Coligny calendar

The Coligny calendar is a bronze plaque with an inscribed calendar, made in Roman Gaul in the 2nd century AD. It lays out a five-year cycle of a lunisolar calendar, each year with twelve lunar months. An intercalary month is inserted before each 2.5 years. This intercalary reiterates the days passed through in the preceding 30 months, with a day from each month passed through. It is the most important evidence for the reconstruction of an ancient Celtic calendar.

Overview of the re-assembled tablet found in Coligny, France

The god found with the Coligny calendar reconstituted by A. André.

Approximately 40% of the original calendar remains in the form of reassemble fragments. The calendar was found in 1897 in France, in Coligny, Ain (46°23′N 5°21′E, near Lyon), along with broken pieces of a bronze statue of a life-size naked male holding a spear, likely Roman Mars. It was engraved on a bronze tablet, preserved in 73 fragments, that was originally 134.8 cm wide by 78.0 cm high. With the rim attached the plate measured 52 by 32 unicae Drusianus (2.75 cm to the unicae). It is written in Latin inscriptional capitals and numerals, but terms are in the Gaulish language. Based on the style of lettering and the accompanying statue, the bronze plaque probably dates to the end of the second century, although the copying errors indicate the calendar itself is much older. It is now held at the Gallo-Roman Museum of Lyon-Fourvière.

Eight small fragments of a similar calendar were found at the double-shrine of Villards-d'Héria. It does not have the holes of a peg calendar that the Coligny calendar does, but otherwise has the same notations. It is now held in the Musée d'Archéologie du Jura at Lons-le-Saunier.

The Importance of an Accurate Reconstruction

Reconstructed Coligny Calendar

Determining an accurate reconstruction of the original Coligny calendar is of the greatest importance. Recent scholars have attempted to generate explanations based on suppositions about how the calendar should have operated, rather than what is actually indicated by calendar’s notation. All calendar systems are created from the shuffling of lunar months and solar days, arranging 28-to-31-day months into 360-to-390-day years in an attempt bring order into the complexity created by their being 354.367 days in 12 lunar months and 365.242 days in a solar year (the year created by the sun cycling through 12 Zodiac constellations). Many of these recent explanations ignore not only the system Olmsted had reconstructed 40 years ago, but also the earlier pioneering work of Thurneysen (1899), MacNeill (1928), and Duval and Pinault (1986), each of whom contributed toward advancing an understanding of the calendar’s abbreviated terminology.

Beside the determination of the patterns inherent in much of the calendar’s notation, the pioneering calendar researchers were avid scholars of Celtic languages. Through combining their linguistic skills with a determination of the function in the calendar of the abbreviated archaic Gaulish terminology, they were able expand many of these truncated terms to their original fuller forms. In any given term, letters were left off arbitrarily from the middle as well as the end of words in a dialect which appears to be archaic Gaulish, half the time corrected to standard Gaulish and half the time left unchanged in the original language of culture of the calendar priests. Olmsted's major contribution, which built upon their earlier work, was to determine the pattern and function of the N-count and TII notations marking out the sonno cingos “the pathway of the sun” in its travel through lunar calendar. These intercalary solstice counting schemes, alongside the notation of the intercalary months determine how the calendar functioned.

The most blatant of the recent erroneous reconstructions have been suggested by individuals with little training in ancient or even modern Celtic languages and little knowledge of basic astronomy or number theory. Some of the recent research ignores the clear notation engraved on the surviving pieces of the broken-up calendar plate, completely transforming the calendar’s operation and leading to a system entirely different from that which the calendar actually preserves.

An examination of the actual surviving notation demonstrates the following about the Intercalary Months. Intercalary-Month-1 QVIMON had 30 days, not the 29 days needed for the Coligny calendar to follow the Metonic cycle. Like Intercalary-Month-2 RANTARAN, Intercalary-Month-1 QVIMON supplies a day for each of the months passed through in each of the preceding 2 ½ years. Since 30 months were passed through in the preceding 2 ½ years, Intercalary-Month-1 QVIMON should contain 30 days. Intercalary-Month-1 QVIMON is stated to be a MATV month. All of the other months labeled MATV have 30 days. The written-out numeral OX[..]ANTIA POG, though somewhat unusual, can be seen as an abbreviation for an archaic Celtic form of 385 and not at all as 384. The term TIOCOB[RIXT] found immediately after the last day of Intercalary-Month-1 QVIMON also indicates CANTLOS day 15 was used in place of non-existent ATENOVX day 15 to give QVIMON its 30th day. The remaining portion of the broken-off second digit of the Roman numeral for the last day has a slant, indicating that it is a V (for XV) rather than an I (for XI[III]). Since days 8 through days 30 of Intercalary-Month-2 RANTARAN survive, it also had 30 days. Without 29 days in one of the intercalary months the Metonic argument falls apart.

Determining the actual means by which the Coligny calendar priests kept track of the position of the sun with relationship to that of the moon is a completely different task from dreaming up new schemes at the caprice of the writer or attempting to impose upon the Celtic calendar the systems of the ancient Greeks, Romans, Egyptians, and even those of the Maya Indians.

One can always make a supposedly simple alteration to the Gaulish calendar described by Plinius which will transform a 30-year calendar-cycle, accurate to lunar time to within 1 day every 199 years, into a Metonic 19-year calendar cycle, accurate to lunar time to within 1 day every 210 years. Having nearly the same accuracy (within 5%), the major difference between the two systems is that solar time and lunar time roughly converge every 19 years in the Metonic cycle, and solar time falls back by one day every 23.8 years in the Celtic calendar described by Plinius. However, one must utilize the actual notation engraved on the Coligny calendar plate to determine its system of operation, rather than make suggested additions or changes to the existing notation to enable one to transform it into the new system one finds preferable. The addition or deletion of a small number of days can transform Plinius’s 30-year cycle into a 19-year Metonic cycle or into the 25-year cycle actually preserved on the Coligny plates by hundreds of surviving TII marks.

