a ey 
1904] SHULL—PLACE-CONSTANTS FOR ASTER 352 
tion only two-thirds as great, the number of heads is much more 
nearly adequate. There are presented five empirical modes, 16, 22, 
26, 31, 33, almost in agreement with Lupwic’s series. 
5. Correlations—The correlation between rays and bracts is 
shown in fig. 14, between rays and disk-florets in jig. 15, and between 
Prt CI BRHF we x 
Cr | rliaiit) 
i 2 SREP Swe el 
im Crt 
EET A HEHEHE 
a sit 
Poorer itatr 
ee Tit ' rriitt 
“SRR RSh Bee ee TLitti ete 
Seeenegt H| SHES E SSS 
40 SSGRR SSeS EERE GS 
SEH HIN EEE 
Sanna al Ate iy COC 
a BE a Vike £ we | 
5 Vip At ‘Lik | os 
geet \ Teh ES 
PLS CT TY] OE 
| y/TT\ aH Titi 
| \ iT cI 
0 it! Xr | AGE |e I 3 
i H N ‘| REGS eee 
ot i rt [ 
ad Hoe oem BR BS Sw: 
5 v 1 Bil fo He! 2 
yy \} ri rr 
iy ‘ NPRGe ee x: 
tf i 1 }: CI 
fC: ee Woerrry 
20 ffi Wd | 
me | He 0 (Ba 
Aq Jin ees: 
ia a Wwe rye 
yg J y te MRME BY: +1 
15 i : if we i 
- Mh 
ite ’ t 
s 8 rr 
joenner suline : = 
Mi : rr 
it H rr 
ipepHi Cry 
[ ie 
t | ¥ 
i p= 
| 
5 20 25 30 35 77) 
Fig 
+ 12, . 
1903 saperps Summation curve for rays 1900 + 1903 and the ray-curves for 1900 and 
the summat; for comparison; all reduced to the same area, 500 units; heavy line 
Mation curve; d 
b otted line the r900 curve ; dot and dash line the 1903 curve. 
ha and disk-4 
with those 
Orets in fig. 16. The coefficient of correlation is 
» being greatest between rays and bracts, and least 
nd disk-florets. The coefficients may be compared 
f 1900 in Table H: 
