You might remember I wrote a little post about a Carolingian list of eclipses from 812 and a ‘mystery eclipse’. It was a strange list because lots of the dates were out a little bit because they had been adjusted to fit with standard lunar tables. It was also surprising because some eclipses mentioned definitely didn’t happen – they had, somehow, been calculated (if badly). But how?

Apart from the list itself – which survives in multiple copies – our principal source for discussion of the eclipses is a letter by the Irishman Dungal of St Denis to Big Charles, the infamous mass-murdering poetry-loving Frankish emperor. The letter reveals that knowledge of the eclipses at court had come partly from reading books and partly from discussion with ‘a certain archbishop of Constantinople’, which is presumably Nikephoros (patriarch 806-15). Big Charles’s concern, it seemed, was that they had heard about two eclipses in the same year, one on 7 June 810 and one on 30 November 810. What he wanted to know was this: was that possible?

Dungal knew that, technically, it was. Pliny, for one, had stated Hipparchus’s observation that the shortest distance between solar eclipses was seven months. Ptolemy, incidentally, had considered five, six or seven synoptic months between eclipses, depending on the time of year and also on the latitude of the observer. More on that later. Dungal did not know that, of course, because Ptolemy was not available in Carolingian libraries, even in the even-by-then lost translation from the Greek of the digested version by Boethius. But Dungal could see that it was seven months between 7 June and 30 November so it was okay.

(Hang on, that’s six months. Maybe don’t forget that 30 November on the Roman calendar is 2nd kalends of December, and June, July, August, September, October, November, December is seven months. Or maybe just don’t think about it too much).

So, Dungal didn’t really know much that would help him beyond a vaguely expressed principle. From where, then, did this ghost eclipse come?

It is worth considering the reference to Nikephoros and the Greeks here more fully, which I admit I hadn’t done before. A delegation had indeed arrived in Aachen that October to talk about peace and no doubt some other interesting things. Now the Greeks, of course, had not lost Ptolemy. Indeed, they also had the so-called Handy Tables, the digested version that offered practical simplifications of some of the ideas in the Almagest, plus they had commentaries. The interesting thing about the 7 June eclipse is that, if you crunch the numbers in the right table (for the 1134th year of Philip, 4th day of Tybi, after the 15th hour from midday, by the way), you would end up with a suggestion that there might have been a conjunction of the sun and moon just before sunrise in the Mediterranean world on 6 June.[1] That would mean that there was no eclipse visible, of course. However, if you understood enough astronomy to be playing this game, you also knew that days were longer in summer further north. Maybe – just maybe – if you were in or from Constantinople and knew someone further north, say in Aachen, you might ask them if they had seen an eclipse early in the morning in question.

Your suspicions might also be heightened because you would know that an eclipse on 30 November was highly probable on the basis of Ptolemy. It was six synodic months (177 days) after the calculated June eclipse, plus calculations using the Handy Tables would reveal it was almost a dead cert, and of course you might even then just see it so you would have a definite point from which to consider other hypothetical eclipses.

Indeed, the Carolingian list of eclipses then adds another non-existent eclipse that could potentially be explained by this scenario: one on 27 April 811, at the first hour of the day (so again, before sunrise in most of the Greek world). This is one that raised concerns when I first posted about the eclipses because there was a real eclipse a month later that could only be seen in the southern hemisphere. Some people – fairly enough – were not entirely convinced that ‘acceptable margins of error’ and the fact that people were aware that some luni-solar conjunctions took place below the equator quite cut it. But what would happen if we went via ninth-century Greek applications of Ptolemy?

Handy Tables in a copy from 813-20 (Leiden BPG 78).

On first look, it isn’t entirely straightforward. If you crunch the relevant numbers in the Handy Tables, you would see from the centre of the epicycle that there would be a conjunction, but that the ‘latitude argument’ would not even be close to producing an eclipse. The Handy Tables were not, however, designed to take into account lines of latitude much north of the Black Sea. In the Almagest, meanwhile, Ptolemy had suggested that one might possibly get two eclipses five synodic months apart (or 148 days) if you went far enough north. Hypothesising an eclipse on 27 April 811 after a visible one on 30 November 810 would exactly fit such speculation. There is no known early medieval Latin model that would produce the same result. All that would have to happen then is for a sufficiently interested Greek scholar, engaged with the Carolingian world, to ask if the hypothetical eclipse had been observed that far north. Which it hadn’t, but that wouldn’t stop such an inquiry slipping into our list, given that it was not a list of observed eclipses.

That eclipses might be discussed in the context of Carolingian-Byzantine diplomacy should not be surprising. Big Charles and his entourage were famously interested in astronomy and they knew that the Greeks had lots to say on the subject. There had definitely been other conversations about computus and chronology too. And of course we know that there was a lot of talk about power, politics, and religion, as there always was. Nikephoros could not avoid all the chat, even at distance, and was even prompted by the 810 delegation to write a letter to Pope Leo III declaring the orthodoxy of his beliefs – a fascinating and surprisingly little discussed document that includes one of the earliest references to Charles’ coronation. The letter mentions nothing about computistical interests, alas. His interests in the area are at least demonstrated by a little chronographical work, which (coincidentally) itself came to be translated into Latin as part of a three-volume work undertaken by Anastasius Bibliothecarius in the mid ninth century. A query or two about eclipses hardly seems out of place.

To summarise all this then: As far we know, there was no way Dungal or anyone else in the Frankish Empire could predict eclipses with any sophistication, certainly not as well as the Greeks with their Ptolemaic astronomy. There are, however, two ghost eclipses in Frankish sources that could be accounted for if they had been raised as hypothetical eclipses by a Greek scholar, using Ptolemy or the Handy Tables, and interested to see the effects of latitude on their calculations. Greek knowledge in the context of diplomatic exchanges potentially explains some problems in our sources.

One final note: Greek ‘sciences’ might not have been well-known or influential in the early medieval Latin world. Exchanges and conversations still happened.

[1] Four copies of the Royal Frankish Annals by the way give the date as 6 June not 7 June. Never rule out copying errors.