ae 
FRETS, THE INDEX CEPHALICUS. 491 
For this Swedish material [ have applicated also the formula of 
PEARSON (1900) to test 
the goodness of fit ac- 
cording totheindications 
of ELDERTON (1901; p. 
155) I find that in this 
case the fit of the nor- 
mal curve is extremely 
bad. Also for my mate- 
rial I find that the fit 
is bad. The normal curve 
of the 900 Bavarian 
skulls (tab. 2) shows a 
very good fit. 
When I see, that for 
the Swedish material the 
discrepancies between 
„actual and normal distri- 
bution, that are revealed 
by the calculation of the 
standard error of sam- 
pling, are grouped regu- 
larly (see tab. 4), and 
that there are similar 
but smaller differences 
between actual and nor- 
mal distribution in my 
material (p. 482, tab. 1 
and diagrams 1—3), so 
I am inclined, ‘to attri- 
bute the fact that the 
curves do not fit, to the 
very large material with 












































































































6006! | ie | 
Te ellen 
sel | [ 
Ve 
| ] SL 
JE En + L 4 
ea 
5000 JL 
Sar il ia 
; Ton 
En ma il 
4300 r 
F HE 
asp 8 EE - 
| N 1 JE 
ie HEE 
mm Pataki 
HET | Ed 
Fe Led all |_| 
060 (2 
+ —+ | k 
CNE | 
1500 En I! = = 
a a 1 DE 
a0 | iE El 
ne et 
= ++ = 
T EIER at LEA 
saal 
EEZ 
"ON hep aD ap He DH AS eH PTH Mod Ga Hh Och wy Goh EE 




Diagram 4. 44941 Swedish recruits. 
Observations. 
hen Normal Curve. 
which we have to do. TocHER (1906) for 4000 males and almost as 
many females gives normal and skew curves. The diagrams show 
clearly that the fit of the skew curves is much better than of the 
normal curves. TOCHER’S conclusion of his calculations is that the 
fit of the normal curve is very bad. He emphasizes that the appli- 
