GENERAL MEETING. 9 
in size with one of the districts above mentioned, that, on the aver- 
age, a given type of weather—e. g., rain, threatening, fair, hot, 
cloudy, clear—can prevail simultaneously over only 85 per cent. of 
the area. If this law holds for the United States, we can hope for 
no better predictions while the existing system of districts is adhered 
to, for our percentage of verification is now approximately 85. 
Mr. Curtis described the method of verification adopted by the 
Deutsche Seewarte. It differs from that of the U. 8S. Signal 
Service in that the predictions are compared for verification with 
the observations at a single representative station in each district. 
Thus, for northwest Germany, the observations at Hamburg are 
employed. The limits for the prediction of stationary temperature 
are taken as + 1°C., on the basis of an investigation by Hann who 
found that the “change in 24 hours at Hamburg, in two-thirds of 
all cases, averages less than two degrees C.” Mr. Curtis showed 
that, for verifications to be directly comparable with respect to skill 
in prediction, the limits for “stationary” must vary in different dis- 
tricts and at different seasons of they ear. Unless such variations 
are adopted, the verifications should exhibit a uniform geographical 
difference, and an annual period, if the method employed possesses 
any scientific accuracy. As any such change of definition would be 
impracticable, it would seem desirable to base the range allowed for 
“stationary” temperature entirely on physiological considerations, 
leaving the question of comparability for subsequent discussion. 
_ In reply to a question by Mr. Paut, Mr. AxnbE said that the rules 
required that a prediction covering 24 hours should be verified by 
the maps compiled at the 8th and 16th hours, as well as by that 
compiled at the 24th. The desirablity of subdividing the geo- 
graphic sections to which the weather predictions apply was dis- 
cussed by Messrs. GILBERT, PAUL, and ABBE; and Mr. ABBE said 
that if any change was made, it would consist in the abandonment 
of specially defined districts and the substitution of individual States. 
Mr. H. A. Hazen remarked that if the prevailing weather in a 
district treated as a unit actually pertained, on an average, to but 
_ 85 per cent. of the district, then only omniscience could attain to a 
success in weather prediction measured by 85 per cent. of verifica- 
tion. 
