On pg. 128, n. 22 I wrote that the first instance of a shortened vowel before final -r appears in Lucilius, but this is not correct. As Ben Fortson points out, Enn. Ann. 396 Sk has a case of shortened o before final -r in the word sūdor. The line is transmitted as totum sudor habet corpus multumque laborat and the o must scan as the first short of the weak half of the second foot. This shortened vowel is an outlier in Ennius who otherwise preserves original long vowels before final -r, except when other processes (Iambic or Cretic shortening) apply. For this reason Lindsay (Early Latin Verse 1922:125) wanted to transpose sudor and corpus so that the line would read totum corpus habet sudor multumque laborat, but this does not seem justified. So a more accurate statement would be that the first instance of shortening of a long vowel before final -r occurs in Ennius, but that the shortened scansion becomes regular by Lucilius' time. Lucilius has a number of instance of short scansions and no certain instance of the long scansion.