GC 6.5 does not ask whether the job was late. It asks who caused each delay, and it sorts causes into separate paragraphs with different consequences. A schedule analysis that produces a single number for total slippage has not answered the clause. The output has to be per event, per paragraph, and dated twice.
Delay and extension of contract time in the standard Canadian stipulated price form is dealt with in GC 6.5, DELAYS, of CCDC 2, Stipulated Price Contract. The current edition is CCDC 2 - 2020, published by the Canadian Construction Documents Committee. It has not been superseded; the CCDC documents released in 2026 editions were CCDC 14, CCDC 15, CCDC 32 and CCDC 33, not CCDC 2. This note describes the structure of GC 6.5 and how to build an analysis that fits it. It does not reproduce the clause text, which is published under a paid, registration controlled licence, and it is not a substitute for reading the executed contract.
01 The paragraphs, and why the structure is the whole point
Within GC 6.5, paragraph 6.5.1 deals with delay caused by the Owner, the Consultant or anyone employed or engaged by them, paragraph 6.5.2 deals with delay caused by a stop work order issued by a court or other public authority, paragraph 6.5.3 deals with delay caused by events beyond the Contractor's control, and paragraph 6.5.6, added in the 2020 edition, deals with the Owner being reimbursed for costs arising out of a delay caused by the Contractor.
That structure is a sorting instruction. The clause is not organised around how much time was lost. It is organised around who or what caused the loss, and the consequences differ by category. It follows that the useful unit of analysis is the event, not the project, and that every event has to be attributed to one of the categories before the time consequence means anything.
Most delay submissions I see fail on this before any arithmetic is done. They are organised as a chronology of everything that went wrong, with a total at the end and an implicit invitation to attribute the total to the other side. That is not a GC 6.5 case. It is a narrative that requires the reader to do the sorting, and the reader who does the sorting is the one deciding the claim.
Attribute each event to exactly one paragraph, and record in the working papers why the other paragraphs were rejected. Paragraph 6.5.3 is residual, so an event should land there only after 6.5.1 and 6.5.2 have been considered and excluded on the documents. A submission that shows its exclusion reasoning is much harder to attack than one that simply asserts the category, and the reasoning takes an hour per event to write down while it is fresh.
What each category actually demands of the record
Paragraph 6.5.1 turns on the actor. A late drawing, on its own, is not a 6.5.1 event. A late drawing issued by the Consultant, or by a design consultant engaged by the Owner, is. The event record therefore has to name the act or omission, name the actor, and cite the document that proves both. Where the actor is a separate contractor or a supplier engaged directly by the Owner, that relationship has to be established from the contract documents rather than assumed from site practice.
Paragraph 6.5.2 turns on the instrument. The trigger is a stop work order issued by a court or other public authority. A verbal hold from an inspector, a regulator's letter expressing concern, a utility's refusal to energise, and an Owner's own stand-down direction are none of them a stop work order on the face of the paragraph. Some of those events may belong in another category. Some belong nowhere. Putting general regulatory friction into the stop work order category is a common overreach, and it is easy to expose because the analysis then has no order to produce.
Paragraph 6.5.3 is the residual category, and it is where discipline slips. Events beyond the Contractor's control is a phrase that will absorb anything an analyst is not willing to think hard about. The test that keeps it honest is whether the Contractor could have controlled the event, not whether it wanted to, and whether the record shows control being attempted. Weather claims live here and are usually lost here, because the record establishes that the weather happened without establishing that the schedule could not absorb it.
Paragraph 6.5.6 points the other way. The Owner being reimbursed for costs arising out of a delay caused by the Contractor is a category that requires the same evidentiary discipline in reverse, and it was added in the 2020 edition. See section 05 below.
02 Two dates for every event, not one
An extension of time provision runs on when a delay commenced. A CPM analysis produces a different date: the date the event began to affect the path that drives a contractual completion date. Those two dates are frequently weeks or months apart, and conflating them is the most consequential technical error in this area because it puts the notice position and the causation position on the same footing when they are not.
Consider a design change issued during procurement of an item with substantial float. The delay commences when the change disrupts the affected activity. The critical effect may not appear until the float is exhausted, possibly two updates later. If the submission dates the delay from the critical effect, it looks like a late notice on its own face. If it dates the critical effect from the commencement, it overstates the time consequence.
State both dates for every event, and label them. Date the commencement of the delay from the physical or documentary event, proved by the document. Date the effect separately, from the schedule update in which the affected path first drives the completion date. Never present the analysis output date as the notice trigger date. If those two dates are inconsistent with the notice that was actually given, that is a problem for counsel to manage, not a problem for the analyst to smooth over by choosing the convenient date.
03 The notice paragraph, and why there is no day count in this note
GC 6.5 contains a notice requirement, at paragraph 6.5.4, running from the commencement of the delay. I am not giving a day count for it here. The secondary sources I checked do not agree with each other on the figure, none of them identifies the edition it is quoting, and I have not read the primary CCDC 2 - 2020 text, which is published under a paid, registration controlled licence. Publishing a number I cannot verify is worse than publishing no number, because a number in a document like this gets diarised.
