BOTANICAL MUSEUM LEAFLETS HARVARD UNIVERSITY PRINTED AND PUBLISHED AT THE BOTANICAL MUSEUM CAMBRIDGE, MASSACHUSETTS BOTANICAL MUSEUM LEAFLETS HARVARD UNIVERSITY VOLUME XIII BOTANICAL MUSEUM CAMBRIDGE, MASSACHUSETTS 1947-1949 TABLE OF CONTENTS NuMBER 1 (September 8, 1947) PAGE Studies in the Genus Hevea I By Ricuarp Evans SCHULTES ....... | An Erroneous Record of Hevea in Colombia By A. H. G. Aston anp Ricuarp Evans SCHULTES ............... 212 NuMBER 2 (November 19, 1947) Commentaries on Spiranthes I By Oakes AMES ............. #2197 NuMBER 8 (December 29, 1947) Chicha, A Native South American Beer By Hueu C. Cur.Ler ann Martin CARDENAS 33 NuMBER 4 (July 27, 1948) Vascular Anatomy of Orchid Flowers Bye. ds SWAMY 3. Sis Goa et ww eC NuMBER 5 (October 20, 1948) Studies in the Genus Hevea II By Ricoarp Evans ScHuurEs ....... 97 NuMBER 6 (November 24, 1948) Notes on Orchids of the American Tropics By CHARLES SCHWEINFURTH ...... . . 188 [v ] NuMBER 7 (January 17, 1949) Maize Granaries in Mexico By Erraim HERNANDEZ Nouocorz!. . . . . 158 Numer 8 (March 4, 1949) New Archaeological Evidence on Evolution in Maize g£ By Pau C. MANGELSDORF AND C. EARLE SMITH JR. ....... 6 ee ww ew ww 6218 NuMBER 9 (April 18, 1949) Plantae Austro-Americanae V By Ricoarp Evans ScuuLTES ...... . 261 NuMBER 10 (June 24, 1949) Plantae Colombianae XI By Ricnarp Evans ScHuLres ...... . 298 New Concepts in Styrax from Eastern Colombia By Jesus M. Iproso anp Ricuarp Evans MCHULTES: «4 «+» ¢ +s *% » a % & do ,s BS [ vi J INDEX OF ILLUSTRATIONS PLATE Ambelania Markgrafiana Monachino . . . XXXVII Bat Cave Maize. ....... . XXIII-XXVIII Begonia lutea Smith & Schubert . . . . . . XXXVI Brassia Allen L. O. Wms. ex C. Schweinf. . . . XII Chicha Industry ............ IV-VII Cunuria crassipes Muell.-drg. . . .... XXIX Cynometra Zamorana PR. EB. Schultes. . . . XXXIV Gongora armeniaca (Lindl.) Reichb.f. var. bicornuta C. Schweinf. & P. H. Allen... . XI Graffenrieda fantastica R. H. Schultes & BOVE -« Te Oe a s. eV, RV Hevea rigidifolia (Spruce ex Benth.) Mie Are. 4 SU oe 2 co VELL, EX SX XIX Lightia lemniscata Schomb. . 2. 2... 2... XXX Maize Granaries. .......... XIII-XXV Map showing range of Hevea rigidifolia . . Fig. 12 Oakes-A mesia cryptantha C. Schweinf. & P.H. Allen X Roupala colombiana R. H. Schultes . . . . XXXII Roupala saxicola Rk. H. Schultes . . 2...) XXXIII Schoenobiblus peruvianus Standl. . . . . . XXXI [ vii ] Septotheca Tessmannii Ulbrich . 2. 2...) | XXIX Spiranthes Beckii Lindl. 2 2. ww dT Spiranthes gracilis (Bigel.) Beck. 2... 2. TT Spiranthes gracilis var. floridana (Wherry) Correll 11, 111 Spiranthes Grayi Ames ........... .I1I Styrax rigidifolius [drobo & R. EH. Schultes XXXVIII Styrax rigidifolius forma yapobodensis Idrobo & R.E. Schultes ...........XXXVIII Vascular anatomy of Orchid Flowers. . . Figs. 1-11 [ viii | INDEX TO GENERA AND SPECIES ABUTA grandifolia (Mart.) Sandw., 263 splendida Krukoff & Moldenke, 263 ACACIA aroma Gill., 35,59 acahual, 182 aca llantu, 52 aca fahui, 47 AECHMEA corymbosa (Mart.) Mez, 294 Schultesiana (Mart. ) Mez, 294 tillandsioides (Mart.) Baker var. Kienastii (FE. Morr. ex Mez) L.B.Smith, 295 AERIDES, 62,68 AGAVE spp., 180,191 AGROPYRON repens Beauv., 234 a-hoon-da, 267 airampu, 40,51,54 algja, 34 ALOPECURUS pratensis L., 233 AMBELANIA Markgrafiana Monachino, 308, 312 Angostura-bitters, 14,15 AMARANTHUS caudatus L., 161,189 APOCYNACEAE, 308 ATRIPLEX canescens (Pursh) Nutt., 215 BALANOPHORACEAR, 264 Balso blanco, 303 Bat cave maize, 213 BEGONIA lutea Smith & Schubert, 304, 312 BrGontraceak, 304 bi, 34,58 BomBacackAk, 276,303 BOMBAX coriaceum Mart. & Zucc., 303, 307,327,332 borra de chicha, 51,52 BRAHEA sp., 179 BRASSIA Allenii L.O. Wms. ex C. Schweinf., 145,146,149 glumacea Lindl., 149 BRoMELIACEAR, 294 BULBOPHYLLUM, 62,67,68, 76 BuRMANNIACEAB, 296 bur-oo-ma, 282 [ ix ] cacao-simaron, 277 cantaros, 36,47,52 CASTILLA Ulei Warb., 298 forma lecithogalacta R.E. Schultes, 298 CATTLEYA violacea (H BK.) Rolfe, 261 caucho negro, 298 cencalli, 172,173 CEPHAELIS barcellana (Muell.-Arg.) Standl., 309 CEPHALOCARPUS Dracaenula Nees, 293,294 chancaca, 45 CHELONANTHUS chelonoides (L,f.) Gilg, 307 CHENOPODIUM Quinoa L., 34 chicha 33 chicha de mani, 34 chicha de quinoa, 34 chicheria, 53 CHONDODENDRON polyanthum, 264 toxicoferum (Wedd.) Krukoff & Moldenke, 264 chua conchu, 46 chuspillo, 40 cincalli, 167,168,173,175,179, 183,185,191 cincalote, 167,173,179,183,185, 191 clavo, 285 [x ] CLUSIA spp., 327 coca, 802 COELOGYNE, 62,68,76 comatl, 163 ComposiTak, 310 coscomate, 181 COUMA macrocarpa Barb. Rodr., 308 CRESCENTIA Cujete L., 45 CUCURBITA ficifolia Bouche, 45 cuescomate, 181 cuescumal, 180,181 cuescomatls, 158,161,163,164, 165,168,172,176,177,178, 183,185,189,190 culli, 36,40, 41 CUNURIA crassipes Muell.-Arg., 121, 265,266,267, 272 Spruceana Baill., 268,269 cunury, 122 CUPHEA cathargenensis (Jacq. ) Macbride, 305 racemosa (L,f.) Spreng., 305 strigulosa HBK., 305 curare, 264,289,291 CUSPARIA trifoliata (Willd.) Engl., 13, 14,15 CYMBIDIUM, 62,63,67,68, 69,75,76,83 CYNOMETRA Zamorana R,E.Schultes, 301, 302,311 CYPERACEAE, 293 CYPRIPEDIUM, 62,65,69,70, 87,88 da kwatai, 57 DENDROBIUM, 62,63,66,67, 68,74,76,83 DICTYOSTEGA orobanchioides (Hook.) Miers var. parviflora (Benth. ) Jonker ex Pulle, 296 di he dure, 34 di he tada, 34 DILLeENIAcEAE, 283 DIPLOCENTRUM, 62,63 do-ko-gay, 308 DUROIA hirsuta (Poepp. & Endl.) K. Schum., 309 e bi teni, 57 e bi data, 57 e bi data du duki, 57 e cua huara, 58 EHRHARTA, 234 aphylla Schrad., 233 e mi he, 34 enramada, 167,168,182,184, 185,192 EPIDENDRUM imatophyllum Lindl., 296 nocturnum Jacq. var. minus Cogn., 296 ERIA, 62,78 ERYTHROXYLON Coca Lam., 302 ESENBECKIA febrifuga (St. Hil.) Mart., 15 e ti hua, 58,59 EULOPHIA, 62,63,67,68,69, 78,87 epidendraea, 78,79 nuda Lindl., 78 ELYONURUS ciliaris HBK., 178 EvupuHorsBiAckak, 265 FALLUGIA paradoxa (D.Don) Endl., 215 FICUS chiribiquetensis Dugand, 307 327 FLAcourRTIACEAE, 284,304 Jorno, 265 GENTIANACEAE, 307 GEODORUM, 62 densiflorum, 80,82 GONGORA armeniaca (Lindl.) Reichb f., 143 var. bicornuta C. Schweinf. & P.H,Allen, 139,140 maculata Lindl., 144 var. latibasis C. Schweinf. & P.H. Allen, 144 GONGYLOLEPIS Benthamiana Rob. Schomb., 310 eriocladia Blake, 310 glaberrima Blake, 310 maroana Baill., 307,310 [ xi ] GOURLEIA spinosa ( Mol.) Skeels, 35,59 GRAFFENRIEDA fantastica R. FE. Schultes, 306, 307,311,312,327 sessilifolia Triana, 306 granero, 165 guyana, 14 HABENARIA, 63,66,71,88 § Euhabenaria, 62,73 monorrhiza (Sw.) Reichb f., 297 § Peristylus, 62,73 S Platanthera, 62,73 hanchi, 45,46,47,51,52 HELICONIA sp., 58 HELIOCARPUS popayanensis HBK., 303 HELOSIS guyannensis L.C. Rich., 264 HERRANIA albiflora Goud., 276,277 kanukuensis R. E.Schultes,277 laciniifolia Goud. ex Triana & Planch, apud Garcia- Barriga 277,281 lemniscata (Schomb.) R.E. Schultes, 277,281 purpurea (Pittier) R.E. Schultes, 282,283 HEVEA apiculata Baill,, 115 Benthamiana Muell.-Arg., 115,116,117,127,128,129, 130,131 var. subglabrifolia Ducke, 128 [ xii ] brasiliensis (H BK.) Muell.- Arg., 2,5,10,115,116,121, 123,131 var. subeoncolor Ducke, 10 camporum Ducke, 108 confusa Hems/l., 117,121, 122,123 cuneata Hub., 115 Duckei Hub., 115 guianensis Aubl,, 12,14,115 var. lutea ( Benth.) Ducke ro Schultes, 113,115,118, 127 lutea (Benth.) Muell.-Arg., 115 microphylla Ule, 1,2,3,4,5, 6,7,8,9,97,112, 116,128 var. major Pax, 1,2,4,9 var, typica Pax, 1,9 minor Hemsl., 1,2,3,4,5,6,7, 8,108, 116,130 nitida Mart. ex Muell.-Arg., 9,10,115,117,118, 123,132 var. toxicodendroides (R.E.Schultes & Vinton) R.E.Schultes, 11,108, 131,307,327 paludosa Ule, 115 pauciflora (Spruce ex Benth.) Muell.-Arg., 117,118,120, 121,122 var. coriacea Ducke, 107, 117,118,119,120, 121, 122, 123,130,132 rigidifolia (Spruce ex Benth.) Muell.-Arg., 2,97,98,99, 101,102, 106,107,108,109, 111,112,113,114,115,116, 117,118,119, 120,121,122, 123,124, 127,128,129,130, 131,132,272 Spruceana (Benth. ) Muell.- Arg., 3,7,8,117, 128,132 **toxicodendroides’’? R.E. Schultes ex P.H.Allen, 11 viridis Hub., 9,10,123 var. toxicodendroides R.E. Schultes & Vinton, 10,11 HippocrRaTEACckAk, 271 horno, 265 huai cha, 51 huatlal, 182 huilcaparu, 36,40 hui na métahi, 57 huittapu, 45 IBIDIUM floridanum Wherry, 27 ir-rro-chee, 290,291 i-spa-na-nai-ambe, 305 JATROPHA elastica, L.f., 13 JUNIPERUS spp., 215 kaima, 51,52 kketacha, 47 kketa ordinaria, 51 ku-pu-iu-mem-ba, 298 LAGENARIA siceraria (Mol.) Standl., 45 layacote, 45 LecuMmINosak, 301 LEITGEBIA ap., 327 LEPTOTHYRSA Sprucei Hook, f., 261,265 LIGHTIA lemniscata Schomb., 278,281 llaitu, 52 llantu, 52 LOCKHARTIA amoena Endres & Reichb,f. var, triangulabia (4. & S.) C. Schweinf. & P.H, Allen, 150 triangulabia A. & S., 150 LoGANIACEAE, 280 LUCUMA dolichophylla Standl. ex L. Williams, 289 LUISIA, 62 LyTHRACEAE, 305 maize, 213 Maize granaries, 153 MANIHOT esculenta Crantz, 33 mano tact, 58 maran, 41 maran una, 41 mate, 45 MAURITIA sp., 34,58 MAXILLARIA, 150 Jenischiana (Reichb f.) C. Schweinf. , 150 rufescens Lindl., 297 sanguinolenta (Lindl. ) C. Schweinf. , 150 MAYNA amazonica (Mart. ex Eich.) J.F.Macbride, 284 [ xiii ] longifolia Poepp. & Endl. , 304 var, phasmatocarpa R. E. Schultes, 304 pacifica Cuatr., 285 var. brachycarpa Cuatr., 285 var. pusilla R.E.Schultes, 285 MELASTOMACEAR, 306 MENISPERMACEAE, 261,263 MICONIA paradoxa Triana, 307 misqui kketa, 46,47,51,58 Morackark, 298 mukeando, 36 muko, 36,41,45,46 na sa puri, 34 NAVIA caulescens Mart. ex Schult f., 295 NEOTTIA gracilis Bigel., 19,23 NEUWIEDIA, 91 NYMPHAEACEAR, 264 OAKES-AMESIA, 133,139 cryptantha C. Schweinf: & P.H. Allen, 134,136 OBERONIA, 62,63,68,76,78 ODONTOGLOSSUM coronarium Lindl., 297 ohuatlapilli, 183 OPUNTIA Soehrensii Britt. & Rose, 40 sulphurea G. Don, 40,54 OrcHIDACEAR, 261,296 orchid flowers, 61 ORNITHIDIUM, 150 Jenischianum Reichb.f., 150 sanguinolentum Lind]l., 150 ORNITHOCEPHALUS, 135, 139 polyodon Reichbf., 151,152 palo de leche, 309 paltal hualtacu, 40,41 PAPHIOPEDILUM, 62,65, 69,71 patcha chenapii, 57 PAULLINIA seaberula R. E.Schultes, 271 PEREBEA lecithogalacta (2. E. Schultes) R.E.Schultes, 298 perol, 36,46,47 52 PHALARIS canariensis L., 233,234 PHOLIDOTA, 62,67,68,76 PINUS edulis Engelm., 215 ponderosa Dougl. ex P. Laws., 215 PLEUROTHALLIS ciliata Knowles & Westcott, 262 PLUMERIA sp., 307 PoLyGonacrkak, 301 PROSOPIS spp., 35,59 PrRoTEeaceaAr, 299 quautli, 189 quina, 264 [ xiv ] RHUS trilobata Nutt. ex T. & G., 210 RHYNCHOSTYLIS, 62,66,68 ROUPALA colombiana R. FE. Schultes, 299,300,311 saxicola R. E.Schultes, 299, 300, 307,311,327 Rusiackak, 309 Ruraceak, 265 SACCOLABIUM, 62 SALSOLA pestifer 4. Nels., 215 SALVIA hispanica L., 161 sa-pé-pa, 263,264 SAPINDACEAE, 271 SAPOTACEAE, 289 sapotilla, 276 SATYRIUM, 62 nepalense D.Don, 73 SAURAUIA rigidissima FR. FE. Schultes, 283, 284 SCHINUS Molle L., 39,51 SCHOENOBIBLUS coriaceus Domke, 289 peruvianus Stand/l., 285,286, 289 se-hé-pa, 290,291 seke, 51 SELAGINELLA, 309 SENECIO clivicolus Wedd., 51 SENEFELDERA, 271 chiribiquetensis R. Fe. Schultes & L.Croizat, 270,307,327 SEPTOTHECA Tessmannii Ulbrich, 272,276 seringa barriguda, 8,9 seringa serapo, 9 sina métahi, 57 SIPHONIA elastica Pers., 12 nitida Mart. ex Muell.-Arg.,9 rigidjfolia Spruce ex Benth., 101 SOBRALIA rosea Poepp. & Endl. , 262,298 SOLANACEAE, 292 SOLANUM apaporanum FR, E. Schuites, 292 soliman, 309 somo, 54,58 SPATHOGLOTTIS, 62,67,68, 76 SPIRANTHES Beckii Lindl., 17,18,19,22, 23,27,28 floridana (Wherry ) Correll, 27 floridana (Wherry) Cory, 23, 27 gracilis (Bigel.) Beck, 18,19, 20,22,23,24,27,28, 29 var, floridana (Wherry) Cor- rell, 19,20,22,23,24,27, 28,30 Grayi Ames, 17,18,19,20,22, 23,24 simplex Gray, 17,23 simplex Grisebach, 17,23 [ xv ] tuberosa Raf., 17,18,20,21, 22,23 tuberosa sensu Fernald, 17,18, 22,23,27 STERCULIACEAR, 276 STRYCHNOS, 261 brachiata PR. & P., 290 Krichsonii R.Schomb., 290 Jobertiana Baill., 290 Peckii B.L. Robinson, 291 rondeletioides Spruce ex Benth., 291 STYRAX, 325 ambiguus Seub., 328 calvescens Perk., 328 rigidifolius Idrobo & R.E. Schultes, 327,328,330,332 forma yapobodensis Idrobo & R.E.Schultes, 330,332 roraimae Peck, 327 sect. Eustyrax ser. Valvatae, 828 sutu, 51 tacu, 58 lada, 57 tambaqui serringa, 9 tapanco, 167,168,174,175,182, 184,185 tecti, 34 TERNSTROEMIA sp., 327 THYMELEACEAR, 285 tiki, 46 TiLtacrar, 303 tinajas, 166 TITHONIA sp., 182 TONTELEA attenuata Miers, 271 tostado, 34 TRIPLARIS Pavonii Meissn., 301 troje, 160,180,183,185,192 trove, 165 tuhuré, 54,58 tumu, 57 tupari, 57 tutuma, 45 tza-he-vee-ko, 285 uchukilla, 40,41 upl, 45,46,47 upi alpi, 47 Urtequillo blanco, 303 VANDA, 62 VANILLA, 62,66,68,76 Vascular anatomy, 61 vasya, 163 VELLOZIA phantasmagoria PF. EF. Schultes, 263,327 VELLOZIACEAR, 263 VICTORIA regia Lindl., 264,265 VOCHYSIA spp., 327 wah-puh, 269 wah-so-né-né, 267 wirki, 36,45,46 [ xvi | XYLOBIUM ZEUXINE, 62,63,76,78 squalens (Lindl.) Lindl. var. gracile (Schltr.) C. Schweinf., 298 giricua, 182,184,185,192 sincolohtli, 179,183 zapoterana, 276 soyate, 179 [ xvii ] ERRATA page 9, line 15 for carriguda read barriguda page 28, line 22 for 994 read t. 994 page 47, line 21 for (centaros) read (cantaros) page 58, line 30 for eti hua read e ti hua page 164, lines 4 and 19 for Cuezcomayxtlahuacan read Cuezcomaixtlahuacan page [201], Plate XVII for A read B; for B read C; for C read A Issued June 27, 1949 [ xix ] BOTANICAL MUSEUM LEAFLETS HARVARD UNIVERSITY CamsBripGre, MAssacHuseTTs, SEPTEMBER 8, 1947 Voi. 13, No. 1 STUDIES IN THE GENUS HEVEA I BY RicHarp Evans ScHuu tes ' 1. THE DIFFERENTIATION OF HEVEA MICROPHYLLA AND H. MINOR In 1899, Hemsley described and illustrated an unusual species of Hevea, naming it H. minor because of its small stature. The species was based upon fruiting material collected by Richard Spruce on the Rio Casiquiare in southern Venezuela. A few years later, in 1905, Ule described Hevea mi- crophylla, basing the species upon fruiting material which he had collected on the lower Rio Negro in Brazil. Five years later, the monographer Pax published variety typ- ica and variety major of this concept. In 1906, Huber (Bol. Mus. Goeldi 4 (1906) 683-634) suggested that Hevea microphylla might be synonymous with H. minor, pointing out several characters in which the two concepts, as described, agree. He admitted, how- ever, that there seemed to be differences in other charac- ters, so he chose ‘‘to consider HZ. microphylla a distinct species for the present.”’ Identifying a flowering collec- tion made by Ducke (Ducke 7027) in the lower Rio ‘Agent (Botanist), Rubber Plant Investigations, Bureau of Plant Industry, Soils and Agricultural Engineering, Agricultural Research Administration, United States Department of Agriculture; Research Fellow, Botanical Museum, Harvard University. eo Negro as representing Hevea minor, Huber published an amplified description of this species, which included a description of the flowers of the Ducke specimens (loc. cit. 684-635). He indicated that the two species appeared to be closely allied, although the flowers of Hevea micro- phylla were not known. In 1918, Huber still maintained them as distinct concepts, including them with Hevea brasiliensis (HA BK.) Muell.-Arg. in his series Intermediae as he had done previously (loc. cit. 622), but intimating that further studies might make it necessary to remove AZ. microphylla and H. minor from series Intermediae (he considered series Luteae and Intermediae to represent provisional classifications and not natural groups) and, together with Hl. rigidifolia (Spruce ex Bentham) Muell.-Arg., to form a new group (Bol. Mus. Goeldi 7 (1918) 202). Apparently accepting Huber’s determination of his flowering collection (Duchke 7027) as Hevea minor, Ducke, who had collected topotype material of HZ. mi- crophylla (Ducke HJ BR 23750) which agreed in all characters with Ducke 7027, reduced H. microphylla and AZ. microphylla var. major to synonymy under HZ. minor (Arch. Instit. Biol. Veget. 2, no. 2 (1985) 242). Re- cently, he has maintained this opinion (Bol. Teen. Instit. Agron. Norte no. 10 (1946) 20). Baldwin enumerated nine species which he accepts as valid: he includes Hevea minor but makes no mention of AZ. microphylla, thereby suggesting agreement with Ducke’s treatment of the latter as representing the same concept as the former (Journ. Hered. 88, no. 2 (1947) 54. W hile engaged in a study of Hevea in the Kew Her- barium in June, 1947, | was able to consult the type of HI, minor (Spruce 3457) as well as a sterile duplicate type (Ule 6025) and a topotype (Ule 60.23) of H. microphylla. It is now apparent that Hevea minor and H. microphylla [2] represent two distinct entities. It is also evident from a comparison of Ducke 7027 and Ducke HJ BR 23750 with the type material that the two Ducke collections represent HI. microphylla. This being so, the flowers which Huber described in his amplified description of Hevea minor (Bol. Mus. Goeldi 4 (1906) 684-635) are actually flowers of HZ. microphylla. Hevea minor is still known only from the type collection which is in fruit. It is unfortunate that the misinterpretation of these two concepts has become so firmly established in Hevea literature. The following observations are presented with the hope that they may lead towards clarification. The seed and capsule of the type of Hevea minor were very accurately illustrated by Hemsley in conjunction with his original description of the species (Hooker’s Icon. Pl. (1899) t. 2572). Perfectly complanate-ovoid with two very conspicuous flat surfaces ventrally and a very pronounced ridge dorsally, the seed measures 15 mm. long, 9 mm. thick, 11-12 mm. wide. The valves of the capsule, which have an unusually thick ligneous endocarp—nearly 8 mm. thick—in relation to their over- all size are only slightly contorted due to dehiscence; they measure 26 mm. long, 11 mm. deep. The epicarp of the capsule is apparently extremely thin in life. The capsule itself is perfectly globose with very little trace of a trisulcate condition, is not apically pointed, and measures about 25 mm. long and 25 mm. in diameter. It is borne ona slender peduncle about 4 cm. in length. Similarly, an adequate description and a clear illustra- tion of the critical structures of Hevea microphylla are available (Engler Bot. Jahrb. 85 (1905) 668, t. 1: k, 1). The seed is rather large, measuring, according to the type description, 20-25 mm. long, 12—15 mm. thick, and would suggest, in many respects, the smaller seeds of Hevea Spruceana (Benth.) Muell.-Arg. It is peculiarly [3 ] obovoid, much wider at the base than at the apex, is sub- quadrangular in cross section, has obsolete flat surfaces ventrally and absolutely no ridge dorsally. The seeds which accompany the topotypical specimen Ducke HJIBR 23750, which is, indeed, a very close match for Ule 6023 and 6025, are in complete agreement with those described by Ule and measure 25-28 mm. long, 15-16 mm. thick, 17-18 mm. wide at the widest point. The valves of Duchke HJ BR 23750 are large, measuring up to 45 mm. in length. They have an extraordinarily thin endocarp which is only 0.8 mm. thick! In dehiscence, they twist very tightly, due probably to the almost papery consistence of the endocarp. If we are to judge from Ule’s figure, the capsule of Hevea microphylla is elongate-ovoid-pyramidal with a very acute apex, slight- ly trisuleate with a conspicuous dorsal keel in each carpel, 40-50 mm. long, 830-40 mm. in diameter. It is borne on a robust peduncle 6-7 cm. long. There appear to be no significant distinguishing char- acters in the leaves of the two concepts except that the leaves of Hevea minor are definitely concolorous, whereas those of all collections of H. microphylla are very dis- colorous. Floral characters which might further separate them will not be available until AZevea minor is found in flower. I believe that the seeds of the two are so utterly distinct in size and shape that we are justified in regard- ing them as distinct species. There are no valid reasons whatsoever for Pax’s crea- tion of Hevea microphylla var. major. Pax gives as his basis for the variety ‘‘foliola majora, angustiora,’’ but Ule 6023 and 6025 as well as Ducke 7027 and Ducke HJBR 23750 show all possible intergradations in the size of the leaflets, and this is known to be a character of little taxonomic value in Hevea. It is clear that the description of Hevea microphylla var. major is the de- [ 4] scription of an individual collection and does not repre- sent a biological entity. Were the confusion which has been created by the re- duction of Hevea microphylla to synonymy under H. minor confined to nomenclature, it would probably not be so urgently in need of clarification. It has led to serious misunderstanding of the fundamental biology of the two plants. When Hemsley described Hevea minor, his basic diag- nosis was: ‘‘pro genere omnibus partibus parvis, semini- bus albis immaculatis,’’ and, following the description, he observed: ‘‘This is so very distinct in the smallness of all its parts, and particularly in its small white seeds, that we have not hesitated to establish it on incomplete material.’’ Later, Pax emphasized this unusual condition of the seed when he wrote, under Hevea minor: ‘‘. semina laevia, immaculata, triangulari-oblonga, alba. ... (Engler Pflanzenr. 4, 147 (1910) 125). Hemsley and Pax, both without field experience in Hevea studies, had not realized that the seeds which they were describing were white and without spots because they were unripe. Huber had pointed this out in 1906 (Bol. Mus. Goeldi 4 (1906) 633), explaining that he had noted in his own field work with Hevea brasiliensis and other species that unripe fruits can ripen and even open after separation from the tree, in which case the seeds do not develop normally. The fact that the seeds of the type specimen of Hevea minor were unripe was also stated by Ducke (Arch. Instit. Biol. Veget. 2, no. 2 (1985) 242). I have examined the seeds carefully and, basing my judgment on field experience in the collection and study of several tons of seed of Hevea brasiliensis, find that, while they were slightly short of complete maturation when Spruce collected the specimen, the hardness of their testa and their fully rounded out appearance are convincing evi- [ 5 ] 99 dence that full size, or very nearly so, and mature shape had been obtained. These, after all, are the characters of fundamental significance; the type of coloration which the seed would have had when ripe is of much lesser importance. It is necessary to consider the mode of dehiscence of Hevea minor and H. microphylla because great signifi- cance has been placed on this in the classification of the genus. Ducke (Arch. Instit. Biol. Veget. 2, no. 2 (1985) 243) interpreted the fruit of Hevea minor as follows: ‘*'This species was created... . on a specimen... . without flowers and with no mature capsule (the seeds are still white!) but sufficiently characterized by the form of the leaves and chiefly by the form, the consistence and the slow dehiscence of the capsule (see the half-opened cap- sule reproduced in Hemsley’s work, a feature which would be impossible in the case of any other known Hevea species).”’ In his key to the species of Hevea, Ducke (loc. cit. 225) says of H. minor: ‘‘Capsule.... opens with a slow dehiscence and lets the seeds fall in the water; the capsule then twists itself and remains for a long time adherent to the peduncle... .°” He separates Hevea minor from all other species on the basis of the presumed slow dehiscence of the capsule of the former as against an explosive shedding of seed in the latter (loc. cit. 221). This erroneous stand results from the belief that Hevea minor and H. microphylla are one and the same. Judging from the structure of the thick, woody valves of the capsule of Hevea minor, the capsule opens, as in all other species with comparably strong valves, more or less explosively. The valves are not strongly twisted, due to their shortness and their extremely strong ligneous endocarp which measures, as stated above, up to 8 mm. [6] in thickness. ‘They may dehisce as in some specimens of Hevea Spruceana where the explosion is less violent than in most other species, but the structure of the valves in- dicates definitely that dehiscence is explosive. The ‘‘half- opened fruit’? drawn by Hemsley’s artist and actually preserved in that state on the herbarium sheet is due, I believe, to the fact that, as shown by the whiteness of the seeds, the capsule was not quite mature and that the artificial heat applied in making the dried specimen caused it to open partially while still under pressure in the plant press. This has often occurred when I have been drying Hevea specimens by means of artificial heat. Unless we can find some definite evidence of slow open- ing and gentle dropping of the seed in Hevea minor, we are making an unsound assumption in believing that such a mode of dehiscence is normal. In Hevea microphylla, however, there is indeed very strong evidence that slow dehiscence is the rule, for the entire structure of the valves is distinct and is such that a violent opening and shedding of the seeds would be mechanically impossible. The valves are provided with an excessively thin—0.3 mm.—endocarp which is coria- ceous, not even woody. Clearly, this structure is too weak to build up the tension necessary for a violent bursting. The valves open slowly, as Ducke has pointed out, and let the seeds drop slowly to the ground, per- sisting for some time on the peduncle (as often is the case in Hevea Spruceana) and then, when the seed is shed, twist themselves rather tightly due to their almost papery consistence. ‘This is indeed unusual in the genus. There would appear to be a strong ecological differ- entiation between Hevea minor and H. microphylla. The former is known only from the dry, sandy scrub-forest or caatinga of the Casiquiare. The latter, so far as avail- able collections indicate, inhabits forests which are peri- [7] odically subject to very deep flooding. Spruce does not give us in his notes any indication of the shape of the trunk of Hevea minor, stating merely: ‘‘Arbor parva, 15 pedalis.’’ If the trunk had deviated from a normal cylindrical shape, this careful observer would certainly have noted the fact. Hevea microphylla responds to its flood habitat by producing a swollen or bellied trunk, and it is called seringa barriguda (=bellied rubber) by the natives of Sao Joaquim on the Rio Negro, the same common name which is applied to H. Spruceana (Ducke : Bol. Teen. Instit. Agron. Norte no. 10 (1946) 21). CLAVIS SPECIERUM HEVEAE MINORIS ET H. MICROPHYLLAE: A. Arbor parva 15 pedalis, cum trunco probabiliter cylindrico, in caatinga occurrens. Folia concoloria, Capsula perfecte globosa, 25 mm. X25 mm., cum pedunculo subgracili, 40 mm. longo; valvis 26 mm. longis, vix contortis, endocarpio lignoso, usque ad 38 mm. crasso. Semina parva, complanato-ovoidea, 15 mm. X 9 mm. X 11-12 mm., carina dorsali conspicua. Dehiscentia prob- abiliter explodens. 1. Hevea minor AA. Arbor parva vel mediocris, cum trunco infra incrassato, in sylvis profunde inundatis occurrens. Folia discoloria. Capsula ovoideo- pyramidalis, apice valde acuta, 40-50 mm. * 30-40 mm., cum pedunculo robusto, 60-70 mm. longo; valvis 45 mm, longis, valde contortis, endocarpio vix coriaceo, valde tenuissimo, 0.3 in diametro. Semina magna, elongato-obovoidea, subquadrangu- laria, 20-28 mm. 12-16 mm. < 17-18 mm., ecarinata. Dehis- centia ut videtur lenta. 2. Hevea microphylla 1. Hevea minor Hemsley in Hooker’s Icon. Pl. 26 (1899) tab. 2572; Pax in Engler’s Pflanzenr. 4, 147 (1910) 125. Venezurta: Rio Casiquiare, “‘in sylvis humilioribus. Arbor parva, 15 pedalis. Siphonia.’’? Richard Spruce 3457 (Tyrus Herb. Kew). [In Spruce’s field note-book, preserved at Kew, the following annotations appear: “°3457. Siphonia—Casiq. Caatinga. Sm. tree 15 ft. Lts. sm. lane. acum. obt. Frt.’?] [8 | 2, Hevea microphylla Ule in Engler’s Bot. Jahrb. 35 (1905) 669, tab. 1: j,k, 1,m; Ulein Kautschukgewinnung (Kolonialwirtsch. Kom. 1905) (1905) 10; Huber in Bol. Mus. Goeldi 4 (1906) 634 pro parte [non accurate, sub Hevea minore|, 636; Pax in Engler’s Pflanzenr. 4, 147 (1910) 125; Ducke in Arch. Instit. Biol. Veget. 2, no. 2 (1935) 241 pro parte, tab. pag. 246, a-f, 247, a—b [non accurate, sub Hevea minore]; Ducke in Bol. Técn. Instit. Agron. Norte, no. 10 (1946) 20 pro parte. Hevea microphylla Ule var. typica Pax in Engler’s Pflanzenr. 4, 147 (1910) 126. Hevea microphylla Ule var. major Pax in Engler’s Pflanzenr. 4, 127 (1910) 126. Brazit: Estado do Amazonas—Rio Negro. ““Inseln Xipard, Sao Joaquim. [Nom. vulg. =] seringa serapo. E. Ule 6023, February 1902 (Herb. Kew).—‘“‘Inseln Xiparé, Sao Joaquim. [Nom. vulg. = ] carri- guda. E, Ule 6024, February 1902 (Typus Heveae microphyllae var. majoris non vidi). —‘“‘Inseln Xipari, Sao Joaquim. [Nom vulg. = ] tambaqui seringa.’’ E, Ule 6025. February 1902. (Typus pupiicatrus Herb. Kew).—‘‘ Barcellos.’’ 4. Ducke 7027.—‘“‘Insula Xiparé, prope Sao Joaquim... . silva profunde inundata. Arbores (3 ex.) parvae vel mediocres, trunco infra incrassato, flor pallide luteis. [Nom vulg. = ] seringa tambaqui vel seringa barriguda.’’ A. Ducke Herb. Jard. Bot. Rio 23750, August 8, 1931 (Tororyeus Herb. Kew; U.S. Nat. Herb.). 2. A NEW INTERPRETATION OF HEVEA NITIDA AND ITS VARIETY Hevea nitida Martius ex Mueller-Argoviensis in Martius FI]. Bras. 11, pt. 2 (1874) 301. Siphonia nitida Martius ex Mueller-Argoviensis loc. cit. Hevea viridis Huber in Bull. Soc. Bot. France 49 (1902) 48; emend. in Bol. Mus. Goeldi 7 (1910) 2385. In his forthcoming ‘‘A Study of Hevea with its eco- nomic aspects in the Republic of Peru,’’ Dr. Russell J. Seibert reduces Hevea viridis to synonymy under A. nitida on the basis of new characters which he has found [9 ] to be of great value in the classification of the species of this genus. Hitherto, (in Bot. Mus. Leafl. Harvard Univ. 12 (1945) 7) I have believed Hevea nitida to represent the concept which Ducke described as HZ. brasiliensis (H BK.) Muell.-Arg. var. suwbconcolor. 1 am now, however, in complete agreement with Seibert’s point of view that Hevea nitida is a distinct concept representing the same species as that which has been known as AZ, viridis. Seibert writes: ‘*They [certain specimens of Hevea viridis| appear to represent topotypical cultivated mate- rial from Huber’s type locality of HZ. viridis. The one flowering collection made by Archer is quite referable to H. nitida in floral morphology, the short-shoots, and in general, the leaflets. The lower leaf surfaces of this and other specimens of the cultivated plant, however, tend to show a minute lepidote condition slightly atypi- cal of H. nitida. The scales, notwithstanding, are neither of sufficient size nor density to affect the concolorous aspect. There remains some question, since the leaflets do show a slight AH. brasiliensis aspect, whether or not Huber’s A. viridis had some admixture of AH. brasilien- sis germ-plasm. ‘Through the excellent photograph (made by the Chicago Field Museum) of the entire type specimen of Martius’ collection deposited in the Herbarium at Mu- nich, it has been possible to identify HZ. nitida with HZ. viridis with some degree of certainty. The presence of interflush short-shoots as well as the glossy under surface of the leaflets leaves little doubt that HZ. viridis should henceforth be referred to HZ. nitida.”’ In view of this new understanding of Hevea nitida, it is necessary to alter the name of the diminutive, shrub- by Hevea, described as H. viridis var. toaicodendroides [ 10 | from the ancient remnant quarzite hills of the A paporis- Vaupés area of Colombia. Hevea nitida Martius ex Mueller-Argoviensis var. toxicodendroides (2. Ll’. Schultes et Vinton) R. E. Schultes comb. nov. Hevea viridis Huber var. towicodendroides R. E. Schultes et Vinton in Caldasia 8 (1944) 25. ‘** Hevea toxicodendroides R.E.Schultes ex P.H. Allen in Mo. Bot. Gard. Bull. 82 (February, 1944) 50; no- men nudum. [ 11 ] AN ERRONEOUS RECORD OF HEVEA IN COLOMBIA BY A. H. G. Ausron’ ano Ricuarp Evans ScHuULTES? In Rees’ ‘‘Cyclopedia’’ (89 (1819) sub Stphonia), there is a most interesting reference to a Mutis collection of Hevea guianensis Aubl. (‘‘ Siphonia elastica’’) from Co- lombia. Sir J. E. Smith, who compiled the section of the encyclopedia on Siphonia,* wrote: ‘‘There is, in- deed, in the Linnaean herbarium, besides the original specimen,‘ marked with this last name [S¢phonia elastica], another from Mutis, which that learned botanist judged to be a distinct species, though affording, as some other trees do, a similar gum. The leaflets in this specimen are larger, more acute at each end, and destitute of partial stalks. The calyx is nearly half an inch long. The ' Principal Scientific Officer, Department of Botany, British Muse- um (Natural History), London, England. * Botanist, United States Department of Agriculture; Research Fellow, Botanical Museum of Harvard University. ® Lady Smith: “* Memoir and correspondence of the late Sir James Edward Smith, M.D.’ 1 (18382) 488-489; B. D. Jackson: ‘‘An at- tempt to ascertain the actual dates of publication of the various parts of Rees’ Cyclopedia’’ (1895) 3. ‘ This statement undoubtedly refers to a specimen of Hevea guianen- sis preserved in the Linnaean Society in Smith’s herbarium. From all appearances of the specimen, it is one of the Aublet collection from French Guiana and represents the type of the genus. [ 12 ] younger Linnaeus, in his Supplementum 422, promised to publish something at a future time respecting the various trees that yield an elastic gum of the same utility as the Caoutchouc; but he did not live to execute his design. ’”’ From the point of view of the taxonomic history of Hevea, the genus of the Para rubber tree, the existence of a Mutis specimen would be of the utmost significance. It would represent the earliest known collection of the genus from Colombia, the northwesternmost sector of its range. A Mutis specimen would also be of extreme importance as it would alter our understanding of the distribution of Hevea, for, so far as we are aware, no Mutis material was collected in those parts of the Ama- zon and Orinoco drainage areas where the genus is known to occur. For these reasons, we consulted the Mutis material to which Smith made reference and which is preserved in Linnaeus’ herbarium at the Linnaean Society. The speci- men, included in the Jatropha folder, is not a Hevea but represents the rutaceous Cusparia trifoliata (Willd. ) Engler,’ the type of which was collected in Venezuela. Comprising several leaves and an inflorescence of fer- tilized flowers from which the corollas have dropped, the specimen is mounted upon paper bearing a Spanish water- mark. In the upper right corner of the sheet, there is, in Mutis’ handwriting, a figure ‘‘89.’’ At the bottom of the sheet, the elder Linnaeus had written ‘‘ Hevea’’ on one line and “‘elastica’’ below it, an epithet which has never been published. The younger Linnaeus scratched out the word ‘*‘Hevea’’ and substituted ‘‘ Jatropha,” but he did not cite the collection in his Supplementwm 422 under Jatropha elastica to which concept he reduced ° Tabula in Humboldt. Plant. Aequin. 2 (1813) t. 97. [ 13 | Aublet’s Hevea guianensis. Elsewhere on the sheet, ap- parently in the elder Linnaeus’ hand, there is an anno- tation ‘‘Gummi elastique.”’ A detailed search through the collection of Mutis’ correspondence with Linnaeus, preserved at the Lin- naean Society, revealed an enumeration of two ship- ments of plants from Colombia to Sweden. The speci- men in question was included in the second shipment. We find that Mutis had made, under ‘‘89’’ in the enu- meration of the specimens of this shipment, the follow- ing interesting annotation: ‘‘Pro Chinchona habita ab incolis guyanae.’” Even though the term guyana was rather loosely employed in this early period to designate much of southern Venezuela and the Orinoco basin, it has been impossible for us to ascertain Mutis’ source for this statement. It is, of course, highly significant as an observation, because Cusparia trifoliata is the source of Angostura-bitters and has been used rather widely in South American folk-medicine as a febrifuge. Humboldt states that ‘‘On the coasts of New Andalusia, the cuspa is considered as a kind of Cinchona.”’ ° Further study of Mutis’ correspondence has failed to shed any light on Linnaeus’ source for his annotation ‘*Gummi elastique.’” It is most probable, in our opin- ion, that the note was added somewhat casually for the benefit of students after the plant had been determined, to Linnaeus’ apparent satisfaction, as representing the concept now known as Hevea guianensis. Cusparia tri- foliata, of course, is not a latex-bearing plant. There is an additional annotation on the sheet. It is in pencil, apparently in the handwriting of Smith, and states that the specimen is markedly different from a sterile specimen from Brazil which Linnaeus had likewise ® Personal Narrative (translated by H. M. Williams) 3 (1822) 27. [ 14 | annotated ‘‘Gummi elastique’’ but which had not been labelled with a Latin epithet. This Brazilian specimen ' is neither Jatropha nor Hevea but, curiously, would seem also to be a rutaceous plant, Msenbechkia febrifuga (St. Hil.) Mart., likewise a source of bitters and often, in an earlier period, used to adulterate or falsify true Angostura-bitters. It naturally does not represent a lac- ticiferous plant, and the annotation with reference to gum elastic is as erroneous as the analogous annotation on the sheet of the Mutis specimen of Cusparia trifolata and was undoubtedly made for the same purpose. We wish to express our appreciation to Spencer Savage, Esq., Assistant Secretary of the Linnaean So- ciety, for his very kind help in the study which has led to the clarification of the erroneous record of a Mutis collection of Hevea. 7 The Brazilian specimens in the herbarium of Linnaeus were prob- ably collected by Joaquim Velloso de Miranda (1733-1815) and sent to Linnaeus by Domenico Vandelli (1735-1816). He collected in Minas Geraes. [ 15 ] WIP al a BOTANICAL MUSEUM LEAFLETS HARVARD UNIVERSITY Voi. 138, No. 2 CampripGr, Massacuusetts, NoveMBER 19, 1947 COMMENTARIES ON SPIRANTHES, 1 BY OakKES AMES I. SPIRANTHES TUBEROSA sensu FERNALD In JANuARY 1946', Professor M. L. Fernald resur- rected from outdated literature the obscure species Spi- ranthes tuberosa Rat. (1888), type specimens of which are not known to exist. In seeking some element in our flora with which this ambiguous Rafinesquian product, interpreted by a twenty-one word description, might be specifically identical, Fernald hit on the ‘‘white lipped S. Becku Lindl.’’ (1840), stating that there was certainly no doubt about it. He failed, however, to explain how he discovered that S. tuberosa had a white lip. Rafinesque certainly gave no clue, for, in his brief description, he dis- missed the lip with just two descriptive terms. He used the words ‘‘cuneato acuto,’’ which do not apply to the lip of any known species of Spiranthes in the flora of the United States. Fernald claimed that S. twberosa not only is identical with and antedates S. Beckit Lindl., but is also the characteristic species described in 1867 by Asa Gray as S. simplex (a homonym of S. simplex Grisebach and a synonym of S. Grayi Ames). In August 1946’, Fernald discussed the identity of Spiranthes tuberosa at greater length. He showed that, 'Rhodora 48 (1946) 10. ? Rhodora 48 (1946) 189, [17 ] from the bibliographic evidence, Spiranthes Becku Lindl. ‘*taxonomically ’” was as hopeless a muddle as could be imagined since it was concocted from elements of several different species. He pointed out in detail how Lindley, at a time when nomenclatural procedure was governed by less rigid rules than are now observed, adopted the specific name ‘‘ Beckii’’ in violation of all obligations to the synonymy cited. Fernald, consequently, quite prop- erly dismissed the name, S. Becki, as being doubly il- legitimate. He also recognized two strongly marked geographic variations in S. tuberosa, a southern one and a northern one. He assigned the species name to the southern element and reduced S. Gray to varietal status for the northern element. Actually and taxonomically Spiranthes Becki is the ancient, well-established and well-known S. gracilis (Bigel.) Beck, which nomenclaturally antedates S. tu- berosa by nine years. When John Lindley, in his ‘*Genera and Species of Orchidaceous Plants’* (1840), described 8. Becku, he cited two collections of specimens as being referable to it: one from Massachusetts (without collector), the other from Louisiana (Drummond 8382). The type sheet of SS. Becki in Lindley’s herbarium at Kew contains these two collections, and also a third con- sisting of two plants from Albany, New York, collected by Beck. As Lindley did not refer directly to these New York plants in his description, it may well be that they were acquired after 1840 and consequently should be ex- cluded from any discussion of the Lindleyan type of SS. Becku. (It may be noted, however, that camera-lucida drawings of the lips of the New York plants, made by Victor S. Summerhayes, show that in outline and neu- ration they exemplify 8. gracilis (Bigel.) Beck). The Massachusetts specimen on the type sheet (col- lector unknown) is accompanied by the following note: [ 18 ] Spiranthes aestivalis var. gracilis Torr. mss. S. gracilis Beck Neottia gracilis Bigelow fl. bost. ed. 2. This hardly differs specifically from S. aestivalis. The common variety not infrequently bears a small ovate radical leaf, Massachusetts. Mr. Summerhayes has also made for me a camera- lucida drawing of the lip of this Massachusetts plant. There can be no doubt about its being typical S. gracilis. Indeed, when Asa Gray described SS. simplex in 1867, in the fifth edition of ‘*The Manual,’’ he was familiar with Lindley’s type of S. Becki and referred the latter species to S. gracilis, ‘‘at least as to the northern plant.’* This reference to the northern plant was made because the type sheet of S. Becki also carried the specimens from Louisiana, In our discussions of the components of Lindley’s illegitimate S. Bechii we must give close attention to these Louisiana collections made by Drummond in 1832 near New Orleans, especially so since Asa Gray believed Drummond’s 382 to represent plants that were different from the northern SS. gracilis. In the Gray Herbarium there is a duplicate of Drummond’s collection annotated by Gray as S. Becki. In 1904, | accepted this Drum- mond plant as being referable not only to S§. Bechii Lindl., but as being conspecific with S. Grayi Ames (S. simplex Gray, not Grisebach). More recent studies have proved that S. Gray? is a distinct species. I am indebted to V.S. Summerhayes again for his great kindness in sending me a carefully executed camera-lucida drawing of the lip of Drummond’s 882 in the Lindley Herbarium. The Kew plant is identical with the duplicate in the Gray Herbarium and is clearly the same as the rather common southern Spiranthes recently recognized as S. gracilis var. floridana (Wherry) Correll. Fernald has annotated the Drummond plants and sev- [ 19 ] eral other similar southern specimens in the Gray Her- barium, clearly referable to S. gracilis var. floridana, as typical S. tuberosa Raf. and these may be taken to indi- ‘ate the strongly marked geographic variant to which he referred in ‘‘Rhodora’’*. From a study of these speci- mens and of Professor Fernald’s remarks, it is apparent that he has misidentified some of the southern represent- atives of JS. gracilis and has treated them as conspecific with the slender forms of S. Gray: which range south- ward to Georgia, Alabama and Mississippi. Actually S. gracilis var. floridana is quite distinct from S. Gray. It differs in the relatively narrower blade of the lip which is usually constricted near the apex and is conspicuously 5-nerved on the claw. The lip ranges between 8 mm. and 5mm. in length. (The lip of the Drummond speci- men, for example, is 8.4-4 mm. long.) The lip of S&. Gray? in relation to its length is broader than the lip of S. gracilis var. floridana. In outline it is commonly broadly ovate, obtuse or rounded-quadrate, the average length being about 8 mm. The blade of the lip is white and rarely more than 8-nerved beyond the middle. It gives the impression of being 8-nerved on the claw and paucinervate above. The lip of S. gracilis var. floridana is yellow on the central portion. In the extreme South S. Grayi is a very rare species and some of the herbari- um records that have been relied upon for reporting Florida stations should be restudied to ascertain whether or not they are actually based on variants of S. Grayi or on variants of WS. gracilis var. floridana. Furthermore, the likelihood of hybridity should not be overlooked where S. Grayi and S. gracilis mingle. In his reduction of S. Gray: to varietal rank, Fernald, as stated above, was influenced by the belief that S. ® Rhodora 48 (1946) 192, [ 20 | tuberosa, as he interpreted it, consists of two strongly marked geographic variations. He stated that essentially all the material from New England belonged to one category, readily recognizable by the relatively close spikes of closely spiralling, often crowded flowers, and a thick, usually solitary vertical “‘tuber.’’ To the other category he assigned all the material in the Gray Her- barium from eastern Texas to Florida and north to South Carolina in which the spikes are strongly secund without or with few spiral twists in the rachis and with the rela- tively few flowers distant and not overlapping. Fernald stated further: ‘‘From North Carolina to New Jersey both variations, with some transitions, occur, the plant often reaching a height of 5.25 dm., while its roots are usually more slender and not infrequently 2 or even 3. This is true S. tuberosa Raf....’’ It is difficult to see why S. tuberosa, described by Rafinesque as ‘‘rad. tu- berosa monorchis’’ should be applied to the southern variant with not infrequently 2 or even 8 roots, rather than to the northern one with its characteristic thick solitary turbinate root. In his recent articles on Spiranthes, Professor Fernald has placed stress on the structure or aspect of the spikes and on the presence or absence of foliage at flowering time, seeming to regard vegetative characters as relia- ble guides to specific differences in the genus. Whether the spikes are congested or slender; whether they are composed of many spirals or of few; whether the flowers are quaquaversal or secund is a manifestation of the na- ture of seasonal growth. The behavior of the flowers in the spikes in their response to torsion does not give rise to a fixed specific or inherited pattern. As should be ex- pected, rachidial torsion varies in intensity from plant to plant, and in the same species varying degrees of spiral- ity in the arrangement of the flowers may be regarded [ 21 ] as the consequence of individual variation or as the effects of climate or environment. Under certain climatic con- ditions and in response to favorable edaphic influences the flower scape may elongate rapidly. ‘Torsion in the rachis then becomes so regulated that secund flowers with closely imbricating bracts are the rule rather than the exception (cf. Plate III, fig. 3). If floral develop- ment is hastened by climatic conditions, the period of anthesis coincides with the maturing of the foliage. In Florida, where the spring approaches with a rush and vegetation becomes suddenly active, S. gracilis var. flor- idana has its commonly secund flowers fully open before the leaves have had time to attain maturity. This hap- pens also in the northern area of distribution for the spe- cies. It would be a very simple matter to pick out dried specimens of JS. gracilis from Nova Scotia and dried speci- mens of var. floridana from Florida which would be so alike in floral and vegetative aspects that even expert botanists, without helpful labels to assist them, would be unable to tell these variants apart. It would seem that in his treatment of Spiranthes tu- berosa Raf. Professor Fernald has added measurably to the ‘‘muddle”’ he attributed to Lindley, inasmuch as S. tuberosa Raf., sensu Fernald (like S. Beckit Lindl.) is ‘*concocted from elements of several different species.” In view of the difficulty experienced in properly apply- ing the name S. tuberosa Rat., without a type specimen to guide us, it would seem preferable to abandon it, and to use 8. Grayi Ames and S. gracilis (Bigel.) Beck var. floridana(W herry) Correll for the two elements involved. [ 22 ] SPIRANTHES BECKI Lindl. (1840) This species was composed of two distinct units, one from Massachusetts, one from Louisiana (both units referable to S. gracilis (Big.) Beck). | aLviIg (Louisjana, Drummond 332) (Massachusetts) . We NEOTTIA GRACILIS Bigelow SPIRANTHES FLORIDANA (Wherry) Cory (1936) (1824) S. TUBEROSA Raf. sensu Fernald (1946) Fernald reduced S. Beckii and S. Grayi to S. tuberosa Raf. He as- sumed that S. Beckit and S. Grayi are synonymous. Then he made S. Grayi a variety of S. tuberosa, and adopted as the type of S. tuberosa, in part, S. gracilis (var. flori- dana). S. tuberosa Raf. sensu Fernald is thus concocted from several different species. ZX SPIRANTHES GRACILIS var. FLORIDANA (Wherry) Correll (1940) SPIRANTHES GRACILIS (Big.) Beck (1833) SPIRANTHES SIMPLEX Gray (1867) SPIRANTHES TUBEROSA Raf. Type unknown. (1833) SPIRANTHES GRAYI Ames (S. SIMPLEX Gray not Grisebach) (1904) EXPLANATION OF THE ILLUSTRATION Prare IL. Sprranrurs Grayi Ames. 1, lip from a Massachusetts specimen, 3 mm. long, 2.5 mm. wide. 2, lip, from a Georgia specimen, 3.5 mm. long, 2.5 mm. wide. SPIRANTHES GRACILIS var. FLoRIDANA (Wherry) Cor- rell. 3, lip from Drummond 332 (from a camera- lucida sketch by Victor S. Summerhayes), 3.4 mm. long, 1.3 mm. wide. 5, lip froma Floridian specimen (Volusia County), 4.5 mm. long, 1.5 mm. wide. Sprrantues Graciiis (Bigelow) Beck. 4, lip from a Canadian specimen, 5 mm. long, 2.5 mm. wide. 6, lip from a Massachusetts specimen, 5 mm. long, 2.5 mm. wide. Figs. 1,2, 4,5 & 6 drawn by Biancur Ames, with the aid of the camera-lucida. [ 24 ] Dpinonthe Graye ee ee Var Bs f, ork Lama Canada. Dpinanthes gnoctha Buy.) Beck Masrachusett ——r Il. SprirRANTHES GRACILIS var. FLORIDANA Spiranthes gracilis (Bigel.) Beck var. floridana (Wherry) Correll in Bot. Mus. Leafl. Harv. Univ. 8 (1940) 76. Spiranthes Becht Lindley Gen. & Sp. Orch. Pl. (1840) 472, ex parte. Ihidium floridanum Wherry in Journ. Wash. Acad. 21 (1931) 49. Spiranthes floridana (Wherry) Cory in Rhodora 38 (1986) 405. Spiranthes floridana (Wherry) Correll in Journ. Elisha Mitchell Sci. Soe. 58 (1987) 159. Spiranthes tuberosa Rafinesque, sensu Fernald in Rho- dora 48 (1946) 10, 189. When Edgar T. Wherry described Ibidium florida- num, he cited Spiranthes gracilis (Bigel.) Beck as a close ally, admitting that a search for morphological differ- ences between these species had not been wholly succes- ful. The only differences of consequence between Ibidiwm Hloridanum and its northern ally are, as Wherry stated, chiefly in the color of the flowers, the dates of anthesis and the geographical distribution. It should be noted that Wherry used the length of the lateral sepals in relation to ‘‘the bend of the lip’’ as a separative character, but this does not appear to be a dependable guide to a con- stant difference between the northern and southern rep- resentatives of the Spiranthes gracilis alliance. Wherry gave the geographical distribution of [bidiuwm floridanum as reaching westward to Texas and northward to South Carolina, with a blooming season extending from Decem- ber to May. He reported specimens from South Carolina, Georgia, Florida, Alabama, Mississippi, Louisiana and Texas. For Spiranthes gracilis (Bigel.) Beck he gave [ 27 ] Aiken County, South Carolina, Alabama and Arkansas as the southernmost boundary, with a blooming season extending from July to September. When growing wild, Spiranthes gracilis var. floridana may be recognized by the yellowish hue of the rachis and floral bracts and in having the central part of the lip yel- low rather than green as in S. gracilis. The lip is variable in outline and length, ranging from 8-5 mm. long and varying from pandurate to narrowly oblong, the apical portion being conspicuously denticulate and variously fluted on the margin. The variety exhibits greater variation than the species, ranging from 1-4 dm. or more in height, with the flow- ers 83-5 mm. long. The flowers in their arrangement vary from quaquaversal to secund. As is true of the typical species in the extreme North, flowers and leaves are usu- ally present simultaneously. In their venation and tex- ture the leaves of specimens from Florida, Canada and New England are very much alike and arouse the sus- picion that the differences in foliar texture between vari- ants of S. gracilis emphasized by Professor Fernald in ‘*Rhodora’’ (48 (1946) 8, t. 9938, fig. 6; 994, fig. 5) are the exception rather than the rule and are surely not of taxonomic significance. Wherry gave the date of the earliest record for the discovery of the variety as 1834, depending on specimens collected by Burrows at Tampa Bay, Florida. In estab- lishing this record he overlooked Drummond’s speci- mens (no. 332) (a component of Spiranthes Becki Lindl.) from New Orleans, collected in 1832. Until 1905, S. gracilis var. floridana remained one of the rarest of our Spiranthes. In that year, collecting for me in Dade and Lee Counties, Florida, A. A. Eaton found an abundance of specimens, blooming in February and March. After prolonged study, during which two [ 28 ] ‘‘herbarium names’*’ came into being, one honoring Eaton, I concluded that the Floridian Spiranthes was puzzlingly close to S. gracilis and at most simply a geo- graphical variety. It may well be that morphological differences sustain- ing Wherry’s conclusions may be discovered, but as yet I have not found them. However, it is indeed clear, that the Floridian variant is much more puzzling taxonomi- cally than the Canadian one which Fernald has separated out because the flowers are usually secund and the leaves are present during anthesis. [ 29 ] EXPLANATION OF THE ILLUSTRATION Pirate II]. Sprranrues GRACILIS var, FLORIDANA (Wherry) Correll, 1, plant, natural size, from Vo- lusia County, Florida, February 4, 1947. 2, spike, natural size, from Volusia County, Florida, March, 1944. 3, part of spike enlarged to illustrate rach- idial torsion and its effect on the arrangement of secund flowers. 4, flower, side view (lateral sepals removed), about eight times natural size. 5, a lat- eral sepal, about six times natural size. 6, dorsal sepal, about six times natural size. 7, a petal about six times natural size. 8, lip, partly spread out, about six times natural size. 9, lip, flattened to show contour, about six times natural size. Drawn from fresh specimens with the aid of the camera- lucida, by BLaNncue Ames. [ 30 } PuaTeE III ——— BOTANICAL MUSEUM LEAFLETS HARVARD UNIVERSITY Vout. 13, No. 3 CHICHA, A NATIVE SOUTH AMERICAN BEER BY Hueu C. Cutter! anp Martin CARDENAS? SIMPLE FERMENTED beverages enlivened the ceremo- nies of most Indian groups in the Americas long before the Conquest. These mildly alcoholic brews were so common in some regions that they may be regarded as furnishing substantial contributions to the diet. Even today this is true of Central Mexico, where pulque, made from the sap of the maguey plant, is the national drink; and of much of Bolivia and Peru, where enormous quan- tities of maize chicha are consumed. When discussing chicha it is necessary to specify the type, for the word ‘‘chicha’’ was spread by the Spaniards so that it is now used to designate both alcoholic and non-alcoholic beverages made from a wide variety of plants and prepared in diverse fashions. Many of these are listed by La Barre (1938). In a relatively simple community of the civilized Takana Indians near Rurren- abaque in the Bolivian lowlands, intoxicating chicha is prepared from mandioca (Manihot esculenta Crantz, 1 Formerly Research Associate of the Botanical Museum of Harvard University. This work is part of that done while a Fellow of the Gug- genheim Foundation, 1946-47. ? Professor of Plant Pathology and Genetics, Universidad Autono- Ct ve . * 6 ma Simon Bolivar,’’ Cochabamba, Bolivia. [ 33 | called e mt he® in 'Takana), from ripe plantains (va sa puri), from fruits of the wine palm (07, a species of Mau- ritia), trom flour corn (di he dure), and from flint corn (di he tada). There are also non-alcoholic chichas made from toasted corn flour or from ground toasted peanuts, the latter probably having been introduced into Rurren- abaque from the southeast. In the highlands and valleys of Bolivia there are two non-alcoholic beverages called chicha. Chicha de quinoa, or alga, is usually merely water in which quinoa (Chen- opodium Quinoa Willd.) has been boiled. Sometimes sugar and cinnamon are added. For chicha de mani, or fectt in the Aymara language, peanuts are toasted and ground, then toasted again to remove the oil. Whenever peanut chicha is to be made some peanut paste and sugar are stirred into quinoa chicha and peanut oil is poured over the surface. Sweetened drinks made of toasted and ground maize or barley are usually called tostado, al- though they are occasionally called chicha. A simple alcoholic chicha can be made by mixing a substance containing starch or sugar with water and al- lowing the liquid to ferment. However, very little chicha is prepared in this way. Most chicha is made by methods which increase the alcoholic content and improve the flavor, the increase in alcoholic content being brought about by converting some of the starches to sugars which ure more readily available for fermentation. An enzyme, diastase, will bring about this change, and in South America the most common source of diastase for chicha- making is saliva. The custom of masticating roots, fruits and grains in the preparation of beverages is widespread. Explorers have found chicha made in this fashion among * Native words are spelled in the fashion of the literate population of the area in which they occur. Pronunciation and accent follow the usual rules for Spanish, [ 84 ] the most primitive as well as among some of the more advanced tribes. The raw materials chewed for fermented beverages have been reported to be mandioca in lowland Ecuador (Flornoy, 1945); mandioca and maize in central Brazil (Roquette-Pinto, 1938); maize and sweet potatoes in coastal Brazil (Métraux and Nimuendajti, 1946); al- garroba (Prosopis spp.), tusca (Acacia aroma Gill.) and chanar fruits (Gourleia spinosa (Mol.) Skeels) in the Chaco of Bolivia, Paraguay and Argentina near the Pil- comayo River (Métraux, 1946); and maize in the high- lands of Peru (Zarate, 18538). Malting (soaking grain in water and then allowing it to germinate) is another method of introducing diastase. Garcilaso de la Vega‘ (1800), José de Acosta (1940), Bernabé Cobo (1890-1893) and Hipélito Ruiz (1981) briefly mention malting. The early writers did not give detailed information on native foods and were generally scornful of indigenous customs. They left no complete account of chicha-making. It is probable, however, that malting is a pre-Columbian development. Malting is common in most of Bolivia and Peru, es- pecially in the highlands, but it is not common in other regions. It is probable that maize and quinoa were first domesticated in Bolivia and Peru, and mandioca in east- central Brazil. The roots of mandioca could not be malted, but they are very well suited to mastication. Where seed crops had been cultivated for a long time there would have been ample opportunity to develop malting and still continue the chewing methods which had been found to be adapted to the pods of algarroba and similar uncultivated raw materials. ‘The dates given are those of the editions consulted, and are not the dates when written, nor necessarily the dates of the respective first editions. The references to Acosta and Cobo were suggested by Dr. John Rowe. | 85 | EXPLANATION OF THE ILLUSTRATION Piare IV. A. Salivated corn flour or muko. (Left) from culli maize; (right) from huileaparu. B. In- terior of a chicha-making establishment in Punata, Bolivia. The steaming pan over the fire is the perol, The large containers are wirkis, for the temporary storage of chicha materials while being processed. C, Salivating maize flour or mukeando. Punata, Bo- livia. D. Cantaros in which chicha is placed near the end of the process. The chicha is sold directly from these Jars. Cochabamba. [ 36 | VEY: PLATE The authors have studied chicha-making in Bolivia at Tiquipaya, Quillacollo, Punata, Sacaba and Cochabamba among the Quecha-white population. We were able to secure little information of value from the civilized Chi- quitano Indians of Santiago de Chiquitos, but from other lowland groups, particularly the Reyesanos near Reyes and the 'Takanas between Reyes and 'Tumapasa, we se- cured quite complete data. The literature on the subject was of very little help except to confirm our observation that chicha-making is an old art, widespread in South America. CHICHA-MAKING IN THE COCHABAMBA VALLEY Because the chicha-making methods in use in the Cochabamba Valley, located at an altitude of more than eight thousand feet in the Bolivian highlands, are the most complex, and because they are undergoing changes, they will be considered in greater detail than the methods employed by the lowland Indians. The valleys of Bolivia are separated by high mountain ranges, and inter-valley commerce was long restricted by both physiographic and ethnic barriers. The consequent existence of many small centers of cultivated plant pop- ulations which had only an occasional interchange of plant material constitutes an ideal situation for the rapid evolution of numerous forms of plants. In the Cocha- bamba Valley, for example, there are many varieties of maize, some with highly specialized uses (Cutler, 1946). Raw Materials In Cochabamba, chicha is prepared from maize. The berries of molle (Schinus Molle L.) and the chewed pods of algarroba are no longer utilized. At slightly higher elevations an alcoholic chicha is sometimes made from malted quinoa grains. [ 89 ] The most highly preferred maize for chicha is chuspillo, a many-rowed sweet corn. Most of this sweet corn is grown on the larger haciendas for use only as toasted grains, or for the preparation of an especially strong chicha. The next choice for chicha is cull, a cherry-red to al- most black maize which contains large amounts of water- soluble anthocyanins. Usually the purple cobs are ground and added to the corn flour. Chicha made from cu//i is a rich Burgundy color. Occasionally the fruits of one of the various Opuntias called arampu (Opuntia sulphurea G. Don in the Cochabamba district (Pl. 1Vc), O. Soeh- rensiu Britt. & Rose near La Paz, and an undescribed species from near Arani) are added to chicha to give it this desired color. The young inflorescences of a purple amaranth are also said to be used to dye chicha. Uchukilla, a maize with small ears closely resembling culli except that it is white, is the third choice for chicha- making in the Cochabamba Valley. Very similar small orange flint ears grown at altitudes of about 2,700 meters are used for chicha in 'Totora, in the Department of Cochabamba. The chicha made from this maize is one of the best. Another preferred source of chicha is pa/tal hualtacu, also somewhat like wehukilla but with larger ears, yellow flint endosperm and an occasional slight tinge of blue in the aleurone. Notwithstanding the fact that chicha made from any of the five types of maize already referred to is of superior quality, the most common source of the beverage in the Cochabamba Valley is huilcaparu the maize most fre- quently grown. This has a rust-red cob with fourteen to eighteen rows of dented grains. The denting varies from small dimples to as much as is found in the most extreme of North American dents. The endosperm is yellow and [ 40 ] the aleurone is a deeper blue than that of pa/tal hualtacu. This combination makes the grains appear brownish, with a silvery sheen produced by minute air spaces under the pericarp. Forbes (1870, p. 249) says, ‘‘In some parts a fermented drink is made by the Indians from the sweet stalk of the young green Indian corn, called “‘huiru’’ (wiru): this is the name of the stalk.*” Although sweet young stalks, usually of cu/lt or uchukilla, are sold today in the markets to be chewed like sugar cane, they are not used in chicha- making. Preliminary Steps The maize grains are usually ground by hand, often with a half-moon-shaped stone rocker (maran una) on a flat stone (maran) as has been done for centuries. The flour is then mixed with saliva. On some of the larger haciendas it is still the custom to have women and child- ren gather in groups to do this (PI. [Vc). The flour is moistened very slightly with water, rolled into a ball of convenient size and popped into the mouth. It is thor- oughly worked with the tongue until well mixed with saliva, after which it is pressed against the roof of the mouth to form a single mass, then shoved forward with the tongue and removed with the fingers. The teeth play very little part in the process. For this reason we prefer the term ‘‘salivation,’’ and use the word ‘‘mastication’’ only when referring to cases where it is necessary for the teeth to macerate the raw materials, as is done when al- garroba, mandioca or sweet potatoes are used. The salivated morsels are dried in the sun and sacked for storage and shipment. They roughly resemble sets of false upper teeth (PI. [Va). Since muko, as the sali- rated flour is called, commands a much higher price than unsalivated flour, property owners try to convert as much flour as possible into muko. [ 41 ] EXPLANATION OF THE ILLUSTRATION PLrare V. A. Grinding corn flour, Punata, Bolivia. B. Chicha sign near Cochabamba. C. Chicha sign in ‘Tiquipaya. [ 42 | PLatTrE V a In some places in the Cochabamba Valley maize is malted. In Tiquipaya this is done by soaking the grains overnight in a pottery jar with enough water to cover them. The next day the grains are spread to a depth of about four inches on some leaves, and covered with sev- eral blankets. ‘he germinating seedlings generate con- siderable heat. The temperature in one such bed was 34°C. When the shoots are nearly as long as the grains they are placed in the sun to dry. In Tiquipaya, and also in Cuzco, Peru, it is a common sight to see these sprouted grains spread out on bright colored blankets in the dusty streets. When dry, the malted grains are ground. Both the germinating grain and the flour pre- pared from it are called huimapu. Brewing The First Day A wide-mouthed earthen pot (wz) is filled about one third full with dried and ground malted grains (huif- apu) or salivated flour (muko). When muko is used, un- salivated maize flour is often mixed with it. Crude sugar (chancaca), or the pulp of a squash (Cucurbita ficifolia Bouché) known as layacote, is occasionally added. The pot, which is about thirty inches high and thirty-four inches in diameter, is filled with water heated to about 75°C. Boiling water is not used as it is said to produce an undesirable pasty consistency. After being well mixed for nearly an hour the mixture is allowed to settle and cool further. Three layers can then be distinguished: a liquid (wp?) above; an almost jelly-like layer in the mid- dle; and coarser particles (hanchi) below. The cloudy liquid wp? is scooped up with a gourd (either Crescentia Cuyjete L.., known in the Quecha language as tutwma; or Lagenaria siceraria (Mol.) Standl., known as mate) and placed in another large-mouthed pot and allowed to [ 45 ] stand. The middle layer is transferred to a large shallow pan (perol) permanently installed on a clay support over a fire (PI. [Vx). During the last few hours of the day this second layer simmers and concentrates. ‘To the han- chi may be added more muko or some roughly broken corn (thi) and more hot water. Rarely quinoa is added instead of the t/t. This mixture is agitated again, al- lowed to settle, and the liquid portion is then added to the original wpe. The Second Day The wp continues to stand. The fire under the perol is rebuilt and the mixture is allowed to simmer for sev- eral hours longer until it becomes caramel-like in color and flavor. This product, called misqui kketa, is trans- ferred to a wirhi and allowed to cool. Most of the by- products of chicha-making are no longer important since their place has been taken by sweets which are more easily made with sugar. A small amount of the misqui hketa may be eaten inthe home where the chicha is being prepared, but nearly all of it is used in the chicha itself. It is probable that the msqui kketa was originally used solely as a sweet and was not added to the chicha. Chicha- makers, like alchemists of old, surround their art with an aura of mystery, and probably emphasize some of their steps in such a way as to confound inquisitive con- sumers. The hanchi left in the bottom of the original open- mouthed werAz is drained of its remaining moisture in the following manner: A cylinder woven of splints is pressed vertically into the layer of hanchi. The particles inside the cylinder are removed and packed about the peri- phery. The last remnant of liquid filters through the woven cylinder. This is a sweet, clear, non-alcoholic liquid called chua conehu which may be drunk in the [ 46 ] house where the chicha is made, but, more commonly, is added to the upz. The further disposition of the hanchi will be discussed later. The Third Day The upz, which is now slightly bitter, is transferred to the perol, the large shallow pan over a fire. A thick white sediment remains in the w/v. This sediment, known as upt alpi, is further treated by adding sugar, cinnamon and some orange leaves and then boiling. It is then eaten as a delicacy in the household where the chicha is being produced. After the wpi has boiled for about three hours in the perol it is transferred to a wirki to cool. In Cochabamba, which is 2,572 meters above sea level, boiling occurs at 89°C. At the end of this third day the misqui kketa is added to the uwpz, an operation locally called Aketacha. The Fourth Day The mixture of wp and misqui kketa has now begun to ferment and bubbles violently. This action soon slows down and the liquid is transferred to narrower-mouthed pots (centaros) at the place where the chicha will be sold or consumed (PI. [Vp). The Fifth to the Tenth Day In the Cochabamba Valley, very few bubbles arise and fermentation is considered to be complete after about six days. At higher and cooler altitudes more days are needed, while in the hot lowlands less than two days may suffice. Before the chicha is drunk, the floating froth (aca fiahut) is removed with the cupped hand. This froth contains oil, much yeast and some extraneous material. It is used as a sort of French polish for furniture. Occasionally [ 47 ] EXPLANATION OF THE ILLUSTRATION Pirate VI. Chicha signs. A. Sacaba. B. Arani. C. Calla-Calla. D. Cochabamba. E. Taquina. F. Cochabamba. [ 48 | PuatrE VI chicha-makers may add some of this yeasty mixture to new batches of green chicha as a starter. Usually, how- ever, this is unnecessary because the porous earthen jars are only slightly washed out and so are normally well supplied with yeast. The jars are washed with ashes and water, swished about with a branch of molle (Schinus Molle L.) or huai cha (Senecio clivicolus Wedd. ). After the chicha has been drunk, a layer of sediment remains in the bottom of the pot. This sediment, the borra de chicha, is laid on a gunny-sack stretched over a small jar. A liquid, the sutu, which resembles chicha in color, but lacks the carbonated sparkle and cidery tang of good chicha, drips through the sacking. Sutw is greatly prized for it is reputed to have a higher alcoholic content than chicha. It is usually consumed at home, sweetened with sugar and dyed a wine-color with airampu (see p. 40). For each forty liters of chicha only about one liter of sutu is obtained. After the sutw is drained off, the re- maining borra is ted to pigs and chickens, or is saved to be used in making a weak chicha to be discussed below. It may also be added to bread dough, for it provides yeast as well as filler. At the end of the second day of chicha-making a coarse sediment, the hanchi, remains. This is often given to pigs and chickens, but it may also be used to make kaima, a cheap and weak chicha. Cold water is mixed with the hanchi and it is allowed to stand. After a short time a sediment settles to the bottom. The upper part of this sediment is smooth and creamy. This kketa ordinaria, as it is called, may be cooked with brown sugar to make a custard, or it may be added to the misqui kketa when it is being concentrated. If Aketa ordinaria is added to the musqui kketa, more borra is left after the chicha has been drunk. The coarser, lower part of the sediment is fed to pigs. The supernatant liquid (se/e) is removed and boiled [ 51 | in the perol for two hours, then transferred to a tall pot (cantaro) and allowed to stand for two or three days. Borra, usually from a previous lot of chicha, is then added. The beverage thus produced is called haima. It is consumed mainly by the Indian field laborers whose work in the hot sun makes them particularly susceptible to alcohol. They prefer Aaima to the stronger chicha be- ‘ause it is possible to drink more without becoming drunk or sleepy. The sediment of the Aaima, also called hanchi, is fed to pigs. Consumption and Marketing Well-made chicha is an attractive drink, clear and sparkling, with a flavor somewhat similar to that of apple cider. Its alcoholic content varies greatly, from about two per cent in new or in weak and watered chicha to as high as twelve per cent in a few exceptional samples. Most Cochabamba chicha has five per cent alcohol or less. When chicha is ready for sale, a sign, the aca llantu (‘‘Chicha flag’’) is displayed. Lé/antu is probably related to Haitu, the word fora fringe which denotes nobility or rulership. The frequent occurrence of fringes and tassels in chicha signs suggests that there may be some basis to the local belief that the better types of chicha were for- merly reserved for the ruling class. ‘The simplest chicha signs are white flags attached to tall poles. These are seen only in the country districts. During fiestas the flags are decorated with flowers and colored ribbons (Pl. Vir). Some of the signs in small settlements are very ornate. Fancy tin stars, wooden condors, reed boats, cloth squares, tissue-paper frills and wooden plaques are common. Some of these are shown in Plates V and VI. In Quillacollo and Cochabamba there are laws which prohibit these fancy devices, and [ 52 ] most signs are simple lettered boards. These usually an- nounce ‘‘Chicha,*’ ‘‘Buena Chicha’? (Good Chicha), “*Chicha Punatefia’’ (Chicha from Punata) or ‘*Chicha Clizena (Chicha from Clissa). Many chicherias also have a parrot in the doorway so that the illiterate may know where to drink. One chicheria places a single parrot in the doorway when ordinary chicha is for sale, but two parrots when the chicha is exceptionally good. Chicha is taxed in Bolivia. Each year the right to col- lect these taxes is auctioned to the highest bidder. ‘The tax is 48 Bolivian centavos (about one cent in United States money) for each bottle of 660 ¢c.c. A bottle of chicha is usually sold for about eight United States cents. The tax is collected when the chicha is made. In Coch- abamba and its suburbs, with a population of about 80,000, according to the records of the Treasury of Cochabamba, taxes were paid in 1946 by 768 licensed chicherias on 4,617,388 liters of chicha. It is rumored that more than twice this amount was actually made. CHICHA-MAKING IN THE BoLIvIAN LOWLANDS Chicha-making in the Bolivian lowlands is a short and simple process. Since fermentation progresses rapidly in the tropical heat, and since the natives have only a few small pots, there is neither equipment nor time for the elaborate methods which prevail in the highland valleys. Among the Takana Indians who live on both sides of the Rio Beni near Rurrenabaque, chicha forms a part of the daily food of most families and nearly all homes con- stantly have some in preparation. Métraux (1942, p. 42) states: “‘None of the Takana tribes is known to have brewed any fermented drink, ....a curious exception in an area where most tribes enjoyed several kinds of beer.”” We believe that fermented drinks have been known in this region for a long time, and were either [ 58 | EXPLANATION OF THE ILLUSTRATION Piare VII. A. Grinding flour corn for chicha near Balas, Rio Beni, Bolivia. B. Germinating corn grains for chicha, near Reyes, Bolivia. C. Airampu (Opuntia sulphurea). The seeds are used to dye chicha, near Cochabamba. D. Pounding grains of white flint corn for somé or tuhuré, Rurrenabaque, Bolivia. Puiatre VII not seen or not recorded by Métraux’s sources of infor- mation.° Most chicha is made from the long ears of dull orange flour corn. White flint maize is seldom used. The ears are kept in the husks, tied in pairs and hung over a pole on the rafters or under the trees until needed. A pointed wooden peg held in the hand is sometimes used to facili- tate the removal of the husks and grains. The grains are soaked overnight in a jar of water, then placed in a basket lined with Heliconia leaves and cov- ered with more leaves (Pl. VII). Inside one basket the temperature was found to be 86°C. When the shoots are nearly as long as the seed (the sprouted grains are called e bi data in Takana), they are crushed with a rocker stone (tuwmu) on a semicircular hardwood board (tada) (Pl. VILa). Usually some of this ground malted grain is salivated slightly by mixing it in the mouth. The malted, ground and salivated meal (e be ten?) is immedi- ately placed in a jar and mixed with warm water. The mixture is then boiled for three or four hours. The boil- ing liquid is called e bi data du duki, the liquid after boiling but still hot s¢za métahi, and the liquid when cold but not fermented hui na métahi. After it has cooled the liquid is strained (da kwatat) through a piece of cotton cloth and left in a narrow-mouthed jar to ferment (patcha chenapui, i.e. to become strong). Although the chicha is at its best after two days, drinking usually begins soon after the liquid cools and before it has fermented, and continues until it is gone. After five days the chicha be- comes vinegar-like, but it has usually been consumed ° The little that is known of the Takana language is summarized by Créqui-Montfort and Rivet (1921-22). Since they list only one word for chicha (tupari), we are including other Takana terms recorded near Rurrenabaque. These should be accepted with caution since they were recorded by one who is not a linguist and does not speak Takana. [ 57 ] betore this. he sediment which was left after straining and that which remains in the pot after the chicha is drunk are thrown to the pigs and chickens. There are no by-products. Just as in the highlands no yeast is added. The liquid is boiled in one pot, however, and fermented in another which is merely rinsed out after each lot of chicha has been made. White flint corn is used to make two drinks in the Rio Beni town of Rurrenabaque. The grains are pounded in 2 wooden mortar (taczv) with a pestle (mano tacu in the Takana dialect of Rurrenabaque) to loosen the pericarp (Pl. VIIb). After winnowing, the grains are boiled in water for about three hours. Sugar is then added. When cool, this product is called somd, probably an introduced term. If allowed to ferment, it is called tuhuré. Reference has been made to the sweet caramel-like paste (misqui kketa) produced in the highland chicha pro- cess. The Takana Indians prepare a similar sweet as the end-product in a process which is exactly like that for chicha up to the time of boiling. The malting, grinding and salivation have been completed and the water added. The mixture is then strained and boiled. However, in- stead of boiling for only a few hours, this process con- tinues for almost a day until the liquid is sticky and caramel-like, and the Takana women judge it has reached its ‘‘point.”” This means that when cool it will harden just as properly made fudge does. While still hot the sticky liquid is poured or scraped onto the broad leaves of any one of several species of Heliconia. Usually it is eaten as a sweet (et? hua; in the Takana dialect spoken near Reyes, e cua huara). Since sugar cane is now com- mon in the area, very few of the Takanas still make the sweet solely as a confection. It is usually utilized as an ingredient for achicha prepared from the di palm, a spe- cies of Mauritia. [ 58 | STEPS IN THE MAKING OF CHICHA AND ITS BY-PRODUCTS IN THE COCHABAMBA VALLEY OF BOLIVIA Ground Salivated Oried Either “muko” “huaghapu” used. or Sometimes algaroba, sugor or white squash added. $4 Germinated Soaked Ground - = | Ferment for Furniture “a “aca fahui os glee Boll 3 hours Add water about 70°C. Stir well and tet settle for one Boil until caramel-like -~ _ Za ( Insert pee ; > = Na cylinder rom ‘ Se s' 2 ’ Brea oa) Add “tiki” or oF) Lan 8 more “muko”, rile | Add cold ; e, o_o Sek e” | Ferment Boil 2 hours for 2-3 days PRELIMINARIES SECOND DAY THIRD DAY FIRST DAY FOURTH TO TENTH DAYS Drawn by Gordon W. Dillon The fruits of this palm are boiled in water until the shell comes off. The mass is then worked with the hands and the corn sweet (e ti hua) added. This mixture is again boiled and broken up with the hands, after which it is strained through a sieve. After standing in pots for two days it is ready to be drunk. Natives claim that this chicha is stronger than that made from maize alone. Chicha made from the palm fruits alone without adding the corn sweet is weaker and is seldom prepared. On the western margin of the Bolivian Chaco near Camiri and Lagunillas, the Chiriguano Indians make chicha by simply mixing boiled and masticated grains of maize with hot water and allowing this to ferment. In Santa Cruz, plain ground corn is sometimes mixed with hot water and left to ferment. This makes a weak chicha which must be consumed as soon as it has fermented. According to Métraux (1946), the Indians near the Pil- comayo River in the Chaco of Bolivia and Paraguay use the fruits of algarroba (Prosopis spp.), chanar (Gourliea spinosa (Mol.) Skeels), and tusca (Acacia aroma Gill.) in the preparation of alcoholic drinks. These fruits, par- ticularly algarroba, are usually masticated as a prelim- inary step in the process. These observations on the manufacture of chicha in Bolivia are presented as introductory material to a pos- sible future study of native American beverages. A com- plete report on the manufacture and use of such beverages together with their associated ceremonies and supersti- tions would yield valuable information as to the spread of food plants and customs. Before such a study can be made, however, it will be necessary for the student to go into the field and collect his own data, since the infor- mation available in the literature, and even in the note- books of ethnologists, will not suffice for more than the roughest of outlines. [ 59 | LITERATURE CITED Acosta, José de, 1940. Historia natural y moral de las Indias. Mexico City. Cobo, Bernabé, 1890-93. Historia del Nuevo Mundo. Seville. Créqui-Montfort, G. de, and Paul Rivet, 1921-22. La famille linguis- tique Takana. Journ. Soc. Amér. Paris, n.s., 13: 91-102, 281-311, 14: 141-182. Cutler, H. C., 1946. Races of maize in South America. Bot. Mus. Leafl. Harv. Univ. 12: 257-291. Flornoy, Bertrand, 1945. Voyages en Haut Amazone. Rio de Janeiro. Forbes, David, 1870. On the Aymara Indians of Bolivia & Peru. Journ. Ethnol. Soc. London, n.s., 2: 1938-3805. Garcilaso de la Vega, 1800. Historia general del Pert 6 comentarios reales de los Incas. Madrid. La Barre, Weston, 1938. Native American Beers. Amer. Anthrop. 40: 224-234, Métraux, Alfred, 1942. The native tribes of Eastern Bolivia and West- ern Matto Grosso. Bur. Amer. Ethnol. Bull. 134: 1-182. Métraux, Alfred, 1946. Ethnography of the Chaco, in Handbook of South American Indians, Bur. Amer. Ethnol. Bull. 143, vol. 1: 197-370. Métraux, Alfred, and Curt Nimuendaji, 1946. The Camacan linguis- tic family, in Handbook of South American Indians, Bur. Amer. Ethnol. Bull. 148, vol. 1: 547-552. Roquette-Pinto, E., 1938. Rondonia, ed. 4, Sao Paulo. Ruiz, Hipdlito, 1931. Relacién del viaje hecho al Reyno del Pert y Chile, Madrid. Zarate, Augustin de, 1853. Historia del descubrimento y conquista de la Provincia del Pert. Madrid. [ 60 | BOTANICAL MUSEUM LEAFLETS HARVARD UNIVERSITY CampripGe, MAssacuuseTts, JULY 27, 1948 VoL. 13, No. 4 VASCULAR ANATOMY OF ORCHID FLOWERS BY B. G. L. Swamy Biological Laboratories, Harvard University Tuar the structural complexities of a flower can be made clear by astudy of floral anatomy in relation to its vascularization is becoming a prevalent thought. While such a study is being concentrated on the members of several other natural orders, orchids, characterized by their ‘‘over-specialized’’ floral organization, still remain to be attacked from this angle. Robert Brown (1831) was the first to elucidate the nature of the orchid flower. He compared it with the structural and other morphological features of the flowers of certain allied monocotyledonous families and inter- preted the labellum as a compound structure made up of the lateral stamens of the outer whorl and the median petal. Darwin, after studying the course of vascular traces in a considerable number of orchid flowers belong- ing to different tribes, incorporated the results in his monograph, ‘“The Various Contrivances by which Or- chids are Fertilised by Insects’’ (ed. 1899). In general, he confirmed Brown’s observations, particularly with re- gard to the compound nature of the gynostemium and labellum. Subsequent to this publication, very little has been [ 61 ] published on the vascular anatomy of orchid flowers. Even this meagre information is confined to the study of only teratological material. The present study is an attempt to understand the structural complexities of normal orchid flowers. Material and Methods Vascular anatomy of the flowers of the following gen- era have been studied; the number of species investigated in this study is indicated by a digit in parenthesis after each genus: Diandrae Cypripedilinae Paphiopedilum (2) Cypripedium (1) Monandrae Ophrydinae Habenaria § Platanthera (1) Habenaria § Euhabenaria (6) Habenaria § Peristylus (2) Satyrium (1) Neottiinae Vanilla (1) Zeuxine (1) Coelogyninae Coelogyne (2) Pholidota (1) Liparidinae Oberonia (2) Phajinae Spathoglottis (1) Cyrtopodiinae Geodorum (1) Eulophia (2) Dendrobiinae Dendrobium (8) Eria (1) Bolbophyllinae Bulbophyllum (1) Cy mbidiinae Cymbidium (1) Sarcanthinae Diplocentrum (2) Luisia (1) Saccolabium (3) Vanda (1) Aerides (2) Rhynchostylis (1) Material was fixed in formalin-acetic-alecohol, and serial microtome sections of the entire flower buds were pre- pared following the customary methods. Basic fuchsin or crystal violet, with a counter stain of light green, was employed for staining the vasculature and ground tissue respectively. This combination gave satisfactory contrast. In addition to sectioning, entire flower buds of some species were cleared first in chloral hydrate and subse- quently in lactic acid. This method was especially suit- able for small flowers (like Oberonia, Diplocentrum, ete. ) and those with delicate floral parts (like Zeuavine, Haben- aria, etc.). But for larger flower buds (Cymbidium, Hu- lophia, Dendrobium, ete.), a certain amount of trimming of the perianth was essential. After observing both serial sections and cleared ma- terial, wire-plasticine models were constructed, and the three dimensional drawings reproduced here have been sketched from such models. The vascular diagrams that accompany the text have been drawn as seen from above; however, the exact places of origin of certain traces have been slightly exaggerated and spread out in one plane in order to bring out the details. Structure of the Flower The orchid flower is bilaterally symmetrical. The ovary is inferior and bears at its upper extremity the other floral organs. There are two whorls in the perianth, each whorl consisting of three members. Often, all the members of the outer whorl are similar in appearance, though some- times the median member (often designated as the ‘*hood’’ or the ‘‘dorsal sepal’’?) may be somewhat more conspicuous than the lateral members either in pattern or size. In the inner whorl of the perianth, the two lat- eral members are usually small and of the same shape as those of the outer perianth members, whereas the median [ 63 ] one is almost always very dissimilar both in size and shape, and is known as the ‘‘lip’’ or ‘‘labellum.’’ Next are two whorls of stamens, which present definite vari- ations of expression in the two great subdivisions of the family, Diandrae and Monandrae. In the former, the two lateral stamens belonging to the inner whorl are function- al, and the median one of the outer whorl is represented by a conspicuous staminode, which assumes various shapes. No indications of other staminal members can be made out externally. In the Monandrae, on the con- trary, the median stamen belonging to the outer whorl is the functional one and no indications of the other stami- nal members are to be seen externally. In both groups the next and innermost whorl consists of three stigmas ; in the Diandrae, all three are usually functional, whereas, in the Monandrae, only the two lateral are functional, the median one being modified as the rostellum. Further- more, it may be noted that the stamens and stigmas in the Diandrae are almost ‘‘free,’” whereas in the Monan- drae they are united into a central pillar-like structure, which has come to be known by the names ‘‘column”’ or ‘‘oynostemium., ” In the present text and the accompanying figures | am using the following terminology and symbols for the various parts of the orchid flower (reference to Fig. 1 may be made): Outer whorl of perianth Dorsal sepal DS Lateral sepals LS Inner whorl of perianth Median petal (labellum) MP Lateral petals LP Outer whorl of stamens Median stamen Al Lateral stamens A2 and A$ Inner whorl of stamens Lateral stamens al and a2 Median stamen a3 Whorl of stigmas Median stigma Gl Lateral stigmas G2 and G3 [ 64 | Dorsal Sepal (DS) Anther (A}) Sligma (confluent, G2 63) Gynoslegium Lateral Petal (LP) Lateral Sepal (Ls) A B Fig. 1. A. A monandrous orchid flower showing its various parts. B. Ground plan of the various vascular traces of a generalized orchid flower to indicate the symbols used in the text. The orchid ovary, as will be seen presently, is trav- ersed by six main vascular traces,* each lying under a perianth member; I am applying to these traces the ab- breviations corresponding to the perianth members. Main Vascular Traces of the Ovary In the inflorescence axis of the orchids the vascular bundles are arranged in the same scattered pattern as in the vegetative axis of monocotyledons. The number of bundles that enter the floral axis and their method of breaking up into the six main traces of the ovary vary considerably in different genera. In Paphiopedilum and Cypripedium of the Diandrae (Fig. 2 A), six vascular bundles from the inflorescence axis deviate into the flower and constitute the corresponding main traces (DS, two LS, two LP and MP) of the ovary. The bract is sup- plied by a seventh independent bundle (Br) from the inflorescence axis. * The term trace is used throughout this text to designate the vas- cular elements in the flower from the moment the vascular bundles of the inflorescence axis enter the ovary. [ 65 ] Among the monandrous orchids, usually three bundles from the inflorescence axis deviate into the floral axis and while doing so break up into the six main traces of the ovary. The exact method of their breaking up varies among the different genera. Further, the different levels at which they break up is also variable to some extent within the same species. In Habenaria (Fig. 2 B), two of the three bundles that deviate from the inflorescence Fig. 2. A to E, Diagrams illustrating methods of the differentiation of the six main traces of the ovary from the vascular bundles of the inflorescence axis. axis directly constitute those main traces of the ovary underlying the lateral petals (LP); the third bundle not only gives rise to the midrib trace of the bract (Br), but also, by further breaking up, constitutes the remaining four main traces of the ovary, underlying the dorsal sepal (DS), the two lateral sepals (LS) and the median petal (MP). In Vanilla, some species of Dendrobium, Rhyn- chostylis and a few others (Fig. 2 C), each of the three vascular bundles of the inflorescence axis that enters the flowers gives rise to two main traces of the ovary, the traces occupying positions under the adjacent perianth members (LLP, DS; LP, LS; MP, LS); the median [ 66 ] trace of the bract (Br) is also given off by one of the three vascular bundles that deviates from the inflorescence axis. In Bulbophyllum, Pholidota, some other species of Dendrobium and a few others (Fig. 2D), three vascular bundles from the inflorescence axis constitute the main traces of the ovary. One of the three breaks up into the median trace of the bract (Br) and the main trace of the ovary underlying the dorsal sepal (DS). The second, on bifurcation, forms the respective main traces underlying the lateral petal (LP) and lateral sepal (LS) of one side, whereas the third bundle gives rise to the remaining three main traces lying under the lateral petal (LP), the lateral sepal (LS) of the other side, and also to the me- dian petal (MP). In Cymbidium, Eulophia, Spathoglottis and some members of the tribe Sarcanthinae (Fig. 2 E), though three vascular bundles from the inflorescence axis deviate into the floral axis, one of them constitutes only the midrib trace of the bract (Br). The second one gives rise to the main trace underlying the dorsal sepal (DS), one lateral sepal (LS) and median petal (MP). The third breaks up into the main traces underlying the lateral petals (LP) of both sides and the remaining lateral sepal. Thus in this instance only two bundles actually take part in the construction of the six main traces of the ovary. The situation described above may be represented in the following tabular form (the vascular supply of the bract is omitted): Tribe Number of bundles deviating Method of further breaking or JSrom the inflorescence axis up of the bundles of the genera into the floral axis inflorescence axis Cy pripedilinae 6 No further differentiation ; (Fig. 2A) but directly constitute the six main traces of the ovary,—DS, LS, LS, LP, ie. Mer: [ 67 ] Ophrydinae 3 1 (LP)+4 (DS, LS, MP, (Fig. 2B) LS)+1 (LP) Vanilla, Dendrobium, 3 2 (LP, DS)+2 (LP, LS) Rhynchostylis, Aerides, ete. +2 (MP, LS) (Fig. 2C) Oberonia, Bulbophyllum, 3 3 (LP, LS, MP)+1 (DS) Coelogyne, Pholidota, +e (LP, LS) ete. (Fig. 2 D) Cymbidium, Eulophia, 4 3 (LP, LP, LS)+3 (MP, Spathoglottis, some LS, DS) members of Sarcanthinae, etc, (Fig. 2 E) From the preceding account it will be clear that, from the standpoint of the origin of traces, there is no basis for distinguishing the main traces that belong to the outer whorl from those that belong to the inner whorl of the perianth. Actually, the bundles of the inflores- cence axis that enter the flower split up in a variety of methods and constitute the six main traces of the ovary. Neither does there seem to be anything in the position of the traces themselves to distinguish the two whorls, because in transverse sections the six main bundles are arranged more or less on one and the same circumference (Fig. 8 A). There is also no difference between the traces of the two whorls with respect to size or degree of ex- pression. Even those main traces that adjoin the placen- tae (LP, Fig. 8 B), which in other groups usually show an exaggerated development either in size or prolifera- tion, in this case look exactly similar to the traces lying on alternating radii (DS, Fig. 3B). In some orchids with large-sized labellum (like Cym- bidium, Hulophia, Spathoglottis, ete.) the main trace underlying this modified petal (MP) shows a double na- ture (Fig. 2). It may be argued by some that this feature may be due to the size-relation of the organ which it is supplying. The weight of this argument becomes [ 68 | invalidated when we consider the fact that the corres- ponding trace in orchids like Cypripedium and Paphioped- lum, which possess labella of much larger size and display than those of Cymbidium or Hulophia, is always single. ee Be i yz ~ B A 5 BDS — (¢ Fig. 3. A. Transverse section of an ovary showing the disposition of the six main vascular traces running in its wall, B. A portion of a transverse section of an ovary enlarged; traces DS and one of its adjacent LP are shown. Another point which may be noted here is that the traces running along the placentae and underlying the inner whorl of perianth members do not show any signs of breaking up and supplying any part of the placentae. Hence the ovules are totally devoid of any kind of vas- cular supply. Types of Vasculature in the Flower The six main traces of the ovary begin to break up further into the various traces of the floral whorls and organs just below the level of insertion of the perianth. Though the general scheme of splitting is essentially the same in all the orchids investigated, three distinct types [ 69 | may be recognized on the basis of the origin and nature of the vascular supply to the staminal and_ stigmatic whorls, rather than on the method of vascular supply to the perianth. Tyre L. (Seen in the members of Cypripedilinae; Fig. 4 A, B and C). The traces to the median and lateral stigmas (G1, G2, G8) are the first to be separated. As can be expected, they arise from the main traces, DS and the two LS, underlying the median and lateral sepals respectively. At a slightly higher level, the same main traces trifurcate, to constitute the median and marginal traces of the respective perianth member. Further, the marginal traces of these sepal members supply the mar- ginal traces to the petals also (Fig. 4 A). The main trace, DS, gives rise to another small trace (A1) at a slightly higher level ; this trace runs vertically into the staminode. The main traces occupying positions under the lateral perianth lobes of the inner whorl, after giving out the traces to the functional stamens (al and a2), merely pass into the respective petal as its median trace. The mar- ginal traces of these, as has been stated, are derived from branches of marginal traces of the sepal members. ‘The main trace, MP, does not split up but continues into the labellum as its median trace. The following points may be especially noted in this connection: 1. Masters (1887) observed in some species of Cypri- pedium that the trace underlying the labellum also tri- furcated to form the median and marginal traces of that petal. However, the flowers here examined do not show this feature. The marginals of the labellum are always given out by the marginal veins from its adjacent sepal members. [ 70 | re fe fe OF ie ‘. 7 \ TNS G Fig. 4. A, Band C. Paphiopedilum. D, E and F. Habenaria. G, H and I. A monandrous orchid A, D and G are vascular diagrams; B, E and H are drawings from wire-plasticine models; C, F and I are transverse sections at the region homologous with the gynostemium, 2. The vascular traces supplying the staminode, func- tional stamens and stigmas, all run only for a very short distance in a common tissue; hence a definite structure like the gynostemium is almost absent or highly reduced in these plants. However reduced this structure may be, the following six vascular traces may always be recog- nized in a transverse section (Fig. 2C): a. Al occupying a dorsal median position, b. al and a2 occupying a lateral position, and ce. G1, G2 and G8, all three together occupying a place just below Al. 3. Link (1849) thought that the fertile lateral anthers of Cypripedilinae originated by the division of the median stamen, the staminode being looked upon as an enor- mously developed connective. Masters’ observations on teratological material have already cast a doubt on the tenability of this view. The present studies also suggest that Link’s interpretation is erroneous. Tyrer II. (Seen in the members of Ophrydinae; Fig. 4, D, Kand F). The main trace, DS, gives rise to the traces that supply the median stigma (G1), the median stamen, Al, and the median trace of the corresponding sepal. The marginal traces for this sepal are supplied by the main traces underlying the lateral petals (LP). The main trace (MP) does not give rise to any other trace excepting the median one of the labellum and its spur, which is merely a downward pouch-like extension of the labellum; the marginal traces are supplied by those of the adjacent perianth lobes of the outer whorl (LS). The main traces underlying the lateral sepals, in addition to forming the median and marginal traces of these lobes, give rise to the traces, G2 and G8, which supply the lateral stigmas. It may also be noted that the marginal traces of the [ 72 ] lateral petals are altogether absent in Habenaria § Pla- tanthera; the marginal traces towards the dorsal sepal alone are present in Habenaria § Peristylus; and in some species of Muhabenaria, their presence is occasional and often they are very poorly developed. A transverse section passing through the gynostemium at the level of the stigmas (Fig. 4 F') shows the follow- ing four vascular traces: a. Al, representing the supply to the functional stamen (median) of the outer whorl, and b. G1, G2 and G3, supplying the median and lateral stigmas respectively. Generally the members of Ophrydinae possess a spur. Morphologically this structure is nothing but a down- ward outgrowth of the labellum in the form of a long tube or sac. Satyrium nepalense has a double spur; it is also dorsal in relation to the inflorescence and flower, as there is no resupination. Where the spur is single, the median trace of the labellum runs throughout its entire length and finally continues upwards into the labellum (Fig. 5 A). The marginal traces of the labellum (that are derived from the adjacent sepal traces), however, do not enter the spur but curve up into the expanded portion of the labellum (Fig. 5 A). On the other hand, where the spur is double, the median trace of the labellum does not enter the spur but runs straight into the median petal as its median trace; the marginal traces (that are given off by the adjacent sepal traces) run throughout the length of the spur on that side and then continue into the median petal as its marginal traces (Fig. 5 B). Type ILI. (Seen in the majority of monandrous or- chids. Fig. 4 G, H and I). The first traces to be sepa- rated off at the level of insertion of the perianth are those to the three stigmas (G1, G2 and G8), given off by the [73 Me LS Hf if uM ‘ LY A B Fig. 5. Drawings from wire-plasticine models to illustrate the nature of the vascular supply to the labellum and spur. A. When the spur is single; B. When the spur is double. main traces underlying the sepals. "Then, in some genera, each of the main traces underlying the lateral petals (LP) gives off a trace (al and a2) into the gynostemium, these representing the lateral stamens of the inner whorl. In some other genera, the main traces of the lateral sepals (LS) each give out a trace (A2 and A3) into the gyno- stemium; these represent the traces belonging to the lat- eral stamens of the outer whorl. The commoner condition is the presence of the traces representing the lateral sta- mens of the inner whorl. In some genera (Dendrobium, [ 74 ] Cymbidium, ete.), both sets of lateral stamens (A2, A3 and al, a2) may be present. It must be stated here that the expression of the staminal traces is highly variable in different genera and is often associated with highly com- plicated types of adnation. (This phenomenon is treated in detail in a later part of this text under ‘‘Comparative Resumé of the Vascular Supply of the Stamens.’’ See page 77). After supplying the vasculature to the stamens and stigmas, the residue of the main traces (excepting the one under the dorsal sepal) breaks up further to constitute the vascular traces of the perianth lobes. (Variations of the vasculature of the perianth members of this type are explained ina later part of this text under ‘‘Comparative Resumé of the Vascular Supply of the Perianth.*’ See page 76). The following points may be noted in connection with this type: 1. ‘The gynostemium (Fig. 4 I) contains the following traces: a. Al supplying the median stamen of the outer whorl, which is the functional one, b. G1, G2 and G8, reaching the rostellum and the confluent lateral functional stigmas, and c, either al and a2 representing the traces of the lateral stamens of the inner whorl, or A2 and A8% representing the lateral stamens of the outer whorl, or both. 2. The traces of the staminal and stigmatic whorls are all embedded in a well-developed pillar-like structure (a gynostemium). Types I and II are consistent; no marked variations were seen. But Type III is subjected to a high degree of plasticity. This is closely associated with increasing cohesion, adnation and other phenomena which are de- scribed in the following paragraphs. [ 75 | Comparative Résumé of the Vascular Supply of the Perianth In minute flowers like those of Oberonia and in the flowers of the semi-saprophyte, Zeuwine, the vasculature is very poorly developed. As a result, the system is not attended by any secondary modifications (Fig. 6, B). In flowers like those of Vanilla, Coelogyne, ete., the mar- ginal traces of the adjacent perianth members lie close together (Fig. 6 C).. In Dendrobium, Bulbophyllum, A B G D E Fig. 6. A to E. Series of diagrams to illustrate the increasing com- plexity of the vascular supply of the perianth members, Only one- half of the flowers is shown; traces to the stamens and _ stigmas are not shown for the sake of avoiding confusion, Pholidota, Spathoglottis, etc., the marginal traces fuse to form an ‘‘arch’’ and from this point again separate out (Fig. 6 D). In Cymbidium and many members of the tribe Sarcanthinae, the marginal traces after forming an ‘arch’? proceed in a fused condition for a considerable distance and then separate (Fig. 6 E). Often some of the marginal traces even enter the respective perianth member in the fused condition (see the marginal trace of the dorsal sepal in Fig. 6 E). It may also be noted that the main trace, DS, does not in general give rise to any of the marginal traces of the dorsal sepal. But when it does split up into the marginal traces, as in Cymbidium and some members of the Sar- [ 76 | canthinae, it invariably anastomoses with those of the adjacent lateral petal. Another feature to be noted is the origin of the mar- ginal traces of the adjacent perianth lobes. Usually the marginal traces of the outer perianth members are de- rived from the traces underlying the perianth members of the inner whorl or vice versa. Comparative Résumé of the Vascular Supply of the Stamens A lack of proper understanding of the vasculature of the staminal whorls, in the past, seems to have been the cause of the confusion prevailing today with regard to the exact composition of the androecium of orchids. In this connection two points need clarification, —(A) the number of staminal traces that are given off and their identity, and (B) the method by which a simple origin becomes increasingly complicated through the interven- tion of adnation. It is difficult to separate the two as- pects which are so closely associated in orchids; but for the sake of convenience of description, I propose to treat the two points under the headings (A) and (B), as men- tioned above. A. It has already been pointed out that in the Cypri- pedilinae the two lateral stamens of the inner whorl! (al and a2) are functional and that the median stamen of the outer whorl (A1) is externally represented by a stami- node, which also receives a vascular trace. Furthermore, it has also been mentioned that in the monandrous orchids it is the median stamen of the outer whorl (A1) that is functional and that quite often the lateral stamens of the inner whorl (al and a2) are represented only by vascular traces. ‘The presence or absence of the traces cannot be made out by any external morphological distinctions. eae It may be recalled that in some species the traces for the lateral stamens of the outer whorl (A2 and A3) may also be represented and that in many instances all of the four lateral stamens may be represented by their corres- ponding vascular traces in varying degrees of expression: that is, the traces themselves may be so rudimentary that they do not run even to the base of the gynostemium, or they run up only a short distance into the structure, or they may even reach the distal end of it. The degree of expression may affect one, two, three or all of the four lateral staminal traces. The latter condition may be fre- quently seen in Zeuaxine, Oberonia, Eria, ete. Clear-cut examples where either the lateral stamens of the inner whorl alone or only those of the outer whorl manifest themselves are seen in Mulophia. In EH. nuda (Fig. 7 A and B) the main traces (LS) underlying the lateral perianth members of the outer whorl, after giving off the traces G2 and G3 to the respective stigmas, sep- arate out correspondingly the traces A2 and A8’. That AiG A> G3 Fig. 7. Drawings of vascular skeletons made from wire-plasticine models illustrating the splitting up of the six main vascular traces of the ovary into the vascular supply of various whorls. A and B. Eulophia nuda; C and D. Eulophia epidendraea. In Band D the staminal and stigmatic traces are shown as rearranged according to their status; vascular traces of the perianth members are omitted in these. [ 78 ] these traces represent only the lateral stamens of the outer whorl and that these cannot be considered to be otherwise is proved by the fact that they originate after the stigmatic traces have been separated off and before the trifurcation which results in the formation of the me- dian and marginal traces of the respective perianth mem- ber. In EH. epidendraea (Fig. 7 C and D), the main traces, LP, underlying the lateral petals, before splitting up to form the supply to the corresponding perianth members, give out the traces al and a2 which obviously represent the traces for the lateral stamens of the inner whorl. That these traces cannot be interpreted otherwise is borne out by the point of their separation from the main trace, LP. B. In other orchids, the degree of expression of the traces for the stamens of the two whorls is complicated by the associated phenomenon of adnation. The series of diagrammatic illustrations in Fig. 8 is intended to convey the increasing complexity in the origin of the staminal traces. In these figures the behavior of one pair of the main traces (LP and LS) underlying the lateral perianth members of one side of the flower is shown. A illustrates the simplest condition where there is no fusion ; the staminal traces a2 (belonging to the inner whorl) and A8 (belonging to the outer whorl) are quite distinct both in origin and further course. In B, the two traces are still distinct; but their separation from the respective main traces is postponed until after the trifurcation of the main traces themselves; thus the position of the origin of the staminal traces is carried up to a higher level in relation to the floral axis. In C, the position of origin of the two staminal traces is carried up still higher; they originate at a point nearer to the anastomosing point of the marginal traces. Furthermore, the two staminal [ 79 | traces come to lie closer soon after separation and seem to fuse with one another and then proceed upwards in the gynostemium. I) illustrates the climax of this tendency. The origin itself of both the staminal members is carried higher to a point where the adjacent marginal traces fuse and thus their distinctness becomes totally obliterated. Further, the marginal traces of the respective perianth au a , a. +A3 az+As m™ A B Cc O Ne ° ° Py eé Fig. 8. A to D. Diagrams to illustrate the evolution of © compound”’ staminal traces, a2-+A3. member are also separated from the same point. Thus the nature of the staminal trace in this instance becomes “compound” by the adnation of a lateral stamen of the inner whorl with that of a lateral stamen of the outer whorl, and as a consequence such traces have been here interpreted and represented as al + A 2 or a2+ A38, as the case may be. The degree of expression of adnation is often seen to vary in one and the same species. Geodorum densiflorum may be taken as a good example to exhibit this phenome- non; furthermore, this orchid also throws some light on the fundamental nature of the androecium of the family. Normally in this species, in addition to A1, the lateral stamens of the two whorls are also represented by ‘‘com- pound’’ traces, al+A2 and a2+A8 (Fig. 9 A and B). [ 80 | As has been stated more than once, the vascular pat- tern is not always strictly fixed, but is somewhat plastic in this species. In all of the three instances of deviations to be described below, the origin and nature of the trace A1 is very consistent, as in the other monandrous orchids. It is the nature of the other vascular traces of the stam- inal whorls that present interesting variations. In the first deviation (Fig. 9 C and D) the staminal vascular traces arise from the point of trifurcation of the main traces, LS. Thus the traces A2 and A8 complete the outer whorl of stamens. In the second deviation (Fig. 9 EK and F), the main trace under one of the lateral sepals give rise to the trace A2 (representing a lateral stamen of the outer whorl), whereas on the opposite side the main trace lying under the lateral petal gives the trace a2 (repre- senting a lateral stamen of the inner whorl). In the third deviation (Fig. 9 G and H, left), there arises a com- pound staminal trace, al+A2, whereas on the opposite side the main trace under the lateral sepal gives rise to A8, representing a lateral stamen of the outer whorl). In addition to these two, the main trace, MP (lying under the labellum), also sends out a slender offshoot, a8, into the gynostemium; this trace is to be interpreted only as representing the median stamen of the inner whorl. ‘Thus, these deviations, when taken together, reveal the pres- ence of all of the six stamens belonging to the two whorls. This also makes it possible to visualize the fact that all of the six stamens were once functional in the ancestral orchid flower. In passing, it may also be noted by a study of these illustrations (Fig. 9) that the plasticity of vasculature is not confined to the staminal whorls alone but is extended to the labellum also, whereby the latter is supplied by supernumerary marginal traces in certain instances. Before proceeding to the general discussion, it 1s im- [ 81 | Fig. 9. A, C, E and G. Vascular diagrams of the normal and three deviation patterns respectively of Geodorum densiflorum. Traces belonging to the stamens and stigmas are shown in solid black. B, D, F and H. Staminal and stigmatic traces of A, C, E and G respectively, arranged according to their status. portant to have an idea of the changes that are brought about in the vascular pattern of the flower during the development of the flower itself. This type of study was confined only to Dendrobium and Cymbidium, but I feel that the results obtained may in general be applicable to other orchids also. Here again, particular attention may be laid upon the development of ‘‘compound’’ staminal traces. In the young condition (when the microsporangium is just differentiated) the marginal traces of the adjacent lateral petal and sepal show the following features (Fig. 10 A): The staminal traces al, a2, A2 and AB are all dis- tinct from one another, although their place of origin is shifted higher up to the point of concrescence of the ad- jacent marginal traces; but actual concrescence is not yet affected. During subsequent development, concres- cence and adnation proceed to such a degree that the distinct origin of the staminal traces becomes totally obliterated so that they are seen only as ‘‘compound’’ Al air Ar a2+A3 is Fig. 10. Vascular diagrams of Dendrobium showing the juvenile con- dition in A and the adult condition in B. [ 83 | traces (Fig. 10 B); thus the traces for adjacent lateral stamens of the one side are represented by the compound trace, al +A2, and on the other side by the similar trace, a2+ A8. General Discussion Ovary. In recent years the nature of superior and in- ferior ovaries has raised a good deal of controversy. The evidence at hand points out that the inferior ovary in at least the majority of instances is a product of adnation of floral whorls, though there seems to be some evidence as to its receptacular nature in a few genera or groups of plants. The orchid ovary, as has been said before, is trav- ersed by six main vascular traces that run throughout the length of the ovary proper without any kind of splitting, and the further division begins only at about the inser- tion level of the perianth and other floral organs. This condition may be looked upon as a result of adnation in its extreme degree of expression. The irregularity seen in the method of origin of the six main traces of the ovary in different genera of orchids may also be correlated in some way with the high degree of adnation. Perianru. The whorls of perianth members exter- nally suggest considerable distinctness. In general, those of the outer whorl are uniform in appearance, whereas those of the inner whorl are variable. Particularly does the labellum (median petal) assume a series of varied pat- terns in size, shape and color. Whatever may be the outward manifestation, internally each of them (sepals as well as petals) receives a median and two marginal traces. Quite often the marginal traces of one whorl of the perianth members are derived from the traces belong- ing to the perianth members of the other whorl. Ana- tomically, all of the perianth members are thus of the [ 84 | same category. The 3-trace supply is more a characteris- tic feature of sepals than of stamens. From these consid- erations, the perianth of orchids seems to be more sepa- line in character. Ontogeny of the flower, in some cases, has been sought to elucidate the nature of the petals—whether they are akin to sepals or stamens. Thus Goebel (1933) correlates the belated appearance of petals during the ontogeny of the flower with their nearness to stamens. In the orchids the ontogenetic order of appearance of floral whorls is centripetal (Swamy, 1946; also unpublished data). And even when Goebel’s correlation is applied to the condi- tion in orchids, it is seen that the petals are not allied to stamens but to sepals. LaBELLUM. This structure of the orchid flower has been the target for much dispute. Brown (1831) put for- ward the doctrine that the labellum is acompound struc- ture resulting from the fusion of some of the staminal members with the lip. However, he did not offer sufh- cient evidence for his opinion and also was not certain as to which of the stamens enter into fusion. He further suggested that such a fusion was especially responsible for the extra-conspicuousness of the labellum with ridges and crests in some orchids. Lindley (1840) followed this view. Darwin (1899), after studying the course of vascu- lar traces in some orchids, came to the same conclusion: ‘*The Orchid flower consists of five simple parts, namely, three sepals and two petals; and of two compound parts, namely, the column and labellum. The column is formed of three pistils and generally of four stamens, all conflu- ent. The labellum is formed of one petal with two pet- aloid stamens of the outer whorl, likewise completely confluent.’’ As Darwin thus elaborated and confirmed Brown's ideas, the authorship of the compound theory [ 85 ] of the labellum is attributed both to Brown and Darwin (see Fig. 11, which gives a diagrammatized interpretation of a monandrous orchid flower according to Darwin’s conception). ar aa Av az Fig. 11. Vascular diagram of a monandrous orchid flower reconstructed according to the description of Darwin, The Brownian and Darwinian conception of the orchid labellum held sway for a long time, influencing subse- quent studies to a great extent. Many reports on tera- tological orchid flowers were interpreted in the light of this hypothesis and were brought forward as evidences in favor of it. Even as recently as 1987, Saunders, while attempting to advocate her theory of ‘‘carpel polymor- phism,’’ fully accepted the old hypothesis, ignoring the opposition brought forward in refutation of the Brownian and Darwinian views of the orchid labellum. Criiger’s views (1865) which are contrary to the com- pound concept of the labellum seem to have been ignored until 1916, when Worsdell brought out his two volumes of ‘Principles of Plant Teratology.’ In this publica- tion, Worsdell has accumulated all of the points adduced in favor of the compound nature of the labellum and has [ 86 | given his own alternative interpretations of the very ex- amples which were previously interpreted as evidencing the old hypothesis. He also summarizes other instances ‘‘which decidedly seem to tend towards disapproving the accepted theory and to support that advanced by Criiger.’’ In addition to these, he presents some of his own original observations on some abnormal flowers of Cypripedium. All of these combine to support the opinion that there is not even a single instance to support the view of the compound nature of the labellum. As can be seen, until now the Brownian and Darwin- ian concept of the orchid labellum has largely been checked only with teratological flowers. Excepting the study of Masters (1887), which deals in part with the vascular anatomy of normal flowers, I am not aware of other subsequent contributions dealing with the anatom- ical studies of normal orchid flowers. The present anatomical study of 40 species of orchid flowers belonging to 24 genera also does not show any evidence in support of the compound theory of the la- bellum. The traces representing the lateral stamens of the outer whorl, whenever present, are distinctly identi- fiable inthe gynostemium. Their independent origin and course is especially clear in forms like Hu/ophia, where no kind of secondary fusion nor complication accompanies their expression. Even in those instances where adnation results in a compound staminal supply as al+ A2 and a2+A38, the traces without an exception pass into the gynostemium but not into the labellum. Furthermore, the labellum is almost always provided with separate mar- ginal traces, given out by the veins belonging to the ad- jacent perianth members. As has already been stated in an earlier part of this text, the labellum receives typically a median and two marginal traces and hence is in no way different anatomically from other perianth members. [ 87 ] From a consideration of these facts, the only reasonable conclusion that can be arrived at is that the labellum is not a compound structure as was supposed by Brown and Darwin. It may also be noted here that those in- stances in which the marginal traces of the labellum were not in evidence or exhibited abnormal behavior when present offered difficulties of interpretation to Darwin himself. For he writes: ‘‘This anomaly is so far of im- portance, as it throws some doubt on the view that the labellum is always an organ compounded of one petal and two petaloid stamens. ”’ Spur. This structure, as has been explained already, is a basal outgrowth from the labellum in the form of a pouch, sac or narrow tube. Where the spur is single, the median trace of the labellum continues into it and then recurves upwards in conformity with the curvature of the spur to constitute the median trace of the same petal. When the spur is double, the median trace of the labellum does not enter the spur but each marginal trace enters into the spur of the corresponding side. In pass- ing, it may be mentioned that Darwin thought the traces that traverse the double spurs were staminal (A2 and A8). That Darwin’s views are untenable has already been made clear. GYNOSTEMIUM. This is a structure peculiar to the or- chid flower. Its exact nature has, until now, been little understood. Chiefly the works of Oliver (1895), Rendle (1930) and Willis (1936) have propagated the idea that this structure is an extension of the floral axis and that on this account it isto be looked upon as being axial. It may be recalled that such a distinct structure is not clearly evidenced in Cypripedium, Habenaria, ete. In the former, it is incipient if it is present at all; serial [ 88 | microtome sections alone can reveal its incipient condi- tion anatomically. In all of the other orchid flowers stud- ied at this time the structure is well-developed; and in all such species, it is an appendicular structure containing the vascular traces supplying the staminal and stigmatic whorls. The morphological apex of the flower, therefore, does not extend to the apex of the gynostemium but ceases at the point of the insertion level of the perianth. Thus, in one and the same ‘‘floral tube’’ (as designated by Wilson and Just, 1939) of orchids, different degrees of adnation are seen, the ovary proper containing the fused vascular traces of all of the floral whorls and the gynostemium proper in which the individual traces of the stamens and stigmas are embedded. STAMENS. Robert Brown (1831) was the first to state in clear terms that the stamens in the orchid flowers are arranged in two alternating whorls, each whorl contain- ing three stamens. Since then Darwin’s investigations (1899) on several other orchids have brought to light a certain amount of evidence to support Brown’s views. But as Darwin also believed that the lateral stamens of the outer whorl were fused with the labellum, he con- templated that only the lateral stamens belonging to the inner whorl were always present in the gynostemium. It is not so easy to determine which particular trace represents which particular stamen just by observing a section of the gynostemium. ‘*‘Compound’”’ staminal traces have been demonstrated in the present study as a very common condition in orchid flowers. Some other salient points concerning the vascular supply to the sta- mens may now be recalled: (1) In the Cypripedilinae, al and a2 are normally present and are functional. AJ1 is transformed into a staminode. In teratological flowers (see Worsdell, 1916) [ 89 | sometimes AY or A8 or even a8 may be represented by the corresponding staminodal outgrowths externally or by vascular traces internally. (2) In the Ophrydinae it has been shown here that Al alone manifests itself as the functional stamen and that no other stamens are represented even by vascular traces. (3) In the remainder of the members of the Monan- drae, as arule, Al is the functional stamen, al and a2 or sometimes A2 and A838, being present as traces. The origin and course of the various staminal traces (except- ing the one pertaining to A1) may be somewhat irregular in some advanced genera. ‘They may be quite clear in the juvenile condition of the flower but later become oblit- erated, or they may be ‘‘compound’’ from the very beginning. In the majority of the genera, however, all of the four lateral stamens are represented by the ‘‘com- pound”’ traces, al+A2 and a2+A38. (4) The expression of a8, though rare, can still be seen in some instances. The above facts, when considered from an ‘‘over-all’’ point of view, account for the presence of all of the six stamens belonging to the two whorls. At one time, per- haps, all of the six stamens were functional. During phylogenetic specialization, the median one of the inner whorl (a) disappeared early, probably in correlation with the differentiation of the labellum. Subsequently, the elimination of the lateral stamens of the outer whorl! was effected first and of the inner whorl later; this is sug- gested by the more frequent predominance of the latter and the less frequent expression of the former among the orchids here studied. However, when the vascular traces of the lateral stamens of both the whorls are present, they form ‘‘compound”’ traces as al+A2 and a2+A3. [ 90 | A study of the published floral diagrams of Neuwwiedia (Pfitzer, 1889) reveals that an early step in the suppres- sion of the three stamens of the posterior half of the flower (A2, a8 and A8) had already started here; hence the functional stamens were the median one of the outer whorl (A1) and the lateral stamens of the inner whorl (al and a2). Such flowers specialized further along two distinct lines: (1) suppression of the median stamen of the outer whorl (A 1), which tendency gave rise to forms like the diandrous orchids, and (2) suppression of the lat- eral stamens of the inner whorl (al and a2) which ten- dency resulted in monandrous orchids. Such an inter- pretation is in conformity with the stand taken by Rolfe (1909-12). (These aspects are fully discussed in my paper, ‘‘Embryological studies in the Orchidaceae, Part II,’’ which is under publication). CarPEL. According to the classical view, the orchid ovary is unilocular and tricarpellary; the ovules are mar- ginal, the margins of the adjacent carpels having fused to such an extent as to obliterate the double nature of the ovule-bearing vascular traces. ‘The median trace of each carpel supplies the respective stigma. The only opposition to the above-mentioned view is that advanced by Saunders (1928, 1987). She contends that each of the main vascular traces of the ovary repre- sents a carpel; that the main traces underlying the outer whorl of perianth are to be considered as solid sterile car- pels bearing stigmas and that the main traces underlying the inner whorl of perianth members are to be considered as representing semi-solid carpels where placentae have approached the median trace of the carpel. There is absolutely no anatomical evidence to favor Saunders’ views. The inconsistencies and irrelevancies of the theory of ‘‘Carpel Polymorphism’’ have been [ 91 ] brought to light repeatedly by several workers in several plant families. So it becomes unnecessary to consider Saunders’ attack with any seriousness. However, one difficulty that may be postulated against favoring the classical view of the orchid carpel is that of the irregularity of the origin and differentiation of the median and marginal traces of each carpel. But this ob- jection, if it can really be called an objection, will be nul- lified when we take into consideration the high degree of anastomoses, cohesion, adnation and other allied phe- nomena accompanying the structural evolution of the flower. SUMMARY The orchid ovary is traversed by six main vascular traces. In the Cypripedilinae, the origin and subsequent development of these six traces are not attended by any secondary modifications. But among the monandrous genera, a series of increasingly complex and diverse pat- terns is witnessed; thus the number of vascular bundles that enter into the construction of the flower from the inflorescence axis is usually three, but this may be re- duced to two; their method of breaking up into the six main traces of the ovary also is highly diverse in different genera. None of the main traces of the ovary show any signs of supply either to the placentae or to the ovules. The main traces begin to break up and supply the other floral members only at the upper extremity of the ovary. Depending largely upon the presence or absence, num- ber and nature of the vascular traces supplying the sta- mens, three types of vasculature of orchid flowers are recognized: Type I, characterizing the Cypripedilinae, Type II, characterizing the Ophrydinae and Type III, characterizing the rest of the Monandrae. [ 92 ] It is Type III that presents an interesting series of increasing complexity. The latter is intimately associated with anastomoses, cohesion and adnation of the individu- al vascular traces of the flower. The increasing trend of complexity of perianth members is seen in the derivation of marginal traces of the perianth members of one whorl from those of the other and their anastomoses. As a re- sult of this, the places of origin of the staminal traces become highly displaced and carried to higher levels, until the adjacently placed lateral staminal traces form ‘compound’ traces, here designated as al A2 and a2 +A8. The effect of adnation on the different whorls of the flower has been described in detail. The inferior ovary is looked upon as due to an extreme adnation of the different floral whorls. The outer and inner whorls of the perianth, though externally distinct from one another, present the same anatomical features and on this account the inner whorl is considered to be more sepaline than otherwise. The labellum is also shown to receive the same vascular supply as the rest of the peri- anth members; it is stressed that there is no evidence to consider it to be a compound structure, as was thought by Brown and Darwin. The two lateral stamens of the outer whorl which were thought by Darwin to have fused with the labellum, are here shown to be represented in the gynostemium, either as individual traces (A 2 and A8) or in their ‘‘ecompound’”’ manifestations (al+A2 and a2+A3), whenever they are present. It has also been shown that the median stamen of the inner whorl is capable of expressing itself occa- sionally. It is suggested that in the ancestral orchid flower all of the six stamens were functional; that in course of time the posterior three ceased to be functional and that it seems as if such a progenitor may have given rise to the [98 | Diandrae on the one hand and the Monandrae on the other. Evidence is presented in favor of the appendicular theory of the gynostemium, in opposition to the views of those who consider it to be a prolongation of the floral AXIS. I am very thankful to Professor R. H. Wetmore for kindly reading the manuscript and offering valuable sug- gestions for improvement in presentation. [ 94 ] LITERATURE CITED Brown, R., 1831. Observations on the Organs and Mode of the Fe- cundation of Orchideae and Asclepiadeae. Trans. Linn. Soc. 16: 685-745. Criiger, H., 1865. A few Notes on the Fecundation of Orchids and their Morphology. Jour. Linn. Soe. 8: 127-235. Darwin, C., 1899. ““The Various Contrivances by which Orchids are Fertilised by Insects.’’ London. Eichler, A. W., 1875. ‘““Blithendiagramme.’’ Leipzig. Goebel, K., 1933. ‘‘Organographie der Pflanzen. Samenpflanzen,”’ 3rd ed. Jena. Lindley, J., 1830-1840. “‘The Genera and Species of Orchidaceous Plants.’’ London. Link, H. F., 1849. Bemerkungen iiber den Bau der Orchideen, be- sonders der Vandern, Bot. Zeitung. 7: 745-751. Masters, M. T., 1887. On the floral Conformation of the genus Cypri- pedium. Jour. Linn. Soc, 22: 402-422. Oliver, F. W., 1895. ‘“‘Natural History of Plants.’’ New York. Pfitzer, E., 1889. Orchidaceae in ““Die natiirlichen Pflanzenfamilien.’”’ Rendle, A. B., 1930. ‘“‘The Classification of Flowering Plants. Vol. I. Monocotyledons.’’ London. Rolfe, 1909-1912. The evolution of the Orchidaceae. Orchid Rev. (Several discontinuous pages in the issues.) Saunders, E. R., 1923. A Reversionary Character in the Stock (Matthiola incana) and its Significance in regard to the Structure and Evolution of the Gynoecium in the Rhoeodales, the Orchidaceae and other Families. Ann, Bot. 37: 451-482. ——, 1987. ‘*Floral Morphology.’’ Cambridge (England), Swamy, B.G. L., 1946. Embryology of Habenaria. Proc, Nat. Acad. Sci. India. 12: 413-426. Willis, J. C., 1936. ‘‘A Dictionary of Flowering Plants and Ferns,’’ Cambridge (England). Wilson, C.S. and Just, T., 1939. The Morphology of the Flower. Bot. Rev. 4: 97-131. Worsdell, W.C., 1916. ‘‘Principles of Plant Teratology,’’ London. [ 95 | BOTANICAL MUSEUM LEAFLETS HARVARD UNIVERSITY CampripGce, Massacuusetts, Ocroser 20, 1948 VoL..13; No.5 STUDIES IN THE GENUS HEVEA II BY RicHarp Evans ScHULTES' THE REDISCOVERY OF HEVEA RIGIDIFOLIA IN 1852, Richard Spruce made the type collection of the extraordinary Hevea rigidifolia near the confluence of the Rios Uaupés’ and Negro in the Amazonas of Brazil. Nearly a century has passed since this achieve- ment, but, apparently until recently, no other collection has been made, and the species has remained known solely from the information derived from a study of the type material. In March, 1944, during an extensive cytogeographic survey carried out in the Amazon Valley for the United States Department of Agriculture and in collaboration with the Instituto Agrondmico do Norte, Dr. John T. Baldwin, Jr. discovered Hevea rigidifolia and introduced it into cultivation at Belém do Para. In 1945, Felisberto Camargo, director of the Instituto Agrondmico do Norte, sent his botanical explorer, Sen- hor Ricardo de Lemos Frées, to the upper Rio Negro Botanist, Division of Rubber Plant Investigations, Bureau of Plant Industry, Soils and Agricultural Engineering, Agricultural Research Administration, United States Department of Agriculture; Research Fellow, Botanical Museum of Harvard University ; Collaborator, In- stituto Agronémico do Norte, Belém do Para, Brasil. >This river is known in Colombia as the Vaupés. [ 97 | and lower Rio Uaupés area, where Spruce had collected so extensively, for an investigation of Hevea and several other genera of economic plants. In July, 1946, in Belém, I had the pleasure of study- ing a number of the specimens of Hevea which Fr6es collected on this trip and which Dr. Adolpho Ducke had determined (cf. Ducke in Bol. Técn. Instit. Agron. Norte no. 10 (1946) 18). The scope and quality of the collection are such that 1 am moved to compliment the collector most highly on his accomplishments and Dr. Camargo for his part in bringing about this valuable con- tribution to our knowledge of Hevea in one of the most isolated regions of the Amazon Valley. Without a doubt the most outstanding achievement of Frées’ trip was his rediscovery of Hevea rigidifolia. Four excellent and extensive collections were made in three localities, all rather near the type locality. These collections greatly increase our understanding of this spe- cies which is certainly (with the possible exception of Hevea microphylla Ule (Schultes in Bot. Mus. Leafl. Harvard Univ. 13 (1947) 1-9) the most distinct of the whole genus. One of the collections includes several valves of the capsule and two seeds, both of which strue- tures have hitherto been unknown; and another collec- tion includes seedling material. In May, 1947, Fro6es recollected Hevea rigidifolia in sterile condition from a locality where he had found it two years previously. According to Senhor Froes, two thousand seedlings of Hevea rigidifolia from this locality (Serra de Tunuhy) have been established at Belterra. The Fro6es collections of Hevea rigidifola are impor- tant from several points of view. They indicate that the species is probably a very highly restricted endemic. It is evident also from a comparison of the collections with the type material that the species is apparently rather [ 98 ] stable, showing variations which are surprisingly slight for the genus. On the basis of a study of this new ma- terial, we can now attempt to discuss the relationships of Hevea rigidifola with other species. Furthermore, the examination of the Froes material has convinced me more firmly than ever that it is imperative to bring this species, which has a number of unusual characteristics found in no other species, into cultivation on a large scale for investigations in selection and hybridization and to study even more intensively its relationships. There have been four descriptions of Hevea rigidifolia, including the original, but they have all been based on the type collection. The original description is very short and has proved to be inadequate, in view of the compli- cations which are presenting themselves in the detailed and critical studies of the genus as to its subspecific con- cepts and phylogenetical relationships. The second’ and third’ descriptions, those of Mueller-Argoviensis, are Be* | |... foliolis coriaceis margine recurvis, alabastris masc. cylin- drico-conicis angustis, calycibus pro 2/3 longitudinis partitis, laciniis triangulari-lanceolatis acute acuminatis, disci masc. glandulis lanceo- lato-subulatis glabris, antheris 6-10 irregulariter biseriatim verticilla- tis, columna supra-staminali integra glabra, stigmatibus breviuscule stipitatis... Arbor 30-pedalis. Foliola circ. 9-12 cm. longa, 4-5 cm. lata, lanceolato-elliptica, acute acuminata, basi acuta, subtus sub- glaucescentia, supra—laevigata, glabra, multo crassiora quam in con- generibus. Costae primariae crassae, secundariae tenuiores, subtus prominentes ; venae validiusculae, subdeplanatae. Paniculae amplae, albido-v. cinereo-tomentellae. Calyces masc. aperientes 4 mm, ae- quantes v. paulo longiores, foeminei aperti usque 7 mm. longi, intus circa fundum minute subulato-glanduligeri; laciniae longe et anguste acuminatae. ”’ #** |. . foliolis coriaceis margine recurvis; alabastris masc. cylin- drico-conicis angustis; calycibus pro 2/3 v. 3/5 longitudinis partitis, laciniis e base triangulari anguste lanceolatis acute acuminatis; disci masc. glandulis lanceolato-subulatis glabris; antheris 6-10 irregulari- ter biverticillatis, columna suprastaminali angusta glabra; stylo brevi distincto, ovario glabro. **ArBor 30-pedalis. Perrott validi, foliola aequantes v. iis longiores, [ 99 | greatly amplified from the original of Spruce and have served until the present time; the third (1874) was re- produced with only slight alterations by Pax.’ In the following notes, I have thought it advisable to offer a new description, based upon the three previous ones, but extended and altered to conform with the ad- ditional information resulting from further observations and studies of the typical material and from an examina- tion of the Baldwin and Frées collections. The fact that it has been necessary to make only minor changes is at once a tribute to the accurate work of Spruce, Mueller- Argoviensis and Pax on the type collection and an indi- ‘ration of the unusual stability of this species-concept. apice valide biglandulosi. Fotro.a cire. 9-12 em. longa, 4—5 em. lata, oblongato-elliptica, acute cuspidato-acuminata, basi acuta, rigidiora quam vulgo in genere, supra laevigata et nitida, utraque pagina glabra, subtus e glaucescente palliodora et validiuscule venulosa, venae vix prominentes; costae majores utroque latere circ. 12, primariae validae et valde prominentes. PANntcuLArk amplae, floribundae, micranthae, cinereo v. fere albo-tomentellae. Catycrs masc, aperientes 4 mm. aequantes v. paulo longiores, feminei aperti usque 7 mm. longi, utri- usque sexus intus inferne praesertim vellereo-tomentelli. GLANDULAR disci florum masc. elongatae, erecto-patulae, acuminatae, basi tantum connatae et more ovarii et columnae suprastaminalis omnino glabrae. Ovarium ovoideum, stylo brevi stigmatibus paulo longiore valido termi- natum, teres et laeve, glaberrimum. Fructus ignoti.”’ > **Arbor ad 10m. alta. Petiolus validus, foliola aequans vel super- ans, apice valide biglandulosus; foliola 9-12 em. longa, 4—5 cm. lata, oblongo-elliptica, cuspidato-acuminata, basi acuta, valde coriacea, margine revoluta, supra laevia et nitida, utraque pagina glabra, subtus glaucescenti-pallida, valide venulosa; costae utroque latere ca, 12. Paniculae amplae, floribundae, cinereo- vel albo-tomentellae. Alabas- tra % cylindrico-conica, angusta. Calyx 4 4 mm. longus, 2? 7 mm. longus, utriusque sexus intus tomentellus; lobi triangulares, acumi- nati; disci glandulae % lanceolato-subulatae, glabrae, acuminatae, basi tantum connatae; antherae 9~10, irregulariter biverticillata ; columna staminalis glabra, ultra antheras longius producta; ovarium glabrum; stylus brevis. Fructus ignoti.’’ [ 100 | Hevea rigidifolia (Spruce ex Bentham) Mueller- Argoviensis in Linnaea 84 (1865) 203; in DC. Prodr. 15, pt. 2 (1866) 718; in Martius FI. Brasil. 11, pt. 2 (1874) 8300—Hemsley in Hooker’s Icon. Pl. 26 (1898) t. 2578, figs. 11, 12, 18—Pax in Engler Pflanzenr. IV, 147 (Heft 42) (1910) 124—Ducke in Arch. Instit. Biol. Veg. Rio Janeiro 2, no. 2 (1985) 235; in Bol. Téen. Instit. Agron. Norte no. 10 (1946) 18. Siphonia rigidifolia Spruce ex Bentham in Hooker's Journ. Bot. 6 (1854) 871. ORIGINAL DESCRIPTION: **S. rigidifolia, Spruce, MS. ; foliolis ellipticis crasso-coriaceis glabris, paniculis pulveraceo-tomentosis, calycibus subacuminatis, antheris 5—8 ad basin columnae irregulariter subverticillatis.— Ramuli glabrati. Foliola breviter petiolulata, 5 poll. longa, 24 poll. lata, acute acumi- nata, margine recurva, basi cuneata, multo crassiora et rigidiora quam in caeteris speciebus, subtus punctis crebis albicantia. Paniculae py- ramidatae, semipedales. Flores pallide flavi, masculi 2 lin., foeminei 3 lin. longi. Antherae S. luteae. Stylus evidentior. **A milky tree of 30 feet in height, from the caatingas of the Rio Uaupés R. Spruce.’’ EXTENDED DESCRIPTION : Arbor parva usque ad triginta ped. alta, 8.5 ad 10 poll. in diametro, latice albo. Petiolt robusti, 8-24 cm. (max- ima pro parte 13-17 cm.) longi, 1-8 mm. in diametro, teretes, cortice glabro rubicundulo-brunneo vel stram- ineo-brunneo, tenuiter striati, basi leviter subcarnoso- dilatati, apice valide biglandulosi. Petioluli robustiores, 3-13 mm. (maxima pro parte 5-7 mm.) longi. Foliola valde reclinata, statu adulto valdissime coriacea (et multo rigidiora quam illa in genere vulgo visa), perfecte oblon- gato-elliptica vel saepe elongato-elliptica, basi cuneata vel saepe valde rotundata, apice acute cuspidato-acuminata, margine conspicue recurvata, 6-19.5 cm. (maxima _ pro parte 10-18 cm.) longa, 8.5-8.5 em. (maxima pro parte 4.5-6.5 em.) lata, supra laevigata, pallide cinereo-viridia, [101 J EXPLANATION OF THE ILLUSTRATION Prare VIII. Hevea ricrmiroira (Spruce ex Benth.) Muell,-Arg. 1, terminal branch with inflorescence, one third natural size. 2and 38, staminate flowers, natural size. 4, stamen, one and one third times natural size. 5 and 7, seeds, one third natural size. 8 and 9, valves of the capsule, one third natural size. 10 and 11, ovary, two and two thirds times natural size. 12 and 13, pistillate flowers, one and one third times natural size. Drawn by A. N. Ferraz [ 102 ] Puate VIII valde nitida, omnino glabra, infra glaucescenter pallidiora crebro-tessellatis punctis albicantia et validiuscule venu- losa, costis subtus prominenter elevatis, glabris (special- iter primariis), secundariis utrinque duodecim ad quin- decim. Paniculae amplae, elongatae, folia subaequantes, aliquid rigidae, floribundae, minute cinereo- vel fere albo- tomentellae. Alabastra masculina anguste et longe cy- lindrico-conica, valde acuminata, calycis segmentis apice valde et conspicue imbricatis contortisque ; feminea mul- to majora, acuminata. Ca/yces pro genere satis crassius- culi, masculini aperientes 3.5—4.6 (plerumque 4) mm. longi, feminei 6-8 mm. longi, extus densissime albo- tomentelli, intus usualiter glabri sed raro vel apicem ver- sus vel fundum versus parce minute pulveraceo-tomen- telli, utroque sexu per 2/3 vel 3/5 longitudinis partem partiti, laciniis e basi triangulari anguste lanceolatis, apice acute acuminatis, marginibus integris involutis, pallide luteis; laciniarum florum femineorum apicibus valde re- tortis. -Antherae plerumque sex (raro septem ad decem), irregulariter biverticillatae, parvae; columna suprastam- inalis angusta, gracilis, glabra, usque ad 2 mm. ultra antheras producta. Disci glandulae florum masculinorum elongato-subulatae, erecto-patulae, glabrae, acuminatae, basi aliquid connatae; florum femineorum magnae, irreg- ulariter elongato-linguiformes, usque ad 1 vel 1.2 mm. longae, valde erectae, glabrae. Ovarium ovoideum, 1 mm. X1.5 mm., glaberrimum sed nune minute punctu- latum, brevi cum stylo stigmatibus; stigmata congesta, subglobosa, crasso-conflata, glabra. Coccorum valvae lignosae, usque ad 5 cm. longae, 1.7—2 cm. latae; epi- carpium tenue sed probabiliter vivo fibroso-striatum ; en- docarpium 2 mm. crassum, sicco valde contortum. Semina regulariter elliptico-ovoidea, valde angulosa, aureo-brun- nea cum maculis irregularibus magnisque, ferrugineo- brunneis, 2.5—2.9 em. X 1.6-1.9 em. X1.8—2.2 em., raph- { 105 | ide prominenti et faciebus satis amplis. In terris saxosis, semi-xerophyticis, in plantarum societate quae in lingua brasiliense ‘‘caatinga’’ nominatur, in locis usque ad circa mille pedes altitudine crescit. Nomen vulgare ‘‘serin- gueira’’ vel ‘‘seringueira da serra’’ est. CoLLECTIONES EXAMINATAE: Brazit: Estado do Amazonas, prope Panuré [Ipanoré] ad Rio Uaupés, Oct. 1852-Jan. 1853, Richard Spruce 2527 (Typr cotiection). Brazit: Estado do Amazonas, Jauareté, Uaupés, Rio Negro. ‘‘Ar- vore de 9 m., 25 cm. [diam.]; folhas coriaceas, um tanto pendentes. “*Seringueira.’? Oct. 22, 1945, Ricardo de Lemos Frées 21258. Brazit: Estado do Amazonas, Iraruca, Rio Igana. ‘*Arvore, 7m., 15 cm. [diam. |. Casca fina, latex branco, abundante, coagulavel. Fol- has pendentes, coriaceas.’’ Nov. 16, 1945, Ricardo de Lemos Frées 21394, Brazit: Estado do Amazonas, adjacencias da serra de Tunuhy, estando-se até a $08 metros dealtitude. *“Rica em latex branco, coagu- lavel.’’ Nov. 14, 1945, Ricardo de Lemos Frées 21401, Brazit: Estado do Amazonas, Serra Tunuhy, [Rio] Icana, Rio Negro. May 5, 1947, Ricardo de Lemos Frées 22303. Brazit: Estado do Pard, Belém do Parad. “*Seedling cultivated in Instituto Agrondédmico do Norte, collected by J.T. Baldwin, Jr., at Montepelago, Rio Uaupés, Estado do Amazonas, March 1944. Ster- ile.’’ August, 1947, Richard Evans Schultes 8663. The distribution of Hevea rigidifolia The small number of stations now known for Hevea rigidifola indicate that it is, with all probability, a highly restricted endemic, for the rather extensive collections of Hevea which have been made in other regions® in the Amazon Valley have never yielded this species. Hevea rigidifola appears to be centered near the con- fluence of the Rio Negro with the Rio Vaupés, or, more precisely, in the triangular sector formed by the lower course of the Rio Vaupés, the Rio Negro above its con- ® Hevea rigidjfolia is erroneously attributed ‘‘am mittleren Orinoko’’ in Planzenr. IV. 147 (Heft 42) (1910) 124, [ 106 | fluence with the Vaupés and the Colombian border. Since this species is endemic to caatingas, however, its dispersal is necessarily interrupted and usually confined to the hinterlands. Therefore, further explorations may and doubtlessly will increase our knowledge of its range. Inasmuch as the Fr6es collections from Jauareté were made exactly on the Colombo- Brazilian boundary, it is safe to include Hevea rigidifolia in the enumeration of Colombian species of the genus (Schultes in Bot. Mus. Leafl. Harvard Univ. 12 (1945) 11). In 1945, I published the following remarks in this connection (here translated from the Spanish) (in Rev. Acad. Col. Ciéne. Exact. Fis- ico. Nat. 6: nos. 22-23 (1945) 386): ‘*This species, col- lected by the English botanist, Richard Spruce, in caatin- gas in the lower Vaupés near Panuré, Brazil, is repre- sented in herbaria only by the type collection. The type locality at Panuré is very near the boundary between Colombia and Brazil and, for this reason, it seems prob- able that Hevea rigidifolia will also be found in the Vaupés of Colombia below Mitt.’ It should be sought in the Rio Querari as well. For the same reason, I suspect that in time we shall also be able to assign Hevea rigidifolia to the Venezuelan flora. [t should be expected to occur in the general vicin- ity of the Piedra de Cocuy. Pittier (Manual de las plan- tas usuales de Venezuela (1926) 262) indicates that Hevea rigidifolia forms a part of the Venezuelan flora, but I strongly suspect that this statement is based upon col- lections of the very thick-coriaceous Hevea pauciflora (Spruce ex Bentham) Muell.-Arg. var. coriacea Ducke, “In the small but critically interesting collection of Hevea which Paul H. Allen, of the Missouri Botanical Garden, made in this gen- eral area in 1943 and 1944 for the Rubber Development Corporation, H., rigidifolia is not represented. (See Baldwin in Amer. Journ. Bot. 34 (1947) 268 for reference to Allen 3049.) [ 107 ] frequent in the Orinoco area, which have been erron- eously determined. In spite of the fact that Hevea rigidifolia may prob- ably occur within the boundaries of three different coun- tries, | am of the opinion that, with the single exception of H. minor Hemsl., it represents the species which is most narrowly restricted in range® and possibly one of the oldest of the genus. It is of interest to note that Hevea rigidifolia is some- times found on caatingas at relatively high elevations. The F’rées 21401 collection from Serra de Tunuhy was made at an altitude of approximately one thousand feet above sea level. I have found that in the peculiar habitat of the caatinga flora of these open sandstone mesas or campinas in the adjacent areas of Colombia, the greater the altitude above the forest floor the more intense the conditions of and response to xerophytism. Thus, the higher one finds the diminutive xerophytic Hevea nitida Mart. ex Muell.-Arg. var. towicodendroides (R. E. Schultes et Vinton) R. E. Schultes (in Bot. Mus. Leafl. Harvard Univ., 13 (1947) 9-11), the more reduced it is in all its vegetative parts. It is significant that the specimens in Foes 21401, trom a tree growing so high above the jun- gle surroundings, shows no special vegetative adaptations correlated with increased intensity of heat, radiation, drought or other xerophytic conditions. According to Senhor Frées, Hevea rigidifolia is found on the flat sandstone mesas as well as on the lightly forested slopes formed by the talus from the eroded cliffs of the hills. The collection F’rées 21401 was made in the light ‘‘talus-forest’” of Serra de Tunuhy, the tree grow- *The range of Hevea camporum Ducke, the specific status of which is now under question by Dr. Adolpho Ducke (in Bol. Téen. Inst. Agron. Norte no. 10 (1946) 19), is not known but may be less extensive than that of any other. [ 108 ] wel A x * ett pe teen * eg PtH H EHH t etter e egg x e) bi + + O- SERRA at + x» ys TUNUHI 4 re © + + + + + + + + Qos Cc FHF HHH Hee egeeeee THE KNOWN RANGE OF HEVEA RIGIDIFOLIA * TYPE LOCALITY (1852) © FROES LOCALITIES (1945) " BALDWIN LOCALITY Fig. 12. Map showing the known range of Hevea riGIpivoLia (Spruce ex Benth.) Muell.-Arg. [ 109 ] ing in gravel and among huge, scattered blocks under severe conditions of chersophytic drought. Senhor Fr6ées described the habitat to me as follows: ‘‘Hevea rigidi- folia in this locality (Serra de Tunuhy) is fair-sized be- cause it is under a light forest; the crown of the slender Hevea protrudes above the forest seeking light. ... I estimate that the mesa-like top of Tunuhy is 1000 feet above the forest floor, but I ascended only about 600 feet, where I collected number 21407.”’ Variability of Hevea rigidifolia The present study of the variability of important spe- cific characters of Hevea rigidifolia is based upon an ex- amination of: (1) the type and a duplicate type in the Royal Botanic Gardens at Kew; (2) a photograph of a specimen of the type collection from the Delessert Her- barium; (8) one duplicate type preserved in each of the following institutions: Gray Herbarium of Harvard University ; the New York Botanical Garden; the Brit- ish Museum of Natural History; the Fielding Herbarium at Oxford University; the Herbarium at Cambridge University, and the Herbarium Amazonicum Musei Paraensis (Museu Goeldi) in Belém do Para; (4) thirty- six specimens of the four Frées collections; (5) a young individual introduced to cultivation at the Instituto Agronomico do Norte by Baldwin in 1944. Further- more, [ have also had the opportunity of questioning Senhor Fr6es in great detail about the habit and habitat of the trees from which his collections were made. It is clearly apparent from a study of these collections from five different localities that Hevea rigidifolia is an exceptionally stable species. It is indeed unusual to find a concept of such stability in this very variable genus. Unless later investigations with large quantities of seed [ 111 ] should indicate otherwise, there seem to be no varieties or forms of the concept, a condition which does not ap- pear to exist in any other well-known species with the possible exception of H. microphylla, All of the collec- tions indicate that Hevea rigidifolia is a small tree, usu- ally between twenty and thirty feet in height. However, there is apparently considerable variation in the diameter of the trunk which may measure, according to Fré6es, from eight to twenty-five centimeters; this variability in the size of the trunk is found in a large number of small trees which grow under the xerophytic conditions of ‘aatingas,. The leaflets of Hevea rigidifolia are apparently always of the same consistency, color and glaucescence; their adult size is very constant, varying usually from 10 to 18 em. in length and 4.5 to 6.5 em. in width; they are al- rays peculiarly and strongly recurved-marginate ; their tips are consistently long and sharply cuspidate-acumi- nate; there is never any trace of pilosity in the axils of the nerves; the number of secondary nerves is twelve to fifteen. Occasionally (as in F’rdes 21394) the strongly rounded base of the leaflets departs from the cuneate condition which is usual in this species; this is, curiously enough, the only vegetative variation of importance which was noted, except for a very slight divergence from a perfectly oblongate-elliptic shape (as in some specimens of Frdes 21253). The inflorescences in all of the collections are peculiarly rigid, are profusely flowered, and are more or less equal or subequal to the leaves. No variations of any sort in the form of the staminate buds is apparent; all are very strongly cylindric-conic, conspicuously acuminate, and densely clothed with a white indumentum. The pistillate flowers, with the spreading and sharply recurved narrow acuminate tips [112 ] of the calyx lobes, show no variation. Size in the flowers is also constant. I have studied five staminate flowers from various specimens of the type collection and thirty-six, taken at random, from the Fr6ées collections. The number of an- thers is extremely constant, being six in all cases but two. In these cases seven anthers were found, the unusual insertion of an extra anther indicating abnormality. I am at a loss to explain Mueller-Argoviensis’ statement that the number can vary between six and ten, unless arare or abnormal condition existed in the material which he studied (necessarily material of the type collection). Hemsley (in Hooker’s Icon. PI. 26 (1898) t. 2573, fig. 11) indicated six anthers, irregularly inserted in two verticils. Spruce, in the original description, described them as they are in Hevea guianensis Aubl. var. lutea (Benth.) Ducke & Schultes, which has most frequently five, less frequently six, but rarely (and probably abnormally) seven or eight. It is evident from a study of the new collections that the number of stamens is normally and quite constantly six. No variation of any appreciable degree was noted in the size, shape, and insertion of the lobes of the stami- nate disk. In all of the flowers examined, they are large, very long-acuminate, and conspicuously spreading. It is of interest to note that the disk lobes in Hevea rigidifolia are larger than in any other species. An examination of three pistillate flowers from the duplicate type at the Gray Herbarium, two from the type at Kew, and fifteen from the several Fr6ées collections has served to emphasize the stability of all the important characters, with the exception of the indumentum of the inner surface of the calyx. The calyx lobes are invariably divided about two thirds or three fifths of their length: the tomentulose indumentum of the exterior of the calyx [ 113 ] lobes is surprisingly constant in density; the interior of the lobes is almost always perfectly glabrous; the ovary is completely glabrous, sometimes very slightly punc- tate; the style is very much abbreviated and thickened ; the stigma lobes are unusually large and carnose, cover- ing the short style; the annular disk is consistently very large and conspicuous, very irregularly deep-laciniate, the tongue-shaped segments often up to 1 or 1.2 mm. in length—probably the largest of any species of Hevea. All but three of the pistillate flowers which I examined were completely glabrous within. In the three excep- tions, | found an extremely sparse tomentulose indu- mentum on the upper half of the inner surface of the ‘alyx, especially near the apex of the lobes. Seibert (in Ann. Mo. Bot. Gard. 34 (1947) 841, pl. 88), has illus- trated the pistillate calyx as very remotely tomentulose on the upper half of its interior surface. Hemsley (loc. cit. fig. 12) published a drawing of an open pistillate calyx of Hevea rigidifolia showing a tomentulose con- dition on the entire inner surface of the lobes, but this was an error. [| have examined the original sketch which is attached to the type sheet at Kew and have found that the interior surface of the pistillate calyx is drawn as tomentulose only near the apex of the lobes. The dis- sected flower on which this sketch was based agrees with the original drawing. It should be noted that the origi- nal sketch of Hevea rigidijfolia was redrawn and, unfor- tunately, altered in preparing the plates for publication. Mueller-Argoviensis (in Martius F]. Brasil. 11, pt. 2 (1874) 800) stated that the calyces of both staminate and pistillate flowers are ‘‘intus inferne praesertim vellereo- tomentelli.’’’ Pax (in Engler Pflanzenr. 4, 147 (Heft 42) °In an earlier description of Hevea rigidifolia, Mueller-Argoviensis (in DC. Prodr. 15, pt. 2 (1866) 718) indicated that the pistillate calyx was “‘intus circa fundum minute subulato-glanduligeri.’’ (See footnote 3.) [114 ] (1910) 124) repeated this, stating: ‘‘calyx.... utriusque sexus intus tomentellus.’’ I find it difficult to reconcile this with my own observations on flowers from both the type and from the new material, but I think we may be justified in assuming that there is some variability in this character and that the normal condition is a glabrous interior of the pistillate calyx. Hemsley (loc. cit. fig. 13) has also published a sketch of the ovary showing a sericeous indumentum covering all parts of the ovary. Mueller- Argoviensis (loc. cit.) and Pax (loc. cit.) both indicate ‘‘ovarium glabrum”’; this coincides with my own observations. Relationships of Hevea rigidifolia There has been some difference of opinion with regard to the closest affinities of Hevea rigidifolia. Mueller- Argoviensis, in his description of Hevea nitida, wrote: ‘‘evidenter affinis H. rigidifoliae et H. brasilensis.’? In a detailed consideration of generic subdivisions in Hevea, Huber (in Bol. Mus. Paraense 4 (1906) 622, 632) in- cluded Hevea rigidifola in the section Bisiphonia, series Luteae, together with HZ. lutea (Benth.) Muell.-Arg.,” H. cuneata Hub.," H. Duckei Hub.,” HH. paludosa Ule, HW. apiculata Baillon”® and H. Benthamiana Muell.- Arg. Huber’s statements regarding the relationships of Hevea rigidifolia should be of interest because of his vast knowledge of the genus, although it is probably true that, due to his confusion of an unusual variant of H. Benthamiana from the Rio Negro with H. rigidifolia, Now considered to represent Hevea guianensis var. lutea. 14 poorly understood concept representing possibly a form of Hevea guianensis. ” Now included by Ducke as a synonym of Hevea Benthamiana. ® Probably a variant of Hevea guianensis. [ 115 ] he did not, at that time, clearly delimit the former from the latter species. He wrote (loc. cit. 682), translated: ‘*Pertaining undoubtedly to the Luteae series because of the disposition of the anthers, Hevea rigidifolia occupies, from other points of view, a special position in this group .... In its whitish pubescence and acuminate staminate buds, the inflorescence strongly suggests Hevea brasilien- sis GHBK.) Muell.-Arg. Furthermore, the segments of the disk of the staminate flower are more strongly de- veloped in this species [i.e. HZ. rigidifolia) than in any other species.”’ Later, the same investigator (in Bol. Mus. Paraense 7 (1918) 202) indicated that the position of Hevea rigidifolia in the Luteae series was open to some doubt; and (loc. cit. 206) that it might be necessary to remove Hi. microphylla Ule™ and H. minor from series Intermediae (he considered series Luteae and Intermediae to represent provisional classifications and not natural groups) and, together with Hi. rigidifolia, form a new group which would be characterized by a strong develop- ment of the segments of the staminate disk. Ducke more recently stated (in Arch. Instit. Biol. Veg. Rio Janeiro 2, no. 2 (1985) 285): ‘‘According to the descriptions, the specie [1.e.: Hevea rigidifolia] dit- fers from H. Benthamiana principally by its glabrous rigid coriaceous leaflets with recurvate margin; the dis- cus and the anthers (6 to 10) do not seem to differ in any essential character from those of the typical Bentham- zana.’’ After having had occasion to see the Frées ma- terial, Ducke (in Bol. Téen. Instit. Agron. Norte no. 10 (1946) 13) wrote in regard to Hevea rigidifolia ‘‘distingue- se, de uma Benthamiana de folhas inteiramente glabras, pela consistencia rigidamente coriacea das mesmas. .. . as flores masculinas sio mais longamente acuminadas e ' Until recently confused with Hevea minor. [ 116 | mais estreitas que em qualquer especimen de Bentham- lana, por mim visto.”’ The extensive Froes material, with its abundance of flowers, enables us to evaluate more critically and more extensively than has hitherto been possible the relation- ships of Hevea rigidifolia. It would appear quite prob- able that Hevea rigidifolia occupies a somewhat inter- mediate position between Hevea pauciflora var. coriacea” and HI. Benthamiana,” having, florally at least, closer affinities with the latter than with the former. The extremely heavy coriaceous nature and the glau- cescence of the leaflets of Hevea rigidifolia as well as the glabrous condition of the lower surface of the leaflets In the present article, I am using the concept Hevea pauciflora var. coriacea in the sense in which it has been established and used by Ducke (in Arch. Instit. Biol. Veg. Rio Janeiro 2, no. 2 (1935) 239). The concept has also been used in this sense in several of my previous papers (in Bot. Mus. Leafl. Harvard Univ. 12 (1945) 1-32; in Rev. Acad. Colomb. Ciéne. Exact. Fisico-Quim. Nat. 6, no. 22-23 (1945) 331-888; in Rev. Fac. Nac. Agron. (Medellin) 6, no. 22 (1946) 18-45). I suspect that what has been considered as Hevea paucifilora var. coriacea represents a concept deserving varietal distine- tion. Baldwin calls this H. confusa (Baldwin: in Journ. Hered. 38 (1947) 54; Baldwin & Schultes in Bot. Mus. Leafl. Harvard Univ. 12 (1947) 335). Until it is possible to investigate the entire Hevea pauciflora complex critically, however, I believe it is advisable to main- tain Ducke’s treatment of it. A complete study of the H. paucjflora var. coriacea— H. rigidifolia—H. nitida—H. confusa problem and of the relationships between these concepts is urgently needed and must be carried out before a clear understanding of the genus as a whole can be expected. Ducke has recently (in Bol. Téen. Instit. Agron. Norte no. 10) (1946) 18) indicated doubt as to whether Hevea confusa should be maintained as a variety of H. pauciflora or incorporated with H. paucifiora var. coriacea. The writer believes that Hevea Benthamiana and H. Spruceana (Benth.) Muell.-Arg. are distantly allied species and that some rela- tionship could, therefore, be expected to exist between H. rigidjfolia and H. Spruceana. Since it appears, however, to be of a minor degree, a detailed analysis of this relationship will not be discussed in the present study. [ 117 ] combine with other general similarities in vegetative characters to indicate a rather close relationship with its companion of the caatingas, AZ. pauciflora var. coriacea. It is worth pointing out, incidentally, that while coria- ceous leaflets are very possibly a manifestation of xero- phytic caatinga conditions in both of these species, as in many other plants, there are sufficient similarities in vegetative characters to indicate that this is due to some affinity and not to an ecological parallelism between Hevea rigidifoha and H. pauciflora var. coriacea. The seed of Hevea rigidifoha does not appear to be similar to that of AZ. pauciflora var. coriacea. It is larger, broader in relation to its length, and much less sharply angular. However, one seed preserved with the collec- tion Mrées 21253 curiously approaches the shape and size which, so far as we know, is usual for Hevea pauciflora var. coriacea, being smaller, narrower, and very sharply angular in cross section. (This seed may have been in- cluded erroneously in F’rdes 21253, because it does not match any other seeds in the same collection. Further- more, rdes 21253 was growing in close proximity to Froes 21249 which appears to be referable to the Hevea pauciflora var. coriacea complex, and a confusion as to the source of this seed would not be difficult.) Ducke (in Bol. 'Téen. Instit. Agron. Norte no. 10 (1946) 13) indicates that the seed of Hevea rigidifolia is similar in shape and size to that of typical HY. pauciflora. Uf this unique seed actually belongs to Hevea rigidifolia, it might suggest some relationship with HZ. gwianensis var. lutea or H. nitida which have similar angular seeds. We cannot draw any definite conclusions, however, on the basis of the few known seeds of Hevea rigidifolia. In floral characters, however, there is less to point to a close affinity between FTevea rigidifolia and H. pauci- flora var. coriacea, This is especially true when the dif- [ 118 | ferences in such critical characters as the number of sta- mens, the size of the anthers, the length of the supra- staminal column, the size and form of the glandular lobes in the staminate flower, and the condition of the annular disk in the pistillate flower are taken into consideration. The very different form of the masculine buds—short, obtuse, even somewhat truncate in Hevea pauciflora var. coriacea, but extremely long and acuminate with the tips of the segments peculiarly spiral-imbricate in HZ. rigidi- folia—likewise would tend to indicate that the affinity between the two might not be exceptionally close. The young leaflets and seedling leaves of Hevea rigid- ifolia, asin some specimens of F’rdes 21258, are strikingly similar to the adult leaflets of H. pauciflora var. coriacea, although the shape and size of the adult leaflets of HZ. rigidifolia are peculiar and cannot be easily confused with those of any other species. In connection with the relationship of Hevea rigidi- folia and H. pauciflora var. coriacea, it is pertinent to discuss several interesting collections which, in some re- spects, would superficially appear to be rather intermedi- ate between the two concepts. As stated above, in March, 1944, Dr. John T. Baldwin, Jr. discovered Hd. rigidifolia and introduced it into cultivation (Baldwin in Am. Journ. Bot. 38 (1946) 215-216). There is one young in- dividual growing vigorously in the collection at the In- stituto Agrondmico do Norte (PI. TX). Baldwin made chromosome studies in the field and prepared herbarium material, but unfortunately there appears to be some confusion in the determination of sev- eral of his herbarium collections. In two articles (Journ. Hered. 388 (1947) 59; Am. Journ. Bot. 34 (1947) 261) Baldwin cites his collections 3669 and 3670 as represent- ing Hevea rigidifolia. 1 have studied a specimen of Baldwin 3670 which is preserved in the Herbarium of (119 J the Arnold Arboretum. It clearly represents one of the numerous variants of Hevea pauciflora. A good photo- graph of Baldwin 3669, a specimen of which I have not seen, has been published (in Am. Journ. Bot. 34 (1947) 264, fig. 2). It represents the same variant of the Hevea pauciflora complex as Baldwin 3670. Although these collections, with fruit but lacking flowers, exhibit simi- larities in certain characters with typical Hevea rigidi- Jolia, it is highly inadvisable, in my opinion, to consider them as representing this peculiarly distinct species. Several valves of the capsule which accompany Bald- win 3670 would appear to be typical of Hevea rigidifolia in the shape and size of the few seeds, for both approach somewhat the analagous fruiting material of /’rdes 21258. The ligneous valves of Hevea rigidifolia are peculiar to that species, being large, long and narrow with unusually thick endocarp, and very strongly twisted due to the explosive opening of the capsule; whereas in Hevea pau- ciflora var. coriacea the valves are smaller, as broad as long and have relatively thick and strong ligneous walls which are not in the least distorted during the explosive shedding of the seed. The peduncle of the fruit, pre- served in Baldwin 3670, is like that of Hevea rigidifolia. However, it is in a study of the leaflets, so extraordi- narily constant in typical Hevea rigidifolia, that doubts arise as to the real identity. Although the under surface of the leaflets of Baldwin 3670 is armed with the minutely tessellate, waxy scales which lend the peculiar dull waxy- gray color to that part of the leaf of Hevea rigidifolia, the individual scale-like mosaic points are much smaller in the Baldwin collection than is normal for Hevea rigid- ifoha. It is, however, principally in the texture and gen- eral appearance of the leaflets that Baldwin 3670 deviates most from typical Hevea rigidifola. The leaflets are long and narrowly lanceolate-elliptic, departing from the very { 120 | characteristic and constant oblong-ovate shape; the tips do not have the very long and pronounced cuspidate acu- men so typical of Hevea rigidifolia; the upper surface 1s, in life, apparently much duller and somewhat less glau- cescent ; the lower surface is not so whitish; the margin is very much less conspicuously inrolled; the petiolules are extremely abbreviated, whereas one of the major charac- teristics of H. rigidifolia is the presence of an unusually long petiolule; and the texture, in general, is far less coriaceous. The leaflets of Baldwin 3670 are very close indeed in shape, size and texture to those of Hevea pau- ciflora var. coriacea. Baldwin (in Journ. Hered. 38 (1947) 56) has stated that Hevea pauciflora ‘‘is probably the prototype of H. brasiliensis and was likely one of the parents of H.rigid- ifolia.”’ Later, the same investigator went much farther and wrote: (in Am. Journ. Bot. 84 (1947) 265) ‘‘... A. rigidifolia (and certain other species) may have originated from intergeneric hybridization of Hevea and Cunwria. For this hypothesis, H. pauciflora (from which it is con- sidered that Hi. brasiliensis evolved) and Cunuria cras- sipes with purplish fruit are selected as possible parents. ”’ I cannot accept these assumptions for which no shred of evidence is advanced. Baldwin (loc. cit. 265) advances, apparently as arguments for his reasoning, a miscellany of observation and speculation when he states: ‘‘At Montepelago.... and at Iraruca.... purplish-fruited Cunuria crassipes grows adjacent to Hi. rigidifolia. The leaflets of this species of Hevea are cunurioid. On top of the hill at Montepelago, is a huge tree referable to H. pauciflora, and that tree is buttressed just as repre- sentatives of Cunuria are buttressed. Seed of AH. rigid- ifolia and of H. pauciflora are much alike. ‘Those of HZ. confusa, though smaller, fall in this group. Under in- fluence of the elements, the seed of Hi. rigidifolia lose [ 121 ] their color pattern and assume the uniform brownish- red hue of Cunuria seed. That the natives recognize in the seed of HI. pauciflora resemblance to those of cunury may be of significance. ’”’ Another very interesting collection is Allen 3049 from the Rio Papuri, an affluent of the Vaupés, which forms part of the Colombo- Brazilian boundary. The leaflets are somewhat like those of typical Hevea rigidifolia, but the seed and capsule valve suggest those of H. pauciflora var. coriacea. The seeds, smaller than usual for the Hevea pauciflora complex, are very strongly angular, and the valves, long in relation to their breadth, are not twisted. The leaflets, less coriaceous than in typical Hevea rigidifolia, are unusually obovate and very shortly cuspidate; they are extremely large. Of this collection Baldwin (in Am. Journ. Bot. 84 (1947) 268) writes: ‘%... is a puzzling specimen. R. J. Seibert has annotated it as a possible hybrid of HZ. pauciflora (or H. confusa) and A. rigidifolia... carpels of old fruit, which seem refer- able to H. confusa Hemsl. and which could have been wrongly associated with the collection.... My inclination is to consider the material. ... as HQ. confusa with in- fluence from Hi. rigidifolia.”’ A third collection of significance in this problem is Froes 21249. 'This likewise represents a variant of the Hevea paucifiora complex, but not the same variant as Baldwin 3669, 38670 and Allen 3049. This extensive col- lection has leaflets which suggest, in several minor char- acters, those of Hevea rigidifolia, but its flowers indicate its true affinities. Upon superficial examination of the leaves, L annotated this collection in July 1946 as Hevea rigidifolia, but later, detailed examination of the flowers disclosed my error. Baldwin (loc. cit. 268) has considered this collection to represent Hevea rigidifolia, stating: “Froes 21249 shows influence of H. confusa. This col- [ 122 ] lection and Allen 3049 resemble certain of my specimens of H. confusa from along the Rio Negro.’’ In May, 1947, he annotated two specimens in the Herbarium of the Instituto Agrondmico do Norte (16736, 16737) as “HL. rigidifolia with influence from H. confusa.’’ It is indeed of interest that F’roes 21249 was found growing in close proximity to an individual of Hevea rigidifolia (F'roes 21253). It is suggestive that some relationship may exist be- tween Hevea rigidifola and H. nitida. Baldwin (in Journ. Hered. 38 (1947) 59) considers Hevea rigidifolia “and H. confusa {| H. pauciflora var. coriacea), H. pauci- Hora and H. viridis |.H. nitida] to be expressions of the same complex.”’” There is not only a similarity in the seed, but also the general coriaceous and reclinate condi- tion of the leaflets could be interpreted as indicative of some degree of affinity. The floral characters of the two species, however, do not approach each other in any es- sential respect. The flowers of Hevea nitida indicate, in my opinion, a very close relationship with HZ. brasiliensis, thus removing Hd. nitida from any immediate phylo- genetic proximity to A. rigidifolia.” The resemblance of the seed of Hevea rigidifolia to that of one form of H. brasiliensis of the upper reaches of the Solimoes must be interpreted as coincidental. The ‘‘similarity’’ between the flowers of Hevea rigidifolia and AZ, brasiliensis which Mueller-Argoviensis intimated when he wrote in the description of the latter: ‘‘facies florum ut in A. rigidifolia sed structura diversa’’ is en- tirely superficial and, from a phylogenetic point of view, probably insignificant. There does appear to be a likeness between the nor- Baldwin (loc. cit. 59) considers H. pauciflora var. coriacea (H. Ore. é ooze ° ee confusa) and H. nitida (H. viridis) so closely allied that they “‘may eventually be combined.’’ I cannot subscribe to this opinion. D128") EXPLANATION OF THE ILLUSTRATION Prare TX. A seedling of Hevea rigiprroiia (Spruce ex Benth.) Muell.-Arg., growing in the garden of the Instituto Agrondémico do Norte at Belém do Parad, Brazil. [ 124 ] PuatrE LX _ mal seed of Hevea guianensis var. lutea and that of Hi. rigidifola, especially in shape and in coloration. The seed of the former is, however, smaller than that of the latter. This similarity may be significant, but we cannot draw any definite conclusions on the basis of the few known seeds of Hevea rigidifolia. The flower of Hevea rigidifolia does indeed show, in several characters, an approach to that of H. gwianensis var. /utea; this was suggested by Spruce in the original description when he stated: ‘‘antherae JS. /uteae.”” None of the floral characters, however, indicates any degree of close affinity. Moreover, the habits and habitats, and therefore probably also the history, of the two concepts are very unlike in all respects, for the one is a widely distributed forest giant growing on well-drained ground along the banks of rivulets or brooks, whereas the other, apparently a highly restricted endemic, is usually a small tree of xerophytic caatingas. In floral characters, Hevea rigidifolia does, as Ducke has pointed out, resemble H. Benthamiana very closely. In Hevea rigidifolia the staminate buds are extremely acuminate and cylindric-conic; in. H. Benthamiana they are commonly acuminate, although the latter does not have the segments apically spirally twisted as does the former. Although the flowers of the former are much larger than those of the latter, the general form of the urceolus of the staminate flower is markedly similar. There are, however, several rather important differences which do serve to separate the two. In Hevea rigidifolia the number of anthers, surprisingly constant, is very low (normally only six), the suprastaminal column is conspic- uously slender and elongate, and the ovary is completely glabrous; whereas, in H. Benthamiana, the anther num- ber is variable (usually from six to nine), the column is short, and the ovary is densely white-sericeous. In both [ 127 ] species, however, the lobes of the disk in the staminate flower are similar, being rather elongate, very acuminate and conspicuously erect-spreading. The lax and irregu- larly spreading long-acuminate tips of the calyx lobes of the female flower of Hevea rigidifolia, however, has no similarity to any of the forms of H. Benthamiana known to me; indeed, the female flower of Hevea rigidifolia, which, upon superficial observation would seem to be somewhat like that of HY. microphylla, appears to be unique in the genus in this interesting characteristic. The characters of the ligneous valves of the fruit have hitherto not been utilized taxonomically, but it is of great interest to note in passing that those of Hevea rigidifolia are very like those of most of the known forms of Hi. Benthamiana in having an unusually thin endocarp. The long and narrow shape of the valves of Hevea rigidifolia is very different from that which is commonly seen in HI. Benthamana, but it does remind one of the shape of the valves of H. Spruceana. Ducke (in Bol. Téen. Instit. Agron. Norte no. 10 (1946) 18-14) indicates that a speci- men collected by him in 1905 on the Rio Negro at Bar- cellos and distributed by Huber as ‘‘Hevea rigidifolia”’ is a form of HI. Benthamiana var. subglabrifolia Ducke with perfectly glabrous leaflets. (See also Ducke in Arch. Instit. Biol. Veg. 2, no. 2 (1985) 235 and in Arquiv. Serv. Florest. Rio de Janeiro 2, no. 1 (1948) 86.) Recently in Belém, I had occasion to study a remarka- ble collection of Hevea Benthamiana from the Rio Ne- gro which Ducke has determined as representing a per- fectly glabrous form of var. subglabrifolia.” 1 believe that further study will possibly indicate that this collec- tion represents a distinct variety. The reclinate leaflets 'S Brazil: Amazonas, Rio Negro infer., loco Acajutuba. ‘‘Igapé secus rivulum ad marginem campinae. Arbor 18-20 m.; latex flavi- dus, parum copiosus.’’ March 22, 1941, Ducke 1963, [ 128 ] are extremely coriaceous with strongly recurvate margins, as in Hevea rigidifolia, and although they are much smaller and more narrowly lanceolate-elliptic, they could very easily be misidentified (without fertile parts) as Hi. rigidifolia. The seeds and capsules of this collection, however, unquestionably belong to the concept Hevea Benthamiana; they are much smaller than in typical Hevea Benthamiana, but they are perfectly ovoid and are yellowish or light golden tan with large reddish brown or brownish black spots. The tree, eighteen or twenty meters tall, had sparse yellowish latex. It grew along a brooklet at the edge of asavanna. According to Ducke, the coriaceous character of the leaflets of this variant was gradually lost as one penetrated farther into the more densely forested igap6. This collection, as the analagous Barcellos material mentioned above, provides us with in- teresting data for a consideration of xerophytic parallel- isms in Hevea. It is entirely possible, too, that this un- usual form of Hevea Benthamiana which, at our present state of understanding, we call a parallelism, may be fundamentally significant as indicative of some actual relationship between Hevea Benthamiana and H.rigidi- folia. Possible economic and scientific importance of Hevea rigidifolia Little is known of the economic value of Hevea rigid- ifolia. Ducke (in Rev. Bot. Appl. 9 (1929) 628) intimated that the species “‘semble fournir du caoutchouc,’’ but naturally he could make no mention at that time as to whether or not it was ever exploited. Huber (in Bol. Mus. Paraense 4 (1906) 633) stated that Hevea rigidi- folia: ‘‘foi designada ao Sr. Ducke como fornecendo bor- racha b6éa,’’ but this statement was made in reference to a collection from Barcellos which was misidentified as A, rigidifolia (Ducke loc. cit.). Jumelle (‘‘Les plantes [ 129 ] a caoutchoue et a gutta’’ (1908) 132) repeated Huber’s statement that Hevea rigidifolia yields a good rubber. La Rue (USDA Bull. 1422 (1926) 8) included Hevea rigidifolia in a list of species yielding ‘‘fair to good rub- ber,’’ but there seems to be no basis for this statement. Spruce (in Hooker’s Journ. Bot. ser. 8, no. 6 (1854) 871) merely remarked that it is ‘‘a milky tree of 80 feet.”’ Pittier (loc. cit.) enumerated three species of Hevea for the flora of Venezuela: HI. Benthamiana, H. minor and H. rigidifolia, but, as stated above, it is doubtful that his report of the last-named species is based on cor- rectly determined material; it refers probably to A. pauciflora var. coriacea. However this may be, Pittier writes that (translated) ‘‘of the three, the two first seem to give good rubber, but we know nothing about the third. It is probable that all take part in rubber produc- tion in the country.”’ The field data annotated on Frédes 21394 indicate merely ‘“‘latex branco, abundante, coagulavel,’’ and Foes 21401 adds that the species can be ‘‘rica em latex branco, coagulavel.”’ In writing of several of his caatinga collections, Frées (‘‘Excursio botanica ao alto Rio Negro’’ Mss. report to Dr. Felisberto C. de Camargo (June 4, 1946) 28) states (translated): ‘‘T'he Heveas of the forest at Javarité (21231, 21249, 21251, 21253) which are also common in the stretches of forest at Caiary-Aiary are low, measuring hardly 5 or 6 meters by 10 to 12 centimeters in diameter at breast height, with heavy and reclinate foliage, with medium-sized fruits as can be seen from the specimens collected in the region of the Rio Icana, in caatingas similar to those of Caiary-Aiary; they certainly must be little known because of the restriction of this species to its peculiar habitat. These forms of Hevea have, in themselves, no commercial importance because besides [ 130 | their small size they are poor producers of latex, and the latex itself is of inferior quality. However, they would be of scientific value because of their rarity. ”’ It would seem from this statement, and from others in the same report which indicate that the greater part of the upper Rio Negro production of rubber is from Hevea Benthamiana, that we are justified in inferring that Hevea rigidifola furnishes none of the rubber pro- duced in the area. This does not, however, assist in the appraisal of the quality of the rubber, for it is clear that the principal reason for not utilizing Hevea rigidifola would be its small size. It might possibly be true that, as in the case of Hevea nitida var. toaicodendroides, its latex could be of high quality even though almost un- obtainable due to the diminutiveness of the plant. The fact that in its native state Hevea rigidifolia is of no commercial interest does not, however, indicate that it is devoid of importance to the scientific program of Hevea-investigation now in progress cooperatively be- tween the United States Department of Agriculture and several Latin American countries. I consider that one of our most urgent tasks is the introduction of extensive living material of Hevea rigidifolia from several different localities to one or several of our experimental nurseries. The differences between Hevea rigidifolia and the spe- cies commonly exploited, especially HZ. brasilensis, are so great that we would be justified in expecting a num- ber of interesting, and perhaps even revolutionizing, characteristics to appear as a result of an intensive pro- gram of selection and breeding. This small tree is native to isolated sterile (cretaceous) sandstone caatingas on the pre-cambrian granitic Brazil- ian shield where the pH is probably rather highly acidic and where conditions of psammophytic and chersophytic drought prevail even during the rainy season. How this [ 131 | species will react in growth as well as in quality and yield of latex when planted on a rich soil with an abundant availability of water is certain to be enlightening.” Selec- tion should be carried out to some extent in the caatingas in order to secure for scientific study the most outstand- ing individuals from several points of view. A still more rigorous and accurate selection can be practiced in nurse- ries if large numbers of seedlings are planted and studied under controlled conditions. The potentialities of this species as a fast grower, outstanding yielder, disease- resistant strain, drought-resistant material, root-stock, crown-budding stock, as well as in numerous other re- spects, are certainly worth investigation. The very fact that Hevea rigidifolia is one of the most distinct of the dozen or more species of the genus is in- dicative of the value which it might have in a plant- breeding program. This is especially true when one con- siders the possibilities which might arise as a result of chromosome incompatibility. It is likewise imperative that we collect ample speci- mens of the coagulated latex from Hevea rigidifolia in its native state for chemical examination and later chem- ical comparison with latex from the same species culti- rated under varying ecological conditions. It is possible that, as in the case of Hevea nitida in the igapés of the Vaupés of Colombia, of AZ. Spruceana and of H. pauci- Jlora var. coriacea, the rubber of Hevea rigidifolia, of itself, will be of little or no commercial value. The rapid progress in the study of the synthetic elastomers, how- ever, is creating a demand for rubbers of varying proper- ties for use as ‘‘fillers’’ and the latex of no species of Hevea should be overlooked as a potential raw material for utilization in the synthetic rubber industry. "The individual introduced by Baldwin in 1944 to Belém do Para, planted on very light white sandy soil, exhibits an unusually slow growth. It is, at three years, only about five feet tall. [ 132 ] BOTANICAL MUSEUM LEAFLETS HARVARD UNIVERSITY Voi. 13, No. 6 CamBripGe, Massacuusetts, NovEMBER 24, 1948 ’ ’ NOTES ON ORCHIDS OF THE AMERICAN TROPICS BY CHARLES SCHWEINFURTH NEW CONCEPTS OF THE PANAMANIAN ORCHID FLORA THE following new concepts became apparent in the course of the critical work undertaken by Mr. Paul H. Allen on the orchid flora of Panama. Oakes-Amesia C. Schweinfurth & P. H. Allen gen. nov. Divisio: Acrotonae. Tribus: Kerosphaereae. Series: Pleuranthae. Subtribus: Ornithocephaleae. Flores par- vi. Perianthii partes liberae, late patentes. Sepala ovata vel oblongo-ovata, mucronata. Petala sepalis majora, late cuneata, apice truncata vel retusa. Labellum medio profunde trilobatum, per medium longitudinaliter tubulo percursum. Columna brevis, supra in processum rostel- larem superne dilatatum et trilobulatum extensa. An- thera perlonga, sigmoidea, supra abrupte semigloboso- concava. Pollinia quattuor, parva, stipiti perlongo affixa. Herba pusilla, epiphytica, Ornithocephalo valde affinis. Caulis perbrevis, foliorum circulo omnino obtectus. Folia equitantia, late patentia, infra vaginis imbricantibus artic- ulata. Inflorescentiae racemosae, axillares, pluriflorae, folia superantes. [ 133 ] Oakes-Amesia cryptantha C. Schweinfurth & P. H, Allen sp. nov. Herba pusilla, epiphytica. Radices fibratae, pubes- centes, numerosae. Caulis perbrevis, foliorum vaginis valde imbricatis omnino obtectus. Folia equitantia, vagi- nis articulata; laminae elliptico-lineares, acutae; vaginae conduplicatae. Inflorescentiae singulae vel tres, in folio- rum superiorum axillis, racemosae, folia paulo excedentes. Flores parvi, cum segmentis liberis patentibus. Sepalum dorsale oblongo-ovatum, concavum, valde mucronatum. Sepala lateralia oblique ovata, coneava, valde mucronata. Petala triangulari-cuneata, abrupte et late truncata vel leviter retusa. Labellum medio profunde trilobatum, valde concavum cum lobis lateralibus et lobi medii later- ibus erectis; lobi laterales late oblongi, apice rotundati; lobus medius expansus subquadrato-cuneatus, apice trun- catus; discus per medium tubulo percursus et supra tu- bulum fascia papillosa ornatus. Columna brevis, apoda, supra geniculata, antice cum processu rostellari elongato trilobato. Anthera operculata, incumbens. Pollinia quattuor. Plant epiphytic, small, up to 5 cm. high. Roots slen- der, fibrous, finely pubescent, numerous. Stem obscure, entirely enveloped by a circle of leaves. Leaves equitant, coriaceous, about six or more in number, articulated to imbricating sheaths; blades elliptic-linear, acute, 1.8-3.9 em. long, 3.5-4.5 mm. wide; sheaths conduplicate, short, 6-8 mm. long. Inflorescences one to three, spring- ing from the upper axils, racemose, erect or arcuate, about 5.5 cm. long, apparently 9-flowered, with the short peduncle and lightly fractiflex rachis narrowly bialate. Bracts spreading, triangular-lanceolate to ovate-lanceo- late, long-acuminate, amplexicaul, dorsally carinate as an extension of the bialate rachis. Flowers small, white, the lip being dark green marked with white. Perianth [ 184 ] parts widely spreading. Dorsal sepal oblong-ovate, con- cave, mucronate, 1-nerved, dorsally lightly carinate, the upper margins being minutely erose, about 2.7 mm. long and 1.25 mm. wide when expanded. Lateral sepals ob- liquely ovate, concave, mucronate, 1-nerved, dorsally carinate, with the upper margins minutely erose, about 2.9 mm. long and 1.5 mm. wide when expanded. Petals triangular-cuneate, abruptly and broadly truncate to lightly retuse at the apex with the upper margins mi- nutely erose, 3-nerved through the center, about 3 mm. long and subequally wide above. Lip complex, deeply 3-lobed in the middle, concave with the lateral lobes and the sides of the mid-lobe erect in natural position, about 3.5 mm. long, the longitudinal center traversed by a narrow tube (with opening at the apex), above which is a linear-oblong fleshy papillose band beset with hairs at the apex ; lateral lobes obliquely and broadly oblong with a rounded apex, porrect; mid-lobe subquadrate-cuneate when expanded with the opening of the tube in the center of the truncate apex and with a high fleshy keel on the under surface forming a prominent mucro when viewed from above. Column short, right-angled-genicu- late near the middle, about 1.5 mm. high; from the sinus of the angle in front extends a porrect elongate rostellar process about 8 mm. long which is dilated above and prominently mucronate (thus 3-lobulate), the dilated portion on each side being decurved and forming a rounded semiobovate lobule; anther imbedded on the rostellar process (i.e. incumbent), elongate, abruptly sig- moid and semiglobose-dilated above; pollinia four, small, affixed to the apex of an elongate triangular-linear stipe. At first sight Allen & Allen 4196 appeared to repre- sent a typical member of the well-marked genus Orni- thocephalus. An examination of the flowers, however, revealed characters which could not be reconciled with [ 185 ] EXPLANATION OF THE ILLUSTRATION Pirate X. Oakes-AmesIA cRYPTANTHA C, Schweinfurth & P. H, Allen. 1, plant, twice natural size. 2, flower, from side, six and two thirds times natural size. 3, dorsal sepal, six and two thirds times natural size. 4, petal, six and two thirds times natural size. 5, lateral sepal, six and two thirds times natural size. 6, lip, from above, six and two thirds times natural size. 7, portion of column, with rostellar process and anther in place, from side, six and two thirds times natural size. 8, anther with pollinia, natural position, from side, six and two thirds times natural size, 9, pollinia with stipe and viscid disc, natural position, from side, six and two thirds times natural size. 10, pollinia with stipe and viscid disc, ex- tended, from above, six and two thirds times natural size. Drawn by Gorpon W, DILion. [ 136 ] PLatE X eryptantha & PA Allen OAKES-AMESIA cc Schwein fi our present conception of that genus. After prolonged study, it has seemed to us the wiser course to recognize this collection as constituting a new genus. A marked point of difference from Ornithocephalus is the sharply 3-lobed lip provided with a central longitudi- nal tube (open in front) and an overlying fleshy papillose linear-oblong band which is pubescent in front. More- over, there is a stout trilobulate rostellar process that is very different from the slender simple attenuate rostellar process of Ornithocephalus. In profile the column with its rostellar process and anther suggests the head and folded legs of a praying mantis. The name Oakes-A mesia was chosen to commemorate the foremost living orchidologist who has recently com- pleted a term of fifty years as teacher, mentor and bene- factor of Harvard University. Panama: Prov. of Coclé, summit of Cerro Pajita, hills north of El Valle de Anton, at 1000-1200 meters altitude, epiphyte on small trees in dwarf cloud forest, January 7, 1947, Paul H. and Dorothy O. Allen 4196 (Tyre in Herb. Ames No. 64880). Gongora armeniaca (Lind/.) Reichenbach filius var. bicornuta C. Schweinfurth & P. H. Allen var. nov. Herba florum colore et labelli hypochilio antice bi- cornuto et medio supra conspicue dentato et columnae apice utrinque angulato a specie differt. Plant rather small, with an abbreviated rhizome. Pseu- dobulbs approximate, broadly ovoid, bifoliate, deeply plurisuleate, 2-3 cm. high. Leaves lanceolate-elliptic, acuminate, shortly narrowed to a subpetiolate base, pli- cate, with 8-5 nerves conspicuously exserted beneath, up to 12 cm. long and nearly 4 ecm. wide. Scape lateral, basal, pendent, filiform below, slightly thickened above, very loosely few- (up to 5-) flowered in the upper part, about 183 cm. long; rachis of raceme lightly fractiflex, [ 139 ] “XPLANATION OF THE ILLUSTRATION PLtare XI. GonGorRaA ARMENIACA var, BICORNUTA C, Schweinfurth & P. H. Allen. 1, plant, natural posi- tion, about natural size. 2, lip, natural position, from above, nearly three times natural size. 3, lip, natural position, from side, three times natural size. 4, column (with foot) and petals, natural po- sition, about twice natural size. Drawn by Dorotuy O, ALLEN [ 140 ] PiateE XI nearly 5 cm. long. Sepals pale cream-color, minutely spotted with red. Dorsal sepal erect and cucullate, obo- vate, acute, about 1.8 cm. long and 1.2 cm. wide above, with the mid-nerve slightly carinate without. Lateral sepals spreading-reflexed, very obliquely elliptic-ovate with the anterior margin nearly semicircular and the pos- terior margin nearly straight when expanded, acute, the mid-nerve distinctly carinate without, long-adnate to the column-foot at the base, about 1.9 cm. long to the pos- terior point of attachment and nearly 1.5 cm. wide. Petals small, dark red, very obliquely inserted on the column, sigmoid, oblong-lanceolate, with an upcurved, acute or acuminate apex. Lip very fleshy, with a broad abbreviated claw, about 1.5 cm. long in natural position, rich waxy orange with a dark red tip to the epichile; hypochile, when viewed from the side, terminating on each side in a conspicuous lightly incurved horn, with a short but prominent tooth near the middle of the upper surface, subquadrate-pandurate when viewed from above ; epichile upeurved and inserted at nearly a right angle to the hypochile, triangular-lanceolate with an uncinate tip when viewed from the side, linear-triangular and sulcate when viewed from above. Column relatively short, pale cream-color dotted with red, arcuate, clavate, with a prominent angle on each side of the apex, about 1.2 cm. long, extended into a stout sulcate pubescent foot. This concept differs rather strikingly from Gongora armeniaca in the color of the flowers, in the cornute an- terior projections of the hypochile, in the conspicuous middle tooth on its upper surface and in the angulate apex of the column. Panama: Prov. of Veraguas, region west of Santa Fé, Cerro Tute, at 1000 meters altitude, June 24, 1947, P. H. Allen 4648 (Typr in Herb. Ames No. 64263). [ 148 ] Gongora maculata Lindley var. latibasis C. Schweinfurth & P. H. Allen var. nov. Herba floribus majoribus atque labelli hypochilio per- lato basi triangulari nec vere cornuto a specie differt. Vegetative portions not seen. Pseudobulbs said to be subglobose, about 5 cm. in diameter, deeply suleate. Leaves about 80 cm. long and 10 cm. wide, grayish green. Flowers (of which I have examined two) some- what larger than those of Gongora maculata; buds tawny yellow; sepals dark blood-red; petals and column green with small red spots. Dorsal sepal broadly ovate-elliptic, acute, adnate to the middle of the column, lightly con- cave, with revolute margins, about 2.3 cm. long and 1.3 cm. wide when expanded. Lateral sepals larger, wide- spreading, obliquely triangular-ovate, acuminate, with revolute margins, the base being long-adnate to the column-foot, about 3 cm. long to the posterior point of insertion and 2 cm. wide. Petals relatively small (as in the genus), long-adnate to the column, the free portion being sigmoid, linear, acute, strongly incurved in natural position, Lip very fleshy, about 2.2 cm. long, adnate to the abbreviated free apex of the column-foot; hypochile (when viewed from above) subquadrate-triangular with a very broad subbasal portion which is destitute of horns, the anterior portion in front of the broad rounded sinus produced on each side into an erect bristle; epichile nar- rowly triangular and acuminate when viewed from above, laterally compressed and broadly triangular with a re- curved apex when viewed from the side. Column arcu- ate, slightly enlarged above, about 2 cm. long. By reason of its large size, this concept appears to be quite distinct from the other members of the genus Gon- gora which have been examined. It is unusual, among the other members of its alliance, in having a lip with a very broad triangular basal portion which lacks any true horns. [ 144 ] Panama: Canal Zone, about 2 miles west of Gatun Dam, at about 184 meters altitude, April 1, 1939 (blooming time), Butcher s.n. (Type in Herb. Ames No, 64244). Brassia Allenii L. O. Williams ex C. Schweinfurth sp. nov. Herba epiphytica, robusta, sine pseudobulbo caulique. Vaginarum foliiferarum fasciculus flabelliformis. Vaginae conduplicatae, imbricatae. Folia oblanceolata vel ob- longo-lanceolata vel lineari-oblanceolata, oblique acuta. Inflorescentia lateralis, axillaris, foliis brevior, laxe pau- ciflora. Flores pro genere parvi. Sepala similia, lineari- lanceolata, supra longe attenuata. Petala quam sepala breviora et latiora, oblique et anguste lanceolata, longe attenuata. Labellum ovatum vel suborbiculari-ovatum, antice abrupte in apicem anguste triangularem produc- tum; discus callis duobus brevibus carnosis apice in den- tem productis ornatus. Columna abbreviata, crassa. Plant epiphytic, more or less robust. Roots fibrous, glabrous, stout, numerous. Stems or pseudobulbs lack- ing, their place being taken by a flabellate cluster of leaf- bearing sheaths. Leaf-sheaths several to numerous, im- bricating, conduplicate, nervose, surrounded on each side by abbreviated often non-leaf-bearing sheaths. Leaves oblanceolate to oblong-lanceolate or linear-oblanceolate, very obliquely acute with one side of the apex much surpassing the other, gradually narrowed to a complicate or conduplicate base, plicate, many-nerved with about seven to nine nerves prominently exserted beneath, com- monly about 18.5-830 cm. long and 2.3—2.7 (rarely 3.8) cm. wide. Inflorescence in the axil of an upper leaf- sheath, loosely 6- to 7-flowered from near the base, dis- tinctly shorter than the leaves, gently recurved. Floral bracts conspicuous, deeply concave, ovate, amplexicaul, acute or acuminate, scarious, many-nerved, slightly sur- [ 145 ] EXPLANATION OF THE ILLUSTRATION Prare XII. Brassta Attentr L. O. Wms. ex C. Schweinfurth, 1, plant, natural position, about one half natural size. 2, flower, from front, about five sevenths natural size. 3, column and base of lip, from side, about twice natural size. Drawn by Dorotuy O. ALLEN [ 146 | PuiatTeE XII ~N > aS = Ni WA \ passing the pedicellate ovary, the lower ones 1.9—2.7 cm. long. Flowers rather small for the genus, non-resupinate in natural position (i.e. with the lip uppermost). Dorsal sepal linear-lanceolate, long-attenuate with conduplicate upper portion, about 3.7 cm. long and 5 mm. wide be- low. Lateral sepals similar but apparently a little longer and narrower (sometimes slightly broader), about 4.3 em. long and 4 mm. wide. Petals obliquely and narrowly lanceolate, long-attenuate above with conduplicate mar- gins, about 3.8 cm. long and 6 mm. wide below. Lip ovate or suborbicular-ovate, with the lobulate anterior margins terminating abruptly in a more or less elongate triangular or linear-triangular mucro, subcordate at base, about 1.5—2.1 em. long and 1.2-1.5 em. wide near the base; disc at the base just in front of the column with a pair of short approximate fleshy keels which are deeply constricted in front forming a pair of conical complanate teeth. Column very short and stout, wingless, about 8 mm. high at the back. This species is extremely variable both vegetatively and florally. The measurements of the most complete specimen, which has been taken as the type, are uni- formly somewhat smaller than those of other collections. Brassia Allenii seems to be related to the Venezuelan Brassia glumacea Lindl., but differs in having a much stouter habit, in having the racemes markedly shorter than the leaves and in its apparently dissimilar lip. This concept was first considered as new and was named by Dr. Louis O. Williams, but has remained un- described until this time. Panama: Prov. of Panama, summit of Cerro Campana near Cam- pana, at about 1000 meters altitude, in cloud forest, October 2, 1947, Paul H, Allen 5150 (Tyrerin Herb. Ames No. 64857); Prov. of Coclé, vicinity of El Valle, at 800-1000 meters altitude, sepals and petals dark reddish brown, lip yellow, fragrant, August 17, 1937, P. H. Allen [ 149 ] 374; Prov. of Coclé, mountains beyond La Pintada, at 400-600 meters altitude, February 17, 1935, 4. A. Hunter & P. H. Allen 592 (old fruit); Prov. of Coclé, region north of El Valle de Anton, vicinity of La Mesa, at about 1000 meters altitude, very common in dark wet forest of higher elevations, November 12, 1941, P. H. Allen 2830; same locality, Finca Mufioz, at 800-1000 meters altitude, Feb. 2, 1942, P. H. Allen 2922; same locality, at about 700-800 meters altitude (flowered by Mr. Elmer Ohlson, September 10, 1942), P. H. Allen 2963; Prov. of Coclé, ‘Loma del Tigre,’’ hills north of El Valle de Anton, at 1000 meters altitude, November 16, 1946, P.H. Allen 3804. Lockhartia amoena Endres & Reichb.f. var. tri- angulabia (4. & 8.) C. Schweinfurth & P. H. Allen comb, nov. Lockhartia triangulabia A. & S. Sched. Orch. 8 (1925) 80. In view of the great variability shown by the lip in the numerous collections of Lockhartia amoena, it appears to us that the concept described as Lockhartia triangulabia represents only an extreme form of the earlier species. We therefore propose the above combination. NOMENCLATURAL CHANGES IN VENEZUELAN ORCHIDS In conformity with our present view of regarding the concept Ornithidium Salisb. as being referable to the large and polymorphic genus Maaillaria Ruiz & Pav., the two following transfers become necessary. Maxillaria Jenischiana (Peichb.f.) C. Schweinfurth comb. nov. Ornithidium Jenischianum Reichenbach filius in Bonpl. 2 (1854) 18. Maxillaria sanguinolenta ( Lindl.) C. Schweinfurth comb, nov. Ornithidium sanguinolentum Lindley Orch. Linden. (1846) 22, no. 117. [ 150 | ORNITHOCEPHALUS POLYODON Ornithocephalus polyodon Reichenbach filius in Linnaea 41 (1876) 33. Since this species, originally found in Ecuador, was described from two inflorescences only, it seems wise to append a more detailed description as afforded by a recent Peruvian collection consisting of three complete plants. Plant rather large and stout for the genus, epiphytic, about 11-19 cm. tall. Stems abbreviated as in the genus, entirely concealed by a circle of nine to eleven crowded imbricating sheaths of the leaves. Leaf-sheaths condu- plicate with a scarious band on the upper margin. Leaves elliptic-lanceolate to linear-lanceolate, more or less acu- minate, usually lightly oblique above, somewhat nar- rowed to an obliquely articulated base, about 3 (outer blade) -16 cm. long, 0.85—-1.4 cm. wide. Scapes lateral in the axils of leaf-sheaths, usually two to each plant, suberect or flexuous to arcuate, about 9.5-19.5 cm. long, entirely glandular-setulose (slightly so below), racemose above; peduncle short, 4 em. or more long, with three or more remote, spreading, semiamplexicaul sheaths; raceme up to 14.5 cm. long, loose or subdense, several- (8) to many-flowered. Floral bracts spreading, suborbicular- ovate, semiamplexicaul, glandular-ciliolate. Flowers rather large for the genus, with membranaceous wide- spreading segments. Dorsal sepal deeply concave, sub- orbicular-obovate, broadly rounded at the apex, irregu- larly glandular-ciliate (especially on the lateral margins), rather densely glandular-setose on the outer surface, 1- nerved, about 6 mm. long and 4mm. wide. Lateral sepals oblong-obovate, broadly rounded above, lightly concave, somewhat oblique, sparingly serrate, rather densely glandular-setose without, 1-nerved and dorsally keeled, [ 151 ] about 5.5 mm. long and 8.8 mm. or less wide. Petals cuneate-flabellate with recurved sides in natural position, oblong-obovate when expanded, broadly rounded at the apex, l-nerved, about 5 mm. long, 8.5—-4.2 mm. wide, 2- to 8-nerved. Lip deeply 8-lobed, shorter than the other segments in natural position, with saccate base, erect lat- eral lobes and recurved mid-lobe, when expanded about 6.5 mm. long and slightly wider across the base; lateral lobes small, divaricate, triangular-oblong, with a recurved convex apex; mid-lobe much larger, subquadrate-ovate when expanded, slightly dilated above and below, rather abruptly narrowed above to an acute tip, 7-nerved; disc with a transverse subelliptical pluridentate callus which extends as a toothed ridge from the base of the lateral lobes, continues in a curved line in front of the lateral lobes and is extended across the base of the mid-lobe. Column very small, clavate, with a reflexed linear ros- tellum. In the Ames Herbarium there is a drawing of Ornitho- cephalus polyodon from the Reichenbach Herbarium con- sisting of two flowering racemes and a rather detailed floral analysis. The latter shows a close agreement with the flowers of the Peruvian collection, having similar oblong-obovate sepals with a broadly rounded apex. The petals are shown as sometimes broadly rounded and sometimes (as described) retuse at the apex. The lateral lobes of the lip (as drawn) appear to be subequal in size to the mid-lobe. The type description, as well as the Reichenbach drawing, shows that there is a toothed up- per margin of the petals, a character which does not ap- pear in our specimen. Ecuapor: (Tyre). Peru: Loreto, Rio Mazan, at 100-125 meters altitude, February 1935, José M. Schunke 391. [ 152 ] BOTANICAL MUSEUM LEAFLETS HARVARD UNIVERSITY ANUARY 17, 1949 VoL. 13, No. 7 CAMBRIDG E, MassacuHUseETTs, J MAIZE GRANARIES IN MEXICO BY ErraimM HERNANDEZ XOLOCOTZI ' I. INrRopUCTION Tuts study attempts to follow the development of the maize granary, an important element of material culture, in Mexico where maize has been and still is the basic source of food. Sufficient information is now available to permit the presentation of a tentative outline of this de- velopment from the pre-Conquest period to the present day. Further anthropological and ethnobotanical studies along these lines may result eventually in the establish- ment of significant correlations between granary types and indigenous cultures. Such correlations would be of great value to scientists engaged in plant exploration, in connection with maize research of either a practical or a theoretical aspect, since Indian cultures have acted in part as isolating mechanisms resulting in the production and maintenance of maize ecotypes. In the Western Hemisphere, maize was the cereal which served as the economic basis for civilization. Ar- chaeological studies have brought to light various cul- tures with an incipient agriculture based on plants other than maize (Zingg 1939 and Anderson 1947); however, it was not until after the domestication and spread of 1 Rockefeller Foundation Fellow from Mexican Agricultural Program and Agrénomo **B’’ de la Secretaria de Agricultura y Ganaderia, Mexico. [ 158 | maize that the South and Middle American civilizations were able to flourish. The widespread adaptation of maize agriculture resulted in a vital interdependence between this domesticated cereal and man. After domestication, maize could no longer subsist without the aid of man, and man himself relied more and more upon the product of this crop as his main source of food. This dependence on maize is still an outstanding characteristic of most of the Latin American countries today and of the Indian pop- ulations in particular. It is probable that the problem of maize storage arose with the beginning of maize agriculture. With an in- crease in population resulting from an initial food sur- plus, maize storage took on an even greater importance. As the culture increased in complexity, this problem took on a group as well as an individual significance. By the time of the Conquest of Mexico, attention was de- voted to the maize granaries by the rulers and priests, by the rich and the poor, by the craftsmen and the peasants. The tribal and family granaries represented a savings account which kept away disaster and maintained social vigor. Failure of the family granary meant slavery ; fail- ure of the tribal granaries meant social disintegration. However, in spite of extensive cultural similarities among the Indian populations of Mexico, the develop- ment of the storage structures have followed various lines. Diversification in granary types resulted from: a) differences in climate, b) differences in materials available for construction, ¢) special cultural traits of the society, and d) variations in the cultural level. Il. THe Propuem or MaIzZE STORAGE Maize became a dominant crop in the agricultural economy of the American civilizations because it pos- sessed the favorable characteristics generally found among [ 154 ] the cereals. hese are as follows: 1) the seed is a highly nutritive caryopsis, 2) the production of seed is highly efficient, 3) the seed can be handled and stored with ease, 4) the maize plant lends itself to the use of simple methods of cultivation, and 5) the plant has a wide toler- ance to many and varied ecological conditions. The nature of the seed favored its storage. However, the storage of the maize crop required special attention due to the following reasons: 1) several pests and dis- eases attack the grain, 2) a surplus stock would prevent the calamities of a poor crop, and 8) climatic variations often resulted in considerable decreases in yield. No doubt the chief interest of the people was the pro- duction of an essential minimum amount of grain each year. This is suggested by the complex religious pattern centered around the maize theme, the interest in astron- omy and its relation to the cultivation of maize, and the use of mixed seed of maize lines adapted to various cli- matic conditions. Once the crop had been obtained, it was necessary to protect it until the following harvest since this was periodical. In addition, the demands of a non-agricultural population, and the use of marginal and sub-marginal lands for agriculture due to population pres- sure, all stimulated the development of some method of storage. Thus it is seen that the problem of storage was con- comitant with the cultivation of maize. The method of storage could vary, but under any condition it necessi- tated the development of a special cultural pattern. The term maize granary is here used to designate the material element designed for the solution of this problem. III. ArcHAEOLOGICAL PERIOD (circa 500-1100 A.D.) Anthropological studies show that the Mexican Indian [ 155 ] cultures developed intricate religious patterns at an early stage. With an increase in populations and the flourish- ing of these societies, this tendency led to the establish- ment of a strong theocracy, later aided by rulers in charge of the secular functions of the group. Under these con- ditions, the cultures followed a dichotomous develop- ment; on the one hand, of the ruling classes and their magnificent material culture, and on the other hand, of the ruled masses with a less spectacular but more stable form of material culture which has persisted to date in many areas of Mexico. It was not until recently that archaeologists began to pay attention to the cultural patterns of this larger part of the Indian population. The increased interest in pot- tery types indicates a tendency toward the study of this rich source of information. As it is, the recorded infor- mation on maize granaries from archaeological remains is extremely meager. The perusal of the extensive liter- ature dealing with the impressive ruins in the Mayan area, San Juan ‘Teotihuacan, Monte Alban, Mitla and La Venta, all undoubtedly of great religious importance, adds very little to an understanding of our particular problem. S. Linné (1984) describes what are probably the oldest remains definitely related to maize storage. He writes of the results of his excavations of some of the smaller out-lying ruins of the ancient religious center at San Juan Teotihuacan, State of Mexico: ‘While purely utilitarian pottery was almost entirely absent among the finds that were made below the floors of the ruin, it appears that almost all the vessels just described have served some practical pur- pose or other, e.g. for serving up food. There are also numerous frag- ments of storage vessels. These were generally of ovoidal shape, with a relatively narrow mouth, and from many of the fragments it is evi- dent that the vessels often were of considerable size.’’ As will be seen later, this method of storage was used [ 156 ] by the farmers of Mexico City at the time of the Con- quest (Pl. XVI, A). Archaeological studies in the Tarahumara region of northwestern Mexico, in the States of Sonora and Chi- huahua, give a more complete picture of the methods used in storing maize. According to the works of C. Lumholtz (1902) and R. M. Zingg (1940), these are as follows: 1) underground cavities, 2) niches of caves walled in, 8) rectangular and cylindrical stone and mor- tar structures, 4) wattle and clay daub containers, and 5) vasiform structures made of coiled grass and clay daub. In 1981, Robert M. Zingg carried out extensive studies in the Rio Fuerte region of southern Chihuahua. From the material obtained, he concluded (1940) that it repre- sented a Basket-Maker and a Cave-Dweller cultural phase of the ancient Tfarahumaras. The methods of maize storage used during the Cave-Dweller phase are those included above under numbers one to three. These are of special interest because they are still followed in part by the modern Tarahumaras, and they are not re- ported from any other group in Mexico. During the Basket-Maker phase, Zingg finds “‘large bowl-like containers, 24” in diameter and 8” deep’’ hav- ing a funerary context but probably related also to an element used in the storage of maize. These containers ‘‘made by coiling rolls of grass imbedded in mud _ as binder”’ appear to Zingg as ‘‘a simplification of the coiled mud and grass storage structures of the Cliff Houses of the Casas Grandes archaeological phase of northern Chi- huahua’’ described by Lumholtz (1902). These latter storage structures were found by Lumholtz at Cave Val- ley, close to Pacheco, and at Aros River, south of Chui- cuichupa, both localities in the State of Chihuahua within the Tarahumara region (PI. XIII, A, B). Lumholtz showed them to be maize granaries which he related to [ 157 ] the vasiform granaries (cwezcomatls) found in use in 1902 in the central Mexican States of Tlaxcala and Veracruz. As this type of storage structure is used at the present time in Tlaxcala and Morelos, it seems that an uninter- rupted sequence is present in the use of this type of granary from the archaeological to the present period. Because of the relationship of these forms of granaries, it is of interest to include the description of the struc- tures (Plate XIII) discovered by Lumholtz (1902): ee, he most unique feature of this cave, however, is the cupola- shaped structure which stands in an open space in front of the house group, near the mouth of the cave, but still under its roof. Its height, measured inside, is twelve feet, and its widest inside diameter is eleven feet. Its walls average eight inches in thickness. It has one aperture three feet wide at the top, another one of the same diameter near the base, and there are several others nearly opposite each other. In the two upper ones are seen distinct impressions of timber in the plaster. *“The building was made by twisting long grass into a compact cable and laying it up one round upon another. As the coil proceeded, thick coats of plaster were laid on inside and outside. This plaster, which is the same material as that of which the houses are constructed, got thoroughly mixed with the straw during the process of building, and the entire structure was finished without any opening except the one at the top. The other apertures were undoubtedly cut out afterward. There is no trace of withes or other binding material to hold the straw cables in place. They are kept in position only by the plaster, which here, as in the houses, is almost as hard as the conglomerate of the surrounding rocks.’”’ Further on he adds: ‘*Two of them were deeply sunken into the floor of the cave, and inside of them we found between the rubbish and debris that filled them, several grains of corn and some beans.’’ A study of the findings of Isabel Kelly (1947) at Apat- zingan, Michoacan, along the Pacific coast of Mexico, suggests the presence of remains which may be the foundations of granaries similar to those now in use along the coast of Guerrero. However, a sufficient knowledge is lacking for accurate determination. [ 158 ] SUMMARY—ARCHAEOLOGICAL PERIOD (500-1100 A.D.) Granary Type Indian Group Distribution 1. Subterranean Tarahumara Sonora and Chihuahua 2, Niches of caves Tarahumara’ Son. and Chih. 3. Rectangular and cylindrical, stone Tarahumara_ Son. and Chih. and mortar . Wattle and clay daub Tarahumara’ Son. and Chih. . Vasiform, coiled grass and clay daub T'arahumara’ Son. and Chih. OQ oO . Clay jars Teotihuacan San Juan Teo- tihuacan, Mex. IV. Pre-ConavEst PEriop (1100-1518 A.D.) For a knowledge of the cultures found in Mexico at the time of the Conquest, two extensive sources of in- formation are available: a) the Codicis, and b) the writ- ings of men who lived in Mexico during the few decades following the Conquest. It seems reasonable to assume that these historical documents refer to cultural traits of a well-established pattern which remained unchanged for some time after the Conquest. For this reason, this in- formation is interpreted as a reflection of conditions ex- isting previous to the Conquest. The Codicis are a form of pictorial writing used by the Mexican tribes to record such information as tribal his- tory, rights to land ownership, commercial transactions and tribute lists. The pictorial technique varied widely probably as the result of a constant development from pure pictorial representation to a semi-phonetic writing. Of the Codicis which escaped destruction, the most im- portant ones to our problem are the Codex Mendoza, the Historia Chichimeca, the Yanhuitlan Codex, the Codex 'Troano-Cortesiano and the Lienzo de Tlaxcala. The information gleaned from these sources may be divided into two categories: a) one referring to the stor- age structures used by the rulers, and b) the other re- [ 159 ] ferring to the maize granaries of the farmers and crafts- men. In most cases the same type of granary was used by both classes, but the small family granary types are the ones that have survived and give a distinctive ap- pearance to many of the Indian communities of present- day Mexico. A. Imperial Granaries Bernal Diaz del Castillo (1568) gives an insight into the organization of the palace of Moctezuma when he states: “*Dejemos de hablar de la costa y comida de su casa, y digamos de los mayordomos y tesoreros y despensas y botelleria y de los que tenian cargo de las casas adonde tenian el maiz.”’ Thus we learn that Moctezuma had overseers and ‘*treasurers’’ who were in charge of the ruler’s pantries, ‘“wine’’ rooms and houses where maize was stored. Fr. Bernardino de Sahagun (1529-1590) amplifies this information in his section dealing with the warehouses of the Aztec ruler wherein he writes: “*Otra sala del palacio se llamaba petlacaleo. En este lugar posaba un mayordomo del sefior, que tenia cargo y cuenta de todas las trojes de los mantenimientos de maiz que se guardaba para proveimiento de la ciudad y republica, que cabian a cada dos mil fanegas de maiz, en las cuales habia maiz de veinte afios sin dafiarse; tambien habia otras trojes en que se guardaba mucha cantidad de frijoles.’’ In addition there were other warehouses (trqjes) for salt, peppers and two kinds of pumpkin seeds. As much of the imperial supplies was obtained as tribute exacted from the towns under the Aztec domi- nation, the Codex Mendoza (Kingsborough 1831-1848, Long 1942) is an important source of information. This Codex was made by order of Don Antonio de Mendoza, Viceroy of Nueva Espafa from 1585 to 1550, and was intended for King Charles V. It was painted by Mexican [ 160 | artists and the pictorial technique shows no Spanish in- fluence. It is believed to be a copy of an older Codex made in 1511 to record Moctezuma’s tribute list. Throughout this document, the measure of the tribute to be paid in terms of maize, beans, chia (Salvia hispan- ica) and quautli (4 maranthus caudatus), is indicated as a large wooden granary of the log-cabin type. Other authors writing at a much later period suggest that the following two types of granaries were also used by the Aztec rulers: large cask-like containers made of wattle and clay daub and kept within the rooms of the palace (Torquemada 1723); and cuezcomatls (vasiform granaries) according to Fray Alonso de la Mota y Esco- bar (Simpson 1984). The following four types of imperial granaries may therefore be recognized: 1. Rooms within the palace. This type of storage is described by Diaz del Castillo (1568) and Sahagun (1529- 1590). 2. Log-cabin granary (Pl. XIV, A). In 18 of the plates of the Codex Mendoza (Kingsborough 1831-1848) a total of forty of these figures are illustrated, and in all cases the Spanish text refers to them as ‘‘troxes.’’ All are strikingly similar and seem to represent a cubical structure consisting of low corner posts, a plank floor, log or board walls interrupted at about the middle by a circle, and a plank roof. The representations of various types of grains are drawn above the figures of these structures, and it is evident, from the rest of the infor- mation given, that these granaries were used as measures of the yearly tribute which had to be given to the Aztec ruler. There is no indication regarding the manner in which the walls were held together at the corners, and [ 161 ] it seems best to regard this lack of detail as a simplifica- tion leading to economy of effort on the part of the artist. The color and position of the circle on the front wall sug- gest a small wooden door similar to the one found in log-cabin structures common among the Mixtee Indians today. Anderson and Barlow (1943) made a special study of these figures, and their findings may be summarized as follows: the figures represent large wooden granaries with acapacity of four or five thousand ‘‘fanegas’’ (about 10,000 bushels); these ‘‘troxes’’ were used as units of measure of the yearly tribute required of the various provinces in terms of maize, beans, chia and quautli; these granaries were built at strategic localities through- out the Valley of Mexico and into Oaxaca as far as Oaxaca City. Finally, at a late date Francisco Saverio Clavijero (1780) gives a full description of this type of granary and his description is included since it has been repeated by Prescott (1843), Brasseur de Bourbourg (1858) and Bancroft (1875). ‘*Eras, y Graneros. Tanian eras para deshojar, y desgranar las mazoreas, y graneros para guardar el grano. Estos eran cuadrados, y por lo comun, de madera. Servianse para esto del qjamet/, arbol al- tisimo, de pocas ramas, y estas mui delgadas, de corteza tenue, y lisa, y de contestura flexible, pero dificil de romperse, y rajarse. Formaban el granero, disponiendo en cuadro, unos sobre otros, los troncos re- dondos e iguales de ojametl, sin otra trabazon que una especie de hor- quilla en su estremidad, para ajustarlos, y unirlos tan perfectamente, que no dejasen paso a la luz. Cuando llegaban a cierta altura, los cu- brian con otra trabazon de pinos, y sobre ella construian el techo, para defender el grano de la lluvia. Estos graneros no tenian otra salida que dos solas ventanas, una pequefia en la parte inferior, y otra grande en la superior. Los habia tan espaciosos que podian contener cinco mil, seis mil, y aun mas fanegas de maiz.”’ He goes on to say that some of these structures were [162 J still to be found in places at some distance from the capital (Mexico City) and that they seemed old enough to have been constructed before the Conquest. 3. Cuezcomatls, vasiform grass and daub granaries. In 1609, Fray Alonso de la Mota y Escobar (Simpson 1984) includes the following statement in his memoirs with reference to San Juan Cuezcomatepec, a village 25 kilometers northwest of Cordoba, Veracruz: “*Hablase en este pueblo lengua mexicana perfecta llamase en ella Cuezcomatepec que suena el pueblo de los troxes porque dizen las tenia aqui Molezuma de mucha cantidad de mahiz que por ser tan frio y seco se conseruaua aqui como en deposito para los tiempos de sus hambres.”’ As will be shown later it seems reasonable to assume that the Indian word cuezcomatl refers to the vasiform type of granary. Most students of the Nahuatl] language indicate that the etymology of the word cuezcomatl is not known. From a very ancient period the word has meant a gran- ary and cuezcomatl has been used as a root word in the composition of other words. However, Siméon (1885) writes in his Dictionnaire de la Langue Nahuatl: **Cuezcomatl, s. Havre-sac, magasin de pain; sommet de la téte, crane. En comp. nocuescon, mon crane; tocuezcon, notre crane, le crane en general. Avec les postp. c, tlan: cuezcomac, dans le havre-sac, dans le grenier; cuezcomatitlan, parmi les havre-sacs.”’ While Garcia Cubas (1888-1891) adds that comatl means vasija or container. It seems then that cuezcomatl might mean a container in the form of a skull. A study of Pl. XIII, the vasiform grass and daub granary of the ancient ‘Tarahumaras related to the present-day grana- ries, suggests the possible reason for the use of this word for these structures. 4. Wattle and daub cask-like granaries. TTorquemada (1723) has a chapter in which he relates the looting of [ 168 ] ‘* Moctecuhcuma’s’* warehouses by the Indian allies of Cortes and by Pedro de Alvarado. Of the containers in which the royal supplies were kept, he says: **Kstaba el Cacao en unas Vasijas hechas de Mimbres, tan grandes como Cubas, que seis Hombres no las podian abarcar; estaban em- barradas por de dentro, y por de fuera, y asentadas por orden, como Cubas; servian de Troxes para el Maiz, y otras Semillas se conser- ’ ’ ’ A vaban bien en ellas. ...’’ B. Family Maize Granaries The information recorded about the family maize gran- aries of this period presents for the first time the extreme diversity which is to be expected in this country of nu- merous Indian cultures. a. Permanent and semi-permanent family granary types 1. Niches of caves. The use of this method of maize storage is not reported for this period, but undoubtedly it was in use among the ‘Tarahumara Indians since it has been found among them during the Archaeological Period and the 19th Century. 2. Rectangular and cylindrical stone and mortar struc- tures. The same may be said for this type of granary as has been said for the previous type. 3. Wattle and clay daub. Probably in use among the Tarahumara, as a similar reasoning holds for this type as for Type 1. 4. Cuezcomatl, vasiform grass and daub granary (PI. XIV, B,C, D). Several of the town names included in the Codex Mendoza (Kingsborough 1831-1848) and the Historia Chichimeca (Boban 1891) are combinations utilizing the Indian word ecuezcomatl. As this is the [ 164 ] name used at present for the vasiform granaries in the States of Tlaxcala and Morelos, aspecial study was made of the town glyphs in the Codicis. Those for Cuezco- matly yacac (Cuezcomatl iyacac), Cuezcomayxtlahuacan, Cuezcomahuacan and Cuezcomatitlan all consist of a dominant vasiform figure with the following characteris- tics: a) the structure seems to rest on a rectangular foundation built on top of roundish stone corner sup- ports, b) the vasiform figure is divided by several hori- zontal lines suggesting a structure made of several sec- tions, and ¢) the top of the structure is shown as an open circular mouth except in Cuezcomatlyyacac in which a sort of round lid is indicated. Wherever interpretations of the town glyphs are made, cuezcomatl is translated as trove, granero or granary. For instance, Clark (1988) gives the following meanings in his analysis of the Codex Mendoza: Cuezcomahuacan: place of granaries Cuezcomayxtlahuacan: place of the plain of bins Cuezcomatitlan: amongst the granaries Cuezcomatliyacac: in the beginning of the granaries It seems then that the farmers of the Mexican central plateau at the pre-Conquest period were using a vasiform maize granary which is related to the archaeological re- mains of the Tarahumara and to the present day cuez- comatls of Tlaxcala and Morelos. Sahagun (1529-1590) suggests the use of this type of granary when he writes: “*De algunos zazamiles de los muchachos que usa esta gente mexi- cana, que son los “que cosa y cosa de nuestra lengua’ (Acertijos).? Que cosa y cosa una vieja que tiene los cabellos de heno, y esta cerca de la puerta de casa? Es la troje de maiz.”’ Torquemada (1723) also seems to refer to this type of granary when, in speaking of the custom of punishing [ 165 ] those who stole maize from the granaries, he states that these “‘troxes”’ are like very large vessels (tinqjas) with their mouths on the uppermost part and anyone wishing to take out the contents must climb into it. The Lienzo de Tlaxcala (Anonymous 1892) includes two perfect rep- resentations of these granaries. 5. Clay jars (Pl. XVI, A). This figure reproduced from the illustrations which form part of the work of Sahagun (1529-1590) depicts the cycle of activities of the good farmer. The storage of the harvest in large clay jars is clearly shown. In the northwestern part of Mexico, among the Cahita- speaking Indians, today represented by the modern Yaqui and Mayo of southern Sonora and northern Sina- loa, Beals (1943) has found that during the pre-Conquest period maize sometimes ‘‘was cached underground in pottery vessels stoppered with clay.”’ 6. Log-cabin type of granary. Plate XVI, B, repro- duced from Sahagun (1529-1590), shows the storage of the maize harvest by a Mexican family. The granary consists of stone corner posts, a base made of boards or planks, and walls of logs or boards placed horizontally. Economy of detail on the part of the artist may account for the lack of detail as to the method in which the cor- ners were formed. ‘Two stones shown at the upper corners of the structure probably refer to rocks used to hold down the thin board or wooden shingle roof. The dimensions of the front wall as shown must be about 5 by 8 feet. In spite of the marked differences already noted be- tween the rulers and the lower economic classes in the Mexican Indian cultures, it seems reasonable to assume that the richer agricultural families tended to use the same type of granaries as utilized by the tribal rulers. For | 166 | this reason, the use of the wooden log-cabin type of granary by the family group is also confirmed by the information given in the Codex Mendoza (Kingsborough 1831-1848) and the Yanhuitlan Codex (Jimenez Moreno and Mateos Higuera 1940) which will be discussed later. 7. Cinealli, highland crib. A characteristic crib-like wooden granary known as cincalote is used today in a limited area to the south of Mexico City (Pl. XVIII, A). Sahagun (1529-1590) seems to refer to this type of granary when, in describing the feasts and ceremonies observed by the lapidaries, he states: eé % . . Al otro que se llamaba Cinteotl tambien le componian como a varon, con una caratula labrada como mosaico,.... y ponianale en un tablado alto de donde estaba mirando, el cual se Ilamaba cinealli, compuesto de cafias de maiz a manera de jacal;... .”’ b. ‘Temporary methods of maize storage 1. Rafters. Sahagun (1529-1590), describing the habits and labors of the Mexican farmer born under the favorable sign Ce Tochtli, says that he was diligent, hard- working, careful of his crops and: “*asi abundantemente cogia de todas maneras de legumbres y henchia su casa de todas maneras de maiz, y colgaba por todos los maderos de su casa sarteles de maiz....”’ 2. Tapanco, loft and interior of the house. This method of storage is reported to have been in use among the Cahita Indians (Beals 1943) who inhabited the dry semi-tropical coastal plains of Sonora and Sinaloa. 3. Enramada. Beals (1948) finds that the Cahita In- dians also stored maize on the cob on enramadas which are raised platforms built as separate structures or as an- nexes to the front of the native houses (Pl. X XI, C,D). [ 167 ] With regard to the Mayan groups, the methods of maize storage during this period are unknown. The Co- dex 'Troano-Cortesiano suggests the use of clay jars, but no definite information is available. SUMMARY—PRE-CONQUEST PERIOD (1100-1518 A.D.) Granary Type Indian Group Distribution A. Imperial Granaries 1. Rooms within palace Aztec Valley of Mex. 2, Log-cabin granary Aztec Valley of Mex. Mixtec Mixteca of Oax. 3. Cuezcomatl, vasiform, Aztec Valley of Mex. grass-daub Mazapan Puebla Tlaxcalan Tlaxcala 4, Cask-like, wattle-daub Aztec Valley of Mex. B. Family Granaries a. Permanent and semi-permanent structures 1. Niches of caves Tarahumara Son. and Chih, 2, Rectangular and cylindrical, Tarahumara Son. and Chih. stone and mortar 8. Wattle and clay daub Tarahumara Son. and Chih, 4. Cuezcomatl, vasiform, Tarahumara Chihuahua grass-daub Aztec Valley of Mex. Mazapan Puebla Tlaxcalan Tlaxcala 5. Clay jars Aztec Valley of Mex. Cahita Son. and Sinaloa Mayan (?) Yucatan (?) 6. Log-cabin Aztec Valley of Mex. Mixtec Mixteca of Oax. 7. Cinealli, highland crib Aztec Valley of Mex. b. Temporary methods of storage 1. Rafters Aztec Valley of Mex. 2. Tapanco, loft and interior Cahita Son. and Sin. of house 3. Enramada Cahita Son. and Sin. [ 168 ] V. IMMEDIATE Post-ConaueEst PERIOD (1520-1600 A.D.) Fundamentally the main social change which took place from the pre-Conquest to the Conquest periods consisted of asubstitution of a Spanish theocratic-colonial government for the native theocratic-tribal ruler organi- zation. As far as the Indian population was concerned, one master was substituted for another. The established rhythm of activities continued almost unaltered, in spite of the localized demands for Indian labor to work the mines. Various crops were introduced at this time, but maize continued to be the main source of food, and un- doubtedly the methods of maize storage of the preceding period were also in use at this time by the family groups. The establishment of the Encomienda system of land holdings placed the favored Encomendado in the position of exacting tribute from the Indian groups under his control. With regard to the storage of maize under the Encomienda, the Yanhuitlan Codex contains some val- uable information. This Codex has been described in detail by Jimenez Moreno and Mateos Higuera (1940), and Anderson and Finan (1945) have added some perti- nent data. It shows a large log-cabin type of structure used as a measure for the amount of maize which the people of Yanhuitlan, Oaxaca, were obliged to produce for the Encomendado Don Domingo in order to satisfy what was prescribed in the required valuation made by the Viceroy Don Antonio de Mendoza on October 26, 1548. This structure is very similar to the wooden log- cabin type of granary still in use in this region of the Mixtec Indians (PI. XVII, D). SUMMARY—IMMEDIATE POST-CONQUEST PERIOD (1520-1600 A.D.) In spite of the Conquest, the agricultural Indian population of Mex- [ 169 ] ico continued to use the same types Of maize granaries as those existing before the Conquest. The granary system of the Indian rulers came to an end shortly after the Conquest, but the new Encomienda system began using some of the same methods of maize storage. The only definite information in this regard is that given for the use of the wooden log-cabin type of granary used in the Encomienda of Yan- huitlan, Oaxaca, among the Mixtec Indians. VI. Lare 19th Cenrury PErRIop With regard to the study of maize granaries, the only important change after the political independence of Mexico was the substitution of the Hacienda system for the Encomienda. In general, the Indian population con- tinued to use its characteristic methods of maize storage. However, the prolonged absence of a definite tribal or- ganization and the increase in the non-Indian population resulted in a gradual mixing of granary types, a degen- eration in building techniques, the introduction of foreign elements, and the disappearance of Indian granary types over large areas of Mexico. A. Hacienda Granaries Very little information is available as to the methods of maize storage under the Hacienda system. It is to be expected, however, that these would show a strong for- eign influence. 1. Conical stone and mortar structures. These re- markable granaries, recently photographed by P. C. Mangelsdorf (Pl. XLX, A,B), are found in a small area in the States of Aguascalientes, Zacatecas and Jalisco (I. Kelly, in correspondence). These conical structures have walls of adobe faced with lime mortar and small stones. They have a winding stairway on their outer sur- face leading to an inlet close to the apex. A large rec- tangular door at the base serves for the removal of the [ 170 | grain. Their conical shape gives them a high mechanical efficiency for the storage of grain. It is estimated that the larger structures have a capacity of about eighty metric tons. Their relationship to indigenous or foreign structures is unknown. 2. Cubical stone and mortar warehouses. LL. Hernan- dez Xolocotzi (in correspondence) reports that during this period, the Haciendas of Tlaxcala used warehouses which consisted of masonry walls one meter in thickness forming large rectangular rooms 20 meters in length and 8 by 8 meters on the side, with ventilating holes close to the roof. B. Family Maize Granaries Three men in particular have published extensive and comprehensive descriptions of the maize granaries in Mexico in their reports of archaeological expeditions during the final years of the 19th Century: A. F. Ban- delier (1884), C. Lumholtz (1902) and F. Starr (1894, 1899, 1899-1900, 1901-1903, 1908). a. Permanent and semi-permanent structures 1. Niches of caves. Lumholtz (1902) reports this method of maize storage among the T'arahumara Indians at Yepochic, Chihuahua. 2. Rectangular and cylindrical stone and mortar struc- tures. This type of granary was used among the Tara- humara of Sonora and Chihuahua, and the Huicholes of Nayarit (Lumholtz 1902). Of the Huicholes at Batista, Nayarit, Lumholtz writes: ips - , The corn when shelled is kept in round store-houses made of stone and mud. It is put in from above, and taken out through an opening near the ground, a stone serving as a door for this aperture, which is [171 ] kept tightly closed. The store-houses are very small, as the Huicholes ‘ . 7 P 1 harvest only from four to five fanegas of corn a year.”’ 3. Cuezcomatl, vasiform grass-daub structure. This type of granary (Pl. NV, A), previously described from the ‘Tarahumara area and the Mexican central plateau, seems to be confined during this period to the States of Tlaxcala, Puebla and Veracruz (/). Its apparent absence from Morelos is probably due to lack of information. Lumholtz (1902) compares the archaeological remains of the Tarahumara area with the vasiform granaries which he saw in Tlaxcala and Veracruz. There is some doubt as to its occurrence in Veracruz, however, as it is not mentioned in use in that State in the writings of Ban- delier and Starr, nor is it found there at the present time. Its use in Tlaxcala and Puebla is reported by Bandelier (1884) at Huexotzingo, Santiago Xaltepetlapan and San Simon Tlalnicontla. It is Starr (1899, 1899-1900), how- ever, who gives a complete picture of this type of gran- ary through his descriptions and numerous illustrations. He found this granary near Cholula, Puebla, and throughout Tlaxcala. He writes (1899): “The cencalli in Tlaxcala is of a characteristic form and bears the special name of cuezcomatl. It is a vasiform construction of adobe or of clay which rises to a height of from five to ten feet. ... . Above this base rises the great rounded and hollow body, open above; over this a neat, little, two-pitched roof of thatch serves to shed rain. The cuescomatl is a storehouse or granary for maize on the cob.’’ 4. Log-cabin type of granary. The use of this gen- eral type of granary is reported from an extensive area in Mexico. It is found among the Tarahumara of Chi- 'Fanega: ‘“The modern Spanish fanega has been equated to 1.6 English bushels, which would be slightly over 2 American bushels. For Mexico 2.5 American bushels is a conservative equivalent.’’ (An- derson and Barlow 1943. ) huahua and the Tepehuanes of northwestern Durango (Lumholtz 1902), the farmers of Puebla, who call it cen- calli or house of ripe corn (Bandelier 1884), and the Mix- tecs at Tilantongo (Pl. XVII, A) and Yodocono, Oaxaca (Starr 1899, 1899-1900, 1908). However, several minor variations occur from place to place. Variations to the general type are as follows: a) the roof circular in outline; among the Triquis of western Oaxaca, in the Districts of Tlaxiaco and Jux- tlahuacan (Starr 1899, 1899-1900). b) the roof on independent pillar posts; among the Zoques at Tuxtla Gutierrez, Chiapas (Starr 1899-1900). c) the whole structure on tall corner pillars (Pl. XVII, B); among the Mixes of Oaxaca, in the towns of Ayutla, Quesaltepec, and Ixcuintepec (Starr 1899, 1899-1900, 1908). d) the whole structure on tall corner pillars and the thatched roof with ridge pole extending into the air (PI. XVII, C); among the Mazatecs of Oaxaca, Districts of Cuicatlan and Teotitlan (Starr 1899-1900), and the Cuicatecs and Chinantecs of Oaxaca (Starr 1899-1900). 5. Cincalli, highland crib. Cincalli means house of maize in Nahuatl. Cincalote, an evident corruption of cincalli, is the name used by the Otomi Indians living just south of Mexico City for their crib-like granaries. As early as 1899, Starr illustrated these characteristic structures from Huixquilucan, State of Mexico (PI. XVIII, A). 6. Cylindrical granary of upright poles and circular roof (PI. XX, A). This type is reported among the Tri- quis of Oaxaca (Starr 1899-1900). [ 173 ] 7. Maguey hut type of granary. This structure, made almost entirely from the various parts of the maguey or century plant, was found in use as a maize granary among the farmers of southern Puebla and northern Oaxaca in a markedly semi-arid area (Bandelier 1884). 8. Cylindrical or square open granaries with upright poles. This type of semi-permanent structure is reported in use among the Tepehuanes of Durango (Lumholtz 1902). b. ‘Temporary methods of storage 1. Tapanco, loft and interior of house. It seems that the following Indian groups used this method of storing maize, as no other type of granary is known among them; Tarascans of Michoacan; Zapotecs, Huaves and Chon- tales of Oaxaca; and Totonacs of northern Veracruz. 2. Conical container made of maize stalks (Pl. XXII, A). While describing the granaries found in Puebla, Bandelier (1884) writes: “‘In the Plain, I have also seen storehouses made of cornstalks, set vertically, and tied to an inner frame, or forming a conical hollow stack. But such frail structures are temporary, and mostly used for maize only.”’ SUMMARY—LATE 19th CENTURY PERIOD Granary Type Indian Group Distribution A. Hacienda Granaries 1. Conical, adobe and mortar Aguas., Zac., Jal. 2, Cubical, stone and mortar Tlaxcala B. Family Maize Granaries a. Permanent and semi-permanent structures 1. Niches of caves Tarahumara Son. and Chih. [ 174 ] 2. Rectangular and cylindrical, Tarahumara ; stone and mortar Huicholes 3. Cuezcomatl, vasiform, grass- Tlaxcalan daub 4, Log-cabin a) circular roof b)roof on independent pillars c) granary on tall corner pillars d)granary on tall pillars with projecting ridge pole 5. Cincalli, highland crib 6. Completely cylindrical 7. Maguey hut 8. Cylindrical or square open structures b. Temporary structures 1. Tapanco, loft and interior of house 2. Conical container, maize stalks VII. Mopern PErRIopD Cholulan Tarahumara Tepehuanes Cholulan Mixtecs Triquis Zoques Mixes Mazatecs ;Cuica- tecs ;Chinantecs Otomi Triquis ? Tepehuanes Tarascan Zapotecs ; Huaves; Chontoles Totonacs Cholulan Son. and Chih. Nayarit Tlaxcala Puebla Veracruz (?) Chihuahua n.w. Durango Puebla n.w. Oaxaca western Oaxaca Chiapas central Oaxaca northern Oaxaca s. Mexico State w. Oaxaca s. Pue., n. Oax. Durango Michoacan eastern Oaxaca northern Veracruz Puebla Many writers have dealt with the anthropological as- pects of the various Indian cultures of present-day Mex- ico, and yet, the published material on maize granaries and methods of storing maize is very meager. The author has obtained extensive information with regard to these problems while engaged in plant exploration work in connection with the plant breeding program of the Ofi- cina de Estudios Especiales of the Mexican Department [175 ] of Agriculture. The greater amount of this data has been obtained from the States of Tabasco, Veracruz, Chiapas, Oaxaca, Guerrero, Morelos, Tlaxcala and Puebla. It is included in this section in its proper place. A point of special interest is the fact that not until this period is reached does the method of maize storage in the extensive area inhabited by the Mayan groups become known. This condition may be the result of the following factors: a) the lack of stimulus in this area to develop permanent granaries due to the possibility of obtaining more than one harvest of maize during the year and of replacing a shortage of maize with sweet potatoes and cassava, b) a lack of remains due to the perishable nature of the structures built for storage, and c) the tendency of archaeologists to concentrate their attention on the more spectacular elements of the Mayan civilization. A. Permanent Hacienda Granaries 1. Cylindrical masonry structure. Eizi Matuda (in correspondence) writes that in the coffee Finca area of southern Chiapas large permanent granaries are neces- sary to store the supply of maize required to supply the needs of the large Indian population hired during the coffee harvest. ‘These granaries consist of cylindrical structures made of mortar and stone. Usually a suitable slope is cut away leaving two terraces. The granary, having dimensions of 1.5 to 2 meters in diameter and 5 to 6 meters in height, is built against the intermediate wall of the terraces. This location facilitates introduction of the grain through an opening at the upper level and removal through a door located at the lower level or terrace. [ 176 ] B. Family Maize Granaries a. Permanent and semi-permanent structures 1. Cylindrical stone and daub. The studies by Ben- nett and Zingg (1935) among the Tarahumaras of north- west Mexico include full descriptions of the maize gran- aries and the technique of construction. They write: **The cornhouses of the gorges appear to be a hold-over of the more ancient type of structure made of stone-in-mud-mortar. This type is seen in the highlands only in archaeological sites. But the corn- houses of the barrancas are not situated in caves. They usually stand upon large bowlders, several near a dwelling. ... With a bowlder for a base, a circular wall of stone-in-mud-mortar is erected. It is about 4 feet in diameter and 5 feet high. Some have rather flattened fronts which contain the door.’’ 2. Cuezcomatl, vasiform grass and daub. ‘Two varia- tions of this type are found and they are here designated as the Tlaxcalan and the Morelos forms; however, they represent a single phase of cultural development. The Tlaxcalan cuezcomatl (Pl. XV, B) is a vasiform structure consisting of thick walls made of grass and daub and the whole capped by a sloping flat or two-pitched roof of woven grass or of pine shingles. The entrance is through a rectangular cut made on the rim of the wall at the mouth, just below the high pitch of the roof. Maize on the cob is stored in these granaries. The Morelos cuezcomatl (Pl. XV, C) is very similar except that the upper part of the structure and the root are made of overlapping layers of grass, with an inverted clay pot often topping the apex (Linné 1988). Entrance is through a small semi-circular gable-like opening near the top. Shelled maize is stored in these granaries and removal is through a small hole close to the base. Both of these highly specialized forms of granaries are dis- appearing rapidly. [177 ] Luis Hernandez XNolocotzi, owner of one of the few remaining 'Tlaxcalan cvezcomatis, writes (in correspond- ence) that these structures are made as follows: first, a solid stone and mortar foundation is built half a meter below the ground and a half meter above; then, a good supply is made of loaf-like units, each about one meter in length, consisting of long grass stems (Z/yonurus eili- aris H BK) well kneaded in clay; these units are placed, while still partly moist, in an overlapping fashion around and around to form the thick circular walls of the gran- ary; finally, the wall is faced on both sides with a heavy coating of daub. He adds that formerly coiled grass ‘ables were used to construct small cuezcomatls. There is no doubt, therefore, that a positive relationship exists between the archaeological granaries of the 'arahumaras, the euezcomatls of the Codicis, the 19th Century reports and the modern structures of Tlaxcala and Morelos. 3. Highland log-cabin. This type of granary is found in use among the ‘Tarahumaras of northwest Mexico and the Mixtecs of northwest Oaxaca. Bennett and Zingg (1985) give the following description of this structure: ‘‘The most compact and carefully made structures of the Tara- humaras are the storehouses for grain, which may be as large as 10 feet high and 8 feet square. The walls are made of well-hewn boards notched to fit very closely. The foundation of these structures is usu- ally made of beams supported by four stones to prevent moisture from entering the interior. ... Above these floor boards, carefully notched and fitted boards are placed on edge to form the side walls in log- cabin fashion. ... Above the walls is a ceiling (pola, Tara.), made of close-fitting boards like those used for the floor. ... Very clever lock- . ing devices are used to prevent easy access to the interior and robbery.’’ The Mixtec form found near Nochistlan is very similar to the one described above. It adds a palm-thatched two pitched roof, thus providing a loft which is also utilized for storage. 4. Yanhuitlan log-cabin. This type (Pl. XVII, D) is treated separately because of its possible relationship to the pictorial representation found in the Yanhuitlan Codex. Only one of these structures was found in Yan- huitlan, Oaxaca. The walls are made of heavy pine boards with interlocking corners. The roof is of mortar and brick. A small rectangular door with a trick locking device is located high on the front wall. 5. Lowland log-cabin. Several granaries of the log- cabin type have been found at Zumpango del Rio, in central Guerrero and at Real de Guadelupe, La Parota and El] Anonal in the southwestern part of Guerrero. It is the common type of granary at Zumpango where it is made of zoyate palm trunks (Brahea sp.), has a palm- thatched roof and les close to the ground. In south- western Guerrero, this type is found only occasionally. Here it is usually built on pillars of medium height and with a roof supported on independent posts. 6. Cincalli, highland crib. The present distribution of this granary, previously mentioned by Starr (1899) among the Otomi Indians, is as follows: Federal District (Isabel Kelly, in correspondence), southern Mexico State and northern Morelos at Tres Cumbres. The amount of maize harvested by a family determines the number and size of the granaries built. As the maize is consumed, the horizontal poles of the walls are taken down as far as the new level, and in this manner, by the time of the new harvest, the entire structure is practically torn down. A new structure is built each year perhaps as a preven- tive measure against the concentration of insects. Red- field (1980) gives the Nahuatl name as zincolohth, and Starr (1899) gives the Otomi name as cincalote. [179 | 7. Completely cylindrical, upright poles. This is the dominant type among the Amuzgos of southeastern Guerrero. It is very similar to the structures found by Starr (1899) among the Triquis directly north of the Amuzgo area. A. Perez Toro (1922) suggests that these circular constructions indicate a cultural influence of the negro population; however, it is clearly of local origin. This type (Pl. XX, B) is reported also from Tlacotepec, Guerrero (Weitlander and Barlow 1944). 8. Maguey hut (Pl. XX, C). This structure is built much the same as a small scale model of a regular hut with a floor slightly raised above the ground and a two- pitched roof. The adaptation to the local semi-arid con- ditions of southern Puebla where it is used is complete, as it is built almost entirely of the different parts of the maguey plant (Agave spp. ). 9. Trqe, wattle and daub sealed structure (Pl. X XI, A, B). This type of granary has its main area of distri- bution in the State of Guerrero (entire Balsas River basin, Zumpango del Rio, and towns of ‘‘la Sierra’’), although in former periods it was used over large parts of the tropical region along the Pacific coast. It is called trge in the lowland plains, and ewescumal in the upland towns. It is the most specialized and efficient type of granary found in the entire tropical region of Mexico, and its use has been extended to the storage of other crops such as sesame. Great variation exists as to shape, size and location of the trae, but the fundamental fea- tures of its construction are constant. Hendrichs (1945) gives a complete illustrated descrip- tion of these structures in his arresting and comprehen- sive study of the region formerly inhabited by the Cui- clatec Indians. The main features of the granary are as [ 180 | follows: the shape may be cylindrical or cubical and it may have a roof which rests upon it or upon independent supports, but in any event, it consists of thick wattle and daub walls, floor and top so that it forms a perfectly sealed unit; the body of the granary is raised some dis- tance from the ground level; the shelled maize is poured in through a small door cut through the top wall and sealed in after all of the harvest has been stored; removal is made periodically through a small hole close to the base of the structure and, after each withdrawal, this aperture is sealed in once more with daub. 10. Cylindrical, wattle and daub open granary. This simpler type of structure is found distributed along the northern limits of Guerrero, in the Mixtec area of east- ern Guerrero (Schultze Jena 1988), the northwest region of Oaxaca (I. Kelly, in correspondence), and the south- western part of Puebla. Throughout this area it is des- ignated as cuescumal, cuescomate and coscomate. 11. Tropical crib. The use of the tropical maize crib, shown in Pl. XVIII, B, C, predominates throughout most of the moist tropical areas of southern Veracruz, Tabasco, northern and southern Chiapas, Campeche, Yucatan, Quintana Roo and eastern Oaxaca. This struc- ture has been amply described in its modern form by Lundell (1933), Wauchope (1988), Perez Toro (1942) and Morley (1946), from the Mayan area of the Yucatan Peninsula. E. Matuda (in correspondence) reports its use in southern Chiapas. I. Kelly (in correspondence) finds it aberrant among the Totonacs of northern Veracruz, while the author has seen it widely used in ‘Tabasco and in the Chimalapa region of the Isthmus of Tehuantepec, Oaxaca. This granary may be built within the yard of the house [ 181 ] ‘ or at the location of the ‘‘milpa’’ or maize planting. It is used to store maize in the husk and this may be thrown in haphazardly or carefully arranged in layers with the ears in a vertical or horizontal position. 12. Granary within the house. This method of stor- age has been found in use among the Totonacs (I. Kelly, in correspondence), the Maya of Tabasco, and the Zap- otecs of the Isthmus of Tehuantepec, Oaxaca. b. Temporary methods of maize storage 1. Tapanco, loft of house. This simple and effective method of storage is widely used throughout the tropical region. It is found among the Tarascans of Michoacan (DD. Brand, in correspondence, G. Foster 1948), the Zap- otecs of the Isthmus of Tehuantepec, Oaxaca, the Toto- nacs of northern Veracruz (I. Kelly, in correspondence), the Mayan groups of southern Mexico, and the farmers of the coasts of Guerrero and Jalisco. 2. EHnramada. Throughout a large part of the coastal areas of Guerrero and Veracruz, where dry winter seasons prevail, the enramada serves for the storage of maize on the cob (Pl. X XI, C, D). 