Early Journal Content on JSTOR, Free to Anyone in the World

This article is one of nearly 500,000 scholarly works digitized and made freely available to everyone in the world by JSTOR.

Known as the Early Journal Content, this set of works include research articles, news, letters, and other writings published in more than 200 of the oldest leading academic journals. The works date from the mid-seventeenth to the early twentieth centuries.

We encourage people to read and share the Early Journal Content openly and to tell others that this resource exists. People may post this content online or redistribute in any way for non-commercial purposes.

Read more about Early Journal Content at http://about.jstor.org/participate-jstor/individuals/early- journal-content.

JSTOR is a digital library of academic journals, books, and primary source objects. JSTOR helps people discover, use, and build upon a wide range of content through a powerful research and teaching platform, and preserves this content for future generations. JSTOR is part of ITHAKA, a not-for-profit organization that also includes Ithaka S+R and Portico. For more information about JSTOR, please contact [email protected]. Vol. 41 No. 10 BULLETIN

OF THE TORREY BOTANICALCLUB

OCTOBER, 1914

Noteson -Vill

PER AXEL RYDBERG

DRYAS Little has been added to the knowledgeof this genus in the last eightyyears. Only one species has been added, Dryas tomentosa Farr. Dryas integrifoliaWahl. Many authors regard this as merely a variety of D. octopetala,as intermediateforms are met with, but these have only been found where the two species grow to- gether and are probably all of hybrid origin. Nathorst* was inclined to regard these forms as hybrids, but he did not call them by a hybrid name, denoting them as D. octopetalaf. inter- media. Hartzt held the same opinion, but apparently by a slip of the pen called them D. integrifoliaintermedia Nathorst. This hybridhas also been collected in Alaska at Orca, Prince William's Sound, I899, Coville& KearneyI191. Dryas octopetalaL. Hartz, loc. cit., admittedseveral varieties of Dryas octopetala,viz. var. minorHook., var. hirsutaHartz, and var argenteaBlytt. The last one of theseseems to be furthestfrom the type, but by no means deserving specific rank. Simmons described a similar form of D. integrifolia,viz. D. integrifolia canescens. Dryas tomentosaFarr. This species is closely related to D. Drummondiiand perhaps not distinct. It may be only a variety

* Ofvers. Kong. Sv. Vet. Akad. Hand. 411: 24. I884. t Medd. Groenl. I8: 32I. I895. ]The BULLETIN for September (41: 435-482) was issued 8 0 I9I4.] 483 484 RYDBERG: NOTES ON ROSACEAE standing in the same relationshipto D. Drummondiias D. octo- petala argenteaBlytt stands to D. octopetala. This species has also been collected in the Canadian Rockies, Macoun 65I25.

GEUM Geum vernum (Raf.) T. & G. was originally described as Stylopusvernus Raf., and the writeris somewhat in doubt if this species should not be regarded as the type of a distinct genus. The habit is that of a typical Geum, but the receptacle in fruit becomes stalked above the hypanthium,and the bractlets are usually lacking. Occasionally, however,there are found minute bractletsin this species, and the receptacleis more or less stalked in Geum rivale. The generic characters of Stylopusdo therefore break down,and it is perhaps betterto regardit a Geum,especially as the generalhabit is not essentiallydifferent. Geum virginianumL. This species has in general been very well understood. Murray, however, applied that name to G. canadenseand redescribedthe true G. virginianumas G. laciniatum Murr. Many have referredthe latter synonymto G. canadense and Scheutz suggeststhat it may be the same as G. agrimonioides C. A. Mey., i. e. G. MeyerianumRydb., but in my opinionit belongs to G. virginianumL. Geum camporumRydb. Some twenty years ago, the writer, when working over the Rosaceae of Nebraska, had trouble in determiningsome specimensof Geum,and did not know whether to referthem to G. canadenseor to G. virginianum. These speci- mens had the thick leaves, the stout stem and branches,and the largefruiting head of the latter,but the receptaclewas not glabrous and the achenes had the pubescence of G. canadense. When workingup thematerial for the North American Flora, he foundthat the was morecommon in the prairieregion of the Mississippi valley than was expected and that its rangeextended from Minne- sota and to Arkansasand Texas. As both G. virgini- anum and G. canadenseare found in the region,G. camporummight be regardedas a hybridof the two. These two have about the same distribution(except that G. virginianumis not found in Mexico), but G. camporumis not foundexcept in the westernpart of theircommon range. It is, therefore,not probable that it is a RYDBERG: NOTES ON ROSACEAE 485

