Julian Calendar (2024)

Stephen P. Morse

This article appeared in the Association ofProfessional Genealogists Quarterly (March 2014).

The Julian calendar is important to historians because itwas used worldwide for over 16 centuries, and in various parts of the world foranother three centuries after that. Andit's important to genealogists because it was used to record events in manycountries as recently as the early 1900s.For these reasons, it's necessary to understand the Julian calendar andto know when and how the conversion to our current Gregorian calendar was done.

On the surface, converting between the two calendars appearsto be straightforward. But furtheranalysis shows that there are several subtle but significant issues. For example, George Washington's birthday isconventionally celebrated on February 22.But as a result of the switch from the Julian calendar to the Gregorianone, his birthday was in fact February 23 in the 19th century, February 24th inthe 20th and 21st century, and will continue to advance in future centuries.

This paper will present the Julian calendar by first givinga historic perspective of the Roman calendars from which it was derived. It will then explain the workings of theJulian calendar, and the reforms that were made to convert it to the moreaccurate Gregorian calendar. It willthen describe the implications of these reforms, and the problems that they cancause for genealogists and historians.

The Calendar Requirement

The calendar has only one basic requirement -- that theseasons don't migrate through the years.We are used to going to the beach in July, and if after a few years itdoesn't warm up sufficiently until November we might not be too happy.

The seasons are determined by the position of the Earth inits orbit around the sun. At one pointin its orbit, the Earth's axis is tilted with the north pole toward the sun,and at the diametrically opposite point it is tilted with the north pole awayfrom the sun. At these two points thenorthern hemisphere is experiencing summer and winter respectively. And approximately midway between these twopoints is spring and fall. This isillustrated in Figure 1.

<![if !vml]><![endif]>

FIGURE 1. The earth'sorbit and the seasons

Why is it so difficult to design a calendar and get itright? The astronomers tell us that theearth goes around the sun once every 365.2422 days. And it is that fractional part (.2422) that causes all theproblems. We wouldn't want to start anew year after a fractional number of days.So we have to compromise and pick some integralnumber of days that is close to 365.2422.

If we pick a number that is too low, the seasons will occurlater and later each year. Let's assumethat we start the year in winter, and that we choose 300 days for the length ofa year. After the first year, the earthwill not yet have reached the correct point for the second winter. So the winter seasonwon't start until 65 days into the second year.And it won't start until 130 days into the third year. On the other hand, if choose too high anumber, say 400, the seasons will come earlier and earlier each year.

Of course we could improve onthings by picking numbers closer to 365.2422, such as 365 or 366. That will certainly slow down the drift butit won't eliminate it. If we wait longenough, even the best designed calendar will start to experience drift.

The seeds of the current Gregorian calendar lie in thecalendars used in ancient Rome. Thefirst Roman calendar was introduced in approximately 753 BCE (before the commonera) by Romulus, the legendary first king of Rome. Romulus's calendar had only ten months witheach month having either 30 or 31 days as follows:

Martius -- 31 days

Aprilis-- 30 days

Maius --31 days

Iunius --30 days

Quintilis-- 31 days

Sextilis-- 30 days

September -- 30 days

October -- 31 days

November -- 30 days

December -- 30 days

Calendar of Romulus, ca 753 BCE

It's not clear where the name Apriliscame from, but the other three of the first four months were named after Romangods. And starting with the fifth month,the names reflect the order of the month in the calendar: the fifth month was Quintilis meaning five in Latin, then Sextilismeaning six, all the way up to December meaning ten.

The total number of days in the ten months was 304, and thatwas certainly a problem. The year wouldend long before the earth was in the right position to start the nextyear. So theysimply waited around until their astronomers determined that it was time tostart the next year. That left about 61winter days unaccounted for, as they were not in any month.

That didn't work for long, and by 713 BCE the calendar wasmodified by Numa Pompilius,the legendary second king of Rome. Headded two months at the end of the calendar, Ianuariusand Februarius, to get rid of the unaccounted-fordays. He also introduced an intercalarymonth that occurred after Februarius in certainyears. These years became known as leapyears. In addition, he deleted one dayfrom all the months that had 30 days so that they had 29 days instead.

