§2.TheRelationtoExperimentalData3Oursystemofaxiomsisnot,however,complete,forinvariousproblems in the
theory of probabilitydifferent fieldsof proba-bilityhavetobeexamined.
TheConstruction
ofFields
ofProbability.Thesimplestfieldsofprobabilityareconstructedas
follows.WetakeanarbitraryfinitesetE
={|
t,£
2,..
.,£*}andanarbitraryset{pltp 2,..
.,
pk
)ofnon-negativenumbers withthesum
Pi
+
p 2+•• •+
Pk—
1.g
istakenasthesetofallsubsetsinE,andweputP{ft
i,^,...,^}=^
i+
fc+v+^.Insuchcases,
p
up 2 ,...
,
p
karecalledtheprobabilities
oftheelementaryevents
$
1}£ 2 ,...
,$korsimplyelementaryprobabili-ties.Inthiswayarederivedallpossible
finite
fieldsofprobabilityinwhich
gf
consists ofthesetofallsubsetsofE. (Thefieldofprobability
is
called
finiteif the set E is finite.) For furtherexamplesseeChap.II,
§
3.§- TheRelationtoExperimental
Data4We apply the theory of probabilityto the actual worldofexperimentsinthefollowingmanner:
1)Thereisassumedacomplexofconditions,
©,whichallowsofanynumberofrepetitions.2) Westudyadefinitesetofeventswhich
couldtakeplaceasaresultofthe establishmentofthe conditionsS. In
individualcaseswheretheconditionsarerealized,theeventsoccur,gener-ally,indifferentways. LetEbethesetofallpossiblevariantsd,&,...oftheoutcomeofthegivenevents.Someofthesevari-antsmightingeneralnotoccur.WeincludeinsetEallthevari-antswhichweregardaprioriaspossible.3)Ifthevariantof theevents which
hasactuallyoccurred4Thereaderwhoisinterested
inthepurely
mathematicaldevelopmentofthetheoryonly,neednotreadthissection,since
theworkfollowing
itisbasedonlyupontheaxiomsin
§1 andmakes
nouseofthepresentdiscussion.Herewelimitourselves
toasimpleexplanation
ofhowtheaxiomsofthetheoryofprobability
aroseanddisregardthedeepphilosophicaldissertationsontheconceptofprobabilityintheexperimentalworld.Inestablishingthepremisesnecessary
fortheapplicabilityofthetheoryofprobabilitytotheworldofactualevents,theauthorhasused,inlargemeasure,theworkofR.v.Mises,[1]pp.21-27.