« A.U.C. 687 = 67 B.C. »

The inscription CIL I2 2511 comes from fragments of a sarcophagus from the vault of the gens Salvius, an ancient Etruscan family whose best-known member is the emperor M. Salvius Otho. The inscription is in poor condition, and two quite different reconstructions have been published.

The inscription clearly equates day 3 of a lunar month to a date in A.U.C. 687 = 67, either a.d. V Id. Oct. or a.d. XV Kal. Oct., which are 24 days apart. The month was apparently also named in a second calendar, presumably the Etruscan one. A. Degrassi, ILLRP2 589 n. 3, suggested that the missing name is XOSFER, which is Etruscan October. A. Emiliozzi, MEFRA 95 (1983) 701 at 710, read it (with some doubt) as GIGNE[..], however this does not correspond to any of the 10 known Etruscan months which correspond roughly to the period Martius-December (J. Whatmough, HSCPh 42 (1931) 157). One would expect CELI, Etruscan September. Whatever the case, it is clear that it is only the months that are being equated; the actual conversion is to a lunar date.

Degrassi's restoration is based on the original publication, since he was not able to locate the actual inscription, while Emiliozzi's is based on personal inspection. For this reason, Emiliozzi's reading is to be preferred, provided a solution exists for a.d. XV Kal. Oct. A.U.C. 687 = Lunar day 3.

It is not known whether the lunar date is based on crescent invisibility or first visibility. Roman months were not functionally lunar, and dates according to lunar months are exceeding rare in Roman inscriptions. It is not known whether they were determined by observation of a new moon, first visibility, or based on a formulaic lunar calendar such as that described in pdem Carlsberg 9.

The next recorded case, and the only recorded case before the third century AD, is CIL IV 4182, a Pompeian graffito dated a.d. VIII Id. Feb. in A.U.C. 813 = A.D. 60, "dies solis, luna XIIIIX", i.e. 6 February A.D. 60, Sunday, lunar day 16. Lunar day 1 in CIL IV 4182 is therefore 22 January A.D. 60, which is the date of a new moon. However, this does not require us to make this assumption for the lunar date of CIL I2 2511, written 126 years earlier. CIL IV 4182 is also the earlest Latin inscription naming a weekday -- but 6 February A.D. 60 was a Wednesday, not a Sunday. Brind'Amour, Le calendrier romain 268ff., ingeniously explains the discrepancy by reference to the Antonine astrologer Vettius Valens, Anthologiarum Libri 1.9.1.10, who explains that each of the 24 hours of the day is governed by one of the seven planets in a continuous cycle, Sun, Venus, Mercury, Moon, Saturn, Jupiter, Mars, and that the governing planet of the astrological day is the deity of the first hour of the night. Valens gives a worked example for 13 Mecheir year 4 of Hadrian = 8 February A.D. 120, showing that it was governed by Mercury, i.e. that it was a Wednesday, which indeed it was by modern reckoning. However, by the same reasoning, the daylight half of the day also has a presiding planet, the planet governing the first hour of the day, 12 hours later. If the presiding planet of the night was Mercury, then that of the daylight hours was the Sun. Hence the daylight half of 6 February A.D. 60 was a Sunday.

This structure clearly reflects the Babylonian astronomical and astrological system, with a day beginning at sunset, rather than at midnight in the Roman style. CIL IV 4182 shows a failed step in the evolution of the modern week, according to which the day and the night had different presiding planets. The Babylonian lunar month surely went through a similar process of assimilation. We might reasonably expect that when the system was first introduced in Rome, in the 1st century, that the lunar month reflected Babylonian practice, i.e. began with first visibility.

In view of these uncertainties, we can can only project the lunar month of CIL I2 2511 to start within a couple of days after an astronomical new moon.

The analyses of A.U.C. 691 = 63 and A.U.C. 689 = 65 indicate that at most 5 intercalations occurred between A.U.C. 688 = 66 and A.U.C. 697 = 57. The following table shows the projected Julian dates of the two possible Roman dates of lunar day 1, a.d. VII Id. Oct. A.U.C. 687 and a.d. XVII Kal. Oct. A.U.C. 687, assuming all possible combinations of 22 or 23 day intercalations for up to 5 intercalations, and compares them to the dates of first crescent visibility. Reconstructions that match or are within two days of the first crescent are bolded. Note that first crescent is in the evening, while a Roman day begins in the morniing (formally, at midnight)

Number of Intercalations   Intercalated days    New moon        a.d. VII Id. Oct.        a.d. XVII Kal. Oct.
     A.U.C. 688-696                                        (Rome)               

              1                         22-23              26 Dec. (4.2%)           7-8 Dec.                13-14 Nov.
              2                         44-46              26 Nov. (1.9%)         14-16 Nov.               21-23 Oct.
              3                         66-69              28 Oct. (3.8%)          22-25 Oct.               28 Sep - 1 Oct.
              4                         88-92              28 Sep. (2.1%)           29 Sep.-3 Oct.         5-9 Sept.
              5                       110-115             30 Aug. (4.0%)          6-11 Sep.                13-18 Aug.

Remarkably, both dates give a match for a lunar month based on the first crescent of the evenning of 28 September 67. For Degrassi's reading, there is only one possible solution: a.d. V Id. Oct. A.U.C. 687 = 29 September 67. This match is obtained by 4 intercalations of 23 days each before A.U.C. 697 = 57, giving a result that is the day after the first crescent in September. This matches exactly if the lunar month is counted from first visibility, in the Babylonian fashion.

Emiliozzi's reading gives two possible solutions: a.d. XV Kal. Oct. A.U.C. 687 = 28 or 29 September 67. The first corresponds to a lunar month based on crescent invisibility, the second to a lunar month based on first visibility. Both solutions imply 3 intercalations before A.U.C. 697 = 57; in the first case, 3 of 23 days, in the second 2 of 23 days and one of 22 days. Only the first solution is compatible with the inferred Lex Acilia, hence this is the solution used here.

Since Emilozzi's reading gives acceptable solutions, it is to be preferred. Since, on either solution, it implies only three intercalations between A.U.C. 687 = 67 and A.U.C. 697 = 57, and one of these was A.U.C. 696 = 58, the other two may be assigned to A.U.C. 690 = 64 and A.U.C. 693 = 61, both for an even model and on the inferred Lex Acilia. By the same criteria, A.U.C. 687 = 67 itself must be an intercalary year.

Website © Chris Bennett, 2001-2011 -- All rights reserved