ON THE PHYLOGENY OF ORCHIDS. 477 
became the now existing family of Zingiberace@ and on the other a 
progeny that developed into the now existing orchids. 
I think the Commelinacee are out of the question. Not because they 
have a superior ovary (a character which is often over-valued in systematic 
arrangement), but because they are a very highly differentiated type, and 
differentiated in quite another way—the calyx and the corolla are quite 
different from each other, and not, as in most orchids, very similar. In the 
second place the inflorescence of Commelinacee is sympodial, not racemose 
as in orchids ; and, if the flowers are not radiate, the plane of symmetry is an 
oblique one, asin Gladiolus. If some stamens are sterile or wanting, as in 
Commelina and Cochliostema, they are not at all identical with those 
which are suppressed in orchids, but their place is also given by the 
oblique symmetry of the flower. I should think it impossible to derive a 
plant with simply racemose flowers and a radiate symmetry, from another 
with the much more complicated sympodial inflorescence and oblique 
symmetry of the flower ; on the contrary I suppose that the Commelinacee 
are one of the most recent and most highly differentiated types of 
monocotyledonous plants. 
Thus it becomes necessary to regard plants similar to the Liliacee and 
Amaryllidacee of present times, as the common ancestors of orchids, 
Scitaminee and Commelinacee. If we take the normal diagrams of 
Amaryllidacee, we have six perianth leaves and six stamens; an inferior 
ovary similar to orchids, but the seeds and embryos are quite apparent. 
The Iridacee may have been derived from the Amaryllidacee by the 
abortion of the inner whorl of stamens, and the Hemodoracee by the 
suppression of the outer whorl. 
We must now examine how far the diagram of Amaryllidacee (in 
some genera) comes near to that of orchids. 
A symmetrical structure of the perianth, in the way that the median 
plane separates two equal portions, is not uncommon in the Amaryllidacee ; 
we may take Hippeastrum, Sprekelia, Alstremeria as examples. But 
in the Conantheree the stamens have a tendency to become transformed 
into staminodes on the same side of the flower as that in which they are 
suppressed in orchids, in the superior half of the not resupinated flower, 
as is pointed out by A. Colla,* J. Miers,r and J. G. Baker.t In Zephyra 
only the two lateral stamens of the outer whorl are staminodial; in 
Tecophilea the median one of the inner whorl also ; in this genus only 
the three stamens are left which occur fertile in orchids. In the same 
genus the two lateral segments of the inner perianth whorl are resupinated 
by twisting, in so much that the perianth also is mediano-symmetrical as 
in orchids. 
In the genus Cyanella, as far as the stamens are concerned, we 
approach nearer to orchids. In this genus, which seems to have convolute 
leaves, there is one species, C. orchidiformis Jacq., which has three posterior 
stamens with long filaments and short anthers, and three anterior stamens 
* “Plant. rarior. in region. Chilens.’’’ Mem. Accad. Torin. xxxix. (1835), 19, 20, 
t. 55. 
+ “On the Conantheree,” Trans. Linn. Soc. xxiv. (1864), 501, t. 53. 
+ “On Colchicacee and the Aberrant Tribes of Liliacee,” Journ. Linn. Soc. Bot. 
xvil. (1878), 493. 