Here is Numa's calendar:

Martius -- 31 days

Aprilis-- 29 days

Maius --31 days

Iunius --29 days

Quintilis-- 31 days

Sextilis-- 29 days

September -- 29 days

October -- 31 days

November -- 29 days

December -- 29 days

Ianuarius-- 29 days

Februarius-- 28 days (23 or 24 days in leap year)

Intercalarius-- 0 days (27 days in leap year)

Calendar of NumaPompilius, ca 713 BCE

This resulted in a total of 355 days in a common year and377 days in a leap year. So it's apparent that they would have to have a leap yearjust about every other year. However there was no hard and fast rule as to when therewould be a leap year. Insteadit was left to the whim of the king, and he often chose the leap years forpolitical gain rather than for sound astronomical reasons. This calendar was certainly unstable, but itwas used for the next 700 years.

Sometime later (certainly by 450 BCE) the starting point ofthe calendar had been shifted from Martius to Ianuarius. All other aspects of the calendar remainedthe same, so it was still basically the Numacalendar. This change of starting pointmeant that the month names were now misnomers, no longer corresponding to theirposition in the calendar. In 450 BCE thecalendar was as follows:

Ianuarius-- 29 days

Februarius-- 28 days (23 or 24 days in leap year)

Intercalarius-- 0 days (27 days in leap year)

Martius -- 31 days

Aprilis-- 29 days

Maius --31 days

Iunius --29 days

Quintilis-- 31 days

Sextilis-- 29 days

September -- 29 days

October -- 31 days

November -- 29 days

December -- 29 days

Calendar of 450 BCE

Eventually the abuse of the leap years became so bad thatthe harvest festival was coming before the summer planting season. How could you reap what you had not yetsowed? So in 45BCE Julius Caesar reformed the calendar and introduced the first stablecalendar. He incorporated fixed rulesfor determining which years were leap years.He eliminated the intercalary month and replaced it with a singleintercalary day. He also had a regularpattern of alternating between 31 and 30 daymonths. And he did a one-time insertionof three months in 46 BCE to give the seasons a chance to catch up. See appendix 1.

Julius's calendar is as follows:

Ianuarius-- 31 days

Februarius-- 29 days (30 days in leap year)

Martius -- 31 days

Aprilis-- 30 days

Maius --31 days

Iunius --30 days

Quintilis-- 31 days

Sextilis-- 30 days

September -- 31 days

October -- 30 days

November -- 31 days

December -- 30 days

Calendar of Julius Caesar

In Julius's calendar the rule for leap years was that everythird year shall be a leap year. It isbelieved that Julius intended for it to be every fourth year but the people whoimplemented it made a calculation error in the way they counted to four (aso-called fence-post error). See Figure2.

<![if !vml]><![endif]>

FIGURE 2. Fence-PostError in counting to 4

But in spite of that error, we now had a year that was365.3333 days on average, a number that is getting close to the correct numberof 365.2422.

Julius didn't have much time to enjoy his new calendarbecause he was killed on the Ides of Martius (March 15) in 44 BCE, one yearafter his calendar went into effect. Buthis successor, Augustus Caesar, made some refinements to the calendar, and itwas Augustus's refinements that would become known as the Julian calendar andbe used world-wide for the next 16 centuries.He changed the leap-year cycle to every four years instead of everythree. He renamed Quintilisto Iulius, to honor his predecessor. And since he was doing that, he decided togive homage to himself as well and changed Sextilisto Augustus. But that would mean thathis month had one fewer day than Julius's month, and Augustus wasn't happyabout that. Sohe removed a day from Februarius and added it toAugustus, making it a 31-day month, same as Iulius. But since September had 31 days as well, thatwould make three consecutive months with 31 days. So he interchangedthe number of days in September and October, as well as interchanging thenumber of days in November and December.Augustus's revised calendar, is as follows:

Ianuarius-- 31 days

Februarius-- 28 days (29 days in leap year)

Martius -- 31 days

Aprilis-- 30 days

Maius --31 days

Iunius --30 days

Iulius --31 days

Augustus -- 31 days

September -- 30 days

October -- 31 days

November -- 30 days

December -- 31 days

Calendar of Augustus Caesar, whichbecame known as the Julian calendar

These month names and number of days in each month have remainedthe same up to the present day.

