102 The lithahitaiits of the Eastern Mediterranean 
TABLE VII. 
Differences hetiveen Coefficients of Variation. Head Lengths and Breadths. 
Locality 
Diffeience 
Leukas 
Meligaki 
Mani 
Ci'cte, Pyrgiotissa... 
„ Seliiios and Sphakia 
„ Whole island 
+ -07 + -36 
-i-\2+-:iS 
- -11 + -24 
- -41 + -34 
+ -73 +-35 
Locality 
Lycian Gypsies 
(_'y])rus, Group I 
„ „ n 
„ „ in 
„ IV 
Gioiqis combined 
Diffeience 
+ 
■52 + 
•28 
■00 + 
•21 
+ 
•30 + 
•16 
•03 + 
•18 
+ 
•16 + 
•26 
+ -28± 
•11 
+ indicates that the coefficient of variation of the head length is greater, - that that of the head 
breadth is greater. 
Upper Facial Index. 
The upper facial index has been reduced for a small series (see Table VIII), but 
after inspection the conclusion I arrived at was that on the whole as a racial test 
TABLE VIII. 
Upper Facial Index. Living Adult Males. 
Provenance 
Number 
Mean 
cr 
V 
Leukas 
42 

52^76 ± -30 
3-09 + 
•21 
5^85+ ^38 
Meligala 
33 
56^85+ ^47 
4-02 + 
•33 
7^08+ -59 
Mani 
98 
57 ■45+ -25 
3^73± 
-18 
6-48 ± -31 
Crete, Selinos and Sphalcia ... 
49 
53-00+ -28 
2-95 + 
-20 
5-56+ -37 
Albania 
138 
48^42+ •IS 
3 •20 + 
•13 
6-61 + -27 
Lycian Gj'psies 
53 
54^23± ^29 
2-88 + 
•22 
5^31 f40 
Gvprus, Group I 
111 
50^10+ ^25 
3-92 + 
•18 
7-82+ -35 
„ II 
221 
48^63+ -12 
2-69 + 
•09 
5 -.53+ -18 
in 
167 
49-71 + ^]9 
3^61 + 
•13 
7-26+ ^27 
„ IV 
87 
50^90+ ^28 
3^87 + 
•20 
7^60± -39 
Groups combined ... 
586 
49-52+ •lO 
3^71 + 
•07 
7 -50 ± -15 
this index did not appear to be of great value, at least within the area under con- 
sideration. The reasons for this conclusion are that the indices fall into three groups: 
(1) Meligala and Mani. 
(2) Crete, Lycia and Leukas. 
(3) Cyprus. 
W^e have previously found that on the basis of other measurements the extremes 
are Lycia on the one hand and the Sphakiots and Selinots on the other. It would 
appear thei'efore that some special factor was at work differing from those that we 
have previously met with. 
It has been suggested by Pearson and Crewdson Benington <-■'' that the personal 
ecjuation in measurement may be responsible for this divergence. Until Hawes' 
