Population geneticists can reconstruct the ancestries of macroscopic populations from polymorphisms in present day individuals. For example, the migration "out of Africa" is recorded in human genome variation in different parts of the world. Here we apply this approach to human colorectal cancer cell populations and polymorphic passenger methylation patterns. By sampling molecular variation from different parts of the same cancer, it should be possible to infer how individual tumors grow because recent clonal expansions should be less diverse than older expansions. Average diversity was different between cancers implying that some cancers are older clonal expansions than others. For individual cancers, methylation pattern diversity was relatively uniform throughout the tumor (right versus left side, superficial versus invasive), which is more consistent with a single, uniform or "flat" clonal expansion than with stepwise sequential progression. Many colorectal cancers appear to invade and expand early, but subsequently stall. Epiallele diversity within individual small cancer gland fragments was high and more consistent with frequent rather than extremely rare cancer stem cells (CSCs). These studies suggest that many human colorectal cancers are relatively old uniform clonal expansions, that cancer cell populations contain frequent long-lived CSC lineages, and that some passenger methylation patterns record somatic cell ancestry.