PROCEDURES 



Observed and predicted rates of fire spread were compared using test fires on an 

 old bed of the Clark Fork River east of Missoula, Montana (fig. 1). 



Plots 8 by 30 feet were prepared using logging slash hauled by truck as the 

 fuel. All other vegetation was cleared from the plots. Slash was used as a test fuel 

 because it (a) often is a hazard and silvicultural problem of considerable concern to 

 land managers, (b) cannot be readily tested in a laboratory, and (c) provides a test 

 of the spread model using heterogeneous fuel for which only theory exists. 



Slash of two species, Douglas-fir (Pseudotsuga menziesii var. glauca (Beissn.) 

 Franco) without needles and ponderosa pine (Pinus ponderosa Laws.) with needles, was. 

 used at three depths for each of two loadings. Each combination of species loading 

 and depth was replicated twice, making a total of 24 plots. Two extra plots were con- 

 structed for training but were not needed for this purpose so that 13 test fires were 

 burned for each species. The loadings and depths are within a range that is often 

 encountered in slash. 



The plots were on a zero slope and were burned under low wind velocities. A test 

 of the spread model at high wind velocities was not possible. Fairly low fuel moisture 

 contents were sought and obtained on all fires. 



The plots were loaded uniformly with weighed amounts of slash providing conditions 

 of approximately 15 and 30 tons per acre. Immediately after weighing and loading, the 

 plot moisture content was sampled by ovendrying needles and branches from several 

 diameter size classes. The moisture estimates (with standard errors within 4.9 and 8.6 

 percent of mean values for Douglas-fir and ponderosa pine, respectively) permitted pre- 

 cise estimation of the actual ovendry weight of fuel. The slash on some plots was 

 buoyed with wire to provide desired fuel depths. 



5 



