THE MICROBIOLOGY OF THE ATMOSPHERE 



that there is a random scatter around the source. In such a distribution 

 the gradient in any direction would have the form of one-half of the normal 

 frequency curve. S. Wright (1943, 1946) studied genetic effects of isolation 

 distance, and his methods were applied by J. W. Wright (1953) to compare 

 dispersion distances of pollens of various forest trees with a view to 

 delimiting a 'neighbourhood' for race formation. Simple sticky-slide 

 traps were exposed at various distances around isolated trees, and pollen 

 counts were used to find the standard deviation of the scatter. Observed 

 values for the standard deviation were as follows: ash, 17-46 metres; 

 Douglas-fir, 18 metres; poplar and elm, 300 metres or more; spruce, 40 

 metres; Atlas cedar, 73 metres; Lebanon cedar, 43 metres; and pinyon 

 (pine), 17 metres. Dispersal data were well fitted by the Gregory formulae, 

 but not by theories which assume that the trajectory of each grain can 

 be calculated from the rate and distance of fall, and the wind velocity. 



Bateman (1950) questioned whether gene dispersal is statistically 

 'normal', and showed by his regression method that many observed 

 distributions — including those of fungus spores, passively borne insects, 

 pollen, and wind-dispersed seeds — were highly leptokurtic, i.e. the peak 

 and tails of the distribution are exaggerated at the expense of the shoulders. 

 Compared with a normal frequency distribution having the same standard 

 deviation ('same over-all degree of inbreeding'), the leptokurtosis charac- 

 teristic of passive airborne dispersal produces more breeding between 

 close relatives and simultaneously more breeding between distant relatives 

 (see also Parker-Rhodes, 195 1). 



180 