hybrid,and it is more likely a distinct species. The following specimensbelong to it: KANSAS: Manhattan, I892, Norton; Riley County, I895, NortonI37; Olathe, I892, Hitchcock;Fort Riley, I892, Gayle522; Emporia, I89I, E. Smith; Cowley County, I899, Mark White54; Lawrence, W. C. Stevens. NEBRASKA: Lincoln, I887, Webber;Nehawka, Sweezey; Sar- gent Bluffs,I853 or I854, Hayden; Franklin, I893, Laybourne; Middle Loupe, I893, Rydberg i6o8; Glenwood, i888, T. A. Williams. SOUTH DAKOTA: Black Hills, Fort Meade, I887, Forwood15. OKLAHOMA:Waugh I75. TEXAS: San Marcos, I898, Stanfield;Crab Apple, Jermey47I; Industry,I893, WUrzlow. MINNESOTA: Fort Snelling, I890, Mearns 346, 347. ARKANSAS: Whippe Expedition,Bigelow. Geumcanadense Jacq. This has usually been knownunder the name Geumalbum J. F. Gmel. The latter was arbitrarilysubsti- tiutedby Gmelin,probably because he thought that it was more appropriate. I say arbitrarilyand without good cause, for he simplybased his species on Jacquin'splate and originaldescription of G. canadense. Geum album (i. e. G. canadense) has been reported fromthe valley of Mexico. As that station was so far remote from the otherwise known range of Geum canadense, the nearest station being in Texas, the writerthought that the specimensfrom Mexico belonged to some other species and that it was merely a case of misidentification. He found, however, in the National Her- barium three sheets collected by Schiede, no. 58o, at San Angel, not far from Mexico City, and these specimens can not be dis- tinguishedfrom specimens fromthe . It was also collected at Chinantla, I84I, LiebmannI743. Geum MeyerianumRydb. C. A. Meyer* gave a very good description of this species, but mistook it for G. agrimonioides Pursh, which is not a Geum at all, but Drymocallisagrimonioides (Pursh) Rydb., untillately usually knownas Potentillaarguta. It was, therefore,necessary to give anothername to Geumagrimoni-

* Ind. Sem. Petrop. II: SupI. 29, I846. 486 RYDBERG: NOTES ON ROSACEAE oides of -Meyerand the writerwith pleasure dedicated it to the botanist who firstdistinguished it. It is strange that this species (or variety) of the Geumcanadense group has not been recognized by any American authors. It was admitted and described also by Scheutz in his monograph.* It is related to G. canadense, but the basal leaves and lower stem-leavesare pinnatifidand the stem is more hirsute. These characters are sometimesfound in G. hirsutumMuhl. [G. florum(Porter) Bicknell],but the petals in G. Meyerianumare white and longer than the sepals, while in G. hirsutumthey are pale yellow and much shorter. The followingspecimens belong to G. Meyerianum: NEW YORK: Fleishmann,Delaware County,I892, Hermannvon Schrenk; Tuxedo, I896, W. H. Lewis; Oneida, 1903, H. D. House. ONTARIO: Battersea, I 898, Fowler. QUEBEC: Danville, I894, A. K. Berg. PENNSYLVANIA: New Danville, Pike Crossing, I90I, Heller; Bangor, I899, Porter; Easton, I899, Porter. DISTRICT OF COLUMBIA: banks of canal, I895, Pollard. Geumhirsutum Muhl. This species has been well characterized by Porter and Bicknell under the names G. albumflavum and G. flavum. It was also described by Fischer and Trautvetter,who mistookit forG. canadense,which they held distinctfrom G. album. The oldest name, however, is G. hirsutumMuhl. Muhlenberg listed it in his catalogue in I8I3. In this publication it is to be regarded as a nomen semi-nudum,but nine years later Link gave it a short description,pointing out the essential characters. Geum decurrensRydb. Some specimens from New Mexico, Arizona and Colorado and named G. strictumseemed so different fromall other specimens of that widespread and rathervariable species, that the writer thought it advisable to propose in the North American Flora a new species based on these specimens The essential characters are pointed out in that work. It is, however,advisable to cite some more specimens. ARIZONA: Bakers Butte, Mogollon Mountains, I887, Mearns 59. NEW MEXICO: Mogollon Mountains, I903, Metcalfe536. COLORADO: Rico, I898, Crandall4109.

* Nov. Act. Soc. Sci. Upsala III. 7: 32. I870. RYDBERG: NOTES ON ROSACEAE 487

Geum oregonense(Scheutz) Rydb. This was originally de- scribedas a varietyof G. urbanumto whichit is not closelyrelated. Its relationshipis with G. macrophyllum. Usually it is well dis- tinguished from that species by its open inflorescence,smaller stem-leavesand smallerand usually paler petals, but intermediate formsare not lacking. Some of these at least may be regarded as hybrids. Geum oregonenseis common throughoutthe Rocky Mountains, as well as the Sierra Nevada and the Cascades. In the southernRockies, G. macrophyllumis not found,but it extends east to northernNew England, where G. oregonenseis not found at all. In California the characters separating the two species seem to be less marked, and considerable questionable material has been seen. Geum perincisus Rydb. Many subarctic specimens which might have been referredto G. oregonensehave the upper leaves deeply incised and the petals broader and in form approaching those of G. strictum. It is hard to say whetherthese should be regarded as a variety of G. oregonenseor as a distinct species. I adopted the latter view, because all these specimens were far northernones and some of them found much furthereast than any of the typical G. oregonense. The followingspecimens belong here: ALASKA: Eagle, Yukon Valley, I902, Arthur Collier 34, 35; Copper River region,i902, William L. Poto I47. YUKON: Fort Selkirk,I899, TarletonI20. MACKENZIE: Fort Simpson, I86I-62, Onion,Kennicott & Har- disty; Fort Resolution, I9OI, Preble & Preble I54. ALBERTA: Cave Avenue, Banff,McCalla 2074. MICHIGAN: Turin, I9OI, Barlow. Geum strictumAit. is a very variable species. Usually the terminalleaflet is moreor less rhombic,as it is commonlydescribed, but not seldom it is rounded or subreniformas it is in G. macro- phyllumor G. oregonense. It usually.can be easily distinguished by its large rounded petals and always by its fruit. The lower portionof the style is never glandular and the upper portionhas hairs about twice as long as those of the othertwo species. Geum scopulorumis the common formof G. strictumin the Rockies, a little less robust than in the East. 488 RYDBERG: NOTES ON ROSACEAE

