698 



SCIENCE. 



[N. S. Vol. XVIII. No. 405. 



Aristida, but of this there are ten species re- 

 corded from the region. 



To gain a general idea of the grass flora of 

 the Transvaal, it is well to note the genera 

 containing the largest number of species. We 

 find the following figures : 



Seven genera have only two species each. 



Thirty-four genera are only represented 

 by a single species. These include the well- 

 known North American genera Paspalumj 

 Brormis and Poa, all three represented by 

 alien species. It is probable' that native 

 species of Paspalum will be found in some 

 parts of the Transvaal. 



In these notes I have used the word species 

 in a broad sense, for convenience, enumerating 

 as species some of Dr. Stapf's varieties. 



Of the four largest genera, three — Andro- 

 pogon, Panicam and Aristida — are well known 

 to North American botanists, being promi- 

 nently represented in the North American 

 grass flora. With the exception of this ap- 

 parent connection, which may not be as real 

 as ajiparent, there is little or no relationship 

 between the two floras. 



Certain well-known North American genera, 

 such as Rotthaellia, Btenotaplirum, Koderia, 

 Trisetum, Avena, Danthonia, Agrosiis, Cal- 

 amagrostis, Stipa, Oryzopsis, Tragus, 8par- 

 tina, Phalaris, Melica, Panicularia Picc- 

 cinellia, Festuca, Poa, Bromus, Lepturus, 

 Agropyron and Hordeum, all of which occur 

 in South Africa, are not represented by a 

 single species indigenous to the Transvaal, 

 and only two of these, Poa and Bromus, by 

 alien species. 



Three of the largest genera of South 

 African grasses, Danthonia with twenty-five 

 species, Ehrharta with twenty-five and Pen- 



faschisiis with thirty-eight, have not, so far, 

 been found in the Transvaal collections. 



With regard to the relative proportion of 

 individuals, a very rapid reconnoissance of 

 the high veldt made during early winter be- 

 fore the grasses had lost their inflorescences, 

 showed that Andropogons, and perhaps in 

 places Anthistirias, are the most characteristic 

 grasses throughout the high veldt. Other 

 genera are represented by scattered individ- 

 uals or by relatively small colonies restricted 

 by peculiar edaphic conditions. 



From what little I have so far seen of the 

 bush veldt and the low country, it would ap- 

 pear that grasses are almost as abvmdant 

 there as on the high veldt, but there they 

 cease to form so characteristic a feature of 

 the vegetation because of the abundance and 

 prominence of the biishes and trees. 



Bermuda grass (Capriola dactylon) is 

 widely distributed along roadsides, in lawns, 

 in cattle kraals and on town lands where the 

 oxen are outspanned, but it does not have 

 the aspect of an indigenous grass. On the 

 open veldt it is usually found at outspanning 

 places, and covering old and deserted ant- 

 heaps, but the patches are rarely connected. 

 It is highly valued as a pasture grass and is 

 closely eaten down by mules. The Boers 

 state that it is readily introduced into the 

 veldt by grazing closely with sheep. 



The reed (Phragmites vulgaris) is com- 

 mon along almost every stream that I have 

 crossed, from the sources of the Limpopo on 

 the high veldt to the Mooi, Malmanie and 

 Marico in the west, not far from Mafeking, 

 and to the falls of the Koomatie at Koomatie- 

 poort on the frontier of Portuguese East 

 Africa. It has been called a distinct species 

 by some writers, but Dr. Stapf does not con- 

 sider it worthy of even varietal ranli. 



The adaptation of the vegetation to peculiar 

 climatic conditions — a literally rainless winter 

 season of four to flve months (May to Septem- 

 ber inclusive) and a fair rainfall (twenty-five 

 to thirty inches) during the rest of the year — 

 is interestingly demonstrated by the develop- 

 ment of the bulb-, corm- and tuber-producing 

 habit. This is not confined to the families 

 Amaryllidacese and Iridacese (which are well 



