342 SPECULATIONS ON CELLULAR ACTIONS 



cells. In any given case, the deviations from normal are determined in 

 one of the following ways : 



1. The observations are made on groups of cells of the same species; 

 for example, the respiration rate of a suspension of irradiated yeast cells 

 is measured and expressed as percentage of the rate of a non-irradiated 

 control suspension. 



2. The observations are made on the individual cells of the sample. 

 a. The end effect is of the all-or-none type; for example, the cell 



divides or it does not. 



h. The effect is a quantitative modification of a normal structure or 

 activity; for example, the number of fat droplets in the cell is increased, 

 or the respiration rate of the cell is decreased.* 



Throughout this paper I shall be discussing primarily the all-or-none 

 effects (type 2a), although many of my remarks may well apply to the 

 other two types. 



A very common type of experiment consists in dividing a population 

 of similar cells into several samples, giving each of the samples one of a 

 series of graded doses of a given radiation, and subsequently classifying 

 each cell on an all-or-none basis. From these raw data can be calculated 

 the fraction of the cells affected in each sample, and this fraction can 

 then be plotted as a function of dose to obtain a dose-effect curve. For 

 convenience we frequently alter this procedure by plotting the fraction 

 not affected against dose, thus obtaining a survival curve. 



If the individual decisive processes occur entirely at random, any 

 survival curve can be theoretically described in terms of w, the number 

 of individual decisive processes of type M which must operate on a pre- 

 cursor to produce a decisive entity; n, the number of decisive entities re- 

 quired to constitute a decisive state; and h, the mean number of in- 

 dividual decisive processes M produced per cell per unit dose. How- 

 ever, in some radiobiological actions, the individual decisive processes 

 operating on a given individual precursor may be linked; that is, r such 

 processes may result from r individual chains of events (Fig. 2) which 

 have their ultimate origin in the same ionization track. Moreover, it is 

 possible that in some actions the decisive processes operating on s (two 

 or more) precursors may also be linked. Accordingly, to make our state- 

 ments as general as possible, we shall replace m with m/r, the number 

 of groups of decisive processes required to change a precursor to a de- 

 cisive entity; replace n with n/s, the number of groups of decisive en- 



* This method is usually laborious and consequently avoided. If desired, the ex- 

 perimental data can usually be handled as all-or-none by using an arbitrary criter- 

 ion; for example, the cells whose respiration rate is decreased to 50 per cent or less 

 of normal may be scored as affected, the remainder as "survivors." 