Geumaleppicum Jacq.* has been regardedas a synonymof G. strictum. Being an older name it should have been substituted. But I think that it is well distinct from the North American plant. So are all specimens fromEurope referredto G. strictum. Whether they should all be included or not in G. aleppicum, I could not tell, but it is evident that G. strictumshould be excluded fromthe floraof Europe and Asia Minor. Geummexicanum Rydb. Specimens in habit much resembling G. macrophyllumhave been collected in southernMexico. They were also determinedas that species. As G. macrophyllumhas not been collected at any station nearer than Sierra Nevada in California, it is very improbable that that species should grow in southern Mexico. A closer examination of these specimens revealed that the petals and the fruitwere essentially those of G. strictum. As the habit is quite different,a new species G. mexicanumwas proposed and based upon these specimens. To this belong the following: VERA CRUZ: Mount Orizaba, I89I, HenryE. Seaton 251. HIDALGO:Sierra de Pachuca, I90I, Rose & Hay 5563; I906, Rose & Rose 11489. Geum urbanum L., a native of Europe and temperate Asia, has been introduced in this country and is well established at several places, especially at Cambridge,Massachusetts. Geum geniculatumMichx. The type locality of this species was given as Canada. This must have been a mistake, for the plant is known only fromthe mountain slopes of North Carolina and eastern Tennessee. Geum rivaleL. This is a native of North America as well as Europe and Asia. Rafinesque thoughtthat the American plant was differentand redescribedit as G. nutans Raf. As there was an older G. nutans,Steudel proposed the name G. Rafinesquianum for the Americanplant, but this was unnecessaryas that and the European one are identical.

HYBRIDS Hybrids in Geum are not unknown in Europe; why should they be in America? Geum intermediumEhrh., a hybrid of G.

* Coll. I: 80. RYDBERG: NOTES ON ROSACEAE 489 rivale L. and G. urbanumL., has been known for a long time. Geum rivaleL. hybridizesalso here in America and the following hybridshave been known. As the flowersof G. rivale are quite differentfrom those of most species of the genus, its hybridsare easily distinguished. Geum rivale X strictum,G. auranthacumFries, was the first one to be recognized. It was described from garden material, but the followingnative specimensare to be referredhere: ALBERTA: Macoun 200I6. NEW YORK: Catskills, near Hunter, I898, Britton; shores of Lake Champlain, I900, N. L. & E. G. Britton. Geummacrophyllum X rivale,G. pulchrumFern., is represented by the followingspecimens: VERMONT: Mendon, Eggleston. QUEBEC: Bic, I905, Williams, Collins & Fernald. The hybridsbetween the species of the G. strictumgroup are not so easily distinguished,partly because the species themselves are closely related and seeminglygrade into one another. As the intergradingforms are found in the region where the ranges of two species overlap, they may representhybrids. As examples of such hybridsmay be given the followingspecimens: Geumoregonense X strictum. UTAH: Logan, I910, GeorgeZundel 206 (in part). : Helena, I908, Butler798. COLORADO: Honnold, I90I, Tweedy4172. Geummacrophyllum X oregonense. BRITISH COLUMBIA: Scagit Valley, I905, J. M. Macoun 699I4. WYOMING: Crevasse Mountain, Yellowstone Park, I902, Mearns 2I9I. MONTANA: Tobacco Mountains, i909, Butler423I. OREGON: Oregon City, I905, Lyon 6o. Geummacrophyllum X strictum. MONTANA: Lake McDonald, I90I, MacDougal 959.

SIEVERSIA This genus was based on Sieversia anemonoides and hence monotypic. In I823 Robert Brown extended the genus to include all the Geums without articulate styles. This limitation was 490 RYDBERG: NOTES ON ROSACEAE retained up to I906, when Dr. Greene segregatedfrom it Acomasty- lis and Erythrocoma. In my opinion the former was rightly taken out, as it differsfrom Sieversia by the same character of the fruitthat separates Anemone fromPulsatilla. Erythrocoma on the other hand I can not regard as distinct genericallyfrom Sieversia. Its species differfrom the type of Sieversia only in the erect instead of spreading petals and a better developed hypanthium. If Erythrocomashould be kept distinct,Geum rivale should be taken out of Geum,as it differsby just the same char- *acters,and still that species frequentlyhybridizes with the other species. Dr. Greene made the followingstatements which are not exactly true: "Sieversia was founded on a Siberian under- shrub, low and slender,with almost rotate calyx and corolla, the formernearly clhorisepalous, and its maturestyles are long,filiform, plumose to the very apex." There are apparently three species which have gone under the name Sieversia anemonoides,of which one can be called an undershrub,the othertwo having the habit of Dryas octopetala,being stronglycespitose with the branchesmore or less covered by soil and moss. I think the type of Sieversia was one of these. The differencebetween this cespitose stem and the branching rootstock of Erythrocomais indeed very small. The styles of Sieversiaanemonoides Willd., or S. pentapetala(L.) Greene, are not plumose to the apex, forthe upper part is naked, soft, withering,as it is in Erythrocoma,the only differencebeing that it is very short, scarcely 2 mm. long. This naked portion of the style is found not only in the type of Sieversia and in S. ciliata and its relatives,but also in the S. montanagroup, to which S. radicata,S. Peckii and S. calthifoliabelong. The naked portion in most is rather persistent,although usually witheringin age, but in some species or even individuals it breaks off. It is how- ever never articulatedto the lower portionof the style as in Geum, where the upper portion, which is usually hairy, falls off very early. C. P. Smith* describes the style of Erythrocomaciliata as follows: "Style hooked, or articulated[italics mine],the terminal portionoften deciduous." I have seen it somewhat bent or even occasionally hooked, but never with a proper articulation. The soft upper naked portion,after it has withered,often falls offat

