[HOME]

Egyptian Dates

Intro page      |      How to Read the Tables      |      The Egyptian Calendars       |       Sources      |      Analysis

Civil dates: (Excel) (HTML) (CSV)                                                   Lunar cycle: (Excel) (HTML) (CSV)

This page gives access to a set of conversion tables for determining the Julian equivalent of Egyptian civil and lunar dates in the Ptolemaic era. Two tables are provided: a table converting civil dates to Julian dates, and a table notionally converting lunar dates to civil dates according to the lunar cycle of pCarlsberg 9.

In this section, several topics are discussed:

• The wandering year
  

• The regnal eras
  
• The Carlsberg cycle
  
• The financial year
  
• The Canopic reform
  
• The Soter era

The Alexandrian Reform

Under Augustus the civil calendar was changed from the wandering year of 365 days to a fixed year by the intercalation of a leap day every four years. Conversion tables for this calendar were published by Theon in the fourth century AD. At that time the fixed calendar intercalated the leap day as a 6th Epagomenal day at the end of the year immediately preceding the next Julian leap day, i.e. on 29 August. As a result, the relationship between Egyptian and Julian dates was absolutely fixed except between 6 Epagomene = 29 August and the following 29 February = 4 Phamenoth (26 February = a.d. V Kal. Mart. (= 1 Phamenoth) in the Roman calendar) every fourth year, when the Egyptian calendar is one day behind the Julian one. This calendar is known as the Alexandrian calendar.

Theon, in his commentary on the Handy Tables of Claudius Ptolemy, noted that the Alexandrian calendar coincided with the wandering calendar between Augustus year 5 = 26/5 and year 8 = 23/2, implying that the Alexandrian calendar started in year 5, with the first leap year occurring at the end of year 8. Theon did not explain why the reform was not instituted until year 5. More seriously, the Roman calendar itself was not operating under correct Julian rules for intercalation before AD 4, so the Alexandrian calendar described by Theon was not phase-locked to the Roman one in the first few cycles of its existence. This raises the question of whether Theon's description of its origins is historically accurate, or just a mathematical projection of the calendar as it operated in his day.

In the early 19th century, Ideler suggested that the Alexandrian calendar should have been based on the era of Augustus' rule in Egypt. He argued that the Alexandrian calendar was not intended to align the civil year to the solar year but to fix the relationship between the Roman and Egyptian calendars, i.e. that the guiding principle was to cause an intercalation at the end of the Egyptian year before the Roman leap year. However, while noting the apparent difficulty of explaining why the calendar only started in Augustus' year 5 rather than year 1, most scholars followed Theon's lead; for example, J. G. Smyly, Hermathena 11 (1901) 81, showed that the contemporary data available to him was consistent with Theon.

The debate was reopened by W. F. Snyder, AJP 64 (1943) 385, who worked through the consequences of Ideler's model in detail. Since the early Julian calendar intercalated every third year instead of every fourth, this model would lead to an Egyptian leap year every third year instead of every fourth when the Alexandrian calendar was first introduced. It would further follow that intercalations were suspended in the Alexandrian calendar when they were suspended by Augustus in the Roman one in 8 B.C. Snyder showed that, according to the standard view of the operation of the early Julian calendar, Ideler's reconstruction caused the Alexandrian calendar to coincide with the wandering calendar between year 1 = 30/29 and year 3 = 28/7 of Augustus, while remaining consistent with the available data for later years, thus resolving the apparent difficulty. Snyder held that this coincidence showed that the Alexandrian calendar was introduced at the beginning of Augustus' reign in Egypt. He regarded it as sufficient proof, in itself, of the correctness of his proposal.

