T. Kirk. — A Revision of the Neic Zealand Gentians. 337 



Hab. South Island : Arthur's Pass, and other localities 

 in the Canterbury Alps ; 2,500ft. -3,000ft. 



Var. vacillata. 



Stems suberect, numerous, very slender, 3in.-4in. long, 

 1-3-flowered. Leaves linear-spathulate ; cauline smaller, nar- 

 rowed below, but not petioled. Calyx cleft nearly to the 

 base ; sepals broadly oblong, obtuse. 



Hab. South Island : Canterbury Alps ; J. B. Armstrong ! 

 Otago— Mount Earnslaw ; T. Kirk. Apparently local. 



Var. divisa. 



Stems very slender, erect, excessively divided from the 

 base. Eadical leaves obovate, membranous, petioles broad ; 

 cauline leaves short, broad, sessile or rarely broadly petiolate ; 

 pedicels narrowly winged. Flowers forming a lax, open 

 corymb. Calyx much shorter than the corolla, cleft nearly 

 to the base, segments linear-oblong, obtuse or subacute ; 

 corolla cleft nearly to the base, lobes rounded at the apex, 

 broad, spreading. 



Hab. ^outh Island : Mountains of Nelson and Canter- 

 bury; T. H. Potts ! N. T. Garrington. Apparently local. 



Var. magnifica. 



Leaves all radical, rosulate, narrow obovate-spathulate, 

 shortly petiolate, the remains of old petioles crowded below. 

 Scapes very numerous, lin.-3in. long, with one or two pairs 

 of linear bracts, having a single large pedicellate flower in 

 the axis of each, the whole forming a globular mass 4in.-8in. 

 or more in diameter. Calyx fully equalling the corolla-tube, 

 cleft for two-thirds of its length, segments broadly oblong, 

 -|in.-Y 3 gin. broad, subacute or obtuse ; corolla deeply divided, 

 lobes broadly rounded. 



Hab. South Island : Eidges near the summit of Mount 

 Captain, Amuri ; 4,500ft. ; T. Kirk. 



A singular and beautiful form, which will probably be 

 found to deserve specific honours. The plants are solitary, 

 and scattered over the slope of the ridge, so that they re- 

 semble large snowballs, and attract the attention of the 

 observer long before he reaches their habitat. No traces 

 of leaves are visible until the plant is dug up, when the 

 crowded remains of dead leaves at the base of the rosette, 

 combined with the stout root, afford evidence of the 

 lengthened period of duration enjoyed by this form. The 

 fknvers become yellow during desiccation. It is nearly allied 

 to G. cerina, Hook. f. 



22 





