H. A. Hazen — Verification of Tornado Predictions. 127 



§ 19. Such fractional units are convenient for the purpose of 

 comparing the varying condensation in species belonging to the 

 respective groups, but it will be borne in mind that in order to 

 construct formulas which shall represent the true equivalent 

 weights of liquid and solid species, we must multiply instead of 

 divide the formulas hitherto accepted as representing the nor- 

 mal species. The combining weight of these must be the unit 

 for fixing the equivalent weights and the true formulas of such 

 liquid and solid species ; which are generated by integration or 

 polymerization from the normal species. This, though known to 

 us in the volatile elements, and in compounds like carbon dioxyd, 

 water- vapor, formic and acetic aldehydes and pentine, is un- 

 known in the case of bodies like carbon, silicon, silicon dioxyd, 

 and most of the solid J oxyds ; as also in the various silicates, 

 carbonates, sulphates and phosphates. For all these we choose, 

 as representing the normal species, the simplest formula which 

 satisfies the relations of valency, and which corresponds to the 

 theoretical gaseous or volatile species. 



The writer has sought in the preceding pages to illustrate 

 and set forth in detail several points which were more briefly 

 noticed in his lately published volume entitled "A New Basis 

 for Chemistry," to which this paper may be considered as a 

 supplement. 



Art. XV. — Verification of Tornado Predictions ; by H. Allen 



Hazen.* 



In taking up the study of the verification of tornado predic- 

 tions, we must carry in mind the exact words of the prediction 

 which are: '"Conditions are favorable for the development of 

 tornadoes in region . . . ." It is generally understood that in 

 these predictions the country east of the 102° meridian is 

 divided into eighteen districts, and tornadoes are predicted for 

 in each of these. For the present study the predictions for the 

 month of June. 1885, are to be discussed. One scheme of veri- 

 fication that has been suggested is that of Prof. Gilbert, of 

 the Geological Survey, published in American Meteorological 

 Journal for September, 1884. Prof. Gilbert divides the pre- 

 dictions and occurrences into three general classes. 1. Suc- 



* In the number of the American Meteorological Journal for July, 1884, Mr. Fin- 

 ley published a statement as to success in tornado prediction claiming over 97 per 

 cent. This statement aroused quite a lively discussion in Science and other jour- 

 nals, notably one by Prof. Gilbert, in which he credited Mr. Finley with 23 per 

 cent. In October, 1885, several persons were desired by Mr. Finley to take up 

 the discussion, and the following paper is now presented as written then, it hav- 

 ing been found impossible to publish it before this, through circumstances beyond 

 the writer's control. — Washington, D. C, July 4, 1887. 



