,

CONT.RIBUTIONS PROM THE GRAY HERBARIUM OF ~RVARD UNIVERSITY.

NF.W SERJES.- LXXTl I.

1. Records preliminary to a general TI'catmcll t of the Eupato-rieae,-IV. By B. L. ROBINSON . . . . . . . . 3

II. New Plants collected by Mrs. Richard C. CUI'ti8 in Portu-guese West Africa. By 1. M. JOHNSTON. . . . . 31

III. On some South American Proteaceae. By 1. 111. JOHNSTON 41 IV. Studies in the Boraginaceae,-II1. By r. lV1. JOHNSTON. 42

l. The Old World Genera of the Borayi",oi".", 42 2. Notes on miscellaneous American Borayinaceac 7;3

PUBLISHED BY

THE GRAY HEIWAIUU1I1 0]<' HARVARD UNIVERSITY CAMBRIDGE, 38, MASS., U. S. A.

1924.

72 JOHNSTON

tlus amplification of the generic limits of Cynoglosbum., but the rollow~ ing species, incorrectly referred to Paracarywn, may be. transferred.

Cynoglossum Thomsoni (Cla.rke), comb. nov. Paracaryu11l. Thom­aoni Clarke in Hook. FI. Brit. India iv. 161 (1883).

55. Rindera rallas, Reise i. 4SG (1771).

M ((Ilia Schult. Obs. Bot. 30 (1S09). Cyphon",Uia Boiss. FI. Orient. iv. 2i2 (lS;5). Bilegn"''' Brand in Fedde, Repert . xiii. 549 (1915).­. \bout 15 species, most of them in western Asia. and adjacent Europe.

50. Tysonia Bolus in Hook. Icon. xx. t. 1942 (IS90).

:\ South African monot.vpe wh ich is unrepresented in the Gray Herbarium. The plant originally figured by Bolus has peculiar sub­disciform, unarmed, and winged nutlets, characters wh ich, sub­stantiated by t he odd corolla-developments, seems to furni sh ample justification Tor the recognition of the genus. The p lant described and figured by Brand, Pflanzenr. iv. Farn. 252, 88, fig. II (1921) is oln'iously quite different in fmiting st ructures, having the depressed, broadly ovoid, densely glochidiatc, wingless nutlets of a Cynoglo88um .

57. Bothriospennum Bungc, Enum. PI. China Bor. 47 (1835). See n oy. Soc. Cat.

A well-nuu'ked genus of 3 to 4 species, all of them occurring in China and one of them ranging from northern India tn :rVlanchuria. and on the Japanese, Phil ippine. Mascarene. and Hawaiian islands . Though this g€'lllI S has been placed in the Eritrichieae it seems clear that its propel" posit ion is in the Gynogloaseae next to 1'hyrocaT1.llls. Bothr£­oSjJl'fm'll1n and this relative are remarkably simi lar in general ha bit and in fl oral structures, in fact appear to differ only in the direction of thei r nutlets and in the degree to which the dorsal margining is dC\'eloped on the latter. The two genera agree in having the attach­ment-surface of their verrucose nutlets proximate to the radicle end of the seed, or in other words have nut lets morphologically apical in attachment. In 1'ltyrocarpua the nutlets are divergent, but in BotliriospcrmuJn they are in\'ertoo, parallel and with their backs to­gether. ln Bothrioapcrnlutn the peculiar scar above t11C outlct-at­tnchment, which has been frequently mistaken for the real attach­ment-scar, is, hence, morphologically t11e equi \'alent of the dorsal cupulate structure on tlle nutlets of Thyrocarpus. The tumid margin of the scar on the back of the nutlets of Bothrioapermu7n is the homo­logue of the outer lobed dorsal margin on the nutlets of ThYTocarplla ,

STUDIES I N THE BQRAGl NAGEA t::

and the carti laginous veil within the tumid rim is the homologue of the inner one of the double margin of Tltyrocarpus. The cartilaginous veil, well developed in Bothrioapermwn, is also present, but early evanescent, in numerous species of OmlJhalodea and Paracaryu.m.

