A land-farming project to remove diesel fuel from soil, which showed e
xtremely variable total-petroleum-hydrocarbon (TPH) concentrations, is
presented. Experiments showed that the variability was not an artifac
t of the TPH analytical method, but reflected the random distribution
of contaminant in the soil. It was found that the data were well fit b
y a lognormal distribution. The implications of the sampling statistic
s on the remediation standard are discussed using the data, the lognor
mal distribution, and the remediation standard imposed on this particu
lar remediation project.