Analytical models used for distribution of load within arch bridges are considered inadequate. They underestimate the strength of many bridges while others are showing signs of distress at loads much lower than analysis would predict. Since the analyses are clearly sound in the two dimensional sense in which they work, any error must result from the input, of which distribution models are a major part. This paper presents a more rational concept of distribution, showing how the concept can be modelled economically and effectively and indicates the likely benefits.