The pattern of platelet disappearance.

S INCE PROBLEMS of estimating the survival of freely mobile cells (mostly the blood cells ) have occupied the attention of investigators, there has been much discussion as to the appropriate mathematical model. Schi#{248}dt,’ considering red cells, envisaged two major possibilities : first, that cell destruction is indiscriminate and age-homogenous so that an exponential curve should result; second, that the probability of destruction of a cell in the next interval of time is a function of the age of the cell. The former model has been likened to the survival curve of glasses in a restaurant where a new glass is as likely to be dropped and broken as an old one. The age-dependent model would be like the survival of a flashlight battery in continuous use : not all batteries would have exactly the same life, but the expectation of survival at any given time nonetheless changes with age. Theoretical aspects of these models have been explored to some extent. Although there has been little hesitation in supposing an age-dependent model for red cells, mainly because of what is known about their physiology and because considerable accuracy in measurement is easily obtained, the same cannot l)C said of platelet survival. Dornhorst pointed out ( and we have recently reemphasized ) that a high coefficient of variation ( i.e., a scatter of survival times of individual cells wide in comparison with average ) must necessarily lead to a total population curve which has at best a short “linear” segment, and with even a little error of measurement such curves may, in some instances, be indistinguishable from the exponential. One such case is illustrated in Figure 1, which shows how good a fit it is possible to get to the same series of points from a platelet survival curve with both linear and exponential models, provided observations arc restricted to a 10-day period. (The result is hardly surprising; it is a simple exercise in mathematics to show that, over a range small relative to the mean, any continuous monotonic transformation,