A comparison of the facts described above for Younger Avestan with those uncovered in the discussion of Gathic Avestan reveals that tmesis in Avestan was the result of either (1) a fronting of the preverb or (2) the application of rules governing the placement of the clitics (especiallycā andvā). The fronting rule allowed movement of the preverb to a limited set of positions, normally initial position in the sentence, second position if initial position was taken up by an element in COMP, and, in metrical texts, to a major metrical boundary. Thus the relationship between a preverb and its verb was easily recoverable by the native speaker.This close relationship between tmesis and fronting is likely to be very old. The majority of hte Vedic and Greek examples can be analyzed as 'preverb-fronting',15 compare also Watkins (1963) for Celtic. It seems likely that the fronting of preverbs at least originally implied some emphasis on the semantics of the preverb, such emphasis being the normal function of fronting in the Proto-Indo-European clause. Tmesis without demonstrable preverb fronting is not found in the Avestan corpus, nor in Vedic prose, but it is not sufficiently rare in the Rigveda (given our current understanding of Rigvedic clause structure) to allow us to say that tmesis and preverb fronting were one and the same process for that language. The Avestan evidence surveyed here will thus prove crucial to the development of a comprehensive account of the diachronic development of tmesis in Indo-Iranian, and, therefore, in the Indo-European language family as a whole. A comparison of the facts described above for Younger Avestan with those uncovered in the discussion of Gathic Avestan reveals that tmesis in Avestan was the result of either (1) a fronting of the preverb or (2) the application of rules governing the placement of the clitics (especiallycā andvā). The fronting rule allowed movement of the preverb to a limited set of positions, normally initial position in the sentence, second position if initial position was taken up by an element in COMP, and, in metrical texts, to a major metrical boundary. Thus the relationship between a preverb and its verb was easily recoverable by the native speaker.This close relationship between tmesis and fronting is likely to be very old. The majority of hte Vedic and Greek examples can be analyzed as 'preverb-fronting',15 compare also Watkins (1963) for Celtic. It seems likely that the fronting of preverbs at least originally implied some emphasis on the semantics of the preverb, such emphasis being the normal function of fronting in the Proto-Indo-European clause. Tmesis without demonstrable preverb fronting is not found in the Avestan corpus, nor in Vedic prose, but it is not sufficiently rare in the Rigveda (given our current understanding of Rigvedic clause structure) to allow us to say that tmesis and preverb fronting were one and the same process for that language. The Avestan evidence surveyed here will thus prove crucial to the development of a comprehensive account of the diachronic development of tmesis in Indo-Iranian, and, therefore, in the Indo-European language family as a whole.