There was one more thing for Augustus to do, and that was tocompensate for the errors introduced by having the three-year cycle for leapyears. See appendix 2.

Below is a summary of the calendar changes from Romulus toAugustus.

<![if !vml]><![endif]>

The Julian Error and the Gregorian Fix

With Augustus's correction, the Julian calendar now had anaverage of 365.25 days per year. That seemedto be as close as they could get to the true number of 365.2422. Let's see just how significant that error is.

A difference of 0.0078 days per year comes to 1 day every128 years. That's about 3 days every 400years. By the 1500s, that amounted toabout 10 days. And the holidays werebecoming noticeably misaligned with the seasons.

To get back in step, in October 1582 Pope Gregory XIIIdecreed that 10 days be stricken from the calendar. But that was only the first part of hisfix. The second part would be to makesure we remain in step. He did that bydecreeing that century years (those ending in 00) not be leap years unless theyare evenly divisible by 400. That meansthat every 400 years we would have 3 fewer leap years, which translates to 3fewer days. That is close to the errorthat we just calculated above.

There is a third part to Gregory's fix, and that will bediscussed later. The calendar, withthese three fixes, would become known as the Gregorian calendar.

<![if !vml]><![endif]>

FIGURE 3: Striking days from the calendar

Figure 3 is a cartoon depicting the striking of days fromthe calendar in October 1582, and the reaction of the populace. The cartoon is cute but it contains twoerrors. The first is probably just atypo -- it shows the stricken days being from the 4th of October to the 14th,for a total of 11 days. That's one daytoo many. In reality, the 4th of Octoberwas not stricken -- the first stricken day was the 5th of October.

The second error is more significant. The cartoon shows the days being crossed outbut still taking up space on the calendar.That means that if October 4th was on a Thursday, the next day would beMonday October 15. And if that weredone, it would break the sanctity of the Sabbath occurring every seven days, asanctity that is observed in several major religions (Christianity, Judaism,Islam). Certainlythe head of the Catholic church was not about to break the continuity of therevered seven-day count.

So rather than having the stricken days remaining on thecalendar and taking up space, they do not appear on the calendar at all. The correct calendar for October 1582 isshown in Figure 4.

<![if !vml]><![endif]>

FIGURE 4. Calendarfor October 1582

Let's not do it all at once

Gregory decreed that the cutover date for the calendarshould be October 4, 1582. And theCatholic world was eager to obey. Onthat day Italy, Poland, Portugal, and Spain all switched over. By the end of that year France, Holland, andpart of Belgium made the switch. Thefollowing year Austria, the rest of Belgium, and Catholic Germany fell inline. And they were joined byCzechoslovakia and Catholic Switzerland in 1584, Hungary in 1587, andTransylvania in 1590.

The Protestant and Greek Orthodox countries were not thatanxious to switch. Protestant Germanyswitched piecemeal during the 1600s.Denmark, Iceland, the rest of the Netherlands, Norway, and ProtestantSwitzerland switched in the year 1700.Canada, Great Britain, Ireland, and the eastern US switched in1752. Japan switched in 1873 and Egyptin 1875. Then between 1911 and 1923Albania, Bulgaria, China, Estonia, Greece, Latvia, Lithuania, Romania, Russia,and Yugoslavia all switched over. And bringingup the rear was Turkey, in 1927.

Even within the land that would become the United States,the cutover was not simultaneous. Itdepended on which country the specific territory was owned by. Texas, Florida,California, Nevada, Arizona, and New Mexico all switched with Spain in 1582(although one report I read stated that the Spanish colonies didn't switchuntil two years after Spain).Mississippi switched with France in 1582. The eastern seaboard switched with GreatBritain in 1752. And Alaska switched in1867 when it became part of the US.

