ON THE CYCLONIC DISTRIBUTION OF RAINFALL. 



Some years ago I had the pleasure of hearing a lecture on 

 weather, given by one of the observers of the United States 

 Weather Bureau. The lecturer discussed the distribution of 

 weather in the extra tropical cyclones in America. He described 

 the conditions which characterize the four quadrants of an area 

 of low pressure. He especiall}' eniphasized the statement we often 

 And in text-books, that the greatest precipitation occurs in the re- 

 gion which lies some distance to the southeast of the center of an 

 area of low pressure. 



A short time after hearing this lecture I had occasion to dis- 

 cuss weather prognostics with a gentleman whose occupation had 

 led him for many years to closeh'' watch the government's fore- 

 casts issued ■ at Davenport, Iowa. This gentleman said he had 

 found that storms would usually arrive from six to twenty houi's 

 behind the time they were due according to the local forecasts. 

 Otherwise he regarded the predictions as quite reliable and valu 

 able. "When a storm is announced", he said, "it will almost al- 

 ways come, but it is apt to be a little behind time." This state- 

 ment corroborated an impression which I had myself received. It 

 is the writer's belief that if a careful comparison were made of the 

 forecasts referred to, and of the actual conditions of the weather 

 at Davenport previous to 1896 — since which time I have given 

 less attention to the matter- it would be found that the forecasts 

 more frequently missed hy announcing storms too earlj', than too 

 late. 



It occurred to me that this delay of the expected storms might 

 be due to some regional or local variation in the features of the 

 passing cyclones, and that it would be desirable to determine, by 

 some statistical method, the actual relation of weather conditions 

 to different parts of the cyclone for this locality. For the purpose 



