HYBRIDISATION AND ITS FAILURES. 77 
What then is a definition of a species? The following may perhaps 
answer the question. A species is known by a collection of, presumably, 
relatively constant characters; which may be taken from any or all 
parts of the plant. But how many features are required to distinguish a 
species from a sub-species or variety is a matter of opinion, and will 
always remain debatable. Indeed, the difference between an “ artificial ”’ 
and ‘‘natural’’ system of classification depends greatly on this point: 
for any group in the former is based on one, two, or very few points of 
agreement ; in the latter it is generally on as many as possible. Though, 
in many cases, a single character may coincide with the strictest affinity, 
such as the tetradynamous stamens of the Crucifere, the papilionaceous 
corolla of a great section of Leguminose, &c.; and when we come to 
other large groups with irregular corollas, we find that systematists 
professing to classify plants on a natural system do not hesitate to drop 
into an artificial one when it suits their purpose. For example, Liliacce 
are separated from Amaryllidacee solely by having a superior ovary. 
Yet elsewhere we can find both inferior and superior ovaries in genera 
of the same order, as in Samolus and Primula of Primulacee ; or, again, 
in species of the same genus, as Saxifraga tridactylites, S. wmbrosa, and 
half-superior in S. granulata. 
But although the two orders mentioned above are separated on 
account of this single character alone; yet, testing it by crossing, no 
known attempt to unite two members of these orders has ever yet 
succeeded, as far as I can hear from experimenters. It would seem, 
therefore, that they have been differentiated at so remote a period that 
they have lost all physiological connection. 
So too with genera; the corolla of Snapdragon only differs from 
Toadflax in having a small pouch at the base, which elongates into a 
spur in the latter. I can hear of no cross raised between them. Now 
Ehododendron, Rhodora, and Azalea are as much entitled to be called 
genera respectively, if systematists may separate genera by such slight 
differences as the above; and there is no reason why they should be 
merged into one, solely because they will interbreed : for if interbreeding 
is to be a test, then those polymorphic forms of one and the same 
species that cannot be intercrossed with complete fertility ought to be 
separated, as of Lythrum; to say nothing of Linwm perenne and some 
Orchids which cannot bear seed with their own pollen. 
Ehododendron jasmviniflorwm has a corolla as unlike that of a typical 
Rhododendron as can well be imagined—indeed, Mr. Burbidge likens it to 
Erica Aitonii—and might be regarded, therefore, with justice as a 
different genus; since systematists separate the genera of plants with 
irregular corollas entirely by that organ in many cases—as in the 
Scrophularinee. Now it will cross readily with R. Javanicwm, which 
has the typically formed corolla; but not with the American, or species 
of other countries. On the other hand, Mr. Burbidge crossed R. jasmini- 
florum with an Indian Azalea as the male parent. 
Let us take as another instance, the “genera” Lelia and Cattleya. 
Species of these two have yielded many so-called “ bi-geners”’ ; but are 
they worthy of the name? Now the variations in the forms of the 
flowers of different species of each of these two genera do not differ more 