The Contemporary Evidence

J. G. Smyly, Hermathena 11 (1901) 81 noted that the phase of the leap year cycle in the Alexandrian calendar in the first century is given by pOxy 1.45, dated 6 Epagomene year 14 of Domitian = 29 August AD 95. He also cited three synchronisms between the Alexandrian calendar and the Roman or the wandering year:

• SB 1.684: Year 17 Tiberius, 18 Tybi ("Greek") = 1 Mecheir ("Egyptian") [= 13 January AD 30]
  

• pLond. 130: Year 3 Titus, 6 Pharmouthi = 1-2 Pachon ("old") = Kal. Apr. [= 1 April A.D. 81]
  
• pParis 19bis: Year 1 Antoninus, 8 Hadrianos(=Choiak) ("Greek") = 18 Tybi ("old") [= 4 December AD 137]

These, and the following additional synchronisms, are available in D. Hagedorn & K. A. Worp, ZPE 104 (1994) 243:

• pRyl 2.381: Year 4 Caligula, 4 Mesore = 20 Mesore ("Egyptian") [= 28 July A.D. 40]
  

• pOxy 31.2555: Year 7 Claudius, 15 Pachon = 5 Payni ("old") [= 13 May A.D. 46]
  
• pFayum 139: Year 1 Marcus & Verus, 5 Mesore ("Greek") = 16 Thoth ("old") [= 29 July A.D. 161]
  
• pKellis inv. 61: Consulate of Tatius Andronicus & Pompeius Probus, 18 Epeiph ("Greek") = 18 Phaophi ("Egyptian") = a.d. IV. Id. Iul [= 12 July A.D. 310]
  
• tKellis inv. A/5/198: Year 89 Diocletian, 21 Pachon ("Greek") = 1 Epagomene ("Egyptian) [16/17 May A.D. 373]

  Additional equations are given by Hagedorn & Worp, but involve restorations of at least one element. Note that the "Egyptian" date of 18 Phaophi in pKellis inv. 61 should be 6 Phaophi. The fact that the "Greek" and "Egyptian" dates have the same day number suggests the scribe was not concerned about the actual Egyptian date, only the month. Similar behaviour is frequently seen in Egyptian/Macedonian double dates.

These equations show that the official Egyptian civil year did indeed have a four year intercalary cycle with the expected and fixed phase relationship to the Julian intercalary cycle in the early imperial era, after the completion of the Augustan reform of the Roman calendar. However, until very recently, the contemporary evidence for earlier times was insufficiently precise to distinguish Theon's model from the Ideler/Snyder model for the operation of the Alexandrian civil year before the Roman calendar was stabilised. W. F. Snyder, AJP 64 (1943) 385 at 392-393 n. 14 noted two objections that Parker had raised against his reconstruction:

• pOxy 4.804 gives the solar and lunar positions on 5 Phaophi year 27 Augustus = 4 B.C; in particular, that the moon was in Pisces. O. Neugebauer & H. B. van Hoesen (Greek Horoscopes 21) note that these show that it was dated by the civil calendar, not the wandering one, from which they inferred astronomical confirmation that 5 Phaophi year 27 = 2 October 4 B.C. Snyder argued that it also matches his reconstruction, in which the civil calendar is phase-locked to the Roman civil calendar.

  This is not quite correct -- Snyder was basing himself on an incorrect dating of AD 4 by the pOxy editors. According to Snyder's model 5 Phaophi year 27 = a.d. VI Non. Oct AUC 750 = 4 October 4 BC. But the moon was in Pisces from 2-4 October 4 B.C., so this papyrus still does not invalidate his model.

• pdem Rhind 1 gives the synchronism Augustus Year 21 = 10/9 B.C., 10 Epeiph = Lunar [II Shomu] 16. Parker noted that according to the lunar calendar of pCarlsberg 9, year 21 = lunar cycle year 23, in which year Lunar II Shomu 16 = 14 Epeiph (wandering) = 8 July 9 B.C. For Snyder, 10 Epeiph = a.d. IV Non. Iul. = 5 July 9 B.C. The four-day difference between the proleptic wandering date and the civil date matches the Alexandrian calendar, and the actual date of lunar conjunction, but the Snyder interpretation is three days in error. However, Snyder argued that Parker's reconstruction depended on the assumption that the pCarlsberg 9 calendar was actually in use in 10/9 B.C. and that he had not demonstrated this.

