Abstract 
Nasalization is a widespread areal feature among the contemporary dialects of China. From a survey of over a thousand ‘diapoints’ emerges a clear distributional pattern of nasal vowels (NV’s): they tend to occupy the lower portion of the vowel space. Three hypotheses are proposed to explain this phenomenon: (1) Nasalization tends to spread from low to high vowels, as a result, low NV’s are the most likely to emerge. (2) Denasalization progresses in the opposite direction; low NV’s are, therefore, the most likely to survive the attrition. (3) NV’s tend to fall; consequently, NV’s tend to populate the lower corner of the vowel triangle. Each of these hypotheses is examined against a broad data base. Hypothesis (1), it will be shown, must be refined so as to take into account an added parameter: nasalization is more likely to be triggered by an anterior nasal (-m,n) rather than a posterior (-n) ending. Hypothesis (1), modified with the qualifier noted above, is strongly supported by an impressive body of dialectological data. Hypothesis (2) receives some empirical support, but the overriding pattern of denasalization seems to be determined by an ‘equal lifespan’, that is the oldest crop of NV’s is also the first to undergo denasalization, the newest the last. The dialectological evidence for Hypotheses (3) is ambiguous, as NV’s shift in unpredictable directions. However, while nasality appears to favor lowering, certain upward movement of NV’s can be regarded as part of the general rising tendency of tense or long vowels (-VN > by compensatory lengthening).