« A.U.C. 708 = 46 B.C. »

This is the year in which Caesar decided on the reform of the Republican calendar. It was later known to Macrobius as "the last year of confusion" (although this was not quite true). The year's length is primarily determined from later literary evidence.

Suetonius, Caesar 40, tells us that this year had one intercalary month according to the Republican custom (i.e. after the Terminalia), and that Caesar inserted two additional months between November and December, for a total of 15 months. However, he does not give the lengths of any of these months. By contrast, Macrobius, Saturnalia 1.14.3, gives the length as 443 or (in one MS) 440 days, but does not given any details about the internal structure of the year.

The most detailed account comes from Censorinus 20.8, who states that Caesar added two months between November and December for a total of 67 days, in addition to the 23 days already intercalated in Februarius, for a total of 445 days in MSS known to modern scholars. However, W. Maude, De Die Natale ("The Natal Day") by Censorinus, 29 n. 14, notes that Lindenbrog, in his 1642 edition of Censorinus, had pointed out that Scaliger had concluded that one of the intercalations in this year was for an even number of days, and that this implied that the correct length should be 444 days.

Accepting that Caesar added 67 days after November, only 444 = 355 + 67 + 22 or 445 = 355 + 67 + 23 days are possible, given the known structure of the Republican calendar.

These sources are consistent, except for the total length of the year given by Macrobius. However, Dio Cassius 43.26 says that only 67 days were intercalated into this year. He is aware of a tradition that more were inserted, but explicitly denies it is correct. This would make the year only 422 days long, unless he understood the year to start on Kal. Mart., in which case it was 365 days long.

Dio's account shows a number of other problems:

Dio's account describes the Julian reform as an imposition of the structure of the Egyptian calendar on an originally lunar calendar. This is a reasonably accurate description of the subordination to the Julian calendar of the lunar Greek calendars of the east which took place under Augustus, the best-documented example of which is the reform of the Asian calendar in AUC 746 = 8. However, it is clearly not an accurate description of the Julian reform of the Roman calendar, and it is apparent that Dio's source did not understand, or was not concerned with, the Republican calendar.


The reality of the intercalation of two consecutive months is confirmed by the date of a.d. V Kal. Int. Pr. A.U.C. 708 given in Cicero, Epistolares ad Familiares 6.14 and the date of Int. Post. for Cicero, Ad. Att. 12.6 etc. Their position between November and December is confirmed by forward references to December in letters that clearly belong to this series such as Cicero, Ad Att. 12.1. However, we are unable to determine the lengths of the individual months.

J. Rüpke, The Roman Calendar from Numa to Constantine 117-118, has recently proposed a specific structure analogous to the procedures for inserting a normal Intercalaris. He notes that Censorinus does not say that the intercalary months were inserted between November and December, but in November and December ("in mensem Novembrem et Decembrem interponeret"). He suggests that on the last day of November, chosen for minimum interference with festivals earlier in the month, the rex sacrorum announced that the date was not prid. Kal. Dec. but a.d. VI Kal. Int. Prior. (analogous to a.d. VI Kal. Mart.). Int. Prior then started 5 days later. The following two months were then of 31 days each, with the last day of Int. Post. being considered as the last day of November.

This solution, though certainly speculative, is ingenious and plausible, and for chronological purposes I have adopted it here. I would quibble with one detail of the proposed procedure. If the announcement was left as late as the last day of November, then all preceding days would have been accounted as a.d. NN Kal. Dec. By analogy, the dates in Februarius in an intercalary year would have been a.d. NN Kal. Mart., but we possess a speech of Cicero's which refers to a date of a.d. V Kal. Int. (Pro Quinctio 79). It seems to me more likely that the dates after Id. Nov. took the form a.d. NN+4 Kal. Int. Prior. This need not disturb the ferial or legal character of any of the days between Id. Nov. and a.d. VI Kal. Int. Prior. (= prid. Kal. Dec. in an ordinary year), just as the renumbering of dates after Id. Feb. as a.d. NN-4 (or 5) Kal. Int. did not disturb the ferial or legal character of any of the days between Id. Feb. and prid. Kal. Int. (= a.d. VI (or V) Kal. Mart. in an ordinary year.


Contemporary data favours a length in excess of 67 days. ps-Caesar's account of the African War states that Caesar struck camp on a.d. VI Kal. Feb. A.U.C. 708 (De Bello Africo 37), shortly before a ferocious hailstorm that occurred on the night of the setting of the Pleiades, i.e. on 8 November 47 (De Bello Africo 47), and that Caesar's troops were ill-prepared for this since he had been moving camp every 3-4 days. This synchronism requires an intercalation of between 87 and 91 days, for a total length of between 442 and 446 days. Dio's confusion probably arose from the fact that Caesar's reform was decreed part way through the year, after the ordinary intercalary month had already occurred.

