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 Carlsberg Cycle

This section discusses the Egyptian data showing use of a lunar calendar alongside the civil one, the reconstruction of the Carlsberg cycle for setting the relationship between the two calendars, and the chronological utility of the reconstructed Carlsberg cycle, with particular reference to temple phyle work schedules. The principal sources for this discussion are:

Civil-Lunar Synchronisms

The following is the complete set of explicit synchronisms between civil and lunar calendars from the Ptolemaic and Roman periods that are known to me. I would be interested to learn of any others.

A second class of synchronisms has been inferred from dates given in temple phyle work schedule records. These are more properly considered as an application of the lunar calendar; see below.

  1. Edfu VII.5: Ptolemy III Year 10 III Shomu 7 = snwt = Lunar [III Shomu] day 6 (= 23 August 237)

  2. Edfu VII.6: Ptolemy IV Year 10 III Shomu 7 = snwt = Lunar [III Shomu] day 6 (= 17 August 212)
  3. Edfu VII.7 + IV.2: Ptolemy VIII Year 28, IV Shomu 18 = jpt Hmt dnjt snnw = Lunar III Shomu 23 (= 10 September 142)
  4. Edfu VII.8: Ptolemy VIII Year 30, II Shomu 8 = Hb jnt snwt = Lunar II Shomu 6 (= 2 July 140)
  5. LdR IV 411: Cleopatra VII Year 6, III Shomu 13 = jxt Hr xAwt = Lunar [III Shomu] day 5 (= 13 July 46)
  6. iBucheum 13: Augustus Year 1, IV Peret 21 = mcpr cn-nw = Lunar [IV Peret] day 16 (= 17 April 29)
  7. pdem Rhind 1: Augustus Year 21, III Shomu 10 = Hbc-tp = Lunar [II Shomu] 16 (?= III Shomu 14 (wandering) = 4 July 9)

Equation (3) is actually a composite of two equations which give the date of dedication of the temple with two different formulae:

  1. Edfu VII.7: Ptolemy VIII Year 28, IV Shomu 18

  2. Edfu IV.2: IV Shomu 18 = jpt Hmt dnjt snnw = III Shomu 23

Equation (7) is subject to two types of uncertainty. First, Parker noted that the name given for the 16th day, Hbc-tp, is unusual for a lunar month, the 16th day of which was normally called mcpr cn-nw. However, in R. A. Parker, JNES 12 (1953) 50, he noted two copies of spell against Seth, pBM 10252 and pLouvre 3129, which, in an otherwise essentially identical passage concerning protection against a lunar eclipse, differ only in the name of the day.

More significant is the uncertainty about the calendar of the civil date. If it is treated as a wandering date = 30 June 9, the equation gives a month starting on 15 June, which is 4 days before the lunar conjunction on 19 June 9. If it is treated as date according to the Alexandrian calendar, the match is exact. However, the exact nature of the early Augustan civil calendar is in dispute, so the synchronism of the civil date to the wandering date is not certain, although it is very likely.

It is apparent from these equations, even excluding no 7, that there was a second Egyptian calendar other than the well-known civil calendar. The first four, from the temple of Edfu, were among the original data items that allowed Heinrich Brugsch to infer its existence in the late 19th century, and to establish its lunar nature. The lunar character of the calendar is clear when day 1 of the months given by these equations are compared to the nearest lunar conjunction:


Day 1 of month

Nearest conjunction

Edfu VII.5

18 August 237

18 August 237

Edfu VII.6

12 August 212

11 August 212

Edfu VII.7 + IV.2

19 August 142

17/8 August 142

Edfu VII.8

27 June 140

27 June 140

LdR IV 411

9 July 46

8 July 46

iBucheum 13

2 April 29

2 April 29

pdem Rhind 1

19 June 9 (?)

19 June 9

Given that day 1 was either the day of conjunction or, in two instances, the following day, it is clear that the second calendar in these equations was a lunar calendar.

The Reconstructed Carlsberg Cycle

The following account of the structure of the lunar cycle defined in pdem Carlsberg 9 follows L. Depuydt, In Mem. Quaegebeur II 1277. The papyrus shows how to relate a cycle of 309 lunar months of 29 or 30 days to 25 wandering years of 365 days each.

The papyrus consists of 5 lists:

  1. The beginning of 5 consecutive 25 year cycles: Tiberius (here probably meaning Ti. Claudius) year 5 [A.D. 44], Vespasian year 1 [A.D. 69], Domitian year 14 [A.D. 94], Hadrian year 3 [A.D. 119], Antoninus year 7 [A.D. 144].

