VI INTRODUCTORY NOTE 



6. One is impressed by these very large negative correlations, that 

 they are not accidental, but are real correlations and full of meaning 

 but reversed because some governing factor is neglected. If they 

 had all been of the correct sign, and without altering the signs of 

 coefficients less than 30 per cent, which may well be of a more acci- 

 dental character, the average correlation coefficients for the entire 

 131 weeks would have risen to over 37 per cent. 



7. Readers of the next preceding serially numbered paper of these 

 Miscellaneous Collections, by JMr. Clayton, will have noted that he 

 has discovered an enormous shift in latitude of cyclones and anti- 

 cyclones depending on absolute values of the solar constant. Alay 

 it not be that Mr. Hoxmark, by further investigation along those 

 lines, will find such additional relations for Argentina as will enable 

 him to obtain correct signs of correlation in the numerous cases of 

 incorrect but high correlations, so that his future forecasts can be 

 notably improved? It would seem that a special study of the con- 

 ditions surrounding" the cases of high correlations of incorrect sign, 

 which he is here reporting, might lead him to new points of view, 

 which would prove very helpful in future. 



