The tax rate that top income earners face is not historically anomalous. It was higher during the Clinton years but lower during the Reagan years.
The second sentence is true, but the first sentence is not. (That is, if you interpret "historically" in the above sentence as "with respect to the history of the income tax"). Here's a graph of the top income tax rate since the 16th amendment was ratified (courtesy of Professors Joel Slemrod and Jon Bakija):
So yes, it's true that if you look at current top rate against the backdrop of the Clinton and Reagan years, it isn't especially anomalous. But compared with just about any other point in the 20th century, the current rate is far lower. (Indeed, for most of the 1950s the top rate was above 90%!)
Let me be clear: this is not evidence that the Obama plan is a good idea. It is not evidence that the tax code taken as a whole was more progressive in the past. And it is certainly not evidence that a top income tax rate of 90% is the way to go. It's just evidence that Mankiw's claim is about 90% wrong.