  2. A list of the zodiacal signs
  3. A list of numbers arranged in 5 lines
  4. A list of the 25 years of the cycle in 28 lines, consisting of:
    • 1 line of heading
    • 3 lines giving civil dates in year 1
    • 24 lines giving civil dates for Akhet in the remaining 24 years of the cycle
  5. A list of the 9 "great" years (i.e. of 13-months): years 1, 3, 6, 9, 12, 14, 17, 20, 23  

The critical lists are 3 and 4. List 3 contains a set of numbers arranged as follows:


























These represent the civil day dates (in Thoth) of day 1 of the first complete month in the corresponding year of the cycle, arranged so that the difference in civil dates between two successive years is always 11 days, except for the years following years with dates listed in blue (i.e. after years 4, 9, 14, 19, 24), in which case it is 10 days.

The key item is list 4. The first three lines give a full list of dates, but the day number is only given after every second month. Neugebauer, followed by Parker, understood this to mean that the day for the start of the corresponding lunar month was only specified for every second civil month, so that the entry for year 1 should be understood to read:

The following 24 lines gave entries only for the first season of each year, e.g. for year 2:

This arrangement means that the start date of the odd numbered lunar months are unspecified, and are generally ambiguous since the distance between the specified dates are generally 59 = 29+30 = 30+29 days. For Parker, this was a problem -- it seemed to imply freedom of choice for the intermediate dates. Parker then filled in the missing dates as far as he could based on the synchronistic historical evidence.

Depuydt, based on grammatical, paleographical, formatting and logical considerations, plausibly argued that the intent of the papyrus was to apply the dates to the months pairwise:

and for later entries, e.g. for year 2:

with the dates for Peret and Shomu following the difference pattern established for year 1.

The ambiguity that led to the two alternate models, which had actually been noted by Volten, Neugebauer's paleographical partner, arises from uncertainty as to whether a demotic sign represents the word sw (day) without a day number, or is a solar determinative attached to a month name.

Depuydt's interpretation allows the papyrus to specify unambiguously the start dates of 300 of the 309 lunar months in the cycle. The remainder are the 13th months of the 9 "great years". One of these -- the last month of year 14 -- can be determined as 5 Epagomene, since the first days of the previous and next months are separated by 60 days. For the four other 13th months at year end, Depuydt opted for a 30+29 sequence. For the four remaining months, which were blue moons, he associated the blue moon with the preceding civil month.

Depuydt's reconstruction seems eminently sensible, and it is the cycle reconstruction used here. The net result is very close to Parker's reconstruction. Most differences concern the start date of the months beginning in Phamenoth (III Peret) and Epeiph (III Shomu), which Depuydt reconstructs as being one day earlier than Parker.

The Regulation of the Lunar Calendar under the Carlsberg Cycle

The question of whether the Carlsberg cycle actually regulated the operation of the lunar calendar in Ptolemaic times has attracted alot of scholarly attention. Parker argued that the correlations to the lunar synchronisms, and the phyle data considered below showed that it did, and attempted to complete the cycle on this basis, though even then many of the entries were chosen somewhat arbitrarily due to lack of data.

Parker attempted to date the creation of the cycle based on this assumption. Since 309 lunar months is actually about an hour shorter than 25 Egyptian years, the astronomical lunar cycle will drift by about 1 day every 500 years against any 309 month cycle. Noting that in the second century AD most of the entries actually corresponded to the first day after lunar invisibility, and supposing that the cycle originally aligned well with the day of lunar invisibility, he supposed it must have been composed c. 500 years earlier, in the fourth century BC; specifically, he proposed that the first Carlsberg cycle began in 357.

Samuel used Parker's reconstruction to derive a cycle, offset from Parker's by one day, that supposedly controlled the operation of the Macedonian calendar under the early Ptolemies. Other scholars, notably Grzybek and Koenen, have argued for fine modifications of Samuel's reconstruction, proposing slightly different relationships between the Carlsberg cycle and its supposed Macedonian derivative. Indeed B. R. Goldstein & A. C. Bowen, Centaurus 32 (1989) 272, have even argued to invert the relationship, proposing that the Carlsberg cycle was actually derived from the Macedonian calendar in the late third century rather than the other way around. Arguments on this topic have bedevilled the study of the Macedonian calendar for over 40 years; they are considered in the section of this site devoted to the Macedonian calendar.

Depuydt, who deduced an almost completely specified cycle that was slightly different from Parker's, suggested that it was never actually instituted; he described it as "a mere arithmetical Spielerei". However, accepting Parker's criterion of lunar invisibility, he conjectured that the cycle had been designed to align with lunar invisibility on I Akhet 1 in cycle year 1. Focussing just on this date, he noted that the earliest year in which this was true was 282, and so concluded that the cycle had been created at that time.

