Uncertainty, in observations

Given these uncertainties, the only rationale way of assessing the accuracy of the prediction methodology is to compare the ranges in observed and predicted crack depths for different welds and stainless steels. This comparison is made in Figure 18.36b for various BWR core shroud welds (H-3, H-4, H-6) manufactured with different steels (Type 304, 304 L). Over this population the agreement between observation and prediction is reasonable, given the uncertainties in observation and definition of the system. 

---