Sie sind auf Seite 1von 5

175

THE. IMl'ERIAL NUNDINAL CYCLE.

According to Macrobius, Saturnalia 1.16.34, the Roman market, or nundinae, was held every eighth
day. The stability of this so-called nundinal cycle in Republican times is strikingly demonstrated by
Dio 48.33.4, which recounts how an extra day was inserted into 41 B.C. (R) in order to avoid the ill
omen of a market day falling on l January 40 B.C. (R): I it was easier to adjust the calendar than it was
to delay tbe market by a day. For this reason, data about the phase of the nundinal cycle in a given year
is an important tool for establishing the exact chronology of the final decades of the Roman Republic,
relative to the introduction of the Julian catendar on I January 45 B.C. (R).2
There is, however, only one Roman market day whose Julian date can be determined from
contemporary evidence." The Pornpeian graffito CIL IV 4182 states that "Sunday" 6 February A.D. 60,
day 16 of a lunar month, was a market day in Cumae and the fourth day before the Pompeian market.
Another Pompeian graffito, GIL IV 8863, gives a cycle of market days which shows that the Roman
market was two days after that at Cumae. Hence 8 February A.D. 60 was a market day in Rome.
If we could safely assume that the nundinal cycle was stable from that date back to Republican
times, this result would suffice to fix the phase of the market day under the Republic. However, Dio
60.24.7 notes that the Roman market day was moved to another day in A.D. 44 "because of some
religious rites" and that the market day had already been changed several times. This DOle proposes an
identification for these rites, and discusses the nature of the interruptions to the cycle that they caused.
Raepsaet-Charlier bas suggested that these interruptions were somehow compensated for soon
afterwards, and so did not permanently affect the phase of the nundinal cycle.' She did not suggest any
method of compensation, and the notion is inherently unlikely. Since a market was held in a different
town on each day in.ute cycle.> a change to the market day in Rome would either require the Roman
market to be swapped with that of another town or the whole cycle to be delayed by a day or more.
Since there were several circuits of market tOWIlSwhich intersected at the Roman market, the second
alternative seems to be less disruptive. Either way, compensating for such an event would have been at
least as disruptive as introducing it, so we may safely assume that any interruption created a permanent
phase shift in the market cycle.
It is well known that each day in the Roman year was assigned a "nundinal letter" in an 8-letter
cycle A-H, repeated throughout the year.6 This provided a simple visual almanac for determining the
dates of future market days in any given year, since they would fall on days with the same letter.
Because 31 December (R) (Julian) has the nundinalletter E, while I January (R) bas the nundinal letter
A, a discontinuity was introduced at the end of the year that resulted in a phase shift of three letters in
the nundinal cycle. Hence, to find the nundinal letter for the market day in the next year one had to
count forward three leuers in the cycle (or backwards five) from the letter in the current year.
In the many surviving imperial fasti, no day is marked for omission from the cycle by the absence
of a nundinalletter. However, the fasti do not show how the market cycle was affected by intercalation.

J On the assignment of this event to 4 L/40 B.C. rather than 40139 B,C.. see P. Brind' AJllOUr, l:e catendrier remain:
recherches chronologiques (Ottawa, 1983) 81-82. Roman civil dates are distinguished from Julian dates in this paper by the
annotation (R).
2 E.g. Briod' Amour, Le calendrier romain, 45·47.
3 A. Deman & M.-T. Racpsact-Charlicr, "Notes dechronologie romaine", Historia 23 (1974) 271-296.
4 M.-T. Raepsaer-Charlier, review of P. Brind' Amour, U caiendrier romain. Phoenix 39 (1985) 292-294.
5 Deman & Raepsaet-Charlier. Historia 23, 272;Brind'Amour, Le calendrier TOmah4271-275.
6 A. K. Michels, The Cotendarof the RomanRepublic(Princeton, 1967) 27-28.
176 Chr. Bennett