58. Thyrocarpus Hance, An n. Sci'. Nat. ser. 4, xviii. 225 (IS62) .

.-\ chinese genus of 3 species .

59. Actinocarya Benth. in Bcnth. & Hook. Gen . PI. ii. 846 (1876) .

An in teresting Illonotype from Tibet. According to Brand, Pflanzenr. iv. Fam. 252, 15 (1921), thi s genus is to be excluded from the Cynoglo3seae a.nd associated with }'/yo90tis, since its nutlets are said to be basifixed. My dissections of the type collection agree with the details in the plate given by Oliver, H ook. I con. xxiii. t. 2257 (1893), and clearly point to a relationship in the Cynogloaaeao. Kot only do the nutlets bear glochidiate appendages and a coroniform dorsal crest suggestive of the C1I"ogI088oao, but the rad icle end of the seed is next to the attachment-surface of the nu tlet and, hence, the latter is unquestionably apical in attachment.

GO. Paracaryum (A. DC.) Boiss. Ding. PI. Orient. scr. I, xi. 128 (1849).

.lfaaiast,m" nmnd in Fedde, Repert. xiv. ]50 (1915) .- " ·itl, 40 to 45 species in the area between the eastern lHedi terrnnean Basin and lndia . Thel'c scems to he no marked difl'el'cncc betwcen Para­caryw1t and ltfauiast.,u7n.

2. NOTES ON MISCE Lr~ NEOUS AMERICAN H ORAGINACEAE.

Cryptantha latifolia, sp. nov., annua laxe ramosa 7-30 cm. alta; l'amis brevitcr gracili terque hispido-strigosis ct sparse sctosis; fol iis subdistantibus fi rmis oblongo-linearibus vel oblongis obtusis integer­rimis )-3.5 crn. longis 2.5-8 (-10) mill . In tis sessilibus concoloribus pustulato-setosis; spicis solitariis vel gcm inatis pauci- \-el mul t i­bracleatis 3-J2 cm. longis unilaterali bus; floribu s ju vcntate congestis biseriatis matllritate remotiSj calycibus maitll'itatc ovoideis subscs­silibus ascendentibus 2-4 111m. Jongis, lobis lincari bus vel lineari­oblongis o!.>tusis breviter hirsutis; corolla conspicua calycem multo superant i 2.5-5 111111 . lata, lobis late ooovutis 1.2- 1.8 1111ll. longis aluis; nuclIlis 4 triangulari-ovatis cOlllpressis rusccsccntiuus hOlllolllorphis

74 J Oll NSTON

] .5-1.7 mm . longis apice acutis basi rotundo-truncatis dorso Uluricatis ventre verrucosis apice acutis, sulcis cIausis vel.infra medium anguste apertis basin versus divaricate furcatisi gynobasi quadrangulari. columnari ca. I mm. alta i stylo ca. 1 mm. longo lluculam ca. . 0.5 mm. superant i.- P ERU: loose stony upper slopes of sea-side ,hills, Chorrillos near Lima, about 150 m. alt., Sept. 15, 1923, Macbnde 5861 (TYPE,

Field Mus. no. 536,865 ; ISOTYPE, Gray Herb.).-Differing from C. lime"si.s (A. D C.) J ohnston in its broad rotate corollas that much surpass the calyces, broad leaves, larger calyces, and acute ovate muricate nut lets that are about half the length of the calyx . From the briefl y described C. granulosa (R . & P.) J onllston it di ffers in its broad leaves, loosely branched hal: it and muricate nutlets.

