Realignment of the year The first step of the reform was to realign the start of the calendar year (1 January) to the tropical year by making 46 BC 445 days long, compensating for the intercalations which had been missed during Caesar's pontificate. This year had already been extended from 355 to 378 days by the insertion of a regular
intercalary month in February. When Caesar decreed the reform, probably shortly after his return from the
African campaign in late Quintilis (July), he added 67 more days by inserting two extraordinary intercalary months between November and December. The number may compensate for three omitted intercalary months (67 = 22+23+22). It also made the distance from 1 March 46 BC, the original New Year's Day in the Roman calendar, to 1 January 45 BC be 365 days.}} These months are called
Intercalaris Prior and
Intercalaris Posterior in letters of
Cicero written at the time; there is no basis for the statement sometimes seen that they were called "
Undecimber" and "
Duodecimber", terms that arose in the 18th century over a millennium after the Roman Empire's collapse. Their individual lengths are unknown, as is the position of the
Nones and
Ides within them. Because 46 BC was the last of a series of irregular years, this extra-long year was, and is, referred to as the "last year of confusion". The new calendar began operation after the realignment had been completed, in 45 BC.
Months The Julian months were formed by adding ten days to a regular pre-Julian Roman year of 355 days, creating a regular Julian year of 365 days. Two extra days were added to January, Sextilis (August) and December, and one extra day was added to April, June, September, and November. February was not changed in ordinary years, and so continued to be the traditional 28 days. Thus, the ordinary (i.e., non-leap year) lengths of all of the months were set by the Julian calendar to the same values they still hold today. The Julian reform did not change
the method used to account days of the month in the pre-Julian calendar, based on the Kalends, Nones and Ides, nor did it change the positions of these three dates within the months. Macrobius states that the extra days were added immediately before the last day of each month to avoid disturbing the position of the established religious ceremonies relative to the Nones and Ides of the month. The inserted days were all initially characterised as
dies fasti (
F – see
Roman calendar). The character of a few festival days was changed. In the early Julio-Claudian period a large number of festivals were decreed to celebrate events of dynastic importance, which caused the character of the associated dates to be changed to
NP. However, this practice was discontinued around the reign of
Claudius, and the practice of characterising days fell into disuse around the end of the first century AD: the Antonine jurist
Gaius speaks of
dies nefasti as a thing of the past.
Intercalation The old intercalary month was abolished. The new leap day was dated as
ante diem bis sextum Kalendas Martias ('the sixth doubled day before the Kalends of March'), usually abbreviated as
a.d. bis VI Kal. Mart.; hence it is called in English the
bissextile day. The year in which it occurred was termed
annus bissextus, in English the bissextile year. There is debate about the exact position of the bissextile day in the early Julian calendar. The earliest direct evidence is a statement of the 2nd century jurist
Celsus, who states that there were two-halves of a 48-hour day, and that the intercalated day was the "posterior" half. An inscription from AD 168 states that
a.d. V Kal. Mart. was the day after the bissextile day. The 19th century chronologist
Ideler argued that Celsus used the term "posterior" in a technical fashion to refer to the earlier of the two days, which requires the inscription to refer to the whole 48-hour day as the bissextile. Some later historians share this view. Others, following
Mommsen, take the view that Celsus was using the ordinary Latin (and English) meaning of "posterior". A third view is that neither half of the 48-hour "bis sextum" was originally formally designated as intercalated, but that the need to do so arose as the concept of a 48-hour day became obsolete. There is no doubt that the bissextile day eventually became the earlier of the two days for most purposes. In 238 Censorinus stated that it was inserted after the
Terminalia (23 February) and was followed by the last five days of February, i.e., a.d. VI, V, IV, III and prid. Kal. Mart. (which would be 24 to 28 February in a common year and the 25th to 29th in a leap year). Hence he regarded the bissextum as the first half of the doubled day. All later writers, including Macrobius about 430,
Bede in 725, and other medieval
computists (calculators of Easter) followed this rule, as does the
liturgical calendar of the Roman Catholic Church. However, Celsus' definition continued to be used for legal purposes. It was incorporated into
Justinian's Digest, and in the English
Statute De Anno et Die Bissextili of 1236, which was not formally repealed until 1879. The effect of the bissextile day on the
nundinal cycle is not discussed in the sources. According to Dio Cassius, a leap day was inserted in 41 BC to ensure that the first market day of 40 BC did not fall on 1 January, which implies that the old 8-day cycle was not immediately affected by the Julian reform. However, he also reports that in AD 44, and on some previous occasions, the market day was changed to avoid a conflict with a religious festival. This may indicate that a single nundinal letter was assigned to both halves of the 48-hour bissextile day by this time, so that the
Regifugium and the market day might fall on the same date but on different days. In any case, the 8-day nundinal cycle began to be displaced by the 7-day
week in the first century AD, and
dominical letters began to appear alongside nundinal letters in the fasti.
Year length; leap years The Julian calendar has two types of year: "normal" years of 365 days and "leap" years of 366 days. There is a simple cycle of three "normal" years followed by a leap year and this pattern repeats forever without exception. The Julian year is, therefore, on average 365.25 days long. Consequently, the Julian year drifts over time with respect to the
tropical (solar) year (365.24217 days). Although Greek astronomers had known, at least since
Hipparchus, a century before the Julian reform, that the tropical year was slightly shorter than 365.25 days, the calendar did not compensate for this difference. As a result, the calendar year gains about three days every four centuries compared to observed
equinox times and the seasons. This discrepancy was largely corrected by the
Gregorian reform of 1582. The Gregorian calendar has the same months and month lengths as the Julian calendar, but, in the Gregorian calendar, year numbers evenly divisible by 100 are not leap years, except that those evenly divisible by 400 remain leap years (even then, the Gregorian calendar diverges from astronomical observations by one day in 3,030 years). == Leap year error ==