cord with the principle of biennial intercalation. We must of course bear in mind that the Macedonian year may have begun some months earlier than is supposed in the following table. But if the known concordances are quite irreconcilable with our scheme, I feel sure that we should get no better results by shifting the starting-point of the year to another month. In other words, the theory that a month was intercalated every second year would have to be rejected.

The reign of Philadelphos ended on the 25th Dios of his 39th canonical year. For the probable equivalent to this date in the Egyptian calendar I must refer to the table of concordances given in the Annales, XVII, p. 223. On the hypothesis that the first day of the Macedonian year was the anniversary of the accession, the 25th Dios of the 39th canonical year would be the 17th of Choiak. But on the hypothesis, which we have seen good reason for adopting, that in the reign of Philadelphos the Macedonian year was in advance of the canonical it would correspond with the 6th of Choiak. Euergetes ascended the throne on the 25th of Dios and his birthday was the 5th of the same month, so that his first Macedonian year according to our scheme would be nearly a complete year. It would be several months behind the canonical year and seven more months behind the financial year.

To test the theory of biennial intercalation I go back to the double dates of Apollonios for the years 29-32 and take these as the basis of the scheme. And as we are supposing the new year of Euergetes to have begun in Dios, I place the intercalated month between Hyperberetaios and Dios and give it a length of 30 days as in the former table. The new table shows the supposed relation between the two calendars during the first two years and the other years which interest us, the intervening ones being omitted. Nor do I consider the cases in which the papyri give only a general concordance between an Egyptian and a Macedonian month (see P. Hib., pp. 343); it is sufficient to say that in only one case is there

342,

any immediate difficulty in reconciling these concordances with those of the table.

The years for which we have definite double dates are 8, 9, 16, 21 or 22, and 25. Of these the most important is the double date of the Kanopos decree, year , Apellaios 7 Tybi 17, as there is little likelihood of error in a monumental decree of this sort. Whether the date in the protocol refers to the Macedonian year by which the Court reckoned or to the canonical year by which the priests themselves reckoned is a question which we do not need to discuss here; for in the month of Tybi these two years were equal, not only according to our table but according to any probable scheme. The difference between the equation in the inscription and that of the table amounts to two days only. Recalling the fact that in the years in which we have definite evidence of biennial intercalation we find small differences of one or two days between the dates of Apollonios himself, I do not hesitate to claim the date in the decree as a remarkable confirmation of our theory.

Turn now to year 25, for which we have the equations Apellaios 11=Pharmouthi 6 (P. Petr., III, 28 b) and Loios 26 = Choiak 13 (P. Magd., passim). The differences here amount to 9 and 10 days respectively if we take year 25 to be a Macedonian year according to our scheme. But it has been already suggested that officials in the nomes were much inclined to follow the custom of their compatriots and to date by the financial year; or, to put the point in a more general form, I have little doubt that the dockets in the above-mentioned papyri are dated by a year that was several months in advance of the canonical year. In that case the differences between our concordances and those of the papyri are reduced to 2 days and 1 day respectively.

For the double date of year 21, or 22, Smyly's commentary on P. Gurob 2 must be consulted. In the two copies of the document the date is variously given as year 22 (?), Dystros 16-Payni 19 and year 21, Dystros 16

=

19. The protocol again is dated

-Pachon while year 22, the trial seems to have taken place in year 21. The most probable explanation of these inconsistencies is, I think, that the official date was year 21, but that the scribe was more accustomed to reckon by the financial year and sometimes wrote 22 inadvertently in place of 21; Pachon is a slip for Payni; and the correct date is year 21 (=financial year 22), Dystros 16 Payni 19. Our table differs from this by four days.

=

An unpublished Tebtunis papyrus (P. Hib., p. 342) gives the equation year 8, Gorpiaios 2 - Phaophi 7. If this is a financial date, it differs from our table by 10 days; if a Macedonian date, by 9 days.

For the 16th year P. Petr., III, 53 (s) gives the equation Gorpiaios 4-Choiak 11. The difference here is considerable, amounting to at least 21 and perhaps to 40 days (1).

What can we conclude from the above evidence? Evy' ἐμὸς Ερμᾶς, says Kallimachos, εὖχ ̓ ἐμός· οὐ παρὰ τὰς εἴκοσι μεμφόμεθα. Like the poet I have no fault to find except for those twenty days in P. Petr., III, 53 (s). The general result is this. From year 22 of Philadelphos to the end of the reign of Euergetes we find that the Macedonian year was on an average four days longer than the Egyptian. As therefore we know that from year 27 to year 33 of Philadelphos a month was intercalated every odd year in the Macedonian calendar, making the average length of the year 369 days, and as we find that the average length of the year for the ensuing period up to the death of Euergetes was exactly the same as this, it seems a

(1) It may occur to the reader, as it has to me, that 16 might be a slip of the scribe for 26, in which case the concordance would suit our table exactly. But it is doubtful if Euergetes was still alive in Choiak of his 26th financial year. According to the canon he died in his 26 canonical year and according to the evidence of the papyri in his 26th financial year (P. Petr., III, 119, verso, col. II, 9; see also Smyly in Hermathena, 1906, p. 115). As it is certain that the financial year was in advance of the canonical in his reign, we are obliged to conclude that he died in the winter of 222-221 B. C.

highly probable inference that the same system of intercalation had remained in force. We have seen again that four of the six double dates mentioned above agree more or less closely with the concordances in our table. In the other cases there are differences of ten and twenty days. But as we know from Zenon's dates that officials in the provinces found it difficult to keep count in both calendars and were apt to get their equations wrong, it seems to me much more likely that these differences are simply errors than that the system of intercalation had been changed.

It may be objected that our concordances are based on a quite unproved hypothesis about the starting-point of the Macedonian year and would work out differently if we shifted the starting-point to another month. That is true to some extent. I do not regard the said hypothesis as more than a possibility. But if we draw up a similar table on any other hypothesis that is at all probable and which does not conflict with such evidence as we possess, I think it will be found that the differences between such a table and the double dates are on the whole much the same as they are in the above scheme.

My conclusion is that the double dates from the reign of Euergetes are sufficiently and most naturally explained by the theory of biennial intercalation. But I do not pretend to understand the relations between the two calendars during the following period.

H

RECUEIL CHAMPOLLION.

See Scheme, pages 130-131.

9