CYCLONIC DISTRIBUTION OP PRECIPITATION. 
BY J. A. UDDEN. 
Plate XVII 
Several years ago I devised a method for ascertaining statistically the 
relation of weather conditions to different parts of a cyclone for a cer- 
tain locality or limited region of territory. Some results obtained by this 
method seem worth the while recording. 
The method followed was: by marking off eight radii in four concen- 
tric circles 1 plotted tw’^enty-five areas in a figure, which could be used to 
represent definite separate tracts in a circular storm. The lengths of the 
radii of the circles had the ratios 1:4:7:10, and could be used to repre- 
sent the same number of hundreds of miles in a cyclone two thousand 
miles in diameter. The radii was drawn at angles of 45° and were not 
extended into the inner circle. There were thus three tracts marked off 
in each octant outside the smallest circle. With this latter representing 
the central region of a cyclone and the figure so oriented as to allov/ 
the four points of the compass to bisect four alternate octans, it wao 
used to delimit twenty-five fixed areas in a cyclone. The construction 
of the figure will be readily understood from the accompanying diagram 
(Pig. 1) where each area is designated by number. 
My method was then to take a sufficient number of observations on 
the weather at the desired locality when this lay in each of the twenty- 
five corresponding tracts in an actual cyclone. 
These were then averaged for each tract separately, and percentages 
of frequency of certain weather conditions were thus obtained for each 
area, such as frequency of precipitation, of cloudiness and of wind direc- 
tion. The station selected for the first study was Davenport, Iowa, and 
the data used were the observations taken at this station at 8 A. M. 
The location of the low areas was taken from the daily weather maps, 
each corresponding day, by using a transparent paper with a diagram 
drawn to the scale of the map. Days, when no low areas appeared within 
a thousand miles of Davenport, were left out. Nearly a thousand obser- 
vations were used, taken from' as many maps. These were distributed 
somewhat unevenly in the twenty-five tracts of the diagram, but it is 
believed the number in each tract was large enough to secure a fairly 
representative average. 
It will be seen that this is nothing but a simple method of averaging 
weather conditions for difCerent parts of a cyclone at a particular sta- 
tion. The results can be plotted on the diagram as a chart. I have 
called such a chart a composite cyclone. Por the morning hour precipi- 
tation was found to be most frequent at Davenport when this station lies 
in the tract numbered eight, which extends from 100 to 400 miles west 
of the central “low”. It was also found unexpectedly high in tract 
numbered twenty-two, which lies from seven hundred to one thousand 
miles south of the “low”. In a southeast direction precipitation decreased 
very rapidly from the center of this composite cyclone. 
Cyclonic conditions were averaged in a like manner tor some other 
places, representing four other climatic regions in the United States. 
It was found expedient to make use of data slightly different from those 
used in the Davenport cyclone. I combined the 8 A, M. observations 
taken at Amarillo, Dodge City, Wichita, and Oklahoma during the years 
( 223 ) 
