250 MORPHOLOGY OF ANGIOSPERMS 



Salvadoraceae, in which there is sometimes distinct polypetaly, 

 but the reduction of the stamens to two in the former family 

 is hardly to be regarded as a primitive character. The Logania- 

 ceae are general in their resemblances, having features in com- 

 mon with the remaining families, and others suggestive of Tu- 

 biflorales and Rubiales. In fact, Engler suggests that the 

 Loganiaceae may be an older type than any of the others, and 

 may have given rise to the Gentianales and Rubiales, in which 

 he might have included the Tubinorales. If this family may 

 hold any such position in reference to these great alliances it 

 certainly deserves careful investigation. The alliance ends with 

 the Apocynaceae and Asclepiadaceae, in which a latex-system 

 is developed, and oilier evidences of high specialization occur: 

 but they are also characterized by distinct carpels, a feature re- 

 garded as primitive. The Asclepiadiaceae form a very peculiar 

 and highly specialized offshoot, the elaboration of floral struc- 

 tures for entomophily reaching a degree of complexity only to 

 be compared with that of the Orchidaceae. 



V. Tubiflokales.'" — This includes Couvolvulaceae, Pole- 

 moniaceae, Hydrophyllaceae, Borraginaceac, Yerbenaceae, La- 

 biatae, Xolanaceae, Solanaceae, Scrophulariaceae, Bignoniaceae, 

 Pedaliaceae, Martyniaceae, Orobanchaceae, Gesneraceae, Colu- 

 melliaceae, Lentibulariaceae, Globulariaceae, Acanthaceae, 

 Myoporaceae, and Phrymaceae, together containing over 14,000 

 species, the large families being Labiatae with nearly 3,000 

 species, Scrophulariaceae with 2,400, Acanthaceae with nearly 

 2,0(>0, Solanaceae with about 1.700, and Borraginaceac with 

 about 1,550. 



This enormous assemblage of forms has been ordinarily con- 

 sidered as representing at least two alliances, the Polemoniales 

 or Tubiflorae including the first four families of the list above, 

 and the Personales or Labiatiflorae including the remaining 

 families. The tendencies of development are so numerous and 

 interwoven that they are difficult to separate, but rather than 

 merge two such alliance- together it might have been better to 

 have broken up the Personales into five or six alliances, espe- 

 cially if the Plantaginaceae are to be set off as a coordinate 

 alliance Plantaginales. To distinguish them definitely would 



* Tubiflorae of Engler. 