J. Beaujeu, Mel. Heurgon I 14, has adduced a second piece of evidence to favour the intercalation of 90 rather than 67 days. Cicero, Ad Atticum 11.20, tells us that C. Trebonius had arrived in Brindisium on a.d. XVII Kal. Sep. after a journey of 27 days from Seleucia Pieria (the port of Antioch), and reported that he had seen Caesar in Antioch. Since Trebonius had left Seleucia Pieria on a.d. XIII Kal. Sex., it follows that Caesar was in Antioch in around early Quintilis A.U.C. 707. Since Caesar was en route to fight Pharnaces in Pontus, his stay in Antioch was certainly brief. The battle of Zela was fought on prid. Non. Sex. A.U.C. 707 (CIL 1.324). Modern calculations of the time it took him to reach Zela from Alexandria (L. E. Lord, JRS 28 (1938) 19) estimate that it took him around a month to get from Antioch to Zela, and that his stay in Antioch cannot have exceeded a few days, and he must have left some time before Trebonius did. Malalas IX records that Caesar arrived in Antioch on 23 Artemisios. Hence Kal. Quin. A.U.C. 707 was somewhere close to 23 Artemisios.

On a 90-day intercalation, Quintilis A.U.C. 707 started on 17 April 47; on a 67 day intercalation it started on 10 May. Malalas himself explicitly equates Artemisios to Maius, which correctly reflects the alignment of the Antiochene calendar to the Julian calendar in his own day, the sixth century AD. But if the source date was a Julian date, it should have been either c. 10 Artemisios = Maius or c. 17 Xandikos = Aprilis, not 23 Artemisios. However, in 47 BC, the Antiochene calendar cannot yet have been aligned to the Julian calendar, since the Julian reform had not yet been promulgated. This suggests that the date is actually lunar.

The two lunar months in question started on c. 25 March or c. 23 April. 17 April 47 is the 24th day of the first month, 10 May 47 is the 18th day of the second. Hence we may reconcile the Roman data with Malalas' date on the assumptions (a) Artemisios was a lunar month beginning c. 25 March 47 and (b) AUC 708 = 46 was 445 days long, not 422.

No recent scholar appears to have considered the possibility suggested by Lindenbrog in 1642, that this year was 444 days long. The choice between 444 or 445 days is determined by the fact that both Kal. Ian. A.U.C. 702 = 52 and prid. Kal. Ian. A.U.C. 714 = 40, on the Caesarian calendar, were market days according to Dio Cassius 40.47 and 48.33.4, and hence a multiple of 8 days apart. The lengths of all intervening years except A.U.C. 708 can be determined exactly from other data, given the correct phase of the triennial error. Given this data, only a length of 445 days for A.U.C. 708 satisfies this condition.


One other theory requires comment here. G. Radke, Fasti Romani 66, accepts that A.U.C. 708 was 445 days long but argues that the triennial leap year cycle that followed Caesar's death is best understood if this year is considered the first Caesarian year, starting on Kal. Mart., with the leap day which he believes was inserted in A.U.C. 709 = 45 being interpreted as having been inserted at the end of that year. Accordingly, he argues that all months after Intercalaris A.U.C. 708 were adjusted to their Julian lengths, i.e. that 1 day was added to Aprilis, Iunius, September and November, and 2 to Sextilis and December, so that the two intercalary months between November and December totalled not 67 but 59 days, which he proposes to break down as 31 days + 28 days.

Aside from the complete absence of support for this in the ancient sources and the unlikelihood of there having being a leap day in A.U.C. 709 = 45, this scenario suffers from the fatal flaw that it requires Caesar to have instituted the reform between Februarius and Aprilis of this year. At that time he was fully engaged in the African campaign against Cato. According to De Bello Africo 98, he did not return to Rome until a.d. III Kal. Sex., so it is highly unlikely that the reform was decreed any earlier than Sextilis.

Since the Julian dates of A.U.C. 709 = 45 are known, this result fixes the Julian dates of A.U.C. 708 = 46.

26 Dec 2005: Comment on Greek relevance of Dio's description of the Julian reform (thanks to Michael Gartsman for pointing out the issue).
19 Noc 2006: Expand discussion of Malalas synchronism
14 Jan 2012: Add discussion of Rüpke's proposed structuring of the 67-day intercalation; note role of Lindenbrog (not Scaliger) in proposing the 444 day year.

Website © Chris Bennett, 2001-2012 -- All rights reserved