It is only recently that scholars have begun to address Parker's assumption that the Carlsberg cycle actually controlled the operation of the Egyptian lunar calendar. There are several objections that can be raised against this theory:

In short, we cannot test the proposition with any confidence, and the indication of pRyl 4.589 is that it is false. This does not mean that the Carlsberg cycle is not chronologically useful, since it is a good estimator of astronomical lunar dates. But at this time that is all we can safely assume it is: an estimating tool.

The Regulation of Phyle Work Schedules

In addition to the explicit lunar synchronisms listed above, Parker noted a number of records specifying the start dates of the work schedules of priests in the five temple phyles. A number of additional dates have been discovered since Parker's analysis. The complete list of such dates known to me is as follows. I would be interested to learn of any others.

  1. pdem Ox. Griffith 411: [Ptolemy VIII] year 40.

    II Akhet 20 = Phyle 4 day 1 = 13 November 131

  2. pdem Cair 308012: undated [Ptolemy VIII year 40?]

    IV Akhet 20 (to I Peret 18?)    [= Phyle 5? day 1?] (?= 12 January 130)
    I Peret 19 to II Peret 18         [= Phyle 1 day 1 - 30] (?= 10 February - 11 March 130)
    II Peret 19 to III Peret 18       [= Phyle 2 day 1 - 30] (?= 12 March - 10 April 130)
    III Peret 19 to IV Peret 17      [= Phyle 3? day 1 - 29] (?= 11 April - 9 May 130)
    IV Peret 18 to I Shomu 16      [= Phyle 4 day 1 - 29] (?= 10 May - 7 June 130)
    (I Shomu 1)7 to II Shomu 16   [= Phyle 5 day 1 - 30] (?= 8 June - 7 July 130)
    (II Shomu 17) to III Shomu 16 [= Phyle 1 day 1 - 30] (?= 8 July - 6 August 130)

  3. gr Medinet Habu 433: Ptolemy XII Year 26 = Berenice IV Year 4.

    I Peret 1 = Phyle 2 day 12 (= 4 January 55)
    I Peret 19 = Phyle 3 day 1 (= 22 January 55)

  4. gr Medinet Habu 444: Ptolemy XIII Year 5

    I Akhet 14 = Phyle 1 day 20 (= 17 September 48)

  5. gr Medinet Habu 515: Cleopatra VII Year 11.

    IV Peret 15? = Phyle 4 day 1 (= 16? April 41)
    I Shomu 15 = Phyle 4 day 31? (= 14 May 41)

  6. gr Medinet Habu 476: Cleopatra VII Year 15.

    II Peret 21 = Phyle 1 day 17 (= 19 February 37)

  7. gr Medinet Habu 487: (Ptolemy X) Year (9) = (Cleopatra III) year 1(2)

    II Shomu(?) 1[6?] = Phyle 1 day 1 (= 30 June 105?)
    III [Shomu?] 14 = Phyle 1 day (29?) (= 28 July 105?)

  8. odem Fs. Zauzich 208: (Augustus) Year 35

    IV Akhet 1 = Phyle 1 day 1 (= 20 November AD 5)

  9. gr Medinet Habu 2289: (Claudius? Hadrian?) Year 10

    I Shomu 16 = Phyle [lost] day 1 (= II Shomu 4 (wandering) = 11 May AD 50)
    [II Shomu] 16 = Phyle [lost] day 31 (= III Shomu 4 (wandering) = 10 June AD 50)


    I Shomu 16 = Phyle [lost] day 1 (= II Shomu 23 (wandering) = 11 May AD 126)
    [II Shomu] 16 = Phyle [lost] day 31 (= III Shomu 23 (wandering) = 10 June AD 126)

  10. iMoscow 14510: Nero Year 12.

    IV Peret 23 = [Phyle X?] day 6 (= I Shomu 15 (wandering) = 18 April AD 66)

  11. oThebes D3111 = odem Fs. Zauzich 30: Commodus Year 30.

    IV Peret 28 = Phyle 1 day 1 (= II Shomu 21 (wandering) = 23 April AD 190)
    I Shomu 27 = Phyle 1 day 30 (= III Shomu 20 (wandering) = 22 May AD 190)

  12. odem Fs Zauzich 3212: Septimius Severus & Caracalla Year 8

    I Akhet 8 = Phyle 1 day 1 (= 12 July AD 199)

The dates given in (2), (10) and (11) were the only ones known to Parker. Item (2) is a centrepiece of Parker's analysis.

