Constantine's Donation and its relationship to the Acts of Sylvester

This rather complex problem was first exposed by Nicolaus von Kues, Lorenzo Valla and the English bishop Reginald Pecock (in his writing against the Lollards). The whole background story of the donation is dependent on the Acts of Sylvester, which narrates the baptism of Constantine by Sylvester after the victory at the Milvian bridge (cf. Wilhelm Pohlkamp, Textfassung, Literarische Formen und geschichtliche Funktion der römischen Sylvesterakten, in Francia 19/1 (1992), pp.115-196; digitally accessible in the Bavarian State Library). The validity of these actus Sylvestri is in turn assured by the pseudo-Gelasian decretum,  a Gallic index of permissible books from the 6th century (cf Erich von Dobschütz, Decretum Gelasianum, in the series Texte und Untersuchungen zur Altchristlichen Literatur, Bd xxxviim 3m  Leipzig 1911, digitally accessible at the Internet Archive) that eventually made its way into the Gratian judicial corpus.

The core problem was that Eusebius of Cesarea, though later condemned as Arian, had himself been the Church historian of choice of Constantin, as Reginald Pecock pointed out, and had narrated no baptism of Constantine in Rome (neither after the battle nor due to leprosy), but placed that even in Nicomedia toward the end of the Emperor's life.

Valla and the temporal Arrow

Thinking about interpretation of historical events for long will make one tend into the direction of increasing under-determination. But there are a few things that work like a hard pruning shear and that are not really debated. These have to do with some of the dimensions of our cognition, to put it in terms of Kantian categories, namely that temporally separate things only have one direction of dependence; and that the same macroscopic object cannot be at two locations (that share no spatial overlap) at the same point in time. 

That was basically all that Lorenzo Valla used to dismantle the claims behind the forged Donation of Constantine to Pope Sylvester. Valla argued that certain phrases were in use during the reign of Constantine and certain phrases were not. For example, during the time of Constantine, there were no satraps. Since the Donation assumes that there were satraps as a political title and role, the text shows no understanding of the time of Constantine and therefore cannot be from the time of the reign. 

What that means for an interpretation then is that we have a situation where we have run out of alternatives and ruled them out. As in the old Peanuts joke, there are no oceans (now) in Kansas, nor in North Dakota, nor in Colorado, nor in Arizona, nor in Utah, nor in Wyoming. Given that there is a finite number of states, we can eventually iterate over all of them and determine if they have oceans or not. Thus with the writings from the time of Constantine; eventually we can look through all of them and come up with something like word counts for satrap for each of them. And we can then make a positive statement of an absence: there are no documents from the time of Constantine that recognize satrap as a political title.

We of course immediately see the problem. How did we get this set of dated documents? By some method like Valla's, involving terminus ante quem and terminus post quem. Presumably involving the same logic that Valla just used. For clearly our set of documents cannot use the Donatio itself. There is then a petitio principii in the sense that we already need to exclude the document under consideration in order to make the determination.

But that may be overplaying the hand of the other side. Surely it is good cross-validation practice to go through a canon of writings and see how well, if one document is withheld during training, the separated document holds up when looking at all the other ones as the training corpus. Or perhaps one can cluster the documents by the language and find that the Donatio is an outlier along several dimensions. There is a way that this determination can be made that is not completely dependent on begging the question.

Perhaps the example of the satrap is not the best one either, because that term did exist before Constantine and had become part of the historical discourse since Herodotus (Book III, 89-93).

Modeling the under-determination

The most important change that we need to make in talking about historiographical interpretation is that we know what is true. We do not know what is true, what really happened, and similar.

At the same time, we have to insist that some interpretations are false. As Umberto Eco pointed out long ago, if no interpretation is false, then every text is the same text, and there is only one text. The phone book of Vienna from 1973 is the same thing as Lenin's writings. If we cannot reject an interpretation, then there is no way to differentiate them, and it is all the same.