* Muhlenbergia8: 7. I9I2. RYDBERG: NOTES ON ROSACEAE 491 the junctions with the hard persistentlower portion,but there is no marked articulationat this point as thereis in Geum. As to the limitationof the species, it is impossible for me to follow Dr. Greene in his segregation. The originaltwo species of Pursh's Geum triflorumand Geum ciliatum are apparently well distinct; the intermediateforms may be hybrids. Erythrocoma campanulata Greene seems also to be a good species. In the North American Flora I also admitted Sieversia grisea (Greene) Rydb., S. canescens (Greene) Rydb., S. flavula (Greene) Rydb. and S. brevifolia(Greene) Rydb. If the firsttwo reallyare specif- ically distinctfrom S. triflora(Pursh) R. Br. and the last two from S. ciliata (Pursh) G. Don is questionable. The other species of Erythrocomaof Greene are nothingbut formsof those mentioned, due to local conditions. Sieversiapentapetala (L.) Greene (S. anemonoidesWilld.) is an Asiatic species reported for America only from the Aleutian Islands. The only specimens I have seen fromthis region were collected by Dall. Sieversiaglacialis (R. Br.) Spreng. This species has the floral lcharactersof the genus ErythrocomaGreene, but the habit suggests the species of Acomastylis. Some of the specimensare as follows: ALASKA: York Plateau, near Ip-muk, Port Clarence, I90I, Walpole I906; Port Clarence, I90I, Collier. SIBERIA: Whalen, I894, J. T. White. Sieversia campanulata (Greene) Rydb. is the best species of Erythrocomadescribed by Dr. Greene. The broad reddishpetals much exceedingthe obtusishsepals and bractlets. It seems to be confinedto the Olympic Mountains. The followingspecimens belong here: WASHINGTON: Olympic Mountains, Elmer 2529; WilkesExpe- dition352, 813 in part. Sieversia triflora(Pursh) R. Br. This is distinguishedfrom S. ciliata (Pursh) G. Don in the broader, less deeply cleft leaflets and the more persistentstyle-tips, as pointed out by C. P. Smith, but there is another character which seems to have been over- looked. In all the other species referredto Erythrocoma,the hypanthiumis roundedor even sunkenat the base, even at flower- ing time, and the bractlets are not much, if any, exceeding the 492 RYDBERG: NOTES ON ROSACEAE sepals in length; in S. triflorathe hypanthiumat least in,anthesis is acute at the base and the bractlets much exceeding the sepala in length. If so characterized,Sieverisa trifloratakes in all the formsgrowing on the prairiesor plain region east of the Rockies and extendinginto Montana and Wyoming. Erythrocomacinerascens Greene and E. affinisGreene I can not distinguishfrom it. The formeris a depauperate form,I take to- be due to a higheraltitude, representedby most specimensfrom the Black Hills and Wyoming. The latter is the high northern form with narrower leaflets and brighter coloration. If held distinct it had two older names than that of Dr. Greene, viz. Sieversia rosea Graham and GeumGrahami Steudel. Sieversia grisea (Greene) Rydb. The following proposed species of Erythrocoma,viz. E. grisea, E. arizonica, E. tridentata- and E. aliena, all by Greene, have many characters common with S. triflora,viz. the habit, leaf-formand toothing,pubescence, and styles,but the hypanthiumis not acute at the base, the petals, are broader and the bractletsshorter, slightly if at all longerthan the sepals. They were all described from Arizona and Chihuahua, but the same formsare common in Colorado and rarely as far north as Montana and Washington. They constitute the mountain representativesof S. triflora. Of these Erythrocomagrisea is the- firstone in the list. The type does not representthe common form,but is a depauperate formwith short leaves, standing in the same relationship to the common form as E. cinerascens- Greene stands to Sieversiatriflora. The commonbetter-developed form representsE. arizonica and is common in Colorado.. E. tridentatais a formwith looser pubescence and narrowerleaflets, otherwise not distinguishable from E. arizonica. E. aliena is, exactly the same as E. grisea. Erythrocomaaustralis I regard as a hybrid between Sieversia grisea and S. ciliata. See my notes in the North AmericanFlora. Sieversia canescens (Greene) Rydb. The forms of Erythro- coma of the mountains of California, Oregon and Washington differvery little fromS. grisea, as here understood,except in a denser pubescence and shorter bracts. It is doubtful if they should be regarded as a distinct species. I retained, although RYDBERG: NOTES ON ROSACEAE 493 withhesitation, the specificrank of these forms,on whichErythro- coma canescensGreene was based. Sieversia ciliata (Pursh) G. Don. This species is confinedto the Rocky Mountains and runningdown into the Cascades of Washington. It is found neitheron the plains nor in the Sierra Nevadas. In most of its range it is associated with S. grisea and intermediateforms are not lacking; they are perhaps to be ex- plained as of hybridorigin. The center of its distributionis in the northernRockies, while that of S. grisea is more common in the southern. Its range barely touches that of S. trifloraand only slightlyoverlaps that of S. canescens. Sieversiaflavula (Greene) Rydb. It was only with reluctance I admitted this as a species in the North American Flora. It is rather local in the mountains of Wyoming and Montana. The followingspecimens have been seen. WYOMING: A. Nelson 829, 750I. MONTANA: Chestnut& Jones io5; Rydberg&- Bessey 44I4. Sieversia brevifolia(Greene) Rydb. This is a derivative of S. ciliata. It is confined to the mountains of central Utah. S. Watson 318, doubtfully referredhere by Dr. Greene, I think should be referredto S. ciliata. The followingspecimens belong here: UTAH: Panguitch Lake, Jones6002g; Fish Lake, Jones5779g, 544I0; L. F. Ward378. SieversiaPeckii (Pursh) R. Br. This and the three following form a distinct natural group, which is represented in Europe by Sieversia montana and other species. S. Peckii has been confused with S. radiatumor has been made a variety of the same. It is, however, perfectlydistinct. It is confinedto the mountains of New Hampshire and Maine. Sieversia radiata (Michx.) R. Br. This is closely related to the northwesternS. calthifolia(Menzies) D. Don. It is con- finedto the regionof which Roan Mountain is the center,and the majority of the specimens in our herbaria are from that very mountain. Sieversiacalthifolia (Menzies) D. Don. This species is rather variable. In the high arctic regionsit becomes stunted and more hairy (var. congesta),and then has a quite differentappearance. Another cause for the many synonymscited under the species 494 RYDBERG: NOTES ON ROSACEAE has been the supposition that there are two species in the north- west representedby Geumcalthifolium Menzies and Geumrotundi- folium Langsd. It is evident that G. calthifoliumMenzies de- scribed in Rees' Cyclopedia was based on the plant known as Geumrotundifolium. Scheutz seemed to be of a differentopinion. He kept Geum rotundifoliumand G. calthifoliumdistinct. His descriptionof the latter (which by the way does not agree with that in Rees' Cyclopedia) seems, however, to be drawn from specimens of S. macranthaKearney or some related species and not fromS. calthifolia. Sieversia macranthaKearney. This species was distinguished by Kearney and his descriptionwas found in the manuscript of the Flora of Alaska, which has been in preparationfor many years by the botanists of the United States Department of Agriculture. As there was no telling when this work would be published, I adopted Kearney's name and description,with such slight modi- ficationsin formas to make it congruouswith the general style used in the North American Flora. I have seen no specimens except those of the type collection. It may be possible that Scheutz drewhis characterizationof G. calthifoliumfrom specimens of this species. ACOMASTYLIS As stated before,the writeragrees with Dr. Greene that the specimensreferred to it should be removed fromSieversia as well as fromGeum. They are closely related and the distinguishing characters are rather trivial, but as these trivial characters are supportedby differentranges, the writerthinks they are of specific value. Acomastylishumilis (R. Br.) Rydb. and A. Rossii (R. Br.) Greene. The formerseems sometimes to grade into the latter, but as its range is much more limitedthan that of A. Rossii, it is at least a geographical species. It is confinedin America to the Aleutian and PribiloffIslands, but is also found in easternSiberia, while A. Rossii extends well into Arctic America as far east as Melville Island. Acomastylisgracilipes (Piper) Greene and A. depressaGreene are only knownfrom the typecollections. They are closelyrelated to A. sericea,but the hypanthiumis more flat,not turbinateas i RYDBERG: NOTES ON ROSACEAE 495

