E. R. Saunders 311 



both cases every pollen grain is carrjang the double character. This 

 view receives strong confirmation from the results of cross-breeding. 

 184 pollen grains were tested by crossing with a pure-breeding form. 

 From the mixed character of the F^ families it was definitely ascertained 

 that 181 of these grains must have been carrying doubleness. The 

 absence of doubles in the 3 remaining families can scarcely be regarded 

 as other than accidental, since if genuine it would presumably imply 

 the production by the double-thro\ving forms of a certain number of 

 single-carrying pollen grains, a condition which is not borne out by the 

 results of self-fertilisation. 



II. Races which occur both in the form of double-throwing 

 and non-double-throwing strains. 



The question now arises as to the behaviour of those races which 

 can be obtained both in a pure-breeding and in a sporting form. Are 

 these d-strains also strictly eversporting ? In these cases is it also 

 impossible to breed out the doubles ? From the results which have 

 now been obtained it would seem that to these questions we may safely 

 return an affirmative answer. It will however be convenient to consider 

 the evidence from the sap-coloured and the non-sap-coloured forms 

 separately. 



Commercial seed of both double-throwing and non-double-tbrowing 

 strains was obtained in the case of the two glabrous non-sap-coloured 

 forms white and cream, and of several sap-coloured forms, viz. very light 

 purple or azure (both hoary and glabrous), light purple, dark purple, 

 marine blue, flesh and copper (all glabrous)^ The seed supplied as 

 giving only singles was found, as previously stated, to answer to 

 description ; in no case were doubles obtained from such seed either 

 when the strains were self- fertilised, or bred together. The strains 

 stated to give doubles were tested both (1) by self-fertilisation which 

 affords the readiest means of detecting the sporting individual, though 

 it leaves undetermined the share in the results to be attributed to 

 pollen and ovules respectively ; (2) by crossing with pure-breeding 

 strains, a method which enables us to sample ovules and pollen inde- 

 pendently of each other. In the latter case the experiment has to be 

 carried to F, before a result is obtained. 



^ Unless otherwise stated all races employed in these experiments were of the Ten- 

 week class. 



21—2 