That combination of there being no true one answer but many wrong answers must strike many people as strange because most people know that not-true is false and not-false is true. Due to this symmetry that setup is then confusing.

The first point to make then is that this switch from true to false using negation only works in the case of a complete disjunction. We already said that we want to assume that not all texts are the same text, so we have at least two interpretations. So if the competing interpretations A and B are up for discussion, and we can show not-A, then the disjunction allows us to conclude B. So if two people Alice and Bea are locked in a room with a hard piece of candy, and the candy is gone, and we can prove that Alice did not eat it, then Bea must have eaten it. 

For most historical events, however, this strategy fails because the disjunction of interpretations remains open. As David Hume pointed out in Dialogs concerning Natural Religion, some of these either-or arguments suffer from a lack of alternatives or become far from clear when other alternatives are considered (famously, that the world might be like a plant). The most obvious present day case is the restricted alternative between evolution and intelligent design, where the problems of evolution only can imply an affirmation of intelligent design if there are no competing interpretations.

So for the first measure of approximation, historiography cannot show what the true narrative is, only that there are false narratives because the disjunction of interpretations is not complete. Notice that psychologically, this may still be a problem, because a very large number of competing interpretations is almost an attack on conceptual resources, an interpretative denial-of-service strategy.

The next question then becomes however, how to show that an interpretation is false. Suppose that our hard candy contains a substance that some people have allergic reactions to. Suppose further that Alice has in the past shown allergic reactions. Then we can mount an argument that either Alice would have had an allergic reaction or that Alice did not eat the hard candy. Very quickly the devil shows itself in all the details as the tree of dependence and justification begins to extend downward into the minutiae of cellular biology, chemistry, and medicine of human food allergies. And that is just the natural science side of the problem. There are also other events that are historical in that chain, that have the same disjunction of interpretations intervening between the chemical processes, say, and the conclusions that we wish to draw for adjudicating between A and B. Perhaps Alice's prior episode was misdiagnosed; perhaps the reaction was caused by another allergenic substance; perhaps the evidence for that episode is suspect (Alice's grandmother told her, because Alice was too young at the time to remember herself). 

However to resolve the conundrum of the interpretative process, what should be clear though is that we need to be ready to accommodate a shift in the truth value of our supports at any point in time. Precisely because the supports of the sub-theories might shift with any new piece of evidence; precisely because none of these matters are monotonic in the logical sense of the word---truth maintenance as part of the ongoing process of rejustification in the face of discovery becomes paramount. 

David Reich and the effect of DNA research on Archeology

In his 2018 book, David Reich makes an important point when talking about the origins of the Indo-European language:

By tracing possible migration paths and ruling out others, ancient DNA has ended a decades old stalemate in the controversy regarding the origins of Indo-European languages.  

This is then the contribution that the DNA gene tracing can make. First, it can identify possible scenarios because of the markers. At the same time it also shows some paths to be impossible. 

In the particular case of the Indo-European languages, DNA analysis of ancient skeletons turned out to be lucky; the paths that Reich et al were able to rule out (prune in the search space view of the problem) axed all paths for a specific theory.

Notice that this does not say what happened. Reich and colleagues can say that the most likely remaining hypothesis points to south of the Caucasus, either Iran or Armenia, because the population DNAs there are the most similar to what they ended up being in the European heartlands:

... ancient DNA from people who lived there [i.e. south of the Caucasus Mountains, perhaps in present day Iran or Armenia, RCK] matches what we would expect for a source population both for the Yamnaya and for ancient Anatolians.

But as Reich had argued with respect to Colin Renfrew's agricultural hypothesis of language spread, that the populations who brought the agricultural revolution had brought Indo-European with them as well, 

... theory is always trumped by data ...

by which he means that new data will rule out previous paths. Data is the way to prune the search space.

And this pruning works both way, because it is a recognition function. It recognizes what it can accept (the prediction view) and it therefore knows what to filter. Sometimes these functions are couched positively and sometimes they will be couched negatively and that can muddle up the issue too.