Structure of the Ehretiaceje and Cor(liacea\ 385 



space runs across the axis anteriorly and ])osteriorly, filled 

 with a compressed plate, which is the columella that supplies 

 the nourishing vessels for the growth ot" ovules and seeds. 

 This growth is constant throughout the Ehretiacefe. The sub- 

 sequent developments of the fruit in the different genera 

 become modified in the manner already described. 



In the Borraijinacea' there exists in the earlier stages a 

 normally bicarpal development very similar to that of Cordia) 

 but during the subsequent growth there is a tendency to a 

 separation of. the whole into four carpels, more or less bi- 

 geminately combined in pairs ; the style remains free in the 

 centre, su])ported upon a common gynobase, upon which the 

 four carpels are affixed, and from which their ovules and seeds 

 derive their nourishing vessels. This constitutes a subfamily 

 marked by many peculiar characters : it requires, however, a 

 thorough reinvestigation. 



In the Heliotrojnaceiej the ovary, normally as well as at 

 maturity, is bicarpellary, and the carpels are seated upon a 

 conical gynobase of half their height. The style is usually 

 very short, thick, and suddenly enlarged into a pulvinate or 

 discoid form ; and this is terminated by two sessile stigmata, 

 more or less abbreviated. The fruit is generally exsuccous, 

 divisible into four single or into two bilocular nuts ; when four 

 nuts are produced, there is a short placentary process that 

 rises from the gynobase, to which the nucules are attached, 

 and which answers the ])urpose of the columella seen in the 

 Ehretiaceu'^ in affording nutrition to the seeds ; they are not 

 bigeminately connected, as in that family. 



Hence it will be seen that the Cordiacexp possess characters 

 which amply distinguish them from the Ehretiacete, Heliotro- 

 piacece^ and Borraginacexe. Nearly all the species of the 

 family have been huddled into the single genus Cordia^ be- 

 cause no one has taken the trouble to ascertain their true 

 characters, their examination having been singularly neglected. 

 It is remarkable that, among the 175 species of Cordt'a enu- 

 merated by De CandoUe in his ' Prodronuis,' the number of 

 cells existing in the fruit is mentioned in only four cases, and 

 utter silence is maintained throughout the whole in regard to 

 the number of cells in the ovary, even in the generic charac- 

 ter ; and the point of suspension of the ovules and attachment 

 of the seeds is everywhere ignored. Prof. Fresenius, in 

 working the monograph of the family for 3Iai-tius's ' Flora 

 Brasiliensis,' contents himself with a few Avords in stating the 

 ordinal character : in regard to its 4-locular ovaiy, he merely 

 says there is an anatropous ovule in each cell, ajipended from 

 the sunnnit (which is not exactly true) ; and in regard to the 



