length (L), leaflet width and length to the widest point. Ratios of 
width to length (W/L) and length to widest point to total length 
(WP/L) were determined. Leaflet number, plant height and sex 
of the sample were also recorded. 
Statistical analyses were made with assistance from Dr. David 
Schrader; these analyses of all materials are on file at Florida 
State University. Ken Womble and Melanie Darst helped with 
graphics. 
A total of 377 samples were measured. Most materials came 
from the Mississippi garden. The University of Mississippi 
School of Pharmacy and Dr. C. E. Turner are thanked for 
making their facilities and the plants available. Some garden 
plantings were maintained for several generations: therefore, 
samples were taken only from original seed sources to prevent 
possibility of garden hybridizations. I collected sixty specimens 
representing thirty-nine different seed sources. That somewhat 
extensive population sample was augmented by the intensive 
sampling from the garden in 1972 by Dr. R. E. Schultes and his 
colleagues (with 237 specimens from thirty-two seed sources). 
Vouchers are preseved at the Florida State University and the 
Botanical Museum of Harvard University, respectively. An 
additional eighty specimens from different wild or naturalized 
populations were studied at the Gray Herbarium and Arnold 
Arboretum. 
RESULTS 
Samples were placed in four categories based mainly on growth 
form. The three major forms are illustrated in Fig. 2. Those 
classed as C. sativa were taken from tall, laxly branched plants (S, 
in the tables). Relatively tall plants with very narrow leaflets and 
small achenes were classed as C. sativa, small-seeded (SS); short, 
compact plants that were densely branched were classed as C. 
indica Lam. (1); and those mature plants that were two feet tall or 
less and unbranched, as C. ruderalis Janisch. (R). 
Mean data for leaf morphology for the four categories and 
three collection sets are given in Table |. The four categories in 
my 1976 sampling were all significantly different for W/L and 
62 
