Chapter 3: Extensions of time

An extension changes one thing: the date by which the contractor must finish. The SCL Protocol, the Society of Construction Law's guidance on delay, sets out what that gives each side (Core Principle 2). The contractor is relieved of damages for delay up to the new date and can replan the rest of the work. The employer gets a new completion date, which prevents time for completion of the works becoming 'at large', and can plan its own activities around it. The Protocol is guidance, not law.

Only delay to completion earns an extension, and the reason is ordinary causation. The clause asks how much later the job will finish because of the employer's event. An event that holds up work with time to spare does not make the job finish later, so it earns nothing. The time an activity can slip before it holds up anything else is its float. To move the completion date, the event has to hit the critical path: the longest chain of dependent activities through the job, whose length fixes the finish. Chapter 2 explains both ideas.

The prevention principle works the same way. It rests on the unfairness of insisting on a date the employer has made impossible, and Hamblen J pointed out in Adyard (Commercial Court, 2011) that this necessarily means prevention in fact ([264]). The act must cause some actual delay ([282]). A breach by the employer that delays nothing gives no time. It may still cost the contractor money, but that is a separate claim (chapter 11).

The model project used throughout this guide shows the arithmetic. It is invented: thirteen activities, a planned finish in week 34 and a contract completion date of week 36. Suppose the employer holds up the cladding by 4 weeks. Cladding has 4 weeks of float, so the forecast finish stays at week 34 and no extension is due. Make the hold-up 8 weeks and the finish moves to week 38. That is 2 weeks past the contract date, and 2 weeks is the extension.

Why 2 weeks and not 4, when the finish moved 4 weeks from the planned week 34? Because the clause gives time for delay beyond the completion date, and the job had 2 weeks in hand. The SCL Protocol says that where a clause is worded that way, total float has to be used up before an EOT will be due (8.2). Contractors often argue otherwise (8.1), and who owns float is discussed in chapter 2.

Most contracts want the answer during the job. Under the JCT Design and Build Contract 2016, the employer fixes a later date when completion is likely to be delayed beyond the current date (cl 2.25.1.2, as quoted in Mace v Baltic [44]). It must review the date within 12 weeks after practical completion, whether or not the contractor notified the event (cl 2.25.5). The first decision is a forecast; the review looks back. Other forms give the decision to a certifier. Chapter 5 covers the decision-maker, and chapter 6 the forecast and the look back.

An extension answers one question: should the contractor pay damages for these weeks? Whether the contractor should be paid for them is a different question. The contract answers the two separately, because the parties can allocate the risks separately. Bad weather is the usual example. The employer cannot fairly charge damages for weeks lost to exceptional weather. Nor does it want to pay the contractor's site costs for those weeks, when it did nothing to cause them. That is why the Protocol gives weather as the most common example of time without money.

The SCL Protocol says the contractor is nearly always required to claim its entitlement to an EOT under one provision of the contract and its money under another (12.2). Time-only events are sometimes called "neutral events", a label the Protocol says misleads. They are only neutral in the sense that one party bears the time risk and the other party bears the cost risk.

It is clearer to sort events by who bears what. The Protocol calls an event at the employer's risk under the contract an Employer Risk Event, and one at the contractor's risk a Contractor Risk Event (Appendix A). A Contractor Risk Event gives neither time nor money. An Employer Risk Event can give time only, or time and money, depending on the contract. That makes three boxes. A fourth sometimes appears: money without time, where a change costs the contractor money but does not delay completion. Chapter 13 deals with that.

The JCT forms make the split visible. JCT DB 2016 lists fourteen Relevant Events that can give time (cl 2.26) and five Relevant Matters that can give money (cl 4.21). A Change, the JCT's word for a variation, is on both lists, and so is any impediment or default by the employer. As Jefford J put it in Clerkenwell (TCC, 2026), the Relevant Matters overlap with, but are not identical to the Relevant Events. Exceptionally adverse weather and strikes give time but not money ([14]).

Other forms draw the lines elsewhere. NEC4, which calls the employer the Client, has a single list of compensation events, and each one can change both the Prices and the Completion Date (cl 60.1, 66.2). Weather is on that list only if it is shown to occur on average less frequently than once in ten years, and only the excess over that level counts (cl 60.1(13)). So NEC gives time and money for rare weather, and nothing for weather that is merely bad. Chapter 16 compares the forms.

Amendments move events between boxes, so read the contract actually signed. In Mace v Baltic (TCC, 2026) an amended JCT form made the contractor responsible for obtaining consents, and said it shall not have nor make any claim for an extension of time ([36]). Check the edition too. The JCT 2016 edition was withdrawn on 31 March 2026. The 2024 edition adds Relevant Events for epidemics and changes in law (JCT, Information Booklet). A practitioner article in JCT News (October 2025) says they take clause numbers the 2016 edition used for other events.

