Coligny calendar
The Coligny calendar is a Gaulish peg calendar or parapegma made in Roman Gaul in the 2nd century, giving a five-year cycle of a lunisolar calendar with intercalary months. It is the most important evidence for the reconstruction of an ancient Celtic calendar. It is written in Latin inscriptional capitals and is in the Gaulish language. The restored tablet contains sixteen vertical columns, with 62 months distributed over five years.
It was found in 1897 in France, in Coligny, Ain, along with the head of a bronze statue of a youthful male figure. It is now held at the Gallo-Roman Museum of Lyon-Fourvière. It was engraved on a bronze tablet, preserved in 73 fragments, that was originally wide by tall. Based on the style of lettering and the accompanying objects, it probably dates to the end of the second century.
A similar calendar found nearby at Villards d'Heria is preserved in only eight small fragments. It is now preserved in the Musée d'Archéologie du Jura at Lons-le-Saunier.
Reconstruction
The Continental Celtic calendar as reconstructed from the calendars of Coligny and Villards d'Heria was a lunisolar calendar, attempting to synchronize the solar year and the lunar month.The common lunar year contained 354 or 355 days.
The calendar year began with Samonios.
Le Contel and Verdier argue for a summer solstice start of the year. Monard argues for an autumn equinox start. As Irish "Samhain", Olmsted corresponds to Irish Midwinter. Olmsted argues that Samonios began at actual midwinter when the original 30-year Celtic calendar first originated around 1000 BC, but through a calendar error by one day every 199 years, after 1300 years of operation Samonios was pushd back to November 1 in its celebration.
As with the celebrations of the first day of spring, fall, summer, and winter, not just the days corresponding to the equinoxes and solstices have been displaced earlier by 55 days in Ireland, but the rituals corresponding to those found elsewhere in European and Indo-Iranian tradition, associated with the solar events, have been displaced as well. Such a displacement can be explained by a simple one-day shift every 23.7 years in the Celtic lunar calendar cycle and its associated IVOS festivals with respect to solar time. After some 1300 years of operation, such a calendar would have displaced the lunar festivals to some 55 days earlier than the actual solar events with which they were associated. With the adoption of Christianity the displaced festivals then became fixed points within the newly acquired Julian calendar. But even in this new calendar the festivals were not fixed points in the solar year. Thus after a similar span of 1600 years from its adoption, when Pope Gregory reformed the Julian calendar, it had become 12 days out of sync with solar time.
Besides the Coligny calendar plate, Classical commentary provides a few important details about early Celtic calendar cycles. Diodorus Siculus, quoting Posidonius, describes a sacrifice which took place significantly every five years among the Gauls, presumably at midsummer, since first fruits and animals were burnt in pyres. Plutarch refers to a festival or expedition every 30 years undertaken in the islands off Britain, presumably the Channel Islands, but perhaps the comment on Hesiod refers to islands to the west of Britain and might relate to Ireland. Plutarch's 30-year festive or expeditionary cycle near Britain may relate to the same 30-year cycle as that referred to among the Gauls by Pliny the Elder, writing ca. 75 A.D. but apparently quoting an earlier source.
Thus classical commentary, probably most of it emanating from Posidonius at the end of the second century BC, describes both a 5-year phase, the time interval between the recurring midwinter or midsummer intercalary months, and a 30-year cycle as current among the Gauls and possibly among the British or Irish as well. One very important aspect of Pliny's statement is that both the year and the 30-year cycle begin on the same fixed day of the moon. For the 30-year cycle to begin on the same fixed day of the moon, the 30-year period must contain exactly 371 lunar months or 10956 days, 1 day less than 30 solar years . Thus, as we shall see, Pliny can only be describing the calendar which must have been the predecessor to the 25-year Coligny calendar. On the Coligny calendar the term ATENOVX "returning night" dividing each month into two halves counting from days I to XIIII or XV as well as the N-counting scheme alongside the TII-counting scheme shows that the 25-year Coligny calendar clearly developed from an earlier, fully lunar calendar. The addition of an extra day making the month Equos have 29 rather than 28 days in year 4 of each 5-year phase causes the calendar to shift from a 30-year cycle which contains 10956 days to a 25-year cycle which contains 9130 days.