Cryptantha Macbridei, sp. nov. , annua basal iter ramosn 5-15 em. alta ' ramis paucis ascendentibus ndpresse hispidis j foliis oblanceolatis firmis in tegerrimis sessilibus saepe aeutis 1-2 em. longis 2-3 mm. latis saepe pustulato-setosis, superioribus reduetis, inferioribus oJ~­posit is majoribus; spieis soli tari is vel geminat is 3-) 2 em. long~s uni lateralibus; Aoribus obscure biseriatis omnibus angusto-hraeteabs roaturita te 2- 5 rom. separatis j ealycibus fructiferis oblongis sub­sessili bus ascendentibus vel divaricat is 4-5.5 mm. longis basi eonieis; lobis calycis oblanccolatis vel linearibus saepe acutis in costa infra medium cum setis fi avescent ibus horridis in marginibus adpresse hispidis, supra medium herbaceis paullo hispidis saepe recurvatis vel ascendent ibus; corolla tubulosa calyce vix longiori ca. 2 rom. ionga, lobis orbicularibus erectis ca. 0.5 mm. longis; uuculis 4 ovato-lance­olatis ca. 1.5 mm . longis fuscescentibus subni tidis verrucosis hetero­morphis margine obtusis apice anguste acutis basi rotundat is vent re 3/5 10ngitudinis ad gynobasem quadrallgulari-columnarem 1- 1. 2 mm. longam ndfixis, suleis c1ausis vel anb'1.lste apertis basi in areolam del­toideam semper di latat is; nucula axi llari persistent i dorso basin versus laeve quam nllculae consimile:i :illblongiori.- P ERU: loose stony upper slopes of seaside hills, Chorrillos near Lima, about 150 m. alt. , Sept. 15, 1923, M acbride 58G9 (TYP E, Field Mus. no. 536~ 873; ISOTYI'.E,

Gray Herb.).-Collected with and somewhat resemblmg C. lati/olta, but differing in its small tu bular corollas, coarsely hIrsute calyx­lobes with spreading herbaceous tips, and narrow~r ver:ucose h~ter~­mOl'phous nu t lets. l"rom the descript ion of C. ltnUmsUl, of whIch It appears to be a very close rela tive, it d in'ers in its much larger (4-5 not 3 mm . long) calyces, spreading calyx-lobes, and verrucose nutlets.

Crypt an tb a. peruviana, sp. nov. , annua grisea subsimplex vel laxe ramosa. 10-20 cm. alta ndpresse villoso-ll ispida inconspicue pustulata; ramis gracilibus ascendent ibus; roliis lanceolato-linearibus vellineari-

1

' ..

STUDI ES I N THE n ORAGI NACEAE 75

bus ascendentibus integerrimis sessiJibus acutis 1-3 cm. longis 1-2.5 mm. Intis adpresse villoso-hispidis costa et marginibus hispido-ciliatis supra viridioribus, superioribus paullo reductis; spicis saepe evidenter gemillat is rare ternatis vel soli tari is ebractent is )-5 em. longis uni­lateralibus; fl oribus uniseriatis maturi tate 2-5 mm. separa tis ; calyci­bus fru ctiferis ovoideis sessilibus ascendentibus 2-4 mm. longis, lobia erect is linearibu5 vel lanceolato-linenribus in costa cum set is Aavescent ibus crassis pungent ibus ornatis et in marginibus hispido­villosis; corolla tu l.mlosa quam calyx sublongiori ca. 2 mm. longa, lobis ol'bicularihus ca. 0.3 mm. longis ; nuculis 4 (rare 1-2 abort is) ca. 1.5 nun. longis lanceolato-ovat is compressis acutis pallidis l'ugoso­yerrucosis dense minu teque granuiat is quam lobi calycis 3 .'5 breviori­hus homomorphis margine acuti s basi truncatis ye.nt re ! longitudinis ad gynobasem quadrangulari-columnarem ca. 1 mm. longam affixa, sulcis aperti s vel clausis; stylo ca . 0.6 mm . longo nuculas subsupcrant i. - P ERU: in loose rocks on dry slopes above Rio Blanco, about 3600 m. al t ., May 8-19, 1922, M acbride & Feathers/(me 671, (TYPE, Field Mus. no. 517,202; ISOTYPE, Gray Herb.); ravines and hillsides on southern slopes of EI Chachani, north of Arequipa, nit . 3355 m. March 1920, M r. & Mrs. F. E. lIinkley 77 (G).- A very <listinct species with an erect habit of growth, short ebracteate spikes, a nd pale tuberculate rugose nutlets. The fi ve species of CnJptantha which are at present known from Peru may he dist inguished by aid of the following

K EY TO THE PEU.UV LA N SPEC IES OF CRYPTAN'l'lIA.

