JEFFREY: EVOLUTION BY HYBRIDIZATION 303 



tation. In monotypic genera such as Ginkgo, Liriodendron, Calla, 

 Spathyema, etc., the pollen grains under normal physiological condi- 

 tions of development are all alike and perfect. In Fig. i, Plate V, 

 is shown the pollen of Zanni'chellia palustris, a species isolated in our 

 northern North American flora. It is clear that the grains are strik- 

 ingly uniform and are all well developed. For comparison with the 

 genus just mentioned which has very few species and consequently 

 cannot be considered as highly variable, let us take the common 

 pondweed Potamogeton, of which there are very many species. Fig. 2 

 shows the situation in the large genus just mentioned. The cells are 

 not by any means all perfectly developed and are conspicuously char- 

 acterized by extreme variations in size. It might be maintained on 

 this basis of the illustrations furnished in Figs, i and 2 alone that 

 variability in size of pollen grains is associated with the multiplication 

 of species or in other words with the phenomenon of mutation. 

 Against this view in the forms under discussion may properly be urged 

 the fact that many natural hybrids between species of Potamogeton 

 are known which manifest the usual morphological features of such 

 forms. 



A clearer elucidation of the situation is furnished by the conditions 

 in large genera, where a number of the species coincide both in geo- 

 graphical distribution and in the time of flowering. As a first illus- 

 tration in this connection may be taken the genus Rubus, which has 

 recently been investigated by Dr. Hoar. Fig. 4 shows the condition 

 of the contents of the anther in R. villosns. Clearly the pollen varies 

 greatly in size and perfection of development. A similar condition 

 has been described by the author just cited in a large number of the 

 species of Rubus. The general situation might be interpreted in view 

 of the very numerous and at the same time very variable species of 

 Rubus as an argument for the correlation of mutation and pollen 

 sterility. When however the facts in species of the genus, which are 

 in some manner isolated, are examined quite a difi"erent light is thrown 

 on the subject. Fig. 3, Plate V, illustrates the pollen of R. odomtus, 

 the flowering raspberry, which opens its blossoms at a considerably 

 later period than the mass of the species of the genus. Care has been 

 taken to include as large a number as possible of the grains in the field 

 of view. It is obvious that the variation in size and frequent defective 

 development of unisolated species of Rubus, are conspicuous by their 

 absence. If irregularities in the development of the contents of the 

 anthers were a feature correlated with mutation in the genus Rubus 

 then we ought to find it equally present in isolated and unisolated 

 species. Since that is not the case, the natural inference is that the 

 sterility present in the pollen of species subject to hybrid contamina- 

 21 



