Mitchell, E. (2006). Calendrics and Astronomy in the Mozarabic Chronicle. PHILICA.COM Article number 57.
Calendrics and Astronomy in the Mozarabic Chronicle

Ethan Mitchellconfirmed userThis person has donated to Philica (Independent Researcher)

Published in

The author argues from astronomical observations and inter-calendric comparisons that the “era Hispanica” used by the Visigoths was calculated from September, not January.

Calendrics and Astronomy in the Mozarabic Chronicle


“The booty don’t stop.”




The anonymous Mozarabic work known as The Chronicle of 754, or the Mozarabic Chronicle, remains the most important primary source for Iberian history in the 8th century.  While later Arabic works are more detailed, and often conflict with the Chronicle, the sequence and timing of events in the Chronicle has been strongly defended, to the extent of invalidating later accounts.{1}   


The author of the Chronicle, who I will call M, was fascinated by calendrics and astronomy.  M makes use of six different calendars in the Chronicle, including the Roman Imperial cycle, the Muslim Caliphal Cycle, the Hispanic Regnal Cycle, hijra dating, and a count of elapsed years since the creation of the world.  His base calendar, however, is the little-studied “era Hispanica” notation that appears in several Visigothic texts.


While the term ‘era’ itself is likely of Iberian origin{2}, by the late Roman empire, era notations were being used all across the Mediterranean.  The era Hispanica seems to have appeared in the 5th century, and its base date is 38 BC.  The reason for this choice of date is not known, and the various theories put forward by ancient authors are not especially compelling.  It has generally been stated that the New Year’s Day for the era Hispanica was January 1st {3}  Though I have not had the opportunity to look through the Corpus Scriptorum Muzarabicorum, I have never come across a definitive statement of New Year’s in Visigothic literature.  I believe that the calendrics and astronomy in the Chronicle of 754 offer strong evidence that the era Hispanica used Sep. 1st as New Year’s Day.


The first of September is the nominal anniversary of the Battle of Actium (31 BC), which was widely used as a calendric reference point in the Roman world.  The Byzantine indictions are dated from September 1st, and Byzantine colonists played a large role in the re-development of South-Eastern Spain under the Visigoths{4}.  If literate Iberians adopted a Byzantine calendric format, it is perhaps not too much to suggest that the set-date of 38 BC could be an early mis-count from Actium itself.

This is the problem I propose to consider in reviewing the Chronicle.  Throughout this paper, I have used Wolf’s translation of the text, and Wolseley Haig’s correspondences for hijra dates (which are in turn based on the Kuwaiti algorithm).  All Christian dates are Julian, unless otherwise mentioned.



The e758 Eclipse


Several astronomical events are mentioned in the chronicle, two of which are dated and described with considerable precision.  The first of these is a solar eclipse mentioned in section 65, and reportedly occurring in Era 758, thirty-four years before M was writing.  It was described as follows:


 “At that time, in the beginning of the era 758, in the one hundredth year of the Arabs, an eclipse of the sun, lasting from the seventh to the ninth hour of the day, was observed in Spain, a number of witnesses seeing stars appear.  Many contend that the eclipse appeared at the time of Al-Hurr’s successor, As-Samh.”


Typically, M dates the eclipse in three different systems, which do not precisely align, and we do not know where the errors might lie.  But unlike most of the other events he makes reference to, we can pinpoint an eclipse with transhistorical accuracy.


There were 94 solar eclipses between 701 and 740 AD.  Espenak’s NASA projections are available for the durations, magnitudes, and angles of these events.  In only eleven cases did the locus at the moment of the greatest eclipse occur between 30 and 50 degrees north.  Of these, there are only three cases where the locus occurs between 40 degrees east or west.  No other eclipses in this period would have been significant from the vantage point of the Iberian peninsula.


The three eclipses in question occur in 713, 718, and 720.  All three were total, and reached greatest eclipse over the North Atlantic, after transiting West across Spain.  All three also occurred in the afternoon: respectively at 13:24, 13:57, and 15:51 U.T.  In classical horology, the day is divided into twelve hours of variable length, starting at sunrise and ending at sunset.  For all three of the dates in question, “the seventh to the ninth hour” would be early afternoon, and consistent with the timing of the eclipses.


However, the eclipse in 720 was the largest of the three, with a 242-km wide path of totality.  It was also the closest to Cordoba, with the locus of greatest eclipse occurring only 1830 km away, as opposed to 2870 km and 2570 km, respectively.  These two facts would likely have made it the most noteworthy of the three, and so we might assume that if the author of the Chronicle is only going to mention one eclipse, it would be this one.


