130 H. A. llazen — Verification of Tornado Predictions. 



Table. 



Prediction, occurrence, and non-occurrence of tornadoes by Mr. Finley, 

 for June, 1885. 



Districts. Success. 



















Success- 



Unsuc- 







Day. 





Predicted for 





Occurred in. 



ful. 



cessful. 



Failures. 



By weight. 



2 



9, 



15 









13 







2 



1 



h hi 



3 



13, 



15 









15, 



13 



2 







*,f 



4 



9, 

















1 





£ 



5 



2 



4 









16 







2 



1 





6 



15, 



16, 



18, 



17, 



12 









5 









7 



8, 



9, 



12, 



13 





15, 



8 



1 



3 



1 



hi 



8 



5, 



6. 



4, 



7, 



12 









5 





0, 0, 



9 



12, 



11 















2 









12 



13, 



14. 



15, 



16 





18, 



16, 15 



2 



2 



1 



hi, 1 



13 



15, 



16 















2 





i 



14 



13, 



9, 



15 







16. 



15 



1 



2 



1 



1,4, i-£ 



15 



15, 



17 















2 





i,hh 



19 



14 



















1 







20 



13. 



16 









14, 



13 



1 



1 



1 



i, i 



21 



8, 



2 









W, 



13, 8 



1 



1 



2 



o, o, i 



22 



1 











12, 



2 





1 



2 





27 



16 











18, 



15 





1 



2 



i 



o, *, f , i 



Total, 

















8 



32 



13 



8, 13, 5, 5 





By 1 



iveie 



>hts 



, Total, 



31 cases, 15^ 



success. 



Verificatio 



h, 49 per 



cent. 



A better knowledge of the degree of destructiveness of the 

 storms would give much more rigid results. I have also refrained 

 from giving the occurrence of a large number in any district as 

 much weight as they should have; this would have given a 

 slightly higher percentage of success. 



There is also another important question that I have not 

 touched upon, it is this: The occurrence of tornadoes in any 

 districts must necessarily be connected with their non-occur- 

 rence, i. e., given a large number of districts in which tornadoes 

 are possible, if one had a perfect knowledge of the laws gov- 

 erning the occurrence of tornadoes he would be forced to state 

 or infer that in certain portions of these districts they would 

 not occur, where to the uninitiated there would be an equal 

 probability of occurrence or non-occurrence. Now, since it is 

 difficult to decide what the probability of occurrence is, it would 

 be well nigh impossible to assign a proper weight to a given 

 occurrence, but one thing would certainly seem eminently just 

 and that is that if in any district contiguous to one in which a 

 tornado is predicted, the occurrence of a tornado not predicted 

 receives full weight against a predictor, then the non-occurrence 

 of one in a neighboring district not predicted for should have 

 equal or nearly equal weight in favor of the predictor. It will 

 be seen that if an allowance of the above nature be admitted 

 the 49 per cent already found will be somewhat increased. 



