C. PeLLEW and F. M. DtTRHAM 173 



pollen or occasionally no pollen. We have no evidence as to the 

 inheritance of the female sterility which is so marked a feature of ||. 

 Cases in which no seed is produced could be ecjually well accounted 

 for by deficiencies in the pollen. 



It is a remarkable feature of this group of plants that, though {f 

 gives little st ed which rarely germinates, the seed of its dca-ivatives 

 germinates freely. 



Dimorphic structure of the flowers. 



The structure of the flowers of || is like that of a thrum floribunda, 

 in that the anthers are at the top of the corolla tube and the style is 

 long, the stigma protruding beyond the anthers. The plant is hetero- 

 zygous, giving thrums and pins on selling and also on crossing with 

 a floribunda which is heterozygous for these characters. Among the 

 offspring of jf thrums and pins occur, but within these two forms are 

 numerous minor variations affecting the position of the anthers and the 

 length of the style. We have sometimes had difficulty in deciding whether 

 a plant was thrum or pin by its appearance, for jjlants occur in which 

 the anthers are only half-way up the tube and the style long. On 

 breeding from these intermediate forms we have generall\' found that they 

 were capable of throwing some pins with the anthers at the base of the 

 tube, and we have therefore accepted them as thrums. Microscopical 

 examination of the pollen gives little help in deciding whether a plant 

 is thrum or pin, for the grains are very irregular in size and shape and 

 many of them are bad. The pollen of pins is generally smaller than 

 that of thrums, but in some pins grains as large as those of thrums 

 occur. 



We have found that the pins breed true exce[)t in one case which 

 we regard as due to an error (p. 1G7): the thrums breed true or throw 

 both forms. The ratio of thrums to pins varies somewhat in different 

 families, but there seems little reason to suppose that the two forms are 

 not members of a simple allelomorphic pair. The irregulai- n umbers 

 obtained are perhaps due to the sterility which exists in var3'ing degrees 

 in all the families derived from -ig and its crosses. 



Flower colour. The two flower colours yellow and i.sabellina occur 

 in various shades. Besides these, a new yellow colour has appeared 

 which we have called lemon. It is very near the pale yellow of certain 

 stva.ms oi verticillata but as a rule diti'ers from these in that the margins 

 of the petals are darker. Before giving the more complicated experi- 

 ments in which the lemon colour is involved, we will consider our 



