318 HARSHBERGER— TAXONOMIC CHARTS OF THE [April 19 



mens and the style with the reduction of the number of fertile 

 stamens to one. 



The Palmac€3e and Araceas stand off probably, as having affini- 

 ties with each other, but not closely related to the other petaloideous, 

 monocotyledonous families. Lemnacese may be considered to be a 

 modified or degraded form of Aracese, while the bromeliaceous 

 plants with superior and inferior ovaries show affinities to the 

 Amaryllidacese and the Liliaceae, and hence, the writer has placed 

 the order Bromeliacese on a branch near where the two upper limbs 

 of the family tree diverge from each other. The complete liliaceous 

 structure without great reductions in the number of whorls, but 

 with generally few ovules in each loculus of the ovary, is found in 

 the Commelinacese, while the Mayacaceas, as a family, is closely 

 allied to the Commelinacese. The Xyridaceae are marsh plants with 

 radical leaves arranged in two rows and short spikes on long stalks. 

 The flowers, as in Commelinaceae, have sepals (which, however, are 

 more chaffy) the petals, but the outer series of stamens is want- 

 ing. The order Eriocaulonacese on another branch is sometimes 

 called the " Compositse among Monocotyledons " with radical and 

 grass-like leaves, while the habit of the plants of the Restionacese 

 is quite similar to the Cyperacese. 



It is a much more difficult task to trace the affinities of the 

 dicotyledonous families of plants. Roughly we may divide the fam- 

 ilies into the Incompletse, the Apopetalse and the Gamopetalse. The 

 plants of the primitive Incomplet?e are all or nearly all of them 

 provided with flowers that are wind pollinated. Such orders as the 

 Salicacese, Myricaceae, Juglandaceae, Fagacese, Betulacese and Cory- 

 lacese are not only wind pollinated, but the staminate flowers are in 

 catkins, thus being advantageously situated for the discharged pollen 

 to be carried away by the wind. The perianth in these orders is 

 absent, or extremely rudimentary. The affinities of these primitive 

 dicotyledons, as the writer has been enabled to determine them, is 

 displayed in the larger of the two accompanying charts. 



The apopetalous families in which the petals are absent or dis- 

 tinct are to be regarded as more primitive than the gamopetalous 

 families. The relationship between the families is a group relation- 

 ship. Thus the Loranthacese, Rafflesiacese, Balanophoraceae, Santa- 



