Cumulus Studies -51- 



"The rainfall data actually show, however, that the rainfall 

 on both October 14, 1948 and July 21, 1949 was exceptionally 

 high and could not have possibly been accounted for as the re- 

 sult of naturally occurring rain. This proof is made by the 

 analysis described in this paper. 



"The map of the State of New Mexico, which represents 

 about 120,000 square miles, was divided into eight octants or 

 45° sectors radiating out from Albuquerque. Then concentric 

 circles having radii of 30, 75, and 125 and 175 miles were 

 drawn on the map. This divided the whold state into 27 regions 

 whose average distances and directions from Albuquerque were 

 known. 



'"By entering on' the map for each of these regions the 

 average rainfall for Flights 45 and 110, a comparison could be 

 made of the distribution of the rain on those two days. An ob- 

 jective way of evaluating the similarity between such two dis- 

 tributions is to employ the statistical device known as the 

 correlation co -efficient. This was found in this case to be 

 +0.78 ± 0.076. The chance that such a high value would occur 

 among these figures if one set of them were shuffled giving 

 a random distribution is only 1 in 10. Such close agreement 

 in the distribution on two days could thus hardly be the result 

 of chance. There must be an underlying cause. 



"We believe that the close similarity in distribution is 

 dependent not only on the rather uniform synoptic situations 

 over the states that prevailed on these days, but also depended 

 on the fact that on both days the probability of rainfall depended 

 on the nuclei that spread radially out from Albuquerque, the 

 concentration decreasing as the distance from Albuquerque in- 

 creased. 



"The next step was to investigate just what characteristics 

 of this distribution were so similar on these two days. On each 

 of the two days, nearly all of the rain that fell occurred within 

 four of the eight octants. If each sector were divided into four 

 to six parts arranged radially so that each would contain equal 

 numbers of observing stations (about eight per region), the an- 

 alysis showed that the average rainfall rose rapidly to a max- 

 imum in intensity about 30 miles from the point of seeding and 

 that in each of the four sectors it decreased regularly as the 

 distance from the source of the silver-iodide smoke increased. 



