COVER TYPE AND FIRE CONTROL 13 
Many of the original reports did not specify the types in which 
the fire originated. To determine the type with substantial accu- 
racy, the location of each fire was referred to a type map prepared 
for the purpose. This map shows the major types and, although 
probably some errors in determining the type in which particular 
fires occurred have crept in, these errors again are probably com- 
pensatory. it has been necessary to charge the acreage and cost of 
each fire to the type in which it originated. This is not invariably 
correct, as for example when a fire starting in chaparral is controlled 
at the edge of the western yellow pine and burns over a small area in 
that type. However, with few exceptions, the major run and charac- 
ter of a fire are determined by the type in which it started. A fire 
starting in chaparral, for example, will ordinarily attain such size 
and such momentum that it will still be very difficult to combat 
after it spreads into western yellow pine timber or into woodland 
areas, and can be classified as characteristically a chaparral fire. 
The use of the data in this manner probably affects the figures for 
total area burned by types, but is immaterial in other relationships. 
The unusual fire year and the few critical fire days of each year are 
of extreme importance in measuring ultimate success in protection. 
‘However, these extreme days and years affect all types within a 
general region simultaneously, but with an intensity varying with 
the type. The quality of data at hand precluded any detailed study 
of these important facts. They will, however, be isolated and 
analyzed in subsequent studies. 
In the analysis of the 10,476 fires the following data were abstracted 
for each fire: Date of start, cause, location, acres burned, class of fire, 
cost of suppression, elements of cost, type of cover, length of time 
from outbreak of fire to attack on fire, and for the years 1918-1920 
the length of time from discovery of fire to attack. These data were 
then assembled in various ways—by cover types, causes, by size of 
fires, etc., and studied in groups. In the analysis of the data, cover 
type was employed as the controlling basis for classification of the 
major observed and recorded facts. Average relationships have been 
used throughout. 
Meteorological data have been taken from the United States 
Weather Bureau records and from material collected by J. A. Mitchell 
and EK. N. Munns, of the Forest Service. 
OMISSION OF SPECIAL-RISK FIRES 
Of the eight recognized causes of fires—railroad, lumbering, brush 
burning, miscellaneous, lightning, camper, incendiary, and unknown— 
those resulting from railroads, logging, brush burning, and freak causes 
(miscellaneous), which may be termed “‘special-risk”’ fires, are every- 
where localized. Their occurrence depends mainly on the presence 
within a known restricted area of specific fire-using agencies. They 
are mainly the result of industrial occupancy. For example, lumber- 
ing has caused many forest fires through the use of fire in its operation 
of wood-burning donkey engines and railroads, and in brush burning, 
but such fires are possible only during the life of the lumbering 
operation. Furthermore, protective agencies and the industry have, 
during this 10-year period, been devising new methods of prevention 
and control for these classes of fires, making any specific conclusion 
drawn from observations to date of types and special-risk fires highly 
