Overview

For the development of a scientific theory, experiments are always useful in those areas where theory has not yet reached; they will indicate, perhaps, unforeseen behaviour which can then be incorporated in relevant analysis. Thus the design of connexions for steelwork or the disposition of reinforcement for concrete are difficult problems amenable to analysis, but illuminated greatly by the results of practical tests. Indeed, progress in design in these areas would not really have been possible without the guidance of experiments.

There was no such compulsion to make experiments for mainstream structural analysis. Navier's formulation, exploited and perfected by nineteenth century engineers, appeared to be logically and self-evidently correct. Experiments to confirm the theory would be otiose if there were disagreement between test and theory, this would indicate merely that the test had been badly made. Thus Mariotte's tests on fixed-ended beams in the seventeenth century which led to the genuine and important observation that the breaking strength of a fixed-ended beam was twice that of the corresponding beam when simply supported were not followed up during the next 200 years. Experiments were indeed made in this period on the strength-of-materials problem the breaking strength of Galileo's cantilever beam but there are only isolated instances of the application of these results to hyperstatic structures. In any case, after 1826 the breaking strength of a structure was not the objective of analysis; Navier had stated clearly that the maximum stress in a structure was not to exceed a certain fraction of the limiting stress of the...