MKK:   Let's talk about the Gallic stuff.   10 minutes left.   I wanted to ask you – you wanted to lead in to a discussion of the Emesa-Treves line, as postulated, with the family of Gregory as an illustration.   I'm defining family, here …

DHK :    … loosely …

MKK:   in a loose sense, as all of those known kin by blood and marriage in every direction.   Do you think that it is worth while, in doing that, to introduce these descents that you have postulated from Gregorius Proculus and Attalus Priscus or would you consider that extraneous to the overall argument?

DHK:    No, I think it is anything but extraneous, because the reason that you postulate the descent from Attalus Priscus is because you find, on the basis of reconstruction of the family, that the Priscus who's known to be of that family had a brother who was an Attalus.   And when you get the two names together, all of a sudden something which, on the basis of one of the two, would not mean much of anything, becomes much, much more convincing, and so if you can find something where you've got that sort of connection, it's really good.

MKK:   This raises another point that I forgot to make yesterday and since the tape is rolling I might as well make it now, although it may not be in aid of any of these points; and that is, one of the basic things that one wants to talk about by way of grundlagen, if I may be [inaudible] to use a pompous term, um, is the overall question – and I talked with you about this either in person or over the phone or both before – of how many families, however defined, and how many people, overall, in any one cross-section in time in Gaul, might we be talking about, how large a population of aristocrats?  

DHK:    Ok, um ...

MKK:   You see the statistics for the probabilities of proper identification as opposed to false or coincidental depend very heavily upon one's conception of how large that prior population in which your samples, your thesis …

DHK:    Well, I have here a map and this has on it the names of all the places that I felt were desirable and necessary for discussion of the Ferreoli.   Therefore, I would say that this is not complete, but it does indicate maybe an order of magnitude, maybe it's half an order of magnitude, because there were quite a lot, a few other places that weren't directly relevant for one reason or another.   So if you say, well, you know, we're dealing with about 30 places and I would suppose that the related families in each of these places, the elite in them, it would probably be on the order of 3 or 4 families for each of these.   So if we double that first figure, 60 and say four families for each, we're talking roughly 240 elite families, is that right?   Ya.   Now if we're dealing with 240 families, I think you fairly often had large families and splitting of families and so this is something that as you go through time, of course you're increasing, and I think we can say that there had been a substantial depopulation in Gaul and my view is that Gaul was more or less like frontier territory and that this is one of the reasons that you had relatively large families rather than the relatively small ones that Settipani would postulate.  

MKK:   Well, there must have been massive depopulation as a result of the events of the 400s and 500s, it's inconceivable that anything else could have happened.   And we know also from Mathisen's work that vast numbers of former aristocrats were driven completely out of Gaul or into poverty or extinction.

DHK:    Yup.

MKK:   So one might be thinking of a total population in any one time in the whole of Gaul – in, let's say, the 500s – of aristocrats to be numbered in the thousands rather than the tens of thousands.

DHK:    Yes.

MKK:   People who could be described as aristocrats.

DHK:    Yes.

MKK:   And the number of adult males in these families would be considerably smaller, of course, in part because – part and parcel of the fact of your conclusion that you reached on other grounds – the families themselves had a good many children at any one point in time – so we would be talking instead about a total population of adult aristocrats, male aristocrats, to be numbered in the hundreds, rather than the thousands.

DHK:    Perhaps a thousand or fifteen hundred.   Even that's probably getting on the large side, but …

MKK:   My feeling is that if you were to take as an example – this of course is a trifle earlier in the period; let's say from the mid-450s to around 480 at the latest – if you were to take the total list of correspondents of Sidonius, who's clearly a very centrally placed individual – not only socially but genealogically – as far as the aristocracy goes, and also an indefatigable letter writer, communicator, and expeditor – if you take his total list of correspondents and of all of the people mentioned either explicitly or by implication as aristocrats therein, I think you probably would have, right there, for that period in time, a not insignificant slice of the whole pie of male aristocrats in this period.

DHK:    And some of the females.

MKK:   And some of the females as well.   And this is important, of course, because connections postulated on the basis of onomastics are inherently statistical in nature – they're saying that this name is so rare – we would not expect, on the basis of its rarity, more than X number of people or X percentage of people to appear, bearing it, in the aristocratic population – the aristocratic population as a whole is so large, therefore, that it is, let us say, unlikely that there were more than 2 or 3 people, at the most, who bore this name at a given time in Gaul among the aristocracy, which is a very different thing than if you're talking about an aristocracy that's ten times as large.  

