Sorghum .] 
CL VII. GRAMINEiE (Stapf). 
105 
About 35 spontaneous species in the tropical and subtropical regions of both 
hemispheres, very few extending into the temperate zones ; one group of forms 
widely cultivated in the tropics, particularly in Africa. 
The treatment of the species of the section Sorghastrum does not call for any 
special remark. Those of the section Eu-sorghum, however, are on a different 
footing, and their adequate treatment for the purposes of classification forms a 
serious and difficult problem. Those which come under consideration in this 
work have with two exceptions ( S . purpureo-sericeum and S. versicolor ) been 
placed by Hackel in one vast species, Andropogon Sorghum, the leading idea being 
that they were all derived from one wild ancestor, the old Holcus halepensis, Linn. 
Piper, however, has recently advanced good reasons why this is extremely im- 
probable. He has pointed out that the Linnean Holcus halepensis ( Andropogon 
Sorghum, subsp. halepensis, var. genuinus , Hack.) is a perennial type almost 
confined to the Mediterranean region (sensu lato) and absent from tropical 
Africa which is the home of most of the spontaneous annual forms and probably 
also the cradle of most of the cultivated races known collectively as Guinea 
corn {Andropogon Sorghum, subsp. sativus, Hack.). To these spontaneous annuals 
and the cultivated forms he confines the name Andropogon Sorghum, and dealing- 
in particular with the former he groups them under 11 subspecies, whilst he 
abstains from attempting to classify the latter. Most of Piper’s subspecies are 
here recognised as definite units, but with the status of species, a procedure 
which seems to have the advantage of simplicity and directness, whilst it leaves 
the door open to any theoretical grouping which may in the future be desirable. 
The same reasoning has been applied to the cultivated forms. Hence the 
breaking up of Hackel’s Andropogon Sorghum, var. sativus. Koernicke, who 
made the first comprehensive attempt to classify them, relied for that purpose 
exclusively on characters exhibited by mature infructescences, especially their 
degree of looseness or contraction, and the colours of the ripe glumes and 
grains ; but Hackel in his monograph introduced characters taken from the 
shape of the spikelets. The grain being in most cases the thing aimed at in 
the evolution of these very numerous races, it is clear that artificially intro- 
duced modifications must from the beginning have tended, in the grain-state, 
to obscure or repress ’the phytogenetically important features in so far as 
they were economically indifferent or undesirable. It seemed, therefore, more 
promising to base the primary grouping on the comparison of the flowering- 
stages, which might be expected to be more or less outside the influence of the 
artificially moulding forces of man. Within these primary groups, which are 
treated here as species, nothing- more than a purely artificial arrangement can 
for the present be attempted. An exhaustive treatment of the hundreds of 
races which have been given distinctive popular names would, even if it were 
possible, be beyond the scope of a colonial flora. But above all, it is not feasible, 
owing to the very rudimentary state of our knowledge and of our collections. 
A certain number of forms have, however, been given scientific names and 
technical descriptions, and it has seemed desirable to account for these, whilst a 
few others represent more or less striking new types quite worthy of being added. 
Unfortunately not a few of the descriptions are so inadequate that in the 
absence of authentic material they cannot be assigned a definite place in the 
classification and are therefore dealt with under the heading- “ Imperfectly 
known species and varieties.” Concerning the species admitted here it may 
be remarked that they have been found to coincide to some extent with the 
groups which the American cultivators have empirically learned to distinguish. 
This seems to justify the hope that in a general way they may prove sound, 
however much correction in detail they may need and will no doubt undergo. 
Primary branches of panicle whorled (at least at the 
lower nodes) ; pedicelled spikelets usually de- 
veloped, although sometimes much reduced (par- 
ticularly in cultivated species) ; lodicules ciliate 
(§ Eu-Sorghum, nos. 1-27). 