The prevention principle is the reason extension clauses exist. Jackson J stated its essence in Multiplex (TCC, 2007): the promisee cannot insist upon the performance of an obligation which he has prevented the promisor from performing ([47]). In a building contract the contractor has promised to finish by a date. If the employer's own act makes that date impossible, the employer cannot hold the contractor to it.

What replaces the date? If the contract has no way to move it, nothing does. The date falls away, and the contractor's duty becomes one to finish within a reasonable time. In the usual phrase, time becomes at large ([48]), and the Court of Appeal adopted that statement in North Midland ([28]). Liquidated damages fall with the date. They are a sum per week counted from the completion date, and once there is no date there is nothing to count from.

The rule is old, and it has always been blunt. In Holme v Guppy (1838) the owner failed to give possession of the site for four weeks. The builders were left at large and forfeited nothing for the delay. In Dodd v Churton (1897) the owner ordered extra work that delayed completion. Lord Esher MR said that an owner who does so is thereby disentitled to claim the penalties for non-completion (North Midland [11]). The reason, he said, was that otherwise a most unreasonable burden would be imposed on the contractor (Multiplex [52]).

Dodd v Churton shows something that surprises people: the employer need not have done anything wrong. Lord Denning MR said so in the Court of Appeal. Conduct that makes finishing on time impossible or impracticable stops the employer insisting on the date, and it may be quite legitimate conduct, such as ordering extra work. Lord Pearson approved that passage in the House of Lords in Trollope & Colls ([1973] 1 WLR 601, 607, as quoted in Multiplex [54]-[55]). So a proper variation is an act of prevention if it delays completion (Multiplex [56](i)). What matters is that the employer caused the delay, not whether it broke the contract.

Peak shows the principle at full strength. Salmon LJ could not see how an employer can insist on compliance with a condition if it is partly his own fault that it cannot be fulfilled. Note the word "partly". The defective piling, which was not the employer's responsibility, started the trouble. The employer's share of the delay was still enough to bring the date down. Unless the contract said otherwise, the employer was left to its ordinary remedy: damages it could prove it had suffered (North Midland [33]).

In modern English law the principle is a term of the contract, not a rule imposed on it. That fits what the principle is: a statement of what parties to a deadline must be taken to have agreed about it. What they actually agree can therefore change it. The Court of Appeal said in North Midland that it can only sensibly operate by way of implied terms ([28]), and that it is not an overriding rule of public or legal policy ([30]). Constable J put it in five words in Tata v DBS (TCC, 2024): It is a creature of contract. That matters because what the contract creates, the contract can shape.

One limit comes from the contractor's own delay. In Jerram Falkus (TCC, 2011) Coulson J held that the principle does not apply if the completion date would not have been achieved anyway, because of concurrent delays caused by the contractor's own default ([52]). That follows from causation: the employer did not prevent completion on a date the contractor would have missed regardless. The Court of Appeal left the point open in North Midland ([50]). Under the JCT clause in Walter Lilly, a contractor delayed by a Relevant Event was entitled to a full extension despite its own concurrent delay ([370]). Different wording, as in North Midland, can say otherwise. Chapter 8 takes this up.

It follows that an extension clause serves the employer at least as much as the contractor. Without one, an employer that causes even a little delay loses its date and its liquidated damages, and must prove its actual loss instead. With one, the date moves by the delay caused and damages run from the new date. Extension clauses exist for the protection of both parties (Multiplex [49]). The Court of Appeal has said they were designed to protect employers, not to give contractors excuses (North Midland [12]). It has also called relief for the contractor their primary purpose ([44]). The two fit together: the clause relieves the contractor, and by doing so keeps the employer's damages alive.

Narrow clauses were once thought to favour employers. As the Court of Appeal said of that belief, in fact it was not ([13]). Peak is the proof. Salmon LJ added that if the clause had provided for delay caused by the employer's own fault, the position would be different. The architect would then have extended the date, and damages would have run from the extended date.

Modern forms therefore carry a catch-all for the employer's own acts. JCT DB 2016 gives time for any impediment, prevention or default, whether by act or omission, by the Employer or anyone for whom it is responsible (cl 2.26.6). The exception is delay to the extent the contractor's own default caused or contributed to it. FIDIC's 2017 Red Book gives time for any delay, impediment or prevention caused by or attributable to the Employer (Sub-Clause 8.5(e)). NEC4 makes any breach of contract by the Client a compensation event (cl 60.1(18)). The SCL Protocol explains why the general words matter: English courts have held that wording such as any other special circumstances does not cover the employer's breaches (5.2).