There was a price to be paid by waiting to do thecutover. The longer you waited, theworse it got. In 1582 the correction was10 days. The year 1600 was a leap yearin both the Julian and Gregorian calendars, so nothing changed until the year1700. From March 1, 1700 to February 28,1800 the error was 11 days. From March1, 1800 to February 28, 1900 the error was 12 days. And from March 1, 1900 to February 28, 2100the error is 13 days. See appendix 3.

Figure 5 shows the calendar for September 1752 in the US,when the switch was made and 11 days were dropped. And figure 6 shows the calendar for 1918 inRussia when it switched over, dropping 13 days.

<![if !vml]><![endif]>

FIGURE 5. September1752 in the US

<![if !vml]><![endif]>

FIGURE 6.January/February 1918 in Russia

Several places on earth had unique cutovers -- namelyAlaska, Sweden, and Nova Scotia.

Alaska got caught on the dateline. Prior to 1867 it was part of Russia. Russia had not yet switched over to the newcalendar. Russian time zones were allwest of the dateline, and in fact the dateline was drawn around Alaska toensure that it was also on the west side.In 1867 Alaska was acquired by the US.The US had already switched over.And US time zones are all east of the dateline. When Alaska changed hands, the dateline wasredrawn with Alaska changing sides. So Alaska lost 12 days due to the switch being in the 1800sbut it repeated 1 day by crossing the dateline (see Figure 7).

<![if !vml]><![endif]>

FIGURE 7. Alaskaloses 11 days and repeats 1

Sweden had a mind of its own. In an attempt to have a gradual conversion,Sweden decided not to have leap years from 1700 to 1740. That way there would be no jolt to thecalendar, no populace demanding to be compensated for 11 lost days. But after skipping the leap year in 1700,they abandoned the plan. This put themout of step with both the Julian and Gregorian calendars, so their dates didn'tcorrespond with anyone else's. In 1712they reverted back to the Julian calendar by having 30 days in February thatyear to make up for the leap day that they missed in 1700. And then in 1753 they gave up and switchedall-at-once to the Gregorian calendar.

And Nova Scotia couldn't make up its mind. It switched to the Gregorian calendar in1605. It then switched back to theJulian calendar in 1710. And thenswitched back to the Gregorian calendar in 1752.

In spite of all this switching, there are places on earth inwhich the Julian calendar is still used.It is used in the Eastern Orthodox Church for calculating Easter andother feasts. It is used by the Berberpeople in North Africa and on Mount Athos.And Ethiopia uses the Alexandrian calendar which is based on the Juliancalendar.

The Start of the Year

There are two distinct events that we associate with a newyear. One is the partying with thedrinking of champagne, the throwing of confetti, and the blowing ofnoisemakers. That day is referred to asNew Year's Day, and has always been on January 1. But then there is the day that the yearnumber changes, and that can be distinct from New Years Day. Let's call that Number-Change Day. In the Gregorian calendar, Number-Change Dayis also on January 1, and coincides with New Year's Day. The same was true for the Romancalendars. But that's not the case forthe Julian calendar.

In the initial Julian calendar, Number-Change Day was onJanuary 1. But when local calendars werealigned to the Julian calendar, each kept its own Number-Change Day. Specifically:

Alexandrian calendar (Egypt):August 29/30

Several local provincial calendars:September 23 (Augustus's birth)

Byzantine year: September 1

Eastern Orthodox Church liturgicalyear: September 1

Russia from 998 CE: March 1

Russia from 1492 CE: September 1

Russia from 1700 CE: January 1

Western Europe during middle ages: December 25 / March 25

England pagan times: December 25

England from 1089 to 1155: January1

England from 1155 to 1751: March 25

The Number-Change Day was reflected in parishregisters. Figure 8 shows a registerwith the year number entered just before March 28 (obviously the year changewas on March 25).