Items (3) to (7) and (9) were published in H.-J. Thissen, Die demotischen Graffiti von Medinet Habu, in 1989. They all come from the Small Temple of Amun at Medinet Habu. Since the phyle start dates in the Medinet Habu graffiti are typically one day later than the dates of the corresponding months in the Carlsberg cycle, Thissen, after consultation with Parker, based his analysis, not on Parker's reconstruction of the Carlsberg cycle, as one might expect, but on the cycle that Samuel had derived from it for analysis of the Macedonian calendar. Thissen supposed that this cycle was adopted after Ptolemy IX's reduction of Thebes in the early 80s. Just why the Ptolemaic government should impose the Macedonian calendrical cycle on an Egyptian temple a century after the regime itself had ceased to use it was not explained. Further, item (9), also from Thebes, does match the Carlsberg cycle, so Thissen's theory appears to require that Theban temple phyle schedules reverted to the Carlsberg cycle at some time after the reign of Cleopatra.

These graffiti have attracted some chronological commentary, notably from M. Chauveau in B. Kramer et al. (eds), Akten des 21. Internationalen Papyrologenkongresses I 163 (items (3) and (4)), with an additional note on item (7) in C. J. Bennett, ZPE 147 (2004) 165 which I now see was jejune. These discussions, and the paleographical comments by M. Chauveau, RdE 46 (1995) 250, have suggested adjustments to some of Thissen's original dates. However, they have not, to date, been subject to calendrical analysis beyond Thissen's surprising and rather implausible conjecture that they were Macedonian lunar months.

The items should all be treated together. They require some commentary, since there are uncertainties associated with most of them.

  1. pdem Ox Griffith 41, from the temple of Soknopiau Nesos at Dime in the Fayyum. The name of the king is not given but is certainly Ptolemy VIII from the year number and the general Ptolemaic context of these papyri.

    The key line (line 4) was read by E. Bresciani, the original editor, as HA.t ir 2 Ax.t sw 19 pA wrS tw-n and interpreted as "Era il giorno 19 di Paofi, la veglia; noi andiammo al templo" (i.e. On 19 Phaophi [it was] the monthly service, we went to the temple...) which suggests a service starting on 19 Phaophi. However, Karl-Theodor Zauzich notes that this reading cannot be right since it duplicates ir (2) and it should be: HA.t 2 Ax.t sw 19 pA wrS tw-n. Noting that HA.t then means "before", and that this is slightly ungrammatical, he translates the intended sense as "[Es war (xpr)] der 19. Paophi, (der Tag) vor dem Monatsdienst, da sind wir in den Tempel gekommen...", i.e. "On 19 Phaophi, the day before the monthly service, we went to the temple".

    My thanks to Prof. Zauzich for bringing this case to my attention and for discussing its interpretation with me.

  2. pdem Cair 30801, from Gebelein, contains an account of grain deliveries giving this list of phyle schedule dates on the recto. All dates except the first are explicitly given as being phyle dates. The date of IV Akhet 20 is given in a summary of the grain deliveries and is presumably also a phyle schedule date. The dates clearly show that the phyle schedule was controlled by a lunar structure, and that the phyles normally observed their work schedules in order of phyle number.

    However, the recto is not dated. Parker argued that

    (a) the mention of a fifth phyle shows that the papyrus postdates the creation of that phyle by the Canopus Decree of 237
    (b) the paleography is of later Ptolemaic date
    (c) the verso contains an account of grain deliveries in a different hand dated to year 41, which must therefore be of Ptolemy VIII
    (d) the recto was always written before the verso
    (e) the even month dates match the pCarlsberg dates for cycle year 13. Therefore the recto dates to the last cycle year 13 before year 41 of Ptolemy VIII = year 26 of Ptolemy VIII = 145/4.

    Steps (a) to (d) are unexceptionable. But step (e) is based on two unproven assumptions: that the phyle schedule corresponds to the first day of the lunar month; and that the lunar month is in fact controlled by the Carlsberg cycle. For this reason, pdem Cair 30801 cannot legitimately be used, as Parker attempted to do, to establish that the cycle controlled the lunar calendar. However, it does clearly establish that phyle schedule dates have a lunar basis. Further, it shows that phyle dates were given for the first and last days of phyle service, at least in retrospective accounts.

    To anticipate: It will be seen that the Medinet Habu graffiti contradict the first assumption. It appears that phyle schedules are in fact best correlated with day 2 of the synodic lunar month, which at this period means day 2 of a Carlsberg month. On this basis, the best match for pdem Cair 30801 turns out to be Carlsberg cycle year 2, and the latest cycle year 2 before Ptolemy VIII year 41 is year 40 = 131/0. This is actually a much more satisfactory dating than Parker's given the ephemeral contents of the papyrus: it implies that the recto and verso are used in consecutive years rather than 15 years apart.

  3. While the date is certain, the second year number was originally read by Thissen as year 3. The correct reading of year 4 was established by M. Chauveau in B. Kramer et al. (eds), Akten des 21. Internationalen Papyrologenkongresses Berlin 13.-19.8 1995 I 163, after examining all double dates associated with Berenice IV.