Where the clause covers the event, the principle has nothing left to do. Acts of prevention by an employer do not set time at large, if the contract provides for extension of time for them (Multiplex [56](ii); North Midland [31]). The date moves, and the contractor must finish by the new date or pay liquidated damages (Adyard [243]). Where the clause is ambiguous, the court should lean in favour of a construction which permits the contractor to recover appropriate extensions of time (Multiplex [57]). That rule makes the machinery work. It gets the contractor its time and lets the employer keep its date.

Time at large sounds like a prize for the contractor, and it is less than it seems. The contractor escapes liquidated damages. It must still finish within a reasonable time, and the employer can still recover the damages it can prove. May LJ said in the Court of Appeal in Shawton (2005) that a reasonable time has to be judged as at the time when the question arises, in the light of all the circumstances ([69]). The party who says the other overran a reasonable time must prove what that time was, leaving out delays the other party itself caused ([33], [72]). Whether a cap on liquidated damages also limits those general damages has no clear English answer; chapter 9 discusses it.

The plea is harder to win than it sounds. Modern clauses are wide, and courts read them so that they work. Two arguments often fail. The first is that a missed notice sets time at large. Jackson J doubted that, because otherwise the contractor could set time at large at its option by ignoring the notice clause (Multiplex [103], obiter). HHJ Stephen Davies took the same view in Steria (TCC, 2007) ([96]). Chapter 4 covers notices.

The second argument is that the certifier's failure to decide an extension that is due sets time at large. The likely answer is that it does not, and that the contractor should have the extension decided by an adjudicator, arbitrator or court. In Multiplex Jackson J held that the machinery remains fully operational despite the delaying events ([88], [91]). But it was still being operated there. I know of no English case that decides what happens when a certifier simply fails to act. The SCL Protocol warns that the mechanism may then fail (5.1). Chapter 5 returns to it.

Because the principle is a term of the contract, the parties can shape it. In North Midland the parties amended the JCT Design and Build 2005 form. Delay caused by a Relevant Event, if concurrent with delay for which the contractor was responsible, shall not be taken into account ([6]). The contractor argued that the prevention principle overrode the clause. Coulson LJ gave five reasons for rejecting that ([29]-[36]). Four said the principle was not engaged at all. It is not a rule of public policy. Time was not at large, because the contract gave extensions for the employer's acts. The old prevention cases had nothing to do with concurrent delay. And the clause merely reversed two first-instance decisions for that contract. The fifth, perhaps the most important of all, was that parties may contract out of some or all of the principle's effects ([36]).

A clause like that does not pretend the employer's act is not prevention. The Court of Appeal still called the event a particular type of prevention ([38]). The clause allocates the risk of it to the contractor, which the parties were entitled to agree ([39]). Chapter 8 covers concurrency clauses.

Once an event has delayed completion, the extension is the delay it caused. That follows from the clause's purpose: the date moves so that the contractor does not pay damages for the employer's delay, and no further. Many clauses, including the JCT forms, say the new date must be "fair and reasonable". Those words give the decision-maker room to estimate. They do not license a discount.

Akenhead J said so in Walter Lilly (TCC, 2012). If an event has actually delayed the works by, say, 10 weeks, an extension of anything other than 10 weeks would be an extraordinary state of affairs ([363]). The words "fair and reasonable" do not imply apportionment; the test is primarily a causation one ([370]). He said that about concurrent delay under a JCT clause, and a differently worded contract can give a different answer. Nor is the grant a favour. Under that clause the word "shall" meant the architect must grant the extension, and no method was prescribed: the architect has to do the best that he or she can ([125]).

The SCL Protocol makes the same point from the other side: the analysis should not start from a position of considering whether the Contractor needs an EOT to avoid liquidated damages (Core Principle 5).

In the Scottish case City Inn (Inner House, 2010), Lord Carloway said the words go to fixing a new date once causation is already determined ([114]). He was dissenting on that point. The majority allowed the delay to be apportioned, which English courts do not follow (Walter Lilly [370]; chapter 8).

The contractor has to prove the delay; in Tata v DBS a claim failed for want of that proof, on that contract's wording ([222]). Chapter 6 explains how delay is proved.

Sometimes the employer's event arrives when the contractor is already behind through its own fault, in culpable delay. The same principle gives the answer. The extension is the delay the event caused to completion, added to the completion date as it then stood. The contractor's earlier lateness stays at its own risk, and the employer's event does not wipe it out by arriving later. The SCL Protocol says an employer delay does not exonerate the Contractor for all its delays prior to that Employer Delay, and that any extension should simply be added to the contract completion date (10.14).

