In the first known attempt to use a sympatric, non-mimetic convergence (‘a naturally controlled experiment in evolutionary morphology’) to quantitatively partition a morphology among its controlling factors, shells of the polygyrid land snails Neohelix major (Binney) and Mesodon normalis (Pilsbry) were analysed in a factorially designed rearing experiment. The complete absence of character displacement or character release confirmed previous field studies. Growth rate, a determinant of adult shell size, was controlled 50% by direct environmental induction (with temperature twice as effective as humidity), 10% by proximate natural selection (with a shorter growing season selecting for faster growth), 10% by genotype-environment interaction (with moisture generally inducing a greater growth surge in a drier climate), and 30% by within-cohort variation (apparently heritable). Spire height confirmed predictions that flatter shells make better estivators (shelterers from drought) and that taller shells make better climbers and tall/flat niches are readily filled: high within-cohort variation (70% of total) provides the raw material for niche-filling (but is reduced in severe climates by stabilizing selection); environmental induction (10%) is strictly by humidity, which makes shells taller for foraging more widely under more favourable conditions; proximate natural selection (10%) produces taller shells in moister climates (with intensity of effect dependent on evolutionary time); and genotype-environment interaction (10%) makes the drought-induction of flat shells stronger in more drought-prone areas. Coiling tightness, allied to whorl-expansion rate, shows distinct evidence of adaptive neutrality: 60% of total variance is explained by phylogenetic constraints despite an estimated 120 million years of genetic isolation and despite the demonstrated ability of related clades to converge; the 10% due to genotype-environment interaction is generally random and nonadaptive in its response; and the remaining 30% appears random, non-induced, and non-selected. This study's protocol of experimentally partitioning a ‘closed universe’ of morphology among a full set of controlling factors is recommended as an alternative to the usual practice of analysing the effects of one or a subset of factors on an ‘open universe’ of morphology. One advantage of this new protocol is that it obviates the ‘nature-nurture debate’ by quantifying the relative contributions of each, which in this snail-shell case differ drastically among the various components of a single morpholocgyy.