is in A. sericeaand A. turbinata. A. gracilipeswas firstdescribed as a Potentilla. In this respect it has had the same fate as A. turbinata. Acomastylissericea Greene has a more western and northern distributionthan A. turbinata (Rydb.) Greene. The following specimenshave been seen. NEVADA: Ruby Mountains, Heller 9I39, 9359; East Hum- bolt Mountains, Watson 320; same locality, M. E. Jones I897. : Salmon River Mountains, Henderson4035. MONTANA:Indian Creek and Pony, Rydberg& Bessey 44I6, 44I7; Spanish Peaks, Flodman 6o8; Belt Mountains, Scribner 40; Rone Mountain, Chestnut9. WYOMING: Wind River Mountains, V. Bailey; northwestern Wyoming,Rose 290. Acomastylis turbinata (Rydb.) Greene. The range of this species covers that of the preceding,but includes also Colorado, Utah, Arizona and New Mexico.

COWANIA Cowania mexicana D. Don is not found within the United States, and not even near the boundary. It is confinedto central Mexico. The plant of southwesternUnited States and northern Mexico should be known as C. Stansburiana Torr., which differs in the formof the hypanthiumand the lobing of the leaves. C. mexicanais representedby the followingspecimens: MEXICO: Durango, I896, Palmer I2, 7I; Nelson 4696; Sierra Madre, Seeman; Hartweg. Cowania Davidsonii Rydb. is closelyrelated to C. Stansburiana, but differsin the elongatedpedicels, the formof the hypanthium, etc. The followingspecimens belong to it: ARIZONA: Blue River, September 8, I902, A. Davidson 754; Castle Creek, Bradstran Mountains, I892, ToumeyI29d. Cowania alba Goodding is unknownto me, except as to a poor fragmentcollected by Purpus and as to the descriptions. Purpus' specimens may well representa hybrid of C. Stansburiana and Kuntzia tridentata. Cowania ericaefoliaTorr. and C. Howardi S. Wats. are the same. The writer has seen the types of the two, which are 496 RYDBERG: NOTES ON ROSACEAE identical, but no other specimens. The species must be very local. FALLUGIA Three species have been proposedin thisgenus, viz. F. paradoxa (D. Don) Endl., F. acuminata Cockerell and F. micranthaCock- erell. Besides, the firsthas had two additional specificnames, viz. Geum cercocarpoidesDC. and Fallugia mexicana Walp. It has been impossible for the writer to distinguish more than one species, for the lobing or not lobing of the sepals is very in- constant, the same individual having both lobed and unlobed sepals, and the size of the petals is so variable, that no line can be drawn between F. acuminata and F. micrantha. In the essen- tially staminate plant the petals are usually larger.

