In order to develop databases of creep information for design purposes for steels and other alloys, large-scale test programmes were undertaken involving several casts , product forms, etc. Little attempt was made to separate materials variables from those due to testing, and, even for simple measures of performance like life to rupture, considerable variability occurred and scatter-bands were large. Design information was obtained using statistical methods and curve fitting to predict long-term performance. The simple design rules incorporated safety factors which, along with the relatively inaccurate materials data , resulted in very conservative designs for power plant and pressure vessels.