  Snyder's point has subsequently been validated by the reanalysis of pCarlsberg 9 by L. Depuydt, In Mem. Quaegebeur II 1277. Nevertheless, A. Jones, ZPE 129 (2000) 159 at 160 n. 4, further argued that the three-day difference between the Snyder and Parker predictions for the start of the lunar month is so large that it supports the view that the civil date in it is given according to the Alexandrian calendar. While there is certainly merit to this objection, since the largest recorded discrepancy between the start of an Egyptian lunar month and the date of conjunction is only two days, a three day discrepancy is not entirely impossible.

I have found one additional papyrus published in the following decades that apparently conforms Snyder's model. SB 18.13849 is a fragmentary papyrus dated in Phaophi of a lost year in which the prefect P. Petronius refers to the birthday celebrations in the 25th of a lost month. R. S. Bagnall, YCS 28 (1983) 85 noted that the papyrus must be early in the Roman period on paleographical grounds, and argued that the birthday celebrations must have been held on 25 Thoth. He pointed out that 25 Thoth was the birthday of Augustus (a.d. IX Kal. Oct. = 23 September (Julian)) in the year following an Alexandrian leap year. A prefect Petronius, variously identified as C. Petronius (Dio Cassius, 54.5.4) or P. Petronius (Pliny, Hist. Nat. 6.35), is known from the literary sources to have governed Egypt in the late 20s; his term of office started in late 25 (Josephus, Ant. Jud. 15.9.4 -- see S. Jameson, JRS 58 (1968) 71). Bagnall accordingly dated the papyrus to Phaophi year 9 = October 22.

While Bagnall did not explicitly discuss the point, this solution only works on the Ideler/Snyder model for the early Alexandrian calendar, given the model of the Roman calendar that was generally assumed at the time. Since the Julian calendar is at least 1 and usually 2 days ahead of that model, it is not possible to date this papyrus to the late 20's on the Theon model of the Alexandrian calendar; nor is it possible on the wandering year. The difficult problem presented by the date of this papyrus is discussed in more detail below.

However, a significant problem arose for the Ideler/Snyder model with the publication of pVindob L.1c. This gives the explicit synchronism a.d. XIIII Kal. Aug. = 27 Epeiph in an unspecified year. The reference to August proves that the papyrus dates to 8 B.C. or later. However the equation 19 July = 27 Epeiph is not valid for the wandering year at that time, hence the Egyptian date must be Alexandrian. But the normal Alexandrian calendar, per the Ideler/Snyder model, gives 19 July = 25 Epeiph. As pointed out by D. Hagedorn, ZPE 100 (1994) 211, this datum alone is enough to invalidate the Ideler/Snyder model, but is correct on Theon's model, and the standard model of the Roman calendar, between 5 and 1 BC.

Although to my knowledge no-one has attempted to do so, it could perhaps be argued that the synchronism of pVindob L.1c was a scribal error or inaccuracy. There is plenty of precedent in the loose equations of Egyptian / Macedonian synchronisms. However, an even more serious problem arose with the publication of pOxy 61.4175, an ephemeris table from 24 B.C. discussed in A. Jones, ZPE 129 (2000) 159. The data in this table shows that the Roman calendar was in sync with the Julian calendar in 24 B.C. But this requires that the phase of the triennial leap year cycle assumed by Snyder for the Roman calendar, and the alignment of the Roman calendar to the Julian calendar, both of which have been almost universally accepted since Scaliger's work in the 16th century, are incorrect.

Jones supposed that this is true only in Egypt, i.e. that the Roman calendar there, and only there, operated on the correct Julian cycle, but it is inherently more likely that the same Roman calendar was used throughout the Roman world. When the relationship between the Roman calendar and the Julian calendar is corrected to account for this data, and other data from this period, the resultant variant of the Ideler/Snyder model is two days out of alignment with the Julian match to the wandering year in 30 BC, which contradicts the basic assumption on which Ideler and Snyder based their model.

One other item is perhaps worth mentioning here. Tab. Amst. inv. 1 is a partial astronomical almanac with data for years 1 and 2 of an unspecified era. A. Jones, CdE 68 (1993) 178, showed that the planetary alignments described are correct for 26/5 and 25/4. In that article, he tentatively suggested that the years belonged to an era based on the year of the Alexandrian reform. In A. Jones, ZPE 129 (2000) 141 he showed that the dates were actually years 1 and 2 of the Fifth Callippic Cycle. However, the observation is interesting and suggests the possibility that the Alexandrian reform was timed to coincide with the start of a Callippic cycle.

