Accurate measurement of dry bulk density is essential for Ore Reserve estimates and tracking the mass and grade of ore during mining operations. This is particularly so in young, tropical nickel laterite deposits, which typically show large variations in dry bulk density depending on ore composition. Diamond core drilling is the method of choice for definition of mineral resources in these deposits. If carefully executed, core drilling can provide excellent sample recovery rates and dry bulk density data. An unusual feature of the core recovery in young tropical nickel laterites is that, when drilling through very soft materials, recoveries of significantly greater than 100 per cent are commonly recorded over multiple contiguous drilled intervals. The cause of this phenomenon has been debated for decades. The critical question is: are these extra-long recovery samples representative of the grade and bulk density of the in situ drilled interval? As part of a study into the measurement and modelling of bulk density within nickel laterite mines at Sorowako in Indonesia, the origin of these extra-long recovery samples was investigated. The methods used for core drilling and density measurement were reviewed. A statistical comparison of the geochemistry of extra-long recovery and normal (recovery less than or equal to 100 per cent) samples was completed. Machine learning was used to predict bulk density from multivariate assay data. Comparison of the measured bulk density and predicted bulk density of the extra-long recovery samples showed that the measured bulk densities are generally reliable. This leads to identification of a physical explanation of the phenomenon of extra-long cores in nickel laterite deposits.