To total 10956 days, Pliny's 30-year calendar must contain an 1801-day phase followed by five 1831-day phases . In this 30-year cycle, since 5 solar years contain 1826 days, during the first 5-year phase the sun falls back 25 days. However, the sun gains back 5 days in each of the next five 5-year phases . Thus the intercalary summer and winter solstices progress at 5-day intervals every 5 years. In contrast, the 25-year Coligny calendar contains an 1802-day phase followed by four 1832-day phases. Since a 5-year period with four 365-day years and one 366-day year contains 1826 days, the sun would initially lag behind the 25-year calendar by 24 days after the first 5-year phase. However, the sun would gain back this initial lag by 6 days for each of the four subsequent 5-year phases. This calendar realigns to within one day of solar time every 25 rather than every 30 years. Containing 9130 days, this 25-year calendar then runs almost exactly 1 day less than the 9131.06 days to be found in 25 solar years and almost exactly 5 days more than the 9124.95 days to be found in 309 lunar months, equivalent to the 5-day week, the cóicde, of the early Irish law tracts.
Indeed, the Coligny calendar indicates which 5-day lunar week reigns over each subsequent 25-year cycle through a series of marks, which also indicate the one-day lag of the sun with each 25-year cycle. TII indicates that the first lunar week begins each month during the first 25 year of operation. ITI indicates that the second lunar week begins each month during the period between year 26 and year 50. IIT indicates that the third lunar week begins each month during the period between year 51 and year 75 from the inception of the calendar. If the first winter solstice begins on day 1 of year 1, the solstice will begin on day 2 in year 26 and on day 3 in year 51. These marks in this sequence, except where shifted or exchanged with other days, are found in groups of three days in a row, each group separated by the 6 days the sun, in each 5-year phase, gains back the initial fall-back of 24 days from the first 5-year phase. In this 25-year calendar the two summer and winter intercalary solstices which occur every 5 years progress at 6-day intervals rather than the 5-day intervals they progress in the 30-year calendar.
The pertinent observation then is that in either a 25-year or 30-year calendar, assuming four 365-day years and one 366-day year every five years, the sun falls behind the lunar festivals of the calendar by 1 day every 23.70 years, a natural consequence of a 365.20-day year compared to the 365.2422 days contained in an actual solar year. The actual IVOS festivals will then fall 1 day earlier with respect to sun every 23.70 years, progressively getting further and further out of whack with solar time. The 25-year cycle is more accurate to solar time than is the 30-year cycle in that the assumption of a one-day shift every 25 years is much closer to the actual one-day shift every 23.70 years than is the assumption of a one-day shift every 30 years. Both calendars keep track of the assumed solar fallback by 1 day in each 25-year or 30-year cycle.
After 1300 years the actual lunar festivals associated with the solstices and equinoxes will take place on average 54 days ahead of the sun. If the Julian calendar was adopted with Christianity around 450 AD, this would give a date of around 850 BC for the beginning of a 30-year pagan Irish calendar, if such existed. But there is nothing exact in this date because of the natural oscillation of the lunar and solar dates caused by the intercalary months and the 11-day difference between the lunar and solar year. This oscillation gives a wide margin of error of + 300 years. Thus from the evidence of the 55-day displacement of the Irish festivals, we might assume that somewhere in the period 550 BC to 1150 BC the original 30-year Celtic calendar cycle was adopted. However, there is another source for dating the origin of the 30-year calendar. The agreement of these two sources then puts our dating on a more solid footing.