The courts call this the net method. In Balfour Beatty v Chestermount (1993) the contractor was already late at its own risk when the employer instructed variations. It argued for an extension running from each instruction for the time each variation needed, what Hamblen J later called the gross basis (Adyard [268]). Colman J rejected that. The question, he said, is whether the event has caused delay to the completion of the Works and, if so, how much delay (as quoted in Walter Lilly [364]). Hamblen J gave an example: on the net basis, a variation needing 2 months would move the date to 12 November 1989 ([269]). That is months before any of the variations was instructed ([267]). So on the net method the new date can fall before the instruction.

The model project shows why the method matters. The contractor's M&E installation overruns by 5 weeks, its own risk, so the forecast finish slips from week 34 to week 39. That is 3 weeks past the contract date. In week 38 the employer instructs extra fit-out work that adds 2 weeks, and the job finishes in week 41. On the net method the extension is 2 weeks, added to week 36. The new date is week 38, and the contractor pays liquidated damages for weeks 38 to 41. On the gross basis the date would run to week 40, 2 weeks after the instruction, and the contractor would pay for 1 week only. The timing of the instruction would have bought back 2 weeks of the contractor's own delay.

Nor can the contractor blame the whole overrun on whatever happened last. One can not identify the last of a number of events and treat it as the cause of all the delay. One looks at what critically delayed the works as they went along (Walter Lilly [365]).

Two neighbouring rules are covered elsewhere. An employer's event that lands inside the contractor's delay and does not move completion at all gives no time: it simply has no effect upon the completion date (Royal Brompton [31]; chapter 8). And an employer that instructs a variation after the completion date, during culpable delay, may lose its liquidated damages. That happens if the contractor then catches up at its own cost, so that the variation becomes the effective cause of delay (SCL 10.13; chapter 10).

Everything above applies one level down. A subcontract has its own completion date and usually its own remedy for lateness, and the main contractor stands where the employer stood. The SCL Protocol says that when it is applied to a subcontract, it is the main contractor that is being referred to as the Employer (Appendix A). So the prevention principle protects a subcontractor against the main contractor. Jackson J said so in Multiplex: The same principle applies as between main contractor and sub-contractor ([48]). Multiplex was itself a dispute between a main contractor and its subcontractor.

The JCT sub-contracts provide for this directly. SBCSub/C 2016 gives the subcontractor time for impediment by the employer (cl 2.19.8), and also for any impediment, prevention or default, whether by act or omission, by the Contractor (cl 2.19.9). A main contractor that holds up its subcontractor, for example by giving late access to an area, owes it time, just as the employer owes time to the main contractor.

The main contractor's other concern is to keep its two contracts in step. If the employer gives it time for an event, it wants the subcontractor to get time for the same event. Otherwise it may owe a subcontractor time it cannot recover from the employer. So many subcontracts give time for any circumstance which entitles the contractor to an extension of time under the main contract (Steria [19], [95]). The subcontract in ISG v FK (TCC, 2024) did the same, excluding events caused by the subcontractor's own breach (clause 9(5)(a), set out at [9]). This is what "back-to-back" means: the back-to-back clause passes the main contract's time down the chain.

Notices are where back-to-back arrangements break. The subcontract's periods are often different from the main contract's. NEC4's subcontract gives the subcontractor seven weeks to notify a compensation event, where the main contract gives eight (cl 61.3). The JCT sub-contract gives the contractor 16 weeks to decide an extension, where the main contract gives the employer 12 (cl 2.18.2; DB 2016 cl 2.25.2). I would check that the subcontractor's notice arrives in time for the main contractor to give its own. In Steria written notice within a reasonable period was a condition precedent to the subcontractor's extension. The judge said a missed notice would not set time at large ([90]-[91], [96]). Chapter 4 deals with notices.

Money follows the same logic, and chapter 13 covers it. Neither JCT 2016 sub-contract has a liquidated damages clause. A late subcontractor pays the contractor's direct loss and/or expense caused by the failure, once the contractor gives notice to that effect (cl 2.21). The main contractor's own liquidated damages pass down, if at all, as part of that proved loss. In Fluor v Zhenhua (TCC, 2018) the judge said a supplier would have been liable for the main contractor's liquidated damages if its delay had been critical. Later events, he added, could not eliminate that liability ([612]). The claim failed because the supplier's work was in float ([614]), so the remark was not needed for the decision.

Notices can make or break an extension, and chapter 4 explains them; chapter 5 covers the person who decides. Proving the delay is chapter 6, and two causes of one delay are chapter 8.