  4. Thissen assigned this graffito to year 5 of Ptolemy XII and Cleopatra V, but noted that the phyle date appeared to lack any correlation with the Carlsberg cycle on this year. M. Chauveau in B. Kramer et al. (eds), Akten des 21. Internationalen Papyrologenkongresses Berlin 13.-19.8 1995 I 163, noted that the queen was not named as "Cleopatra Tryphaena" and that the royal couple are described as "Philopatores" but not as "Philadelphoi". For these reasons he identified them as Ptolemy XIII and Cleopatra VII, and noted that in this year the date of the inscription matched Samuel's modification of the Carlsberg cycle, as Thissen had applied it to the other dates of this series.

  5. Thissen read these two dates as IV Akhet 19 and I Peret 19. These readings are problematic, since the date of the end of service is the 31st day of service. This could only be correct if at Medinet Habu the last day of service corresponded to the first day of service of the next phyle, but such a convention is not known in any other source (cf. no 1). In any case, M. Chauveau, RdE 46 (1995) 250 at 253 no. 51, corrected the reading of the end date to I Shomu 15, and the starting month to IV Peret. Accepting Thissen's theory that the phyle cycle was regulated according to Samuel's extension to the Macedonian calendar of Parker's reconstruction of the Carlsberg cycle, Chauveau proposed IV Peret 17 for the start date, but considered the restoration paleographically doubtful. In personal correspondence, he noted that the two day numbers appeared to be identical, implying a 31st day of service regardless of the value.

  6. There are no concerns with the date of this graffito.

  7. The date of the graffito is the most problematical in the series. The months are recovered from traces of the season of the start of service, read as "Akhet" by Thissen. The day number of the first day must be based on the assumption that the 14th was the last day of service, but it could have been a 29 or 30 day interval, or even 31 days per Thissen's reading of gr. Med, Habu 51, implying 14, 15 or 16.

    As to the year, Thissen read the second regnal date as "year 10" and restored the synchronism as:

    Year (21?) Ptolemy (VI?) and Cleopatra (II?) = Year 10 (Ptolemy VIII?), II Akhet 1[5] to III [Akhet] 14 (= 15 Nov - 14 Dec 161).

    Year 21 Ptolemy VI corresponds to Carlsberg cycle year 22, in which the lunar month start dates are Phaophi 14 and Hathyr 13 -- again, one day earlier than Parker's dates. However, there was no coregency at this time, and a proleptic date of Ptolmy VIII requires justification and support. At the time Thissen wrote, it was apparently supported by a date of year 20=9, but M. Chauveau, RdE 50 (1999) 272, noted that the demotic numeral for "9" was easily confused with that for "5", and showed that this date should be read year 20=5, which is a perfectly normal date for Cleopatra VII.

    In personal correspondence, Michel Chauveau noted that Thissen's reading "year 10" is in error, and corrected it to "year 11"; several other demoticists I have consulted on the point concur. This changes Thissen's candidate date to year 22 (Ptolemy VI) = year 11 (Ptolemy VIII). Not only does this date suffer from the same objection, but the Carlsberg cycle year, year 23, has start dates of 3 Phaophi and 2 Hathyr, which are clearly not compatible.

    The obvious alternate double date is year 14 Cleopatra III = year 11 Ptolemy X. This corresponds to cycle year 4, in which the Parker start dates are 28 Phaophi and 27 Hathyr, which is obviously, neither of these alternatives are close to the dates given by the graffito. In personal correspondence, Karl-Theodor Zauzich has suggested another possibility, that Ptolemy X and Cleopatra III are named and dated in the reverse of the usual order, dated to 18 Hathyr year 8 (of Ptolemy X) = year 11 (of Cleopatra III). This corresponds to cycle year 1, in which the cycle start dates are 1 Phaophi (Parker) or 30 Phaophi (Depuydt) -- again not remotely close to the mid month date given by the graffito.

    A second difficulty is that the dating formula in the graffito names a king Ptolemy before a queen Cleopatra. This is not a problem with Thissen's solution, since Ptolemy VIII was not actually resident in country at the time, nor with Zauzich's proposal, but it does create difficulties for any ordinary double date of Cleopatra III and Ptolemy X.

