Appraisals of the two levelings that formed that southern California field test for the accumulation of the atmospheric refraction error indicate that random error and systematic error unrelated to refraction competed with the systematic refraction error and severely complicate any analysis of the text results. If the fewer than one-third of the sections that met less than second-order, class I standards are dropped, the divergences virtually disappears between the presumably more refraction contaminated long-sight- length survey and the less contaminated short-sight-length survey.