Curolla. a. t o.nthesis twice lenglh of calyx, limb broad . r~ave~ o.blong-lincur or oblong; nutlets muricate or verrucose ... C. lali/olia. Leaves hnear; nuliets granulose .......................... . C. granulosa.

CoroUa tubu lar, nlways shorter than or barely exceeding calyx. Spikes naked, usually geminate; nutlets pale; leaves linear or

lance-li near, very obscurely pustuin.te ..... ... ......... C. pen,viana. Spikes bracteate, usually solitarYi nutlets dark; leaves oblong

or oblaneeolnte, conspicuously pustulate. Calyx 4-'; nun. long, lobes spreading or recurv ing above

middle; nutlets verrucose . .. ...... . ....... . .... , .. . C. Macbridei. Calyx 3 mm. long, lobes erect; nu tlets granulate .......... . . C.limenai.s.

Am sinckia hispida (R . & P .), comh. no\". Lithospermm" hispidu11t R. & P . ]II. P eru v. ii . 5 (1 i99).- Ruiz & Pavon's specific name is the oldest ill th e genus, and hence should replace A. anguslifolia Lehm., cf. Contr. Gray Herb. n. s. Ixx. 44 (1924). The type of A . hispida was collected in the arid belt to the north of L ima, P eru.

Harpagonella. Palmeri, var. arizonica., n. var., cornibus calycis quam en formne typicae grncilioribus longioribus ca. 4 mm . longi-

76 J QII I\'STO:\

tudinc; Iltlcul is palilio grandiorihlls.-A HIZQX'.-\ : plain~J ] ,Qwell, l\iuy 3, 1884, 11" . F. P<tr·i .• 1t IG~ (TYP E, Gray Herb.) ; near Tucson, April J881 , Pringle J63; Tucson, Lemmon; Tucson, 1877, G'rcenc JIID.-The plant of Arizona Cl11'1'cntly referred to 11 . Palmeri Grtly differs from that of California and Lower California in having more elongate cornutc processes 0 11 t he fruiting calyx and noticeably larger nutlets. The type of II . ]J(tllJlcri, coll ected on Guac/elupe I sland by Palmer in 18 7.=), is qui tt· indis t inguisha.b le from the plant of coasta l Southern Ca.lifornia and adjacent Lower California.

l/ ur7j(L{jolu'fla waS placed in a spcciul trihe by Gi..irke, E. & P. Nat. Pflanzenf. i \', .\b t . JU, 1:30 ( I 95), and was so treated in a p':LI>er of ro," own, ContI'. Gray Herb. n. s. Ixx. 5 (1924). A rc('cnt. detai led stud,\' of lIaf/lfl(JolU'l/n, ho\\,('\,er, has cOIwjnced me that it has been tre<llC'd " 'ilh too milch (lil;nily, and that it clearly fall s in to the Cynoyl06'8t:ar n('xt to l 'eciocar,l}CL as first indicated by Gray, Proc .. \111, .Icad . xi. s,~ ( 1 ~7U) .

