[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 Alexandrian reform
  
• The Soter era

The Canopic reform

On 7 Appellaios (Mac.) = 17 Tybi (Eg.) year 9 of Ptolemy III = 7 March 238, a conclave of the Egyptian priesthood held in Canopus at the command of the king issued the Canopic Decree (OGIS 56) which, among other things, changed the civil calendar from the wandering year of 365 days to a fixed year by the intercalation of a leap day at the end of the year in every fourth year, so that the heliacal rising of Sothis (Sirius) would occur on a fixed date, 1 Payni:

And whereas feasts of the Benefactor Gods are celebrated each month in the temples in accordance with the previously written decree, the first (day) and the ninth and the twenty-fifth, and feasts and public festivals are celebrated each year in honor of the other greatest gods, (be it resolved)

for there to be held each year a public festival in the temples and throughout the whole country in honor of King Ptolemy and Queen Berenike, the Benefactor Gods, on the day on which the star of Isis rises, which is reckoned in the sacred writings to be the new year, and which now in the ninth year is observed on the first day of the month Pauni, at which time both the little Boubastia and the great Boubastia are celebrated and the gathering of the crops and the rise of the river takes place;

but if, further, it happens that the rising of the star changes to another day in four years, for the festival not to be moved but to be held on the first of Pauni all the same, on which (day) it was originally held in the ninth year,

and to celebrate it for five days with the wearing of garlands and with sacrifices and libations and what else that is fitting;

and, in order also that the seasons may always do as they should, in accordance with the now existing order of the universe, and that it may not happen that some of the public feasts held in the winter are ever held in the summer, the star changing by one day every four years, and that others of those now held in the summer are held in the winter in future times as has happened in the past and as would be happening now, if the arrangement of the year remained of 360 days plus the five days later brought into usage (be it resolved)

for a one-day feast of the Benefactor Gods to be added every four years to the five additional days before the new year, in order that all may know that the former defect in the arrangement of the seasons and the year and in the beliefs about the whole ordering of the heavens has come to be corrected and made good by the Benefactor Gods.

(trans. as given by R. A. Bagnall)

This reform is essentially identical to the Alexandrian reform undertaken under Augustus, except that it fixed the Julian equivalents of the reformed Egyptian year to those of the wandering year of the early 230s BC rather than those of that of the mid-20s BC. Thus, under the Canopic calendar, 1 Thoth = 21-23 October, while under the Alexandrian calendar 1 Thoth = 29/30 August.

This page considers the phase of the leap year cycle of the Canopic reform, and the evidence for its longevity provided by the Macedonian/Egyptian double dates of the later Ptolemies.

The Intercalary Phase

The decree does not explicitly state which year of the four year cycle starting in year 9 of Ptolemy III would end in a 6th epagomenal day. However, we are told that the reform was intended to fix the date of the heliacal rising of Sothis at 1 Payni, and that a festival to be celebrated in honour of Ptolemy III and Berenice II and lasting for 5 days was to start on that day, with an additional festival on the 6th epagomenal day. We may therefore proceed by investigating what is known about these events.

Since the date of the Sothic rising was said to advance by 1 day every four years against the 365-day Egyptian calendar, we may speak of a "Sothic quadrennium" of 365+365+365+366 days. This is the same quadrennium that underlies the Julian leap year cycle, but the two are not necessarily in phase. Noting that 237 was a Julian leap year (as highlighted in green), there are four possible relationships between Canopic quadrennia and Julian quadrennia, as shown in the following table. The 6-day Epagomenes, and the Julian dates of 1 Payni, are highlighted. For reasons discussed below, the most likely cycle is that in which 6 Epagomene fell at the end of cycle year 1 or 2.

A second rising of Sothis is reported on 1 Thoth = 20 July AD 139 in Censorinus 21.10, writing in AD 238. He notes that it marked the start of the Egyptian "Great Year", and occurred exactly 100 years before the birthdate of his patron (counting inclusively) on 20 July AD 238. This gives us the phase of the Sothic quadrennium: AD 139 is the first year Sirius rose on 1 Thoth. Hence the Sothic quadrennium was fixed at 19, 19, 19, 20 July in the second century AD (expressed in phase sync with the Julian cycle, i.e. from AD 136-139, as is conventional).