Two possibilities suggest themselves: that the bissextile day was included in the cycle; or tbat the
bissextile day was omitted from the cycle.'
Including tbe bissextile day would leave the length of the cycle that covered it unchanged, at 8 days,
but would cause the nundinal letter of the market day to move backwards by one thereafter for the
remainder of tbe year. This possibility is supported by Dio 48.33.4, which requires exactly this
behaviour for an intercalation in 41 B.C. to cause the market day to move from 1 January 40 B.C. (R) to
3J December 41.B.C. (R).
Omitting the bissextile day would increase the length of the particular cycle that covered it from 8
days to 9. This omission would arise naturally if the bissextile day and 24 February (R) were treated
calendrically as a single 48-hour day with two parts, as described by the jurist Cclsusf While the phase
of the cycle would slip by one day each leap year as a result, the nundinalletter of the market day in the
leap year would be unchanged. In view of Dio 48.33.4, Tibiletti suggested that this rule took effect as
part of the Augustan reform of 8 8.C.9 He did not, however, seek to validate the hypothesis.
Brind' Amour sought to test it by inferring a second imperial nundinal date from CIL IV 8863.tO
This graffito has the foUowingstructure:
DIES NUNDINAE
SAT POMPEIS X[VIIII) VITI NON XV XXVIIII
SOL NUCERlA X[VIfI] Vll VllU If [XlVI XXX
LUN ATILLA XV[ITj VI VUI ID XVJ.I
MAR NOLA XV[I] [V] vn IV XVIlI
MER CUMIS XV [I]V VI V XVlUI
JOV PUTIOUS XIV [1)11 V VI XX
VEN ROMA XIII PRJ IV VlI XXI
CAPUA XII K III VITI XXII
XI NON PRI vnu xxm
X VII IDUS X XXIV
VIIIl VI XI XXV
XIJ XXVI
xm XXVII
XIV XXVITI
The graffito evidently gives (a) the days of the sabbatical week, starting witb Saturday (b) a market
circuit of 8 different towns, starting with Pompeii (c) a countdown from a day X[Vllll1to the start of
the next Roman month, followed by an apparently somewhat.confused count of days in the next.month,
up to the Ides, and (d) a count frOIDI to XXX. which may reasonably be understood as a lunar month.
Brind' Amour supposed that CIL IV 8863 was a ready reckoner for a specific lunar month in which
day I, a Pompeian market day, fell on a Saturday, the 14th day of a 31 day month with an Ides on the
13th. Since day I of the lunar month in ClL IV 4182, 22 January A.D. 60, was a new moon, he argued
that the lunar month of CIL IV 8863 was also based on lunar conjunction. He supposed that the Roman
month in question was either August or December. There is one date before the Vesuvian eruption of
A.D. 79 that meets all these conditions: Saturday 14 December A.D. 76. From CIL IV 4182, we know
that 10 February A.D. 60 was a market day in Pompeii. Since the distance between these two dates is
6,152 = 8"769 days, he concluded that the nundinal cycle was stable between A.D. 60 and A.D. 76,that

7 G. Tibiletti, "Qualche Problema Nundinario". RSA 611 (197611)27·34.


8 Cited in Justinian. Digest 50.16.98.
9 Tibitelli. RSA 617,29.
10Brind' Amour.l-A: calendrier romain, 268-271.
The Imperial Nundinal Cycle 177