    Next, there is the remote possibility that the double date represents a series that is otherwise unknown. On this hypothesis, the first year is most likely a year 1. We are then looking for a single regnal year 11 which corresponding to a cycle year in which lunar II Akhet, II Peret or II Shomu start on 14, 15 or 16 of the coresponding civil months. For II Akhet, those are cycle years 8, 19, and 22; for II Peret, cycle years 5 and 19; and for II Shomu, cycle years 2, 5, and 16. There is one, and only one, candidate on this hypothesis: year 11 of Cleopatra VII, cycle year 16, II Shomu (Payni) 15 = 13 June 41. Lunar conjunction occurred at 00:19 on 13 June 41, indicating that invisibility was 12 June, so the date matches theory perfectly. In possible support of this, one might note that the recorded service is for phyle 1. This is only two months after the phyle service for phyle 4 recorded in gr Medinet Habu 51 (item (5)), so phyle 1 is the phyle that we would normally expect to serve.

    Although this is a good match to our numerical data, it is open to two very serious, if not fatal, objections:

    1. The hypothetical double dated series is not only not documented anywhere else, it is not documented in gr Medinet Habu 51 (item (5)).

    2. We would still have to explain the remarkable precedence of Ptolemy XV over Cleopatra VII.

    The only reason I am willing to consider this solution as remotely possible is Iseum stele 1970/52 = H. S. Smith, RdE 24 (1972) 176, 186 n. 20. This stele is dated to year 11 of Ptolemy "pA Wynn" -- Ptolemy "the Greek", and names a mother of Apis who is otherwise only known from stele H.5-1887 dated 18 Mesore year 11 of Cleopatra VII = 15 August 41. The two stelae name the same workman and both record the opening of the catacomb "in one night", hence they have the same date, which is very shortly after the putative date of gr Medinet Habu 48. Conceivably, the double date could reflect a short-lived era of Ptolemy "the Greek". (For his identity see discussion under Ptolemy XV).

    One further, and in my view more likely, possibility for the date of this graffito has been suggested to me by John Gee (pers. comm. May 2006): that "11" is actually a partially-erased "1[2]". Steve Vinson and Karl-Theodor Zauzich note that this requires that Edgerton failed to notice that there was a partial erasure, which they consider unlikely. Further, while Gee and Zauzich consider that the surviving traces of the first year number could be compatible with a [9] (though not a [15]), it would have to be an unusual form of the numeral. However, the double date year 9=12 would be compatible with Ptolemy X being named first. More importantly, on this reading there is a reasonable lunar solution for the phyle service date: year 9=12 of Ptolemy X & Cleopatra III, cycle year 2, II Shomu (Payni) 16 = 30 June 105. According to PLSV 3.0, lunar invisibility at Thebes was on June 30 105. III Shomu 14 = 28 July 105 was the date of last visibility for that month.

    At this time, the date of this graffito has to remain uncertain. However, I feel fairly confident that the correct solution is year 9=12 of Ptolemy X and Cleopatra III. Further examination of the graffito in situ may confirm that the "11" is partially erased, although I am told by Christina di Cerbo that it is currently covered by a thick layer of mud which has washed in from the roof of the small temple.

    My thanks to Christina di Cerbo, Michel Chauveau, John Gee, Richard Jasnow, Heinz Thissen, Steve Vinson and Karl-Theodor Zauzich for their great kindness in discussing this graffito with me at various times. For any demoticist reading this who is interested in trying his or her hand at reconciling the date of this graffito, I reproduce here Edgerton's facsimile of its first few lines.

  8. This item is a 2-month lease of temple service from 1 Choiak to 30 Mecheir of year 35. It is dated by Kaplony-Heckel (in H. Hoffmann & H. J. Thissen, Res Severa Verum Gaudiam, 283 at 300 and n. 89) to year 35 of Ptolemy IX, on the grounds that the paleography is first century and the names on the contract are typically Ptolemaic rather than Roman. However, there is no close match to the start of a lunar month or the Carlsberg cycle for Ptolemy IX (or Ptolemies II, VI or VIII), so Augustus is the only possible solution.

    Note that this item is solved using the calendar of the wandering year, not the Alexandrian calendar.

  9. This item was omitted from Thissen's list of double dates. The graffito also includes dates from years 19 and 9, but there is no clear indication of which king is involved; nor even that is of Ptolemaic date. Other graffiti from the same location (Room "S" of the small temple -- including item (7)) are clearly of late Ptolemaic date.

    The closest match I have found within the first century for this graffito is 16 Pachon year 10 of Claudius = 11 May AD 50 if the Alexandrian calendar is assumed. The nearest previous year 19 is year 19 Tiberius = AD 32/3. This corresponds to 4 Payni of Carlsberg cycle year 6 in the wandering year, which is day 2 of the Carlsberg month starting on 3 Payni. Conjunction was at 16:19 on 9 May AD 50, giving invisibility on that day. This match appears to be a day late, unless the month was based on final observation of the setting crescent on 8 May, in which case day 1 might be set at 10 May.

