Observing Bede's Anglo-Saxon Calendar

by John Robert Stone © 1997

In 1973 and again, in 1976, Kenneth Harrison considered a possible reconstruction of St. Bede's Anglo-Saxon calendar. In both places, Harrison employs an anciently known intercalation cycle that spans eight years or 99 lunations. It is called the octaëteris or ogdoas. It corresponds to the intercalation pattern at the start of the Metonic cycle, namely OOEOOEOE, where O represents an ordinary, or hollow year, of twelve lunar months and E an embolismic, or swollen, year of thirteen. This article does not wish to challenge Professor Harrison's efforts, but to identify an alternative family of intercalation rules of a generally more empirical model.

Let us begin by restating Bede's topics and principal assertions in Chapter 15 of De Temporum Ratione ("On the Reckoning of the Seasons").

• The English month is a lunation, a cycle of lunar phases.
• The month names are listed with approximate Julian correlates. (See month chart.)
• The beginning of the English year falls on Christmas Eve, which was called by the heathens "Mothers' Eve."
• The year was divided into two seasons, winter and summer. Winter began at Winterfylleð, which Bede explains as "Winter-full-moon".
• In intercalary years a third month named Liða was inserted in summer and made a lunar year of thirteen months, which Bede designated a "year of three Liðas".
• The origins and meanings of the month names are speculated upon.

The Main Argument.

Solstitial Clues.

The Beginning of the English Month.

Both the Jewish and the Muslim religious calendars begin their days at sunset. While it seems counter-intuitive to begin a new day at the end of the work day, it is nevertheless a telltale trait of societies that have used, at one time or another, lunar months and have reckoned them from the first sighting of the lunar crescent after new moon. The Babylonian calendar, the progenitor of the Jewish, also began its months "when the thin crescent of the new moon was first visible in the sky at sunset" (Parker and Dubberstein 1956:1).

If "new crescent" means "new month," it reasonably implies "new day" as well, because a new month should commence on a separate day. Hence, the day is reckoned to start at sunset. Here is how it works in practice. Sunset is observed and a new day begins. As civil twilight deepens, the sighting of a new crescent moon, shimmering in the west, announces whether this day belongs to a new month, as well.

Old English dictionaries, such as Clark Hall's, document that the English began their days at sunset because their day names changed at sunset. At sundown of "Tiw's day" (Tiwesdæg) Tuesday became "Woden's eve" (Wodnesniht). At sunrise, "Woden's eve" gave way to "Woden's day" (Wodnesdæg), but, at sundown, Wednesday became "Thunor's Eve" (Ðunresniht), and so on. (See the chart of the English week.) The change of the "planet" name at sunset shows that conceptually the day had changed.

Further, three Christian feasts, Christmas, All Saints' Day (All Hallows), and St. John's Day, still remember an evening component, although Halloween and St. John's Eve (Midsummer Eve) have lost any church significance.

Also, the meaning of night as 'eve', when it occurred in combination with a name, persisted into classical New English as two of Shakespeare's titles attest.¹ Twelfth Night according to the Oxford English Dictionary is the same as Twelfth Eve, January 5, the eve of Twelfth Day (Epiphany). A Midsummer-night's Dream is a dream meant for Midsummer Eve, not the night of June 24 when the enchantment is over!

Interestingly, the English counted a number of days as so many "nights." It is still heard in the British "fortnight", two weeks, and formerly heard in "se'nnight," one week. A well-known passage of Tacitus, from which we shall quote later, documents this practice in a broader German context. (Moreover, just as the English counted their days as so many nights because the day began at evening, so they counted their years as so many winters because the year began at Midwinter. For example, in Richard II, Shakespeare has John of Gaunt say, "What is six winters? they are quickly gone.")

From this indirect evidence we conclude that the English, and other German peoples too, changed months upon the observation of a new lunar crescent. Sighting the crescent meant that the new day, begun not half an hour before, belonged to this new month. The crescent was the knife of time, which cut one month from another.

Three Sample Rules.

What follows are three examples of intercalation rules that keep the "after" months after their respective solstices. The calendars generated by these rules work about the same. They may insert the third Liða in different years, but, over time, they keep apace of each other.

Rule #1. The next month is intercalary if the first crescent of the after Liða is observed on or before July 4, the eleventh evening after Midsummer Eve (June 23).

The reason that the rule employs a zone of eleven evenings is because a lunar "year" of twelve lunations amounts to only about 354 or 355 days, eleven or ten days shy of the length of the tropical year, the year of the (solstice-controlled) seasons. Therefore, in every tropical year, the lunar year slips back about ten or eleven days. In the case above, if the after Liða crescent is seen on or before July 4, then it threatens, in the next year, to fall before the solstice, after which the lunar months will no longer represent the season.

