It has become an orthodoxy in employment law that an employment contract requires personal service, which a valid substitution clause can defeat. The same approach has been read across to ‘worker’ status, where the various statutory definitions all refer to contracts under which an individual undertakes to ‘perform personally any work or services’ for the other party, and to the extended definition of ‘employment’ in the Equality Act 2010. The prize for an employer of a carefully crafted substitution clause is, therefore, very valuable: no minimum wage, no discrimination claims, no working time limits or holiday pay, no unfair dismissal or redundancy rights – in fact, no employment rights of any kind.


(Photo from

In recent years ‘no mutuality’ clauses, purporting to negate any obligation on the part of the employee to work, have ceased to work the same magic as substitution clauses. Mutuality in the strict sense is an essential element of employment status but not the status of ‘worker’ (Cotswold Developments v Williams [2006] IRLR 181, though cf. Windle [2016] ICR 721.) Following Autoclenz [2011] ICR 1157, the courts may well find such clauses are shams because they do not reflect how the parties conducted themselves in practice; and, as Elias LJ has explained more than once, even if there is no mutuality present in weeks when no work is done, once a worker agrees to work (or has begun working), a particular assignment, a micro-contract with mutuality almost invariably arises.

Little wonder that the attention of employers’ lawyers has now shifted to substitution clauses as a more effective means of boundary manipulation. Deliveroo, for example, recently re-drafted their cyclists’ contracts to include detailed, unrestricted substitution clauses. Though rarely exercised in practice, the clauses were sufficient to persuade the Central Arbitration Committee that the cyclists were not workers for the purpose of s.296 of the Trade Union and Labour (Relations) Consolidation Act 1992 (which is in similar terms to other definitions of ‘worker’), meaning that their union’s application for compulsory recognition was dismissed. Moreover, by imposing requirements on individuals to provide a continuous, uninterrupted service for long periods, an employer can pretty much ensure an individual must use a substitute at times, undermining any ‘sham’ argument. Evidence before the recent BEIS Committee inquiry into the Taylor Review, for example, showed that Hermes required its drivers to work 21 days without a break during the busy Christmas period. The paradoxical result is that by imposing excessive working hours an employer can take its workforce outside the scope of the legislation meant to protect against such abuses, the Working Time Regulations.

I’ve always been troubled by this area of law, and the large black hole it opens up in the protection of very fundamental social rights. The contract of employment is, of course, personal in the sense that it cannot be assigned to a third party, as the House of Lords held long ago in Nokes v Doncaster Amalgamated Collieries [1940] AC 1014. While the earliest cases, such as R v Negus (18872-5) L.R. 2 C.C.R. 34 and Yewens v Noakes (1880) 6 QBD 530, focused on control alone as the test for a servant, the issue of substitution arose in Braddell v Baker (1911) 104 LT 673. A jobbing gardener (coincidentally named Taylor) worked four days a week but it was agreed that he could work for other employers, and if he could not attend on a particular date he would send a qualified substitute, as he had done on occasions. The Kings Bench Division upheld the justices’ decision that he was not a ‘servant’ for the purpose of tax legislation. Only one of the three judges, Hamilton J, addressed the effect of the substitution clause, with the others looking at the overall picture or the absence of control. Hamilton J drew a distinction between a servant who agreed to render personal service and was not obliged to procure a substitute if a lawful reason prevented him performing his contract, and a tradesman who contracted to supply a service and was therefore free (or required?) to provide a substitute if he could not attend personally.

Braddell was cited in argument in the classic case of Ready Mixed Concrete v Minister of Pensions [1968] 2 QB 497, and McKenna’s judgment soon came to be treated as the source of the rule that potential substitution could defeat an obligation of personal service. It spawned later authorities, beginning with Express & Echo v Tanton [1999] ICR 693, in which a delivery driver’s contract allowed him to appoint another suitable driver if he was ‘unable or unwilling to perform the services personally’, delivering a knock-out blow to employment status. It was followed by many cases, some of which were cited by Etherton MR in the Court of Appeal in Pimlico Plumbers v Smith [2017] ICR 657. The logic of Braddell was soon forgotten, as clauses allowing substitutes only where an individual was unable to perform were treated as insufficient to negate personal service. In addition, a test originally developed for distinguishing employees and the self-employed was adopted without modification for the statutory definitions of ‘worker’, even though these are expressed in different terms and serve different purposes, and the term ‘worker’ can include those who are self-employed.

