40 (jXLiii. coMMELiXACE.E (clarke). [Comjuelina. 



Mozamb. Dist. Grerman East Africa : Kilimanjaro ; at Maranj^u, 5000- 

 5500 ft., Volkens^ 1244 ! 2324 ! British Central Africa : Urungu ; Fwambo, Carson, 

 25 ! Kambole, Houth-west of Lake Tanganyika, 5000 ft., Nutt ! Xyasaland ; marshy 

 promontory of Lake Shirwa (Chilwa), Meller ! Zomba Kock, Whyte ! Mount Sochi, 

 Scotf-Mliot, 8525 ! Mount Mlanji, Whyte ! and without precise locality, Buchanan, 

 338 ! Ngamiland : edge of the rivers near Lake Ngami, McCahe, 24 ! along tlie 

 River Chobe, McCahe, 47 ! 



Frequent also in the Mascarene Islands and Extratropical South Africa. 



This is a very common species; and the plentiful material shows that, on the 

 same plant, the length and acumination of the spathe varies excessively, and that the 

 leaves vary considerably from oblong to elliptic-oblong. Where thei'e is no fruit and 

 the colour of the iiowers is not noted, I have (in general) not cited the example in 

 the foregoing geography. The plant is frequent in Abyssinia, and Schweinfurth 

 gives many localities which are not copied here, though I do not doubt that his 

 numbers are either C. africana, or one of the species following (C. edulis, C. iavolu- 

 crosa), v/hich I am not sure to differ. In C. Elliotii, the capsule is noted to be 

 5-seeded, but tlie plant is otherwise very exactly C. africana, in whicl) there are 

 always 5 ovules. As to Schimper, 590, it is larger than C. africana usually is in 

 leaf, spathe, and capsule, and it may be A. Richard's C. involucrosa ; but, if so, I 

 fear C. involucrosa, A. Rich., is not distinct from C. africana. 



23. C. edulis, A . Rich. Tent. Fl. Ahyss. ii. 841. Leaves elliptic, not 

 lanceolate; stems and sheaths marked by a longitudinal line of hairs; 

 three interior sepals heart-shaped and equal ; otherwise as C. africana. — 

 Schweinf. Beitr. Fl. Aethiop. 29.') ; C. B. Clarke in DC. Monogr. Phan. 

 iii. 105; Durand&Schinz,Conspect. Fl. Afr.v. 425; Engl. Hochgebirgsfl. 

 Trop. Afr. 155. C. heccariana, Martelli, Fl. Bogos. 87 ; Schweinf. in 

 Bull. Herb. Boiss. ii. Append, ii. 54. 



srile Ziand. Eritrea : Mogod Valley, 4500 ft., Schweinfurth Sf Eiva, 1608 ! 

 Abyssinia : Tigre ; Mount Sholoda, near Adowa, 6700 ft., Schimper, 60 ! Wojerat 

 district, Fetit ■ Shoa, Petit I 



The above is condensed from A. Richard's description. The only part of the dif- 

 ferences alleged that appears distinctive is the broader leaves. The specimen of C. edulis 

 in herb. Kew. from herb. Franqueville (believed to be a piece of A. Richard's type) 

 agrees very well with the description ; the upper leaves are If by f in., more elliptic 

 than as in C. africana. The s])athe in this type specimen is less than f in. long, 

 not acuminate, but hardly difPers from some spathes to be found in C. africana. 

 This type shows no capsule : but it is either closely allied to C. africana, or to be 

 united therewith. Buchanan, 6360, from Nyasaland, and Scott-Elliot, 8525, from 

 the Shire Highlands, have the leaves elliptic at the base, but they do not match 

 C. edulis, and are rather forms of C africana. 



24. C. involucrosa, A. Rich. Tent. Fl. Abyss, ii. 342. Very large. 

 Leaves sheathing at the base, sessile, lanceolate, acute, pubescent and 

 ciliate when young. Spathes 2-2J in. long, long.-peduncled, simply 

 folded, hairy ciliate on the margin. — Schweinf. Beitr. Fl. Aethiop. 295; 

 C. B. Clarke in DC. Monogr. Phan. iii. 165 (excl. Schweinf. 590) ; 

 Durand &l Schinz, Conspect. Fl. Afr. v. 425. 



Nile ]Land. Abyssinia: Tigre; Tchelalchekenneh, Quartin-Dillon. 

 No type of this has been seen ; the above is condensed from A. Richard,' who was 

 ■" disposed to believe the flowers yellow." Assuming the flowers yellow, and noting 

 the large size of the plant, A. Richard's C. involucrosa was probably the plant of 