The 720 eclipse occurred on October 6th.  To make this “the beginning of the era 758” would certainly suggest a New Year’s Day in September, not January.  If that hypothesis is correct, then the calendar offset ought to be 37 years 4 months, not 38 years.  Furthermore, if that hypothesis is correct, we can corroborate that M has accurately described the date and time of this eclipse.


Given those provisional assumptions, we should take the second event mentioned seriously.  It is much harder to classify.



The e788 Anomaly


“In the Sixth Year of Constantine’s rule, in the era 788, on the nones of April, Sunday, during the first, second, and part of the third hours, with all the citizens of Cordoba watching, three suns, shining in wondrous manner and fading into a crescent of emerald fire, were seen.”


We should note that in this case, M is a describing an event only four years after the fact, and quite possibly was an eyewitness.  The strange phrase “emerald fire” initially suggests an aurora borealis.  The idea might gain some weight from the fact that the event occurred in April, when the earth’s magnetic axis is aligned with the Parker Belts, and auroras are more common.


Currently, an aurora at such a great distance from the magnetic pole would be very exceptional.  Feldshtein has argued that there is no literary evidence that auroras were visible from Scandanavia prior to the thirteenth century{5}.  However, Isidore of Seville records what appears to be an aurora in 451, as a portent of the Battle of the Catalaunian Fields.  (Isidore’s History of the Goths, 26) This description is in turn based on Hydatius (149).  The magnetic poles move, and it is possible that auroras were visible from Spain in this era.


A more serious problem with the aurora theory is in the details of the anomaly.  Few auroras, even from directly under a “corona” pattern, could be described as a “triple sun,” especially during and after dawn, when the viewer is facing the Eastern horizon, not the zenith.  Moreover, while there have been exceptionally bright auroras—such as the one on September 2nd, 1859 (Gregorian), seen throughout the Northern hemisphere and apparently luminous enough to cast shadows and allow newsprint to be read—it is hard to imagine that any such aurora could remain visible three hours after dawn.  Moreover, if such an aurora were visible, it should have been noted by astronomers all around the world, as the 1006 supernova was.  Yet I cannot find any other references to an anomaly in this period.


Laying aside M’s description of “emerald fire,” then, we turn to the other details: a tripled sun, at and just after sunrise.  This closely fits the pattern of a “fata morgana” mirage caused by super-refraction.  Such mirages occur when still, thermostatic air refracts light rays around the horizon without (much) distortion.  Doubled or tripled images are not uncommon, and they can be stable enough to last for several hours.  Again, the occurrence of the anomaly near the horizon makes more sense.  The thermal inversions necessary to create Fata Morganas are typically found over ice, or cold water.  They can, however, be seen over deserts; I have seen one in the Bolivian Altiplano.  Whatever the exact cause of the anomaly, it seems likely that it was a localized, low-atmosphere phenomenon.  This suggests, in turn, that M was either present in Cordoba, or spent enough time there to give weight to the local anecdotes.


The Calendars


There are 79 different direct date comparisons in the Chronicle.  Many of these are of little value, however.  Where the events being dated are far removed from M, either temporally or geographically, he makes errors and has inconsistencies of several years, so there is no hope in interrogating the data for a more precise result.

Like earlier authors, M uses the succession of Roman emperors as a base-line calendar, counting the years of each imperium, and dating each of them back to the creation of the world.  An entire block of these, comprising five dates, are offset by 3-5 years.  Since it is almost certain that M was working with a list of the Roman imperia, we should interpret this as a single error, not a series of errors.  Indeed, in the closing passage of the Chronicle, M seems to identify this offset, but leaves it is a problem for the reader:


“If you wish, you may subtract four of these years in accordance with certain historians who diligently affirm this, computing the fifty-sixth year of the reign of Octavian to have expired in the 5,210th year of the world, and asserting that Christ was born in the 42nd year of the emperor Octavian.”


This is the solution to the error that M had been making on the earlier Roman dates, but he does not fix it.  In a moment of perspective quite incredible for the chronologically obsessed, he philosophizes:


“With regard to this particular discrepancy in years the opinions are not very different from one another—we have added four years in accordance with the many who contend that Christ was born in the year 5200, lest we stray too far from the path about which so many distinguished men agree.  Over such a large period of time, the addition or subtraction of four years would not seem to be damning….”


But you and I are not so wise as that, and so we pass on to the rest of M’s dates to see if we can corroborate the proposed offset of four months.