Some personal performance in fact by the claimant is an invariable feature of the cases where substitution clauses have defeated personal service. The focus of the law, in theory, is not on who in fact performs the service but how the contractual right to exercise a substitution clause affects the individual’s obligation to perform services. If the substitution clause permits the individual never to perform any duties personally – in effect, permitting perpetual sub-contracting of the service – it should be inconsistent with an obligation of any personal service.

So far, so simple. But the problem, of course, is that clauses come in many shapes and sizes and, especially since Autoclenz, courts and tribunals are enjoined to examine what happens in practice to determine the ‘real’ obligations of the parties. The tension between written norms and empirical facts can make the legal waters very murky indeed, and Etherton MR’s attempt to clarify the principles in Pimlico served mostly to highlight their uncertainty. First, he accepted that an unfettered right to appoint a substitute is inconsistent with an obligation of personal service. Second, where on its face the clause is conditional, he said its effect on personal service ‘will depend on the precise contractual arrangements and, in particular, the nature and degree of any fetter on a right of substitution or, using different language, the extent to which the right of substitution is limited or occasional’.

The second principle implicitly acknowledges a considerable degree of conceptual uncertainty and discretion on the part of the particular tribunal, as the fine lines drawn in the cases cited by Etherton MR demonstrate. On its own terms it blurs the line between the norm (‘degree of fetter’) and factual practice (‘limited or occasional’). But even the first proposition turns out to be less clear than at first blush. Take an apparently unrestricted substitution clause in a contract between an employer and worker A. If A never uses it, does that suggest it is a sham, as was the case in Autoclenz itself, or simply that A hasn’t needed to use it? What if another worker, B, who is employed on identical terms, exercises the clause a couple of times? Does B’s practice defeat any sham argument for A and all other workers employed on those terms? Does it matter that B only exercised it because of exceptional pressing family reasons, or to go on holiday, or that she informed or asked the employer beforehand?

If we started from scratch, and without the baggage of the case-law, it might be preferable to ask a single and more straightforward legal question: does the contract, as interpreted in light of practice, impose any obligation on the employee to do any personal work? At least in relation to ‘worker’, the wording of the statutory definitions supports this approach. For example, s.230 of the Employment Rights Act 1996 refers to an undertaking ‘to perform personally any work or services’ (my emphasis), not an undertaking to perform ‘the work or services’, lifting the terminology from the Employers and Workmen Act of 1875. It implies it should be sufficient that the putative worker agrees to perform personally some work, even if on other occasions she has the right to engage a substitute.

Alternatively, courts might have adopted a similar approach to how they have analysed the closely-related doctrine of mutuality of obligation. There, judges have not hesitated to find an obligation to work once an individual agrees to perform work for a particular week, day or shift, even if he or she has the right to refuse to work beforehand. This reasoning has not spread into substitution clauses, though its extension is hinted at by Underhill LJ in Pimlico (at §130). Its logic would be that a worker with a right to substitute is nonetheless still a worker in respect of any individual assignment or shift he or she agrees to work unless, perhaps, the right extends to providing a substitute even for a particular shift that the individual has agreed to work (or has already begun working).

But the cases summarised by Etherton MR in Pimlico have crystallised into different principles, in which the focus is on the width of the power to appoint a substitute, obscuring the very specific historical origins of the doctrine, the different wording of the statutory definitions, and the purposes of the legislation. Lord Wilson’s judgment in the Supreme Court in Pimlico [2018] UKSC 29 falls short of reappraising the fundamentals of this area of law: for example, in common with the earlier cases it takes for granted that the same test applies to personal service in employee or worker cases (§20), despite the very different statutory concepts engaged. It nonetheless offers fresh help in cutting through this particular legal Gordian knot.

