Gregory Clark (who you should remember from the excellent "A Farewell to Alms") has some new research out looking at the historic rates of social mobility. A good summation of that is here, in the NY Times. Worth reading that but the short version is that social mobility has been very low across societies and history. The current day is not real change from those historical numbers.
It's not quite clogs to clogs in three generations, more like that in three centuries.
However, while I accept the result that he's found and also think it most interesting, I'm not absolutely sure that it's entirely right. The problem, to my mind, comes from the method he's used to work all of this out.
He looks at surnames, marks out certain surnames at certain points in time as being markers of either upper or lower class, then looks at how those markers change over time. Do formerly lower class names move up into the upper classes? Do upper class names decline in social place? The answer is yes but it takes time, those centuries.
Any number of possible explanations come to mind: family networks, genetics, inheritance of goods and opportunity. But the one thing that I worry about here is that the study of surnames is by definition the study of male lineage. And that strikes me as a problem.
For example, Middleton, under Clark's system, is not going to be recorded as one of those names that made the leap from middle class to royalty. Nor, as has happened in the past, those of the meteoric rise of Amy Lyon or Nell Gwynn. Yes, OK, those are exceptions (and I am not placing the former Ms. Middleton in the same class as the two grande horizontales) but I have a feeling that it's not quite that much of an exception.
I have a feeling, but cannot prove and have no idea how you would prove, that social mobility through marriage is something rather more common in women than it is in men. Thus by looking solely at surnames Clark is underestimating that social mobility. Capturing it accurately for men but perhaps not for women: or rather, capturing it accurately for the descendants of men but not for the descendants of women.