the bissextile day was included in the cycle, and therefore that Tibiletti's conjecture was incorrect. It
would also follow that the reasons for the nundinal shift reported by Dio were probably arbitrary and
unrecoverable. I I
However, this analysis is open to two objections, even if we accept the premise that the graffito is a
ready reckoner. First, January is also a 31·day month with ao Ides on the 13th. Second, ClL IV 4182
says that 6 February A.D. 60 was a Sunday, but by modern reckoning it was a Wednesday.
Brind' Amour himself provided a plausible explanation for this discrepancy, but the same argument
implies that the "Saturday" of CIL IV 8863 should correspond to a modem Tuesday. 12 If we widen the
search for new moons Oil the 14th day of a 31·day month with an Ides on the 13th in the decades before
thc eruption of Vesuvius to include Januarys, and if wc look for Tuesdays rather than Saturdays, the
only solution is Tuesday 14 January A.D. 66. But the distance between this date and 10 February A.D.
60 is 2,165 = 270*8+5 days. This date would mean that the phase of the market cycle slipped by five
days against the Julian year in only six years.
In any case, it is very doubtful that CIL IV 8863 is a ready reckoner for a specific month.
The search of a synchronism initially depended on the assumption that tbe "NON" following the
Kalends is meant to represent the name of a specific month. Della Cone supposed that it was a badly
written NOV, while Degrassi supposed it was a very badly written lAN.13 But NOV is impossible,
since the Ides of October fell on the 15th of that month, while IAN can also be ruled out since the
Kalends of January was followed by a.d. IV Non. Ian., which cannot be reconciled with the VII that
follows the NON in the graffito. There is no way to reconcile "NON" with either September or
February. Thus the premise that "NON" represents a specific month is rather unlikely.
Further, if the author of the graffito was trying to construct a ready reckoner, he was apparently not
very familiar with the Roman calendar. There is no month where the day following the Kalends was
a.d. VII Non., nor is there any month in which the Nones is followed by a.d. vun
Id, However, the
Kalends itself is a.d. vn Non. in March, May, July and October, while the Nones is always a.d. Vllii Id,
This might suggest that we are actually looking for a 31·day month with an Ides on the 13th that is
followed by a month with an Ides on the 15th. But no such month exists.
Finally, if the graffito is a ready reckoner for a specific month, it is a remarkable coincidence that aU
four cycles listed apparently start on the first day of the cycle: the first day of the week, the first town in
tne market cycle (from a Pompeian perspective), the first day in which a new month name is used, and
the first day of a lunar month. It may also be noted that the lunar month is arranged in two columns of
14 days followed by the two extra days, a structure Which clearly shows how to relate it to the week.
In light of all this, I suggest that ClL IV 8863 is not a ready reckoner but a universal monthly
calendar, a compressed version of the more familiar annual fasti. The "NON" in the fourth column is
exactly what it appears to be: the Nones. The middle three columns are not intended to list the days of a
specific month but to show the general structure of the Roman month, just as the other columns give a
generic sabbatical, nundinal and lunar cycle. Thus, tile rust day in which a month name was used could
appear as early as the 19th day before the Kalends, the Kalends was followed by a similar countdown to
the Nones, in which the Kalcnds could be as early as the 7th day before the Nones, and finally the
Nones itself was always the 9th day before the Ides.
If this explanation is correct, then CIL IV 8863 does not imply any synchronism between the four
cycles and so cannot be used to recover a specific market date.

IJ The table of.imperial nundinae at Bond' Amour. Le catendrier romain, 332':133 supposes a single interruption in
.1\.0.44. This model must be regarded as schematic, since Die 60.24.7 is clear that several interruptions had occurred before
A.D. 44. and there is no reason to suppose thal that one-was the Ja.~t.
12Bri.•d' Amour, Lecalendrier romain, 268-269.
13Brind' Amour, i.e calendrier romain, 270.
178 Chr. Bennett

The nearest known Roman market date is that provided by Dio 48.33.4. However. tbis date cannot
be used directly to test the hypothesis. Opinions differ on whether the intercalation Dio describes had a
permanent effect on the intercalary cycle. I. or was a minor perturbation that was corrected without
fundamentally altering the cycle. IS In the former case, the market day in question actually fell on 31
December 41 B.C. (R); in the latter, the intercalation can be ignored. so that the market day effectively
fell on I January 40 B.C. (R). Also. there are several different models for the early Julian calendar.
Each one predicts a specific Julian date for the market day. However, having selected a model, the
number of days between the predicted Julian date and 8 February A.D. 60 can be calculated. The
remainder, after division by 8. represents the actual net phase shift in the nundinal cycle between 40
B.C. and A.D. 60 according to the model selected.
If the bissextile day was combined with 24 February (R) for nundinal accounting in the Augustan
reform of 8 B.C.. the nundinal cycle would slip by one day each leap year after the reform. and the net
phase shift in the nundinal cycle would be tile remainder. after division by 8. of the number of leap
years between the Augustan reform and A.D. 60. Hence. ifTibilerti is correct, the phase shift given by
the number of leap years after the Augustan reform should equal the phase shift determined from the
selected model of tbe early Julian calendar.
Three of the existing models for the early Julian calendar provide the required match:
I. Leap years in 39, 36 .... 12,9 B.C., A.D. 8, 12 .... 16
Market day of Dio 48.33.4: 1 January 40 B.C. (R) = I January 40 B.C.
=
Distance to 8 February A.D. 60 36.197 days. Net nundinal phase slip 5 days =
=
Number of post-reform leap days 13. Predicted phase slip 5 days =
2. Leap years in 38, 35 .... 14, II B.c., A.D. 4. 8, 12 .. .'7
Market day of Dio 48.33.4: 3J December 41 B.C. (R) = 31 December 41 B.C.
Distance to 8 February A.D. 60 = 36,198 days. Net nundinal phase slip = 6 days
Number of post-reform leap days = 14. Predicted phase slip = 6 days
3. Leap years in 38, 35 .... 11,8 B.C.. A.D. 4. 8, 12 ... IS
Market day of Dio 48.33.4: 31 December 41 B.C. (R) = 30 December 41 B.C.
Distance to 8 February A.D. 60 = 36.199 days. Net nundinal phase slip = 7 days
Number of post-reform leap days = IS. Predicted phase slip = 7 days
These results support Tibiletti's hypothesis, although they do not uniquely distinguish the correct
model for the early Julian calendar, which must be decided on other grounds.
Even if the bissextile day was omined from the nundinal cycle, however. this is not sufficient to
explain Dio's statement. which is clear that the market day was actually moved to avoid a conflict with
religious rites. l.n order to understand this, we need to examine how the nundinal cycle operates once
the bissextile day is excluded from it.
The net effect is that tbe nundinal letters of the market days followed the sequence B, E. H. C. F. A,
D, G, repealing every 8 years. Since A.D. 60 was a leap year and the nundinal letter of 8 February (R)
is G, the nundinalletter of the market day in leap years alternated between C and G. But G is also the