The entry TRINOX SAMO SINDIV "three-nights of Samonios today") on the 17th of Samonios suggests that, like the Irish festival of Samhain, it lasted for three nights. The phrase *trinoxtion Samonii is comparable to a Gaulish festival mentioned in a 1st-century AD Latin inscription from Limoges, France, which mentions a "10-night festival of Grannus" VERG. Olmsted indicates that TRINOUX SAMO is actually an abbreviation for TRINOUX "the Third Night of Samonios", just as IUX RIVROS "the fourth night of Rivros. Both of these days are displaced by the transference of lucky MAT days into unlucky ANMAT months. Their original positions are indicated by the ordinal numerals "TRI- and PETI-.
The solar year was approximated by the insertion of a 13th intercalary month every two and a half years. The additional months were intercalated before Samonios in the first year, and between Cutios and Giamonios in the third year. The name of the first intercalary month is not known with certainty, the text being fragmentary.
In a suggestion first made by Schmidt, the name of the first intercalary month is probably Quimonios, found in the final verse of the gnomic line at the end of the month, OXANTIA POC DEDOR TON IN QVIMON, emended to OXANTIA POC DEDOR TON IN QVIMON "Three hundred eighty and five are given this year through Quimonios".
The name of the second intercalary month is reconstructed as Rantaranos or Bantaranos, based on the reading of the fifth line in the corresponding fragment.
A gnomic verse pertaining to intercalation was taking up the first two lines, read as CIALLOS B SONNO CINGOS. The term sonno cingos is interpreted as "sun's march" = "a year" by Delamarre.
The months were divided into two halves, the beginning of the second half marked with the term atenoux or "renewal". The basic unit of the Celtic calendar was thus the fortnight or half-month, as is also suggested in traces in Celtic folklore. The first half was always 15 days, the second half either 14 or 15 days on alternate months.
Months of 30 days were marked MAT, months of 29 days were marked ANM.
This has been read as "lucky" and "unlucky", respectively, based on comparison with Middle Welsh mad and anfad, but the meaning could here also be merely descriptive, "complete" and "incomplete".
There is no indication of any religious or ritual content.
The Coligny calendar as reconstructed consisted of 16 columns and 4 rows, with two intercalary months given half a column each, resulting in a table of the 62 months of the five-year cycle, as follows :
In spite of its fragmentary state, the calendar can be reconstructed with confidence due to its regular composition.
An exception is the 9th month Equos, which in years 1 and 5 is a month of 30 days but in spite of this still marked ANM.
MacNeill suggested that Equos in years 2 and 4 may have had only 28 days, while Olmsted suggested 28 days in year 2 and 29 days in year 4.
The following table gives the sequence of months in a five-year cycle, with the suggested length of each month according to Mac Neill and Olmsted:
The total of 1831 days is very close to the exact value of 62 × 29.530585 = 1830.90 days, keeping the calendar in relatively good agreement with the synodic month, but the aim of reconciling the lunar cycle with the tropical year is only met with poor accuracy, five tropical years corresponding to 5 × 365.24219052 = 1826.21 days.
As pointed out already by Ricci, based on the mention of a 30-year cycle used by the Celts in Pliny's Naturalis historia, if one intercalary month is dropped every thirty years, the error is reduced to 30 – = 1.27 days in a 30-year period.
This proposed omission of the intercalary month once in 30 years also improves the accuracy of the lunar calendar, assuming 371 lunations in 10,956 days, or an assumed synodic month of = 29.53010 days, resulting in an error of one day in 195 years.
Steinrücken has proposed that Pliny's statement that the Celtic month begins on the sixth day of the month may be taken as evidence for the age of this system: assuming that the month was originally aligned with lunations, a shift of five days corresponds to a period of 975 years, suggesting a starting date in the 10th century BC. Omsted in a similar argument proposes an origin around
"850 ± 300 BC".
In the Coligny calendar, there is a hole in the metal sheet for each day, intended for a peg marking the current date.