<![if !vml]><![endif]>

FIGURE 8. A parish registershowing the year number change in March

As mentioned before, there were three steps to a Gregorianconversion, but so far we have talked about onlytwo. We said that to get back in stepwith the seasons, the calendar had to suddenly skip ahead by 10+ days. And to remain in step with the seasons, thecalendar had to stop observing leap years three times in every 400 years.

The third step was to unify the Number-Change Day. As part of the switch, the year number was tobe incremented on January 1, along with the New Year's Day celebration.

Double Dating

Dates between New Year's Day and Number-Change Day have twoyears associated with them. The first isthe year in the Julian calendar, which was the calendar in use in thecountry. The second is the year that itwould have been if the year number had changed on January 1. To avoid confusion, both years are sometimesentered into a record. This is calledDouble Dating or Dual Dating.

Here is an example of a (fictitious) parish register in whichdouble dating is used.

1661

March 26: John Smith

...

December 31: Mary Jones

1661/1662

January 1: Tom Brown

...

March 24: William Anderson

1662

More realistically, Figure 9 shows a tombstone that liststhe date of death as "Febrey 15th Anno1703/4." That means it was 1703 inthe Julian calendar in use, but it would have been 1704 if they had changed theyear number on January 1.

<![if !vml]><![endif]>

FIGURE 9. Tombstonewith double dating

One more example is that of George Washington's birth recordfrom the Washington family Bible (Figure 10).It lists his date of birth as "11th day of February 1731/2."

<![if !vml]><![endif]>

FIGURE 10. GeorgeWashington's birth record

George Washington's Birthday

We just saw George Washington's birth record, and it reportsthat he was born on February 11, 1731/2 in the Julian calendar. The Gregorian switch occurred in the US onSeptember 2, 1752. And George would turn21 five months later, on February 11, 1753.

But if we count the days of his life up to February 11,1753,we see that he had been alive 11 days shy of 21 years. That presented a problem.

To solve the problem, the switch-over decree made specialprovisions to use the Julian calendar for computing time durations that startedbefore the switch occurred. So George's 21st birthday was on February 11, 1752/3 Julian,which is February 22, 1753 Gregorian.He'd have to wait 11 more days before they would sell him alcohol.

George died in 1799, but had he lived to 1801, his birthdaywould have been on February 11 Julian which would now be February 23Gregorian. Note the change in hisGregorian birthday from February 22 to February 23. And in the 1900s and 2000s, his birthdaywould be February 24.

Now George is not unique.You might have ancestors who were born at a time and place that theJulian calendar was used. How do youenter their dates of birth on your family tree?If you enter it using the Julian calendar date, there is noconfusion. But then it would beinconsistent with other dates in your tree that are Gregorian. If you enter it as the converted Gregoriandate, which converted value do you use?The converted value changes each time the century changes. I'm not going to answer this question, but Ido want to make you aware of the problem.It's up to you to decide which date to use.

Birthdays were not the only problem caused by theconversion. Suppose you pay rent on thefirst day of every month, and you made a payment on September 1, 1752. Due to the switch-over, September 2 of thatyear is followed immediately by September 14.Your next payment is due on October 1, but that's only 19 days from yourprevious payment.

Special provisions were made stating that monthly or yearlypayments would become due "at and upon the respective natural days andtimes as the same should ... have been payable ... in case this act had notbeen made" (Provision 6 of the switch-over decree). That translates into saying that you continueto use the Julian calendar to determine when such payments are due.

The Gregorian Error

As was previously mentioned, the average Julian year is365.25 days and the true length of a year is 365.2422 days. That difference caused the Julian seasons toadvance 1 day every 128 years.

The average Gregorian year comes out to be 365.2425days. That difference causes the seasonsto advance 1 day every 3,333 years. So all we've done is kick the can down the road, although wedid kick it quite far this time.

