51 MR. 0. RENTHAM OX TERNSTRCEMIACE2E. 



The union of the base of the petals into a perfect coralline tube 

 is indeed sometimes very marked, especially in several species of 

 Saurauja, Ternstrcemia, and Eurya ; but in a large number of others 

 the margins of those petals which are external are free, or can be 

 traced in prominent lines to the base ; and from these to the free 

 petals of some Frezicras the passage is very gradual, showing 

 that the union must be considered in the same light as that which 

 occurs in so many Malvaceae, and that the affinities of the whole 

 group are essentially among Eialypetalce. The close connexion 

 of Ternstrozmiacece with Guttiferce and Ilypericinece, and their 

 rather more distant affinity with Tiliacece and Dipterocarpece, have 

 long since been pointed out, nor has any more intimate connexion 

 been suggested. Among Gamopetalce the affinity with Sapotaccce 

 suggested by Lindley, with Ericaceae pointed out by Planchon, 

 and with Ehcnaceas insisted on by Choisy, may in some cases be 

 traced, but it is evidently always much slighter than those above 

 mentioned among Dialypetalce. 



Among the genera collected together by Choisy to form his 

 Gamopetalous order Ternstroemiacece, the genera Ternstrcemia, 

 Adinandra, Cleyera, Freziera, and Eurya are indeed so closely 

 connected with each other that their distinct separation by positive 

 characters is very difficult. All have a similar habit and inflores- 

 cence, adnate anthers, the fruit either perfectly indehiscent or 

 with a slight tendency to the loculicidal dehiscence, and the seed 

 with a long cylindrical curved or horseshoe embryo, in a varying 

 quantity of albumen, sometimes rather thick and fleshy or even 

 granular, more frequently thin or even reduced to a scarcely per- 

 ceptible pellicle lining the testa. But with this tribe Choisy 

 unites Saurauja, which has a different habit and inflorescence, 

 versatile anthers, and a comparatively small straight embryo in a 

 more copious albumen. On the latter account, as well as for some 

 peculiarities in the foliage, Lindley had proposed to remove it to 

 Dilleniacece, next to his new genus Actinidia, with a suggested 

 affinity with Cldetra among Ericaceae; and it is certain that we 

 have here connecting links with both of these otherwise widely 

 distant families. The connexion, indeed, of Saurauja with Acti- 

 nidia (Trochostipna, Zucc.) is so close that the two must be taken 

 together as explaining in some measure each other's position in the 

 system. Of the two, Actinidia is without doubt the nearest to 

 Dilleniacece in its scandent habit and foliage and less united car- 

 pels ; yet even here the characteristic anthers and seeds of the 

 latter Order are wanting, and although the carpels do not always 