The reliability of Censorinus' account has recently been challenged by P. F. O'Mara, JNES 62 (2003) 17, who argued that Censorinus' prime purpose was to exalt the birthday of his patron, and therefore that his data need not be historically accurate. He noted that Censorinus does not elsewhere show any familiarity with Egyptian practice, and that various ancient authorities give a number of dates for the heliacal rising, ranging from 19 July to 22 July. He supposed that Censorinus selected the date most appropriate to his purpose, as occurring exactly a century before the birthday of his patron. In particular he notes that Censorinus' date is unsourced, and that he states that Sothis "customarily" rises on 21 July (a.d. XII Kal. Aug.) in Egypt, which is conventionally amended to 20 July (a.d. XIII Kal. Aug.), but the centennial anniversary he describes actually requires that Sothis rose on 19 July for 3 years out of 4. He concludes that Censorinus' account could well be explained by his having simply subtracted 25 days from 20 July to find the Julian equivalent of 1 Thoth in AD 139, and then selecting the date given by an authority for the Sothic rising that fits that result. Therefore, he concludes, the date of the Sothic rising given in Censorinus cannot be assumed to be correct. By way of illustration, he suggests that Censorinus would have reached exactly the same conclusion if he had been writing in AD 235, 236 or 237 -- i.e. he would have using exactly the same reasoning to date the start of the Great Year to AD 136, 137 or 138.

O'Mara makes some interesting and worthwhile points about Censorinus' motives, his lack of expertise in Egyptian matters, and his understanding and manipulation of the facts he presents. However, he does not show (or argue) that Censorinus changed any of his source material, only that he selected material that met his need; nor does he show that Censorinus selected any facts which are actually incorrect. The most he shows is that Censorinus makes some questionable use of his facts on his client's behalf. Hence the more relevant question is how reliable were Censorinus' sources.

Censorinus' known sources (Varro, Suetonius etc.) being reputable (or considered so in his time), one might argue against O'Mara that, since Censorinus is not deeply familiar with Egyptian material, but does generally present accurate material, the fact that he chose this particular datum shows that it was actually well known in his time. In any case, given the minor role Egyptian material plays in his work, it is hard to imagine that he plucked the coincidence out of the air, as O'Mara implies: he got it from somewhere. The fact that he slightly misrepresents the significance of the datum, by explaining that 2[0] July was the "customary" date, rather than the date on which the rising was observed, strongly suggests to me that in fact the datum itself is accurate.

As to relevant errors in Censorinus' account, O'Mara cites two:

As noted, Censorinus gives 21 July as the "customary" date for the rising, not 20 July. However, O'Mara himself accepts the usual MS emendation from 21 to 20 July, since 21 July is not consistent with the equation 1 Thoth = a.d. VII Kal. Iul. (25 June) which Censorinus gives, correctly, for 1 Thoth in AD 238 (and it is easy to see how "a.d. XIII Kal. Aug." could be corrupted to "a.d. XII Kal. Aug.").

The standard emendation goes back to Scaliger (J. J. Scaliger, De Emendatione Temporum, 138). In view of the debate over whether the Egyptian day began at dawn or sunrise, it is perhaps worth noting that 21 July could in fact be the correct date for the heliacal rising, if 1 Thoth began at sunrise on 20 July AD 139. However, Censorinus' wording does suggest that the equation 1 Thoth = 2[0] July in AD 139 was retrocalculated from the equation 1 Thoth = 25 June in AD 238 (if not necessarily by him), and 21 July is not the correct result for that. Moreover, 20 July is consistent with modern astronomical retrocalculations (see below).

Censorinus equates AD 238 with year 267 of the era of the Roman rule in Egypt (starting in 30 BC), when, in O'Mara's view, it should have been year 268. But the number 267 is in fact correct, as has long been known (cf. A. T. Grafton & N. M. Swerdlow, CQ 35 (1985) 454 at 454). The point is to understand how it was calculated. While Censorinus states that he calculates his Roman years based on a new year of 1 January, he also clearly states that Egyptian years start on 1 Thoth. He sets the epoch for the Egyptian era by stating that Roman rule in Egypt started two years before Augustus took the imperium, in February 27 BC. In Julian terms, Roman rule in Egypt started three years before Augustus took the imperium, but in Egyptian terms Augustus' assumption of imperium fell in his Egyptian year 3. Therefore, in this instance Censorinus is calculating the year of the Egyptian era according to the Egyptian (Alexandrian) calendar, and 20 July AD 238 falls in Egyptian year 267 of an era starting on 1 Thoth in 30 BC, not year 268. It is O'Mara, not Censorinus, who is wrong.