The middle of each month is marked atenoux, interpreted as the term for the night of the full moon.
There is an additional marker prinni loudin in 30-day months, at the first day of the first month, the second day of the second 30-day month, and so on. The same system is used for 29-day months, with a marker prinni laget. In Olmsted's interpretation, prinni is translated "path, course", loudin and laget as "increasing" and "decreasing", respectively, in reference to the yearly path of the Sun, prinni loudin in Samonios marking summer solstice and prinni laget in Giamonios marking winter solstice.
Sample month
The following table shows the arrangement of a complete month SAMO. This is the only month out of 62 that has been preserved without any gaps.Each month is divided into two half-months or "fortnights", divided by the word atenoux. Within each half-month, the arrangement is tabular, beginning in the first column with the Roman numeral of the day of the half-month.
In the second column are occasional "trigrams" of the form +II, I+I or II+, and sometimes the letter M, of unknown significance.
In a third column, each day is marked by the letter N or D. In the final column, days are marked with additional information, such as IVOS,
INIS R,
AMB, among others.
In the month Samonios depicted above, the 17th day is marked TRINVXSAMO, corresponding to TRINOSAM SINDIV in Samonios of year 1.
The name of the following month, DVM, is mentioned several times. Conversely, the following month marks days 1, 8, 16 and 17 with SAMON. This "exchanging of days" in odd months with the following, and in even months with the preceding month is also found in other parts of the calendar.
List of months
The names of the twelve months as recorded are 1 samon-,2 dumann-, 3 riuros, 4 anagantio-, 5 ogronn-, 6 cutios, 7 giamoni-, 8 simiuisonna-, 9 equo, 10 elembiu-, 11 edrini- / aedrini-, 12 cantlos.
Seven of these names have no clear etymology. The five month-names that do have language cognates are samon- and giamoni-, which are the stems of the words for "summer" and "winter", respectively; months equos and cantlos might associated with Celtic words for "horse" and "song", respectively; and the name ogronn- is interpreted by Birkhan as a word for "cold".
| month name | days | etymology | interpretation | notes | |
| 1 | Samonios | 30 | "summer's ". | June–July | trinoxtion Samonii on 17th Samonios presumably marks the full moon closest to midsummer. |
| 2 | Dumannios | 29 | tentatively compared with Latin fūmus: "smoke". | July–August | - |
| 3 | Riuros | 30 | tentatively compared with Old Irish remor: "stout, thick, fat", Welsh rhef: "thick, stout, great, large" | August–September | - |
| 4 | Anagantio | 29 | unknown | September–October | - |
| 5 | Ogronnios | 30 | "cold month" | October–November | - |
| 6 | Cutios | 30 | unknown | November–December | - |
| 7 | Giamonios | 29 | "winter's " | December–January | 17th Giamonios, the day opposite trinoxtion Samonii is marked NSDS |
| 8 | Semiuisonna | 30 | unknown | January–February | |
| 9 | Equos | 30, 28, 29 | unknown | February–March | - |
| 10 | Elembiu | 29 | compare to the Celtic word for "deer" and the Attic Έλαφηβολιών "deer-hunting month". | March–April | - |
| 11 | Aedrinios | 30 | compare with Old Irish aed "fire", "heat" | April–May | - |
| 12 | Cantlos | 29 | compared with Welsh cathl, Old Irish cétal "song". | May–June | 15th Cantlos is marked TIOCOBREXT |
The names of the twelve regular months can be reconstructed with some certainty in spite of the fragmentary state of the calendar, as each of them was repeated five times. The two intercalary months occur only once each, and their names are consequently reconstructed with much less certainty.
The name Quimonios is obtained from reading the very end of the first segment as QVIMON, The reconstucton of either *Rantaranos or *Bantaranos is based on reading ANTARAN in the fifth line of the 32nd segment. Olmsted gives a tentative explanation of *Rantaranos as "the count in between".