KUNTZIA vs. PURSHIA Dr. Greene,* when adopting the name Kuntzia instead of Purshia, made the followingremarks: "A well-knownrosaceous type of Rocky Mountain and Californianshrub, at firstreferred to the South American genus Tigarea, was taken up by the elder De Candolle in i8i8, as a new genus, under the name Purshia. Sprengel,who about a year earlier,had himselfpublished a genus Purshia, soon afterproposed Kuntzia forthe name of the Candol- lean Purshia; and this will apparently be the proper name for the westerngenus now called Purshia, which latter name is more than once revertible;for Rafinesque had a Purshia in print as early as I8I3. I findno record of any earlier Kuntzia than this of Sprengel,which most writerswho have mentionedit, say was substituted for the Candollean Purshia in Sprengel's Systema, i825; but I findit fouryears earlierthan that, in the firstedition of Steudel's Nomenclator (I82 I). " There is m6re than one correctionto be made in the above statement. Purshia DC. was not published in I8I8. The publication is usually cited as Purshia DC., Trans. Linn. Soc. Bot. 12: I 57. I 8 I 8. The titlepage ofthat volume bears the year I 8 I 8, but the firstpart of it, in which Purshia appears, was published in i8 I 7. The firstpublication of Purshia Sprengel was neither in his Systema (I825), nor in Steudel's Nomenclator, but in

* Pittonia 2: 298. I892. RYDBERG: NOTES ON ROSACEAE 497

Sprengel's Anleitungen,*I8I7. Then comes the question which was the earlier,Purshia DC. or Purshia Spreng. Fortunately, Sprengelhimself when proposingKuntzia gave underPurshia DC., which he replaces, a referenceto the Supplement to Lamarck's Encyclopedie Methodique,t where Poiret publishes Purshia for De Candolle, the year before its technical publication by De Candolle in the Transactions of the Linnean Society. There is, therefore,no question concerningPurshia DC. antedatingPurshia Spreng. But how about Purshia Raf.? I have been unable to findit mentionedin any of Rafinesque's writingsof I8I3. The Kew Index gives the publicationof Purshia Raf. as "Am. Month. Mag. (I8I9) I9I." At the place referredto we find only the followingremarks under Onosmodium:"Sprengel has since given it the name Purshia, which had already been applied to another genus." Nobody can tell if this refersto an earlierPurshia Raf. It may just as well referto the earlier Purshia DC. Apparently the firstappearance of Purshia Raf. was in Journalde Physique: in I8I9. It is evident that Purshia DC. can not be thrownout on any other ground than by regardingBurshia Raf. (i8o8) an error in orthography. It was not a misprint,for Rafinesque states that it was named afterMr. Bursh and on the unpublished plate of Rafinesque's,the originalspelling is Burshia.

CHAMAEBATIA Chamaebatiaaustralis (Brand.) Abrams has been collected in Lower Californiaby Orcutt and Miss Irish and in southernCali- fornia by Pringlein I882 and by Chandler,no. 5214.

CERCOCARPUS Cercocarpusmacrophyllus C. K. Schneideris the most common- of the Mexican species of . Most of the material labeled as C. fothergilloidesbelongs to this species. Some of the specimensare cited below. VERA CRUZ: Orizaba, I892, J. G. Smith 199; Cuerta de San Juan del Estada, Liebman I7I9. HIDALGO: Pachuca, I905, Purpus I139.