The Early Regnal Years of Augustus

T. C. Skeat, ZPE 132 (2000) 240, while not attempting to challenge Jones' astronomical analysis, argued that his own research into the early regnal years of Augustus showed that the Ideler/Snyder model must be correct, since it preserved the correct Egyptian / Roman synchronism for a festival that was to be celebrated on the date of the fall of Alexandria: 8 Mesore year 22 = Kal. Sex. AUC 724 = 3 August 30.

The weakness in this argument is that the festival is supposed to have been celebrated on Kal. Sex., not 8 Mesore. While the Ideler/Snyder model fixes the relationship, the festival could be celebrated on Kal. Sex. (Kal. Aug.) on other Egyptian dates if the relationship was not fixed. Skeat did not adduce any proof that this festival was celebrated on any specific Egyptian date in the years before the Roman calendar was correctly synchronised to the Julian one.

The issue of the early Augustan regnal year is important to what follows, so it is worth recounting Skeat's development of his ideas in some detail.

T. C. Skeat, JRS 43 (1953) 98, had pointed out that the Canon of Ptolemy assigned 22 years to Cleopatra, but since her reign ended before the end of her year 22 we would normally have expected 21 years, with year 22 Cleopatra = year 1 of Augustus. The correctness of the Canon is proven by pOxy 12.1453, an annual contract for lamplighting for year 1 of Augustus from 1 Thoth to "[5?] Mesore (Epagomene)" which continued a contract for the previous year 22=7 of Cleopatra.

The date of pOxy 12.1453 used in that paper was the restoration of the original editors, Grenfell & Hunt. Returning to the problem in T. C. Skeat, ZPE 53 (1983) 241, he noted that there was no space for a reference to Epagomene, and that Dio Cassius 51.19.6 states that the Senate had proclaimed the anniversary of the fall of Alexandria would "be taken by the inhabitants of that city as the starting-point in their reckoning of time" --- i.e. that it was the start of the regnal year. He restored the surviving traces of the numeral as [7?], hence arguing that the contract lasted to the end of the full first regnal year of Augustus. It follows that Augustus instituted an anniversary-based regnal year, which in Skeat's view ran from 8 Mesore to 7 Mesore. Assuming the Snyder reconstruction of the Alexandrian calendar, that relationship would be preserved through Roman leap years.

The pOxy 12.1453 synchronism is not completely solid, since the numeral is almost all lost, only the top of the loop survives. While Skeat restored "z" (7), partly on the basis of his chronology, Grenfell and Hunt restored "e" (5) -- partly on the basis of their's. But they also noted that "V" (6) was possible, and "V" was regarded as paleographically equiprobable by G. Geraci, Genesi della provincia romana d'Egitto 160.

In T. C. Skeat, CdE 69 (1994) 308, he adduced additional evidence for this model: a retrospective reference to "Day 19 of Caesar" in the Temple of Dendara, and another contract pRyl. 4.601, an annual lease ending on 7 Mesore year 4; the 7 was a later insertion, indicating uncertainty about the correct date. Finally, he noted that J. R. Rea, JEA 68 (1982) 277, had published SB 16.12469, the lease of a cow in year 5 = 26/5 which runs from Hathyr to 30 Mesore without a change in year number. He argued that this evidence showed that the attempt to implement the Senate decree was still ongoing in year 4, but had failed no later than year 5, and that the regnal year was realigned to 1 Thoth starting in year 6, thereby explaining the later chronographical tradition; year 5 was then 392 days long. Retrospective references to dates in late Mesore were then realigned to the new convention. This left the problem of how to make unambiguous retrospective references to dates between 8 Mesore and 5 Epagomene in year 22 = 1, since it could be politically unwise to assign them to Cleopatra. The Dendera inscription shows how this problem was solved.

