Changes

A Celtic lunisolar calendar has been reconstructed from fragments of two second AD calendars found in the French towns of Coligny and Villards d'Heria. The first month on this calendar is called ''Samonios''. The Celts celebrated a three-day festival that began on the 17th of ''Samonios'', according to the calendar.<ref name="Olmsted"/>

A Celtic lunisolar calendar has been reconstructed from fragments of two second AD calendars found in the French towns of Coligny and Villards d'Heria. The first month on this calendar is called ''Samonios''. The Celts celebrated a three-day festival that began on the 17th of ''Samonios'', according to the calendar.<ref name="Olmsted"/>

With a leap day every five years, the Coligny calendar gives the length of a year as 365.20 days. The true length of a solar year is 365.2422 days.<ref name="Olmsted"/>

If Samhain was originally the date of the winter solstice, it could migrated to its current November 1 location due to a combination of calendar drift and round off error. Roman sources record that the Celts used a thirty-year calendar cycle. The Coligny calendar is five years long. A thirty-year cycle can be constructed with six Coligny calendars end-to-end, but with one month of 29 days dropped. That would give the Celtic year an average length of 365.2333 days. The true length of a solar year is 365.2422 days. The calendar would have drifted by one day every 113 years. Placing the start of a season on the first of the month has obvious advantages in terms of convenience.

 

The amount of drift can be used to estimate the age of a calendar. According to [[Pliny the Elder|Pliny]], Celtic months started on the sixth day of the lunar cycle in the period 120-100 BC. As the lunar cycle drifted by one day every 199 years, this suggests that the calendar had been operating without correction for 975 years.<ref name="Olmsted"/>

 

If Samhain was once on the winter solstice, it could have gradually moved to its current location due to calendar drift. Assuming Samhain and Beltaine transferred to the Julian calendar around AD 200, that gives us 1,300 years of solar calendar drift (1100 BC to AD 200), or 50.6 (1300/23.7) days worth of drift. November 1 is 51 days before the solstice on December 21. That this rough calculation yields the desired date exactly is surely an artifact. But placing the start of a season on the first of the month has obvious advantages in terms of convenience.

==References==

==References==