XXIV INTRODUCTION. 
whatever. It is true that in some instances I have 
continued to employ it; but rarely otherwise than as a 
secondary character. Under any circumstances it is 
to be suspected. I have seen canina with hairy and 
smooth leaves on the same plant. In that species and 
tomentosa there is every gradation from perfect nudity 
to the most dense pubescence. R. carolina has hairy 
or smooth leaves; so have spinosissima, semperflorens, 
arvensis and many others. Yet I never met with hairy 
leaves on R. fraxinifolia nor naked ones on cinna- 
momea. 
The Stipule proceed from each side of the petiole 
at its base, to which they always have some degree of 
adherence. Certain species, such as bracteata and in- 
volucrata, and even canina occasionally, explain the 
nature of these appendages by producing them in the 
form of leaves, differing from the one to which they 
are attached only in being smaller and less perfect. 
Their modifications are not numerous, but when occur- 
ring, extremely important. The section Banksiane is 
chiefly characterized by their being subulate, nearly 
distinct from the petiole, and deciduous, as in Peaches 
and Nectarines. The pectinate stipules of multiflora 
neatly distinguish it from those in its vicinity. They 
are narrow in majalis; broad in cinnamomea; flat and 
waved in lucida; convolute in carolina and Woodsii; 
with a continuous direction in spinosissima, &c.; sud- 
denly divaricate at the end in sulphurea. In berberi- 
folia they become confluent to the exclusion of the 
leaves, and, their function being altered, assume a 
much more firm and rigid texture than is usual. 
Leaves are always pinnate. Their density may 
sometimes be employed; as in sempervirens, which 
