76 CONTRIBUTIONS FROM THE NATIONAL HERBARIUM. 



of them divided in two part way or to the base, and some 12 to 15 mm. long ; 

 body of the bur thick-walled, the lobes mostly 12 to 15, erect or spreading, simi- 

 lar to the spines ; spikelets 4 to 7, more or less distorted by the pressure of the 

 rigid involucre, 7 to 9 mm. long, 2 to 3 mm. wide; first glume very narrow, 

 usually wanting ; second glume and sterile lemma slightly shorter than the 

 acuminate fruit, obscurely puberulent on the middle internerves. 



The bur of C. palmeri is larger than that of any other known species of the 

 genus. A second specimen of Palmer 689 with yellow-spined burs is mentioned 

 by 'Vasey in the original description as a " yellow colored variety." 



DISTRIBUTION. 



In dry sands near the coast, Sonora and Lower California, Mexico. 



Lower California : Carmen Island, Palmer 14 in 1S70, 865 in 1890. Calmallf, 

 Orcutt in 1899. Magdalena Bay, Brandeyce in 18S9. San Jose del Cabo, 

 Purpus 519. San Felipe, Goldman 1161. Between Santo Domingo and Ma- 

 tancita, Nelson & Goldman 7276. 



Sonora : Guaymas, Palmer 271 and 689 in 1887. Adair Bay, Sykes 58. 



DOUBTFUL SPECIES. 



The following names, based on North American plants, the writer 

 has not been able to identify : 



Cenchrus carolinianus Walt. Fl. Carol. 79. 1788. No locality is given, but 

 so far as known Walter's plants were collected in the vicinity of his home, 

 which was on the south side of the Santee River, in the northern part of 

 Berkeley County, South Carolina, to the east of Eutaw Springs, and near the 

 mouth of the old Santee Canal 1 . No specimen of Cenchrus was found in the 

 Walter Herbarium, now in the British Museum 2 . The brief diagnosis is as 

 follows : " Involucrum eehinatum bifiorum, spica glomerata, glumis globosis 

 muricato-spinosis laevibus." This was meant apparently to distinguish the 

 species from Linnaeus's " glumis femineis globosis muricato-spinosis hirsutis," 

 that is, C. tribuloides. Walter's diagnosis does not apply to any known 

 species. Our only species with smooth burs is C. gracillimus, which is not 

 found north of Florida. When the American grasses in the Linnaean Her- 

 barium were examined by A. S. Hitchcock in 1907, it was found that C. 

 tribuloides was the coast form currently called C. macrocephalus. The name 

 C. carolinianus was then applied to the common inland species previously 

 known as C. tribuloides. That species, however, is not known to occur in 

 South Carolina. It has been found in North Carolina and Georgia but appears 

 there to be an introduced weed. Two species of Cenchrus are known from 

 South Carolina, C. tribuloides, confined to the coast, and C. incertus in the 

 coastal plain. Of these two only C. incertus is known to occur in Walter's 

 region. His statement " glumis [bur] laevibus " better applies to C. incertus 

 with its finely pubescent burs than to C. tribuloides with conspicuously villous 

 burs. Since the diagnosis is inadequate and the type specimen nonexistant, 

 the name can not be applied with certainty and is therefore rejected. 



Cenchrus gracilis Beauv. Ess. Agrost. 57. 157. 1812. A name only for a 

 specimen sent by Bosc, presumably from the Carolinas. 



Cenchrus hirsutus Spreng. Neu. Entd. 3: 15. 1822. " Hispaniola." The de- 

 scription, which suggests a species of Pennisetum rather than Cenchrus, does 

 not agree with any species known to us. 



1 See Brainerd, Bull. Charleston Mus. 3 : 33. 1907. 



2 See Hitchcock, Ann. Rep. Mo. Bot. Gard. 16: 48. 1905. 