O'Mara's hypothetical scenario, intended to show how Censorinus might have set the start of the Great Year to any year in the interval AD 136-139 if he had been writing in any year in the range AD 235-238, is difficult to sustain. The purpose of Censorinus' tract was to glorify his patron, here by noting that his birthday marked the start of year 100 of the Egyptian Great Year. While Censorinus is the earliest surviving author to explicitly attest to a Sothic Great Year, he surely did not invent it. His primary source appears to have been Varro's lost Antiquitates Rerum Humanarum. Varro certainly made use of Egyptian data and may well have been mentioned the Sothic Great Year. The concept of the "Great Year" was well-known in his time, and it is evident from the Canopic Decree that the specific concept of the 1460/1461-year Sothic cycle, which defines the period of the Great Year, was well understood by Alexandrian astronomers. Therefore there is every reason to believe that the concept of the Sothic Great Year is much older than A.D. 139. The actual date of the start of the Sothic Great Year would certainly have been observed and would have been known to Censorinus' astrologically literate contemporaries. If he had got it wrong, he would have opened up his patron not to praise but to ridicule.

Once we accept that the Great Year was not Censorinus' invention, it is easy to see how he could have reached his result without doing any calculations at all. He leads up to the Sothic equation by giving the (correct) current year number in multiple different epochs: Ol. 254.2 = A.U.C. 991 = Caesar 283 = Augustus (Julian) 265 = Augustus (Alexandrian) 267 = Nabonassar 986 = Philip 562; this 7-calendar synchronism was, for Scaliger and other early modern chronologists, one of the most important data items they used to establish ancient chronology. Censorinus also goes into great detail about the start date of each year, in preparation for explaining the anniversary of the Great Year. The last two of the dates, those of the eras of Nabonassar and Philip, are based on the same wandering Egyptian calendar as the Great Year. All three of these eras are astronomical; indeed, Censorinus is the only surviving ancient source for the first two outside the Almagest. Almost certainly, he got all of these dates from a single source, and since this source included two other astronomical dates it is surely also the source of the third.

Finally, not discussed by O'Mara, but long ago noted by L. Borchardt, Die Annalen und die zeitlichen Festlegung des alten Reiches der ägyptischen Geschichte 55, Alexandrian coins of years 2 and 6 of Antoninus Pius = AD 139 and AD 143 are known showing a phoenix and bearing the legend AIWN. While these coins do not specifically mention Sothis, they fit perfectly with the notion that they mark the beginning and the end of the quadrennium in which the heliacal rising of Sothis fell on 1 Thoth, marking the start of the Sothic cycle, and no other explanation is known. If this event received so much public attention that it was commemorated in coinage, it is quite reasonable to suppose that it was still well-known a century later.

In short, O'Mara has not proved his case, and the conventional analysis of Censorinus is much more likely than not. The remainder of the discussion here assumes that the standard interpretation of Censorinus is in fact correct and that it accurately reflects what happened.

Given the phase of the Sothic rising against the Julian quadrennium in AD 139, and assuming, with the ancients, that the period of the Sothic cycle against the Egyptian wandering years was exactly fixed at 1460 (Julian) years, one would expect the phase of the Sothic rising at the time of the Canopic Decree to be the same. However there is an immediate problem: 238 BC, like AD 139, is the last year of a Julian quadrennium, but in this year 1 Payni fell on 19 July, not 20 July.

One possible explanation of the discrepancy between Canopus and Censorinus was advocated by L. Borchardt (as reported by R. Krauss, Sothis- und Monddaten, 59): that the Canopic decree was based on observations made to the south of Alexandria/Canopus. In general, the heliacal rising in Egypt occurs about 1 day earlier for each additional degree of latitude south of the Mediterranean coast. Borchardt's preferred sites, Memphis or Heliopolis, are about a degree south of Alexandria, hence a heliacal rising of 20 July in Alexandria generally corresponds to a rising of 19 July at these sites. However, R. Krauss, Sothis- und Monddaten 58, has calculated the likely Julian quadrennia for the heliacal rising of Sothis at the latitude of Memphis assuming an arcus visionis of between 8° and 9°. His model showed a cycle of 18, 18, 18, 19 July at the time of Censorinus, as expected, which lends confidence in his result for the time of the Canopic Decree: 17, 18, 18, 18 July. Hence it is very unlikely that the Canopic observations were made at Memphis. Indeed, the predicted cycle for Alexandria/Canopus in 238 on Krauss' model is 18, 19, 19, 19 July, which is entirely consistent with the Canopic decree. Nevertheless, the possibility that the Canopic date is not based on an observation at the latitude of Alexandria/Canopus cannot be completely excluded; it just does not seem to be necessary to consider it.