Just like Etherton MR below, Lord Wilson sought to clarify the boundary separating substitution clauses which were consistent or inconsistent with personal service. For this purpose, he accepted that Mr Smith had the right to substitute another Pimlico operative in a wide range of circumstances, not limited to when he was unable to do the work but including when he found more lucrative work elsewhere (§28). Although Lord Wilson declined to rewrite the statutory test, his formulation of the relevant question comes very close to doing just that (§32):

But there are cases, of which the present case is one, in which it is helpful to assess the significance of Mr Smith’s right to substitute another Pimlico operative by reference to whether the dominant feature of the contract remained personal performance on his part.

In light of the other provisions of the contract which were addressed to Mr Smith personally, and the restriction of the substitution clause to other Pimlico operatives, Lord Wilson concluded the tribunal was entitled to find Mr Smith was a ‘worker’ (and, therefore, was employed for the purpose of the Equality Act).

Lord Wilson’s test leaves some important issues unclear or unresolved. He considered it significant that the substitute had to be another Pimlico operative who would also be bound by a ‘suite of heavy obligations’ (§34). But the attributes of the substitute tell us little about the obligation owed by Mr Smith to work personally – unless, that is, Lord Wilson’s ‘dominant feature’ test overwrites the statutory wording. We are left guessing, too, when it is ‘helpful’ to apply the ‘dominant feature’ test or how this test would have affected earlier cases. Mr Smith’s power of substitution doesn’t look very dissimilar to Mr Taylor’s right in Braddell, for example. Nor is it clear why Lord Wilson considered Tanton as a ‘clear case’ (§21). In Tanton the relief driver had to be ‘trained and suitable to undertake the services’, illustrating that the company was concerned about the substitute’s attributes, just as in Pimlico. The answer may be that in Tanton the contract clearly contemplated the possibility of no service at all being performed by Mr Tanton, illustrated by the fact that one stage a substitute covered his duties for six months while he was ill. But if one were to apply the ‘dominant feature’ test, it is plausible that Mr Tanton was an employee.

Regardless of these uncertainties, a focus on whether the ‘dominant feature’ of the contract is personal performance by the individual probably reduces the potential for substitution clauses to negate employment protection. It makes the answer in a case such as Autoclenz more obvious, and diminishes the significance of the occasional exercise of a substitution clause. It comes close to Mark Freedland’s suggested test, in The Personal Employment Contract, of whether the contract intends the work will be predominantly carried out by the worker in person. A contract may envisage some use of substitutes in a wide range of circumstances; but still its dominant feature may be personal performance by the other party. A recent ET decision, Leyland v Hermes, relied in part on the ‘dominant feature’ test in finding Hermes couriers were workers.

The effect of Lord Wilson’s judgment may soon be tested further in the forthcoming judicial review of the CAC ruling in Deliveroo. True, on its face the written clause in the cyclists’ contracts contemplated them never performing any services for Deliveroo. But in practice substitution was rare for the obvious reason that if riders didn’t want to do jobs they simply didn’t bother to log on at all: hence the CAC’s puzzlement about why the substitution clause existed (save for the obvious reason of removing employment rights). It hardly looks like a ‘clear case’ à la Tanton. Despite the valiant drafting of the clause, the factual background doesn’t suggest that the contract envisaged no personal performance by the individual cyclists. Interpreted in light of practice, the ‘dominant feature’ of the contracts appears to be personal service by the individual cyclists, with the use of the substitution clause confined to exceptional and very occasional circumstances.

michael fordAbout the author: Michael Ford QC is a Professor of Law at the University of Bristol, a QC at Old Square Chambers and a fee-paid Employment Judge. He is also Co-editor of the UK Labour Law Blog.

(Suggested citation: M Ford, ‘Pimlico Plumbers: Cutting the Gordian Knot of Substitution Clauses?’, UK Labour Law Blog, 19th July 2018, available at