This paper explores the statistical nature of the mechanisms of fracture in structural materials and the loads that initiate them. The result of employing either a weak link or bundle model in the application of statistics to problems of fracture is examined. Experimental data is presented for the fracture of metal chains and the fracture of sheet aluminum containing machined cut-outs and cracks. The scatter in fracture loads for the chains and fracture toughness for the sheet specimens is statistically analyzed utilizing four statistical distributions: normal, lognormal, extreme value, and two-parameter Weibull functions. Based upon this work there appears to be no general a priori justification to utilize either the weak link or the bundle model in the statistical assessment of the fracture of typical engineering structural components.