The most likely source of the discrepancy, and one which appears to be sufficient to explain it, is slippage of the astronomical Sothic quadrennium against the Julian one between 238 and A.D. 139. The astronomical Sothic cycle is slightly different from the cycle of 1460 Julian years assumed by classical authors and changes over time due to factors such as the precession of the earth about its axis. M. F. Ingham, JEA 55 (1969) 36, calculated that the cycle ending c. AD 138 was about 1452 or 1453 years long. This means that the date of the heliacal rising slipped 7 or 8 days against the Julian calendar over the course of that cycle. If one assumed a constant rate of change, this would be roughly 1 day every 181 or 207 years. Since the period of the astronomical Sothic cycle is itself getting slightly shorter over time, the chances are that the actual intervals between slips of a day between 238 and AD 139 are slightly shorter than these averages.

The actual dates of the slippages will depend on second-order factors which are difficult to determine precisely. However, we can say that the most likely aggregate slippages over the 376 years from 238 to AD 139 are 1, 2 or 3 days, depending on the phase of the slip against this interval, with by far the most probable slip being 2, then 3, then 1, and a slip of 4 or more days being very improbable. Any discrepancy greater than 3 days probably requires additional factors, most likely including a change in the place of observation, but for 3 days or less no additional factors are required.

Ingham's model is not unchallenged. B. E. Schaefer, JHA 31 (2000) 149, based on his own model of atmospheric effects and assuming arcus visionis of around 11°, estimates a net slip of only 0.1 days against the Julian calendar at Memphis between c. 500 BC and A.D. 1 and 2 days between A.D. 1 and A.D. 500, i.e. about 0.6 days between c. 238 B.C. and A.D. 139. Since this is an average figure, and since a slip certainly occurred, the effective slip on Schaefer's model will be 1 day.

Working backwards from AD 139, the candidate Canopic quadrennia in 238 at the latitude of Alexandria and Canopus are therefore:

19, 19, 19, 20 July in AD 136-139
19, 19, 19, 19 July for a slip of 1 day between 241-238 BC and AD 136-139
18, 19, 19, 19 July for a slip of 2 days between 241-238 BC and AD 136-139
18, 18, 19, 19 July for a slip of 3 days between 241-238 BC and AD 136-139

The cycle for a 2 day slip matches the cycle predicted by Krauss' model for Memphis in 238, discussed above.

The question of how these slips would have been reflected in the Canopic calendar has been much discussed: was the Canopic leap-year cycle observationally based or schematic? R. Krauss, Sothis- und Monddaten 54ff, concludes that the calendar was schematic. He presents two arguments for this. The first, based on his analysis of the language of the decree, concludes that the Canopic leap year cycle was the same as that of Censorinus, i.e. 19, 19, 19, 20 July, a cycle which could not have been observationally based at that time. This (in my opinion highly improbable) argument is considered further below. The second, more briefly stated, is that the Canopic decree is the product of Graeco-Roman astronomy, and that all other statements made in classical authors about the Sothic cycle clearly refer to a fixed schematic cycle of 1460 solar years, showing no evidence of understanding that an occasional phase slip would occur. In my view, this argument is rather more persuasive.

In any case, these phase slips are very rare. Nevertheless, the case of a net slip of 3 days between the Canopic and Censorinus sightings is potentially important because the first such slip would most likely have occurred within a very few years of the promulgation of the reform. L. E. Rose, Sun, Moon and Sothis, 188f, suggests that this is exactly what happened, and that the Canopic reform failed for precisely this reason. However, while we cannot exclude a net slip of 3 days, it seems to me rather unlikely that the reform failed for this reason.

• First, granting the scenario for the sake of argument, the first few instances of the slip might easily be put down to observational errors, especially since they would only occur one year in four even if viewing conditions were perfect.
  

• Second, the decree itself contained a provision that recognised that the Sothic rising might not occur on the expected day. The implications of this provision are considered further below.
  
• Third, the evidence of the Macedonian double dates considered below suggests that in fact the reform was not abandoned for a considerable time.

Returning to the phase of the Canopic intercalary cycle with this background, I have found three positions in the recent literature.

