94 
PHILOSOPHICAL SOCIETY OF WASHINGTON. 
31st Meeting. May 25, 1887. 
The Chairman presided. 
Present, sixteen members and one guest. 
Mr. M. H, Doolittle continued his paper on 
ASSOCIATION KATIOS, 
which consisted chiefly in a review and criticism of an article* by 
Mr, G. K. Gilbert on Finley’s Tornado Predictions. The follow¬ 
ing is an abstract of Mr. Doolittle’s paper: 
By Mr. Gilbert’s notation: 
s = whole number of predictions. 
jp = number of positive predictions. 
0 — number of occurrences. 
c — number of verifications of positive predictions. 
2 ^' = s — p~ number of negative predictions. 
o' ~ s — 0 == number of non-occcurrences. 
0 — c = “ measure of the failure in inclusion.” 
p — G== “ measure of the failure in exclusion.” 
We should also have for consistency 
c = measure of success in inclusion. 
c' = s — 0 — p G= measure of success in exclusion. 
He says, “If inclusion and exclusion are equally important their 
measures bear the same weights.” This is as true for success as for 
failure, but Mr. Gilbert gives to the measure of success in exclusion 
no weight whatever, and entirely disregards it. 
He has 
0 + p — 2c = measure of the general failure of prediction. 
We should also have 
s — 0 — p 2 c ~ measure of the general success in prediction. 
The latter measure consists of the “ favorable cases; ” the former 
* American Meteorological Journal. 8°. Detroit, Micli., 1884, Sep¬ 
tember; vol. 1, no. 5, pp. 166-172. 
