Stem prostrate and rooting, laxly branched, c. 2-3 mm. diam., almost or quite glab. Lvs ± distichously arranged, not crowded; lamina coriac., 10-25 × 8-20 mm., broadly ovate or rounded rhomboid, crenate-serrate, glab.; petiole > lamina, broad, flat, fringed along edges with long white hairs and with some sparse hairs on back. Peduncle erect, 10-15 cm. tall, almost or quite glab., purplish. Bracts in up to 5 pairs (occ. threes), ovate to ovate-cuneate, subsessile and ciliate along lower margin, crenate-serrate above. Pedicels > fls, almost or quite glab., hairs if present non-glandular. Calyx-clefts unequal; lobes oblong to slightly cuneate, us. slightly notched, glab. except for occ. marginal hairs. Corolla up to 20 mm. diam.

DIST.: S. Vicinity of Arthur Pass; Hokitika Valley; mountains near Lake Wakatipu.

Type locality: Mt. Alexander, 5000 ft. Type: W, 4154 "Plant from Mt. Alexander, Teremakau Valley, grown in my garden at Dunedin" D. Petrie.

The above description, based on that given by Petrie, fits well the type specimens (erect stems only, two with fls, one with frs); also (a) several sheets of Cockayne's No. 5993 "Source of R. Otira, under cliffs of Otira Glacier on grassy slope, 1280 m. in shade. 25 Jan. 1898"; (b) Hokitika Valley, Forest Service Survey 1957-58; (c) four distinct collections made from 1923 to 1947 from the Routeburn-Lake Harris-Lake Howden track; (d) specimens of cultivated plants coming originally from Arthur Pass, Mt. Earnslaw and Cosmos Peaks, Lake Wakatipu. Petrie says: "The present species has grown very freely in gardens at Dunedin. It is easily cultivated, and forms a showy plant. Its large bracts easily distinguish it from the other species native to New Zealand." Recorded also from Upper Clarence River (Simpson and Thomson, T.R.S.N.Z. 71, 1941, 92).

In general lvs are glab. except on petioles and adjacent margins, but in some plants some lvs have short hairs covering part or all of the upper surface. Galls, similar to those associated with abnormally hairy lvs and stems in O. caespitosa var. gracilis, have been seen on some plants, but these could not be firmly correlated with lf-hairiness.

G. Simpson pointed out that these plants could be hybrids between O. caespitosa and O. macrocarpa; they resemble the former sp. in habit, in shape of calyx-lobes and in the predominantly paired bracts; the latter sp. is suggested by greater size and more evenly crenate lf-margins than in O. caespitosa, notching of calyx-lobes, and occ. occurrence of more than two bracts at a node, though three is the largest number seen; neither of these spp. would account for sporadic hairiness of lvs. Plants grown by W. B. Brockie from seed of O. cockayniana from Mt. Alexander showed some diversity, but are represented as herbarium specimens only by a series of detached lvs showing between them little more variation in size, shape, and hairiness than is often seen in different parts of one plant. Both Simpson and Brockie report that O. cockayniana is us. found only as a single plant, but Cockayne (A 8502) notes of his No. 5993 "forming large patch"; he mentions as other localities Kelly's Hill and rock bank of Lake Minchin, Upper Poulter. The localities for O. cockayniana all lie within the area of distribution of the putative parents, but no consistent difference is obvious between those from near Arthur Pass where O. macrocarpa is represented by var. calycina and those from further south where only var. macrocarpa is known.

A series of specimens from Wilmot Pass, Routeburn and Humboldt Mts are not unlike O. cockayniana in size and habit, but in general hairiness of lvs, bracts, stems, peduncles and calyx-lobe margins, as well as in the un-notched and only slightly uneven calyx-lobes, they suggest a hybrid of O. caespitosa with O. crosbyi rather than with O. macrocarpa.

Pending further investigation the name O. cockayniana is retained for the rather uniform group of plants recognized by Petrie.