R. A. Parker, in Fs Hughes, 177 at 186, following a comment by G. H. Wheeler, JEA 9 (1923) 6, argued that the language of the Canopic Decree ("on the day on which the star of Isis rises ... which now in the ninth year is observed on the first day of the month Pauni") implies that a Sothic rising had already been observed on 1 Payni of the ninth year. Given the date of the Decree (7 Appellaios = 17 Tybi year 9) this would only be possible if the ninth year is the Macedonian year, in which case the decree itself tells us that 1 Payni year 8 (Eg.) = 19 July 239 was also a Sothic rising. Hence, Parker's analysis implies that the Decree gives us that the Canopic Sothic quadrennium was X, X, 19, 19 July, which concurs with the above analysis of the possible quadrennia derivable from Censorinus, but does not refine it. The argument, if correct, does show that a gain of 4 days between 238 BC and AD 139, giving a Canopic quadrennium of 18, 18, 18, 19 July, is not possible, but in any case this is a very improbable number.

A. J. Spalinger, in idem., Three Studies on Egyptian Feasts and Their Chronological Implications, 31 at 40, argues that syntactical anaysis of the language of the Decree actually implies that it was simply stating an accepted fact, similar to the statement that the birthday of the king was on 5 Dios, and that the general usage of calendars in the operational statements of the decree requires the Egyptian calendar to be in use. He notes in particular that the priests to be enrolled in the fifth phyle included "those who are to be assigned until the month Mesore of the ninth year"; the use of the future tense can only apply if the year is the ninth Egyptian year. It would follow that we cannot conclude that the decree is stating that a rising was observed on 19 July 239, although we also cannot exclude it. (L. E. Rose, Sun, Moon and Sothis, 154, also disagrees with Parker's interpretation, though he misidentifies the text on which it was based.)

Spalinger does not conclude that Parker was wrong about the Canopic cycle, only that the Decree itself only allows us to infer X, X, X, 19 July. However, 1 Payni equates to 19 July in both year 8 and year 9 of Ptolemy III (Eg.) = 239 and 238. If Sothis did not rise on 19 July 239, the only possible alternative for year 8 (Eg.) is 18 July 239 (=30 Pachon year 8). In this case, Spalinger's argument would require a gain of 4 days between 238 and AD 139. While this is not absolutely impossible, especially if the Canopic observation was not made in Alexandria, this assumption does not otherwise seem to be necessary. Therefore, even if the language of the decree does not offer irrefutable proof that risings occurred on 19 July 239 (and 238), rather than 18 July, this remains the most likely sequence of events.

L. E. Rose, Sun, Moon and Sothis, 178ff, argues (p181) that the slip between 238 and AD 139 must be precisely three days, on the grounds that a slip of two days (or one) would imply a Sothic rising on 1 Payni = 19 July in 240. He supposes that the Decree shows that the rising on 1 Payni in 239 was the first such rising on that date, which would imply that the previous rising was on 30 Pachon year 7 (Eg.) = 18 July 240.

However, the Decree says no such thing, and Rose presents no argument that this interpretation is required.

R. Krauss, Sothis- und Monddaten 54ff, notes that, shortly before introducing the reform, the decree states "but if, further, it happens that the rising of the star changes to another day in four years, for the festival not to be moved but to be held on the first of Pauni all the same". He connects this to the fact that there was no possibility of a 6 Epagomene between the date of the promulgation of the reform (17 Tybi) and the next 1 Payni thereafter, and infers that the reason for the provision is that it was expected that the heliacal rising would in fact not occur on 1 Payni in 238, but on 2 Payni, and would occur on 1 Payni thereafter because a 6 Epagomene was inserted at the end of that year. If so, the Julian cycle implied by the decree is 19, 19, 19, 20 July -- the same as the cycle of Censorinus! Since, however, this cannot be the astronomical cycle in Alexandria/Canopus or anywhere in Egypt at this time, he concludes that the Canopic cycle is completely schematic, and is not tied to any observed rising of Sothis. He argues that this cycle was a fixed offset from a schematic cycle of either 17, 17, 17, 18 July or 18, 18, 18, 19 July, which was set in Memphis in the early 1st millenium.