3. Ziricua, cylindrical reed container (P]. X XII, B,C). Some variation occurs as to the kind of material used in building these containers. The most outstanding exam- ples are found in Guerrero where they are called ziricua and huatlal. In the Balsas River Basin, the zi7ricua is usually made of stalks of Tithonia sp. (acahual, Hen- drichs 1945). On the mountain slopes of Guerrero where Nahuatl-speaking groups have invaded the area, the huatlal is made of split bamboo canes (Hendrichs 1945). This type of granary is also found at Tixtla, Guerrero, among the Mixtecs at Cahuatachi, Guerrero (Schultze [ 182 ] Jena 1938), and among the Amuzgo Indians of south- eastern Guerrero. The Nahuatl-speaking groups of Morelos use a similar structure called ohuatlapilli (Red- field 1980) made of cornstalks. 4. Non-specialized storage structures. Finally, due to the economic level of the population or to the breakdown of former patterns, several methods of maize storage which do not involve the use of specialized storage struc- tures have been adapted in various regions. Some of these methods are: enclosures within the house, enclosures against the walls of the patio, sacks or mats, and stringers of maize ears hung on trees. SUMMARY—MODERN PERIOD Granary Type Indian Group Distribution A. Permanent Hacienda Granaries 1. Cylindrical, masonry southern Chiapas B. Family Maize Granaries a. Permanent and semi-permanent structures 1, Cylindrical, stone-daub Tarahumara Son. and Chih. 2, Cuezcomatl, vasiform, grass- Tlaxcalan Tlaxcala daub Nahuatl (?) northern Morelos 38. Highland log-cabin Tarahumara ; Son. and Chih. Mixtec n.w. Oaxaca 4, Yanhuitlan log-cabin Mixtec Yanhuitlan, Oaxaca 5. Lowland log-cabin ? central and s.w. Guerrero 6. Cincalli, highland crib, Otomi (?) s. Mex. State cincalote, zincolohtli Nahuatl (?) n. Morelos, D.F. 7. Completely cylindrical, Amuzgo s.e. Guerrero upright poles p s. Guerrero 8. Maguey hut ? s.w. Puebla 9. Troje, wattle-daub sealed Cuiclatec (?) w. Guerrero structure Nahuatl (?) n.w. Guerrero [ 183 ] 10. Cylindrical, wattle-daub Mixtec e. and n. Guerrero, open structure ? n.w. Oax., s.w. Puebla 11. Tropical crib Mayan Yucatan Peninsula, Tabasco, Chiapas, s. Veracruz, Oaxaca Totonac n. Veracruz 12. Granary within house Mayan Tabasco, Totonac n. Veracruz Zapotec e. Oaxaca b. Temporary methods of maize storage 1. Tapanco, loft of house ‘Tarascan Michoacan Zapotec e. Oaxaca Totonac n, Veracruz Mayan southern Mexico t Guerrero, Jalisco 2, Enramada rg Guerrero, Veracruz 8. Ziricua, cylindrical reed Cuiclatec (?) w. Guerrero, Nahuatl (?) n. Gro., n. Morelos, Amuzgo s.e. Guerrero 4, Non-specialized struc- ? Jalisco, Guerrero, tures Oaxaca, etc. VILL. ResuMEN El objecto del presente estudio es el de seguir el des- arrollo de los graneros para maiz durante la historia de México, en este pais adonde el maiz ha sido y sigue si- endo la base principal del alimento humano. Como ya existen a la presente fecha suficientes datos para poder presentar un esquema preliminar de este desarrollo, es de esperarse que en un futuro no muy lejano se puedan establecer correlaciones significativas entre los tipos de graneros y las culturas indigenas de México. Estas cor- relaciones serian de gran utilidad para los investigadores agricolas puesto que las culturas indigenas han actuado en parte como mecanismos aislantes favorables para la produccion y conservacion de ecotipos de maiz. Con la introduccién de este cereal a la economia agri- [ 184 ] TENTATIVE CULTURAL AND TEMPORAL DISTRIBUTION OF MAIZE GRANARY TYPES TABLE I IN MEXICO Granary Types Archaeological Period (500-1100 A.D.) Pre-Conquest Period (1100-1518 A.D.) Post-Conquest Period (1520-1600 A.D.) 19th Century Period Modern Period A. Large Permanent Granaries a. Imperial Granaries 1 1 z 3 ~] 16. ij. . Rooms . Log-cabin . Cuezcomail, vasiform . Cask-like, — wattle-daub b. Encomienda Granaries . Log-cabin c. Hacienda Granaries . Conical, adobe-mortar . Cubical, masonry . Cylindrical, masonry B. Family Granaries a. Permanent and semi-permanent . Subterranean . Niches of caves . Rectangular and cylin- drical, stone and mortar 4, Wattle-clay daub . Cuezcomatl, vasiform | . Clay jars . Highland log-eabin a) roof round b) roof on inde- pendent posts ¢) cabin on posts —d) cabin on posts and projecting ridge pole . Yanhuitlan log-cabin . Lowland log-cabin . Cincalli, highland crib . Completely cylindrical . Square, upright poles . Maguey hut . Troje, daub-wattle, sealed . Cylindrical wattle-daub, open top Tropical crib Granary within house b. Temporary and non-specialized . Rafters . Tapanco, loft . Enramada . Conical, corn-stalks . Ziricua, reed . Non-specialized Tarahumara Tarahumara Tarahumara ‘Tarahumara Tarahumara Teotihuacan Aztec Aztec ; Mixtec Aztec; Mazapan; Tlaxcalan Aztec Tarahumara Tarahumara Tarahumara Tarahumara Aztec Tlaxcalan Mazapan Aztec ; Cahita ; Mayan ? Aztec Mixtec Aztec ? Aztec Cahita Cahita Mixtec Tarahumara Tarahumara Tarahumara Nahuatl Tlaxcalan Mazapan Mixtec Mixtec Spanish ? Spanish ? Tarahumara Tarahumara Huicholes Nahuatl ? Tlaxcalan Cholulan ? Mixtec Tarahumara Tepehuan Cholulan ? Triquis Zoques Mixes Mazatecs Cuicatecs Chinantecs Nahuatl Otomi Triquis Tepehuan ? Puebla Chontales Tarascan Zapotec Totonac Huaves Cholulan? Non-Indian Tarahumara Nahuatl ? Tlaxcalan Mixtee Tarahumara Tepehuan Cholulan ? Mixtec ? Guerrero Nahuatl ? Otomi Amuzgos ? Puebla Cuiclatee ? Nahuatl ? Mixtec ? n.w. Oaxaca s.w. Puebla Mayan Totonac Totonac Mayan Zapotec Mayan Tarascan Zapotec Totonac ?, Guerrero Veracruz Cuiclatec ? Nahuatl ? Amuzgo ? Jalisco Guerrero Oaxaca, etc. It has not been possible to establish definite cultural determinations in every case. This fact accounts for the inclusion of such terms as Tlaxcalan and Cholulan, and for the shift from the term Aztec during the Pre-Conquest period to Nahuatl for the later periods. In the column under the heading Modern Period, wherever it has not been possible to fix the cultural element to a specific Indian group, the distribution is given. cola de los grupos indigenas de México, estos iniciaron su €poca de florecimiento cultural. Pero el cultivo del maiz traia en si involucrado el problema de su almacen- amiento. Apesar de las afinidades entre los numerosos grupos indigenas, las construcciones que se desarrollaron para solucionar este problema siguieron una evalucién variada que dié como resultado el sin nimero de formas de graneros que hoy se encuentran en las distintas regi- ones rurales de México. La informacion sobre los graneros de maiz en México, obtenida de la literatura y las observaciones del autor, se ha dividido en cuatro épocas como sigue: I. Periodo Arqueologico (circa 500-1100 AD); IIL. Periodo Ante- rior a la Conquista (1100-1518 AD); III. Periodo In- mediato a la Conquista (1520-1600 AD); IV. Periodo del Siglo XIX; y V. Periodo Moderno. En cada época, se han dividido los diferentes tipos entre los comunales (tales como los del Imperio Azteca, de las Encomiendas y de las Haciendas) y los familiares (permentes y tem- porales). Adonde ha sido posible, se ha sefialado bajo cada tipo de construccién la regi6n adonde se encuentra y los grupos indigenas que lo utilizan. Los resimenes bajo cada periodo indican la distribucién cultural y geo- grafica de los varios tipos de graneros, mientras que la Tabla I presenta, en forma preliminar, la distribucién cultural y temporal. Los principales tipos de graneros familiares son como sigue: a) los cuezcomatls de Tlaxcala y Morelos; b) las cabanas de la Mixteca; c) el cincalli o cincalote de la zona elevada al sur del Estado de México; d) la traje de la region tropical de la Cuenca del Rio Balsas, Guerrero ; e) el jacal tropical de la zona Maya del sur de México; f) la zivicua de la region tropical de la Cuenca del Rio Balsas, Guerrero; g) el tapanco y la enramada de las partes tropicales de ambas costas. [ 185 ] BIBLIOGRAPHY Anderson, E. 1947. Corn before Columbus. Pioneer Hi-Bred Corn Co., Des Moines, Iowa. Anderson, E., and Barlow, R. H. 1943. The Maize Tribute of Moc- tezuma’s Empire. Annals Missouri Bot. Garden 30 (4): 413-420. Anderson, E., and Finan, J. J. 1945. Maize in the Yanhuitlan Codex. Annals Missouri Bot. Garden 33 (3): 861-368. Anonymous. 1892. Lienzo de Tlaxcala, (Published by A. Chavero.) Bancroft, H. H. 1875. The Native Races of the Pacific States of North America. Bandelier, A. F. 1884. Report of an Archaeological Tour in Mexico in 1881. Papers, Archaeo, Inst. Am., Am. Ser. II. Beals, R. L. 1943. The Aboriginal Culture of the Cahita Indians. Ibero-Americana 19: 1-86, pls. I-III. Bennett, W.C., and Zingg, R.M. 1935. The Tarahumara, an Indian Tribe of Northern Mexico. Univ. Chicago Press, Chicago, III. Boban, E. 1891. Documents pour servir A l’historie du Méxique, Cata- loque raisonné de la collection de M. E.-Eugéne Goupil. Paris, Ernest Leroux, printer. Brasseur de Bourbourg, L’Abbé. 1858. Historie des Nations Civili- sées du Mexique et de 1’ Amérique-Centrale. Clark, J.C. 1938. Codex Mendoza. Clavijero, F. S. 1780. Historia Antigua de Mexico. (Spanish trans- lation from the Italian by José J. de Moro, pub. 1826.) Cubas, G. 1888-1891, Diccionario Geogrdafico, Histérico y Biografico de los Estados Unidos Mexicanos. Secretaria de Fomento, Mexico. Diaz del Castillo, B. 1568. Historia Verdadera de la Conquista de la Nueva Espafia. Editorial Pedro Robredo, Mexico, D.F. 1939. Foster, G. M. 1948. Empire’s Children, the People of Tzintzuntzan. Smithsonian Institution, Inst. Soc. Antero. Publ. no. 6. [ 186 ] Hendrichs Perez, P. R. 1945. Por Tierras Ignotas. Editorial Cultura, Mexico, D. F. Jimenez Moreno, W., and Mateos Higuera, S. 1940. Cédice de Yan- huitlan, Mexico. Museo Nac., Inst. Nac. de Antropol. e Hist., Mexico, D. F. Kelly, I. 1947. Excavations at Apatzingan, Mexico. Viking Fund Publs. in Anthrop. no. 7. Kingsborough, Lord. 1831-1848. Mexican Antiquities. Linné, S. 1934. Archaeological Researches at Teotihuacan, Mexico. The Ethnographic Museum of Sweden, N.S., publ. 1. Linné, S. 1938. American roof-apex caps of clay. Ethnos 3 (1): 18- 82, Stockholm. Long, R. C. 1942. Codex Mendoza. Lumbholtz, C. 1902. Unknown Mexico. Charles Scribner’s Sons, New York. Lundell, C. L. 19338. The Agriculture of the Maya. Southwest Re- view 19: 65-77. Morley, G.S. 1946. The Ancient Maya. Stanford Univ. Press, Palo Alto, Calif. Perez Toro, A. 1922. Influencia de la raza negra en Mexico. Ethnos 1: 215-219 (Mexico). Perez Toro, A. 1942. La Milpa. Publs. del Gobierno de Yucatan. Prescott, W. H. 1843. Conquest of Mexico. Redfield, R. 1930. Tepoztlan; a Mexican village. Univ. Chicago Press. Sahagun, Fr. B. de. 1529-1590. Historia General de las Cosas de Nueva Espafia. Editorial P. Robredo, Mexico, D. F. (Publ. in 1938). Schultze Jena, L. 1938. Bei den Azteken, Mixteken und Tlapaneken der Sierra Madre del Sur von Mexico in Indiana v. 3(Jena, Germany). Siméon, R. 1885. Dictionnaire de la Langue Nahuatl ou Mexicaine Imprimérie Nationale, Paris. [ 187 ] Simpson, L. B. 1934. Studies in the Administration of the Indians in New Spain. I. The Laws of Burgos of 1512. II. The Civil Congre- gation. Ibero-Americana 7: 1-129. Starr, F. 1894. Notes on Mexican Archaeology. Chicago Univ., Dept. Anthrop., Bull. 1: 3-16. Starr, F. 1899. Indians of Southern Mexico, an Ethnographic Album. Chicago. Starr, F. 1899-1900. Notes upon Ethnography of Southern Mexico. Davenport Acad. Nat. Scien. 8: 102-198, figs. 1-72. Starr, F. 1901-1903. Notes upon the Ethnography of Southern Mex- ico, expedition of 1901. Davenport Acad. Nat. Scien. 9: 63-172. Starr, F. 1908. In Indian Mexico. Forbes & Co., Chicago, III. Torquemada, Fr. J. de. 1723. Los Veinte iun Libros Rituales i Mon- archia Indiana, Wauchope, R. 1938. Modern Maya Houses, Carnegie Institution of Washington Publ. 502: 133-134, fig. 49g, pl. 37¢. Weitlander, R. J., and Barlow, R. H. 1944, Expeditions in Western Guerrero. Tlalocan I (3, 4): 364-375. Zingg, R. M. 1939. A Reconstruction of Uto-Aztecan History. Univ. Denver, Contr. to Ethnography II: 1-274. Zingg, R. M. 1940. Report on Archaeology of Southern Chihuahua. Univ. Denver, Contr. to Ethnography III. [ 188 ] EXPLANATION OF THE ILLUSTRATIONS Piate XIII. A. “‘Casas Grandes’’ archaeological remains of vasiform type of maize granary made by twisting long grass into a compact cable and laying it up one round upon another and then covering the wall with thick coats of plaster on in- side and outside; Tarahumara Indians of Sonora and Chihuahua. B. ‘‘Casas Grandes’’ archaeologi- cal remains of vasiform type of maize granary ; Tarahumara Indians of Sonora and Chihuahua. (A and B reproduced from C. Lumholtz 1902.) Pirate XIV. A. Imperial granaries of the log-cabin type depicted in the Codex Mendoza as units of measure of the tribute required of the various con- quered towns by the Aztec rulers during the pre- Conquest Period ; figs. 21 and 22 show two of these structures, with representations of maize, beans and quautli (Amaranthus caudatus) seeds above ; town glyph number 6 represents Cuezcomahuacan. B. Cuezcomatl, vasiform grass and daub granary used by the agricultural families of the Aztec Indians during the pre-Conquest period. The glyph of the conquered town number 11 shows this type of structure and signifies Cuezcomatlyyacac or ‘‘in the beginning of the granaries.’’ C. Cuezcomati, vasi- form grass and daub granary depicted in the glyph of Cuezcomayxtlahuacan, conquered town number 19 and interpreted as “‘the place of the plain of bins.’’ (A, B and C reproduced from the Codex Mendoza, Kingsborough 1831-1848.) D. Cuezco- matl, vasiform granary used by the Aztec Indians during the pre-Conquest period as shown in the Codicis; the glyph corresponds to the town of Cuezcomatl Iyacac. (Reproduced from the Historia Chichimeca, Boban 1891.) [ 189 ] Pirate XV. A. Cuescomatils, vasiform maize gran- aries used by the farmers of Tlaxcala during the later part of the 19th Century. (Reproduced from Starr 1899.) B. Tlaxcalan cuescomatl, vasiform type of maize granary used at present in decreasing numbers among the farmers of Tlaxcala; found at San Bernabe Amaxac de Guerrero, Tlax. C. Morelos cuezcomat/, vasiform granary used to store shelled maize ; entrance is through the small gable- like opening near the top; the apex is usually ter- minated by an inverted clay pot. Pirate XVI. A. storage of maize in large clay jars among the Aztec Indians during the pre-Conquest period. B. Highland log-cabin type of granary used by the Aztec Indians during the pre-Conquest period; note that the ears of maize shown are very similar to the Mexican pyramidal race which is still dominant in the area near Mexico City today. (A and B reproduced from Sahagun 1529-1590.) Piate XVII. A. Log-cabin type of family granary used by the Mixtec Indians of Tilantongo, Oaxaca during the later part of the 19th Century. B. Log- cabin type of family granary used by the Mixe Indians of Oaxaca during the later part of the 19th Century ; the entire structure is built on four cor- ner pillars of medium height. C. Log-cabin type of family granary used by the Mazatec Indians of Huautla, Oaxaca during the later part of the 19th Century; the entire structure is built on tall cor- ner pillars, and the projecting ridge poles of the roofs are very characteristic. (A and C reproduced from Starr 1908; B from Starr 1899.) D. Yan- huitlan log-cabin type of family granary very sim- ilar to the ones depicted in the Yanhuitlan Codex; found rarely at present among the Mixtec Indians of Yanhuitlan, Oaxaca. [ 190 ] Pirate XVIII. A. Cincalli, cincalote, highland crib- type of maize granary used by the Otomi Indians of Huixquilucan, Mexico, during the later part of the 19th Century ; the number and size of the struc- tures varies with the amount of maize harvested by the family during the year. (Reproduced from Starr 1899.) B. Tropical crib, dominant type of maize granary used at present over an extensive area including southern Veracruz, Tabasco, Chia- pas, Oaxaca, and all of the Yucatan Peninsula; Mayan Indians. (Photographed by Dr. A.J.Sharp, at Finca La Gloria, Oaxaca.) C. Crib-type of maize granary, variant form made of sturdy palm trunks, found close to Veracruz City, Veracruz. (Photo- graph by Dr. P. C. Mangelsdorf. ) PLatre XIX. A. Hacienda granaries, conical adobe- mortar type found over a small area of Zacatecas, Aguascalientes and eastern Jalisco; photograph shows the spiral stairways leading to the aperture near the apex. B. Hacienda granary, conical adobe- mortar structure showing the outlet door at the base. (Photographs by Dr. P. C. Mangelsdorf. ) Pirate XX. A. Completely circular structure used for storage of maize by the Triqui Indians of Oaxaca during the later part of the 19th Century. (Repro- duced from Starr 1899.) B. Completely circular maize granary used by the Amuzgo Indians of southeastern Guerrero; Modern Period. C. Ma- guey hut type of maize granary used at present by the farmers of southeastern Puebla; the entire structure with the exception of the corner posts is made from different parts of the maguey plant (Agavesp.). (Photograph by Dr. P.C. Mangelsdorf. ) [ 191 ] Prare XXI. A. Traje, wattle and daub sealed structures used to store maize at present in the tropical lowland area of the Balsas River Basin, Guerrero; the photograph shows the first steps in the construction of this type of structure. B. Taye, wattle and daub sealed structures; photograph showing the application of the daub on the inner walls; Balsas River Basin, Guerrero. Modern Period. C. Enramada, temporary structure used for maize storage over a large part of the tropical coast- al areas of Mexico where dry winter months pre- vail; near Petatlan, Guerrero. D. Enramada, va- riant form used at present near Veracruz City, Veracruz. (Photograph by Dr. P. C. Mangelsdorf. ) Pirate XXII. A. Temporary conical structure made of maize stalks and used for maize storage near Cholula, Puebla, during the later part of the 19th Century. (Reproduced from Starr 1899.) B. Ziri- cua, temporary structure made of maize stalks and used by the farmers of the Balsas River Basin, Guerrero for storage of maize. Modern Period. C. Cylindrical reed structure similar to the stricua and used at present for maize storage among the Amuzgo Indians of southeastern Guerrero. [ 192 ] : 0 PLATE ° PLatE XIV EAA SEES PLATE XV Paci ji i PuaTE XVI Pi Ms CEA PLaTE XVII athe terete Pe: PLaTE XVIII XX ’ PLATE PratE XX] re - XXII PLATE Ca MBRIDGE, Massacuusetts, Marcu 4, 1949 BOTANICAL MUSEUM LEAFLETS HARVARD UNIVERSITY NEW ARCHAEOLOGICAL EVIDENCE ON EVOLUTION IN MAIZE BY Paut C. MANGELSDORF AND C. EARLE SMITH, JR. THe problem of the origin of maize, which for several generations past has commanded the attention of bota- nists and archaeologists alike, is appreciably nearer the point of final solution as a result of the discovery of im- portant prehistoric plant remains in New Mexico during the summer of 1948. An expedition sponsored by the Peabody Museum of Harvard University’ and led by Mr. Herbert W. Dick, a graduate student in anthropol- ogy, uncovered many cobs and other parts of maize from the accumulated refuse in an abandoned rock shelter, known as Bat Cave. This material not only furnishes direct proof of the nature of primitive maize but also pro- vides for the first time tangible evidence of a well-defined evolutionary sequence in this important American cereal. Anticipating the evidence to be presented in detail later in this paper, it can be said here that direct archae- ological evidence is now available to show: (1) that prim- itive maize was both a pod corn and a pop corn; (2) that ‘This expedition, the Early Man Division of the Upper Gila Ex- pedition of the Peabody Museum of Harvard University, is under the general direction of Dr. J. O. Brew and is financed in part by the Viking Fund of New York City; the excavation of Bat Cave was in cooperation with the University of New Mexico. The general aims of the Expedition are set forth by Brew and Danson (1948). [ 213 ] maize did not originate from teosinte; (8) that much of the variation in modern maize is the product of introgres- sion from teosinte which occurred as a later step in the evolution of maize. The prehistoric maize remains which permit us to draw these far-reaching conclusions comprise a considerable part of the total vegetal remains isolated from the refuse in Bat Cave by Mr. C. Earle Smith, Jr., the junior au- thor of this paper, who served as botanist to the expedi- tion. The vegetal remains other than maize, which are also of considerable ethnobotanical interest, will be de- scribed separately by Mr. Smith. Suffice it to say here that, in addition to maize, seeds of both kidney beans and squashes were found at all except the lowest levels of the Bat Cave refuse in which maize itself occurs. Description OF Bat Cave Location and Climate. Bat Cave is located in Catron County, New Mexico, on the edge of the Plains of San Augustin, the basin of an ancient lake of considerable size Which has been described by Powers (1989). The longitudinal axis of the basin which lies in a northeast to southwest direction is about 60 miles. The basin varies from six to twenty miles in width. There are no natural streams permanently flowing into the basin and the plains are dry for the better part of the year. ‘The arroyos and canyons carry water only during the rainy season in July and August, when afternoon showers sometimes precipitate large amounts of water in a short period of time. Records kept at Aragon, a village about 25 miles from the site, indicate an average annual rainfall of approximately 14 inches; the precipi- tation, at present, on the plains of San Augustin, is probably a few inches less. Present Vegetation. The paucity of water in the re- [ 214 ] gion greatly restricts the natural vegetation. In the im- mediate vicinity of the site at Bat Cave Atriplex canes- cens (Pursh) Nutt. provides the major cover on the plain with species of Sporobolus and Heliotropium playing a secondary role. There is a distinct change in the vegeta- tion as one leaves the old lake floor and ascends the gravelly beach terraces which mark the ancient shores of the lake. Russian thistle, Sa/sola pestifer A. Nels., is common on the edge of the plain. Many species of Com- positae, Labiatae, Chenopodiaceae, Amaranthaceae, Boraginaceae and Gramineae are found on the terraces themselves. Where the terraces are especially rocky, Fallugia paradowa (I). Don) Endl. occurs as a common shrub with Rhus trilobata Nutt. ex Torr. & Gray and species of Ribes and Forsteria. During the dry season this vegetation has a distinctly xerophytic appearance accented by occasional plants of Opuntia and Mammil- laria. In the rainy season the vegetation becomes quite green, and a number of short-lived annual flowering plants, as well as species of Selaginella and Notholaena, make their appearance. The vegetation remains more or less the same for some 200 to 400 feet up the slopes surrounding the basin. At this point a distinct change in the vegetational pattern occurs. Pinus edulis Engelm. and Juniperus spp. be- come the dominant features in the vegetation. In one ‘anyon near Bat Cave several trees of Pinus ponderosa Dougl. ex P. Laws. occur, while, in some protected areas of this pifon-juniper belt, trees of various species of Quercus are found. Vegetation in the Past. The species identified among the vegetal remains found in Bat Cave would seem to indicate that the natural vegetation which occurs today is not substantially different, except in a few character- [215 | istics, from that of the past. The region is not one where such plants as maize, beans and squashes, none of them species notable for drought resistance, would be expected to occur in the wild, and certainly at the present time conditions are scarcely suitable for their cultivation. At an earlier period, however, when Lake San Augustin was a permanent body of water with a water level con- siderably higher than the present level of the plain, maize, beans and squashes could have been, and un- doubtedly were, grown on its shores. Description of the Site. Bat Cave is situated on the eastern edge of the Plains of San Augustin about 165 feet above the level of the lake floor and at an altitude of about 7000 feet. Carved out of an almost vertical cliff, apparently by wave action, Bat Cave is actually a series of six more or less separate caves. The largest of these is about 100 feet deep and 75 feet from floor to roof. A second is approximately 40 feet deep and 80 feet from floor to roof. Four smaller caves lying between these two are 10 to 20 feet deep and 10 to 15 feet high. It is from these smaller caves that the material described in this paper was excavated. Eecavation of Material. The refuse was removed in successive strata each about twelve inches in thickness. The lowest stratum which contained plant remains is designated as I and the uppermost as VI. The bulk of the material was carefully excavated im situ by hand trowelling on a vertical face. The depth of the specimen was carefully noted in all cases. All the botanical material was packaged in separate lots for each area in the excavation. The total depth of the deposits in which maize was found averaged between five and six feet. The entire [ 216 ] deposit bore artifacts without hiatus to the bottom where it rested upon a layer of water-washed gravel. Dating. Dr. Ernst Antevs, who is preparing a sepa- rate report on the geology of Bat Cave and the surround- ing region, estimates that the cultural deposits containing the maize had their beginning not later than 2500 B.C. This estimate is based on the fact that the refuse rests upon a layer of windblown sand and dust which repre- sents a period decidedly drier than the present. Antevs believes the sand and dust were deposited between 5500 B.C. and 2500 B.C. The level representing this very dry age rests upon old beach gravels which were laid down during a preceding pluvial period. The gravel rests on bedrock. The top level of the maize-bearing deposit is dated by pottery as 500 A.D. to 1000 A.D. Thus, the deposit containing maize covers a span of not less than 3000 years. Exact dating of the site is not, for our pur- pose, of overwhelming importance. Much more signifi- cant is the fact that the characteristics of the maize from Bat Cave are such as to make it the most primitive which has so far been discovered; and, even with conservative dating, it is the most ancient. Maize Remains. The maize remains isolated from the refuse comprised a total of 766 specimens of shelled cobs, 125 loose kernels, eight specimens of husks, ten of leaf sheaths, and five of tassels and tassel fragments. Most of the specimens, even those from the lowest stratum, were extraordinarily well-preserved, with the long plant hairs on the epidermis of the glumes and the surface of the cupules still intact. Indeed, most of the vegetal material isin such an excellent state of preservation that botanists would be inclined to question the geologists’ estimates of its antiquity did not the specimens of maize from the different strata exhibit a distinct evolutionary sequence. [ 217 ] Tue Evo.turionary SEQUENCE IN THE Coss Shelled cobs comprised by far the largest fraction of the maize remains found in Bat Cave. These included 52 separate lots, totaling, as has already been stated, 766 specimens. Of these, 471 were sufficiently well-preserved to be studied for a number of distinct botanical charac- teristics. Our method of studying the prehistoric cobs was to snap each cob in two, examining the two surfaces thus exposed under a dissecting microscope. ‘The number of characteristics which can be studied in the maize cob is truly astonishing. Those which we considered are listed in Table I. Further studies of the maize cob will un- doubtedly reveal other characteristics which can be uti- lized in comparing maize populations. Lenz (1948), for example, describes a ‘‘rachis flap’’ which is prominent in some varieties and relatively inconspicuous in others. The rachis flap was so erratically distributed in the Bat Cave specimens that we gave no serious attention to it. Nor did we attempt to measure rachilla length, another characteristic in which maize varieties differ considerably, according to Lenz (1948) and Cutler and Cutler (1948). However, some of the other characteristics which we did investigate, such as length and shape of the glumes, are strongly correlated with rachilla length, so that indirectly we undoubtedly took this character into account. The characteristics in which the specimens from the six strata are compared are described below. Length of Cob. This is a simple measurement made on all cobs which were intact with respect to length or so nearly intact that their original length could be esti- mated. The data in Table I show that there is little dif- ference in the mean length of the cobs in the first three strata. Thereafter, there is a more or less progressive in- [ 218 | TABLE I. CHARACTERISTICS OF THE PREHISTORIC MAIZE OF BAT CAVE COMPARED BY STRATA Stratum Characteristics I II III IV Vv VI No. of separate lots : 4 8 6 8 24 Total No. of cobs 25 55 77 146 118 345 No. cobs scored 19 23 38 15 78 238 No. cobs intact in length 4 2 5 13 i7 25 Length intact cobs —em. max. 9.5 10:8. -8.5.- 18.8. -45 16.5 sie “se “© min. 6 7.2 6.5 5.1 5 4. iy oe is ** mean 7.3 8.4 7.7 8.5 9. 9.3 Mean diameter of rachis mm. ind 8.6 7.9 8.4 9, 11.2 e “cob 148 O° {57-096 8.9 168-617 17.5 Index cob/rachis—max. 2.5 3 2.8 2.6 2. 2.8 eo ee Ss min. 1.6 15 1.4 1.2 Ls 1.2 er oi mean 1.9 oe ane 1.9 i 1.6 Rows of grain-——max. 14. 16 16 20 16 20 Pee, se sein. 8 8 8 4 8 4 ee eé ee ae mean 10.7 12 11,2 12 | 11.4 Rachis tissue 9% spongy 28 88 58 43 29 38 ss 7 * horny [ke 12 39 39 28 24 ia st * bony 3 18 43 38 Surface of cupules % glabrous 19 4 8 8 8 aes i Pe Raa ** hispid 3 13 25 15 ae i ** hirsute 81 96 89 79 67 63 , Texture 9% fleshy 97 91 71 56 25 40 = ** horny 3 9 26 36 18 Zi 8 me bony 3 8 oT 39 me Surface ** glabrous 100 96 95 79 92 85 = ** hairy + 5 21 8 15 ° a Venation * slight 10 44 18 5 6 16 * none 90 56-82 9 94 84. Shape % boat 20 13 16 12 7 8 ** intermed. 80 61 47 59 38 49 ** collar 26 «87 29 55 43 % Texture ‘* chaffy 100 48 47 61 37 49 = * fleshy 5258 39 59 48 oe S horny 4 3 2 Surface ** clabrous 100 96 89 71 74 78 ee . . a hispid 3 8 3 3 a “* hairy 4 8 21 23 19 Venation ‘* strong 30 30 29 56 21 26 ‘““intermed. 53 26 55 28 41 43 ** none 17 44 16 16 38 31 Tunicate alleles % tu 1 a nm ** tuw 8 1 33 55 sc AS ** «toa 80 92 82 73 61 4.2 ab a ** tuh 20 8 10 26 5 3 Teosinte introgression 9% none 100* 100f 73 43 9 27 ca i % intermed. 29 52 58 61 ie _ % strong 3 5 oS 12 Total No. 2 Q 10 20 67 24 , Av. length 1/1000in. 275 298 280 812 3802 3816 7 Av. thickness SS 198 247 228 210 236 217 4 = Width max. 1/1000in. 258 301 340 376 398 428 ‘© min. re BBO 229 218 «214 = 197 ~~ 201 ‘* mean “eS 844 265 265 284 308 338 * There is one cob with a cob/rachis index of 1.2 which also shows evidence of strong introgression of teosinte. Since this cob is atypical of the stratum in other characteristics as well it is undoubtedly an intrusion and is not included in this datum. { There is one cob in this stratum which shows unmistakable teosinte intro- gression but which is atypical in other respects and is probably an intrusion. crease in mean length from stratum to stratum. Even more interesting than the mean length of the cob is a comparison of the maximum and minimum lengths. Beginning with the fourth stratum there is a progres- sively greater divergence between the shortest and the longest ears in each stratum. Part of the increase in range of variation may be attributed to a progressive increase in the size of the samples available, but there is no doubt that there is a true increase in the range of variation be- yond that resulting from sampling. This is reflected not only in the size of the cob but in many other characters as well. Diameter of the Rachis and Cob. When a cob of maize is snapped in two it is quite easy, in examining the cross section, especially of the lower half, to distinguish be- tween an inner zone made up of the rachis and an outer zone made up of the glumes. These zones represent two concentric circles whose diameter is easily determined with a pair of calipers. The data in Table I reveal that there are no significant differences in the diameters of the rachis and the entire cob in the three lowest strata. Thereafter, however, there is a progressive increase in mean diameter of the rachis and in mean diameter of the entire cob from stratum to stratum. The Cob/Rachis Index. Much more impoicant than the actual diameter of these two concentric circles rep- resenting the rachis and the entire cob is the ratio of one to the other. This ratio, called the cob/rachis index and obtained by dividing the larger figure by the smaller, may prove to be one of the most useful measurements yet employed in studying and classifying maize varieties, for it has a clear-cut and definite genetic basis involving changes in alleles at the 7'u-tw locus on the fourth chro- mosome. The existence of four alleles—tunicate ( 7'w), [ 219 ] half-tunicate (tuk), weak-tunicate (tww), and non-tunicate (tuw)—at this locus has already been demonstrated (Man- gelsdorf, 1948). In addition, a condition intermediate between weak-tunicate and half-tunicate is found in some races of maize including the Chapalote maize of Mexico. Since this form of tunicate, which has not yet been studied genetically, produces approximately the same effect when homozygous as does half-tunicate when het- erozy gous, it may appropriately be called fourth-tunicate and tentatively given the symbols tw/, subject to verifi- ‘ation by later genetic tests. The importance of these 7'w alleles in connection with our present studies lies in the fact that a change from a higher to a lower allele in the series is almost invariably accompanied by a drastic change in the cob/rachis ratio. In other words, when the glumes become shorter, the rachis, other factors remaining constant, automatically becomes thicker. This is easily demonstrated in modern isogenic stocks which are segregating for two alleles in the Ju series. A comparison of three such stocks is shown below: Stock Cob/Rachis Index Inbred A158 tuhtuh 2.9 _ “* tuhtu 2.1 ve “tutu 1.5 The cob/rachis index, therefore, is in substantial part a simple estimate of the degree to which a variety of maize has departed from a primitive tunicate form to- ward the modern non-tunicate condition. It must be realized, however, that in addition to changes in both the diameter of the rachis and that of the entire cob re- sulting from changes in 7'w alleles, there are other differ- ences resulting from the action of a vast number of modifying factors. [ 220 | It is interesting to note that the cob/rachis index in the Bat Cave maize varies from 3.0, a condition in which the rachis is slender and the glumes are prominent, to 1.2, a condition in which the rachis is large and the glumes are relatively short. The cob/rachis index is, however, relatively constant through the three lower strata, but shows some change in the upper three. It is of particular interest to note that, although new types with a low cob/rachis index appear in the upper strata, the original types with a high index, though becoming somewhat less frequent, do not disappear. If human se- lection had been operating in favor of the lower alleles at the T'u-tu locus, it certainly had not, in the period rep- resented by the uppermost strata, succeeded in com- pletely eliminating the higher alleles. There has, as in other characteristics, been primarily an increase in the range of variation. Rows of Grain. There has been no constant change in the number of rows of grain from stratum to stratum. An increase in the circumference of the cob should re- sult either in an increased number of rows of grain or in wider grains. It will be shown later that there has been a progressive increase from stratum to stratum in the average width of the grains. Similarly, although it is not clearly revealed by the summarized data, there has been an increase in the number of rows as a consequence of the increase in the circumference of the rachis. Indeed, the principal variation in the two lower strata has been con- cerned largely with the diameter of the rachis, the prom- inence of the glumes and the number of rows. Of 41 ears which were scored for both row numbers and 7'v allele, 16 were classified as fourth-tunicate, 8 as half- tunicate and 17 as intermediate between these two con- ditions. The average row numbers of these three groups is shown below. [ 221 | Type of Har Average Row Number Half-tunicate 10.5 Intermediate 112 Fourth-tunicate 11.9 Here, then, is another factor not previously recog- nized which is involved in the complex problem of row number in maize. It can now be said that, other things being equal, the lower the allele of Zu, the higher will be the number of rows on the ear. Rachis Tissue. It is a common observation both of practical farmers and students of maize that there is a great variation in the hardness or stiffness of individual cobs. Lenz (1948) has recently reported that the texture of the cob depends to a large extent upon the relative amount of heavily schlerenchymatized tissue. A_histo- logical examination of the nearly 500 cobs included in this study was out of the question, but it was possible to score the cobs with respect to the tissue of the rachis by testing with a dissecting needle the surface exposed when the cob was broken. The tissue immediately below the epidermis of the rachis was used for this purpose ; the cobs were classified as having spongy, horny or bony tis- sue. Bony cobs, such as would be expected from an in- trogression of ‘Tripsacum or teosinte, do not make their first appearance until Stratum III where they occur to the extent of three percent. The percentage of bony cobs reaches a maximum of 48 percent in Stratum V and falls off slightly to 88 percent in Stratum VI. Surface of Cupules. The alveoli or cupules in the ra- chis from which the paired spikelets arise vary consider- ably in the nature of their surfaces. Some are virtually glabrous, others are beset with short stiff hairs (hispid) and still others are lined with long slender hairs (hirsute). Glabrous and hirsute types occur in all strata, but cupules | 222 | with hispid surfaces do not make their appearance until Stratum III. Thereafter, they increase progressively to Stratum V and fall off again in VI, a pattern quite simi- lar to that exhibited by bony cobs and probably for the same reason. A strong introgression of teosinte into maize, as in some segregates from maize-teosinte hy- brids, results in glabrous cupules. This is not surprising, since the inner surface of the bony shell of teosinte, the counterpart of the cupule in the rachis of maize, is al- ways glabrous. But a smaller amount of teosinte intro- gression may cause the hairs in the cupule merely to be shorter and stiffer. Thus the hispid condition encoun- tered in later strata of the Bat Cave material may well be a manifestation of teosinte (or Tripsacum) introgres- sion. Lower Glumes. The lower glumes of an ear of maize vary: (a) in texture which may be chaffy, fleshy, horny or bony; (b) in surface which may be glabrous, hispid or hairy; and (c) in venation which may be apparent to the eye or obscured by the thickness of the tissues. The data in Table I reveal that there are no truly chaffy or membranaceous glumes such as are found in modern pod corn. Neither are there bony glumes such as those found in teosinte until we reach Stratum III. Thereafter, the percentage of bony glumes increases progressively to Stratum V and falls off slightly in Stratum VI. The in- duration of the glumes is almost a certain sign of teosinte or Tripsacum introgression. Types with bony glumes similar to those found in the upper strata of this material are readily synthesized experimentally by substituting one or more chromosomes from teosinte for the corres- ponding maize chromosome. There is little doubt that the strong, bony glumes evident in much of the Bat Cave material is the product of introgression from teosinte or [ 223 ] Tripsacum. The significant fact is that this introgression is not apparent in the lower strata. The fact that all glumes in the lowest stratum are glabrous may be of some significance. Cutler and Cutler (1948) report that hairs on the pistillate glumes are occa- sionally present in numbers proportional to those found on the leaf sheath. Insofar as glabrous pistillate glumes are associated with glabrous leaf sheaths, we may infer that the early Bat Cave corn was glabrous. Completely glabrous husks, found in Stratum II and described later, support this conclusion. The venation of the lower glume may be obscured because the glume is fleshy or because it is indurated as in teosinte. Both conditions are encountered in this ma- terial. Glumes on 90 percent of the ears in the lowest stratum show no venation because they are fleshy. In Stratum VI, however, the lack of venation occurs on both fleshy and indurated glumes. Upper Glumes. The upper pistillate glumes, easily studied on the exposed end of the lower half of a broken cob, possess several interesting characteristics. They vary in shape from those which resemble one end of a boat and have obviously once enclosed a kernel, to a short, stiff half-collar which once surrounded the base of a kernel. Forms intermediate between these two extremes are common. Boat-shaped upper glumes characteristic of true pod corn are found in all six strata, but are most com- mon in Stratum I. Collar-shaped glumes are most com- mon in Stratum V. The upper glume is most commonly membranaceous or chaffy, but may be fleshy, and, as in teosinte and Trip- sacum, even horny. All of the ears in the lowest stratum have chaffy upper glumes. Horny upper glumes occur only in Strata V and VI. [ 224 ] The surface of the upper glumes, like that of the lower, may be glabrous, hispid, or hairy, although the two glumes are not necessarily identical with respect to these characteristics. Only glabrous upper glumes were found in Stratum I and glabrous glumes continued to predom- inate throughout all strata. Hairy upper glumes made their first appearance in Stratum II and increased in fre- quency in successive strata up to V. Hispid upper glumes first appeared in III and occurred in small percentages in the remaining strata. The visible venation of the upper glume is strongly but not completely correlated with its shape and texture. The data on venation of the upper glume in this material are generally in agreement with the data on shape and texture. Strongly veined glumes, like those which are boat-shaped or chaffy, have their lowest frequency in Stratum V. Tunicate Alleles. An attempt was made to classify the cobs with respect to the degree to which they were tunicate or non-tunicate. As has already been mentioned, four degrees of tunicate— Tu, tuk, tww, and tu—are known and a fifth, tw intermediate between fu’ and tuw, is believed to occur. No cobs with glumes as strongly developed as those of modern pod corn of the genotype Tutu were found, but true pod corn of the half-tunicate type occurred in each of the six strata. Pod corn of the fourth-tunicate type also occurred in all strata and indeed was the most common type in all except the uppermost. Weak-tuni- cate first appeared in Stratum III and became the most common type in VI. Only one ear regarded as non- tunicate was encountered. These data indicate strongly that one of the principal factors involved in the evolution of maize has been a eee | gradual replacement of the higher alleles in the T'w-tu series by the lower. Mangelsdorf (1948) has shown that both Zw and tuk have relatively high mutation rates. It is evident from the Bat Cave material that mutations at the 7'u-tu locus have occurred and that there has been a gradual decline in the frequency of the higher alleles and an increase in the frequency of the lower. This, in turn, has had far-reaching effects upon the size of the ra- chis, on the size of the entire