A deadline is a piece of arithmetic before it is anything else, and the arithmetic is not intuitive. Whether the first day counts, whether Saturdays and holidays are inside the period or outside it, whether a month means thirty days or the same date next month, and what happens when the last day falls on a closure are all answered by rules that differ between forums. This subject sets out how a period is actually counted and where the counting goes wrong.
Every time limit has a trigger, and the trigger is defined by the rule rather than by common sense. Service, issue, receipt, demand and completion are all different events, and a period counted from the wrong one is wrong by however many days separate them. Finding the trigger in the text is the first step in any deadline question.
A period expressed in months runs to the same day-number in the later month rather than by adding a fixed number of days. Where the later month is shorter and has no corresponding date, the period generally ends on its last day. Periods expressed in days are counted in days regardless of how many months they span, and mixing the two conventions is a common source of error.
Where the last day of a computed period falls on a weekend, a holiday or a day the filing office is inaccessible, the period generally extends to the next day it is open. The extension applies to the end of a period rather than to days inside it, and an office being busy, understaffed or closed to visitors is not the same as being inaccessible.
A period counted back from a future event is computed in the opposite direction: the day of the event is excluded, the days are counted backwards, and a last day falling on a closure moves earlier rather than later. Additional time for service also moves the deadline earlier. The arithmetic is simple and the instinct to apply the forward conventions is strong.
Deadlines expire at an hour as well as on a date, and the hour depends on how the step is taken. Paper filing usually ends at the office's closing time, electronic filing frequently runs to midnight in the forum's time zone, and service on a party may have its own cut-off. A step taken after the cut-off is treated as taken the following day.
Within a period counts from a trigger and produces a date; by a date and not later than a date name one directly. The practical differences are whether the named day is included, whether the counting conventions apply at all, and what happens when the day falls on a closure. Reading which construction a rule uses settles a question that would otherwise be argued.
A deadline provision can be reduced to four questions: what starts it, how long it runs, how the days are counted and what follows from missing it. Reading a rule with those four questions in mind takes a few minutes, produces a date that can be checked by somebody else, and catches the assumptions that would otherwise be carried in from another forum.
Where more than one instrument sets a date for the same step, the usual order of precedence puts a specific order above a general rule and a statute above both. Where the position is genuinely unclear, the earlier date is the safe one. The real risk is not the conflict but the failure to notice that a second date exists.
Where a period runs from service and the document was posted, many systems add a fixed number of days to compensate for transit. The addition applies only to periods triggered by service, only to the methods the rule lists, and it is applied after the ordinary computation. Electronic service has narrowed it substantially in several forums.
Where a system serves filers across several time zones, it records submissions on its own clock. A party in a different zone has correspondingly less of the last day, and the shortfall is silent: the interface shows a date rather than a boundary. Service between parties in different zones raises the same question separately.
A period expressed in years runs to the corresponding date in the later year, with the twenty-ninth of February as the single recurring complication. The arithmetic is straightforward. What causes difficulty is evidence: long periods are usually limitation periods, and the trigger date sits years in the past with no contemporaneous record of it.
The general rule excludes the day of the triggering act and includes the last day of the period. Some periods count differently, particularly those measured backwards or those fixed by a specific date rather than a length. The convention is easy to state and easy to get wrong, and a single day is enough to make a filing late.
