As Clackson points out, I made a mistake in representing the statistics for the 3rd plural perfect endings. I followed the presentation of Bauer 1933, but Bauer lumped together -ērunt and -ĕrunt in his counts whereas I separated them. This creates the misleading impression that -ĕrunt is not attested in Vergil or Horace, but that is wrong. According to Pye 1963 -erunt is attested at Verg. E. 4.61, G. 2.129, 3.283, Aen. 2.774, 3.48, 3.681 and 10.334. So the figures for Vergil should be corrected to -ērunt 22 vs. -ĕrunt 7. In Horace according to Pye there are three examples of -ĕrunt (Epod. 9.17, S. 1.10.45, Ep. 1.4.7). So the figures for the Satires should be corrected to -ērunt 10 vs. -ĕrunt 1. The figures for Plautus, Terence, and Juvenal are correct.
See D. Pye. 1963. Latin 3rd plural perf. ind. act. endings in verse usage. TPhS:1-27.