Hijra Overlaps


There are 29 identifications of era with hijra in relation to events in Spain, which we might expect M to have a more accurate grasp of than he did Rome or Arabia.  Three of these reference the actual beginning or ending of the Muslim year; sections 88, 90, and 91.  These sections let us compare a known lunar date to the two possible frames for the era.  Sections 88 and 91 do not shed much light on the problem, since either frame would contain their respective dates.  In section 90, however, M offers even more precision:


At this time, in the beginning of the era 783, the first year of Constantine, with the one hundred twenty-seventh year of the Arabs about to expire, everyone in the land learned that Yazid Walid had died a natural death…


The qualifiers make this passage invaluable to addressing our question.  Depending on the algorithm used, hijra 127 ended on October 1st or 2nd of Julian 745.   If this was the beginning of era 783, the New Year’s Day must have occurred in the autumn.


From this tantalizing piece of evidence, we turn to the remaining 26 era-hijra comparisons.  In each case, we have a known lunar year to compare to the two possible frameworks of solar years.  Since we do not know the precise dates of the 26 events in question, we can work probabilistically from the amount of overlap between a given lunar year and the putative solar years.


Unfortunately, this method is of no use for the dates given in the Chronicle.  The average overlap between a given hijra year and a January-based era is 101.6 days.  For a September-based era, the average overlap is 102.4 days—exactly equivalent.  (The numbers are low because in several cases M has erroneously tried to identify two years that do not overlap at all.)


However, if we simply look at the possibility of overlap, the picture is rather different.  Using a September-based era, 16 of the 26 events can be placed into the hijra year that M mentions.  Using a January-based era, only 12 of the 26 can be placed correctly.  If we leave out the cases where neither frame works, there are five events that can only fit into a September-era, and one event that can only fit into a January-era.


This is a bit past the limit of our resolution, for we are now trying to compare the uncertainty of M’s errors to our uncertainty about the era notation.  Nevertheless, in the dating of the e788 eclipse and the death of Yazid Walid, there seems to be strong evidence that M is using a Byzantine New Year’s, and there is nothing in the remaining dates to cause us to reject that impression. 




On the balance of the evidence, it appears that the author of the Mozarabic Chronicle used a New Year’s day of September 1st, in the Byzantine fashion.  He likely lived in Cordoba, and seems to have been quite an accurate and meticulous observer of dates, although he does make some errors.


The only dates in the Chronicle which are of much interest to non-Iberianists are the date of the Muslim conquest, and the date of the Battle of Poitiers/Tours. 

Musa’s entry into Spain is described as “In the era 749, in [Justinian’s] fourth year as emperor, the ninety-second of the Arabs, and the fifth of Walid.”  If this statement is accurate, and if M is indeed using a Byzantine New Year’s, then it provides a very narrow window for this crucial date.  M would have Musa enter Spain between Sep. 1st, and Oct. 18th, 711—which is quite in keeping with the ancient practice of raiding during the harvest. 


In sections 79-81, M places the Battle of Poitiers/Tours between era 769 and era 772.  Thus, the window given is Sep. 1st, 731, to Aug. 30, 735.  This minor shift conforms to the arguments of recent historians that the battle took place rather later than once thought.



New Haven, Vermont, November, e2044



Notes and Bibliography

{1}Collins, 1989, pp. 26-32

{2}Blackburn and Holford-Stevens, p. 765

{3} Ibid, p. 767)

{4}Blackburn and Holford-Stevens, 766, 771

{5}Feldshtein (1963).  I have not personally seen this text.




Blackburn, Bonnie and Holford-Stevens, Leofranc (1999) The Oxford Companion to the Year.  Oxford University Press, Cambridge. 


Espenak, Fred (2006)  Six Millennium Catalog of Solar Eclipse Paths.  NASA., accessed November 11, 2006.


Feldshtein, Y. (1963). "Some problems concerning the morphology of auroras and magnetic disturbances at high latitudes," Geomagnetism and Aeronomy, 3.


Haig, Lt.-Colonel Sir Wolseley (1932) Comparative Table of Muhammadan and Christian Dates.  Luzac, London.


Wolf, Kenneth Baxter, (2000) Conquerors and Chroniclers of Early Medieval Spain.  Liverpool University Press, Liverpool.



Information about this Article
This Article has not yet been peer-reviewed

Published on Wednesday 22nd November, 2006 at 14:02:31.

Creative Commons License
This work is licensed under a Creative Commons Attribution 2.5 License.
The full citation for this Article is:
Mitchell, E. (2006). Calendrics and Astronomy in the Mozarabic Chronicle. PHILICA.COM Article number 57.

Website copyright © 2006-07 Philica; authors retain the rights to their work under this Creative Commons License and reviews are copyleft under the GNU free documentation license.
Using this site indicates acceptance of our Terms and Conditions.

This page was generated in 0.0230 seconds.