Chapter 4: Notices and time bars

A notice exists to protect the party who will pay, and the judges say so. Jackson J in Multiplex (TCC, 2007) said that prompt notice enables matters to be investigated while they are still current. Constable J in Tata v DBS (TCC, 2024) added mitigation: a notice regime gives the employer the opportunity to engage in the mitigation of delay it knows it will be blamed for.

The Privy Council, in Uniform Building Contractors (2026), added certainty: claims for more money should be clearly set out and promptly made ([69]). The SCL Protocol, which is industry guidance rather than law, recommends dealing with the time impact of employer risk events, so far as possible, as the work proceeds (Core Principle 4).

Those purposes explain why notice clauses come in different strengths. A clause can do one of three things. It can make the right depend on the notice, so that no notice means no claim: a condition precedent. It can impose a lesser sanction, cutting the award down to what timely notice would have produced. Or it can simply tell the contractor to give notice and say nothing about what happens if it does not. The next section explains how to tell which you have.

The lesser sanctions work like damages inside the contract. The contractor keeps its claim but bears the cost that its own silence caused. NEC4 says that if the contractor failed to give an early warning it should have given, the compensation event is assessed as if the Contractor had given the early warning (clause 63.7). FIDIC's 2017 forms do something similar for failures other than the 28-day notice and the 84-day statement of the legal basis. The award must take account of how far the failure prevented or prejudiced proper investigation of the claim (Sub-Clause 20.2.7).

A bare obligation is the weakest kind. A late notice is then a breach of contract, but the claim survives. In principle the employer could recover damages for whatever harm the silence caused. No English decision has awarded them. In Steria (TCC, 2007) the judge thought that in many cases such a loss would be difficult if not impossible to prove, and left the point open.

Even where lateness is not fatal, a thin notice costs something. The certifier, whoever the contract names to decide extensions, cannot award what it has not been shown, and an extension is hard to withdraw. So the SCL Protocol expects a certifier without the information to award only the minimum EOT that can be justified at the time (4.5). Notices and records are one discipline (chapter 14).

Why should a court be slow to read a notice clause as a bar? The reason is that a bar takes away a valuable right: time the contractor has earned, or money for the employer's own risk events. Parties do not usually give up valuable rights without saying so, and the courts look for clear words. But parties are also free to allocate risk as they choose. If they have clearly agreed that no notice means no claim, the court enforces that bargain, even against a claim that is good on the merits. Both halves point to the same place: the words.

The Court of Appeal set out the test in 2025 in DBS v Tata. It is the appeal from Tata v DBS, the same dispute with the parties' names reversed. Coulson LJ drew five points from the cases, at [26].

The five points are these. The answer turns on the precise words used, read in their contractual context. There must be something that makes the relief conditional on the requirement. Clear words will usually be needed, but the clause need not call itself a condition precedent. The link will usually be expressed in the language of obligation, "shall", but that is not enough on its own. And the requirement need not be a fixed number of days: "timely", "promptly" or "within a reasonable time" can do ([26] point (e), [50]-[51]). A Court of Appeal decision binds the High Court. The contract was for an IT system, not a building, but the test is general contract law, drawn from cases on many kinds of contract ([26]).

Lewison LJ, agreeing, gave the most useful handle. A sentence whose structure is "if-then" is the paradigm of conditionality. The part after "then" is not reached until the "if" is satisfied ([75]). "Unless ... not" does the same work in the negative ([81]). So read a notice clause for its grammar. Find the words that link the notice to the right, and ask what sits on each side of them.

The cases the Court of Appeal reviewed show the test at work. The sub-contract in Steria extended time if delay arose from listed causes, then in any such case provided the Sub-Contractor shall have given within a reasonable period written notice of the circumstances. That was treated as a condition precedent, even though it gave no express warning of the consequence ([91]). By contrast, a force majeure clause in Scottish Power v BP told a party claiming relief what it "shall" do. It did not say, as it easily could have, that the party must do those things to get relief. It was not a condition precedent ([19]-[20]).

The test cuts both ways. In DBS v Tata it was the employer that lost. The Disclosure and Barring Service had engaged Tata Consultancy Services to modernise its IT systems. Clause 6.1 said that if a milestone was missed through TCS's default, DBS was to issue a Non-conformance Report promptly, and would then have a set of options. One was to demand Delay Payments, akin to liquidated damages. DBS never issued a report. It lost its claim to about £1.592m of Delay Payments, because the failure to provide one was fatal ([3], [70]). Employers under FIDIC face the same risk. The 2017 forms put both parties under one claims procedure, and the employer may claim payment or set off only by complying with this Sub-Clause 20.2. JCT employers need their own notices in the right order before they can deduct liquidated damages (chapter 9).

