58TheIslamicateContextexistbeforeIslam.Italso
explainswhyso
manyoftheslavesol-dierscamefrom
steppeanddesert
regions.Yeteventhe
stirruptheorycorrelatesfarlesswellthanIslamdoes.Ifitwasthedecisivefactor,why,then,
didtheothernon-horse-breedingareas(Hindu
IndiaandChinain700-1400
c.E.)notalso
developmilitaryslavery?Furthermore,manymilitary
slaveswerefootsoldiers,liketheblack
troopsinEgypt
andtheJanissaries,therenowned
elite
corpsoftheOttoman
Empire.Something
besidesthe
stirrupmusthavebeeninvolvedintheroleofslavesoldiersinIslamdom.The
strikingcorrelationbetweenIslamand
militaryslaveryshownin
chapter2 raisesthe
possibilityofacausalconnectionbetweenthisreligionandthis
patternof
militaryrecruitment.However,
correlationalone,asstatisticiansandlogicians
nevertireof
pointingout,doesnot
implycausation.Somecorrelationsare
merelyunrelated;nooneclaims
significanceforthe"near-perfectcorrelationsbetweenthedeathratein
Hyderabad,Indiafrom 1911 to 1919,andvariationsinthe
membershipofthe InternationalAssociation ofMachinists
duringthesameperiod.’’9Others arecounterlogical:statisticsindicatethatthepresenceoffireenginesatfirescorrelateswithmoredestructivefires;canoneinferthatfireswouldbelessdestructiveifnofire
engineswere
dispatched?10High
correlationbetweenIslamand
militaryslavery"canserveonlyasthestartingpointforfurtherinvestigationandanalysis.’’11In
general,"toestablishthe
reg-ularisticcausal
propositionthatXcausedY,three
thingsmustbedemonstrated.First,theremustbeacorrelationbetweenXandY.Second,theremustbea
propertemporalrelationship
intheiroccurrence,
Ximustoccurbefore
Y.Third,theremustbeatleasta
presumptiveagencywhichconnectsthem.’’12InthecaseofIslamand
militaryslavery,correlationdoesexistandthetem-poralsequenceis
proper(Islamprecededmilitaryslavery);
to9.D.H. Fischer,Historians’Fallacies:Towardthe
Logic
ofHistorical
Thought(NewYork,1970),pp.
168-69.10.R.M.MacIver,SocialCausation(Boston,1942),
p.92.11.Ibid.,p.93.- Fischer,
ttistorians’
Fallacies,p.169.