Embedded within the notation of the final 25-year-cycle of the Coligny calendar indicated by the solstice counting schemes is an earlier 30-year-cycle calendar. The more accurate later 25-year-cycle lunar calendar was developed simply by giving EQVOS 29 rather than 28 days in year-4. Although the last eight days of this month in year-4 are completely missing from the fragments of the calendar plate, the six-days separation in the occurrence of the triple cluster of these marks (TII, ITI, IIT) demonstrates that the calendar followed a 25-year cycle in which the midwinter intercalary solstice dates where separated by six days rather than the five days they would have been separated had they followed a 30-year cycle. The 30-year cycle can only occur by giving EQVOS in Year-4 the 28 days it had in the earlier calendar described by Plinius. Since the TII marks demonstrate that six days separate the solstice dates occurring in the midwinter intercalary month (occurring every 5 years), 5 times six-days separation equals 30 days, and the calendar recycles after 5 such five-year phases. The separation of 5 years in the intercalary midwinter solstice dates of the 30-year cycle would need 6 such five-year phases since 6 times five-days separation adds up to 30 days.

The notation on the Coligny plate indicates that at its initiation an original constant-lunar 30-year-cycle calendar system had each month begin on the first day of the new moon. The inherent error of approximately one day every 200 years, after 1000 years each month would begin on the sixth day. The new 25-year calendar was created to take into account this 5-day shift. The new 25-year calendar shifts to a new 5-day-week phase of the moon at beginning of each new 25-year cycle. The symbol TII indicates the first 5-day-week phase of the moon, ITI the second five-day-week phase beginning after 25 years of operation, and IIT the third five-day-week phase beginning after 50 years of operation and two 25-year cycles through the calendar. The clusters of these three different marks occur on three sequential days, a result of the sun’s falling back relative to the calendar by 1 day in each 25-year cycle.

As noted above and repeated here for closer analysis, Plinius states (Naturalis Historia, XVI: 250) that the months and years of the 30-year-cycle Gaulish calendar began on the sixth day of the moon.

Est autem id rarum admodum inventu et repertum magna religione petitur et ante omnia sexta luna, quae principia mensum annorumque his facit, et saeculi post tricesimum annum, quia iam virium abunde habeat nec sit sui dimidia (Zwicker 1934: 55).

Plinius’s record goes further, however, than simply indicating that the months began on the 6th day of the new moon. Plinius indicates, moreover, that the beginning of each 30-year cycle also commenced on the 6th day of the new moon. We should recall that the function of the 30-year cycle is to allow the lunar and solar phases to come back into alignment. Over this 30-year term the lunar and solar phases must come within a few days of each other for the calendar to recycle the alignments.

There are 29.5306 days in a lunar month and 365.242 days in a solar year. The only way this 30-year period could occur on constant lunar alignments, as Plinius indicates, is for the 30-year period to contain 371 months or 10956 days, 0.15 days longer than the 10955.85 days in 371 lunar months and 1.27 days shorter than 10957.27 days in 30 exact solar years (table 46b). The 30-year calendar assumed the solar reconning to be 1 day shorter, and thus it was 0.27 days in error every 30 years. Pliny's statement is sufficient uniquely to characterize the same 30-year calendar outlined in this research as preceding the 25-year Coligny calendar. Plinius's statement must therefore represent a genuine and accurate observation of the earlier Gaulish calendar.

The elder Plinius lived between 24 and 79 AD. If Plinius's took his account from another author, it could have originated from Poseidonius (135-51 BC) travels in Gaul near 90 BC. Of course, Plinius may have obtained his information from a contemporary familiar with the Gaulish calendar. Thus, the observation that the months began on the 6th day of the moon could have been made any time between 90 BC and 80 AD. Since the lunar phases fall later by 1 day every 200 years of the Gaulish calendar, the calendar would have commenced operation sometime between 1090 BC and 920 BC. In its initial period of operation around 900 BC to 1100 BC the months would have begun on the 1st day of the new moon. These potential starting dates indicate that the shift in solar time by 1 day every 23.8 years would have reached 55 days (the time between December 25 and November 1) around after 1300 years of the calendar’s operation. The Irish then could have given up the 30-year Celtic calendar and adopted the Julian calendar somewhere between 200 and 400 AD, overlapping with the missions of Patrick and Cianrán between 350 AD and 450 AD. Thus, the reason given to Plinius's observer for beginning the Gaulish months on the 6th day of the moon (that the moon by that date “has considerable strength”) may be seen as a rationalization. The months simply began by that first-century date on the 6th day of the moon through the gradual progressive displacement of the moon and the calendar due to the error in the lunar reckoning.

In a paper presented to the 2007 Celtic Congress in Bonn, Garrett Olmsted explained the shift in the date of Irish Midwinter from December 25 to November 1 due to a drift in solar time with respect to the cycles indicated on Coligny calendar plate (Olmsted 2009: 193-204). The paper dealt with the implications of this shift to the important connection between the Irish festival Samain and the Coligny-calendar month-name SAMON. The error in the lunar reckoning of the earlier Celtic calendar (with a 30-year cycle) explains Pliny’s observation that the months of the Celtic calendar began on the sixth day of the moon. The term ATENOVX on the Coligny calendar indicates that at its inception the full moon took place on the preceding day 15. Thus, the new moon must have occurred on day-1 of the calendar.

A starting date around 1000 BC for the inauguration of the 30-year Celtic calendar would explain both the 55-day shift in Irish Quarter Festivals accumulated by the end of the fourth century AD and the 5-day shift in the monthly lunar reckoning accumulated by the end of the first century BC. The 30-year Gaulish calendar described by Pliny (24-79 AD) was out of whack with lunar time by 1 day every 199 years. Considering the oscillation of lunar time with respect to solar, one could suggest that a calendar reform to correct the lunar error brought in the 25-calendar in Gaul during the Augustan period. At this point the 25-year calendar entered a 200-year long transmission process finally to be engraved on the bronze plate found at Coligny. Besides being out of whack with lunar time by one day every 199 years, the 30-year Celtic calendar was out of whack with solar time by one day every 23.8 years. This one-day fallback of the sun was noted and kept track of by the calendar. However, the calendar assumed the solar fallback took place over 30 years leading to a progressive error one-day every 112 years in the calendar’s reckoning. The Coligny calendar’s greater accuracy was due to its assuming this one-day fallback occurred after 25 years, much closer to the actual 1-day fallback every 23.8 years. Also, the festivals to celebrate the solstices and equinoxes progressively moved earlier by on day every 23.8 years. A later calendar reform in Ireland brought in the Julian calendar at the conversion to Christianity. By then, the festivals had shifted 55-day with respect to solar time, and the festival for midwinter was celebrated on November 1.

