118 
As a third example Poa pratensis L. may serve. This species appears 
in a very large number of forms with panicles of widely different types, as 
illustrated in Plates XVIII and XIX, top rows. These forms were all 
collected by the writer in a small, unoccupied block in the city of Hull, Que. 
The above examples, which perfectly match the panicle variations in 
oats, Avena saliva L., not only illustrate the variability of the panicle in 
grass species belonging to different genera, but also the striking fact that 
the variations run along parallel lines or, as Vavilov (41) has expressed it, 
that they are homologous. 
It should be emphasized that the existence of large numbers of forms, 
having panicles of different types, in Dactylis glomerata, Bromus inermis, 
and Poa pratensis, cannot be ascribed to any mysterious influence of 
cultivation. A similar range of variation, although, as far as it is known 
so far, perhaps somewhat less extensive, is found in other grasses which 
have not been subjected to domestication, for instance Bromus Pumpel- 
lianus Scribn., Bromus Hookerianus Thurb., and related species, Puccinellia 
species, Poa palustris L., Poa compressa L., and others. Whether a grass 
is cultivated or not has no other bearing on its variability than that in the 
former case the closer study of it more readily reveals the existence of the 
various forms. As Vavilov (41, p. 50) says: 
“There is no essential difference in this respect between wild and cultivated plants. 
Wild Linneons, like . . Agropyrum cris latum, Agropyrum repens, . . Alopecurus pratensis . . 
studied in detail at Russian Experimental Stations by plant breeders (Roudzinski, Lorch, 
Jegalov, Bogdan), proved to be no less variable than the cultivated wheats, barleys, oats, 
and peas. The monotypic nature of many wild Linnean species is kept only as long as 
they are studied by a few specimens in the herbarium. The individual study in cultures 
of many samples of the same Linneon inevitably discovers its polymorphic nature." 
Whatever the phylogenetic explanation of the parallel or, if you 
prefer, the homologous variation so commonly found in the family of 
Gramineae, the outstanding fact of interest in this connexion is that many 
Linnean species belonging to different genera display a most remarkable 
similarity in the range and nature of variation in the inflorescence, a 
similarity which, it seems, decidedly points to a common, general law of 
variation. From a taxonomic point of view, the recognition of such a law 
in grasses is of the greatest importance. It means that if certain groups 
of forms, such as Dactylis glomerata , Bromus inermis, Poa pratensis, Poa 
palustris, and Poa compressa, notwithstanding the very wide range of 
variation in the shape of the panicle found in each of them, are universally 
recognized as “species” by taxonomic botanists, there should be every 
reason to exercise great care in basing “species” in general, of panicle- 
bearing grasses, on the characteristics of the panicle alone, a fact fully 
realized by, among others, Fernald and Weatherby (6, p. 3), in the study 
of the genus Puccinellia. 
The forms of the Agrostis stolonifera group display a range of variation 
which, although not fully as wide, as far as it goes is quite parallel to that 
of Dactylis glomerata, Bromus inermis , and Poa pratensis. In the circum- 
stances it seems only reasonable that anyone accepting the latter three 
as species cannot very well, if he is consistent, segregate species within 
the Agrostis stolonifera group on the characteristics of the panicle alone. 
The second outstanding character used by some authors to differ- 
entiate between “species” within the Agrostis stolonifera group is the 
presence or absence of surface runners or stolons. 