    Two other possible matches exist in the second century. While this seems rather late, compared to the other graffiti, the Small Temple of Amun at Medinet Habu was in active use throughout that century, with major work being done under Antoninus Pius. These matches are:

    • Year 10 of Trajan = 11 May 107 = 18 Payni (wandering), conjunction at 10 May 107 at 12:28. Carlsberg cycle date: 17 Payni year 13. However, the nearest previous year 19 is still Tiberius, which probably eliminates this solution from consideration.

    • Year 10 of Hadrian = 11 May 126 = 23 Payni (wandering), conjunction at 10 May 126 at 05:47. Carlsberg cycle date: 22 Payni year 7. The nearest previous year 19 is year 19 Trajan = AD 115/6.

    Note that this item is solved in all cases by using the Alexandrian calendar, not the calendar of the wandering year.

  10. This item is the only one that falls within the period explicitly covered by pCarlsberg 9.

    There is one possible ambiguity with this inscription. From the original publication in W. Spiegelberg, ZÄS 66 (1930) 42, it is clear that no phyle was mentioned, rather the date is simply given as 23 Phamenoth year 12 = "pA hrw n mH 6 n p3 wrS" -- the 6th day of the wrS. For Spiegelberg, wrS indicated a lunar date tout court. R. A. Parker, The Calendars of Ancient Egypt 18 noted that the same term appears repeatedly in pdem Cair. 30801 (item (2)), and there clearly indicates a period of phyle service. The same is true in pdem Ox. Griffith 41 (item (1)).

    Parker argued that it should be understood the same way in this inscription. However, the absence of reference to a phyle in the inscription opens up the possibility that the term delimits a period of service directed by the author of the inscription, Parthenios, who directed works in the temple of Koptos from Tiberius to Nero, which could be unrelated to phyle service. Nevertheless, K-T. Zauzich, pers. comm., assures me that in his experience the term is exclusively used for phyle service, and should be so understood here.

    As noted by Parker, this item is solved using the Alexandrian calendar, not the calendar of the wandering year.

  11. This item postdates the period that is explicitly covered by pCarlsberg 9. The year was read as 12(?) in the original publication but Parker regards "30" as "perfectly clear"; although faint in the photograph, I can see why he says this. Commodus' Egyptian years are based on the epoch of his father's accession rather than his own. Parker ignored the end date in his analysis since, as he correctly noted, it was not possible to tell whether it was the last day of the work schedule or the first day of the next.

    As noted by Parker, this item is solved using the Alexandrian calendar, not the calendar of the wandering year.

  12. This item is dated by the Pharaohs Sebastoi. It is dated by Kaplony-Heckel (in H. Hoffmann & H. J. Thissen, Res Severa Verum Gaudiam, 283 at 314ff.) to year 8 of Septimius Severus and Caracalla = AD 199/200, rather than year 8 of Marcus Aurelius and Verus = AD 167/8. This is certainly correct. This ostracon, together with odem Thebes 221 = odem Fs Zauzich 33, dated to year 11 of the Pharaohs Sebastoi, and odem Fs Zauzich 34, dated to year 12 of an unspecified pharaoh, all record leases of temple service to the same individual: the god's father Chapochonses, son of Horus; the highest year of Marcus Aurelius and Verus reigning jointly was year 9 = AD 168/9.

    The month of service is dated to 8 Thoth. The year is unspecified. If the lease was agreed in the first few days of Thoth it could be 8 Thoth year 8. Otherwise, the lease was agreed at the end of year 8 for a month of service starting in 8 Thoth year 9. Only 8 Thoth year 8 matches the lunar cycle.

    Note that this item is solved using the calendar of the wandering year, not the Alexandrian calendar.

In summary, of the 19 dates available for the start of phyle service, only 8 are secure, and even 3 of these (items (1), (10) and (11)) are not entirely without difficulty. Of the 12 dates available for the end of phyle service only 2 are secure. Since there is clear evidence for each of the two possible conventions for marking the end of phyle service (nos. (2), (5) and (10)), it is not safe to infer dates of end of service from the dates of start of service of the next phyle, or vice versa.

The following table shows, for each of the 8 secure dates of start of service (per the wandering year), the corresponding Carlsberg cycle year, the first day of the nearest Carlsberg month, and the difference between the two.