Besides explaining the origin of the Irish Quarter Festivals as well as Plinius’s observation that Celtic months began on the sixth day, the other major contribution arising from Olmsted's study of the Coligny calendar is the determination in the pattern and significance of the triple marks: TII, ITI, and TII. These TII-signes trigrammes are engraved at hundreds of dates on the calendar. Georges Pinault (Goulven Pennaod), one of the coauthors along with Duval of RIG: III: Les Calendriers, accepted Olmsted's reconstruction of the original pattern in the distribution of these TII marks and their associated terminology (in a review in Gnomen (1996), vol. 68: 706-710), although indicating reservations about some of the etymologies for rarely-occurring or uniquely-occurring terminology indicated on the calendar. Pinault accepted as well Olmsted’s suggestion that these marks functioned as a lunar/solar counting scheme.

Il est évident que P.-M. Duval était conscient que ... ces signes [trigrammes] avait bien une raison d’être dont il n’avait pu trouver la clé. L’idée de relier ces trigrammes (qu’il désigne commodément par TII comme sigle général) aux notations PRINNI LOVDIN et PRINNI LAGET et d’en faire des indicateurs en rapport avec les solstices est, sans aucun doute, l’idée la plus féconde de l’Auteur [Olmsted]... En raison de l’abondance relative des TII (de l’ordre de 200), il a été possible d’élaborer sur ordinateur différents schémas de distribution qui ont abouti à son tableaux 28 et aux 29-32 permettant de déterminer la date des solstices dans le cycle de 25 ans. C’est là semble-t-il un apport capital à la compréhen¬sion de la mesure du temps chez les Gaulois. Cela s’obtient au meilleur prix: dans DP [Duval et Pinault] 411-415, j’avais tendé de justifier l’hypothèse de Mac Neill sur un mois EQVOS, pourtant cave, de 30 d aux années I, III, V, mais 28 d aux années II et IV. Si cela apparaît certain en I, II, III et V, c’était moins évident en IV et comme l’hypothèse d’Olmsted d’un EQVOS IV de 29 d, scripturairement possible, permet d’expliquer l’ensemble de l’économie du calendrier, je crois qu’il convient de s’y rallier.

Because of its long evolutionary development with earlier stages still embedded within the later calendar, the Coligny calendar gives a unique window into the astronomical capabilities of a supposedly barbarian people, the Celts of pre-Roman Gaul. The calendar also contains a large number of abbreviated terms describing the day-to-day operation of the calendar, much of it in a seemingly archaic dialect of Celtic. Most of these terms have a clear functional context so that their meaning is not only discernable but verifiable. Because of its significant astronomical and linguistic implications, the Coligny calendar is undoubtedly the most important inscription from Celtic Europe.

Since only 40% of the original five-year Coligny calendar survives as a fragmentary mosaic , the reconstruction of the original whole calendar must depend upon recognizing repetitive patterns in the daily notation and filling in the missing sequences in these patterns which can be projected onto the lacunae. For a full discussion of the determination of these patterns as well as the tables illustrating their functioning, see Olmsted’s 1992 work, The Gaulish Calendar. The patterns apparent in the recurrent notation on the Coligny calendar already have been set forth accurately in this publication, which has been made available recently on academia.edu.

The most significant of these patterns is that discerned in the schemes of the TII and the N lunar/solar counting marks and their associated notation. Unlike the N notation, the pattern in the TII marks (plate 1) only became clear after first determining the original positions of the various shifted days, some of which are specified by ordinal numerals indicating their original day positions (see JIES: XVI (1988), nos. 3-4, pp. 267-339; also see review by Claude Lamoureux in Études celtiques: 30, 1994, 313-315, who compared the importance of this article to that of the earlier article of MacNeill in Eriu X: 1928, 1-67).

At the time of Olmsted’s first study of the calendar 40 years ago, there was no readily-available means of verifying that the reconstruction of the notational patterns of the calendar in typescript (Olmsted 1992: 137-168) actually did fit into the space provided by the lacunae in the fragmentary calendar mosaic without cramping. Any convincing reconstruction must fill in the missing letters to the same size and spacing as the surrounding lettering, while still aligning with the surviving notation and partial notation engraved on the calendar plates. By its very nature, working from a typed transcript, even with the photos of the fragmentary months in hand, is one step removed from the actual reconstruction process and thus potentially prone to error.

Using a photo-processing program (Adobe Photoshop V) in 2001 segments duplicating the missing notation were copied from surviving fragments of the Coligny calendar and then were utilized to fill in the missing sequences on the calendar maintaining the original spatial integrity of the fragmentary mosaic (taken from RIG: III at ¾ scale, but shown in plates 2 and 3 at about ¼ scale and elsewhere at ½ scale). Indeed, the original fragmentary mosaic (plate 2) is still embedded in the digitally-reconstructed whole calendar (plate 3). Thus, the fragmentary calendar was brought to photographic completion utilizing the original wording and engraving to be found on the surviving fragments.

As the photographic reproduction published in 2001 in JIES actually preserves the original fragments of the calendar, the typescript reconstruction presented there and in Olmsted 1992 study is shown to be one which fits within the parameters of the original calendar and lines up with the surviving notation. The reconstruction of the calendar based upon the date patterns of the surviving notation presented in my previous studies do indeed fit convincingly within the lacunae. Furthermore, these reconstructions fit within the alignments for the notation worked out by the original engravers of the calendar.

List of months

Detail of Samonios (year 1).

The names of the twelve lunar year months are reconstructed as Samonios, Dumannios, Rivros, Anagantios, Ogronios, Cutios, Giamonios, Simivisonnios, Equos, Elembivios, Edrinios, and Cantlos. The names occur in the form SAMONI (gen.), DUMANNI, RIVRI etc. in the internal notations of the calendar. The name of the first intercalary month is listed at the end of the month as QUIMON for Quimonios, the second is reconstructed as[ S]antaran[...], [R]antaran[...], [B]antaran[...], or Antaran[...].

