It used to be thought that, just as word-initialfl… and fr… behaved likepl…, pr…, tr…, etc., in not producing a long syllable when following a word-final short vowel, just so word-internal …fl… and …fr… allowed both the short and, except for the pre-classical scenic poets, the long scansion. It was implied that these clusters oscillated with the same degree of freedom which is the well-known characteristic of the stop-and-liquid clusters. The difficulty is, of course, that evidence can be no more than minimal since in truly Latin (i.e. neither dialectal nor foreign) material f occurs only at the beginning of words or after a compounding seam. In fact, the argument, explicit or implicit, has turned on Horace, Sat. 1.2.98: custodes lectica ceniflones parasitae; Horace, Sat. 2.2.131: ilium aut nequities aut uafri inscitia iuris; Ov. Ars 3.332: cuiue pater uafri luditur arte Getae; Martial 6.64.26: stigmata nec uafra delebit Cinnamus arte; 12.66.3: arte sed emptorem uafra corrumpis Amoene; Phaedrus 2.6.14: inducta uafris (cj. Festa) aquila monitis paruit; Silius 8.566: et quos aut Rufrae (cj. Heinse) quos aut Aesernia quosue, and Martial 4.71.1: quaero diu totam Safroni Rufe per urbem