Two Weeks Gone, and the Completion Date Has Not Moved

Six wet days, an eleven-day wait, and nothing on paper

The dig ran into the back half of February and lost six working days to standing water — not to rain falling, to a formation that would not take plant the morning after it stopped. Separately, the revised setting-out for the steel was needed by the fourth and arrived on the fifteenth, and the frame could not be released against a drawing that was going to change. The job is a fortnight behind, everybody on site agrees it is a fortnight, and the completion date is where it was in December.

That last clause is the whole problem. A completion date does not drift because a job drifts. It moves only when somebody with the power to move it says so in the way the contract requires, and until then every day past it is a day on which damages accrue at the rate the contract fixed. Nobody has to prove you were at fault: liquidated damages are payable on the fact of late completion, which is why they exist at all.

So the fortnight is not one problem. It is two events with different names, different tests, different clocks and — the part that catches most contractors — different outcomes, because one of them will buy you time and nothing else. Writing both up in one letter as delays to date is how a good claim and a weak claim get assessed together at the level of the weak one.

Time and money are two claims and they fail separately

The JCT vocabulary does the sorting if you let it. Relevant Events entitle you to an extension of time; Relevant Matters entitle you to loss and expense. The lists overlap heavily but they are not the same list, and here the weather is what separates them: exceptionally adverse weather conditions sit among the Relevant Events and not among the Relevant Matters. That is the drafter saying the employer will carry the date and you will carry the cost. Six days of standing water may well move your completion date. On an unamended form it will not pay for the six days.

The late setting-out is the other case. Information issued later than it should have been is both — it moves the date and opens the loss and expense door — because its cause lies on the employer's side of the line rather than in the sky. Which is the practical reason to serve two applications rather than one: put them in the same document and a weak weather half drags down the half that was worth money.

The other forms draw the line elsewhere. NEC4's core clause 6 collapses the distinction on purpose: a compensation event is quoted for as a change to the Prices and to the Completion Date in one document, forecast forward and settled while it is happening. AIA Document A201 keeps time in Article 8 but pushes any claim for additional time or money through Article 15, inside a fixed window — twenty-one days in the current General Conditions. FIDIC is hardest of the four: twenty-eight days from when the contractor became aware or should have become aware, drafted so a late notice loses the entitlement. Obrascon Huarte Lain v Attorney General for Gibraltar put that second limb under a microscope, and it reads badly for anyone who waited until they were sure.

The clock started before you noticed it had

Almost every form measures the notice period from awareness, not from measurement, and that is the most expensive misunderstanding in this area. The trigger is the point at which it became reasonably apparent that progress was being, or was likely to be, delayed — not the point at which you finished working out by how much. On the steel, awareness arrived on the fourth, when the drawing did not. Not on the fifteenth when it came, and certainly not in April when erection finally slipped.

Whether lateness is fatal depends on the form and on what has been done to it. The unamended JCT notice is not generally a condition precedent: the assessor's duty is to grant a fair and reasonable extension, and lateness goes to what you can still prove rather than to whether you may ask. FIDIC's is a condition precedent and says so. NEC's eight-week bar is one in effect. And an amendment can turn a permissive JCT notice into a hard bar without using the words — Steria v Sigma Wireless is the authority that clear conditional wording is enough. Read the amendment schedule, not the printed form.

So split what you send. The notice is short, goes out immediately, names the event and the date it arose, and contains no arithmetic. The particulars follow when they exist. Several forms start the assessor's own response deadline from receipt of the particulars rather than from the notice, so a notice with nothing behind it also stops the other side's clock from ever starting.

1. The event in one sentence, using the contract's own words for it — not "bad weather" but the term the form actually lists.
2. The date it arose and the date it became apparent that progress would be affected, as two dates, because they are not always the same one.
3. Which operations on the current programme it affects, named as they appear on that programme.
4. That particulars and an estimate of the delay will follow, under the clause cited from your own copy rather than from memory.
5. If money is also being sought, a separate letter, so the two applications can be assessed apart.

Exceptional, or just February

"Exceptionally adverse" is a comparative, and the comparison is not with the weather you hoped for. It is with what the record says that place does at that time of year: six wet days in a February fortnight in the north of England is a normal February, and the same six days in July are not. So it is a records exercise, not an argument. Met Office historic station data, and the MIDAS observations behind it in the CEDA Archive, give daily rainfall and temperature for the nearest usable station going back decades; NOAA's Local Climatological Data is the American equivalent. Pull the same fortnight for twenty years, set yours beside it, and you know before writing whether you have anything.

The NEC weather test is worth borrowing even off NEC, because it is the only common form that defines the comparison instead of leaving it to an adjective. It names a weather measurement place in the Contract Data, names the weather data the measurement is compared against, and asks whether a monthly measurement occurs on average less frequently than once in ten years. The refinement most people miss is what you then get: the entitlement is assessed on the excess over the ten-year value, not on the whole month. A February that beat the ten-year rainfall by a fifth does not buy you February. It buys you the fifth.