Samonios Mid Samonios (Gaulish {lang|xtg|samo-}},< *sṃHo-3) is the name given to solstice festival preceding the half year when the sun is daily raising in the sky and the days are getting lighter while Giamonios Mid Giamonios (Gaulish giamo-) refers to solstice festival preceding half of the when the sun is daily falling in the sky and the days are growing shorter. These two months divide the calendar into summer and winter seasons of six months, each season led off by a festival of several days marked with IVOS. This indicates an early version of the same traditional seasons as seen in later Celtic contexts: “For two divisions were formerly on the year, viz., summer from Beltaine (the first of May), and winter from Samuin to Beltaine”. The medieval Irish Glossary Sanas Cormaic does refer to Samain as the first day of winter, which agrees with the Coligny calendar where the solstice of the sun at its yearly low point occurs in the first month SAMON(IOS). The agreement of these month-and-festival names in itself suggests that the pre-Christian Irish calendar had much in common with that originally used in Gaul.

It is not possible to align the Coligny lunar months accurately with modern solar months. According to those who argue that Samonios must be the summer solstice simply because it contains *samo- the month of MID SAMON began around May–June. According to Olmsted the MID SAMON contains the festival of solstice in December and January.

Value	Name	Days	Etymology	Interpretation
I-1	Unknown	30	Unknown meaning	Intercalary One
1	Samonios	30	Samo- is Gaulish for summer.
2	Dumannios	29	Compare to Latin fūmus. Delamarre suggests "month of fumigations."
3	Rivros	30	Compare to Old Irish remor (stout, thick, fat) and Welsh rhef (thick, stout, great, large). Delamarre suggests "fat month."	J
4	Anagantio	29	Delamarre suggests "month of ritual ablutions."	Olmsted suggests the name means "Offspring of the Goddess Ana .
5	Ogronios	30	Delamarre suggests a month of cold or winter	Olmsted seeing the month at the opposite side of the year suggests *og- "to grow, to fruit".
6	Cutios	30	Delamarre suggests a month of invocations
I-2	[.]antaran[...]	30	Delamarre	see "intercale" Intercalary Two
7	Giamonios	29	This name is derived from giamos, the Gaulish word for winter. mi gam is Old Irish for November, according to Cormac's Glossary.	However, Olmsted sees GIAMON as indicating the half year from the Summer Sostice to the Winter Solstice when the sun is daily lowering in the sky and the days are getting shorter.
8	Simivisonnios	30	Simi could mean half, so "half the course of the sun."
9	Equos	29 or 30	Possibly a month of horses or livestock.	Jan–Feb
10	Elembivios	29	Month of the stag.
11	Edrinios	30	Compare with Old Irish áed (fire, "heat).
12	Cantlos	29	Delamarre suggests "month of chanting."

The lunar month

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. The 5 years of the calendar plaque is part of a Metonic cycle of 19 years, although it could also be extended to a 30-year cycle. The full length of the calendar is still being debated. Helen McKay has attempted to readapt the Coligny calendar to a 19-year Metonic cycle through theorizing a copying error in the notation. She has gone so far as to indicate that her suggestion is now proven. To fit the Coligny Calendar to a Metonic cycle she has to theorize that Intercalary One held 29 days, not the 30 that has until now been presumed, and that the anomalous IVOS run is the result of a copying error. .

IC1 1.

Riu 4.

Gia 8.

Aed 12.

Riu 16.

Gia 20.

Aed 24.

Riu 28.

IC2 32.

Equ 35.

Sam 39.

Ogr 43.

Equ 47.

Sam 51.

Ogr 55.

Equ 59.

Ana 5.

Sim 9.

Can 13.

Ana 17.

Sim 21.

Can 25.

Ana 29.

Ele 36.

Dum 40.

Qut 44.

Ele 48.

Dum 52.

Qut 56.

Ele 60.

Sam 2.

Ogr 6.

Equ 10.

Sam 14.

Ogr 18.

Equ 22.

Sam 26.

Ogr 30.

Gia 33.

Aed 37.

Riu 41.

Gia 45.

Aed 49.

Riu 53.

Gia 57.

Aed 61.

Dum 3.

Qut 7.

Ele 11.

Dum 15.

Qut 19.

Ele 23.

Dum 27.

Qut 31.

Sim 34.

Can 38.

Ana 42.

Sim 46.

Can 50.

Ana 54.

Sim 58.

Can 62.

Each lunar year has a 12 lunar months, six months of 30 days and five of 29 days, although not in 29/30 pairs, and one variable month of 29 or 30 days. A synodic month has 29.53 days, so the calendar overcomes any slight slippage or temporary imbalance by the month of MID EQVOS having either 28&NSBP: 29  or 30 days

Olmsted notes that Intercalary month 2 indicates that year-3 has LAT CCCLXXXV "385 days". So Equos has to have 30 days in year-3. Since the end of Equos survives in years 1 and 5, Equos also has 30 days in Years 1 and 5. So Equos has 30 days in years 1, 3, and 5. Since Equos is an ANMATUS month it must have only 28 or 29 days in at least one of the missing years 2 and 4. Duval and Pinault (1996) suggested 28 days in both of these years to enable the calendar to fit Pliny's statement that the Gaulish calendar followed a 30-year cycle. However, if Equos has 28 days in year 2 and 29 days in year 4, it would follow a 25-year cycle.

