The report presents the results of two series of tests which were made to investigate the commonly accepted criterion of instability for shell bodies. This criterion, as generally postulated, states that the total potential energy must be the same before and after buckling. It is an analytical consequence of this contention that the critical load at constant end shortening must be greater than at constant stress. Thus, there should be a difference between tests made in rigid and flexible machines. The experiments recorded provide very strong evidence that this is not so. In the research reported, no evidence of machine influence on initial buckling conditions could be found.