ANGIOSPERMS. MONOCOTYLEDONS. 439 



In the large division of Liliiflorae, in some Spadiciflorae, in many Enantio- 

 blastae, Juncagineae and Alismaceae 1 this formula is at once obtained empirically; in 

 most other cases 'single members or whorls are wanting, but their suppression is 

 usually at once suggested by the position of the parts that are present. In the Scita- 

 mineae with only one or even with only half an anther the rest of the members of the 

 androecium are present in an imperfect condition, having been changed into petaloid 

 staminodes. The flower of our European Gramineae has no perianth but is enclosed 

 in two paleae, the lower of which is the bract, the upper the bracteole of the flower. 

 The two small scales, known as the lodicules, which force the paleae apart by their 

 strong turgescence and thus cause the flower to open, have been regarded as a rudi- 

 mentary perianth. It is true that they are by their origin lateral portions of one leaf- 

 rudiment, as the history of development in some cases shows 2 . But they are not a 

 perianth, but the flower is enclosed in a number of bracts arranged in two rows, the 

 two lowest of which are the paleae, while one of the upper ones is rudimentary and its 

 lateral portions (or in place of these two smaller independent rudiments) are the lodicules, 

 and the fourth is usually quite abortive, but in the Stipeae is developed as a pos- 

 terior lodicule. The arrangement of the perianth-leaves is different in some of the 

 tropical Gramineae. In Oryza for instance the second staminal whorl is developed, 

 in Bambusa also and in some others. It has been already pointed out that the flower 

 of the Orchideae may be referred to the pentacyclic trimerous type ; the following 

 theoretical diagrams will show the same thing in the case of the more important of 

 the other families. 



If we take the pentacyclic flower with the formula Sn Pn Sfn -f- n Cn ( + n) as 

 typical for Monocotyledons, it appears that in the great majority of families in which 

 the numerical relations deviate from the type 3 , the difference is simply that single 

 members or whole whorls are wanting, but the typical position of those that are 

 present is not disturbed ; it is to the effects produced by abortion therefore that the 

 variety in the forms of flowers in this class is chiefly due ; the cases consequently are 

 not rare in Monocotyledons, in which abortion is carried to such a point that nothing 

 is at last left of an entire flower but a single naked ovary or a single stamen, as is the 

 case in many Aroideae 4 , in which this mode of explaining the present state of the 

 flowers is rendered easy and obvious by the occurrence of typically constructed 

 flowers and of a variety of intermediate forms in which partial abortion only has taken 

 place. It is chiefly in small and closely crowded flowers that a great reduction of 

 the typical number of parts is observed, as for example in the Spadiciflorae, while in 

 large flowers placed further apart the whorls are usually present in full number or in 

 more than their full number (Butomus, Hydrocharis]^ and any deviations consist 

 chiefly in the appearance of petaloid staminodes in the place of fertile stamens, as in 



1 The dimerous flowers of Potamogeten Si Pz St2 + 2 C 4 (see Hegelmaier in Bot. Ztg. 1870, 

 p. 287) differ only in the simultaneous appearance of the four carpels, which are placed diagonally 

 to the previous pairs. The leaves of the perianth are formed in this case as outgrowths (scales from 

 the connective) of the stamens, as in the case undoubtedly in Ruppia. 



2 Hackel, Untersuch. ii. d. Graser in Engler's Jahrb. I. p. 336. 



3 Compare what is said on the subject of abortion on p. 414 and in the introduction to the 

 Angiosperms. 



4 Engler, Araceae in De Candolle's Monographiae Phanerogamarum, Vol. II. 1879. 