J/ Clr,J(I!IOIlt'llfl, has only two O\'u les. The nutlets which den.:·!op fl'0111 Ll lcse art! 5011l(' \\'11at difrel'cnt in shape, attachment, and pu­hesel'm'c. The axial nutlct (that next the dist inct calyx-lobes) is pubescent on all faci.'s, and is s l ightl ~' short('1' and morc loosely affixed to thc Kynoha se t han is the ahaxial nutlet. The abaxial nutlet is cnclosed b~' the peculiarly modified abaxial calyx- lobes, and is puhescent on ly on the face proximute to the axial nu tlet , In hoth nutlets the radicle-end of the secd is next the att.achment...cnd of the Illlth:t, and h('Jl('{', morphologicall ~' at least, the nutlets are apically aU:\clwd. It is qlli tc ('\'id{'nt, t herefore, that the nutlets must stand inverted with t hei l' Imeks ju xtaposed. .\ closc examination I'evt'als a llIargin surroundinK t he pubcsccnt inward ly fa.cing side of the abaxial nll t lt·t which furtht'!' slIgJ!('sts that th is is, indeed, morphologically the hack of the nu t let. The sty le, gynobase, and (.'orolla, the indur­al(·d I'ccllrn'd pedicel:;;, and all the \'('gct..'l.tive characters of IIar-1)(If/Olill/" arc t hose of PcclocaryCi, a genus " ,hich also has very elongate nutl ets apica lly attached, and hence therc seems to be every reason for rctUl'ning to titt" treatmcnt of Gray, I. c., and Bentha.m & Hooker, G(,Il . PI. ii. 8 1G ( lR7G), placing t he genus n('xt to Pcctocnrya and con­sidering' it no more than a h igh l ~' specialized and anomalous member of t he C,l/IIof/loss, 'Cfc.

P~:c.:TO(,A lty ,\ LA' /'EHlf'LORA (La m.) D C. Prod r. x. 120 (1846). I ' f/HL('ili~, \"al' . IJolirill1w J ohnston, Cont I' . Gray Herb. n. s Ixx.37 {192·1).- Thc study of thc matcria l of Pectocarya, recently collected hy ;\1 .... J. F. ~Iucbl'id e in P eru, has cOJl\' inced me that P. grnciHa. var. bolit:imut is synonymous with P. lnter-ijlora, since the habit-

STUD! t:S I N THE nORAGlNACEAE 77

differcnce, by \\ ll ich P (JT(tc'il.ia and p, lateriflora were distinguished, has proved illusory. The three species constituting the § EU]lcctO­canJu appear to be remarkably similar in habit, differing only in the shape and arming of the Jlutlets. PcctocanJu laterijlora is character­ized by its obo\'ute, rather than parallel-sided oblong-linear nu t lets, and appears to represcnt its section in P eru and Bolivia. "" eber­bauer, Engler & Drude, Veg. Erde xii . 136, fig. 5 (19JJ ), has given a suggesti\'e, but not accurately detai led illustration of P. laterijlora,.

CYNOGLOSS"M Ll.m'NSE WiJld. Sp. PI. i. 762 (1798) .- This species is based enti rely upon a plate and description given by F euillee, J our. Obs. Phys. ii. 765, t. 40 (17J4), who found the plan t in the "\'alh~e d'Ylo." As clearly s110wn by Feuillee's maps and di scussion , the locali ty known to him as "Ylo" is the same as the port near the southern boundary oC Peru now known under that name. The type of C. li11l C118C, hence, was not collected near Lima as Willdenow's inappropriate specific name, and Brand's statement, .Pflanzenr. iv. FRill . 252, 142 (1921), would suggest. As the plant has not been reported frol11 within 500 miles of Ylo it is possible that its repu ted oceurrencc there is the resul t of some confusion of data.

CYNOGLOSSUM TRlA NAEUM Werld. Chi or. And. ii. 90 (1859).-A fine plate of this was published by Oliver, H ook. I con. xxv. t. 2458 (1896) . Brand, Pflan zenr. iv. Fam. 252, 136 (1921), apparently intended to cite this illustration, but part of his citation has been omitted and the remainder, the plate-number, wa.s added to thc

. citation to Weddell's Chloris Andina. CORDIA GE RA SCANTII liS L . Syst. ed. 10,936 (17.59); not of Griseb.