Rule #2. The next month is intercalary if the first crescent of the after Liða is observed before Midsummer.

This version of the rule, with no detection zone, is simpler, but does not regulate the lunar months as strictly. The after Liða will rise before the solstice marker in intercalary years.

The twelve days of Christmas make a detection zone of perfect position and length.

Rule #3. The next summer will contain a third Liða if the first crescent of the after Geola is observed within the eves of Christmastide (December 24 to January 4).

Another, similar set of intercalation rules could have been generated by using the observed summer or winter solstice instead of the English quarter days Christmas and Midsummer, but Bede specifically mentions Christmas as the start of the year. So Midsummer was picked for consistency with Bede. The other English quarter days are Lady Day (Annunciation of Mary), March 25, and Michaelmas, September 29. The English months that would generally fall after these are, respectively, Eastermonað (first month of the English summer), and Winterfylleð (first month of the English winter).

Difficulties.

The principal difficulty of Bede's calendar lies in this very reference to Christmas. Not only does the "Mothers' Eve" vigil stand apart from the lunar months, but also it refers to a date in another calendar - the Julian calendar of the Roman Empire. This implies that the Julian calendar must have been at hand and running in parallel. Perhaps Bede's Christmas New Year's is the result of cultural influence from the Roman New Year's, which fell, as it does today, on January 1.

The following argument explains how such an influence could have arisen. First, Caesar's Julian calendar was established in 46 B.C. and proved to be, for the time, an accurate rendition of the tropical year. Second, various German tribes adopted the Roman week and substituted their gods' names for the Roman planet names. This occurred possibly around the third century (E.O.G. Turville-Petre 1964: 101), a time when the Roman Empire employed many German legionnaires. This clearly shows Roman-to-German acculturation. Finally, the Mithraic cult was widespread throughout the Empire, especially in the army. Recognition of December 25 as the Mithraic celebration of Sol Invictus, the rebirth of the Invincible Sun, must have also been widespread. After all, December 25 became the date favored for the celebration of Christmas in the Western Empire, as opposed to January 6, favored in the East.

The foregoing leads to the conclusion that, from about the third century on, December 25 was simply a popular day for acknowledging the winter solstice in the Western Empire. If so, Germans, perhaps through their former soldiers, could have adopted this date well before the migration into Britain in the fifth century. Before this putative adoption, the German tribes might have used a truly observed solstice or some other marker for one.

For this model, the Mothers' vigil is simply part of the solar component in a lunisolar design, but plays no driving role². With the months properly regulated through the summer intercalation, the vigil, however, would normally precede the æfterra Geola crescent. Perhaps the vigil inaugurated a New Year's transition that ended on the sighting of the crescent.

A Seasonal Aid, Not a Dating Tool.

These suggestions equip us with an English calendar capable of bestowing a seasonally-appropriate name to a given lunation. They do not imply, I think, that the English numbered the days of the month like "17 Solmonað." It would have been impossible (as it is even today) to predict the visibility of the next new crescent. So, the months would have had an indeterminate length; the English could not have created dates with this model.

Further, had dates been created, they would be unusable after a certain time. Old dates are meaningful in calendars where the months have fixed lengths based on rules. Probably, the English planned and predicted only in the short term by the moon's quarter phases or by - what is essentially the same - weeks.

The historian Tacitus, writing in A.D. 98, lends support here.

[The German chiefs] assemble, except in the case of a sudden emergency, on certain fixed days, either at new or full moon; for this they consider the most auspicious season for the transaction of business. Instead of reckoning by days as we do, they reckon by nights, and in this manner fix both their ordinary and their legal appointments. Night they regard as bringing on day. (Germania, Chapter 11, Church and Brodribb translation)

A Working Example.

The "Sample Reconstruction" chart illustrates how Rule #1 above would handle the seasons between 1991 and 2002. While calculating the date and time of the new moon is mathematically easy, calculating the likelihood of sighting the first crescent is hard. It can involve many factors, some of which can deal with ophthalmology and atmospheric physics, as in the theory of Bradley E. Schaefer. By lucky coincidence, a computer program, made available for beta testing, was able to perform the computations needed for this study. It was called "Moon Calculator" and was created by Dr. Monzur Ahmed of Birmingham, England. It was downloaded from the Internet in July, 1996.³

"Moon Calculator" provided, by default, the criterion of Mohammed Ilyas for the visibility of the crescent - specifically the Ilyas A criterion. This simply compares the altitude of the moon to its angular separation from the sun. The complete sample was run for Greenwich, England, at latitude 51N32 and longitude 00E00 between 1900 and 2052.

