676 
cases, vortices have occurred about a mile apart [12]; 
and in extreme cases, only 150 or 100 ft apart. On the 
other hand, in some of the most destructive tornadoes, 
no funnel clouds were recognized. Hither the pendants 
were much wider than their vertical extent or they were 
obscured by darkness at night or even during the day 
because of heavy precipitation or dust, a very low 
general cloud base, hills, or buildings. A higher per- 
centage of funnel clouds are seen in the Great Plains 
than in the eastern part of the United States [12]. 
The axis of a tornado initially may be nearly vertical, 
but since the top and bottom usually progress at 
different velocities, the axis becomes tilted away from 
the vertical in the direction of the vertical shear until 
it may be nearly horizontal. The base usually drags 
behind, because the translatory wind is reduced by 
friction at the surface. Finally, it may become sepa- 
rated entirely from the cumulonimbus cloud originally 
over it. 
Audibility and Scent. A tornado reaching the ground 
produces a roaring or buzzing sound which has been 
heard as long as one hour before it arrived [11] and as 
far as 25 mi away; this distance is comparable to that 
at which thunder can be heard, but is much greater 
than the 44 mi at which the sound of heavy hail can 
be heard. As this noise still occurs when a whirl is 
aloft (though to a lesser extent), it is not due entirely 
to the destruction being caused by the wind, but is 
due also to vibrations created by frictional effects in 
the strong wind shear of the whirl. Such sounds are 
augmented by long rolls of thunder, which may overlap 
to make a nearly continuous background of rumble. 
During one tornado, the air seemed to have a suffo- 
cating and burning tendency. Odors during a tornado 
are rarely noticed. They have been reported as re- 
sembling the smell of ozone or burning sulphur. These 
odors are more likely to be due to the effects of lightning 
discharges [17] than to the effects of the tornado. 
Paths. The diameter of a tornado, or width of a 
tornado path, which averages close to 250 yards, varies 
from zero up to about 1 or 2 mi. The length of a tornado 
path averages about 414 mi, and ranges from only 
about 100 ft up to 300 mi. “Cyclonic” tornadoes usually 
have much longer tracks than ‘‘convective” tornadoes, 
because they move faster and last longer. The most 
common path length is only 1144 mi, probably because 
the vortex frequently rises and skips over a stretch of 
land before starting a new path. Repeated lifting of the 
funnel may even occur while the vortex is stationary 
(Miami, April 5, 1925). 
About 90 per cent of all Northern Hemisphere tor- 
nadoes move in a direction from between south and 
west because they are embedded in and consequently 
move with the warm moist air [16], which usually 
blows from this quadrant. In hurricanes, tornadoes 
usually move from the east because of their location 
north of the center. Tornadoes sometimes move from 
the northwest, especially in Texas, where they may be 
steered by an upper wind blowing around the high 
aloft, characteristic of the southwestern United States 
in summer. A translation from unusual directions [12] 
LOCAL CIRCULATIONS 
may be found where the parent cyclone, if present, is 
not northwest of the tornado. Sometimes tornadoes 
within a single system move in widely different di- 
rections [12, 16]. 
The translation speed of a given tornado is not uni- 
form [17]. Also the average translation speeds of dif- 
ferent tornadoes vary, ranging from nearly zero to 
nearly 150 mph, with a mean close to 35 mph. The 
average tornado durations are computed to be 14 min 
at a single point and 8 min on a path along the ground. 
An extreme duration of over 7 hr along the ground has 
been observed (Illinois, May 26, 1917). 
The tornado path is often nearly straight when it 
extends for a considerable distance over flat country. 
For example, it may not deviate more than a mile or 
two in over 150 mi. Yet such a deviation would make 
the most common path (144 mi long) appear to be 
strongly curved. There is no preferred direction of 
curvature as the vortex zigzags along. However, ex- 
perimental research indicates that a vortex rotating in a 
cyclonic sense is forced to the left of a horizontal current 
(looking downwind). Possibly the path curves to the 
left if the tornado is moving faster than the accompany- 
ing secondary low, or to the right if moving slower. 
Hills and even buildings make tornado paths more 
crooked. Like a dust devil [14], a tornado on the slope 
of a hill or ridge tends to be deflected upslope. When 
ridges and valleys are oriented at right angles to the 
direction of translation, the tornado tends to skip, 
but skipping can also occur over water or flat land [17]. 
It appears that leeward slopes of ridges are more often 
severely hit than windward slopes, and that valley 
bottoms [18] are no safer than hilltops. However, some 
meteorologists do not agree with these effects of topog- 
raphy on tornadoes [8]. Rough terram destroys many 
tornadoes, but it sometimes creates whirlwinds by oro- 
graphic convergence. 
Many cases are reported of two or more whirls com- 
bining into a single whirl [11], or of one whirl breaking 
up into several whirls. Under these conditions there is 
usually only one major whirl. 
Theories on the Generation and Maintenance of 
Tornadoes 
Energy for Vertical Motion. A major problem in 
explaining the formation of a tornado is to find the 
source of the potential energy and the manner in which 
it is converted mto kinetic energy. 
In the nimeteenth century, Hare [10] theorized that 
the chief source was electrical because of lightning in 
and around the funnel, the fiery appearance [4] of the 
funnel, the smoky appearance [11] of the dust column, 
and the scorched appearance [4] of vegetation (owing to 
the effects of the sun on broken plants) along the path. 
Hare [10] supposed that the surface air was electrically 
charged and, being attracted upward by an oppositely 
charged thundercloud, started a rapid upward motion. 
Such a flow would tend to wipe out the vertical gradient 
of electric potential, as lightning does in a thunder- 
storm. Electrical phenomena are usually considered a 
result [17] of severe convective activity rather than a 
