KXOAYLEDGE OF THE MUTATING 0EN0TIIEEA9. 41 



general way as do 0. Lamarcldmia and 0. rubrinervis. And more particularly, tbey 

 differ in the same as do the " twin " types first deserihed by DeVries, which are obtained 

 in the Ej when certain small-flowered forms, such as O. biennis and 0. mu?^iGata, are 

 pollinated from O. LamarcMana or its derivatives. It, therefore, seems probable that 

 this twin hybrid behaviour depends upon a double alternative (broad- or narrow-leaved) 

 condition in the wild state. My cultures of 0. muricata from such widely separated 

 regions as Middleton, N.S., St. John, N.B., and Winnipeg, Manitoba, in Canada, have 

 revealed in each case two distinct races, the two being respectively identical from all 

 three localities. This shows not only the wide distribution of these particular types 

 East and "West on the continent (though probably confined to a narrow range of 

 latitude), but since the two types were found in each case from a single collection of 

 seeds, and since they occur in strictly self-pollinating forms having the flower-characters 

 of O. muricata^ it is evident that this twin behaviour can hardly be due to pollination 

 from large-flowered forms. This is especially true since no large-flowered forms are 

 known to occur in the regions where these seed collections were made, and they are 

 certainly absent at least from the Nova Scotia region. 



In short, then, we have to account for the occurrence of broad-leaved and narrow- 

 leaved " twin " types in races which it is reasonably certain were purely self-pollinated. 

 And the same behaviour was repeated in collections of seeds from three widely separated 

 areas. All races of 0. muricata do not show two types, however, several collections 

 of seeds having yielded uniform races with only one very constant type. 



The broad-leaved and narrow-leaved types from my collection of seeds from Winnipeg, 

 are represented respectively in PI. 4. figs. 54 and 55, while fig. 56 shows a plant of the 

 broad-leaved race from Nova Scotia, in flower. Its rosette is identical with that of 

 fig. 54. I have as yet been unable to test the behaviour of the twin races, except to 

 show that the broad-leaved type breeds true. Seeds of O. muricata sent to me from 

 the Bremen Botanical Garden were found also to contain two types which differed but 



V 



slightly if at all from the plants from the three localities in Canada already mentioned. 

 The total numbers of broad- and narrow-leaved individuals from these four sources were 

 respectively 30 and 27, indicating possibly a tendency to appear in equal numbers. It 

 must also be said that the two types were very distinct, as represented in the photographs, 

 without any tendency to be obscured by variation. 



Another lot of seeds, received from the Kiel Botanical Garden under the name 

 O. biennis, yielded 37 plants belonging to two very distinct types : type (1) (26 plants) 

 had very broad, large, light-green rosette-leaves; type (2) (11 plants) had narrow, 

 dark-green leaves with long unmargined petioles. These two types were both unlike the 

 previously mentioned " twin '* types, and were so unlike that it seems probable they may 

 have been thrown together by an accidental mixing of seeds. 



On the other hand, I have collected seeds near Woods Hole, Mass., which yielded a 

 single uniform race of O. muricata, apparently corresponding with the var. canescens 

 of Robinson, and remaining true in the Fa- Seeds kindly sent by Mr. E. L. Rand, 

 from near the beach at Seal Harbour, Maine, also yielded a single, uniform race with 



SECOND SERIES. — BOTANY, VOL. VIII. 



G 