In general, I find this a compelling argument. It has been challenged in detail by Grzybek, but I think that Skeat has the stronger case. The one difficulty I have with the reconstruction, as Skeat has presented it, is that it is not obvious why the reform of year 5 was needed. He says that it "must have caused immense administrative problems", but Egyptians had used anniversary-based regnal years throughout the New Kingdom without apparent difficulty. The uncertainty adduced in pRyl. 4.601 does not appear to be about whether the year was anniversary-based or calendar-based, but about which specific date of the year constituted the anniversary. On Skeat's model, this date was fixed at 8 Mesore, and after 3 years of operation it is unclear why there should have been any doubt.

Alternate Models of the Early Alexandrian Calendar

Thus, both the reconstructed Roman calendar and the direct papyrological evidence invalidate the Ideler/Snyder model of the early Alexandrian calendar. However, this still does not amount to positive proof that Theon's account is historically accurate.

The first step to the solution lies in Skeat's analysis of the Augustan regnal years, which is substantially correct, and which argues for a reform in year 5, just as Theon implies. On the reconstructed Roman calendar used here, Alexandria fell on Kal. Sex. A.U.C. 724 = 1 August 30 B.C. = 6 Mesore year 22=1. The next two Roman leap years occurred in A.U.C. 725 = 29 and A.U.C. 728 = 26. The first of these was certainly taken into account in determining the end of year 1. Since the Egyptian calendar was still the wandering year, years 2-5 of Augustus therefore began on Kal. Sex. A.U.C. 725-8 = 7, 7, 7, 8 Mesore. This analysis implies that the date of pOxy 12.1453 should be restored to [6?] Mesore. More significantly, the reason for the confusion Skeat detected in pRyl. 4.601 is now apparent: the start of the regnal year changed at the end of year 4, unexpectedly from an Egyptian viewpoint. In this light, Theon's account that the Alexandrian reform took place in year 5 is now not only substantiated but also motivated.

If the reformed calendar began to diverge from the civil calendar in year 5, there must have been 6 Egyptian leap years before that of A.D. 3, the first in which the nominal relationship between the Alexandrian and Julian years was actually effected in the Roman calendar. There were also 6 Roman leap years in this period (23, 20, 17, 14, 11, 8 BC), so one might entertain a modified form of the Ideler/Snyder model starting in year 5, with leap years at the end of the preceding Egyptian years, i.e. that the sequence was

24, 21, 18, 15, 12, 9 B.C., A.D. 3, 7.....

corresponding to the ends of regnal years 6, 9, 12, 15, 18 and 21. But this (or any phase-shifted variant of it) is still excluded by pVindob L.1c. We may therefore assume a quadrennial intercalary period for the Alexandrian calendar at this time, as Theon implies.

However, there is still a significant objection to assuming the phase of the intercalary cycle implied by the Theon model. Even on the corrected Roman calendar, it is not possible to solve the equation of SB 18.13849 (25 Thoth = a.d. IX Kal. Oct) in the late 20s with this phase. The following table shows the Julian dates of a.d. IX Kal. Oct. and 25 Thoth during the prefecture of Petronius, 25 - c. 20 B.C.:

It will be seen that there is no match. After 21 B.C., the Roman calendar lags behind the Julian calendar until 1 B.C., at which time the correct alignment of Alexandrian and Roman calendars begins to be effective.

This analysis assumes that the Egyptian date reflects the birthday of Augustus in the same Roman year. On that assumption, I see three possible solutions to this problem.

• The birthday celebrations began in the evening of the Egyptian date. All of these alignments, except that of 23 B.C, are separated by only one day, with the Roman date being the day after the Egyptian date. Since Egyptian days ran from dawn to dawn while Greek ones ran from dusk to dusk, it could be argued that the equation is valid for the evening preceding the day of Augustus' birthday.

  This dodge is frequently encountered in Graeco-Egyptian calendrical studies. It has the great attraction of allowing us to declare victory and walk away. However, while it is demonstrably true that the phase difference in the definition of the day between the Egyptian and Macedonian calendars can affect synchronisms given in administrative documents from the reign of Ptolemy II, by this time the administrative calendar was based on the Egyptian one. The phase difference is only encountered in astronomical and astrological texts, and in these an evening event is clearly distinguished as happening on days N to N+1, which is not the case here. Hence we have no reason to distinguish the phase of the Egyptian day from that of the Roman day in this document.

