This is a most stimulating article, rich in findings about the dynamics of shifts in national attitudes on matters sociopolitical. The analytic separation of conversion and replacement effects produces an interesting counterpoint, and the remarkable dearth of any noteworthy interaction effects allays at least a substantial part of our concern about the underlying indeterminacy of the age/period/cohort paradigm. My only motivation for this postscript is to ask what these data mean on the broadest possible canvas-the evolution of liberal/conservative feelings in the American population. It is possible that they mean exactly what the article implies. There are, however, discordant notes within the data themselves that give one pause. The first discordant note has to do with the fact that the time tracks of these 42 items go every which way from one time period to the next, even in the relatively aggregated assembly of the data being used. The presence of a lot of item movement but very little synchronization of it across items is not a configuration that cries out for some unitary trend descriptor for all 42 items such as "liberal" or "conservative." To be sure, there seems to be somewhat more synchronization if we drop out the two middle periods and look at end point data only (period 1 vs. 4, as in tables 3 and 4). Moreover, the netting out of all this movement, once items are keyed liberal/conservative, does not work out exactly to zero, but goes faintly in a liberal direction. There is nothing wrong with looking at end points in this way: indeed, it is indicated if we are concerned to distill small but true trends of climate from large but short-term blips of weather. And to top matters off, the faintly liberal resultant of all this movement locks in nicely with what is later shown to be a steady cohort replacement pressure in a rather unanimously liberal direction. This is almost a smoking gun.