The short answer: every date in the Gregorian calendar has one fixed weekday, and any of three methods will find it. Zeller’s congruence, the Doomsday rule, and Sakamoto’s algorithm all compute the same number by different routes. The date 2026-11-14 is a Saturday, and no time zone can change that — only a wrongly parsed timestamp can.

Key takeaways

  • Weekdays are modular arithmetic. Seven days repeat, so the answer is always a remainder.
  • Three good algorithms. Zeller for formulas, Doomsday for mental math, Sakamoto for code.
  • January and February are special. Most formulas treat them as months 13 and 14 of the previous year.
  • Leap years shift the calendar. The same date moves one weekday forward, or two across a February 29.
  • Numbering is not universal. ISO calls Monday 1; JavaScript calls Sunday 0.

Every date has a weekday, and it is fixed

The Gregorian calendar is a deterministic rule, so a date and a weekday are locked together. 2026-11-14 is a Saturday in every country that uses the Gregorian calendar, and it was a Saturday in 2026 whether or not the time zone had reached midnight. The only way to get a different weekday is to change the date — which is exactly what a careless timestamp conversion does.

Because the cycle is seven days long, the problem reduces to arithmetic modulo 7. The useful part is not the modulus but the corrections: month lengths, the leap-year rule, and the century exceptions that break the naive “365 days is one weekday” shortcut.

Numbering the days: ISO, JavaScript, and spreadsheets

Before comparing formulas, fix the numbering. ISO 8601 numbers weekdays from Monday = 1 through Sunday = 7. JavaScript’s getDay() returns Sunday = 0 through Saturday = 6. Python’s weekday() returns Monday = 0, and its isoweekday() returns Monday = 1. Excel’s WEEKDAY defaults to Sunday = 1 unless a return type is supplied. Zeller’s congruence returns Saturday = 0.

Five conventions, five chances to be off by one. The fix is to convert once, at the boundary of the calculation, and to name the convention in the variable — iso_weekday, not day.

Zeller’s congruence

For a date with day q, month m, and year y, compute:

h = (q + ⌊13(m + 1) ÷ 5⌋ + K + ⌊K ÷ 4⌋ + ⌊J ÷ 4⌋ + 5J) mod 7

K is the year within the century (y mod 100), J is the century (⌊y ÷ 100⌋), and January and February are counted as months 13 and 14 of the previous year. The result maps to 0 = Saturday, 1 = Sunday, 2 = Monday, and so on to 6 = Friday.

Worked example for 2026-11-14: q = 14, K = 26, J = 20, and m = 11, so ⌊13 × 12 ÷ 5⌋ = 31. The sum is 14 + 31 + 26 + 6 + 5 + 100 = 182, and 182 mod 7 = 0, which is Saturday. The two divisions by 4 are integer divisions, and 5J is there because a century shifts the anchor by five weekdays across the Gregorian cycle.

The Doomsday rule

The Doomsday method is the mental-arithmetic version. Each year has a “doomsday” weekday, and several easy dates always fall on it: 4/4, 6/6, 8/8, 10/10, 12/12, the pairs 5/9 and 9/5, 7/11 and 11/7, 3/14, 2/28 (or 2/29 in a leap year), and 1/3 in a common year or 1/4 in a leap year.

Find the doomsday for the year from the century anchor: 1900 is Wednesday, 2000 is Tuesday, and 2100 is Sunday. Add the last two digits of the year, integer-divide them by 4, and add both to the anchor, then reduce mod 7. For 2026: 2 (Tuesday) + 26 + 6 = 34, and 34 mod 7 = 6, so Saturday is the doomsday. November 7 is a doomsday, and November 14 is seven days later, so it is also a Saturday.

The rule is fast because the anchor dates cluster: once you know the doomsday for the year, any date is a short offset.

Sakamoto’s algorithm

For code, the compact version uses a lookup table t = [0, 3, 2, 5, 0, 3, 5, 1, 4, 6, 2, 4] indexed by month, subtracts one from the year when the month is January or February, and computes:

weekday = (y + ⌊y ÷ 4⌋ − ⌊y ÷ 100⌋ + ⌊y ÷ 400⌋ + t[m − 1] + d) mod 7

The result is 0 = Sunday through 6 = Saturday. For 2026-11-14: 2026 + 506 − 20 + 5 + 2 + 14 = 2,533, and 2,533 mod 7 = 6, which is Saturday. The table encodes the day-of-week offset of the first of each month, and the three divisions encode the leap-year corrections. It is the same arithmetic as Zeller, rearranged for a computer.