There is perhaps some uncertainty as to the translation. Krauss' argument requires that the passage means that a change could occur within the four years of the intercalary cycle. However, A. J. Spalinger, in idem., Three Studies on Egyptian Feasts and Their Chronological Implications, 31 at 36 translates the same passage as "But if it occurs that the heliacal rising of the star changes to another day at the interval of four years" and comments "This is an obvious condition of fact". If so, the whole basis of Krauss' argument would disappear; instead, the passage would simply be motivating the substance of the reform. I cannot comment on the Egyptian or the Greek, but I have to wonder why, if the clause is, as Spalinger says, a statement of fact, it would be constructed as a conditional premise. Further, the actual motivation for the 6th epagomenal day is very clearly spelled out in the following section, that it is intended to ensure that the feasts which are fixed in the Egyptian year remain fixed against the seasons. It seems to me much more reasonable to interpret the clause in question as fixing the calendar date of the Sothic festival even if the heliacal rising is not observed on that date, just as Krauss suggests.

Assuming that the interpetation used by Krauss is correct, therefore, it follows that the Decree allows for the rising not to occur on 1 Payni within a four year cycle, and I agree with him that this is clear evidence that the cycle was intended to be schematic in its phase alignment to Sothis. However, I cannot agree that we must therefore infer that the rising in fact occurred on 2 Payni in 238 -- especially since by Krauss' own admission such a rising was not real! If in fact the cycle is schematic to such an extent that it is even based on virtual heliacal risings, as Krauss supposes, it seems to me we are back to square one: why was it necessary to provide for the possibility of a feast occurring in a year in which a virtual rising occurs on a different date? It would be known with certainty, by design, whether or not the virtual rising would occur.

The notion that the Canopic cycle was based on a Memphite cycle set early in the first millenium BC also strikes me as highly improbable. As Krauss himself points out, the Sothic cycle of 1460 years is only known as a schema of Graeco-Roman astronomy. There would be no reason to fix a quadrennium except within the context of such a schematic astronomy, and there is no evidence whatsoever that the Egyptians were practicing such an astronomy early in the first millenium BC. Rather, we should expect the Canopic cycle to be a schematic cycle grounded in the observational reality of the time the decree was passed, i.e. the third century BC, and almost certainly invented by the Greeks. But, even this possibility does not match Krauss' model: his schematic cycle would not converge with reality for nearly 4 centuries.

Moreove, there is another way to interpret this provision of the decree, which Krauss does not consider, but which is in line with his premise and is not nearly so speculative: not that the rising could be late in 238, but that it might be early. If the actual rising in 239 was on 30 Pachon = 18 July, the statement that "in the ninth year [it] is observed on the first day of the month Pauni" would be predicting a 366-day interval. If the interval turned out to be 365 days, the rising might then be observed again on 30 Pachon = 18 July in 238. So, one could interpret the provision as an indication that the Julian quadrennium was actually 18, 18, 18, 19 July, possibly just after the slip from 18, 18, 18, 18 July to 18, 18, 18, 19 July. However this interpretation is open to the same objection noted above that it implies 4 days slip between 238 and 139 BC.

While my German is poor enough that I may have misconstrued some aspect of Krauss' argument, it seems to me to be more grounded in his belief that Sothic dates were set at a national observatory (now Memphis, previously Elephantine) than in the actual evidence of the Canopic Decree. Although it is not clear why it was felt necessary to provide for the possibility of a Sothic rising that did not fall on 1 Payni, there is no obvious reason why it should necessarily be tied specifically to the Sothic rising of 238, rather than being (as it appears to be) general insurance for a possible event.

The Canopic calendar is grounded in the quadrennial model, which implies that the quadrennial phase shift was clear enough in the observational records on which the decree was based for there to have been confidence in its predictability. Nevertheless, there are many sources of variation that could affect the date of any particular observation in any particular place, of which poor observation is only the most obvious. B. E. Schaefer, JHA 31 (2000) 149 at 151 estimates that variations in the estimated extinction coefficient of one standard deviation alone is enough to cause variations of two days in the observed heliacal rise day. Whether his particular model is right or not, his underlying point, that an isolated observational datum may not be trustworthy, is surely correct. The fact must also have been known to the astronomers who devised the calendar, and this provision is most likely intended to accommodate it.

Unless Censorinus is not reliable, it seems to me that the three quadrennia listed above are the only realistic possibilities, although a 4-day discrepancy might just be possible if the Decree was based on observations made to the south of Alexandria/Canopus. I conclude that the Canopic leap year was almost certainly not at the end of cycle year 4, and was most likely at the end of cycle year 1 or 2. However, no possibility can be completely excluded at this time.

