Fifty simulations for the numbers and sizes of metastases. In these simulations the primary tumors were presumed successfully resected from each host. All biological properties of the tumors and hosts were assumed constant. The only sources for deviation were random chance and heterogeneities in regional organ blood flow. In each simulation there were no clinically detectable metastases at the time of surgery, and the individuals were restaged after a time to allow for 12 volume doublings. a. Simulated numbers of metastases. Using the same parameterization as employed in Fig. 1, six cases sustained no metastases whereas the remaining cases sustained anywhere from one to just under forty metastases. b. Simulated sizes of metastases. The resultant sizes of metastases from each case are plotted here semi-logarithmically. The right axis provides the numbers of cells per metastasis; the left gives the corresponding diameters of the metastases. Many of these metastases would likely be below the threshold of conventional clinical detection (arbitrarily defined here as at 109 cells). Given additional time for growth the subclinical metastases would have also become overt.