Why do months have different numbers of days?

Months have different lengths because the calendar you use today is a Roman compromise that is roughly 2,000 years old, and every fix applied to it since has been a patch rather than a redesign. February is the short one because in the earliest Roman calendar, the year ended there. It was the last month standing when the arithmetic didn’t work out, so it absorbed the shortfall — and it has carried that shortfall ever since.

The Roman year

The record shows the Roman calendar originally began the year in March, not January. That single fact explains several things that otherwise look arbitrary. September, October, November and December carry names meaning seven, eight, nine and ten — because they were the seventh through tenth months when counting started from March, not the ninth through twelfth as they are now. February sat at the far end of the year, after January, as the last month before the cycle restarted in March. When the Roman year needed to reconcile its count of days against the actual solar year, February was where the adjustment landed, simply because it was the month closest to the boundary. Being last made it the one nobody had already fixed a length for, so it became the one that flexed.

Caesar’s patch

Historians date the next major change to Julius Caesar’s reform of 45 BC, which produced the Julian calendar. It fixed month lengths close to the ones still in use today and introduced the leap day as a regular fix rather than an occasional improvisation. One popular story claims Augustus later stole a day from February to lengthen August to match Julius’s July — historians doubt this account, and the record does not support it as fact. The lengths of the months, February’s shortness included, were substantially settled by Caesar’s reform, not by a later emperor’s vanity.

The Julian calendar’s leap-year rule assumed a year of exactly 365.25 days. That was close, but not exact: the true solar year is a little shorter, so the Julian year ran long by about 11 minutes annually. Eleven minutes sounds trivial, but it accumulates. Left uncorrected, it drifted the calendar by roughly one full day every 128 years — a slow error, but a compounding one, and after many centuries it had pulled the calendar noticeably out of step with the seasons.

Why nobody fixed it since

By the 16th century, that drift had become hard to ignore. The record shows that Pope Gregory XIII’s reform of 1582 produced the Gregorian calendar, the one most of the world uses now — and notice how little it actually changed. It corrected the leap-year rule so the average year is closer to the true solar year, and it clawed back the accumulated drift with a one-off skip: in the first adopting countries, October 1582 lost ten days outright, with the 4th followed directly by the 15th. That is the whole reform — a rule tweak and a one-time correction. It did not touch month lengths, did not rename anything, and did not move the start of the year to March, where it had once been. Adoption spread over centuries rather than happening at once: Britain and its colonies did not switch until 1752, and by then the accumulated drift meant they had to skip eleven days, not ten, to catch up.

That pattern — correct the smallest possible thing, leave everything else alone — is the real reason nobody has redesigned the calendar since. Any deeper reform means renumbering or reshuffling dates that already exist in contracts, birth records, historical accounts and religious calendars going back centuries. Skipping ten or eleven days once was disruptive enough to cause real unrest at the time. Redesigning the month lengths from scratch would mean doing that kind of disruption to every date in every document ever written. The cost of a clean redesign has only grown with every year that passes, so the incentive has only ever pointed towards smaller patches, never a rewrite.

The pattern that remains

What’s left is the pattern still taught today: months alternate roughly between 31 and 30 days, with February as the standing exception at 28 (29 in a leap year). The alternation is not perfectly regular — it breaks around July and August, a visible seam left over from the centuries of patching described above — but it is regular enough that a simple trick captures nearly all of it. Make a fist and count knuckles: starting at the knuckle for January, every knuckle is a 31-day month and every dip between knuckles is a shorter one. Run out of knuckles at July, and start again at August on the same hand. It works because the underlying calendar, whatever its odd history, settled into a fixed sequence of lengths under the Julian reform and has not been touched since — only patched around, in the leap-year rule and the one-off skips of the Gregorian reform, ever since.