If a hard number is needed, read GC 6.5.4 off a registered copy of CCDC 2 - 2020 and confirm it against the project's supplementary conditions. That order matters. Supplementary conditions amend GC 6.5 more often than they leave it alone, and notice periods are among the first things owners shorten.
On any file with a time claim, produce a one page notice map before the analysis starts: the clause as executed, including supplementary conditions, the period it gives, the addressee it requires, the form it requires, and then every notice actually given with its date and its addressee. Do this from the contract in front of you, not from the standard form and not from memory of the standard form. The map either supports the claim or tells you which events are in trouble, and it is much cheaper to learn that at the start.
04 What the schedule has to be, for GC 6.5 purposes
The measuring instrument is the schedule the parties were using
The analysis should run on the contemporaneous update sequence, update by update, with the data dates identified. A baseline reconstructed after the fact invites the obvious objection, which is that the analyst built the instrument that produced the answer. Where the contemporaneous schedules are poor, say they are poor and analyse them anyway, with the defects stated. A stated defect is an argument about weight. A silent repair is an attack on credibility.
Calendars decide the number, and they are usually wrong
An extension of contract time is expressed in the contract's own unit of time. Scheduling software carries its own calendars, and on most projects at least one activity is sitting on a calendar nobody chose deliberately. If the analysis measures on a calendar that does not match the contract, every day count in the submission is arguable even where the causation is sound.
The most common technical defect that sinks an otherwise good GC 6.5 claim is a calendar mismatch. It produces numbers that look unanswerable and are wrong by a predictable margin, and the correction always runs against the claimant, because the claimant chose the calendar. Identify the calendar used for the measurement, list its working days and holidays, and demonstrate that it matches the contract's definition of the relevant unit of time as executed, including any supplementary condition that changes it. Do that before quoting a single number of days.
Only movement of a contractual date is an extension of time
GC 6.5 is about extension of contract time. The only movement that supports it is movement of the contract completion date, or of another date the contract actually fixes. Consumption of float that does not move a contractual date is a different subject: it may support a disruption or productivity position, and it may explain why a later event became critical, but it is not itself an extension of time. Blending the two produces a submission where the total does not reconcile to any contractual date, and that is fatal in cross-examination because the arithmetic can be walked back to the point where the blend happened.
Constraints, open ends, and the honesty of the model
Two schedule features quietly decide the result of an insertion analysis. The first is a date constraint on the completion milestone, which absorbs delay and reports none, so an analysis run on the schedule as found will understate the impact and an analysis run with the constraint removed will differ. The second is an open ended activity chain, which cannot transmit an effect to the completion date at all. Both need to be disclosed, and both need a sensitivity run showing the result with and without the feature, because the opposing analyst will do exactly that.
The minimum record
- The update sequence as issued, with data dates, and a log of logic, duration and calendar changes made between updates.
- The document that dates each event, and the document that identifies the actor for a 6.5.1 event.
- The actual start and finish record for the affected activities, from the updates rather than from recollection.
- The calendar file or definition used for the measurement, with its holiday set.
- The contract as executed, including supplementary conditions, for the clause and any defined terms the measurement relies on.
Where a piece of that record does not exist, the honest output is a narrower opinion, not a filled gap. An analysis that silently substitutes a reconstructed date for a missing record is unusable once the substitution is found, and it is always found, because the substituted dates are the tidy ones.
05 Paragraph 6.5.6 makes the analysis two-sided
Paragraph 6.5.6, added in the 2020 edition, deals with the Owner being reimbursed for costs arising out of a delay caused by the Contractor. Whatever the argument about its scope on a given file, its existence changes the posture of the analysis. Contractor caused delay is no longer only a defence to the Contractor's own claim for time. It is a category with a price attached, and the as-built record proves it with the same tools.
The consequence for how the work is done is direct. An analysis built by tagging only the events the Contractor wants to rely on produces a picture in which nothing else happened. That picture will be run again by the other side using the same updates, with the Contractor's own delays tagged instead, and the second run is usually easier to do because the Contractor's delays are recorded in its own progress data.
Run the event list in both directions on every file, whichever side you are on, before forming a view. Identify the Contractor caused delays in the same schedule, on the same basis, with the same dating discipline. If the resulting position is worse than the client hoped, that is information they need at the start, when they still have choices. Bringing it to them after the submission is filed is worse for them and worse for you.
06 Concurrency, and what the clause structure does not decide
Once both directions are analysed, periods will appear in which a delay in one category overlaps a delay in another. The paragraph structure of GC 6.5 sorts causes. On its face it does not supply a rule for apportioning a period in which two causes operate, and I am not going to tell you that it does.