Olmsted notes the evidence that Intercalary-Month-1 QVIMON had 30 days, not the 29 days needed for the Coligny calendar to follow the Metonic cycle theorized by McKay. Like Intercalary-Month-2, Intercalary-Month-1 supplies a day for each of the months passed through in each of the preceding 2 ½ years. Since 30 months were passed through in the preceding 2 ½ years, Intercalary-month-1 should contain 30 days. Intercalary-month-1 is stated to be a MATV "good, complete" month. All of the other months labeled MATV have 30 days. The written-out numeral OX[..]ANTIA POG, though somewhat unusual, can be seen as an abbreviation for an archaic Celtic form of 385 and not at all as 384. The term TIOCOB[RIXT] found immediately after the last day of the Intercalary-month-1 also indicates CANTLOS day 15 was used in place of non-existent ATENOVX day 15 to give intercalary- its 30th day. The remaining portion of the broken-off second digit of the Roman numeral for the last day has a slant, indicating that it is a V (for XV) rather than an I (for XI[III]). Since days 8 through days 30 of Intercalary-Month-2 survive, it also had 30 days. Without 29 days in one of the intercalary months McKay’s whole argument falls apart.

According to McKay at the end of the 19-year Metonic cycle, the calendar has overrun the 62-month lunar point by 0.312 days. This would be fixed by reducing an EQUOS month from 30 days back to 29 once every 61 years.

Olmsted contends that each intercalary month contained the 30 days clearly indicated. He also suggests that the Gaulish calendar originally followed the 30-day constant lunar calendar indicated by Pliny, accurate to lunar time to 1 day in about 200 years. Day 15 is followed by the term ATENOVX "Increasing night", indicating that at one point day 15 was the full moon. If at its inception the months began on the first day of the moon after 1000 years of operation the days would begin on day 6 as indicated by Pliny. Counting schemes keep track of a solar fallback by one day every 23.8 years, assumed to be 1 day every 30 years, resulting in a predictive accuracy of solar positions of 1 day ever 112 years. By simply giving Equos 29 days in year 4 rather then 28 as in the 30-year calendar, the calendar increases the accuracy of its solar predictions to 1 day every 455 years. Taking into account that the moon had already shifted 5 days with reference to the calendar, the 25-year cycle has the moon shift to a new 5-day phase of the moon every 25 years. Lunar tracking accuracy increases to one day every 520 years. The shift to the 25-year calendar is indicated by over 200 surviving marks of the type TII, ITI, IIT indicating the phase of the moon dominating each cycle.

A full cycle of 19 or 30 years

The calendar can perform as a 30-year cycle, by extending the 19-year Metonic cycle to use six 5-year cycles, with a 30-day intercalary month dropped once every 30 years.

Pliny stated that the Celts treated 30 years as an ‘age’,

In a 30-year calendar, the moon finishes only 0.1515 days earlier than the calendar, requiring a day to be removed from EQVOS every 199 years. But the lunar/solar difference is larger at 1.4172 days, requiring a 30-day month to be skipped every 198 years. This relatively fast slippage against the solar year would also add to the already large luni-solar swing, for a total of 75 days before a possible adjustment, further aggravating the solar discrepancy, and displacing seasonal festivals by up to two and a half months. This slippage and inaccuracy indicates that the Coligny calendar indicates the reason why the 30-year was replaced by the 25-year calendar

The start of the lunar month

The calendar month is broken into two halves with the term ATENOVX between them. The first half-month has 15 days (called a cóicthiges ‘fifteen-days’ in Old Irish, coicís in modern Irish). The second half-month has either 15 days, or 14 days with the term DIVERTOMV placed over the space for the 15th day. The notation patterns act as though this 'virtual' 15th day is present.

Pliny reported that the Celtic month began on the ‘6th day of the new moon’.

The mistletoe, however, is but rarely found upon the oak; and when found, is gathered with rites replete with religious awe. This is done more particularly on the sixth day of the moon, the day which is the beginning of their months and years, as also of their ages, which, with them, are but thirty years. This day they select because the moon, though not yet in the middle of her course, has already considerable power and influence; and they call her by a name which signifies, in their language, the all-healing.

Classical writers counted from the day of the first visible moon, so the 6th day would be the first quarter moon, Day 1, the start of the calendar's month. The quarter moon with its D-shape is the only moment in the lunar phase that is easily identifiable by eye. The internal notations of the calendar confirm Pliny's statement, with a focus on the middle triplet of days in each half-month, days 7-8-9 (the full moon) and days 7a-8a-9a (the dark invisible moon).

Full Reconstruction

A full reconstruction of the calendar by McKay (2020) includes the latest information about the intercalary notations and the triple marks. Olmsted (2001) offers a previous reconstruction, which usefully aligns the notations with photographic images. RIG III (1986) presented an earlier in-depth description of terms with a reconstruction.

Sample month

MID SAMONIOS of year 2 is the only month out of 62 that has been preserved without any gaps. Currently, most of the patterns of the various notations are known, even if their significance may not be understood. Because of this, most days on the calendar can be reconstructed with confidence.

M SAMON MAT
◎ I		N	DVMAN IVOS
◎ II	ıƚı M	D	IVOS
◎ III	ƚıı	D	DVM IVO
◎ IIII	M	D
◎ V		D	AMB
◎ VI	M	D
◎ VII		PRIN     LOVDIN
◎ VIII		D	DVM
◎ VIIII	ııƚ M	D
◎ X	M	D
◎ XI		D	AMB
◎ XII	M	D
◎ XIII	ƚıı M	D
◎ XIIII	ıƚı M	D
◎ XV	ııƚ M	D
ATENOUX
◎ I		D	DVMAN
◎ II	ııƚ	D	TRINVX SAMO
◎ III		D	AMB
◎ IIII	ƚıı M	D
◎ V	ıƚı	D	AMB
◎ VI	ııƚ M	D
◎ VII		D	AMB
◎ VIII		N	INIS R
◎ VIIII		N	INIS R
◎ X	ƚıı M	D
◎ XI	ıƚı	D	AMB IVOS
◎ XII	ııƚ M	D	IVOS
◎ XIII		D	AMB IVOS
◎ XIIII	M	D	IVOS
◎ XV		D	AMB IVOS

Drawing of month 14 (Samonios of year 2) by de Ricci.

The month begins with M[ID] SAMON[I] MAT, the 'month of SAMONIOS lucky'.

The double circle "◎" in the table indicates the peg-hole for marking the current day, followed by a Roman numeral for the day's number in the half-month.

All days here were originally marked as M D 'lucky day' because SAMONIOS is marked as a MAT month in its header, but this will often be subsequently overwritten by other notations D 'day' (neutral), D AMB 'unlucky day', or N 'night' as they are added in turn.