The Longevity of the Canopic Reform

There is no doubt that the Canopic reform failed. The wandering year continued to be in general use, and the Alexandrian reform proves that the Canopic reform had been abandoned before the Roman conquest. This does not mean that it was never implemented. Like the Alexandrian calendar, it may well have coexisted alongside the wandering year for some time. However, since Lepsius first published the Canopic Decree in 1866, the question of how long it was followed by any segment of Ptolemaic society has received very little attention, presumably for apparent lack of evidence.

C. R. Lepsius, Das Dekret von Kanopus, 14, speculated that the feast of the Benefactor Gods to be held on the 6th epagomenal day was probably celebrated during the reign of Ptolemy III, in 238, 234, 230, 266 and 222 (i.e. in Canopic cycle year 1) and abandoned thereafter. This view has recently been endorsed by S. Pfeiffer, Das Dekret von Kanopus (238 v. Chr.) 257. The speculation is completely reasonable and may well be correct, at least for the reign of Ptolemy III, but, as a matter of logic, even if there were evidence for the celebration of the feast it would not actually be relevant to the calendrical issue unless that evidence were explicitly dated. Indeed, as noted above, the Decree contains a provision that anticipates that the heliacal Sothic rising might occur on a day other than 1 Payni, and fixes the date of the festival at 1 Payni (Canopic), so the only proof we could hope for is a date for the start of the festival that was not 1 Payni.

Even a date would need to be handled with care. On the one hand, it would have been perfectly possible for the festival to be celebrated on the wandering calendar even during Ptolemy III's reign, with the date advancing by one day every four years. On the other hand, even if the feast was abandoned by the priesthood after the death of king, that need not necessarily imply abandonment of the 6th epagomenal day, even by the priestood, let alone by the Ptolemaic bureaucracy.

The de facto position of modern Egyptology on this question is summarised by A. E. Samuel, Ptolemaic Chronology 76, who bluntly stated "There is no evidence that this decree ever had any effect on the Egyptian calendar". This view effectively denies that the reform was ever implemented at all, a position which seems rather unlikely given the effort that went into propagating the decree and the fact that other aspects -- notably the creation of a fifth priestly phyle -- certainly were implemented.

The only attempt I have found to address this question directly is a limited analysis by S. Pfeiffer, Das Dekret von Kanopus (238 v. Chr.) 250f. Pfeiffer sought to bound the life of the Canopic reform by looking for occurrences of epagomenal dates in Canopic cycle year 1, which he supposed to be the Canopic leap year, following Lepsius. He found that pTebt 3.2.841 explicitly referred to the 5 epagomenal days in 114, which in his view was a candidate Canopic leap year. He concuded that the reform must have been abandoned by this time.

This approach is inherently flawed, since (a) as noted above we don't know the phase of the Canopic leap year (and Pfeiffer did not choose the most probable phase) (b) the evidence of the Alexandrian calendar suggests that the Canopic and wandering years most likely coexisted, and we have no definitive criteria for deciding when to expect a given date of Egyptian form to be given according to the Canopic year or the wandering year. In any case, since only 1 day out of 1461 is a leap day, the probability of any given surviving document being so dated is vanishingly small. For comparison, the earliest known document dated to 6 Epagomene (Alexandrian) is pOxy 1.45, dated 6 Epagomene year 14 of Domitian = 29 August AD 95 -- 120 years after the introduction of the reform. Similarly, the earliest known document dated by the bissextile day of the Julian calendar is CIL VIII 6979, which records a temple dedication on the day after the bissextile (leap day) in AD 168 -- 212 years after the Julian reform.

Contemporary evidence for the analogous Alexandrian reform suggests two other methods for demonstrating use of the Canopic reform: Dates explicitly distinguishing the Canopic and wandering calendars, or double dates between Egyptian dates according to the Canopic calendar (whether so stated or not) and a second calendar.

No Ptolemaic-era dates have been found that are explicitly identified as being according to either the Canopic or the wandering calendar (or at least there are none recognised in any literature that I have seen). Again, the Alexandrian calendar suggests this is not surprising. The earliest such date recorded in D. Hagedorn & K. A. Worp, ZPE 104 (1994) 243 is BGU 3.957, a horoscope (probably) dated under the ["old"] (i.e. wandering) calendar in 10 BC, and many of the early examples of such formulae are from similar astrological texts, which are not yet in evidence in the early Ptolemaic period.

