THE STUDY OF THE EVOLUTION OE THE ELOWER. 277 



group unite to form one (m), so that four (dl, d2, ml, m2) traverse the ovary-wall 

 (Text-fig. 8, B). Each of the bundles m forms an ovular branch o (C), and then extending 

 upward divides to form two (E, m', m''), which become the opposite marginal bundles 

 of the free styles (F) ; dl and c? 2 become the median bundles of the corresponding 

 styles (F). 



The ovary corresponds closely to that of Hamamelis and other Hamamelidacese, con- 

 sisting of two carpels with their respective traces, the adjacent marginal bundles being 

 congenitally united. 



Nature and Origin of the Flower, 



m 



Notwithstanding his interpretation of the morphology of Gar?ya, Lindley * found some 

 difficulty in assigning it a systematic position. He first selected the " Urtical Alliance," 

 but later decided in favour of the HelwingiacesB, then recently established by Decaisne, 

 and accordingly associated the monotypic family Garryaccse with the IlelwingiaceaB to 

 form the alliance or cohort Garryales. Both families agree, writes Lindley, in possessing 

 an epigynous calyx and a minute embryo embedded in copious endosperm. 



Garri/a, bearing the hall-mark of epigyny, is hencRforth compared with epigynous 



forms, even by Baillon, who sceptically regarded Lindley's calyx. Baillon f included 

 Garrya in his Cornacese. He supposed that the ovary, from an evolutionary point of 

 view, is intermediate in structure between those of Aucuba and Cornus. This Aucaha- 

 Garry a-Cornus series of his is analogous to that which he believed could be traced 

 between the Combretacese and bilocular forms of Marlea. 



It is now evident that the true nature of the flower has long been masked, owing to 

 the vagaries of bracts. Bracts are inserted on the ovary-wall in several species, and in 

 the common species, G. elliptica, reach an epigynous position. In G. Fadyenii, where 

 the folioles do not exist, the perianth was supposed to have disappeared ; thus Endlicher % 

 states, "perigonii in floribus pistilligeris vestigium nullum." Bracts, again, caused 

 Bentham to err when he described opposite stamens in G. laurifolia : the stamens are 

 alternate with the perianth segments. Finally, the close proximity of a pair of bracts 

 to the perianth segments in the male flower led Harms astray, who believed that they 



represented a calyx. 



The loss of the epigynous calyx or pe7^ianfh, which had been considered homologous 

 with the perianth of Cornacege, involves a total reconsideration of the whole problem. 



Wangerin, influenced by the amentaceous inflorescence, dicecism, apetalous and 

 hypogynous flowers, compares Garrya with Salix, Betiila, Juglans, and Myrica, and 

 removes it from the Cornaceje to the restored Garryaceae. This family, in Engler's 

 revised Syllabus, is re-established in the Garryales, now situated between the Salicales 

 and Myricales. Both Wangerin and Engler seem to have regarded certain characters 

 as of little consequence— for instance, the basal " orthotropous " ovules of Juglam and 



• J. Liudley, Veg. King. 2nd ed. 295. 



t H. Baillon, Histoire des Plantes, vii. 82 



t S. Endlicher, Gen. PI. Suppl. ii. 30. 



