A. B. Seymour, in June, 1883. This is probably the same as the C. Abietis of Europe, although, as the spores were not quite ripe, one cannot be certain. If there is a difference, it is to be found in the fact that the teleutospores are arranged in threads which branch less than in the European form. But at a later stage of development this supposed difference might disappear. At the same time and place Mr. Seymour found another interesting species of Uredineæ also on Abies Canadensis, not on the same branches as the species last mentioned, nor on the same trees, as far as can now be ascertained. Spermogonia were abundant on both sides of the leaves, on whose under surface were elliptical or elongated sori of a pale yellow color, arranged in two rows parallel to the midrib.

The spores were globose or somewhat elliptical, about 13-17 μ in length, and appeared to be borne in chains composed of a small number of spores. It is possible that this form is Caoma Abietis-pectinata Rees, of which I have seen no specimens. From the description of Rees, however, his species has larger spores than ours, and no mention is made of spermogonia. It may be well to designate our form under the name Caoma Abietis - Canadensis, until more exact information can be obtained. Prof. J. Macoun has found the interesting Melampsora sparsa Winter on Arctostaphylos alpina, on the island. of Anticosti. Visitors to the White Mountains should search for the fungus there.

#

XIV.

CONTRIBUTIONS TO AMERICAN BOTANY.

BY SERENO WATSON.

=

Communicated January 14th, 1885.

1. A History and Revision of the Roses of North America.

History. Perhaps the earliest notice of our wild roses is found in the account of Gosnold's voyage in 1602, where the "eglantine" is noted among other plants growing on Cuttyhunk. Higginson (1630) and Josselyn (1672) also make mention of wild "single damaske roses " in New England, "verie sweete." The first botanist to mention an American species is Parkinson, in his Theatrum Botanicum, in 1640, where he describes his "Rosa sylvestris Virginiensis; the Virginia Bryer Rose," with "divers as great stemmes and branches as any other Rose, set with many small prickles and a few great thornes among them, the leaves very greene and shining, small and almost round." He does not state upon what materials his description was based, but, as it accords rather more nearly with R. lucida than with any other species, they were probably from New England, which was included in the region then known as Virginia. The name and essentially the same description are given by Ray in the Historia Plantarum (1693), but without any additional information.

The next description is by Dillenius, in the Hortus Elthamensis (1732), of a species introduced into England by James Sherard in 1726, which he figures and describes under the phrase "Rosa Carolina fragrans; foliis mediotenus serratis." From the name it may be inferred that it was originally from the Southern States, and it may be quite clearly identified with the R. humilis of Marshall (the later R. parvifolia of Ehrhart). He also describes and figures a second species as raised by Sherard from seeds received from New England. This, however, is evidently not an American rose, but European, as was perceived by Linnæus, who made it the basis of his R. pendulina. Too much confidence in Dillenius's statement of its origin led later botanists into much needless trouble in attempting upon no other ground to retain it among American species.

In 1736 Clayton sent two roses to Gronovius from Virginia, one as a Dog Rose, "Rosa canina," the other as a "Sweet Bryar," which were published by Gronovius in 1739 in the Flora Virginica under Clayton's exceedingly brief descriptive phrases. Linnæus, whose good judgment led him to be very prudent in recognizing species in this genus, in the first edition of his Species Plantarum (1753) makes no reference to these species of Gronovius, nor to any but those of Dilleuius, and solely upon the figure and description of the "Rosa Carolina fragrans" of the latter made his original Rosa Carolina. In the Systema Naturæ of 1760, however, and still more fully in the second edition of the Species Plantarum (1762), he redescribed the species from specimens raised in the Upsal Garden; but, noticing the discrepancies, he now cited Dillenius with a doubt. The specimens preserved in the Linnæan herbarium satisfactorily identify this first established species, as Linnæus here and always afterward defined it, with the R. Carolina as it was figured by Wangenheim (in 1787) and others, and as it has long been generally understood in this country.

Miller, who was the authority of his day upon the cultivated plants of England, in the seventh edition of his Gardeners' Dictionary (1759) described the "Wild Virginia Rose, kept in gardens for the sake of variety, and growing naturally in Virginia and other parts of North America," and, considering it the same as the Rosa sylvestris Virginiensis of Parkinson and Ray, he adopted their specific phrase, and accordingly in the next edition (1768) named it after the Linnæan method Rosa Virginiana. It is impossible to identify it from the description. The name was afterward taken up by DuRoi (1771 and 1772), Wangenheim (1787), and by Gmelin (1796), and applied by them to a form of R. Carolina, or perhaps to the same garden form that was described by Ehrhart in 1789 under the name R. lucida, to which species it was referred in Martyn's edition of Miller's Dictionary in 1807.

Gronovius in the second edition of the Flora Virginica (1762) adds a third species from Virginia, Clayton's "Rosa alta palustris” (his previous species being "sylvestres" or Wood Roses), and refers all three to Linnæus's R. canina, as varieties. Dr. Gray found in 1839 in Clayton's herbarium, preserved at the British Museum, two specimens, one our Rosa parviflora," and the other near to it. Clayton's tall swamp rose was probably R. Carolina.