The notations are usually visually aligned on the D or N. Terms are often shortened, and the spelling is non-standard and often varies.

Next are occasional triple-marks of the form ƚıı ıƚı or ııƚ, in that order before major movements and overwriting. These follow the same offset pattern as the PRINNI notations, and likely divide the daytime into three periods.

Days 5 and 11 in the upper coicise and each odd day (except day 1a) in the lower coicise are marked with D AMB 'inauspicious day'. Day 9a will end up having its D AMB overwritten by N INIS R.

The notation N INIS R occurs in this month on days 8a and 9a. The significance of this nighttime term in unknown.

The name of the following month, DVM(ANNI), is marked on days 1, 3, 8 and 1a. This tracks the swapping of these days' notations (all of them) with the following month DUMANIOS days 1, 8, 1a and 2a where the notations from SAMONIOS have SAMONI added in their turn. Day 2a, first swapped with DUM day 2a, then undergoes another anomalous swap with SAM day 3. Days with notations that have been moved are always marked with their originating month's name (and day name if different).

The notation PRINNI LOUD sits in months marked MAT, at the first day of the first month (Samonios), the second day of the second MAT month (Rivros), and so on for 8 instances. Another PRINNI LOUD originally at SAMONIOS day 1 has been swapped with DVMANNIOS day 1 below it .

The Day 2a(17) is marked with TRINVX SAMO, and this term also has SINDIV IVOS 'festival this (one) day' added to it in years 1 and 4. This means that this day's notations have been swapped with day 3 (TRINVX) of SAMONIOS, after first being swapped with DVMANNIOS day 2a, whose notations now sit at SAMONIOS day 3  ƚıı D DVM IVO. (SAMONIOS day 2a's original notations are found in turn at DUMMANIOS day 2a).

Days 1–3 are marked with a sequence of IVOS, a term interpreted as "festival". This run of IVOS started on the last two days of the previous month CANTLOS, days 13a-14a, so the whole festival lasts for 5 days. This probably equates with the festival of Beltaine, although these sorts of specific terms are not used on the calendar, festivals only being marked with runs of IVOS.

Finally, Day 1 has its 'day' terms overwritten by a single N, without changing the rest of its notations. Originally, it started off with ƚıı M D, was swapped with DUM day 1 receiving D DUMANI, had an IVOS added to give D DUMANI IVOS, and now has that D overwritten by a single N to end up with N DUMANI IVOS. This single N indicates that the notations of SAMONIOS day 1 in this year 2, D DUMANI IVOS, have been used to help create the notations of Intercalary Two day 1.

The Notations

Several different notations, each with their own pattern, are placed sequentially on the 12 lunar months of the calendar, interacting according to certain rules with the notations before them, often replacing them. After the basic notations are set, many days’ notations are then moved to other days, creating visual chaos. Finally, the days of the intercalary months are filled with notations copied from certain days in the 12 yearly months.

The notations, their patterns and interactions have gradually over the last century been identified by several key researchers, and what follows is a general, but not comprehensive, overview of each notation.

Numbering the days

Each month has two halves. The first half has days numbered from I to XV (1 to 15). The second half has either I–XV (1–15), or I–XIIII (1–14) with the 15th day marked with DIVERTOMU. The term ATENOVX is placed between the two half-months. The patterns of the notations act as though the 30th day is always present. This means that in practice some months only have 29 days, but conceptually, all months have 30 days.

MAT and ANM months and their days

Months of 30 days were marked MAT (except EQVOS), months of 29 days were marked ANM(AT) (except Intercalary One). MAT and ANM(AT) have been read as "lucky" and "unlucky", respectively, based on comparison with Middle Welsh mad and anfad and Old Irish mad and ni-mad.

the MAT and ANM months

	SAMONIOS season			GIAMONIOS season
1	mat	SAM	7	anm	GIA
2	anm	DUM	8	mat	SIM
3	mat	RIV	9	anm	EQU
4	anm	ANA	10	anm	ELE
5	mat	OGR	11	mat	EDR
6	mat	CVT	12	anm	CAN

Six months are marked in their header as MAT ‘good, auspicious’, and six months as ANM[AT] ‘not good’. The summer season has 4 MAT months, and the winter season only has 2 MAT months. The summer season needs more auspicious time for all its activities. The months do not run in 29/30 pairs.

For months marked as MAT, all days are initially given M D, a good or auspicious day. Days of the months marked as ANM are given just D, a neutral day. The terms M D and D refer to daylight hours and are in apposition to N for night. Any type of notation marked with N (night) will overwrite the full daytime notation, including the triple mark, M D, D, or D AMB.

The notation D AMBRIX RI

the D AMB pattern (orange) for a lunar month

D AMBRIX RI, usually shortened to D AMB, denotes an inauspicious day. It occurs only on Days 5 and 11 in the upper half-month, that being the period when the moon is more than half full, so it's mostly left free of inauspicious days. In the second half-month, D AMB is placed on every odd numbered day except Day 1, but this is explained by the traditional view that the unit 1 is neither odd nor even. The use of odd numbers as inauspicious is also seen with most months of 29 days being ANMAT ‘not good’. It is symptomatic of Celtic cultures, as the Romans held the reverse view, that odd numbers were auspicious.

The triple marks

the base pattern of the triple marks

The triple marks are a series of ogham-like marks. They are first lain down each month in triplets over three days, ƚıı, ıƚı, or ııƚ, followed by three days with none. As they only occur with days marked with D (for daytime), and never N (for nighttime), they likely divide the daytime into three divisions.

The triple marks are by far the most complex notations, composed of three main patterns. They do not always repeat across the years. The first pattern assigns possible triplet positions which start on the same offset as the first PRINI term in the month, moving down a day in each of the following MAT or ANM months. The first triplet starts on Days 1-2-3 of SAMONIOS in Year 1, Days 2-3-4 in RIVROS, and so on following the MAT sequence of months. The equivalent sequence starts on Days 1-2-3 of GIAMONIOS in Year 3 and follows the ANM months, so mirroring one intercalary period to the other.

