NO. 7 ARGENTINE WEEKLY WEATHER FORECAST HOXMARK II 



perature curves, the predicted rain is indicated by small dots, and the 

 observed precipitation by short heavy Hnes. 



The predicted and observed temperature values at 8 a. m., and 

 8 p. m., for each day from May 29 to August 20, 1924, will be 

 found in tables 7, 8, 9 and 10, with their corresponding weekly 

 correlation coefficient. The solar^radiation data received by cable 

 from the Montezuma Observatory, Calama, and the solar faculae 

 observations from the Pilar Magnetical Observatory for the same 

 period, are shown respectively in tables 11 and 12. 



A negative correlation is not always indicative of absolute failure 

 of a forecast. This is amply illustrated by the two weeks, July 10-16 

 and July 17-23, 1924. 



Glancing at the graph, figure 4, we can see that the observed 

 curve on the eleventh of July shows the maxima indicated by the 

 forecast, though the temperature had fallen several degrees under. 



In the second case, on only two days, namely the 19th and the 20th, 

 were the observed values contrary to the forecasted ; and yet the 

 correlation coefficients for those weeks turn out to be —.49 and — ■'JZ'. 



