Remarks on Weekdays in Late Antiquity Occurring in Documentary Sources

This article tries to find an answer to the question, whether weekday indications occurring in documentary texts from late antiquity match modern computations backwards and whether they can be relied upon as helpful, reliable evidence for establishing the precise date of incompletely preserved documents.

NB: Given the wide range of sources to be scrutinized I cannot claim, of course, that these lists are complete. In order to draw a line somewhere, I have omitted from my lists all texts later than A. D. 700, though I am aware of the fact that among these there are quite a few texts (also in Coptic, Old Nubian and Arabic) which offer some interesting disagreement between the various dating elements 6 . Furthermore, the label 'Eastern' is used here only for reasons of convenience in order to make texts written in Greek standing out vs. the evidence written in Latin (the origin of which is restricted to the Western part of the Roman empire). In fact, some of the Greek inscriptions listed below were actually found in the Western part of the Roman empire (Gaul, Italy, Sicily). The majority of our 'Eastern' sources shows agreement, but at the same time it is astonishing that so many texts offer conflicting data. In attempting to explain these conflicts one may suppose that in some cases the commissioner(s) of a grave inscription simply made an error when, e. g., one had to remember on what day a death or a burial had occurred". In other cases a misreading may be involved, e. g. in the case of confusing " There is another death recorded earlier on the same stone (11. 3-5) as having occurred on 13 April = Friday in a fourth indiction. If this indiction is A. D. 540-41, 13 04. in the year 541 would have fallen on a Saturday, i.e. one day off (13. 04. 511 = Wednesday, 13. 04. 526 = Sunday, 13. 04. 556 -Thursday). 12 The credit for the correct interpretation of T\ Ç = rj(uépa) C goes to Wessel. 13 I owe this reference to the kindness of D. Feissel who will publish an improved reading of the text in a future article.
14 For this text see Bull. Epigr. 1989, 939. 15 According to the editor, the numeral for the weekday, a = 4, could possibly be a stonecutter's mistake for A= 1 in his original; if so, there is no conflict between modern computation and ancient indication, as 7. 09. 609 fell on a Sunday; for a similar case cf. below, fn. 16. 16 See ed.'s commentary; an older edition reads the numeral of the day in the month as '19', whereas the present edition reads '15'. In both cases the weekday numeral has been read as an A (= 1). If the date were 19. 10. rather than 15. 10. (see the drawing of the stone), there would be a perfect match on a Sunday; if one sticks to a date to 15. 10., the weekday numeral should be a A (= 4). For a comparable case cf. above, fh. 15. l or rounded E = 5 with 0 = 9 (cf. fnn. the weekday-numeral A = 4 with an A 16).
As to the way the weekdays in these Greek inscriptions were indicated, at first the days bore names: Sunday = rjuépa 'HXiou / Kupiaicf) Monday = fjuspa £eXf|VT|ç Tuesday = f||jépa "Apscoç Wednesday = rjjiepa 'Epuoù Thursday = f||i£pa Aioç Friday = fiuépa 'AtppoSunc / riapacnCEUii Saturday = f||iÉpa Kpóvou. Only by the middle on the 5 th century finds the first instance (in nr. 10) of a numeral being used to indicate the day in the week. only 1 text apparently showing a 2-days difference), but at the same time it would seem hazardous to conclude that the commissioners of the 'Eastern' stones, or the people who executed these, were markedly sloppier than their 'Western' counterparts; after all, 5 texts out of the 26 'Eastern' weekday indications have a clearly Western origin (cf. nrs. 7 and 9 from Gaul, nrs. 6 and 8 from Sicily, nr. 10 from Italy) and the difference between the number of Western and Eastern matches may be nothing more than sheer coincidence 32 .
As to an explanation of the non-matches, the same factors will have been at work as those mentioned above in the analysis of the 'Eastern' evidence. Moreover, the use of Roman numerals in datings like XIIII Kal. Aug. will have been another factor in creating errors by omitting or adding one stroke; such errors were, of course, easily made.
