An excellent question. (One for which I have no answer.Just Leo said:Let me be the first to introduce the problem that plagues the bigger engines, vibration. Roy Fedden did not use dynamic balancers on the Mercury, because it didn't need them. The Hercules and Centaurus didn't either, although the Centaurus had some harmonics problems that delayed launch of a satisfactory production example, and led to adoption of metalastic motor mount bushings. Wright did some important work with the R-2600 in dynamic balancers, which aided P&W in resolving torsional vibration problems with the R-2800. If you consider that the R-2800 is a close analogy to the twinned Mercury or Pegasus, what are the chances of such twinned engines avoiding the shakes?
Given the design for a twin is likely to encounter vibration in testing before acceptance, where a "twinned" single may not, this could inhibit "twinned" single-rows.
OTOH, might a "twinned" single see a cancelling of vibration problems? That is, the second row, operating "out of phase" to the first, means the vibration disappears? (It does seem more likely to amplify than cancel...)
Adding a second row also offers opportunities to manipulate vibration issues by changing the firing order, doesn't it? (Or does it?