While the Ilyas criterion rates some pretty young crescents as visible, it is still adequate to demonstrate the model's seasonal aptness. Yet one should be aware that a date stated in the example chart may be one day too early for practical viewing of the crescent. Consider each date as a minimum baseline for a particular lunar sighting. Suppose an intercalation was missed in a year when it should have been made. The model does not then become "seasonally unhinged." The observer will simply realize, as the months parade by, early in season, that intercalation will be necessary during the next summer. After that, the calendar should be caught up.

Harrison's ogdoas model, the Metonic cycle, and the intercalation rules stated here were tested against these dates. What is most astonishing is that the old calendar cranks along about the same no matter which rule is used!

Charts

The English Months

The English Week (based on Clark Hall's dictionary, 1960)

Sample Reconstruction of Bede's Anglo-Saxon Calendar

The lunar months commence at Greenwich, England, on the dates shown, or on the evening following. Dates causing intercalation and the resulting third Liða are shown in pale red.

Notes

¹ It must be stated, however, that the sense of "night" as "eve" is so obscure that it does not rate a separate mention in either the OED, nor in the standard Old English dictionaries (including the new thesaurus by Roberts, Kay, and Grundy). Only Pollington's dictionary/thesaurus gives the equation "eve niht" (Pollington 1993: 53, 231).

² Two Old English references, however, deny this statement by implication. In his dictionary, under Geohel-, Joseph Bosworth quotes twice from The Shrine (1864-1870) by the Rev. Oswald Cockayne. While I lack the means for judging both the age and reliability of these citations, I can state unequivocally that both misconstrue the lunisolar calendar described by Bede. They imply that Yule is a kind of solar month, like January.

Perhaps, in a late context, when the Roman month names became universally established, the native names became misunderstood as quaint alternatives.

On þone forman dæig on geare þæt is on þone ærestan geoheldæig eall cristen folc worþiað cristes acennednesse.

on the first day of the year, that is, on the first day of Yule all christian folke honour Christ's birth (The Shrine, p. 29, l. 26).

On þone eahteðan geohheldæig bið þæs monðes fruma þe mon nemneð ianuarius.

on the eighth day of Yule is the beginning of the month that is called January (The Shrine, p. 47, l. 13).

The last quotation can be accounted for when we realize that the Roman convention designates Christmas Day as "eight days before the Kalends of January." One can then simply reverse the reading to say that January begins on "the eighth day of Yule."

³ The December, 1997, Sky & Telescope, on page 73, ("Software Showcase"), announced that Dr. Ahmed's splendid program has been released in a new version. It is now called "MoonCalc" and is available for download from the internet.

The range of program has now been extended back to A.D. 600, in order, probably, to encompass the Muslim era.

Sources

Baeda (St. Bede)

"XV. De mensibus Anglorum," De temporum ratione, A.D. 725. (See Jones).

Bosworth, Joseph, and Toller, T. Northcote

An Anglo-Saxon dictionary; Oxford: Oxford University Press, 1898, 1983.

Hall, J. R. Clark, with a supplement by Meritt, Herbert D.

A concise Anglo-Saxon dictionary, Fourth Edition; Cambridge: Cambridge University Press, 1960; reprinted at Toronto, University of Toronto Press (in association with the Medieval Academy of America), 1984.

Harrison, Kenneth

"The primitive Anglo-Saxon calendar," Antiquity, Vol. XLVII, 1973, 284-7.

"The moon and the Anglo-Saxon calendar," The framework of Anglo-Saxon history to A.D. 900 (Chapter 1); London: Cambridge University Press, 1976.

Jones, Charles W., ed.

Bedae Venerabilis opera, pars vi, opera didascalica 2, Corpus christianorum series latina, CXXIIIB, Turnhout: Brepols Publishers, 1976, 329-332 (Chapter XV "De mensibus Anglorum," De temporum ratione).

Nilsson, Martin P.

Primitive time-reckoning; Lund: C. W. K. Gleerup, 1920, pp. 292-298.

Parker, Richard Anthony, and Dubberstein, Waldo H.

Babylonian chronology 626 B.C.-A.D. 75; Brown University Studies XIX; Providence: Brown University Press, 1956.

Pollington, Stephen

Wordcraft, concise dictionary & thesaurus, Modern English-Old English; Middlesex, Eng.: Anglo-Saxon Books, 1993.

Tacitus, Publius Cornelius

Germany and its tribes [Germania], The complete works of Tacitus, New York: The Modern Library, ed. by Moses Hadas, 1942; pp. 701-732, trans. by Church, Alfred John, and Brodribb, William Jackson.

Turville-Petre, Edward Oswald Gabriel

Myth and religion of the north; New York: Holt, Rinehart, and Winston, 1964; reprinted at Westport, CT, by Greenwood Press, 1975.

Return to Calendar Index