There is accumulating evidence that the germline mutation rate is not a clocklike constant, but is better regarded as an evolving physiological trait. This rate places a fundamental limit on how quickly populations can evolve, and it is not well understood how quickly this rate is itself altered by evolutionary forces like selection and genetic drift. Here, I look at dynamics of mutation rate change during great ape evolution by summarizing evidence of changes in the mutational spectrum: the relative mutability of different genomic regions as a function of local sequence context. These changes can be detected by looking at the mutational spectrum as a function of allele frequency, a proxy for allele age. I find evidence that several mutation spectrum changes occurred relatively early in primate history and fixed in certain lineages; for example, an increase in the rate of A→T transversions appears to have fixed in the common ancestor of humans and chimpanzees. More surprisingly, I find instances in multiple great ape species where the mutational spectrum appears to have changed and then reverted back to its original state, as though a mutational process acted for a short period of time and then subsided. Such a pulse appears to have affected Europeans between 10,000 and 20,000 years ago, producing a previously observed excess of transitions in the context TCC→TTC. The relative abundance of this mutation is greatest among SNPs that segregate around 1% frequency in Europeans; the mutational spectrum of rarer European variants is closer to that of Asian and African variants. A few other mutation types appear to be minor components of the same pulse, peaking in relative abundance around 1% frequency in Europeans, most interestingly the dinucleotide mutations GC→AA and TC→AA, which are putatively associated with error-prone DNA replication by polymerase ζ.