Plant fecundity is typically correlated with size (e.g. biomass); however, in plants whose reproduction is meristem-limited, there should be a stronger relationship between fecundity and architecture (the number and arrangement of subunits, or modules, composing a plant). In a reciprocal common garden experiment, I measured the effects of latitude-dependent flowering date on fecundity, biomass, and architecture in a photoperiodic annual (Perilla frutescens) to test the hypothesis that branch number limits flower production. If it does, the relationship between architecture and fecundity should remain constant across environments, even if the size-fecundity relationship varies. I planted Perilla seedlings from populations at 39°15'N and 33°05'N into a common garden at each latitude. Plants within a garden flowered on the same day, two weeks earlier in the south than in the north. Within a garden, size and flower number were positively correlated, but the slope of the relationship differed between gardens: plants of a given size made more flowers in the south than in the north. By contrast, the relationship between architecture and flower number was constant both within and between gardens. These data suggest that in meristem-limited plants, architecture is better than biomass at predicting fecundity across environments. Considering architecture, in addition to plant size, should enhance our understanding of environmental effects on plants whose reproduction is potentially meristem-limited.