DHK:    Yes.

MKK:   The big difference between intent to retrieve [surely not what I said – ‘two or three'??] people on the one hand, or 20 or 30 people on the other.   And it's that kind of argument that nobody ever explicitly makes, some people quibble endlessly without ever bothering to attempt to quantify these things as to how likely or unlikely they are.   And what I would like to see somebody do is, sit down and say ok, let's see exactly how likely, is it reasonable to assume this is being …

DHK:    Yes.

MKK:   My tendency is to estimate everything quantitatively on the basis of what information I have.   I once sat on a 4th-of-July night across the river on one of the roofs of MIT, staring at the Prudential Center, trying to estimate how many hundreds of thousands of tons it would weigh on the basis of its presumptive dimensions, the total volume of concrete and steel and how much I would estimate concrete and steel to weigh, as well as various other factors, and I came to the conclusion that it probably weighed on the order of 800,000 tons.

DHK:    Did you ever check it?

MKK:   No, I didn't, I didn't [laughs].   Of course, there was a large, it was a large spread around that – the point is, people don't usually estimate things at all, and they have no real concept, therefore, of relative magnitudes.

DHK:    And most people – historians – dislike statistics, although some of them are getting to where they like them, for certain purposes.   But, I don't know – I think extremely few genealogists have used statistics in any way.

MKK:   And you can't think accurately or reasonably, I should say, about these issues, and come to a reasonably likely conclusion, unless you think in those terms.  

DHK:    That's right.

MKK:   There was a term used in a Reader's Digest article when I was a little boy, it was called something like "Kentucky Windage " – basically, aiming a gun, firing it, and basically taking into consideration, where you estimated wind speed and direction, how many feet per thousand do you think you'd probably deflect the bullet, how much the bullet weighed, how far away the target was, various other considerations …

DHK:    And you do it this way, boom …

MKK:   Ya, and the essential point was, that even if you are wrong, to some extent, in each one of your estimators, you should take all of them into consideration – assuming that the errors average out, you are a lot better taking all of those estimated magnitudes into consideration than you would be if you didn't.   And the same thing applies essentially to genealogy as well.   ––Um, I also mentioned earlier, another point that applies to this question of how large the aristocracy was, and something I mentioned before is, take into consideration the very easily quantifiable consideration of how many ancestors you have, each generation back, and estimating an average percentage of inbreeding on the basis of similar aristocracies in the later periods.    I think a very good parallel would probably be the English aristocracy of the 1700s.  

DHK:    Well no, I think there's also, there, the – you can cite you know – if Cleopatra left permanent issue, we are descended from Cleopatra in probably millions of different ways.

MKK:   Precisely so.

DHK:    And if what we are trying to do is show an example, it's not ...

MKK:   Exactly.   And this is another point that has to be made as well, though I'm not quite sure what the best way to make it is.   The argument that has been mounted against descents of this sort is to say, well, this particular example you have given, for this reason and the other reason and yet another reason, is shaky and unlikely, but my reply is – this is basic in the hypothesis – you have to take into consideration, before anything else, the prior probability of some such descent being true.   That completely alters, and should alter, how probable or improbable you see any particular postulated working-out of this descent to be.   And along those lines, I suppose that we can take an aristocrat in Gaul who lives in the year 600 – we know he's identified as an aristocrat, we know also, with tolerable certainty, on the basis of our knowledge of how the social classes intermix in that period – we know that the majority of the aristocrats at that point, at least those of Gallo-Roman origin in Gaul, had primarily aristocratic ancestry; there was much less vertical mixing in the classes in that period than in many other periods.

DHK:    I think there was probably a great deal of vertical mixing but it didn't go up, it went down.  

MKK:   Exactly.   So if you're looking at a particular person who was already an aristocrat, the majority of his ancestors

DHK:    Would be aristocrats …

MKK:   In most cases would also be aristocrats.

DHK:    But if you are looking at a slave or freemen or a lower rank individual, it's quite probable that somewhere along the line, one of his female ancestors had been used in some way by ...

MKK:   A parallel, a somewhat parallel example being afforded, of course, by the average New Englander today.   I come from complete – I mean, from all appearances, as far back as I knew from my own experience – which is true of most of the clientele of this library – I came from totally plebeian ancestors, what Gary would uncharitably refer to as “trash ancestors;” I'm the Northern end of what you'd call poor white – or even off-white – trash, and yet the probability is, even for anyone of my own social class in immediate origin, if you trace back [his entire family tree], it has a great deal of colonial New England ancestry.   One or two or three of those people are likely to have royal ancestry because they come from the gentry ...