• The original phase of the Alexandrian leap year was different from the phase seen in the first century. The above table assumes that the Alexandrian leap year was regnal year 8. It is clear that if instead the leap year was year 5, 6 or 9, a match would exist in 25, 24 or 21 B.C. respectively. If we number the four possible phases of the quadrennial intercalary cycle from 0 to 3, with phase 3 being the phase of Theon's model, the solutions for phases 0 and 1 are:

The sequence of Alexandrian leap years could then be one of:

• 25, 21, 17, 13, 9, 5 B.C., A.D. 3, 7.... or
  

• 24, 20, 16, 12, 8, 4 B.C., A.D. 3, 7.... rather than
  
• 22, 18, 14, 10, 6, 2 B.C., A.D. 3, 7.... as suggested by Theon

with the possible dates for SB 18.13849 highlighted. Either of these solutions require that there was a phase shift in the Alexandrian intercalary cycle, presumably as an indirect response to the Augustan reform of 8 B.C.

Neither pOxy 4.804 nor pdem Rhind 1 are sufficiently precise to distinguish between these solutions. However, the phase 0 solutions are again positively eliminated by pVindob L.1c. On this phase there is only one Alexandrian leap year after Sextilis was renamed Augustus in A.U.C. 746 = 8 and before the establishment of the normal cycle in A.D. 3. Therefore it is not possible to solve the equation of pVindob L.1c (a.d. XIIII Kal. Aug. = 27 Epeiph rather than 25 Epeiph) on this cycle. On the phase 1 solution, however, there are two leap years, in 8 and 4 B.C., hence pVindob L.1c could be dated specifically to 8 B.C., and SB 18.13849 to Phaophi year 7 = October 24 B.C. (Because of the correction to our model of the Roman calendar implied by pOxy 61.4175, the solution on phase 3 (Theon's model) is moved from 5-2 B.C. to 8-6 B.C.)

Against this date, the ephemeris table of pOxy 61.4175 explicitly shows that year 6 = 25/4 was a regular year of 365 days. Because the table shows both Egyptian and Roman dates, this might imply that it uses the official civil calendar rather than the wandering year. Moreover, Jones noted in A. Jones, ZPE 129 (2000) 159 that later ephemeris tables were based on the Alexandrian calendar. However, the phase 1 solution predicts year 6 = 25/4 to be the first Alexandrian leap year, of 366 days. This would appear to exclude a leap year in 24 B.C.

On the other hand, the wandering year continued to be used for other astronomical calculations in later times, and it is only around this time that it began to diverge from the Alexandrian year, so it is perfectly possible that the Egyptian dates in an ephemeris table of such an early date are still those of the wandering year. If so, the first Alexandrian leap day would be overlooked by the table.

The prefect P. Petronius of SB 18.13849 is not to be identified with the prefect (C.) Petronius of the late 20s. In support of this, one might note the uncertainty about the prenomen of the latter in the literary sources. It is easy to place P. Petronius elsewhere, and to suppose that Pliny was lazy about checking the prenomen of the prefect who had invaded Ethiopia. The list of Augustan prefects is very incomplete, and the prefects of the first decade A.D. are largely unknown. The equation of this papyrus is solved by the ordinary Julian and Alexandrian calendars in A.D. 3 and A.D. 7.

This solution implies the following list of Augustan prefects (after G. Bastianini, ANRW II 10.1 (1988) 503)

It should be noted that even if this solution for SB 18.13849 is correct it doesn't actually prove that Theon's model is correct. It only removes the objection to that model arising from that papyrus.

In other words, it seems we have traded one possible alternative to Theon's solution for another. We have to make the rather odd choice between an otherwise-undocumented phase shift in the intercalary cycle of the Alexandrian calendar and an otherwise-unknown prefect. Both suggestions seem possible, but both are radical conclusions to draw on the basis of such slender evidence.

A more reasonable solution to the problem appears when we consider the evidence adduced by W. F. Snyder, Aegyptus 18 (1938) 197 and W. F. Snyder, Aegyptus 33 (1964) 145, two studies devoted to the emerai sebastai, days marked as "Sebaste" days ("Augustan" or "imperial" days) in Egyptian documents of the first century A.D. Certain days in every month are so marked. Snyder showed that most of these could be tied to known imperial birth days or to other dates of imperial significance recorded in Roman fasti.

