Intercalary month in the Old English calendar, added in summer in some years
For an overview of the Old English calendar and links to the other months, see the summary page.
The sixth and seventh months of the year, corresponding approximately to the Roman and modern months of June and July, were called Litha. Like its counterpart in the winter, Giuli (from which we get the word Yule), Litha was a double-length month, or two months of the same name, placed either side of the midsummer solstice. See the article on Litha for more details.
Bosworth and Toller's Old English Dictionary says that the first Litha month (corresponding roughly to modern June) was called "se aerra Litha" and the second one was called "se aeftera Litha". Kathleen Herbert says that the information in Bosworth and Toller comes from Bede's treatise On the Reckoning of Time, and from later English scholars who commented on it such as Aelfric and Bryhtferth (Herbert 1994). The distinguishing terms 'aerra' and 'aeftera' aren't in Bede's account, so I presume they come from one of the later commentaries, but I haven't verified the source. As I understand it, the meaning is closer to "the earlier Litha" and "the later Litha", rather than "the month before Litha" and "the month after Litha", so "Litha" refers to the name of the months and not to a date that occurred at the junction between them.
Bede, writing in 725, tells us:
When an embolismic year occurred (that is, one of 13 lunar months), they assigned the extra month to summer, so that three months together bore the name "Litha", hence they called the year "Trilithi".
--Bede, On the Reckoning of Time, Chapter 15. Translated by Faith Wallis.
This neatly demonstrates both the problem with a lunar-solar calendar and the early English solution to it. A solar year refers to one complete cycle of the sun from one midwinter solstice (or any point of your choice) to the next. This is a natural way to reckon time in an agricultural society living at temperate latitudes, where day length would be an important determinant of agricultural activities. But it is rather long to be the only unit used to measure time.
The lunar month, covering a complete cycle of the moon from one full moon (or any other point of your choice) to the next, is a shorter unit of time, conveniently intermediate between the long unit of the solar year and the short unit of the solar day. Thus a lunar-solar calendar gives you three units of time each of a different order of magnitude, so you don't have to express time periods either in tiny fractions of a year or in very large numbers of days. Great.
Unfortunately, the problem is that none of these natural units of time are exact multiples of each other. A lunar month is 29.53 days. A solar year is 365.24 days. There are 12.37 lunar months to a solar year. So the lunar months won't line up neatly with the solar year. Suppose you start your lunar-solar calendar at a time when the full moon also falls on the midwinter solstice, so both the year and the first month of the year start on the same date. The second month of the year starts at the next full moon, the third month starts at the full moon after that, and so on through the year. But 12 lunar months only take 29.53*12 = 354.37 days to complete. So by the time the next midwinter solstice comes round, at 365.24 days, the moon is already 10.87 days past the full. What do you do? Do you start the new year when the moon was full? In which case the year won't match the solstice. Or do you start the first month of the new year at the solstice? In which case the months won't match the phase of the moon.
I can imagine priests, druids and learned folk tearing their hair out over this infuriating astronomical feature. Some cultures settle on a purely solar calendar and let the months go out of phase with the moon (our modern Western calendar does this), others settle on a purely lunar calendar and let the year go out of phase with the sun (the Islamic lunar calendar does this). Others adopt a hybrid system, adding an extra month when necessary to bring the lunar months back into line with the solar year - a sort of "leap month", if you like, in the same way as the modern Western calendar adds a day (almost) every four years to keep the calendar synchronised with the solar year. This extra month is called an intercalary month.
Clearly the early English applied this hybrid approach, adding an extra month to Litha to keep the lunar months in line with the solar years. This would happen every two or three years (every 2.72 years to be precise), so a "Trilithi" year would be pretty common. It could have been decided by calculation, by observation, or a mixture of both. If you kept a count of the observed full moons starting at the midwinter solstice each year and the second Litha full moon happened before the midsummer solstice, you would know it was a Trilithi year. If you also kept a count of the days and alternated between 29 and 30 days for a lunar month, you could calculate the date of the full moon even if the weather was too cloudy for a direct observation.
Bede: The Reckoning of Time. Translated by Faith Wallis. Liverpool University Press, 1999, ISBN 0-85323-693-3.
Herbert, Kathleen. Looking for the Lost Gods of England. Anglo-Saxon Books, 1994. ISBN 1-898281-04-1.