The thesis of accentual rhythm, presented in this study for the first time, according to which normal word accents are responsible for the inner rhythmical structure of Latin poetry, opens a completely new path in metrical research by examining verses from a musicological point of view. Ancient testimony proves that word accent (according to modern linguistics which are not seriously challenged nowadays it was a stress accent) is fundamental to the Latin language, and that Latin poetry was generally recited to listeners rather than written for readers. In verse-analysis acoustically perceptible factors therefore are of particular interest. An impressive argument is provided by the remaining poetry itself. Regulations of verse clausulae and caesurae have already been widely discussed in metrical research. They all concern word boundaries in the verse. But what was the purpose of those regulations? If they should act upon the ear rather than upon the eye, then it is safe to assume that they were aimed at the regular distribution of word accents, which according to the law of the penultimate depend on the word endings. The resulting rhythms however are unfamiliar to modern occidental ears, in that they do not correspond to the common scheme of two or three beats to the bar. If one recites the long and short syllables in the correct ratio of 2:1 and uses the word accents as downbeats, the result thus derived is additive rhythms (comparable to the Central-European Zwiefacher, Indian Talas, West African Timeline-patterns, etc.), which are variable according to the variously located word boundaries and accents (e.g. the hexameter of 24 time units regulated by clausula and caesurae has only three main types with different basic timelines). This variability is consistent with ancient concepts of rhythm: according to Quintilian 9.4.55 the rhythm of Latin prose and verse (!) never runs on monotonously without variety, therefore any kind of ictus (the general stressing of first syllables of the verse foot, whereas word accents are ignored; in school it is useful to distinguish long from short syllables) in normal reciting of verses must be vehemently rejected. The quantitative system of verse feet is only a grid into which the words are inserted, but which in itself is not perceivable in reciting. It can be compared to a knotted carpet: The pattern is visible on the front side, and the underlying structure can still be seen on the back side. Anybody giving weight to the ictus would therefore be apt to display a carpet backwards. One of the fascinating results of this study is the fact that accentual rhythms are connected to the content and form of the text. Thus Ovid signals phrase endings inside the hexameter by certain rhythms, which take on the function of modern punctuation.