After this survey of the evidence the following remarks may be permitted: a. As there is, both in the 'West' and in the 'East', a sufficient, i.e. more than 50%, level of agreement between the weekdays actually indicated on the stones and those reached at by modern computation, there is no reason to think that the system of weekdays in late antiquity and our own modern computations are completely divergent categories.
b. Given the actual provenance of some of our 'Eastern' inscriptions it does not seem likely that in late antiquity completely divergent systems of weekday counting were in use in the main lands of the Eastern vs. those in the Western part of the Imperium Romanum and that, e.g., a Thursday in some Eastern town like Alexandria in Egypt would have corresponded with a Tuesday in Syracuse on Sicily, vel sim.
As to the value of modern restorations of dating elements on the basis of partly preserved other elements it should be kept in mind that they all rest upon modern computations. But it is begging the question, whether -given the frequently enough occurring discrepancies between computed data and actually indicated data -the 'historical' date of an individual inscription, if its text were completely preserved, matched the modern restoration or differed from it. If, e. g., a Latin inscription from A. D. 343 (by consuls) contains still recognizable parts of a dating like III NON(ae) a n d a weekday like DIE MARTIS, while the month is lost, it is easy to reconstruct the month name on the basis of a modern calculation: it should be IVLIAS (cf. ILCV 4394). But at some future moment the lost part of the inscription might turn up showing the month name as IVNIAS; as III. NON. IVN. ( = 3. 06.) in A.D. 343 would fall on a Friday rather than on a Tuesday, one would suddenly face a discrepancy of 3 days. Likewise, if the complete stone turned out to have in reality MAIAS rather than the restored IVLIAS, III. NON. MAI. (-5. 05.) would fall in A.D. 343 on Thursday, i.e. a discrepancy of 2 days. Still, these discrepancies would be within a familiar range and the only consequence would be that in retrospect the death of some person actually occurred some month(s) earlier (or, for that matter, later) than was assumed before 33 .
Things become, however, more complicated if one wishes to restore in an incompletely preserved text, e. g., some consular name on the basis of a combination of some only partly preserved calendaric data; cf. the case of ILCV 4384, where the dating part has been preserved as [

IA1NVAR. D < I> EIO VIS CONSf VLA TV) FL(A VII) [ L]VN(A) PRIM(A).
The calendaric date has been restored to [VIIKAL. IA]NVAR., the consular name as [BASILI V. C. L]VN(Aj,as according to modem computations '26. 12.463' coincided indeed with both a Thursday and the first day of the lunar cycle. But it remains to be seen, whether this coincidence is enough to warrant the insertion of such an important element like a consular name into the text and to argue that this inscription really is from A. D. 463 and that it may be taken as historical evidence for this year. If a full text of the stone would ever become available, it might well be that, after all, the consulate on the stone was that of, e.g., A.D. 493 (FL. ALBINI V. C.) on [VIIID. IA]NVAR., i.e. on 7. 01; according to our modern computations this is a Thursday indeed; true enough, the numeral of the LVNA shouid be, then, III rather than I, but it is a regular phenomenon to find a conflict between the modern computation of the 'Luna' date and the date actually indicated on the stone (for such 'Lunar' inscriptions cf. the literature cited by G. Alföldy, Eine frühchristliche Inschrift aus Rom**, 461 n. 4).
Likewise, in some cases it has been assumed that the date of a burial should be assigned to a particular year, even if that involved the restoration of a really significant element in a consular formula likep(ost) before cons(ulatum), because otherwise a conflict between the indicated weekday and other dating elements on the stone would arise; cf. ILCV 693 and 4400B. Given the fact, that such conflicts are attested frequently enough, it seems wiser in such cases, too, to leave the texts as they are, rather than to strain an argument, especially if the resulting creation of a postconsulate would create some new problem of its own (cf. CLRE 661 s. a. 398 ad ICUR n. s. I 309 [ = ILCV 4400B] and 665 s. a. 405 ad ICUR I 558 = ILCV 693).