Once a clause is a condition precedent, the court applies it fairly, not harshly. Akenhead J in Obrascon saw no reason to construe the FIDIC clause strictly against the contractor, and reason to read it reasonably broadly, given its serious effect on what could otherwise be good claims ([312]). The employer relies on the bar to defeat the claim, so it must prove the facts that trigger it: that the notice was late ([313]).

Ambiguity also helps the party that would lose the right, usually the contractor. It now does so through the need for clear words, not the old rule that ambiguous words are read against the party relying on them. In Triple Point (UKSC, 2021), Lord Leggatt, with Lord Burrows, adopted in a concurring judgment the rule that the more valuable the right, the clearer the language will need to be. The Court of Appeal in DBS v Tata applied that reasoning to conditions precedent by analogy ([26]). The judge in Steria had reached the same place earlier: genuine ambiguity about whether a notice is a condition precedent means it is not one ([89]). As with his other notice rulings, the point was not needed for the result.

A notice must do its job: tell the other side that a claim is coming and what it is about. The test of content is practical. Under the 1999 FIDIC clause no particular form was needed. The notice had to be in writing to the Engineer, describe the event or circumstance, and be recognisable as a "claim" (Obrascon [313]).

That is why the January letter in Obrascon failed. It warned the Engineer about ponded water, so ponded water is what the Engineer would have looked into. It gave no warning of the delay the contractor later proved, from rain falling on critical work in December. A notice protects the employer only for the delay it describes, so it preserves the claim only for that delay. A progress report that mentions a problem is not a claim at all.

The clause sets the content, so read it. Under the Steria sub-contract the notice had to identify the circumstances and say they had caused delay ([81]). It did not have to make it clear that it is a request for an extension of time. Some clauses require the contractual basis of the claim. The Hong Kong Court of Appeal held in Maeda v Bauer (2020) that the notice must then state the basis relied on, but it may identify more than one basis in the alternative ([53]). That decision is persuasive only in England, but the practice costs nothing.

When does the clock start? Under FIDIC it runs from when the contractor became aware, or should have become aware, of the event or circumstance (Obrascon [311]). The 2017 edition uses the same test for either party (Sub-Clause 20.2.1). A party that ought to have known is treated as if it knew, so not looking does not stop the clock.

Akenhead J read the 1999 clause generously. An extension is due where completion is or will be delayed, so it can be claimed either when it is clear that there will be delay or once delay has started. He held that notice need not be given until there actually is delay, although a contractor can give it earlier with impunity. The "event or circumstance" can mean the incident or the resulting delay ([312]). That is one first-instance judge's reading of one form, and I would not rely on it. Start counting from the day you knew, or should have known, of the event. Give notice early, and give it again if the picture changes.

A time bar removes only what its words reach, because a party keeps any right it has not clearly given up. So read what the bar takes away as well as what triggers it: time, money, the contractor's defences, or damages for the same delay claimed outside the clause.

Tata v DBS shows how much turns on this. Clause 5 of the IT contract required TCS to notify likely delay as soon as reasonably practicable and then serve a draft Exception Report within 5 Working Days. Clause 5.6 said DBS would not be liable to compensate TCS for delays said to be DBS's responsibility unless TCS had complied. Constable J held that this condition precedent applies only to DBS's liability to compensate TCS. It did not stop TCS getting relief from the delay, or defending itself against DBS's claims for Delay Payments or damages, if it proved the delay was DBS's responsibility ([76]).

The bar did reach damages. That is a question of construction ([77]). The words were wide and tied to the delays themselves, not to one contractual right. Allowing the same claim at common law without notice is uncommercial, the judge said. A clause linking the notice only to the express right to loss and expense would have been different ([79]-[80]). As it turned out, DBS was estopped from relying on the missed notice, and TCS's claim then failed on proof ([163], [222]). The ruling on damages is first-instance and was not appealed.

JCT's loss and expense clauses are drafted the other way. Clause 4.23 of Design and Build 2016 says that the loss and expense provisions shall not limit or affect any other rights and remedies of the Contractor. That is the narrower kind of link Tata had in mind, so a damages claim may survive a missed JCT loss-and-expense notice. In WW Gear (TCC, 2010) the employer conceded as much on a different JCT form, and the judge approved the concession. No court has decided the point for the current forms. The Final Statement may still shut the claim out at the end (clause 1.8.1.3; chapter 13).