Item number

Egyptian Date

Carlsberg Year

Carlsberg Month

Difference (Days)


20 Phaophi yr 40 Pt VIII


20 Phaophi



20 Choiak yr 26 Pt XII


19 Choiak



19 Tybi yr 26 Pt XII


18 Tybi



30 Mesore yr 4 Pt XIII


29 Mesore



5 Mecheir yr 15 Cl VII


4 Mecheir



10 Pachon yr 12 Nero


10 Pachon



21 Payni yr 30 Commodus


21 Payni



8 Thoth yr 8 Severus


7 Thoth


The following table shows, for each of the 8 secure dates of start of service, the corresponding Julian dates, the nearest dates of conjunction and invisibility, and the difference between the Julian date of the start of phyle service and the date of lunar invisibility. Dates of invisibility are calculated using PLSV 3.0. Note that the dates of invisibility calculated for items (10) and (11) are a day later than those calculated by Parker, representing improvements in our knowledge of key parameters, notably DT, the correction for secular acceleration in the earth's rate of rotation in the last two millenia.

Item number

Julian Date


Difference (Days)


13 November 131

12 November



24 December 56

23 December



22 January 55

21 January



29 August 48

28 August



3 February 37

2 February



13 April AD 66

12 April



23 April AD 190

21 April



12 July AD 199

11 July


The size of the dataset is still too small to permit definitive conclusions, and the match is not perfect. Item 11 is probably due to a false observation of a last crescent. The lunar phase at dawn on 20 April AD 190 was 4.3% and 1.2% on 21 April. B. E. Schaefer, QJRAS 37 (1996) 759, estimates that the error rate for reporting new crescents is about 15%, although lower for trained observers; presumably the error rate for old crescents is comparable. Nevertheless, these results suggest that the start of phyle service was not intended to occur on the date of the new moon, as Parker assumed, but on the following day. If this conjecture is correct we can also explain why the phyle months of the Medinet Habu graffiti are consistently a day late on the Carlsberg cycle while the Roman era dates match: the Carlsberg cycle never was intended to apply to pre-Roman dates. The papyrus itself provides no indication of use before AD 19 or 44, and the Medinet Habu graffiti appear to indicate that extrapolation back to Ptolemaic times does not match the phyle data.

As noted in the discussion of items (2) and (7), this model permits solutions to be derived for these graffiti. In the case of item (2), the derived solution, year 40 of Ptolemy VIII = 131/0, is clearly preferable to Parker's, since it immediately precedes the record of year 41 on the verso.

The phyle dataset contains one other possible chronological indicator that Parker was unable to make use of: the phyle number associated with a month of service. Parker could not use this data because each of the three items he studied came from different temples. However, the Medinet Habu graffiti all come from a single temple at Medinet Habu and those which are certainly dated cover a period of about 20 years. We may therefore test the natural conjecture that the phyles served in cyclic sequence throughout the period covered by the graffiti. If this conjecture is correct then, given knowledge of one month of service of one phyle, we should be able to predict the month of service for each of the other graffiti. Conversely, if the phyle cycle was maintained without interruption, a date of phyle service that is somewhat uncertain may be resolvable on the basis of the phyle number alone.

The following table shows the expected and actual phyle numbers for the service months recorded in the Medinet Habu graffiti, and the size of the discrepancy, if any. The graffiti cover service from phyle 2 in month 4 (Choiak) of cycle year 2 (item (3)) to that of phyle 1 in month 6 (Mecheir) in cycle year 20 (item (6)). For each graffito, the expected phyle is calculated based on the actual phyle of the previous graffito.

Item Number

Carlsberg Month

Expected Phyle

Actual Phyle



Year 9 Month 13





Year 16 Month 8





Year 20 Month 6




A second opportunity is provided by the set of temple service contracts published by U. Kaplony-Heckel in H. Hoffmann & H. J. Thissen, Res Severa Verum Gaudiam, 283. In addition to numbers (8) and (12) listed above, these contracts include another which is not usable for analysing the start date of a month of phyle service, but which records the phyle number of an identifiable month of service. Number 32 (item (12) above) identifies Phyle 1 as the service file at Djeme in month 1 of cycle year 6 = year 8 of Septimius Severus. Number 33 is a contract dated 1 Mesore of year 11 of Septimius Severus for service of Phyle 3 in Mesore at Djeme = month 13 of cycle year 9. This is 49 months later; the expected phyle is phyle 5. Thus, again we see an interruption in the expected phyle sequence.

Since the number of lunar months between any two phyle service months is independent of the Carlsberg cycle, these discrepancies cannot be explained by supposing that a different cycle was in use. It appears that the cycle of phyles was interrupted from time to time. The following model is exempla gratia. Between items (3) and (4), it appears that a phyle took service in two consecutive months, or, perhaps less likely, phyles were excused service for a total of 4 times. Between items (4) and (5) it appears that a phyle was excused service on one occasion. It is possible that the two-month lease of item (8) illustrates exactly such an interruption in the order of phyle servicw.

It remains possible that the order of phyle service, while not continuously cyclic, was determined by some other systematic algorithm. However, unless such an algorithm can be determined, it appears that a phyle number cannot safely be used to distinguish between two alternatives for phyle service dates.

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