HORIZONTAL DIFFUSION 



if it is to describe spore dispersal over a wide range of conditions (Gregory, 

 1945). First it will be necessary to re-examine Stepanov's results in the 

 light of present knowledge of eddy diffusion. 



TABLE VI 



DISPERSAL OF MIXED SPORES OF Tilktia carks AND Bovistci pliimhea. 

 Experiment 2 (Stepanov, 1935) 



Number of spores per cover-glass 18 x 18 mm. (average of 3) 



Tilktia Bovista 



Stepanov's observational data enable us to test whether the standard 

 deviation, o-, of the spores from their mean position agrees with Sutton's 

 form: o^ = |Ox"', or with the older diffusion theories where a- = iKt. 

 The data also allow us to estimate the parameters m and C, which can 

 then be compared with values obtained by meteorologists for similar 

 conditions. Examination of Tables V and VI shows that the spores at any 

 one distance do not lie in a smooth normal frequency distribution, but are 

 significantly clumped. This is probably because the duration of the dis- 

 persal operation was insufficient to smooth out the action of a few large- 

 scale eddies. 



The standard deviations of spores lying at each distance from the 

 source have been calculated for Table VII, where for convenience the 

 deviations from the mean position at each distance were measured along 

 the arc with the point source as centre. The standard deviation at each 

 distance was calculated from the usual formula, a = -\/[(x — x)"/(n — i)]. 

 This is not strictly legitimate, because the trapped spores are a systematic 

 instead of a random sample of the population and should be regarded as 

 estimates of the ordinate of a normal frequency curve. However, the 

 formula clearly gives a useful approximation — which would have been 

 better if the traps had extended farther laterally and if data for some of the 

 intermediate radii had not been missing. 



Both experiments were done in the same place, and with comparable 

 wind velocities, and when values for log a are plotted against log x the 

 points are found to lie reasonably close to a straight line. The slope of this 

 line is not unity, as it would have been with the older diffusion theories, 



53 



