266 CANNA EDULIS 



the other bearing an anther on one side in contact (in the bud) 

 with the front of the style, round which this staminode is folded, 

 the upper part recurved, the base fused with the posterior surface 

 of the style ; anther apparently 1 -celled, dehiscing down the 

 centre, pollen yellowish- white. Ovary inferior, oblong-ovoid, 

 green, papillose, 3-celled, with numerous axile ovules in two rows 

 in each cell ; style 2 J inches long, horizontally flattened, sub- 

 petaloid, scarlet, erect in the posterior part of the flower; stigma 

 terminal, truncate, linear. Fruit, a roundish capsule about J inch 

 long, crowned with the withered remains of the flower, pericarp 

 thin, membranous, greyish-brown, muricated with deciduous 

 filiform processes, dehiscing loculicidally into 3 valves. Seeds 

 several, nearly globular, dark brown; testa thin, marked with 

 minute distant punctuations ; embryo club-shaped, the narrower 

 radicle at the hilum in contact with the testa and covered for a 

 short way down with a sheath prolonged from the inner seed-coat. 

 Habitat. G. edulis is cultivated in Peru, where it was originally 

 observed by Euiz and Pavon; it was first grown in England in 

 1823 from seed collected by them JO years before. It is also 

 supposed to be the species cultivated in the "West Indian Islands 

 (especially St. Kitts) for the starch of its rhizomes, but this has 

 not been made out with certainty. 



The species of Canna have been as yet little studied, and their 

 limits are ill- defined ; they are also said to vary much under culti- 

 vation. The plant figured and above described has been grown 

 at Chelsea Gardens under the above name for many years, but it 

 cannot be said to agree well with the authentic specimens of 

 C. edulis in the Lainbertian* herbarium (now in the British 

 Museum). These have smaller (and paler?) flowers and narrower 

 staminodes, the bracts are much larger, and the habit of the plant 

 more slender. We have not, however, been able to refer the 

 Chelsea Gardens plant with more certainty to any other described 

 species, though G. Lamberti Lindl. (figured in Bot. Reg. t. 470, and 

 Roscoe's 2nd plate) appears to have several characters in common 

 with it. 



Ruiz & Pavon, PI. Peruviana, i, p. 1 ; Roscoe, Scitaminese ; 



