Is it rare to elide the final vowel if it is long?

Question

In the Ars Poetica we find the line:

posse linenda cedro et levi servanda cupresso

which pedecerto scans as

Pṓssĕ lĭnḗndă cĕdro‿ḗt lēuī́ sēruā́ndă cŭprḗsso

In C.O. Brink's commentary on Horace we find the comment:

cedro et leui: an isolated case in the Ars of a final long vowel elided before short, according to A. Michaelis, of. cit. (above 63-9 n.), 428. This over-simplifies a complex phenomenon, cf. 137, 330, 419, 427, not to mention ‘prodelision’ before est

Other than prodelision of 'est', how does this oversimplify things? Is this rule true and are there other exceptions to it?

Answer 1

According to Sturtevant:

It is often stated that long vowels suffer elision but rarely before a short. We find, on the contrary, that more than four-fifths of all elided long vowels and diphthongs stand before short vowels! Since only about one-fourteenth of them stand before short vowels in open syllables, it would be quite correct to say that long vowels suffer elision but rarely before short syllables. There is, however, no great difference between the treatment of long and short vowels, and so a less misleading statement would be that elision is most frequent before short vowels of syllables long by position.

This article is from 1916, so it may well have been superseded by later scholarship. But I'm assuming his numbers are correct, indicating that quantitatively:

• Long and short vowels are elided about the same amount
• Elision is less common before a light syllable (a short vowel with no coda)

Answer 2

Elision of final long vowels is not that rare, at least not as a general matter. "Elision and Hiatus in Latin Prose and Verse", by E. H. Sturtevant, says that it's more accurate to say that "elision is most frequent before short vowels of syllables long by position" (page 43; The Classical Journal Vol. 12, No. 1 (Oct., 1916), pp. 34-43 (10 pages) https://www.jstor.org/stable/3287750).

Since "et" comes before "levi", it would constitute a syllable long "by position" and containing a short vowel, so this is in fact the context where Sturtevant says elided final long vowels are most frequently found. If Sturtevant's description is accurate, it doesn't seem to be an exception at all to the general tendencies (although Sturtevant is not talking specifically about Horace).

In this case, there's the additional complication that some authors have proposed that "et" could be subject to prodelision, like es or est. As far as I know, the hypothesis of prodelision before et is far less accepted than in forms of sum, and I'm not sure how plausible it is. I've only seen it raised as a possibility in "Hypermeter and elision in Virgil", by G.P. Goold (2002, in Vertis in usum: Studies in Honor of Edward Courtney); Goold (page 89) cites F. W. Shipley (1924) and Soubiran as supporting "obstipui, steteruntque comae't vox faucibus haesit" instead of "obstipui, steteruntque com'et vox faucibus haesit" for Virgil Aeneid 2.774.