Note that the error is close to 1 day in 4,000 years. That suggests a rather simple modification tothe leap-year rule to fix the problem.Namely skip the leap years in millennium years that are divisible4,000. We haven't had such a year yet(at least not since 1582) and we won't have one until the year 4000. So we don't have tomake any decisions on this for another 2,000 years. And at that time there will be no major joltto the system, no 10 days to account for, and nobody will complain about whentheir rent will be due. All that willhappen is that the month of February in the year 4000 will have only 28 days,and nobody is going to be harmed by that.

If such an amendment were made, what would the new errorbe? In that case the average length of ayear would become 365.24225. That'sgetting very close to the magic number of 365.2422, and the difference wouldnow cause the seasons to advance 1 day in 20,000 years.

That amendment is very simple. There is a more complicated (but moreaccurate) change that was proposed by the Greek Orthodox Church in the 1920s,but they never implemented it. Theyrealized that they could do better than the century rule which drops three leapyears every 400 years. Instead they proposed dropping 7 leap years every 900years. They would do that by requiringthat in order for a century year to be a leap year, it needs to give aremainder of 200 or 600 when divided by 900.Of course that was the complicated part, andnobody was going to go along with that.You can probably see immediately that the year 1600 was a leap yearaccording to the Gregorian rules, but it will take you more effort to figureout that it was not a leap year under the proposed Greek rules.

Although it's too complicated to be viable, let's see whatthe error would be under the Greek rules.Under those rules, the average length of a year is 365.24222 days. That's awfully close to the true value of365.2422 days. In fact, the differenceis small enough that the error can no longer be calculated because the 365.2422number has some slight variance in it (it is not the same every year).

Telling One Year From Another

Throughout thispaper we have been designating years with numbers. For example, we said that the Julian calendarcame about in 45 BCE. But of course people in the year 45 BCE didn't refer to the yearthat way because they had no way of knowing that anything special was going tohappen 45 years later. So how did theytalk about their years?

Consular Dating (510 BCE to 541 CE):

The years in theRoman calendars and the early Julian calendar were designated by giving thenames of the two consuls who took office in that year. For example, the two consuls who took officein 59 BCE were Caesar and Bibulus. Sothat year was referred to as the year of Caius IuliusCaesar & Marcus Calpurnius Bibulus. That was quite a mouthful.

This method ofyear-dating started with the first year of the Republic in 510 BCE andcontinued until 541 CE when the emperor stopped appointing consuls.

A list of the consulpairs for each year can be found in wikipedia at http://en.wikipedia.org/wiki/List_of_Roman_consuls

Regnal Dating (541 CE to 800 CE)

Regnal dating refersto specifying the year number of the presiding monarch's reign. It is said to have started with Augustus, whoinformally counted how many times he had held office as a consul or anemperor. It continued in this informalmanner until about 200 CE when the emperors started talking about their regnalyear openly. But it wasn't adopted asthe official year designation until 541 CE, when consular dating ended for lackof consuls.

The then-currentJulian calendar was used worldwide, and different regions had differentmonarchs. Soalthough they were all using the same Julian month and day, they all haddifferent numeric values for the year depending on how long their monarch hadbeen in power. This obviously led tomuch confusion.

Common Era Dating (800 CE to present)

A system ofnumbering the years using BC (Before Christ) and AD (Anno Domini) was devisedby Dionysius Exiguus in the year 525 CE. But it was not widely used until Saint Bedementioned it in his Historia Ecclesiastica in 731CE. By 800 CE it had pretty-muchreplaced regnal dating.

The BC/AD notationwas religion specific since it referenced Christ. So later religion-neutral synonyms for BC andAD were introduced, namely BCE (before the common era) and CE (common era). This is the notation that is used throughoutthis paper.

Conclusion

As this paper illustrated, there are several non-obviousissues that historians and genealogists need to take into consideration whenworking with dates in the Julian calendar.To simplify this in some measure, I have created a tool (figure 11) onmy website for doing conversions between the two calendars. The address of that tool is http://stevemorse.org/jcal/julian.html. The tool adjusts by the required number ofdays and also takes the year-number-change day into consideration.

<![if !vml]><![endif]>

FIGURE 11. Tool for converting between Julian andGregorian calendars

Appendix