66

In 1772 DuRoi described two roses, then in cultivation in German gardens, under the names R. Virginiana and R. Carolina. The first appears to have been the true R. Carolina, and the latter Ehrhart's

1

326

PROCEEDINGS OF THE AMERICAN ACADEMY

R. parviflora, and Willdenow so understood them twenty-four years

later.

In 1785 we find the first determinations of American roses by an American botanist, in the Arbustum Americanum of Humphrey Marshall. He mentions and partially describes four species as found in Pennsylvania, R. Carolinensis, R. palustris, R. humilis, and R. Pennsylvanica plena. The first, or the "Wild Virginia Rose," is given in all probability simply to cover the ground already occupied, for the description, as evidently as the names, is taken wholly from Linnæus and Miller. The second species, the "Swamp Pennsylvania Rose," is the true R. Carolina, and his description is a better one than had before been made. The R. humilis, or "Dwarf Peunsylvania Rose," is without doubt what has usually been regarded as Ehrhart's R. parviflora, and his name has every right to recognition in preference to the latter. It is not easy to account for the double-flowered form of this, of which he makes his last species.

Wangenheim, who as a captain in the Hessian forces had been from 1777 to 1780 in this country, after his return to Germany published in 1781 a description of some North American trees and shrubs, and in 1787 a more elaborate account in his Beytrag zur teutschen Forstwissenschaft. In the first he describes only the R. Carolina, but in the latter, in addition to a good description and recognizable figure of that species, he gives descriptions of "R. Virginiana" and of "R. Pennsylvanica flore pleno, mihi," the first not recognizable, the other a myth as respects any wild native species.

Walter in his Flora Caroliniana (1788) mentions only R. Carolina, with varieties " calycibus laciniis brevibus integris; et longis, laciniatis." This is the earliest reference to the rose which Michaux afterwards described as R. setigera.

Aiton's Hortus Kewensis (1789) refers everything previously published to R. Carolina, under a half-dozen varieties. Here first appears R. blanda, with a good description, stated to be a native of Newfoundland (where it was collected by Sir Joseph Banks) and Hudson's Bay, cultivated since 1773 by James Gordon. In the same year Ehrhart published, in his Beitrage zur Naturkunde, three species then in cultivation in the gardens at Hannover; viz., R. parviflora, of which he knew only a form with double flowers, R. corymbosa, which he rightly considered the same as the Linnæan R. Carolina, and R. lucida. He gives R. Carolina of DuRoi as a synonym of R. parviflora, and appears to have had DuRoi's description in mind in drawing up his own. There was probably some connection between this double-flowered

species of Ehrhart and the double-flowered R. Pennsylvanica of Marshall and of Wangenheim. Its origin is of course uncertain, but it may have been the survival in European gardens of the Rosa Carolina fragrans of Dillenius, the single form having become rare through neglect. Willdenow knew it only as double; Lindley also, in 1820, had never seen it single; and the only cultivated specimen in the Gray herbarium, from Jard. Luxembourg (1814), is double. This specimen differs little otherwise from native specimens of the species for which I have adopted the earlier name of R. humilis. Ehrhart, however, described it as having the calyx-lobes entire, which is not the case in R. humilis. His R. lucida, as it was more fully described by Willdenow, who probably knew what Ehrhart meant, may be with some certainty identified with the common New England species with dark shining leaves, and the name may be accepted as the earliest, and as appropriate.

Willdenow in 1796, in his Berlinische Baumzucht, describes these three species (R. parviflora, R. lucida, and R. Carolina) more in detail, as again in his edition of the Systema in 1799, where he contrasts what he considers to be their distinctive characters, but overlooks the most important differences. In the latter work he adds

Aiton's R. blanda.

Borkhauser in 1790, in an account of the shrubs of Hesse-Darmstadt, described a Rosa fraxinifolia, which Gmelin, in his Flora Badensis (1806), says was then common in cultivation and suspects to be the same as Aiton's R. blanda. There is no reason for doubting their identity.

Salisbury took occasion in his Prodromus (1796) to substitute arbitrarily the name R. fragrans for the Linnæan R. Carolina.

The next decided advance was made by Michaux, who in 1803, in the Flora Boreali-Americana, published his R. setigera and R. lævigata, the latter an introduced species that had been in cultivation in Georgia for over twenty years. His other species, R. Caroliniana and R. Pensylvanica, are shown by his herbarium to be, the first R. humilis, and the second a mixture of R. Carolina and R. blanda. Willdenow in his Enumeratio (1809) added R. nitida and R. gemella. The first is a well-marked species that had been found by Sir Joseph Banks in Newfoundland, and had become introduced into England and upon the Continent, though Willdenow was ignorant of its origin. R. gemella was also a cultivated species, described as having curved infra-axillary spines, and as intermediate between R. lucida and R. Carolina, but the leaves not at all shining. This is referable, with