FI. Brit. W. Ind. 478 (1861), nor Chodat, Bull . Soc. Bot. Geneveser. 2, xii : 209 (1920) . GCT<lacanthlla P . Browne llist. Jamaica 170, t. 29, fig. 3 (J 756). C geraac<tntiloidra HBK. Nov. Gen. et Sp. iii. 69 (18J8). -In 1910 l lrban, Sytnb. Antil. iv . • 1) 16, indicated that, as then used, the binomial, Cordia geradcanthm L., was incorrectly applied to the widely distribu ted trec with canesccnt, densely stellate calyces, and that the name is properlyapplicable to the relatively localized species of the 'West lndies and southern 11exico which has glabrous or spar­ingly hirsute calyces and larger {lowers, and which was described Bnd current as C. gcrMcantho,:dea HBK. Ten years later, in hi s paper on Cordia § G'crascanthus, Chodat, I. c., declared U rban's interpretation of C. geraacanthua L . to be incorrect and used the name in t11e tradi t ional sense, applying it to the widely distributed plant with stellate ca lyces. Further examination of this matter has recently been made to determine the correct speci!ic name fol' use by

78 JOHNSTON

Dr. \V. M. \oVhcclpr in h is puhlications on myrmecophytes. For the convenience of others the results of this study are here put on record.

Curd'ia ycrascnnlh1t8 L. is based upon the Jamaican plant which Patrick Browne, I. c. , described and figured under the name "Gera~ SC(LJlt1I1t$." nrownc's illustration, sllowi ng only the Aoral structures, pnrtrays a corolla of large size which has broad short obtuse lobes with conspicuous pinnate veining, a broad saucer-shaped throat, a stocky weakly ri bbed calyx, ano deltoid calyx-lobes. These char­acters definite ly a.<.;sociate Browne's plant with C. germcanthoides I-II3K. and prohibi t the use of the Linnean name fol' the plant with stel late calyces. I t is to be also noted that not only does Grisebach, J. c. , cite Browne's figure under" C. gerascant/toides HBK.," but he gin!s C. yerascanthoides HBK. as It common in the lowlands and mountains" of Jamaica, a.nd gives the plant with stellate calyces (under C. gerasccmth'll8 J acq .) as H rare" on tha.t island. Browne's plant was 1I0t rare, for he speaks of it as follows, It This tree grows in many parts of .Jamaica, and is generally esteemed as one of the best t imher woods in the island; it rises to considerable height, ... , especially in the low-lands, where it is most common , ..... " It is significant that concern ing thc J amaican occurrence of the plant with stellate calyces, Urban, I. c. (under C. aUiodow Cham.), com­ments.parenthetica lly as follows, " fortasse a d. "" ilson in t roducta ex d. Stapf. in li t." Since the identity of C. gerascanthoidr.a HBK. and GCTC!scan,t/lu8 Browne is certain from a study of Browne's plate a nd desc ription, and from distribut ional considerations, it is evident that GonNct gcra.scanth'll8 L. is, indeed, improperly appl ied to the widely d istributed plant with stellate calyces. Among its close relatives in the "" est Indies and Central America, C. gcrascanthm L. is readily recognized by its lu.rJ;e f1o\.".c rs, saucer-shaped tluoat hirsute or gla­hrescent stout weakly ribbed calyx-tube, and deltoid ca.lyx-lobes. 1t is known only from Cuba!, Isle of Pines !, Jamaica!, southern }\rI ex i co~ , and northern Central America. As Urban, Symb. Anti!. iv. 5 16 (I910) and vii i .• 174 (I921), has pointed Ollt, Cordia allioMra (It & P.) Cham. is the correct name ror the widely distr ibuted plant with stellate calyces, or, in other words, (or the one incorrectly current as If C. gCTaaCanthu3." Cordia aUiodora ranges from .Mexico and the \'Vest Indies southward along the Andes to Bolivia. A number of l: rit icn l species, doubtfu lly distinct from it, have been described from southern Brazil . adjacent Pa.raguay, and Argentina.