14H. Matzar. RiJm;s(.·he Chronologie I (Berlin. 1883) 15. Die's "compensation" was 10 omit the bisscxrile day
scheduled for 40 B.C.(R) 011 Caesar's calendar; in effecr, the bissexnle day was shlfred by a year.
15Brind' Amour, Le calendrier romain, 45. This requires that 40 B.C.(R) was somehow shortened to 364 days. As far
as I know, no method of doing this has ever been suggested by proponents of this view.
16Brind' Amour, Le catendrier remain, 11-15.45,
11Matzat, Rdmische Chronologie J Il~ 18.
IBe. J. Bennett. "Tile Early Auguslan Calendars in Rome and EgYPI".ZP£ t42 (2003) 221-240; idem. above pp. 165
-168. The model predicts thai 8 B.C. was itself a leap year, but this does not imply Ihal the Augustan reform was
promulgated after the bisscxtile day. The match implies 1M' the reform was actually instituted shortly before it. in time for
that day to be omiued from the nundinal cycle (which was, after all, a local Roman matter} but too late to cancel it.
throughout the empire.
The Imperial Nundinai Cycle 179

nundinal letter of the bissextile biduum. Therefore, in the absence of other considerations, the market
day would fallon the first day of the biduum every alternate leap year, including both A.D. 44 and A.D.
60, and the second day would be skipped. However, the first day of the biduum was also the festival of
the Regifugium, which Ovid regarded as the first of the sacred year. 19
This observation suggests that Die's "religious rites" was the Regifugium. It also suggests how the
market day was actually shifted and why it. was done. There is a well-documented superstition against
having the market day fall 00 the first day of the civil year in the late Republ.ic.W Dio 48.33.4 describes
a deliberate attempt to avoid this in the Julian calendar through the intercalation of 41 B.C. (R). As a
result of this action. market days on I January (R) became impossible on the triennial leap year cycle,
since this cycle is an exact multiple of 8 days.21 However, once Augustus restored tbe correct quadren-
nial leap year cycle, it again became possible for a market day to fall on I January (R); indeed, it was
unavoidable.
It appears that the problem was solved by transferring the prohibition from the first day of the civil
year to the first day of the sacred year. We can simply explain Die's statement by supposing that the
market day was moved from the first to the second day of the biduum, thc actual bissexrile day. in every
alternate leap year, in order to avoid' a conflict with the rites of the Regifugium. In this way. the biduum
became a day of ill-omen for public affairs, a belief which is attested as late as A.D. 364.22

Sao Diego Chris Bennett

19Ovid, Fasti 2.49-50.


20Dio 40.47.1, 48.33.4: Macrobius. Saturnalia 1.13.17.
21The triennial period is 365+365+366= 1096days: 1096= 8*137. Hence. havingchanged the date of the market day
to 3 1 December 41 (R). the market day will recur on 31 December (R) everythird year.
22 Ammianus Marcellinus 26.1.7

Das könnte Ihnen auch gefallen