28 
enclosed within the flowering glumes and consequently caught a sub- 
stantial portion of the released pollen. When, after the pollination had 
taken place, the glumes had closed again, one or more of the empty anthers 
were in many cases found hidden inside the glumes, and not uncommonly 
one of them was found squeezed in between their edges, but never hanging 
outside as is the case in most other grasses. The pollination was studied 
in quite a number of the Western Rye grass forms which occur in the 
Edmonton district and, as a result, the writer became satisfied that, in 
the first place, self-pollination apparently is the rule and, secondly, that 
the chances of the pistil of a flower being fertilized by pollen from either 
another flower of the same plant, or from a flower of another plant, are 
rather small. In other words, in the matter of pollination and fertilization 
Western Rye grass behaves like cultivated wheat. 
After these observations it was only natural to suspect that the num- 
erous forms of Western Rye grass might normally breed true to type. In 
order to ascertain whether they actually breed true, seed w f as collected, at 
Calgary, Alberta, in the autumn of 1913, from nine individual plants 
representing conspicuously different types. The nine plants were found 
growing together on a vacant lot in the outskirts of the city, so close together 
that the writer gathered seed from several without moving. The seed thus 
collected from each individual plant was sown the following spring, at the 
Central Experimental Farm, Ottawa, in separate pots. From the seedlings 
secured, quite a number were picked out at random and planted in rows 
some distance apart, each row representing the progeny of one single plant. 
A year later, when the plants had reached full development, all doubt about 
the various forms breeding true to type was removed. There was not the 
slightest variation in the progeny from any one of the nine mother plants, 
every row being perfectly uniform. The results obtained have since been 
amply substantiated, and numerous experiments on a larger scale have 
established the fact that the large number of forms of which the so-called 
Agropyron tenerum is composed, normally breed true to type. According 
to Kirk (5, page 240), who has made extensive investigations, segregation 
may, however, occasionally occur, indicating that some plants are hetero- 
zygous for one or more characters. Kirk, therefore, concludes that some 
natural crossing takes place, but the extent to which it occurs has not been 
determined. 
Although at the time primarily interested in the awnless of short- 
awned A. tenerum, the opportunity to test the related, native, long-awned 
plants going under the names of A. caninum (L.) Beauv., A. Richardsoni 
Schrad., A. caninoides (Ramaley) Beal, etc., was not lost. As was expected, 
these also proved to be normally self-pollinated and bred true to type. 
At the same time experiments were also conducted with the European 
A. caninum , seed from individual plants of which was secured through the 
kind co-operation of Dr. H. Witte, of the Plant Breeding Institution, 
Svalof, Sweden. During these experiments the observation w'as made 
that, during an thesis, the European A. canmum behaved conspicuously 
differently from its North American relatives. In the European A. cani- 
num the spikelets became spreading during anthesis, forming an angle of 
about 45 degrees with the axis of the spike, somewhat after the fashion of 
A. repens (L.) Beauv. After flowering, the spikelets moved back towards 
the axis, but did not, in the forms under observation, become wholly 
