CUMATE AND FOREST FIRES 975 



cant. The suppression energy necessary to corral a fire varies directly 

 as the perimeter, and for that reason it will be used in this study as the 

 criterion of rate of spread. 



On level land', with no wind, and with uniform cover, a fire spreads 

 in a circle and the perimeter varies directly with linear distance 

 traveled. Wind and slope will, however, modify the shape of a fire, so 

 that it tends to become longer in one axis than the other, though still 

 retaining a generally oval shape. In other words, for a given geomet- 

 rical figure, equal increments of time give equal increments of peri- 

 meter. In the actual experimental fires it is found that, with slow rate 

 of spread, the perimeter time relation is approximately a straight line 

 up to a period of two hours ; with more rapid spread, under wind, the 

 increments of perimeter increase from period to period and, instead of 

 a simple arithmetical progression, perimeter on time tends toward a 

 geometrical series. Two independent factors are active in the case of 

 rapid spread : First, the ratio between linear spread and perimeter 

 (which is, of course, 3.14 for a circle) tends to increase, so that peri- 

 meter increases more rapidly than distance traveled. Second, the 

 release of a large amount of heat, in a short period, results in convec- 

 tional currents of air, which increase the wind velocity and hence the 

 rate of spread; or, to put it more simply, a fire creates its own draft. 

 The extent of this increase is difficult to measure instrumentally. In 

 one case, in which the wind velocity 300 feet from the fire was 4.4 

 miles per hour, the velocity at the front of the fire was 6.3, or roughly 

 40 per cent higher. 



Without many more data than are now available, it is impossible to 

 formulate any definite law of the relation of perimeter to elapsed time. 

 Table 4 shows, for 23 experimental fires, the average perimeter in- 

 creases, by 5-minute intervals. 



Table 4 



Average ... 76 