Leap-year traps

The leap rule is divisible by 4, except centuries not divisible by 400. The trap is not the rule itself but where it is applied. In Zeller and Sakamoto, January and February belong to the previous year for the purpose of the leap check, so a date in January 2028 must use 2027 in the year term. Skip that adjustment and every January and February date is wrong by a day.

The second trap is the year-to-year shift. A common year is 365 days, which is 52 weeks and one day, so the same date next year moves forward one weekday. Across a year that contains February 29, it moves forward two. 2026-11-14 is a Saturday, 2027-11-14 is a Sunday, and 2028-11-14 is a Tuesday.

The third trap is the century. 1900 was not a leap year, so the 28-year pattern that holds within a century breaks at the boundary. 2000 was a leap year, which is why the cycle continues cleanly through it.

Daylight saving is irrelevant to all of this. It moves the clock by an hour, never the calendar date, so it cannot change a weekday.

The 28-year and 400-year cycles

Within a century, the pattern of weekdays repeats every 28 years because 28 is the least common multiple of the 7-day week and the 4-year leap cycle. The Gregorian calendar as a whole repeats every 400 years: 400 years contain 146,097 days, which is exactly 20,871 weeks, so dates and weekdays line up again with no remainder.

That exact division is why the Gregorian calendar drifts against the seasons by only about a day every 3,000 years, and why a date table for 2026 works unchanged for 2426.

Common mistakes

  • Using JavaScript’s getDay() (Sunday = 0) where ISO weekday numbering is expected.
  • Forgetting that getMonth() is zero-based while the day of the month is not.
  • Parsing 2026-11-14 as UTC midnight and then reading the weekday in a negative-offset zone, which returns Friday.
  • Not shifting January and February in Zeller or Sakamoto.
  • Using the wrong output map for Zeller, where 0 means Saturday rather than Sunday.
  • Assuming every year shifts the weekday by one, ignoring February 29.
  • Treating 1900 as a leap year because it is divisible by 4.

Worked examples

Date Weekday Check
2000-01-01 Saturday Century starts on a doomsday
1900-01-01 Monday Common century year
2024-02-29 Thursday Leap day exists
2026-03-08 Sunday U.S. daylight saving starts
2026-11-01 Sunday U.S. daylight saving ends
2026-11-14 Saturday ISO week 46, day 6
2100-01-01 Friday Century non-leap year

Each row can be checked two ways: with a formula and with the day-shift rule. From 2026-03-08 to 2026-11-01 is 238 days, and 238 mod 7 = 0, which is why both dates fall on Sunday.

Why the weekday matters

Scheduling systems use it constantly: a recurring meeting on the second Tuesday, a payroll week that closes on Saturday, a billing cycle anchored to the first business day. Those rules are only stable if the weekday is computed from the date rather than from a timestamp. Compute the date, then the weekday, and never the other way around.

When the answer looks wrong, check the date before the algorithm. A date that is one day off produces a weekday that is one day off, and the formula will faithfully report it.

Frequently asked questions

What day of the week was 2026-11-14?

A Saturday. It is ISO week 46, weekday 6, and the 318th day of 2026.

Why does JavaScript return 6 for Saturday?

Because getDay() numbers Sunday as 0. Saturday is the sixth index, not the seventh.

Does a time zone change the weekday?

Not for a calendar date. It changes the weekday of a timestamp when that instant is formatted in another zone.

How often does the calendar repeat?

Every 400 years exactly, because 146,097 days divide into 20,871 weeks. Within a century the pattern repeats every 28 years.

Is 2100 a leap year?

No. It is divisible by 100 but not by 400, so February 2100 has 28 days.

What is the fastest way to find a weekday by hand?

The Doomsday rule. Memorize the anchor dates and the century anchor, then count the offset in weeks.

Further reading

Keep exploring

Read next Date Countdown: Days Until, Inclusive, and Time Zones Count the days between two dates with exclusive and inclusive conventions, weeks, weekdays, and months, and see why… Read next Age Difference: Exact Years, Months, and Days Get the exact gap between two birth dates in years, months, and days, including month-end clamping, leap days,… Read next Age Calculator: Exact Age, Leap Years, and Time Zones Age is completed years, not elapsed days divided by 365. The subtraction rule, leap-day birthdays, total days, and…

Decision path

Use this guide in a decision

Open Date & Time Decision Center Open companion solver