383 



braud to R. raccmusa. which is a syuouym of R. HumbohJliaiid. The plants 

 which Hillebrand refers to R. Humboldtiana have long lanceolate oblong leaves on 

 long petioles, and racemes about 7 cm long; he also says floriferous from the 

 base. Anyone on examining CJandichauds plate can see that the raceme is not 

 floriferous from the base, but at the apex. 



Hillebrand also states upper anthers ciliate; they are not at all ciliate. In 

 Gaudiehaud's plate the drawing of the undeveloped raceme measures about 12 

 em. As the plant is evidently considerably reduced, the specimen from whieh 

 the drawing was made must have had a raceme about 16-18 cm long. The some- 

 what erect habit of the raceme is in all probability due to its being an immature 

 one. as there are no open flowers on it. 



^lann's siiecimen of his Dilissca raccmosa was without flower and conse- 

 quently he was unable to place it in Rollandia. 



"Wawra justly described a Rollandia KaaJar from ilt. Kaula, whence ^laun's 

 and Hillebrand 's specimens were eollected and which were referred by them to 

 7i*. HiDiiholdliaiia. Hillebrand reduced Wawra's Rollandia Kaalac to what he 

 considered R. Iliiinboldliaiia. Wawra himself states that R. KaaUu is closely 

 related to Rollandia lanceolata, and so it certainly is. if not a mere form of that 

 variable species. There is no other Rollandia save Rollandia raccnidsn which in 

 habit can be compared with Gaudichaud's Rollandia Hinnboldtiana. There is 

 no doubt that these tv.o i)lants are identical, and as the specific name Hinn- 

 boldtiana is the older t>ne, II. raccmosa must be reduced to it. In regard to the 

 calycine lobes, it may i)e remarked that the writer found exactly the same 

 calycine lobes as described by Maiui ( iiardly exceeding a line in length) and 

 intermediates ranging up to Id iinn hiug: they are acuminate, munded, acute and 

 truncate: in one word, exceedingly variable. 



DOUBTFUL SPECIES. 



Lobelia piiiuatifida Cham. 

 Lobelia •nnbifnia Cham. 



The above two species are in all pniliahility Rollandias; the first one is perhaps 

 a young plant, as neai-ly all the Hawaiian Rollandia have lobed leaves in their 

 young state. Jjobelia pinnatifida is, of course, not a true Lobelia, as in the de- 

 scri])tion it is .stated '' raccmis axillaribus." It may be related to Wawra's 

 T'dlhiiidia Innijiflora, whicli keeps its siiuiately lobed leaves even in the flowering 

 stage. 



The second species, which is Rollandia.' antbigua G. Don. may be identical 

 with Rollandia H iiinlxildl iinin, but tiiat is a mere conjecture. 



SPECIES EXCLUDENDA. 



Rollandia Fauriei Levi, in Fedde Repertor. Spec. nov. XII :506. 1913. = Cyanca 

 spafhiilata (Hillebr.) Heller. 



