168 HETEEOSTYLED TEIMOEPHIC PLANTS. CHAP. IV, 



sterile with its own pollen; -whilst a long-styled plant of 

 L. regince, though growing by itself, produced fruit. I examined 

 dried flowers from two plants of L. parviflora, both of which 

 were long-styled, and they differed from L. Indica in having 

 eight long stamens with thick filaments, and a crowd of shorter 

 stamens. Thus the evidence whether L. Indica is hetero- 

 styled is curiously conflicting : the unequal number of the short 

 and long stamens, their extreme variability, and especially the 

 fact of their pollen-grains not differing in size, are strongly 

 opposed to this belief; on the other hand, the difference in 

 length of the pistils in two of the plants, their sterility with 

 their own pollen, and the difference in length and structure of 

 the two sets of stamens in the same flower, and in the colour of 

 their pollen, favour the belief. We know that when plants of 

 any kind revert to a former condition, they are apt to be highly 

 variable, and the two halves of the same organ sometimes differ 

 much, as in the case of the above-described anther of the 

 Lagerstroemia ; we may therefore suspect that this species was 

 once heterostyled, and that it still retains traces of its former 

 state, together with a tendency to revert more completely to it. 

 It deserves notice, as bearing on the nature of Lagerstrcemia, 

 that in Lytlirum hyxsopifolia, which is a homostyled species, some 

 of the shorter stamens vary in being either present or absent ; 

 and that these same stamens are altogether absent in L. thymi- 

 folia. In another genus of the Lythraceae, namely Cuphea, three 

 species raised by me from seed certainly were homostyled ; 

 nevertheless their stamens consisted of two sets, differing in 

 length and in the colour and thickness of their filaments, but 

 not in the size or colour of their pollen-grains ; so that they 

 thus far resembled the stamens of Lagerstrcemia. I found that 

 Cuphea purpurea was highly fertile with its own pollen when 

 artificially aided, but sterile when insects were excluded.* 



* Mr. Spence informs me that phic: but he did not notice the 



in several species of the genus length of the pistils. In the 



Mollia (TiliaoesB) which he col- allied Luhea the outer purplish 



lected in South America, the stamens are destitute of anthers, 



stamens of the five outer cohorts I procured some specimens of 



have purplish filaments and green Mollia lepidota and speciosa from 



pollen, whilst the stamens of the Kew, but could not make out that 



five inner cohorts have yellow their pistils differed in length 



pollen. He therefore suspected in different plants ; and in all 



that these species might prove those which I examined tho 



to be heterostyled and trimor- stigma stood close beneath the 



