M. HOSNY 
283 
are most numerous ! And a superficial argument might be used to condemn the 
character of education in Egypt, or education in general. But it will be clear 
on examination of the isolated values that the observed high correhitions arise 
solely fi'oin the urban character in Egypt of both criminality and education. We 
have endeavoured therefore to correct this by finding the partial correlations for 
constant density of population. 
There now result 
^re,s= --9554 + -0143, 
^rei = - -9231 ±-0242, 
^ri,5= + -7480 ± -0721. 
Thus, while there still remains a quite considerable relation between the prevalence 
of literacy and scholars for constant density, we find that for a constant degree of 
urban conditions, the greater the literacy and the greater the amount of education 
the less will be the criminality. The negative correlations are now even higher 
than the uncorrected positive ones and of course are markedly significant. While 
admitting the slender nature of the Egyptian data, we think that this swinging 
over of the relation of crime and education when we correct for density is sug- 
gestive, and it would be of interest to work out similar correlations for states 
in which the statistics are of a more ample character. It does, however, appear 
reasonable to assert that there is no evidence to indicate that education leads 
to criminality — rather the reverse — in Egypt. 
We will next consider the influence of the presence of foreigners in Egypt. 
We find : 
Means Standard Deviations Correlations 
mo= 99-53, (t,,- 195-60, roc = + -8425 ± -0473, 
mE= 86-88, 183-70, r^^c =+ "9546 ± '0145, 
mG = 119-41, <r,,= 256-61, rgp- + -9429 ± -0181, 
nij = 68-24, ai = 152-04, rj,. = + -9192 + -0254. 
Correlations : 
ru„ = + -9575 ± -0136, r^E = + -9844 + 0050, 
r^g = + -9491 ±-0162, rM=+-9617 ±-01 23. 
Here, if we judged by the raw correlations only, we must assert that the corre- 
lations of crime with the presence of foreigners are so high, that the foreigners 
must be corrupting the Egyptian population. But again the association only 
arises because the criminals and foreigners are both prevalent in the big towns. 
If we correct for density of population, we find the results are very different. Thus 
we have : 
i,ryc=--9811 +-0061, i;r£c= + '1692 ± -1591, 
j^Tqc = + -3524 ± -1433, ^r^c = - "0713 + -1628. 
It is now obvious that the correlation of Europeans other than Greeks and 
Italians with criminality has become insignificant having regard to its probable 
error ; the correlation of the presence of Italians and criminality is now negative, 
but less than its probable error. Thus of Christians only the presence of the 