DHK:    It's much more ...

MKK:   I mean provably so, but I would agree with you on the reality – it would probably be more than that – but even provably, the average person who comes here, no matter what immediate circumstances they may come from, if they have a lot of New England ancestry, it's quite easy for them to find a small handful of ancestors, early on, who are of the English gentry.   Once again, if you have somebody who is of the English gentry around the year 1600, what is the probability that he or she was actually a descendant of Edward III?

DHK:    No, it's about 1/4 th.  

MKK:   It's very good.   And for Edward I, it's probably three times that.

DHK:    For Edward I, I think you can probably call Edward I the father of his country.  

MKK:   But that kind of example gives you some idea of how likely it is that you are descended at a remove of 10 or 15 generations from any given aristocrat of that earlier period.   If the ancestry is aristocratic and if you know the size of the aristocracy from which you were trying to draw your ancestors –

DHK:    But there's another point there, which is that the royalty and aristocracy and gentry in England were very sharply differentiated and the royalty – there were periods where you had a substantial number of marriages in the aristocracy, but there were a lot of others where you didn't.    The, ah, I think that the odds shift very remarkably in some of these, and the probability of a decent from an aristocrat with royal, well, a descent from Edward I is very high, but probably it is not the same as the probability for a descent from a particular nobleman of the same period, which, and of course, in Scotland, the situation is extremely different.   And, I didn't realize how different it was until I got to working on the Adairs, who are one of my wife's ancestors, and the particular pedigree given is entirely plausible but there is one person who was supposedly born in Ireland to particular parents, married in Pennsylvania to a particular wife and went to North Carolina where he died.   And there is no record of him in Ireland that I've found, there is no record of him in Pennsylvania that I have found, and there is no record of him in North Carolina that I have found.   As a connecting link, he is pretty much a phantom at the present time.   I don't think he was an invention, but I'm not even 100% sure of that.   But everything about the relationships are plausible.   So, if it was an invention, it was a good one.   So they go back to the Edmondsons, and there's a pedigree, the Edmundsons went for about 10 generations, they remained simple knights, none of them was ever higher status than a knight, and nonetheless, 3 of them intermarried into the ...[end of side 1 of tape]

MKK:   I wanted to give an example of what I was talking about – suppose we have a Gallo-Roman aristocrat living in the year 600, we want to consider the probability that he was descended, at random, from an ‘average' aristocrat, however defined, of the period X number of generations before we could have 1024 separate ancestors theoretically..

DHK:    Say around 100 A.D.   Around 100 A.D. for the Gallo Roman –

MKK:   From 600?

DHK:    No, it would only be around, back around 350 or 450.  

MKK:   So, let's assume, however, that, in reality, at that generational level because of the factor of inbreeding, he might only have a half or a third of that number of separate, discrete ancestors,

DHK:    I would assume a substantially higher percentage.  

MKK:   I would think it probably would be, but I'm trying to be, trying to be fair here, make his pool of ancestors no larger than need be at that point, if we could assume that in that period, let's assume … that we could come up with, that in that period he had 300 or so male ancestors – leaving the women out of it for a moment because it's easier to find the men – if the total size of the aristocracy at that remove in time amounts to, oh, let us say, 5-10 thousand people, and we have 1000 to 1500 aristocratic males, the probability of descending from any one of them is between 20 and 30%.

DHK:    Um, hum.

MKK:   That in itself is not, is not a negligible quantity.   And you see, that is the kind of point that you can make, it is a legitimate one to make, these are reasonable assumptions, you see, this is the ineluctable consequence of simple statistical analysis of the demographics involved.   And nobody does that.   We might as well do that ourselves.   I just don't see it –

DHK:    I've mentioned it once or twice, but who else?

MKK:   Exactly.   It's a kind of reasoning that cries out to be applied to these problems, in particular in the dark ages and classical and ancient genealogy, because there you need to lay more weight upon that in order to get some sense of how likely or unlikely as a priori probabilities any postulated descent might be.   Well anyway, I hope it's been worthwhile talking to you.   I shall see you tomorrow afternoon if you are going to be here.

DHK:    I will be here tomorrow afternoon.    I may not be here Monday morning.