A second pattern, again following the MAT/ANM sequence, determines which triplets of the first pattern will manifest from year to year. This means the triple mark on a day/month of one year may not be found on the same day/month in another year.

A third pattern adds another IIT on Day 21(6a), the last day of the visible moon, adding to another mark if already there, resulting in each Day 21 holding either TIT, ITT, or IIT.

the final triple marks of year 4 (after all exchanges)

The triple marks undergo many changes as other notations are added. Days with N forms of notation overwrite the whole ‘day’ notation, e.g. IIT MD becomes just N, while ITI D AMB becomes just N. Days are moved and exchanged, often overwritten and lost, intercalary borrowed days are marked with N, and so on. The result turns a complex pattern of triple marks into visual chaos.

Olmsted notes the following more fully explained in work The Gaulish calendar . Embedded within the notation of the final 25-year-cycle of the Coligny calendar indicated by the solstice counting schemes is an earlier 30-year-cycle calendar. The more accurate later 25-year-cycle lunar calendar was developed simply by giving EQVOS 29 rather than 28 days in year-4. Although the last eight days of this month in year-4 are completely missing from the fragments of the calendar plate, the six-days separation in the occurrence of the triple cluster of these marks (TII, ITI, IIT) demonstrates that the calendar followed a 25-year cycle in which the midwinter intercalary solstice dates where separated by six days rather than the five days they would have been separated had they followed a 30-year cycle. The 30-year cycle can only occur by giving EQVOS in Year-4 the 28 days it had in the earlier calendar described by Plinius. Since the TII marks demonstrate that six days separate the solstice dates occurring in the midwinter intercalary month (occurring every 5 years), 5 times six-days separation equals 30 days, and the calendar recycles after 5 such five-year phases. The separation of 5 years in the intercalary midwinter solstice dates of the 30-year cycle would need 6 such five-year phases since 6 times five-days separation adds up to 30 days. The notation on the Coligny plate indicates that at its initiation an original constant-lunar 30-year-cycle calendar system had each month begin on the first day of the new moon. The inherent error of approximately one day every 200 years, after 1000 years each month would begin on the sixth day. The new 25-year calendar was created to take into account this 5-day shift. The new 25-year calendar shifts to a new 5-day-week phase of the moon at beginning of each new 25-year cycle. The symbol TII indicates the first 5-day-week phase of the moon, ITI the second five-day-week phase beginning after 25 years of operation, and IIT the third five-day-week phase beginning after 50 years of operation and two 25-year cycles through the calendar. The clusters of these three different marks occur on three sequential days, a result of the sun’s falling back relative to the calendar by 1 day in each 25-year cycle.

The notations PRINI LOUD and PRINI LAG

the PRINI LOUD and PRINI LAG pattern

month		MAT months PRINI LOUD day		ANM months PRINI LAG day
1	SAM	1	7
2	DUM				5
3	RIV	2	8
4	ANA				6
5	OGR	3
6	CUT	4
7	GIA			1	7
8	SIM	5
9	EQU			2	8
10	ELE			3	9
11	AED	6
12	CAN			4

PRINI LOUD has the same MAT month offset, and PRINI LAG the same ANM month offset, as the triple marks. If it falls on a triple mark, it replaces it, along with any M D, D, or D AMB. The PRINI LOUD of SIM 5 is later overwritten by N INIS R. Exchanges will lead to some PRINI LOUD ending up in ANM months, and vice versa. PRINI LOUD does not only occur in the SAMONIOS season, and PRINI LAG does not only occur in the GIAMONIOS season – the SAMONIOS season of 6 months, and the GIAMONIOS season, both contain examples of PRINI LOUD and PRINI LAG.

The notation N INIS R

the N INIS R pattern

The term N INIS R is scattered across the lunar year. The significance of its distribution is undiagnosed. All but three instances occur in the seven months of the SAMONIOS season plus the month of GIAMONIOS. It avoids the days marked with IVOS ‘festival’. As it occurs on seven nights when the moon is absent in the sky (the dark moon of 7a-8a-9a), and avoids the critical moments of the full moon of day 8 and the first visible moon of day 10a, it possibly refers to prognostication associated with stars.

The notations IVOS and SINDIV IVOS

the notations IVOS, SINDIV IVOS, and TIOCOBRIXTIO

The term IVOS ‘festival’ occurs in several runs of days of between three and nine days each, considered to mark each day of a festival. In all but two cases these festivals run from the end of one month into the beginning of the next. Four of these IVOS runs break the year into four-quarters, just as the four main Celtic festivals do in historic times, only here they are centered on Day 1 every three lunar months, rather than Day 1 of every three solar months as today.

There are also three other IVOS festivals on the calendar.

The term SINDIV IVOS ‘this day a festival’, occurs only three times – DUM 2a, SIM 9, and AED 25. These three special festival days must indicate something of exceptional importance in the year.

The notation TIOCOBRIXTIO

TIOCOBRIXTIO is an exceptional term which only occurs on three days in the year – SIM 7, AED 8, and CAN 15. Whatever its significance, it marks days of exceptional importance. Olmsted explained it as T(R)IOCO(NT)O-BRIXTIO "A day in place day 30". It indicates a day used to substitute for a day, such as Cantlos day 30. Cantlos, being an ANMATUS month only has 29 days so day 15 with this notation was substituted for the missing day 30.

Movement of notations between days

At this point, most notations have been assigned their base position on the calendar. What happens next is a major feature of the calendar, the movement of one day's notations to a different day. This visually breaks up the patterns of the notations, making the calendar seem quite random. This exchanging of days according to several different patterns, is a major aspect of the calendar, involving a total of 870 days over 5 years.

EXCHANGES: swapping notations between two days

the pairs of swapped days

There are several patterns in which two days swap their notations.

• The first pattern only involves Day 1 in four pairs of months.
• The second pattern involves days other than Day 1, and uses a different set of four pairs of months to swap between. Days are swapped with the same day of a neighbouring month.
• A third pattern is called the anomalous swaps, where days are swapped between a different day of a month. This occurs just three times per year: between SAM 3 and SAM  2a, between RIV 4 and RIV 10a, and between RIV 8a and ANA 4.

examples of the same and different day swaps

As the notations of one day are moved to another, they take the information with them about their original position (presumably so that one day can be used to prognosticate for its swapped partner). As most movements are to the same day of the month, the day information is redundant, so only the month name (in the genitive) is added. But anomalous swaps between different days require both their original day name and the month to be added.

