cerro Alonso [2008] ICR 145, para 58, and Angé Serrano v European Parliament (Case C-496/08P) [2010] ECR I-1793, para 44.”

 

This was the approach which the ET followed. 

 

44.            While a contractual term can be justified, even if no thought was given to the relevant equality principles at the time when it was adopted, it will be more difficult for an employer to justify the proportionality of means chosen to carry out its aims if it did not conduct the exercise of examining the alternatives, or gather the necessary evidence to inform the choice of the time (see O’Brien, paragraph 48).  Hence, the relevance of the ET’s findings in paragraph 39 of its Reasons.

45.            Thus far, the ET has, to my mind, followed correct principles relating to justification, but, I have come to the conclusion that the ET erred in law in its reasoning in one important respect.  The ET was, in my judgment, wrong in law to regard the statistical evidence produced by the parties as irrelevant to the question of justification.  I am satisfied that this was its approach (see paragraph 43 of its Reasons), coupled with a lack of any reference to the statistical evidence in the section which deals with justification.

 

46.            The ET, having accepted that the unfavourable treatment was in pursuance of a legitimate aim, was, to my mind, bound to make a practical assessment of the impact of the unfavourable treatment, when deciding whether the treatment was appropriate and necessary for achieving the objective pursuit.  Part of the case of the Respondent on this question depended on the statistical evidence.  It argued that the impact on the part-time worker was limited, because the statistics showed that the part-time worker, once the bidding process was undertaken, was in practical terms not required to work more hours than her full-time comparator.  Hence, even if there was, as so the ET found, unfavourable treatment, the Respondent said that the statistics showed that its impact was minimal.

 

47.            I confess that I am sceptical about the Respondent’s “swings and roundabouts” argument based on the statistical evidence.  I find it difficult to see why, if the part-time worker had to be available for a greater number of days, this should not work its way through into a significant impact for the employee, both in terms of days of availability and hours worked.  However, once granted that the ET found the unfavourable treatment to be in pursuance of a legitimate aim, I think the Respondent’s case about the practical impact required to be addressed.  The ET appears to have thought that the mere fact that the Claimant would have to be available to work on proportionately more days meant that the statistics were irrelevant.  I do not think this was correct.  The fact that the Claimant had to be available to work on proportionately more days was the feature that required to be justified.  Its existence did not rule out an enquiry into the extent to which it impacted on her so that a conclusion could be reached on whether the measure was proportionate.

 

48.            It also seems to be impossible for the ET to advocate, as it did, an increase in salary as a simple, non-discriminatory way of achieving the same aim, without asking whether the unfavourable treatment, in terms of days of availability, did work its way through into the amount of work the Claimant did.  If it did not, then it is far from obvious that an increase in salary was an alternative way of achieving the legitimate aim, which the Respondent could be expected to adopt.  It might, indeed, be out of proportion to the impact of the disparate treatment on the Claimant.

 

49.            I am told that this feature, the ET’s suggestion of a salary increase, did not figure largely in the hearing and was not put to the Respondent’s witnesses.  If it had been, I think the ET would have understood more clearly the importance of addressing the statistical evidence which it had before it.

 

50.            It follows that on that ground, the appeal will be allowed.  The ET’s finding of less favourable treatment will be upheld but the question of justification will be remitted.

 

51.            I turn now to the question of remission.  If justification on the first way in which the Claimant put her case is to be remitted, then as a matter of elementary justice the ET will also have to consider the second way in which she put her case.  So the matter is remitted for the ET to consider afresh justification in respect of the first way in which the Claimant put her case, and all aspects of the second way in which she put her case.

 

52.            I have to consider whether remission should be to the same ET or to a differently constituted ET.  In reaching this decision, I take into account general guidance given in Sinclair Roche & Temperley v Heard [2004] IRLR 763.  In this case, I do not doubt the professionalism of the ET, but the ET would have to grapple, and well over a year after the original hearing, with some quite difficult statistical evidence and with its memory of the impact of witnesses which it heard over a year ago.  It also will have to consider the second element of the case, which it did not address at all on the last occasion.  I think that will be a difficult exercise for an ET and, while I do not doubt the professionalism of this ET, I think, on the whole, the best course of action is for the matter to be remitted to a freshly constituted Tribunal.

 


BAILII: Copyright Policy | Disclaimers | Privacy Policy | Feedback | Donate to BAILII
URL: http://www.bailii.org/uk/cases/UKEAT/2017/0291_16_0108.html