Reexamination of Experimental Strength‐vs‐Grain‐Size Data for Ceramics

The Orowan, Petch, and Knudsen equations were examined against 46 sets of strength‐vs‐grain‐size data. For the 30 sets which are most discriminating, represented by 229 averaged‐data points, the variances of the Orowan‐Petch and Knudsen treatments are ∼4.5 and ∼17.1 kpsi 2 , respectively. Statistical considerations give preference to the Orowan‐Petch treatment at a high confidence level in these 30 cases, showing that the Knudsen equation does not remove systematic variations of strength as a function of grain size. The Orowan‐Petch treatment also appears to provide a sounder basis for extrapolation. The identification of the Orowan and Petch equations with postulated physical models permits interpretation of data in terms of probable underlying causes. Several examples are discussed and in most cases agree at least qualitatively with present understanding. A satisfactory physical model for Petch behavior in “nonyielding” ceramics is needed.