EXCHANGES: dragging notations between months

examples of dragging notations

YEAR 1 month/day	pre-drag	post-drag
GIAM 7	PRINI LAG	MD SIMIVIS TIOCOBREXTIO
GIAM 8	D	MD SIMIVIS
GIAM 9	N INIS R	MD SIMIVIS SINDIV IVOS
SIMIVIS 7	MD TIOCOBREXTIO	D EQVI
SIMIVIS 8	MD	PRINI LAG EQVI
SIMIVIS 9	MD SINDIV IVOS	D EQVI
EQUOS 7	D	D ELEMB
EQUOS 8	PRINI LAG	D ELEMB
EQUOS 9	D EQVI	D ELEMB

For the 12 lunar months after an intercalary month, the notations of the triplet of days 7-8-9 (the full moon) and 7a-8a-9a (the dark moon) in each month are dragged sequentially upwards to the previous month, like beads on a string. Their original month name is then added to the notations.

examples of dragging IVOS

YEAR 1 month/day	pre-drag	post-drag
OGRON 28	D AMB	D AMB IVOS
OGRON 29	MD	MD IVOS
OGRON 30	D AMB	D AMB IVOS
CUTIOS 1	MD	MD IVOS
CUTIOS 2	MD	MD IVOS
CUTIOS 3	MD	MD IVOS
CUTIOS 28	D AMB IVOS	D AMB
CUTIOS 29	MD IVOS	MD
CUTIOS 30	D AMB IVOS	D AMB
GIAM 1	MD SIMI IVOS	MD SIMI
GIAM 2	MD IVOS	D
GIAM 3	MD IVOS	D

The notation IVOS is also sequentially dragged upwards a month in the post-intercalary year. However, it does not take all the other notations with it. This keeps the festival runs marked with IVOS intact. The same also applies to SINDIV IVOS.

The notations of the intercalary months

days copied to the intercalary months

The notations on the days of the intercalary months are created by a complex series of copies and merges of notations from certain days in the normal lunar months. Each day of an intercalary month sequentially copies a lunar month and the same day number, with its source month name added. At first 30 days are copied, and for days 1 to 18, their day number is replaced with a single N at the copied site. Secondly, a sequence of days 1 to 6 is again copied from a different year, and these are merged with the first. Thirdly, the days 7-8-9 and 7a-8a-9a which have been dragged from the following month are again merged with the copied notations. At which point, the calendar's notations are complete.

Bibliography

• Delamarre, Xavier (2003). Dictionnaire de la langue gauloise: une approche linguistique du vieux-celtique continental. 2nd edition, Paris, Editions Errance. ISBN 2-87772-237-6.
• Dottin, Georges, La langue gauloise : grammaire, textes et glossaire (1920) no. 53, pp. 172–207.
• Duval, Paul-Marie and Pinault, Georges (eds) (1986). Recueil des inscriptions gauloises (R.I.G.), Vol. 3: Les calendriers de Coligny (73 fragments) et Villards d'Heria (8 fragments). Paris, Editions du CNRS.
• Hitz, Hans-Rudolf (1991). Der gallo-lateinische Mond- und Sonnen-Kalender von Coligny.
• Joyce, P.W. (2000). "Old Celtic Romances". The pursuit of the Giolla Dacker and his horse. Wordsworth Editions Limited, London.
• Laine-Kerjean, C. (1943). "Le calendrier celtique". Zeitschrift für celtische Philologie, 23, pp. 249–84.
• Delamarre, Xavier (2003). La langue gauloise. Paris, Editions Errance. 2nd edition. ISBN 2-87772-224-4. Chapter 9 is titled "Un calendrier gaulois".
• Le Contel, Jean-Michel and Verdier, Paul (1997). Un calendrier celtique: le calendrier gaulois de Coligny. Paris, Editions Errance. ISBN 2-87772-136-1
• Mc Cluskey, Stephen C. (1990). "The Solar Year in the Calendar of Coligny". Études Celtiques, 27, pp. 163–74.
• McKay, Helen (2016). "The Coligny calendar as a Metonic lunar calendar". Études Celtiques, XLII, pp. 95–122.
• McKay, Helen (2018). "Defining the systematic patterns for the triple marks of the Coligny calendar". Études Celtiques, XLIV, pp. 91–118.
• McKay, Helen (2022). "Building the Intercalary Months of the Coligny calendar". Études Celtiques, XLVIII, pp. 55–78.
• Mac Neill, Eóin (1928). "On the notation and chronology of the Calendar of Coligny". Ériu, X, pp. 1–67.
• Monard, Joseph (1996). About the Coligny Calendar. privately published monograph.
• Monard, Joseph (1996). Découpage saisonnier de l'année celtique. privately published monograph.
• Monard, Joseph (1999). Histoire du calendrier gaulois : le calendrier de Coligny. Paris, Burillier. ISBN 2-912616-01-8
• Olmsted, Garrett (1992). The Gaulish calendar: a reconstruction from the bronze fragments from Coligny, with an analysis of its function as a highly accurate lunar-solar predictor, as well as an explanation of its terminology and development. Bonn: R. Habelt. ISBN 3-7749-2530-5
• Parisot, Jean-Paul (1985). "Les phases de la Lune et les saisons dans le calendrier de Coligny". Études indo-européennes, 13, pp. 1–18.
• Pinault, J. (1951). "Notes sur le vocabulaire gaulois, I. Les noms des mois du calendrier de Coligny". Ogam, XIII, pp. 143–154
• Rhys, John (1910). "The Coligny Calendar". Proceedings of the British Academy, 4, pp. 207–318.
• Thurneysen, Rudolf (1899). "Der Kalender von Coligny". Zeitschrift für celtische Philologie, 2, pp. 523–544
• Zavaroni, Adolfo (2007). On the structure and terminology of the Gaulish calendar, British Archaeological Reports British Series.

External links

• The Gallic calendar – Lugdunum Museum