* Ed. 2, 2: 450. I817. t V. 4: 623. i8i6. t 89: 257. I8I9. 498 RYDBERG: NOTES ON ROSACEAE

JALISCO: Leon Dequet. OAXACA: I894, Pringle587I. GUERRERO: Chipancingo,I903, Nelson 7068; between Chilapa and Texla, I894, Nelson 2168. Cercocarpusfothergilloides HBK. To this species the follow- ing are to be referred. PUEBLA: Cerro de Paxtle, Purpus 4200. VERA CRUZ: Orizaba, F. Mueller. CercocarpusTraskiae Eastw. has not been collected outside of Santa Catalina Island, California. Cercocarpusmojadensis C. K. Schneider. To this belongs the followingspecimens: COAHUILA: Sierra Mojada, I892, M. E. Jones 134; Sierra de Plata, I905, Purpus 0o59; San Lorenzo Canyon, i905, Palmer537. Cercocarpus Pringlei (C. K. Schneider) Rydb. This was originallydescribed as a varietyof C. mojadensis,but I thinkit is specificallydistinct. Cercocarpusmacrurus Rydb. This is the C. parvifoliusof the Klamath and Siskiyou region of California and Oregon. It is nearest related to C. montanusRaf., differingin the longerleaves with more numerouslateral veins and the large fruit. Although the charactersseparating it fromC. montanuson one hand and C. Douglasii on the other are not so striking,it is geographically separated fromboth, the formerbeing limitedto the Rockies and the latter to central and southern California. The following specimensbelong to it: CALIFORNIA: Siskiyou Mountains, i88o, Engelmann (fruit); same locality,I866, Henderson(flowers); Ashland, I866, Henderson 259 (flowersand young fruit); Klamath River at Keno, Cusick 2835; Klamathon, I903, Copeland3504; no locality, WilkesExpe- dition1167. OREGON: Weiner, I898, Walpole 8i; Barclay Springs, Modoe Point, Coville1523 in part. Cercocarpusmontanus Raf. There have been two species con- fused under the name of C. parvifoliusNutt. even fromits first descriptionin the Botany ofBeechey's Voyage by Hooker & Arnott. These authors adopted Nuttall's manuscriptname, but included in theirdescription not only Nuttall's type fromthe Rockies but RYDBERG: NOTES ON ROSACEAE 499 also specimens collected by Douglas in California. The latter were the base of the descriptionand figurein Hooker's Icones, plate323. It is a questionwhether Hooker & Arnottdid not have Douglas's specimens in mind when they drew the descriptionof C. parvifolius. It would, therefore,be some doubt as to whichthe name C. parvifoliusshould be applied, the low shrub of the Rocky Mountains or the more -like one fromsouthern and central California. Sargent referredboth to C. parvifolius,while C. K. Schneider, who without question has done the best and most critical work on the genus, referredthe latter to C. betulaefolius, yes, even made it the "var. typicus" of that species. Schneider laid more stresson the formand size of the teeth of the leaves and in this respect Douglas's plant is more like C. betuloides. If the pubescence and leaf-formare considered,it resembles more the plant of the Rockies. As said before,it is doubtfulwhich of the two should be regarded as C. parvifolius. As Nuttall's plant is included in the original descriptionit must be regarded as the type, but unfortunately,or ratherfortunately, it (i. e. the Rocky Mountain shrub) had already a name, C. montanusRaf., based on C. fothergilloidesTorrey, not that of Humboldt, Bonpland and Kunth; and C. parvifoliusbecomes a synonym. Hence the more tree-likespecies of California,represented by Douglas's specimens, was left without a name, and the writer proposed the name C. Douglasii in the North AmericanFlora. CercocarpusDouglasii Rydb. See the discussion above. Of this species I have seen many specimens. They are all from California,except the following: ARIZONA: Jucumba Hot Springs,I 894, Schoenfeldt335. LOWER CALIFORNIA: Nachaguere Valley, I894, Schoenfeldt 3432 and Mearns 3390. Cercocarpusrotundifolius Rydb. This is related to the pre- ceding species, but differsin the small broad rounded-ovalleaves; in C. Douglasii the leaves are obovate or oblanceolate, distinctly cuneate at the base. To C. rotundifoliusbelongs the following specimens: CALIFORNIA: Los Angeles County, I90I, Grant 3488; i850, C. C. Parry; I890, H. E. Hasse. LOWER CALIFORNIA . mountains,I882, Pringle. 500 RYDBERG: NOTES ON ROSACEAE

Cercocarpus alnifolius Rydb. This is the same as C. betulaefoliusBlancheae of C. K. Schneider,mainly; although that author had includedsome specimensof C. Douglasii mixed in with the type. As a species had already been named after Mrs. Blanche Trask, viz. C. Traskiae, it would be bad taste to propose another one. Besides personally I dislike the use of the given name of a person as formingthe specific name. I, therefore, adopted the more appropriate C. alnifol'iusinstead of the varietal name Blancheae. This species is also endemic to Santa Catalina Island, California. Cercocarpusbetuloides Nutt. Hooker in his Icones changed the name to C. betulaefolius,which form has been more commonly adopted than the original. It has also been regardedas a variety of C. parvifoliusNutt., i. e. of C. montanusRaf., but is evidently distinct. Cercocarpusminutiflorus Abrams. Nothing furtherhas been added to the knowledgeof this species since its publication. Cercocarpusflcabellifolius Rydb. In Utah and northwestern New Mexico is found a mountain mahogany, having the pubes- cence of the Californian C. betuloidesand the coarse toothingof the leaves of C. montanus. It has obovate leaves as the latter, but much broader and with more flaringteeth. This character is best seen in the type collected by L. F. Ward. The following specimensare to be referredhere: UTAH: Glenwood, i875, Ward I22; western slope of La Sal Mountains, I9II, Rydberg& Garrett8566; south side of Abajo Mountains, 9275; Juab, I902, Gooddingio73; Silver Reef, I894, Jones 5i63, 5i63b; Laccolite, Jones 5663, 5204e; Marysvale, Jones 5405d; Cedar City, Jones 5208, 5404d; Salina Canyon, Jones 544Im; Fish Creek Canyon, I909, Garrett2523. NEW MEXICO: Aztec, I 899, Baker 384. Cercocarpusargenteus Rydb. This is related to C. montanus, but the pubescenceis appressed-silky,the leaves narrower,toothed above the middle with smaller teeth. The followingspecimens belong here: TEXAS: Randall County, Eggers; Llano Estacado, Bigelow (Whipple Exp.); Guadalupe Mountains, I90I, Bailey 436. NEW MEXICO: El Capitan Mountains, I900, Earle 209; Same RYDBERG: NOTES ON ROSACEAE 501