Aside from astrological texts, the earliest contemporary evidence for the Alexandrian calendar consists of double dates with a second calendar. The earliest explicit double date between the Alexandrian and wandering calendars is SB 1.684, which is dated to 18 Tybi = 1 Mecheir "according to the Egyptians" in year 17 Tiberius = 13 January AD 31, 57 years after the Alexandrian reform. However, much earlier double dates are known with other calendars: the Egyptian lunar calendar in 10 BC (pdem Rhind 1), and the Roman calendar in (most probably) 8-6 BC (pVindob L.1c) and in either 24 BC or, perhaps more likely, AD 3 or AD 7 (SB 18.13849) -- all within 30 years of the reform.

This suggests that evidence for use of the Canopic calendar, if it exists, is most likely to be found in Egyptian/Macedonian double dates after year 9 of Ptolemy III. The double dates of this period have long resisted a clear resolution. The Zenon papyri clearly demonstrated that the official Macedonian calendar under Ptolemy II was a lunar calendar, although the lunar phase of the first day of the month was only determined to within a day or two's precision. Also, provincial use, exemplified by the dockets generated in Zenon's office, shows a tendency towards simplification through a fixed alignment to the Egyptian calendar. The known simplifications are to equate months directly or to place them 10 days out of phase, though even the Zenon papyri contain double dates from Zenon's office that are not lunar but also do not conform to either of these algorithms. For this reason, the best matches should be expected from documents that can be tied to an Alexandrian source, while greater latitude may be allowed from documents from other sources.

T. C. Skeat, JEA 34 (1948) 75, was able to propose an arrangement of the known double dates of Ptolemy III which preserved an approximate lunar alignment, but one that was apparently less precise than that prevailing under Ptolemy II. However, in order to do so he needed to assume that some of these documents were dated according to the financial year. This assumption was strongly attacked by A. E. Samuel, Ptolemaic Chronology 81, whose analysis has not been challenged since, so far as I can determine, although his individual arguments can all be overturned.

Samuel was not able to propose a better model than Skeat's, and indeed concluded "we must admit that the regulation of the Macedonian calendar during the reign of Euergetes was varied and uncertain at best". This conclusion should in itself have been an indication that there was a problem with Samuel's approach. In light of the apparent weakness of his arguments, the only significant objection I see to Skeat's empirical approach is that the lunar alignment appears to be weak after Ptolemy II. But this is exactly what we would expect to see if the Egyptian side of an Egyptian/Macedonian double date was a Canopic date which was misinterpreted as a wandering date -- indeed it should appear to become weaker over time.

So far as I can determine, no one has attempted to analyse these double dates while allowing for the possibility that some of the Egyptian dates may be based on the Canopic calendar. The closest is a paper by M. L. Strack, RhMP 53 (1898) 399, who attempted to analyse the double dates known to him against the wandering year and a Sothic year based on 1 Thoth = 19 July. This approach was immediately, roundly, and rightly rejected by his contemporaries, since there is no reason to suppose that the Sothic calendar exists, and his paper has been completely ignored since. However, his basic insight that a second Egyptian calendar may be involved in addition to the wandering year deserves consideration, since we know that such a calendar did in fact exist. Strack rejected the Canopic calendar as a basis for analysis since the evidence shows that the wandering year was unaffected by the reform. Although he supposed that the Sothic year existed alongside the wandering year, he seems never to have considered that the same might be true for the Canopic year.

The following table contains an analysis against the wandering year and the Canopic for the double dates known to me of Ptolemy III and later which appear to show independent operation of the Macedonian calendar (as always, if you know of others please email me). It assumes that the Macedonian dates are lunar. Consequently, the year number is interpreted according to the C[ivil], F[inancial] or M[acedonian] scheme(s) that give the best lunar match. Where it is not certain whether the ruler involved is Ptolemy III, IV, V or VI, the date is likewise assigned to the king who gives the best lunar match. Each double date is linked to a discussion page giving the issues related to that item.

It will be seen that most, though not all, of those double dates which do not match the wandering year can be resolved by assuming that the Egyptian date was according to the Canopic year. From this table, we may conclude that the Ptolemaic administration probably used the Canopic calendar intermittently until the end of the reign of Ptolemy VI, indeed until the Macedonian calendar was fully assimilated with the Egyptian calendar under Ptolemy VIII, though not consistently. The most interesting feature of the table is that the Canopic calendar and the wandering calendar appear both to have been used by the Ptolemaic administration, even in Alexandria. The significance of this is unclear.

NB: This version is preliminary. Research into individual items is still ongoing at this time.

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