CLIMATE AND FOREST FIRES 



977 



Series II, on the other hand, was burned on the day after a rain, as 

 soon as the htter was dry enough to ignite. 



For each series the fires having the same wind velocity were then 

 averaged. Thus, velocities of l.G to 2.5 miles per hour were thrown 

 together, 8.6 to 3.5 miles, etc., and the perimeter measurements were 

 also averaged. After this had been done, it was found that the average 

 values in Series I and III were practically identical, but the size of the 

 perimeter for a given wind velocity was much lower in Series II than 

 in the other two, due, of course, to greater moisture content of the 

 litter. 



In order to compare all fires by the same standard, the basic rate of 

 spread — that at 0.0 miles wind velocity — was taken as 100, and for each 

 series the values at different velocities were referenced to this base. 

 The values thus derived give, of course, not actual perimeters, but rela- 

 tive indices. 



Table 7 shows the final line-up of the average derived index figures. 



Table 7 



It will be seen that there are considerable differences in the values 

 derived for the different series, and, indeed, there is no intention of 

 claiming that this preliminary study is the final word on the subject. 

 An examination of the last three columns is, however, very instructive 

 in showing a possible mathematical law, which expresses rate of spread 

 as governed by wind velocity. Column 5 gives the average relative 

 perimeter for wind velocities of 1 to 5 miles an hour; column 6, the 

 square of wind velocity ; and the last column, the index figures secured 

 by dividing column 5 by column 6. The last figures are practically 

 constant for the velocities tested and for the data. We may say that 

 rate of spread in perimeter varies as the square of wind velocity. It is 