For the present purposes, the most important fact demonstrated by Snyder's data is that the emerai sebastai were fixed: they fell on the same day every Egyptian month, and every year, even in the interval between 6 Epagomene and a.d. bis. VI Kal. Mart. This points out a hidden, and flawed, assumption behind the above analysis of SB 18.13849, as well as in Bagnall's: both assumed that the Egyptian dates tracked the Roman ones from year to year (and implicitly from month to month). In fact the Egyptian date of a Roman dies fasti was only accurate at the time it was decreed. It was fixed thereafter. Therefore, we can only say that 25 Thoth = 23 September(R) in the year that the date of the birthday celebration was determined. On the model of the early Julian calendar used here, this was true in 29-27 B.C. So the date of SB 18.13849 can be simply explained by supposing that the date was set in these years, and there is no longer any basis for supposing that there was a phase shift in the leap year cycle of the Alexandrian calendar.

A secondary issue is that neither the 25th (the date of Augustus' birth day given in SB 18.13849, and possibly also in IM 8, sometimes interpreted as giving the birthday of Ptolemy XV Caesarion) nor the 26th (the date we would expect on the Alexandrian calendar) are "Sebaste" days. Snyder argued that Augustus' Egyptian birth day was the 27th, the Sebaste date with the longest duration in the record (from A.D. 37 to A.D. 73), and a date with no other known correlative in the fasti. He proposed to explain it by noting that in certain Roman documents (e.g. Suetonius, Augustus 57) Augustus' birthday was celebrated on two days, 23 and 24 September (a.d. IX Kal. Oct. and a.d. VIII Kal. Oct., which was 23 September in the pre-Julian Roman calendar), corresponding to 26 and 27 Thoth. That Augustus had a monthly feast day was conclusively demonstrated by F. Perpillou-Thomas, Fêtes d'Égypte 173, who noted that pMich 5.244 referred to such a monthly feast day in A.D. 43. Almost certainly the 27th was his Sebaste day.

There was, however, an unnumbered Sebaste day no later than Choiak year 26 = 27 Nov.-26 Dec. 5 B.C., when it is mentioned in pTebt 2.459. No numbered Sebaste day is known from the reign of Augustus. Although unnumbered Sebaste days are known from later reigns, when they almost certainly mark the first of the month, it is unlikely that his birthday would have been ignored during his reign after Sebaste days had been introduced. For this reason, Snyder, Aegyptus 18 (1938) 197 at 231, plausibly argued that those from the reign of Augustus almost certainly were for his birth day, i.e. they were on the 27th.

In any case, the general proposition that the date represents the correspondence between the Roman and Egyptian calendar at the time of the reform argues that the 27th was fixed before 1 B.C., when the Alexandrian and Roman calendars finally reached the correct Julian alignment. For this reason, we can rule out the possibility, suggested above, that P. Petronius was prefect in the first decade A.D. -- indeed SB 18.13849 is now evidence that it cannot be so.

On the model of the early Julian calendar used here, 27 Thoth = 23 September(R) in about half the first 25 years of Augustus' reign, including year 22 = 9/8. This year is the most likely date for the switch from 25 Thoth to 27 Thoth is year 22 = 9/8, since it is also the year of the Augustan reform of the Roman calendar, which also saw the honorific renaming of Sextilis as Augustus and the reform of the Asian calendar, in such a way that the Asian new year began on 23 September(R), and the first month, Dios, was renamed Kaisareios. It seems to me virtually certain that the 27th of each month was honorifically renamed "Sebaste" in the same same year.

Accordingly, I have adopted the Theon model for the conversion tables presented here. Theon has been proven right on every point we can test so far, and there is not a single item of evidence that we cannot simply explain on his model. The fact that the Alexandrian calendar started in year 1 of a Callippic cycle is probably just a fortunate coincidence, but it is certainly one that the designers would have been well aware of, and may well have been factored into the choice of intercalary phase.

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