Within the larger framework of documentary texts in general, datings are 'individualistic' elements. If a dating formula is not preserved completely, one must try to restore it as far as reasonably possible on the basis of parallel documents. Sometimes a rather complete restoration of a dating may seem possible and even plausible, but one must always be on one's guard to avoid circular reasoning and, as there are enough instances of inscriptions showing conflicting data, it seems wiser to abstain from the restoration of very specific parts like numerals, names, etc. within a printed text; one may point out to the restoration in the commentary, of course, but that is as far as one may go 35 . discussion (cf. above) of the frequently enough occurring discrepancies between calendaric data and weekday indications it follows that the restoration FEBRU]ARIAS is not necessarily correct. I do not think, however, that in line 5 the single zêta ( = 7), linked by a simple copula Kai to the 10"" (day) in the month of August mentioned just before, can refer to the day in the week. The expected article if) before the numeral Ç is lacking, as is a word like n,uEpa (to be expected on the basis of parallel texts), and it is disturbing, too, that there would be no second copula Kai between the indication of the day in the week and the day in the lunar month. Though it is undoubtedly true that the Greeks used to refer to weekdays with the help of a numeral, all these lacking elements are, taken together, hard to explain. Rather than accepting, therefore, Alföldy's reading of an abbreviated 5EKO(Tat<;) 37    The consequence of this is, of course, that one is dealing with a 17 th day in August which coincides with a 1 9 th day in the lunar month, while there is no longer any day of the week mentioned in this inscription. As A. Ferrua already remarked (cf. fn. 38), this prevents us from establishing a more precise date for this text, as there are far too many instances where 17. 08. coincided with a 1 9 lh day in the lunar month 40 for us to be able to propose any precise date.
wonders what other evidence there is for the supposed date of the pavement; moreover, the 10 th indiction under Justin's reign ran in Palestine from September 576 until September 577 and if the pavement should date from Justin's reign, the burial recorded on it should be dated to 20. 02. 577. At the same time, one would be facing a problem, then, in that this date did not fall upon a Thursday 43 . Of course, one may speculate about an error in any of these (conflicting) data on the pavement, but this seems to be a premature hypothesis, as long as there is some question about the chronographical date of the text under review. Now, the 'burial inscription' referred to for establishing the date of this text, is not the text under review, but a Greek text on another mosaic pavement, also found in Jericho. It is mentioned by M. Avi-Yonah in his list of mosaic pavements in QDAP 2 (1932) 162 nr. 98 (with bibliography on p. 163); this text dates from A.D. 566 (11. 12.), i.e. from under Justin's reign indeed. Though obviously some relationship between both pavements has been supposed by the Ovadiahs, they must be in error about this, as the other pavement was unearthed at a different place in Jericho. It may be attractive to assume that the pavement's text under review dates from the 6" 1 century, but that is only a rather broad dating and one may well ask in which year a date to 20. 02. during a 10 th indiction year in the 6" 1 century (A. D. 502, 517, 532, 547, 562, 577, 592) would correspond with a Thursday. In fact, consulting Grumel's tables (op. cil. [fn. 2] 316-317) one finds that none of these years offers the requested correspondence 44 . The conclusion must be, then, that with the data being taken at face value the text does not date from the 6" 1 century A. D. But if the text would date from the 7" 1 century, a satisfactory correspondence can be found; if the chronological data on the pavement are taken at their face value, the date of the burial fell on 20. 02. 637 or 682. For the moment I see no way of expressing a preference for any of these two dates and, of course, it is another question, whether the pavement's archaeological context matches with such a late date; unfortunately, I cannot answer to that. If it would not match, one should accept the situation that here, like frequently enough elsewhere, there is a conflict among the chronological data on the pavement itself and that no exact date for this text can be proposed.