locality, Plummer; Winsor's Ranch, I908, Standley4I04; Ratan Mountains, I903, Griffiths5497; White Mountains, I907, Wooton & Standley 3606; Sacramento Mountains, I899, Wooton;Las Vegas, I89I, Dewey; Folsom, I903, A. Howell I7I. COLORADO:North Cheyenne Canyon, I894, E. A. Bessey. Cercocarpuspaucidentatus (S. Wats.) Britton. This spccies was based on Cercocarpusparvifolius paucidentatus S. Wats. In order to determinethe type of the species, we must findthe type of the variety. The latter was based on ShaffnerII4, Parry & Palmer 224* fromSan Louis Potosi, and Wright I056 fromTexas or easternNew Mexico. The first,Shaffner II4, mustbe regarded as the type, but Parry & Palmer 225 is the same. Upon this very number C. K. Schneider based his C. Treleasii, which therefore becomes a synonym. Wright I056 belongs to another species, the same as Wilcox's specimen fromwhich Brittonmainly drew his descriptionof C. paucidentatus. This was withouta specific name, and I adopted Schneider's varietal name for it. The fol- lowing specimens belong to C. paucidentatus(S. Wats.) Britton or C. Treleasii C. K. Schneider. SAN Louis POTOSI: ShaffnerII4, 476, 635; Parry & Palmer 225. HIDALGO: Ixmiguelpan,I905, Purpus I383. Cercocarpus eximius (C. K. Schneider) Rydb. This is C. paucidentatusBritton, mainly as to the description,but not the type. Sargent regardedit as the same as C. brevifoliusA. Gray. Schneider firstadmitted it as a variety eximius, but afterwards adopted Sargent's views. He, however, did not have a clear conception of the same, for Rusby I25 and other specimens with better developed and more toothed leaves, he referreddoubtfully to C. betulaefolius. Leaves toothed above the middle are not uncommon and sometimesfound togetherwith perfectlyentire- marginedleaves on the same bush. It is to be admitted that it is closely related to C. breviflorus,but differsin the spreading pubescence and the longer hypanthium. It is also much more common than C. brevifoliusand its range extends throughNew Mexico, Arizona, Chihuahua and Sonora.

* This is evidentlya misprintfor 225, because Parry &' Palmer 224 is a species of Rubus and also cited by Watson under Rubus trivialis. It is now known as R. oligospermusThornber. 502 RYDBERG: NOTES ON ROSACEAE

Cercocarpusbreviforus A. Gray. This has nearly the same range as the preceding,but is more eastern, being found also in westernTexas and Coahuila. It is, however, lacking in Sonora and westernArizona. Cercocarpus ledifolius Nutt. is the most widely distributed species of the genus. Its characters are rather constant. It varies, however, in the width of the leaves and in the margin being more or less revolute. Cercocarpusledifolius intercedens subglabraC. K. Schneider is either an extremelynarrow-leaved formor else a hybridwith C. intricatus. Cercocarpushypoleucus Rydb. This has been mistaken for both C. ledifcliusand C. intricatus,but the villous pubescence of the lower surface of the leaves should exclude it from either, though it may cause some confusionwith C. arizonicus. It has smaller, narrowerleaves with sharper petioles than C. ledifolius, but largerleaves, less enrolled,and largerfruit than C. intricatus and C. arizonicus. The followingspecimens are referredhere: MONTANA: Melrose, I895, Rydberg 2695; Shear 32i6; Red Rock, Shear 3349; Helena, I908, Butler7i3, 774; Tobacco Moun- tain, Butler4236; Montana, Kelsey; Lombard, I900, Blankinship. WYOMING: Wolf Creek Canyon, A. Nelson 2292; between Sheridan and Buffalo,I900, Tweedy 3236; Big Horn, I899, Tweedy 2540; Powder River, I90I, Goodding 252; I893, Evermann; Big Horn Basin, I893, V. Bailey; Tongue River, I898, Tweedy39. IDAHO: Salmon River,Henderson 3I43 and 3790; V. Bailey 49. UTAH: Raft River, S. Watson 3i3 in part. OREGON: Snake River, I897, Sheldon 820I. Cercocarpusintricatus S. Wats. M. E. Jones reduced this to a varietyof C. ledifolius,claiming that they grade into each other; and still he proposed a new species C. arizonicus,which is much closer to C. intricatusthan C. ledifoliusis. I have seen the speci- mens fromWillow Spring, Arizona, on which C. arizonicus was based and these are identical with Jones's own specimensfrom Deep Creek, determinedby Jones as C. ledifoliusintricatus. It is evident that Jones did not distinguishthese species very well. Some of Jones's specimens are evidently hybrids. See below. The range of C. intricatusincludes parts of Utah, Arizona and Nevada. A specimen from California, but with much shorter RYDBERG: NOTES ON ROSACEAE 503 leaves, scarcely 5 mm. long, is referreddoubtfully to this species. It is VernonBailey 20I9 of the Death Valley expedition. Cercocarpusarizonicus Jones. This species is much more local than C. intricatus. It is closely related tQ it, perhaps not distinct If C. hypoleucuswere found in the regionthis might be a hybrid between that species and C. intricatus. UTAH: Deep Creek, I89I, M. E. Jones; Tropic, I894, Jones. NEVADA: Rock Mountains, I898, Purpus 6336?.

HYBRIDS Mr. Coville has collected specimenswhich are without doubt a hybrid between C. ledifoliusand C. macrurus. A specimen of the hybridand one of each of the two parents is mounted on the- same sheet, his no. i523 in the National Herbarium. When these two species hybridize,it would be expected that some of the more closely related species might do so. This may explain some of the intermediateforms between C. betuloidesand C. Douglasii, between C. montanusand C. flcabellifolius,and between C. ledifolius and C. intricatus. Also a specimen collected by M. E. Jones at Silver Reef in I894 seems to be a hybrid of C. ledifoliusand C. arizonicus. A specimen of the latter is included under the same number,5I49k, and C. ledifoliusis found in the region. NEW YORK BOTANICAL GARDEN