I was skeptical about the new techniques that supposedly let us figure out the past via genetic analysis, so I was excited to find this book. It does nothing to calm my suspicions.
First, the book has its origin in blog entries. The author has a huge bibliography and has kept up with the material. But it's mainly a rehash of what the articles say, taken at face value. That is, there's very little defense of the actual methodology, and a lot of ipse dixit assertions with simply a footnote with a reference to some article that supposedly backs up the assertion. Apparently, the author has no doubts. Also, there's a certainly lack of cohesion in the book. Basically the chapters are simply chronological, starting with the mesolithic and ending with the Vikings. Sometimes there's a feeling that stuff gets stuck where it was because there was a blog entry about it and it gets stuck in the relevant chronological chapter.
As for the argument itself, there are two basic principles. First, the general notion is that genes change through mutation at a more or less constant rate, and where you get the highest piling up of mutations in the present day is where the genes started out, while groups that move out from the center have fewer mutations. The idea that the rate of change is predictable is manifestly false (e.g., "the evolutionary effective mutation rate… generally overestimates ages dramatically" p. 231). So much for that premise. Furthermore, the idea that present-day distributions can tell you anything about what expansions of populations took place in the distant past is on the surface of it ludicrous, and the results are often absurd. For instance, the "haplogroup" J1 is supposedly associated with the spread of agriculture from the Near East, and one reflection of this is the "fact" that it's particularly dense among Palestinians. Wait, what? The thin neck of habitable land connecting Egypt with ancient Syria is supposed to reflect the population there from something like 5000-8000 years ago? Oh, so all those Egyptian, Near Eastern, Persian, Roman, Hellenistic, Arab, Crusader and Ottoman armies that marched back and forth over this small area left no effect? Who could believe such an interpretation?
It used to be that students of the "pre-historic" period (i.e., before the development of written records) tried to associate linguistic/ethnic groups with particular physical remains (mainly pottery). With this method, you'd think you could march back from the historically attested levels to a clear break in the archaeological record, and that break would represent the arrival of the historically attested population. Turns out that wasn't so easy to do (e.g., it wasn't possible to figure out when the Greek-speakers showed up in Greece). Now, we associate historical populations with "haplogroups" and "subclades" and can confidently plot the march of the Celts or Sarmatians across Europe (see, for example, the maps of distributions in Ch. 9 that "show" the spread of the Indo-Europeans via gene distribution or the "march" of the Celts in Ch. 10). But the fallacy of associating modern distributions with ancient movements is shown by Map 82 on p. 180. The distribution of Y-DNA R1b-U152 is associated with the "Iron Age Celts" for basically no better reason than the fact that it supposedly corresponds to their greatest extent, as if this has been left undisturbed for the past 2500 years or so. But it doesn't even really fit. First, the "density" argument indicates that the Celts came from north-western Italy and Corsica (!), which is ridiculous, and we also have the special treat that the invasion of Asia Minor by the Celtic Galatians is still represented by a nice bubble of this sort of DNA in Asia Minor. The only problem is that even if one were willing to imagine that it had sat there undisturbed by all of the population changes there since the third century BC (and there have been a lot!), the unfortunate fact is that that bubble of Y-DNA is not in fact where ancient Galatia was, which is located noticeably to the east. So I guess either the DNA is not in fact a vestige of the Galatians or they moved a bit to the west as a mass. I'm going with the "whatever it represents it ain't the Celts" version.
For another example of the absurdity of this methodology, look at Map 109 on p. 233. This supposedly shows a "Slavic" gene. While it is true that the area covered has a slight similarity to the eastern Slav linguistic group, even this isn't very exact (e.g., the gene covers southern but not northern Poland). But it also includes a lot of non-Slavic territory (e.g., the Hungarians, the Albanians, the Greeks). Indeed, it spills over into eastern Asia Minor. The author is reduced to mentioning the Janissaries (Ottoman troops consisting of children taken from Christian populations in the Balkans and forcibly converting to Moslem soldiers in Constantinople), but if that's the case, why did it spread to only western Asia Minor? Furthermore, the "density" argument would lead you to believe that the Slavs started on the Dalmatian coast of what used to be Yugoslavia and spread from there to the the Balkans *and* the southern stepped of Russia and the Ukraine, which is absurd. If one weren't determined to make the association of gene and ethnic group, a much different picture emerges. Rather than representing the "spread" of the Slavs from the steppes into the Balkans, the more obvious interpretation of the map is the spread of a trait that starts in Dalmatia and spreads out, like the increasingly weaker waves from dropping a rock into water at the edge of a pool, to the east, becoming weaker as it spreads.
One really bizarre characteristic of the methodology is that whereas we get really detailed claims about the exact course of expansion of specific cultures across Europe during the pre-historic period, when it comes to the interesting questions of the movement and arrival of historically attested populations, the genetic data is incapable of explaining anything. When did the Greeks arrive in Greece? Don't know. What of the gap between the Mycenaean and "Dark Ages"? Isn't even mentioned. Where did the Etruscans come from and when did they arrive in Italy? No idea. What of the break between the Bronze Age and Iron Age settlement patterns in Italy and the arrival of the Latins? Not even mentioned. What about the spread of the Germans and (their subgroup) the Vikings? Nope. Turns out there is no genetic distinction among these widespread groups operating over many centuries of time.
One truly astonishing admission is that the present-day population of Hungary is indistinguishable from the surrounding Slavic populations (and presumably the Romanians), whereas of four bodies from the time of the arrival of the non-Indo-European Hungarians more than 1000 years ago, two have a haplogroup (N1c) characteristic of Uralic populations in the east (p. 240). Surely what this means is that according to the underlying methodolgy, the Hungarians do not exist as a linguistic/ethnic group distinct from their neighbors. This is manifestly not true, and this genetic "swamping" of the Hungarians by native populations of eastern Europe shows the absurdity of attempting to correlate present-day distributions with ancient movements. (Be it noted that the supposed ancient movements are at times dependent on so-called aDNA, that is, ancient material, and to the extent that this is the material being used, that's more plausible. But this ancient material is constantly equated with modern distributions, and in any event, I'm suspicious about how much DNA evidence actually remains and the extent to which it's representative. That is, how much aDNA is actually attested archaeologically and is this material really extensive enough to bear the weight put in it, like tracing movements by a century in the distant pre-historic period?)
Because the DNA evidence is (bizarrely!) so incapable of telling the story of population movement when it is known, a lot of the chapters about the Romans, the "Wandering" period, and the Vikings degenerates into potted history. And a lot of this is rather dubious. A Roman military unit called the *ala Frontoniana* is apparently associated with the word "frontier" (no, sorry, it means the unit was commanded by somebody named Fronto; p. 195); the author apparently thinks Constantinople was renamed Byzantium (what a historical blunder! p. 205). Perhaps mistakes like this are in and of themselves not too significant, but they do show the extent to which the historical information that's being correlated with genetic is treated in an amateurish way.
Anyway, if this method can't explain the situation when you do know from historical sources what's going on, there's absolutely no reason to believe the conclusions reached about periods when there is no external control over the information. Fundamentally, it may well be that this genetic information will be able to provide useful new data about the distant past. But it will take a good while, it would seem, until initial enthusiasm is replaced with a calmer and more prudent assessment of the methodological limitations of the technique.