SYSTEMATIC POSITION OF MITRASTEMON. 209 



hesitate to refer this plant to the Rafflesiacese, are its having an entirely 

 superior ovary and mitraformed united stamens (HAYATA I., p. 112 ; Makino 

 II., p. 253). As to the difference of the staminal structure, however, the 

 connection of Mitrastemon with rafflesiaceous plants is to be found in the 

 stamens of Pilostylis (SoLMS-LAUBACH I., p. 13 ; LOTSY p. 881) where the 

 one-celled anther-cells are arranged in a few series around the stigma- 

 tic disc, and in those of Mitrastemon where one-celled anthers are arranged 

 in many series and in an irregular way on a staminal tube covering the 

 style and stigma. Regarding the difference of the position of the ovaries, 

 the relation is also to be found in ApodantJies which sometimes has a 

 semisuperior ovary (BENTHAM et HOOKER III., p. 118). Moreover, the 

 position of the ovary cannot by itself to be considered as a sufficiently 

 reliable character to determine the systematic position of a family. As we 

 see in many families in every division of phanerogamous plants, many cases 

 occur in which plants having a superior ovary and others having an inferior 

 ovary are included in one and the same family, and there are even cases in 

 which plants having ovaries which are both superior and inferior are 

 included in one and the same genus, as we see in Asarum, a genus of the 

 Aristolochiacese. Among gamopetalous plants (BENTHAM et HOOKER H.), 

 the following families are mentioned as those which include plants with 

 superior and inferior ovaries : Apocynacese, Campanulacese, Goode- 

 niacese, Myrsinese, Primulacese, Composite and Styracacese. 

 Among Polypetalse (BENTHAM et HOOKER I.) : Bruniacese, Celastrinese, 

 Ficoidese, Hamamelidacess, Holoragacese, Melastomacese, Myrta- 

 cese, Olacinacese, Portulacacese, Rhamnacese, Rhizophoracese, 

 Rosacese, Samydacese, Saxifragacese and Connaraceee. Among Monoch- 

 lamydeae, (BENTHAM et HOOKER III.) : Balanophorese Chenopodiacese, 

 Cupuliferee, Lauraceae, Monimiaceaa and Santalacese. Among 

 Monocotyledones (BENTHAM et HOOKER HI.) : Bromeliacese and Liliacese. 

 As we have stated above, Mitrastemon is closely related to the Raffle- 

 siacese in several points, and even the points which led us previously to 

 regard it as different from the Rafflesiaceae (HAYATA I. p. 112 ; MAKING H. 

 p. 252) do not present sufficient characters to warrant the erection of a 