I am also not going to import a rule from elsewhere. Approaches to concurrency differ between jurisdictions and between the leading practice texts, no Canadian authority on the point is cited in this note because none has been verified for it, and counsel should not assume that an approach they have seen applied in a foreign arbitration reflects Ontario law. That is a genuine gap in this note, and it is a gap in the standard form as well.
What the analyst can do without deciding the law is to make the factual position unambiguous:
- Identify each period of overlap by start and finish date, on the stated calendar.
- Show both driving chains for that period, activity by activity, from the update in force.
- State what the completion date does if each cause is treated as operating alone, and what it does with both.
- Do not present a single apportioned number as the answer unless the contract or the instructions tell you which apportionment rule to apply, and say which rule you were given.
A concurrency section written that way lets counsel argue the law on findings that will survive whichever way the law goes. A concurrency section that applies an unstated rule has to be re-run the moment the rule is challenged.
07 Time and money are different claims
An extension of contract time under GC 6.5 changes the date. It does not, by itself, produce a cheque. The recovery of costs arising from a delay runs on different contract language and on proof of the costs themselves, and the two claims fail independently: a well proved extension of time attached to unproved costs recovers nothing, and a well documented cost record attached to an unproved time position recovers nothing either.
I have not summarised the cost machinery here, because the relevant provisions vary with the edition, with the supplementary conditions, and with what actually happened on the file. The point for the schedule analyst is narrower. The analysis has to produce the period, on dates, that the cost claim will be built on, and it has to produce it in a form the quantum work can use without re-interpretation: start date, finish date, calendar, and what the period represents. A time analysis that hands over a total number of days and nothing else forces the quantum exercise to invent the period.
08 A worked mapping
The table sets out what each GC 6.5 category needs from the analysis. It is a checklist for building the submission, not a statement of the clause text or of entitlement.
| Paragraph | Subject as stated in the clause structure | What the analysis must establish, and from what |
|---|---|---|
| 6.5.1 | Delay caused by the Owner, the Consultant or anyone employed or engaged by them | The act or omission, the actor, and the relationship that puts the actor inside the category, each from a document. Then the commencement date, the update in force, and the effect on a contractual date. |
| 6.5.2 | Delay caused by a stop work order issued by a court or other public authority | The order itself, its issuer, its date, its scope, and the date it was lifted. Then the effect on a contractual date over that period. Absent an order, the event is not in this category. |
| 6.5.3 | Delay caused by events beyond the Contractor's control | The event, its dates, and why control was absent, with the exclusion of 6.5.1 and 6.5.2 reasoned on the documents. Then the effect on a contractual date, and what the schedule could have absorbed. |
| 6.5.6 | The Owner being reimbursed for costs arising out of a delay caused by the Contractor, added in the 2020 edition | The same discipline in reverse: Contractor caused events, dated, attributed, and measured against a contractual date on the same calendar, with concurrency periods identified rather than netted. |
4 rows, complete. Paragraphs described in the verified structure only. Other paragraphs of GC 6.5 are not shown.
09 What is uncertain, and what I have not verified
- The notice period in paragraph 6.5.4 is not stated in this note, for the reasons in section 03. Read it off a registered copy of CCDC 2 - 2020 and confirm it against the supplementary conditions.
- I have not read the primary CCDC 2 - 2020 clause text for this note, and no clause wording is reproduced. Everything here about the structure of GC 6.5 is limited to the subject matter of paragraphs 6.5.1, 6.5.2, 6.5.3 and 6.5.6, and to paragraph 6.5.4 being a notice provision running from the commencement of the delay.
- No case law is cited. There is Canadian authority on delay, notice and concurrency, and none of it has been verified for this note, so none is named.
- The 2020 edition is the current edition of CCDC 2. Contracts executed on earlier editions are still being performed and still being litigated, and paragraph 6.5.6 in particular is a 2020 addition. Check the edition on the file before applying any of this.
- Supplementary conditions govern. Nothing in this note survives a supplementary condition that says otherwise, and the analysis has to be built against the contract as executed.
- Whether an extension of contract time carries any cost consequence on a given file depends on provisions this note does not address.
10 Practical points
- Confirm the edition and read the supplementary conditions before forming any view on GC 6.5.
- Build the notice map from the contract as executed, before the analysis.
- Make the event, not the project, the unit of analysis.
- Attribute each event to one paragraph, and write down why the others were excluded.
- For 6.5.1, name the actor and prove the relationship from the contract documents.
- For 6.5.2, produce the order. No order, no 6.5.2.
- Give every event two dates: commencement, and first critical effect, each labelled.
- Identify the calendar, prove it against the contract, then quote days.
- Count only movement of a contractual date as an extension of time, and keep float consumption in its own section.
- Disclose completion date constraints and open ended chains, and run the sensitivity both ways.
- Run the Contractor caused events as well, whichever side you are on, because 6.5.6 gives them a price.
- Present concurrency as dated findings with both driving chains shown, and do not apply an apportionment rule you were not given.
- Hand the quantum exercise a period with dates and a calendar, not a total.