Bimodality of blood pressure levels

I examine distribution of systolic blood pressure by means of maximum likelihood fitting of various statistical models to the extensive data of Bøe et al., and find evidence for a compound or mixed log-normal distribution, rather than the single log-normal distribution proposed by Pickering. I find a second distribution with a mean blood pressure above the ordinary range. I demonstrate further evidence for the existence of this subgroup in the abnormalities of regression to the mean found in the same data set.