320 
The most important substance that they produce is amylaceous matter, 
which exists in great quantity in some of them, which hence become of incal- 
culable value as aliment for man ; such are all the Com tribe. Plantains, and 
some Palms, which contain it in the fruit : the Sago and other Palms, in which 
it occupies the trunk ; and the eatable Araceous plants. Orchises, Yams, &c., 
in which it is found in the root. Sugar, gluten, oil, and aromatic principles, 
are also frequently met with in Monocotyledons ; but, as Humboldt well re- 
marks, acids, bitters, resins, camphor, tannin, milk, or poisonous matter, are 
either w'hoUy granting or very uncommon. The latter chiefly exists in Araceae, 
some Amar}dlidaceae, and Melanthacese. 
Tlie orders of Monocotyledons are given in the state in which they now 
exist ; but it must be confessed that the characters and limits of many of 
them are far from satisfactory. The whole of those which border upon 
Lihacese require to be reconsidered by some botanist who is in possession 
of the means of examining them in great detail ; their actual condition is, 
no doubt, attributable to the partial -vdew that has hitherto been taken of 
them. 
It seems to me that they naturally divide into six principal groups, of which 
one has the ovary adherent with the calyx (Epig}Tios0e) ; another the same 
character with the stamens and styles combined into a sohd central column 
(Gynandrosae) ; a third with a superior ovary, and the floral envelopes for the 
most part petaloid (H)q)Og}Tiosae) ; a fourth with Exogenous leaves (Retosae) ; 
a fifth with the floral envelopes either altogether absent, or in a very imper- 
fect state, analogous to those of Incompletae among Dicotyledons (Spadicosae), 
and a sixth with the true floral envelopes altogether absent, their place being 
supplied by alternate bracts, which are packed one over the other about the 
ovary (Glumosae). And I think, whatever may be the changes that the limits 
of the orders themselves may undergo, such groups must be considered 
natural. If their mutual affinities are attentively considered, it will be 
seen that as usual no linear arrangement is capable of expressing their re- 
lationship. Let us begin their comparison where we wdU, it can only end in 
their arrangement in a circle. Suppose, for example, omitting Retosae, which 
do not afibct the illustration, we take Epig^mosae to start from. That group 
contains the genus Gladiolus, which, if its stamens and style were consolidated, 
would be almost an Orchidaceous plant ; therefore, Epigynosae and Gynan- 
drosae are in contact. On the other hand, the Bromelial alliance in Epigynosae 
is so allied to the Pandanal in Spadicosae, that we cannot doubt these two 
groups also touching. G^mandrosae are also in like manner placed as it 
were between Epigvmosae and Hypog}Tiosae, Scilla, or some such regular- 
flowered Asphodeleous plant, being a clear aUy of such genera as Thely- 
mitra among Orchidaceae. Now let us see what further affinities can be traced 
from Spadicosae and Hypogynosae elsewhere. The latter contain the Juncal 
alliance, in both whose orders the perianth is so nearly glumaceous as to leave 
no doubt of their connection with Glumosae ; on the other hand, the Typhal 
alliance in Spadicosae approximates so very nearly to Glumosae, as for instance 
Typha to Pennisetum, and Sparganium to many Carices, that the connection 
of H\'pog}Tiosae and Spadicosae may be said to be nearly equal wdth Glumosae. 
If these affinities are expressed on paper by the position the groups really 
occupy, we shall have the latter thus — 
Epig\Tiosae, Gynandrosae, 
Spadicosae, H}q)ogynosae, 
Glumosae. 
But this is a kind of connection that cannot be preserved in a linear arrange- 
ment. In my Key to Structural, Physiological, and Systematic Botany, I at- 
tempted to manage by beginning with Epig^mosae and ending wdth Gynan- 