Is the JCT loss-and-expense notice itself a condition precedent? Clause 4.19 makes the right to be paid subject to ... compliance with clause 4.20, the notice clause. That is the language of condition, and on the DBS v Tata test it leans towards a condition precedent. No case has decided it. In Walter Lilly (TCC, 2012), on an amended JCT 1998 clause, both counsel accepted that a timely application was a condition precedent to loss and expense, and Akenhead J applied that ([463]-[464], [486]).

The JCT time clause leans the other way. The contractor must give notice forthwith when progress is being or is likely to be delayed, identifying any Relevant Event (clause 2.24.1). But after practical completion the employer must review the completion date. It may then fix a later date whether or not the Relevant Event has been specifically notified (clause 2.25.5.1). A contractor that missed a notice still has a route to its extension at the end. The same was true of the 1963 JCT form (1971 revision). As the Court of Appeal described it, Vinelott J held in Merton v Leach that the architect had to extend time whenever he knew of qualifying delay, notice or no notice (DBS v Tata [21]). I read both as pointing against a complete time bar in the JCT time clause. That is an inference; no court has decided it.

NEC4 is explicit. If the contractor does not notify a compensation event within eight weeks of becoming aware that it has happened, the Prices, the Completion Date or a Key Date are not changed (clause 61.3). The exception is an event arising from the Project Manager's or Supervisor's own instruction, notification, certificate or changed decision. The employer's side made that change, so it already knows. The contract's changes are the parties' only rights in respect of a compensation event (clause 63.6); whether that excludes damages for the same event is undecided. No English court has ruled on how 61.3 operates. In NIHE v Healthy Buildings, on the NEC3 Professional Services Contract, whose exception is worded differently, the Northern Ireland Court of Appeal assumed the eight-week rule was a bar. It held that an employer that should itself have notified the event could not rely on it ([30]). The decision is persuasive only in England.

FIDIC's bar is the widest. Under the 2017 forms, a claiming party that misses the 28 days loses the extension or payment it claimed, and the other party is discharged from any liability in connection with the event (Sub-Clause 20.2.1). The 1999 form said the same ([311]). There is one gap. An extension for a Variation needs no requirement to comply with Sub-Clause 20.2 (Sub-Clause 8.5(a)). That makes sense: the Engineer ordered the Variation, so nobody needs to be told about it.

The contractor's first instinct is usually the prevention principle. If the employer caused the delay, can it still hold the contractor to the original date because a letter was late? The principle says that a party cannot insist upon the performance of an obligation which he has prevented the other from performing. If it applied, time would be at large: the completion date would fall away, replaced by a duty to finish within a reasonable time, and no liquidated damages could run.

In England the answer so far is no, and the reason is ordinary. Extension clauses exist to avoid the operation of the prevention principle (Multiplex [49]). The contract gave the contractor a way to move the date for the employer's delay. The contractor lost it through its own failure to give notice, not through anything the employer did. If a missed notice set time at large, the contractor would be better off breaking the notice clause than keeping it. Steria called that a commercial absurdity ([95]). Chapter 3 explains the principle itself.

The contrary view came from Australia. In Gaymark (Northern Territory, 1999) the contractor's extension for 77 days of employer delay was barred because it had not strictly complied with the notice rules. An arbitrator held that the employer's delay nonetheless set time at large, so the employer lost its liquidated damages. Bailey J refused leave to appeal, finding no manifest error of law (Multiplex [97]-[98]). Jackson J in Multiplex had considerable doubt that Gaymark represents the law of England, since otherwise a contractor could set time at large at its option ([103]). His view was obiter: on his sub-contract, losing the extension did not automatically expose the sub-contractor to delay damages ([104]). The judge in Steria held that a missed notice does not put time at large ([96]), though the point was not needed there. No English appeal court has decided it. The Court of Appeal in North Midland (2018) did hold that where the contract provides an extension for the employer's events, time was not set at large. It also held that parties may contract out of the principle's effects ([31], [36]). That was a case about concurrent delay, not a missed notice, but the reasoning points the same way. The English view rests on two first-instance statements, neither needed for the result, which no court has doubted.

The real routes back are narrow and turn on the facts. A party can give up a right, which is waiver. Or it can be stopped from relying on a right it led the other side to believe it would not use, if the other side relied on that to its cost: estoppel. A contractor's complaint that the employer should not keep extra work without paying for it is usually based on waiver and estoppel.

