Auveust 30, 1918] 
crops and the latter bad for the soil. Data 
for a few hundred stations were taken from 
Bulletin Q of the U. S. Weather Bureau, which 
although it brings the records down only to 
the end of 1903, and contains a few typograph- 
ical errors in the figures, is easier to handle 
than any later publication covering the same 
ground, and is probably accurate enough to 
base a working hypothesis on. 
The resulting map differs considerably from 
any other precipitation map, but instead of 
publishing it in its present imperfect state a 
brief description will be given. The line of 
equilibrium, where there is no difference be- 
tween early and late summer rainfall, crosses 
the St. Lawrence River in northern New York 
and extends in a general southwesterly direc- 
tion, with various sinuosities (perhaps due 
largely to differences in altitude and exposure 
between neighboring weather stations in the 
Appalachian region) to the vicinity of New 
Orleans, thence westward, passing between 
Houston and Galveston, to near Del Rio on the 
Rio Grande, northwestward across the Staked 
Plains to the Rocky Mountains in Colorado, 
westward to Monterey County, California, and 
then southeastward just back of the Coast 
Ranges into Mexico. Another part of it sepa- 
rates the northern half of Michigan, north- 
eastern Minnesota, and part of Wisconsin from 
the states to the southward, then passes north- 
westward into Canada and dips back into the 
United States just enough to cut off the north 
end of Idaho and the northwest corner of 
Washington. The borders of the United 
States are mostly in the region of late summer 
excess, while approximately three fourths of 
the country, including almost the whole area 
drained by the Mississippi River, has an early 
summer excess. The greatest late summer ex- 
cess, about 11 inches, is on the east coast of 
Florida,? and the mouths of the Mississippi 
and Rio Grande are not far behind. The 
Black Hills have an early summer excess of 
half in late summer, and perhaps get still greater 
contrasts. 
7 Nassau in the Bahamas, about 180 miles 
farther east, has a late summer excess of nearly 
13 inches. 
SCIENCE 
209 
about 6 inches, and the area having more than 
4 inches extends all the way from Montana to 
Alabama. If ratios instead of differences had 
been used the position of the zero line would 
have remained the same, but the gradient 
would have been steepened in the drier parts 
of the country. 
The map here described suggests at once 
some very interesting correlations. Consider- 
ing other climatic factors first, nearly all our 
tornadoes occur in the region of considerable 
early summer excess of precipitation, and our 
hurricanes in that of considerable late summer 
excess, while regions where the difference is 
not more than an inch or two either way rarely 
suffer much damage from wind. Both torna- 
does and hurricanes usually occur during the 
period of greatest rainfall in their respective 
regions.$ 
The late summer rains commonly come in 
the form of showers in the daytime, while the 
8 There is a tornado frequency map of the United 
States by J. P. Finley in Professional Paper No. 
7 of the U. S. Signal Service (1882), reproduced 
on a smaller scale with fewer details by R. DeC. 
Ward in Quart. Jour. Roy. Meteorol. Soc., 43, 323, 
1917. This is based on 600 tornadoes occurring 
between 1794 and 1881, but is a little misleading, 
because in the early part of that period the re- 
gions where tornadoes are most frequent were 
practically uninhabited by civilized man, and thus 
the apparent frequency of such phenomena in the 
older states is exaggerated. Of the 40 tornadoes 
prior to 1850 reported by Finley, 9 were in New 
York, 5 in Ohio, 3 in Connecticut, 3 in Georgia, 
and none in Kansas and Illinois, which now lead 
the list. A map based on records from about 1870 
to date would be more accurate; but even Fin- 
ley’s map shows a fair degree of correspondence 
between tornadoes and early summer rainfall ex- 
cess. 
It appears from Bulletin X. [not 10] of the 
U. S. Weather Bureau, on Hurricanes in the West 
Indies, by Dr. O. L. Fassig (1913), that there are 
only three well-marked hurricane regions in the 
northern hemisphere, all lying between latitudes 
5° and 30° N., and all having a maximum storm 
frequency in September. In the United States 
hurricanes are most frequent in Florida, but they 
are occasionally felt as far north as the coast of 
Massachusetts. 