Causation is the second half of the test and it fails more claims than the meteorology does. Rain that fell while the gang was inside first-fixing is not delay, and nor is rain on a Saturday you were not working. What counts is weather that stopped or measurably slowed an operation that was critical at the time — which means knowing what was critical that week from an updated programme rather than from recollection. Be honest about consequences, too: six days here went to a formation that would not take plant, which is a ground condition the weather produced, and under some risk allocations that is the stronger route.

The drawing you cannot prove you asked for

Late information is the stronger of the two events and the easier to lose on procedure, because the entitlement almost always depends on having asked. The forms put it differently — an information release schedule where one exists, a duty to have requested the information neither unreasonably early nor unreasonably late where one does not — but the effect is identical. An architect who was never chased is not late, they are uninformed, and a contractor who has been ringing rather than writing has nothing to hand an assessor.

So the request-for-information register is not administration, it is the claim: every request numbered, with a date sent, the date the answer is needed by and why that date, the date it arrived, and what changed. "Why that date" is the column people leave blank and the one that does the work — the fourth was the last day the frame could be released and still arrive in week nineteen, and that reasoning has to have been written at the time, not reconstructed in October. Pricing the instruction belongs to the guides on instructing and on pricing a variation; what this one adds is that the two clocks expire at different times, and the notice goes first.

Concurrent, and which half of the claim it kills

The awkward fact about this fortnight is that the two events overlap. The steel was waiting on a drawing across days when the dig could not have progressed anyway, and an assessor will notice that before you do. Concurrency is where contractor claims quietly lose their money half while keeping their time half, and knowing that in advance changes what you ask for.

The mainstream English position runs from Henry Boot Construction v Malmaison Hotel: where an employer-risk event and a contractor-risk event each independently cause the same period of critical delay, the extension is granted notwithstanding the concurrent cause. Time is given. Money generally is not, because you cannot show the employer's event caused a loss your own event would have caused anyway. That asymmetry is the reason the two applications should never have shared an envelope.

Two qualifications matter enough to look up. In Scotland, City Inn v Shepherd Construction opened the door to apportioning a concurrent delay between its causes rather than granting the whole of it. And North Midland Building v Cyden Homes confirmed that a contract may allocate concurrency by drafting: a clause saying no extension is due where a contractor-risk delay runs concurrently with a relevant event is enforceable and does not offend the prevention principle. If that clause is in your amendment schedule you need to know now, not in the assessment meeting.

The move here is to separate the events by date and by operation before anyone does it for you: days on which only the weather was operating, days on which only the information was, and days on which both were. That third column is what will be argued about, and being the person who drew it beats being the person who denies it exists.

Records that will still be worth something in eleven months

Walter Lilly v Mackay gets quoted at contractors for a reason: the claim that succeeded was the one supported by contemporaneous records, and the court was blunt about what their absence costs. A record made on the day is evidence. The same fact reconstructed eleven months later out of memory and invoices is an assertion, and it gets assessed as one.

The set that matters is small and boring. A daily diary with weather, who attended, what they did and what stopped. Labour allocation sheets by operation, which is what turns "we lost time" into hours against a named activity. Photographs with reliable dates. The RFI register with its date-needed column filled in. Delivery notes and the fabricator correspondence. And underneath it a programme — a baseline the employer has seen, with updates saved at intervals rather than overwritten, because an analysis that cannot show what the plan was before the event has nothing to compare the after against.

The vocabulary of that comparison is worth knowing even on a job that will never see an expert. The Society of Construction Law's Delay and Disruption Protocol argues throughout for assessing entitlement while the facts are still visible. AACE International's recommended practice on forensic schedule analysis names what each recognised method needs in order to be run at all, which doubles as a checklist of the records you should have been keeping.

What actually makes liquidated damages fall over

Contractors reach for the word penalty far too early. The old formulation in Dunlop Pneumatic Tyre v New Garage — that the sum must be a genuine pre-estimate of loss — was replaced by the Supreme Court in Cavendish Square Holding v Makdessi and ParkingEye v Beavis with a different question: whether the clause imposes a detriment out of all proportion to the innocent party's legitimate interest in performance. That is a much harder test to fail, and a rate that merely looks generous is unlikely to be struck down for being generous.

The provisions that do fall over fall over on machinery rather than on amount. The prevention principle is the one to know: where the employer causes delay and the contract has no adequate extension of time machinery for it, time can be set at large, the fixed date disappears, the obligation becomes one to complete within a reasonable time, and the damages go with the date. Peak Construction v McKinney Foundations is the source; Multiplex v Honeywell drew its limits, including the unwelcome half — a contractor who fails to operate a valid notice provision cannot generally use their own failure to put time at large.

Two further points travel into the meeting. The deduction has a procedure of its own — extension assessed, non-completion notified or certified, written notice of intention to withhold served before money moves — and skipping a step frequently defeats it even where the delay was real; that procedure belongs to the guide on settling a final account. And where a contract is terminated before completion, Triple Point Technology v PTT settled that liquidated damages accrue up to termination rather than evaporating.

None of that is a reason to run the argument. It is a reason to serve a valid notice, because almost every route by which liquidated damages come off depends on the extension of time machinery having been operated properly — and the party who operated it properly is you.