Research on syllable weight in generative phonology has focused almost exclusively on systems in which weight is treated as an ordinal hierarchy of clearly delineated categories (e.g. light and heavy). As I discuss, canonical weight-sensitive phenomena in phonology, including quantitative meter and quantity-sensitive stress, can also treat weight as a gradient interval scale in which (1) differences between syllable types are matters of relative degree rather than strict domination, and (2) there is no clear segregation of syllable types into categories, but rather a continuous distribution of types along a continuum of weight. In a meter sensitive to gradient weight, progressively heavier syllables are progressively more skewed towards metrically strong positions, all else being equal. Gradient weight is likewise evident in a stress system when syllables vary along a continuum in their propensities to attract stress, again controlling for distributional confounds unrelated to weight. The dissertation consists of four parts. Part I comprises corpus studies of six quantitative meters, namely, Kamban's Tamil epic meter, the Homeric Greek hexameter, the Latin hexameter, the Finnish Kalevala meter, the Epic Sanskrit "sloka," and the Old Norse "dróttkvaett." All six are widely held to treat syllable weight as exclusively binary. I demonstrate that, in addition to distinguishing light and heavy syllables, the poets in all six traditions exhibit sensitivity to a continuum of weight within the heavies, to the extent that I am able to derive some of the most detailed scales of syllable weight yet documented for individual languages (e.g. at least nine levels in the first case study). Moreover, across the six languages, the weight scales are strongly correlated, both with each other and with the crosslinguistic typology of weight-sensitive phenomena, supporting and shedding new light on the universal phonology of syllable weight. Part II first addresses the universal principles of weight, e.g. complexity and sonority, that motivate the features of the scales in part I. Particular emphasis is given to Tamil, including its violation of the sonority principle: C[subscript 0] V CR (R = rhotic) is lighter than all other C[subscript 0] V CR[subscript ?R] despite the rhotics being highly sonorous. Prosodic minimality in Tamil also diagnoses C[subscript 0] V CR as being lighter than C[subscript 0] V CR[subscript ?R]. I argue that this discrepancy is motivated by the short durations of rhotics relative to other codas in Tamil. More generally, judging from a phonetic corpus of Tamil, duration of the rime (or energy integrated over duration) correlates tightly with the weight continuum inferred from meter. Building on these empirical findings, a generative analysis of gradient weight mapping is proposed in a maximum entropy constraint framework. In it, categorical and gradient constraints (the latter being violated to real-valued degrees supplied by the phonetics; Flemming 2001) interact to generate the weight mapping typology. This typology includes fully categorical systems, fully gradient systems (directly reflecting the phonetics), and systems exhibiting various degrees of incomplete categorization, in which the phonology is polarized towards categories but remains sensitive to the gradient phonetic interface of weight within categories. Part III treats the contributions of onsets to syllable weight. While onset structure is irrelevant to weight categorization in all the languages examined, it contributes consistently to weight as a statistical effect, in that more complexity is associated with greater weight (e.g. in Old Norse, onset Ø < C < CC < CCC[subscript 1]). Even in Tamil, in which complex onsets are illicit, mean duration of the onset correlates significantly with metrical weight. Finally, part IV considers gradient weight in stress assignment in English. The distribution of stress in extant disyllables follows the same (universal) principles established for meter: First, the complexity of the coda correlates monotonically with stress propensity, such that Ø < C < CC (<) CCC[subscript 1] (as seen in both nouns and verbs). Vowel length is also important, such that, taking the rime as a whole, the hierarchy V < V C < VV < VVC is observed for both nouns and verbs. As in meter, onset complexity also contributes significantly to stress propensity, as observed independently in nouns and verbs as well as in initial and final position in disyllables. Experimental evidence is presented supporting the productivity of the universal V < V C < VV < VVC hierarchy in English stress, as well as the onset effect. [The dissertation citations contained here are published with the permission of ProQuest LLC. Further reproduction is prohibited without permission. Copies of dissertations may be obtained by Telephone (800) 1-800-521-0600. Web page: http://www.proquest.com/en-US/products/dissertations/individuals.shtml.]