It rarely works after the event, and Uniform Building Contractors shows why. The contractor laid 28.43 km of water pipeline in Trinidad on the 1999 FIDIC design-and-build form ([1], [5]). It said four items of work were variations, knew of them by the end of 2007, and made no claim under clause 20.1. The Court of Appeal of Trinidad and Tobago awarded it about TT$13.9m. The Privy Council restored the trial judge's dismissal: the contractor could make no claim, and the employer was discharged ([2], [66]). Waiver and estoppel were raised for the first time on appeal, which was much too late. They must be pleaded and proved by clear evidence ([72]). Getting on with the works without claiming was more consistent with there being no representation, reliance or detriment ([71]). Privy Council decisions are persuasive in England, not binding.

Estoppel can succeed where the facts are strong. In Tata v DBS TCS missed the 5 Working Days, but both parties had proceeded on precisely the same assumption, that the five-day requirement had fallen away. The assumption was communicated, TCS relied on it to its detriment, and DBS had not reserved its position ([148], [159]-[163]). DBS was estopped from relying on the missed report. In ISG v FK (TCC, 2024) a main contractor had engaged with a sub-contractor's early warnings for months without complaint. The sub-contractor had an arguable case of waiver and/or estoppel, which the court declined to decide on that application ([45], [47]).

Only the parties can waive a contractual requirement, or someone with authority to act for them (Uniform Building Contractors [84]). The FIDIC Engineer is not a party. The 1999 form gave him no authority to amend the contract or, except as the conditions stated, to relieve a party of its obligations. So the Engineer had no authority to vary or waive the claims procedure, whatever he said ([77], [79]). The 2017 form keeps that rule (Sub-Clause 3.2). A friendly engineer's reassurance is not a waiver.

What FIDIC 2017 adds is a route inside the contract. If the Engineer does not say within 14 days that a Notice of Claim was late, it is deemed to be a valid Notice, though the other party may still dispute that (Sub-Clause 20.2.2). If the Engineer does say so, the claim is still determined. The determination decides whether or not the Notice of Claim shall be treated as a valid Notice (Sub-Clause 20.2.5). The review may take into account why the notice was late, whether the other party would be prejudiced, and what that party already knew, but it is not bound by them. That brings the bar back to its purpose: an employer that knew all along and lost nothing has less reason to rely on it. The Privy Council said, in passing, that the 2017 clause is slightly more flexible, but still a condition precedent ([62]). No English court has yet applied 20.2.5.

Termination does not help. Rights that accrued before termination survive it, and so does the loss of a claim that was barred before it. In Uniform Building Contractors termination could not in law resurrect claims that had not been made in time ([67]-[68]).

The same event can be safe under one standard form and lost under another. Take the model project used throughout this guide, an invented four-storey office building (illustrative only). It is planned to finish in week 34, against a contract completion date of week 36. The employer is to supply the roof design. In week 14 the contractor's site manager sees that the design will be late and will hold up the roofing.

Under the FIDIC 2017 Red Book, the contractor gives a Notice of Claim. It is due as soon as practicable, and no later than 28 days after the contractor became aware, or should have become aware, of the event. A fully detailed claim follows within 84 days (Sub-Clause 20.2.4). If that claim does not include a statement of its contractual or legal basis in time, the Notice of Claim is deemed to have lapsed. The 1999 Sub-Clause 20.1 is broadly the 2017 Sub-Clause 20.2.1. Quotations are from the 2017 first printing; the 2022 reprint differs in places.

Under NEC4, either side gives an early warning as soon as either becomes aware of a matter that could delay completion (clause 15.1). The contractor then has eight weeks to notify the compensation event (clause 61.3). If the late design had come from the Project Manager's own instruction, the bar would not apply.

Under JCT Design and Build 2016, the contractor gives notice forthwith of the material circumstances, including the cause of the delay, and identifies any Relevant Event. Particulars, including an estimate of the delay, follow as soon as possible (clauses 2.24.1-2.24.2). There is no fixed number of days, and "forthwith" does not settle whether the clause is a condition precedent: under DBS v Tata [26] a condition needs no fixed period. JCT says that in the 2024 edition the drafting for notification provisions has been improved, and the employer now has 8 weeks, not 12, for the interim decision.

These are the unamended forms. Check for amendments: a bespoke notice clause can turn any of them into a strict bar. Chapter 16 compares the forms in more detail.

A good notice protects the extension that chapter 3 explains and the certifier decides. The records that make a notice good are the subject of chapter 14.