cob, on the space available for the vascular system, upon the length of the rachillae which bear the kernels and undoubtedly upon the ulti- mate yield of grain. ‘The importance of the changes in relative frequency of the different 7'w alleles can scarcely be over-emphasized. Teosinte Introgression. The hypothesis of Mangels- dorf and Reeves (1989) with respect to the origin of maize postulates that teosinte is the product of a natural hy- bridization of maize and ‘Tripsacum which was followed by an introgression of teosinte (or Tripsacum) germplasm into maize. The origin of teosinte as the progeny of maize and ‘Tripsacum has not yet been finally proved, although much additional evidence in support of the hypothesis has been accumulated since 1989 (cf. Mangelsdorf, 1946). For the purpose of this paper, however, it is not neces- sary to debate the question of the origin of teosinte. We are here concerned primarily with the question whether there is evidence of such an introgression of teosinte into maize. The answer is that there is strong evidence of such introgression, but only in the later strata of the Bat Cave material. The scores for teosinte introgression represent esti- mates involving personal judgment which in turn is based upon an extensive experience in studying derivatives of maize-teosinte hybrids produced experimentally. When 226 ] a cob has bony rachis tissue, bony lower glumes, upper glumes lacking in venation and a hispid cupule, it is al- most certainly a product of teosinte introgression. If it has some of these characteristics, but not all, it is assumed to represent teosinte introgression of a lower order. In each of the first two strata there was only one cob which showed evidence of strong introgression of teo- sinte. These cobs were atypical of the strata in almost all of their characteristics and are probably ‘‘intrusions, ”’ a convenient term employed by archaeologists to describe specimens which are found in any part of a cultural de- posit in which they obviously do not belong. With rats and other rodents digging burrows in ancient refuse heaps, it is almost inevitable that a specimen from a high- er stratum will occasionally find its way into a lower. In- trusions, therefore, have a real basis in fact and are not merely a device for conveniently explaining exceptions. Cobs showing evidence of teosinte introgression occur in somewhat greater frequency in Stratum III, but are not common. These, too, may be intrusions or they may result from the fact that the line of demarcation between the strata is entirely arbitrary. In any case, not until Stratum IV is reached can we be certain that teosinte- contaminated maize has become a conspicuous part of the population. The amount of teosinte introgression is greatest in Stratum V and drops off perceptibly in Stratum VI. Two possible reasons for this decline come to mind at once. The first is that introductions of non-tripsacoid maize from other regions have ‘‘diluted’’ the teosinte germplasm; the second is that some of the more obvious effects of teosinte have become modified through the de- velopment of an appropriate modifier complex. Since there is no clear-cut evidence of an introduction of non- tripsacoid maize on a large scale, the second possibility Bryal appears to be the more plausible one. The senior author (1946) suggested several years ago that much of the re- cent evolution in maize has been a matter of absorbing the morphological assets of teosinte while suppressing its morphological liabilities through selection for appropri- ate modifier factors. The material from Bat Cave is al- most direct evidence in support of this thesis. The interaction between the various alleles of Ju and the introgression of teosinte germplasm is an interesting one which is clearly revealed by some of the cobs in Strata IV and V. In a recent paper (1948) the senior author illustrated a spike of tunicate teosinte produced by introducing the 7 gene into teosinte through re- peated back-crossing. The rachis was reduced to a slen- der disarticulating stem. The prominent glumes, quite distinct from the bony glumes of teosinte in size and structure, had, nevertheless, taken on a coriaceous quality very different from the papery or chaffy glumes of pod corn. They seemed almost to be a product of distributing over a large volume of tissue all of the factors for hard- ness and stiffness ordinarily found in a small volume. Teosinte introgression has similar effects upon geno- types representing lower alleles in the 7'w-tu series. Plate XXVI, Figs. F, G, H, illustrates three cobs from Bat Cave which are similar, if not identical, with respect to their alleles of Z'u, but which differ decidedly in the amount of teosinte introgression. Evidence for introgression of teosinte germplasm into maize was not confined to a general increase in the hard- ness of the rachis and outer glumes, since individual char- acteristics of teosinte also made their appearance. Three cobs were found with the distichous arrangement charac- teristic of teosinte. These occurred in Strata IV, V, and VI. ‘Two cobs were found which bore single spikelets, also a teosinte characteristic. These occurred in Strata [ 228 ] IV and V. All of the evidence combined points to two conclusions with respect to the rdle of teosinte in the origin and evolution of maize: (1) Maize did not originate from teosinte; (2) The course of evolution of maize in its later stages has been strongly affected by an intro- gression of teosinte germplasm. If it should ever be es- tablished beyond a reasonable doubt that teosinte is a hybrid of maize and Tripsacum, then a third important conclusion would be possible; namely that the hybridi- zation between maize and Tripsacum which produced teosinte must have occurred not later than 500 B.C. and perhaps considerably earlier. EvoLutTiIoNn OF KERNEL SIZE The primitive people who grew the maize whose re- mains were found in Bat Cave were evidently extremely efficient in their corn-shelling operations. Not a single cob among the 766 recovered from the digging bore even one normal kernel. However, a total of 125 loose ker- nels, all reasonably well-preserved, were uncovered in the refuse and some of these were found in each of the six layers. The kernels in each of the two lower strata are small and corneous, and were undoubtedly capable of popping. Thus, the primitive maize of Bat Cave was, as Sturtevant (1894) supposed primitive corn to be, both a pod corn and a pop corn. The kernels, like the cobs, show a progressive increase in mean size from stratum to stratum. ‘They were meas- ured in three dimensions and, since the calipers used registered in thousandths of an inch, these units were also used in computing the mean as well as the maximum and minimum dimensions which are set forth in Table I. It is apparent from the data in Table I that, while there is little change in the mean thickness of the kernels from stratum to stratum, there is a noticeable if somewhat [ 229 | irregular increase in mean length and a marked and pro- gressive increase in average width. Since kernel width has increased proportionately more than length, it fur- nishes the most useful dimension for detailed study. Kven more striking than the increase in average kernel width is the progressive increase in the maximum width (width of widest kernel) from stratum to stratum. This increase is virtually linear, but the linearity is probably spurious, since it is scarcely to be believed that the sev- eral strata represent identical periods of time, or that ge- netic changes of a given order always produce the same change in absolute dimensions whether acting upon small or large kernels. Furthermore, part of the increase in maximum size is nothing more than a consequence of increasing the size of the sample. This factor, however, does not account for all of the increase in maximum size, since there is a substantial increase from Stratum IV to Stratum VI, although the samples from the two strata are approximately equal in size. Furthermore, the cobs, even though they lack kernels, furnish indirect evidence of a progressive increase from stratum to stratum in the mean and maximum. width of the kernel. Although there is a progressive increase in kernel width from stratum to stratum, the smaller-kerneled types with which the series began did not disappear from the scene when the new types with larger kernels came into existence. What has actually occurred is that there has been an appreciable increase in the range of variation with respect to kernel size. This is illustrated by Plate XXIV, Fig. A. The marked increase in variation in kernel size in a relatively short period of time provides some indication of the way in which variability accumulates in a man- made environment in which the pressure of natural selec- tion has been reduced or released. Here, perhaps, is the [ 230 ] key to the enormous diversity found in so many species of cultivated plants and domestic animals. Here, too, perhaps, is a clue to the rapidity with which organisms, in nature, can evolve in the face of drastic changes in the environment. EVOLUTION OF THE Husks Although only eight specimens of husks were found in the Bat Cave remains, these differ sufficiently in their characteristics from stratum to stratum to allow some far-reaching inferences to be drawn with respect to the nature of the primitive maize inflorescence and the evo- lution of husks during domestication. No specimens of husks were found in the lowest stra- tum. Only one specimen was found in Stratum IT, but this single specimen, although it represents the husk covering of less than half of the ear, has an extremely important story to tell. The husk appears to be intact with respect to length and reaches a maximum length of 24.5 em. Four leaf sheaths are included, and the number may tell some- thing of the relative height at which the ear arose. If primitive maize resembled either of the maize relatives, teosinte or Tripsacum, its lateral branches would have had a node number approximately equal to the node number on the primary stalk above the point of origin of the branch. This allows us to infer that the ear was borne at the fifth node below the tassel, but it tells us nothing about the number of nodes which might have occurred below the ear. All husks have prominent parallel veins widely spaced. There is no anastomosing venation. It is primitive, un- differentiated husks of this type which produce the “‘stri- ations’’ on the surface of the kernels in ears which are tightly enclosed in husks (Anderson 1944a). Striated ker- [ 281 J nels may therefore be regarded as a primitive character. Perhaps the most important characteristic of this husk is that it shows no evidence of ever having been shaped around an ear. Furthermore, the stem or shank on which it occurs is quite slender, so slender indeed that, of the cobs which occur in Strata I and II, only a cob similar to that illustrated in Plate X XVII could have been borne on it. ‘The husks, as shown in this Plate, are far longer than needed to enclose an ear of this type. All of these facts combined point to the almost ines- capable conclusion that the husks at this stage in the evolution of maize did not enclose the ear. Instead, they were nothing more than an involucre of leaf sheaths, not greatly modified, subtending and surrounding the base of the pistillate inflorescence, but by no means com- pletely enclosing it. The remaining seven specimens of husks together are not as interesting as the single specimen found in Stra- tum II. One of the specimens found in Stratum V ap- pears to be intact or almost so and has a maximum length of 10.5 em. It includes five leaf sheaths, all glabrous and lacking in leaf blades and ligules. The veins are parallel, but are more closely spaced than in the sheaths found in Stratum II. The husks in Stratum V differ from those in Stratum IT primarily in their shape, which gives definite evidence of having once enclosed an ear. Two other specimens found in Stratum V differ from the one described immediately above in having scattered hairs and numerous prickles on the outer sheaths and anastomosing venation on the inner. Both show evidence of having once enclosed an ear. The four specimens found in Stratum VI are too frag- mentary to be of great value. Two of the specimens, however, show evidence of having once been shaped around an ear. ‘lwo of the four specimens are glabrous. [ 282 ] The third has its inner surface beset with short prickles while the outer bears scattered hairs and prickles. The last specimen bears numerous short hairs on the inner surface and between the veins on the outer surface. The veins themselves bear longer and more numerous hairs on the outer surface. Only parallel venation was found on three of the four fragments; anastomosing venation was apparent on the fourth. Cutler (1946) has pointed out that the husks of maize are not fundamentally different from a variety of similar structures found in other genera of grasses, and he com- pares them specifically with a kind of fasciation some- times occurring in the genus Trichachne in which the lateral inflorescence is enclosed by the leaves of the culm on which it is borne. The husks found in Stratum II of the Bat Cave remains would indicate that Cutler’s com- parison is valid. Furthermore, the later evolution of the husk covering is quite consistent with the modification of leaf sheaths which occur in other grasses. The leaf sheaths which comprise the husks obviously became wider during the evolution of maize and event- ually completely enclosed the ear. The leaf blade and ligule were lost. The outer and inner sheaths became differentiated, the latter becoming thin and papery with anastomosing venation and were no longer capable of producing striations on the kernels. These changes are quite in keeping with the morphological pattern in other genera of grasses. Arber (1934), speaking of modification of leaf sheaths in the grasses, makes the following state- ment: An exaggeration of the size [ width] of the leaf-sheath, and a corres- ponding reduction in the limb, occur as part of the normal development in the uppermost foliage leaf enclosing the inflorescence in many grass- es, such as Phalaris canariensis L. and Alopecurus pratensis L. In the xerophytie South African grass, Ehrharta aphylla Schrad., this type of change has gone so far that the lamina is reduced to a mere point.’’ "Reference to figures omitted in this quotation. [ 233 ] The figures that accompany Arber’s statement are particularly interesting in illustrating the marked broad- ening of the sheath in Phalaris canariensis and the re- duction of the leaf blade to a mere point in Hhrharta. Arber also describes and illustrates a broadening and shortening of the leaf sheaths and a reduction in the lam- inae in Agropyron repens Beauv., resulting from injury by insects. And Hitchcock (19385) illustrates a number of species of grasses in which the sheath or sheaths im- mediately below the inflorescence have become shorter and broader and have largely lost their laminae. It is evident, therefore, that the development of the husks in maize is nothing more than an exaggerated form of a morphological development which is widely distributed in the Gramineae. In several recent papers (1946,1948) the senior author described the evolutionary steps which were thought to have occurred in maize during evolution under domesti- cation. In the light of the direct evidence now bearing on this point the views previously expressed are subject to possible modification. For example, it was assumed that wild maize bore mixed staminate and pistillate in- florescences terminally on lateral branches whose sheaths and laminae were essentially identical with those on the main stalk. Although this assumption has not yet been shown to be erroneous, another possibility is now appar- ent. It is quite possible that wild maize already had dis- tinct staminate and pistillate inflorescences, and that a shortening of the internodes immediately below the pis- tillate inflorescence had already occurred, giving rise to an involucre of leaf sheaths which surrounded the base of the ear but did not enclose it. There is no reason why a plant with such an inflorescence could not have existed in a wild state. There is, at least, no doubt that such a plant did exist in the early stages of domestication. [ 234 J One more important and clear-cut conclusion can be drawn from the evidence furnished by the prehistoric husks. Had maize originated from teosinte, as some bot- anists still suppose, then the early primitive races of maize, only slightly removed from teosinte in their characteris- tics, might be expected to possess short husks scarcely covering the ears. The early evolution of the husks would then have become a matter of a progressive increase in length to enable the husks to enclose progressively longer ears. In the Bat Cave material the evolutionary sequence in the husks is exactly the reverse of this. The earliest husks are the longest. Indeed, they are several times longer than the longest ear in the same stratum. The early evolution of husks has been a matter of their be- coming shorter and wider rather than longer. Thus, the accumulated circumstantial evidence against the hypothesis that maize originated from teosinte is finally reinforced by direct evidence which is almost conclusive. LEAF SHEATHS In addition to the husks which are largely modified leaf sheaths, ten specimens of true leaf sheaths were found—one from Stratum IV, the remaining nine from VI. The specimen from Stratum IV is completely gla- brous; of the nine specimens from VI, four are com- pletely glabrous on both inner and outer surfaces, three have hairs and prickles on both surfaces, one is glabrous on the outer surface and beset with numerous short hairs on the inner, and one bears short hairs on the outer sur- face but is glabrous on the inner. The significance of these variations is not yet clear. Only one fact is certain. At least part of the maize found in Bat Cave possessed glabrous leaf sheaths. ‘True gla- brousness is not common in maize varieties today, except [ 235 ] among those grown at high altitudes in Ecuador, Peru and Bolivia, although it is occasionally encountered in Guatemala' and Mexico. The inheritance of glabrousness (unpublished data) does not appear to be particularly complex and it is possible that the characteristic has ap- peared more than once in the evolution of maize under domestication. On the other hand, the fact that there is a large center of glabrous maize at high altitudes in South America may not be without significance in connection with a consideration of the early glabrous maize from Bat Cave which also grew at a relatively high altitude. "TASSELS Anderson and his co-workers (1942, 1944b, 1948) have repeatedly emphasized the importance of studying the maize tassel as a means of understanding the morphology of the ear, and in studying the tassel they have discov- ered an array of characteristics which promise to be in- valuable in the classification of living maize varieties. ‘Tassels, however, are by no means as enduring as ears and cobs, nor are they so likely to be collected in great numbers by primitive peoples or deposited by them in refuse heaps or graves. Only five specimens of tassels were found in the Bat Cave remains, one in Stratum III, three in TV and one in V. The specimen in Stratum III consisted of a fragment of a lateral tassel branch. This included five nodes, of which three each bore an extra spikelet. This condition, first described by Cutler (1946) and called ‘‘multiplica- tion,’ iscommon in the maize varieties of South America. ' Mangelsdorf and Cameron (1942) described as glabrous many va- rieties from Guatemala. This is an error in terminology. Most of the varieties described by them as glabrous, although lacking the promi- nent hairiness so characteristic of high-altitude varieties of Mexico and Central America, are actually hispid. [ 236 ] It is superficially similar to, but actually quite different from, a condition which Anderson (1944b) has designated as ‘‘condensation,’* in which there is a telescoping of in- ternodes to the extent that two or more pairs of spike- lets appear to arise at the same node. Condensation is common in the maize of Mexico and Central America. The spikelets on the single specimen from Stratum III had their glumes sparsely covered with prickles. The three specimens from Stratum IV comprised two fragments and one intact tassel. The first, a fragment of a central spike, had its glumes covered with short hairs and prickles and showed neither condensation nor multi- plication. The second, a fragment of a branch, had its glumes sparsely covered with hairs and prickles and showed both multiplication and condensation. It in- cluded 12 nodes upon which were borne 37 spikelets, 13 more than should have occurred in the absence of both multiplication and condensation. One of the extra spike- lets was the result of multiplication; the others, 6 pairs, were the consequence of condensation. The condensa- tion ratio is, therefore, 1.5. The intact tassel in this stratum comprised a lax cen- tral spike and three branches. Its glumes were covered with hairs and prickles, and the tassel showed neither multiplication nor condensation. It is the kind of tassel, which, among living varieties, is found only in early- maturing types. The single specimen found in Stratum V_ had its glumes covered with long hairs. The fragment included 21 nodes and was free of both condensation and multi- plication since there were no extra spikelets at any node. No pistillate spikelets were found in any of the frag- ments or in the intact tassel. So far as these few tassel specimens have any bearing upon the problem of evolution in maize, they suggest, [ 287 ] but certainly do not prove, that (1) multiplication pre- ceded condensation in this sequence, and (2) the maize grown by the Bat Cave people was early-maturing. Primitive Maize Reconsrrucrep From a study of the cobs found in the lower strata of the Bat Cave material, supplemented by evidence sup- plied by fragments of husks, sheaths, and tassels, it is now possible to reconstruct a primitive maize with a sub- stantial measure of confidence in the reliability of the reconstruction. The earliest Bat Cave material, which was probably an early-maturing race, had glabrous leaf sheaths. Its tassels bore lax central spikes and were sparsely branched. There was no condensation of the spikelets. Its ear, borne several nodes below the tassel, was surrounded, but not enclosed, by an involucre of slightly modified leaf sheaths (some bearing ligules and rudimentary laminae) which were at least twice as long as the ear itself. The small ear arising from the center of this involucre was a typical grass spike bearing paired pistil- late spikelets on a slender rachis. The phyllotaxy was definitely spiral. Indeed, it was so strongly spiral that, in some cases, the cupule from which the spikelets arose was not at right angles to the vertical axis of the rachis but was more or less parallel to the sloping path of the spiral.’ The lower glumes of the spikelets were fleshy, slightly flattened and lacking in conspicuous venation. They were quite similar to the glumes of sorghum and, like them, they did not completely enclose one half of the kernel. The upper glumes were somewhat longer than were the lower, a condition common among grasses, and were glabrous, chaffy, strongly veined, boat-shaped, and enclosed half of the kernel with length to spare. The ‘In a previous paper the senior author (1945) expressed the opinion that spiral phyllotaxy in maize is largely the product of introgression from teosinte. He could scarcely have been more wrong. [ 288 ] kernels were small, slightly longer than wide, and corne- ous. This maize was both a pop corn and a pod corn. An artist’s reconstruction of this primitive maize, ex- ecuted by Mr. Gordon W. Dillon to whom we are in- debted not only for his artistry but also for his botanical acumen, is shown in Plate XXIV, Fig. B. In its phyl- lotaxy and the nature of its glumes the reconstruction is based upon an actual specimen from Stratum [ illus- trated in Plate XXIV, Fig. C. Since this specimen was not intact with respect to length, the reconstructed ear was drawn to correspond in length to an intact specimen in the same stratum. The size of the kernels in the re- construction was determined by the shape of the upper glume which, as shown by examination under the mi- croscope, had obviously partially enclosed a kernel. The kernels on the reconstructed ear are somewhat smaller than any which were actually found in Stratum [], but no smaller than several found in later strata. It is possible that the kernels in primitive maize were not so completely enclosed by glumes as the artist’s re- construction would suggest and were actually more like kernels of varieties of grain sorghum which, although shorter than their glumes, are large enough to push the lower and upper glumes apart. In all other respects, however, Mr. Dillon’s drawing represents an accurate and valid reconstruction of actual remains found in the lowest stratum of the Bat Cave material. Facrors INVOLVED IN THE EVOLUTION OF MAIZE The evolutionary steps required to transform the prim- itive Bat Cave maize into a modern Corn-Belt Dent are no longer difficult to visualize. Indeed, many of them are clearly illustrated by the vegetal remains described in this paper. Elucidation of the remainder requires little more than a judicious use of the imagination. [ 239 ] The three principal factors in the evolution of maize have probably been: (1) a reduction in the pressure of natural selection; (2) a change in the alleles at the T'w- tu locus; (8) an introgression of teosinte germplasm into maize. The principal effect of the first factor has been a tre- mendous increase in total variation. Variations, which in nature would have been rigidly selected against, survive and accumulate in a man-made environment where the struggle for existence is reduced to intra-specific compe- tition. There is no evidence that human selection was an important factor in the early evolution of maize under domestication. Thus, in the absence of natural selection, and before human selection became a factor, the trend was one of creating a wider and wider range of variation. The changes in the alleles at the 7'u-tw locus have al- ready been mentioned, but the full significance of these changes cannot be appreciated without considering the present-day products of this evolutionary trend. It has already been pointed out that changes from the higher to the lower alleles at the 7'u-tu locus result in a reduc- tion in the size of the glumes and an increase in the size of the rachis. These are the primary effects. The second- ary effects, which are perhaps even more important, in- clude a substantial increase in the vascular system of the ear, and a reduction in the length of the rachillae upon which the spikelets are borne. Cutler and Cutler (1948) state that the rachilla in teosinte, and in some species of Tripsacum, is compacted. They state further that in races of maize with long, slender rachillae the grains yield to pressure and can be pushed into the cob, while in races with compact rachillae the grains are firm. All of these characteristics can easily be studied, with no more equipment than a pair of calipers and a rule, by an examination of the cross section of an ear exposed by [ 240 ] breaking the ear in two. It has already been mentioned that the cross section of a shelled cob comprises two con- centric circles, the inner enclosing the zone of rachis tis- sue, and the outer the zone of glume tissue. The four upper diagrams in Plate X XVIII show cross sections of four cobs from Bat Cave and illustrate a sequence of changes in the cob/rachis index. When the kernels are still attached to a cob, the cross section comprises three instead of two concentric circles, the diameter of the third circle representing the distance to the dorsal surface of the kernels. In true pod corn (tuhktuk and Tutu in Plate XXVIII) the third circle may lie within the second, but in all other types of maize it lies without. Now, if the average length of the kernels themselves is ascertained (for example, by measuring ten kernels laid end to end), the point at which the base of the kernels is attached to the rachillae is easily deter- mined by measuring in from the third circle a distance equivalent to the average length of the kernel. ‘These distances are shown by broken lines within the glume zone in the diagrams in Plate XXVIII. All of these are based upon actual diameters of the three concentric cir- cles described above and, although fundamentally simple, are highly significant. They reveal, for example, that in the three genotypes tuAtuh, tuktu and tutu from a relative- ly isogenic stock derived by repeated back-crossing to In- bred A158, there is a progressive increase in the diameter of the rachis, a progressive decrease in the size of the glume zone and a progressive shortening of the distance between the surface of the rachis and the base of the ker- nels. A similar comparison between J'utu and tuwhktu, also in a relatively isogenic stock resulting from back-crossing to Inbred P39, illustrates a similar situation. In both of these series the changing relationship of rachis, glumes and kernels is entirely a matter of alleles at the T'u-tu [ 241 ] locus. There is no doubt, however, that in nature teo- sinte introgression also has had its effects superimposed upon these. One of the most important of these effects is a toughening and hardening of the tissues of the glumes and rachis. Had this not occurred, the massive rachis and the short glumes of modern varieties could scarcely have come into existence. The diagram of modern Dent corn (Plate XXVIII, fourth row) illustrates the lengths to which the evolu- tionary trends revealed by the Bat Cave cobs have now gone. ‘The rachis has become massive and capable of en- closing an enormous vascular system. The glume zone has been reduced to a minimum, but the glumes have become more or less indurated. The distance between the surface of the rachis and the base of the kernels has been drastically shortened. The entire ear has become an extremely efficient grain-bearing structure, one which, to be sure, would scarcely be capable of surviving in the wild, but which in a man-made environment has no peer among cultivated cereals. Maize is, in many respects, the most efficient producer of foodstuffs among the cereals, and one of the factors in that efficiency is undoubtedly the nature of its pistillate inflorescence. THE PLACE OF ORIGIN oF MAIZE The evidence from Bat Cave sheds no new light upon the place of origin of maize as a wild plant. It seems scarcely possible that maize could have existed as a wild plant in the vicinity of Bat Cave since the present and past vegetation of this region is made up largely of xero- phytic or near-xerophytie species. Furthermore, there is no more reason for suspecting that maize originated here than for believing that beans and squashes also had their origin in this region. Mangelsdorf and Reeves (1939) concluded that maize [242 ] had its origin as a wild pod corn in the lowlands of South America. The origin of maize from pod corn is well estab- lished by the Bat Cave remains, but this does not prove that maize originated in South America. The fact that weak alleles of pod corn still have their highest frequency in South America may be of some significance in this connection. Also significant may be the fact that the earliest Bat Cave maize is glabrous and that there is a center of glabrous maize in the highlands of Bolivia, Peru and Ecuador. If an exact dating of the Bat Cave remains can be agreed upon, perhaps a correlation of the Bat Cave maize with prehistoric maize of the same period in South Amer- ica will throw new light on the problem. In the mean- time, the question of where maize originated as a wild plant is not answered by the new evidence. SUMMARY 1. Remains of maize isolated from a cultura] deposit in Bat Cave in New Mexico reveal a distinct evolutionary sequence. 2. The remains are conservatively estimated to repre- sent a span of approximately 8000 years, beginning not later than 2500 B.C. They may actually have begun much earlier. 3. The maize excavated from the lower strata is the most primitive maize so far known. It is both a pod corn and a pop corn. The ear is not enclosed in husks, but is surrounded at its base by an involucre of leaf sheaths. 4. This early maize is clearly not derived from teosinte. 5. Beginning about midway in the sequence, there is strong evidence of an introgression of teosinte germ- plasm into maize. [ 243 ] 6. There is a progressive increase in cob and kernel size from stratum to stratum. 7. Since ancient types did not disappear completely when new types came into existence, there is a progres- sive increase in total variability from stratum to stratum. This factor is believed to be of particular significance in the evolution of cultivated plants in general. 8. Changes in alleles at the J'u-tu locus and an intro- gression of teosinte are believed to be the two most im- portant factors in the evolution of maize. These involve a progressive increase in the size of the rachis and the vascular system of the ear, a reduction in the size of the glumes, a shortening of the distance between the rachis and the base of the kernels and a hardening and stiffen- ing of rachis and glume tissues. 9. The problem of where maize originated as a wild plant is not solved by the new evidence. ACKNOWLEDGMENTS It would be difficult to acknowledge the full extent of our indebtedness to Mr. Herbert W. Dick, whose cap- able leadership made possible the success of the 1948 Bat Cave Expedition and whose recognition of the impor- tance of vegetal remains is directly responsible for the archaeological material upon which this paper is based. We are also grateful to Mr. Gordon W. Dillon for his reconstruction of primitive maize and to C. Parker Mangelsdorf for the diagrammatic illustrations of cross sections of various types of ears. [ 244 ] LITERATURE CITED Anderson, Edgar. 1944a. Maiz reventador. Ann. Mo. Bot, Gard. 31: 301-314. Anderson, Edgar. 1944b. Homologies of the ear and tassel in Zea Mays. Ann. Mo. Bot. Gard. 31: 325-342. Anderson, Edgar and William L. Brown. 1948. A morphological analysis of row number in maize. Ann. Mo. Bot. Gard. 35: 323- 336. Anderson, Edgar and H. C. Cutler. 1942. Races of Zea Mays: I. Their recognition and classification. Ann. Mo. Bot. Gard, 29 : 69-88. Arber, Agnes. 1934. The Gramineae: A study of cereal, bamboo, and grass. Cambridge, England. Brew, John Otis and E. B. Danson. 1948. The 1947 reconnaissance and the proposed Upper Gila expedition of the Peabody Museum of Harvard University. El Palachio 55: 211-222. Cutler, H.C. 1946. Races of maize in South America. Bot. Mus. Leafl. Harv. Univ. 12: 257-291. Cutler, H.C. and M.C. 1948. Studies on the structure of the maize plant. Ann. Mo. Bot. Gard. 35: 301-316. Hitchcock, A. S. 1935. Manual of the grasses of the United States. U.S. Dept. of Agriculture Mise. Publ. No. 200. Lenz, L. W. 1948. Comparative histology of the female inflorescence of Zea Mays L. Ann. Mo. Bot. Gard. 35: 353-376. Mangelsdorf, P. C. 1945. Origin and nature of the ear of maize. Bot. Mus. Leafl. Harv. Univ. 12: 33-75. Mangelsdorf, P. C. 1947. The origin and evolution of maize. In Ad- vances in Genetics, Vol. I, pp. 161-207. New York. Mangelsdorf, P. C. 1948. The rédle of pod corn in the origin and evolution of maize. Ann. Mo. Bot. Gard. 35: 377-406. Mangelsdorf, P. C. and J. W. Cameron. 1942. Western Guatemala; a secondary center of origin of cultivated maize varieties. Bot. Mus. Leafl. Harv. Univ. 10: 217-252. Mangelsdorf, P. C. and R. G. Reeves. 1939. The origin of Indian corn and its relatives. Texas Agr. Exper. Sta. Bull. 574. Powers, William EK. 1939. Basin and shore features of the extinct Lake San Augustin, New Mexico. Journ. Geomorphology 2: 345- 356. Sturtevant, E. L. 1894, Notes on maize. Bull. Torrey Bot. Club 21: 319-343, 503-523. [ 245 ] EXPLANATION OF THE ILLUSTRATIONS Prare XXIII. The largest cobs, about two thirds natural size, found in each of the six successive strata in the Bat Cave material. Although there has been a progressive increase in maximum size from stratum to stratum, the type found in the first stratum has persisted throughout the series. Pirate XXIV. A. Variations in kernel size in the six successive strata of the Bat Cave material, nat- ural size. The kernels, like the cobs, have increased in mean size from stratum to stratum, but the small- est size has persisted throughout the series. B. An artist’s reconstruction, natural size, of the primi- tive pod-pop corn found in the lowest stratum of the Bat Cave refuse heap. This primitive maize shows no evidence of having been derived from teosinte. C. The cob from Stratum I, natural size, upon which the reconstruction in B is based. Pirate XXV. A. A cob, natural size, with promi- nent horny glumes produced experimentally from a maize-teosinte hybrid. The cob has been soaked in sulphuric acid to simulate aging. B, C, D, E and F are cobs from the upper strata of the Bat Cave showing teosinte introgression, natural size. Pirate XXVI. Bat Cave cobs, natural size, show- ing individual teosinte characteristics. A, B, C. Distichous spikes from Strata IV, V and VI, re- spectively. D, E. Cobs bearing some single spike- lets, from Strata IV and V, respectively. F, G, H. Three cobs which are similar in their Jw alleles, but which differ in the degree of teosinte intro- gression. [ 246 ] Pirate XXVII. A fragment of husk found in Stra- tum II and the kind of ear which it must once have covered, about two thirds natural size. The husk consists of four leaf sheaths which show no evi- dence of ever having enclosed an ear. They proba- bly formed an involucre around the base of the ear. Pirate XXVIII. Diagrams illustrating the cross sections of cobs of Bat Cave maize and ears of mod- ern varieties. Upper Row. Four cobs from Bat Cave showing a progressive increase in the size of the rachis accompanied by a progressive increase in the row number and a decrease in the glume zone. Second Row. Diagrams of three degrees of half-tunicate in an isogenic stock showing how the lower alleles increase the diameter of the rachis, reduce the glume zone and shorten the distance between the rachis and the base of the kernels. Third Row. Diagrams of two degrees of tunicate in an isogenic stock illustrating the same phenom- ena described above. Fourth Row. Three ears of modern maize illustrating various stages in the evolutionary trend toward a massive rachis, short glumes and compact rachillae. [ 247 ] Piste “weil a PLATE XXIV XXV PLATE PuratE XXVI Pirate XXVII PLateE XXVIII tutu tutu CHAPALOTE HIDATSA MODERN DENT BOTANICAL MUSEUM LEAFLETS HARVARD UNIVERSITY CampripGr, Massacuusetts, Aprit 18, 1949 VO0ln 1o,-NO.o PLANTAE AUSTRO-AMERICANAE V DE PLANTIS PRINCIPALITER COLOMBIAE OBSERVATIONES BY RicHarp Evans SCHULTES A NUMBER of recently made collections, chiefly from Amazonian Colombia, represent hitherto undescribed concepts or range extensions or are the basis of interest- ing ethnobotanical observations. It appears advisable to present the following miscellany of notes on the signifi- cance of these collections as a contribution to our under- standing of the South American flora, especially that of the northwesternmost reaches of the Amazonian system. It is with pleasure that I thank Dr. B. A. Krukoffand Mr. Joseph Monachino of the New York Botanical Gar- den for determining the collections of Menispermaceae and Strychnos; Mr. Charles Schweinfurth of the Botan- ical Museum of Harvard University for identifying the orchids; and Dr. Adolpho Ducke of the Instituto A gro- nomico do Norte (Belém do Para) for naming the col- lection of Leptothyrsa Sprucei. ORCHIDACEAE Cattleya violacea (H BK.) Rolfe in Gard. Chron. ser. 8, 5 (1889) 802. Cattleya violacea is well known from Amazonian Bra- ail, from southern Venezuela and British Guiana. In [ 261 J Colombia, it has previously been collected (aught 2810) at San José de Ocune on the Rio Vichada, to the northeast of the locality of Schultes 5380. Cotompra: Comisaria del Vaupés, Macaya-Ajaju River confluence: Mount Chiribiquete. Quartzite base. Summit 800-1200 ft. above for- est floor, 1700-2100 ft. above sea level. ‘‘Epiphyte. Sepals deep vio- let, lip deep purple. Spicy fragrance.’’ May 15-16, 1943, Richard Evans Schultes 5380. Pleurothallis ciliata Knowles & Westcott Flor. Cab. 1 (1887) 39. This interesting little orchid has hitherto not been recorded from Colombia. It is known from Trinidad, British Guiana and Peru. Cotompra: Comisaria del Vaupés, Macaya-Ajaju River confluence: Mount Chiribiquete. Quartzite base. Summit 800-1200 ft. above for- est floor, 1700-2100 ft. above sea level. ‘‘Epiphytic on Hevea in for- est. All parts of flower yellow.’* May 15-16, 1943, Richard Evans Schultes 5418. Sobralia rosea Poeppig & Endlicher Nov. Gen. ac Sp. 1 (1886) 54, t. 98. Known from the eastern Andean slopes of Ecuador and Peru, Sobralia rosea has previously been reported from Colombia on the basis of a collection from Norte de Santander (Schlechter in Fedde Repert. Beihefte 7 (1920) 212). The locality of Schultes & Smith 3049, though very distant from the only other Colombian col- lection, fills in an extensive gap in our knowledge of the range of this beautiful orchid. Cotompra: Comisaria del Putumayo, Cerro de Portachuelo, entre Mocoa y Sacchamates, alt. ca. 1600-2000 m. “‘Large clumps with erect stems 8-12 ft. tall. Flowers white; lip fringed with purple.”’ December 9, 1942, Richard Evans Schultes & C. Earle Smith 3049,— Comisaria del Putumayo, Rio San Miguel 6 Sucumbios, Santa Rosa y los alrededores ; alt. 880 m.? ‘‘Flowers white, tips of column purple. ”’ April 7-8, 1942, Richard Evans Schultes 3559. [ 262 ] VELLOZIACEAE Vellozia phantasmagoria FP. I’. Schultes in Bot. Mus. Leafl. Harvard Univ. 12 (1946) 180, t. XDX, XX. The sterile collection cited below establishes the occur- rence of the curious V’ellozia phantasmagoria far to the west of its hitherto known range. It was found by Schultes on Cerro Chiribiquete (its type locality) in the upper Rio Apaporis drainage-area ; and by Paul H. Allen on Cerro Yapoboda, not far from Mitu, in the lower course of the Colombian Vaupés. Cotompra: Comisaria del Vaupés, Rio Nequieni (affluent of Rio Guainia), Cerro Monachi ‘‘ Hanging on rocky crags.’’ June 1948, Richard Evans Schultes & Francisco Lopez 10067. MENISPERMACEAE Abuta grandifolia (Mart.) Sandwith in Kew Bull. (1937) 397. Abuta grandifolia, widespread throughout the Ama- zon Valley and adjacent areas of northern South America, has previously been known from the Vaupés, Caqueta and Putumayo drainage-areas of Colombia (KKrukoff and Moldenke in Brittonia 8 (1988) 60; in Bull. Torrey Bot. Club 68 (1941) 240). The Siona Indians of the Putumayo utilize an infusion of the leaves of this plant in treating fevers. Cotomsia: Intendencia del Amazonas, Rio Putumayo, carretera entre Caucaya (Puerto Leguizamo) y La Tagua, alt. 225 m. ‘‘Small tree.’’ May 17, 1942, Richard Evans Schultes 3740. Abuta splendida Krukoff & Moldenke in Bull. Tor- rey Bot. Club 68 (1941) 241. Described from Bolivian material, 4buta splendida has not hitherto been known from the Colombian flora. Cotomsr1a: Comisaria del Putumayo, Rio San Miguel 6 Sucumbios, Nonejo y los alrededores, frente a la Quebrada Conejo. Nombre kofan : sa-pé-pa. 2-5 abril, 1942, Richard Evans Schultes 3525. [ 263 ] Chondodendron toxicoferum (Wedd.) Krukoff & Moldenke in Brittonia 3 (1939) 838. This species is well known as an ingredient of arrow- poison in the Amazon Region. Known from Amazonian Peru and Brazil (as Chondodendron polyanthum), it was first collected by Castlenau at Pebas in Loreto, Peru, where the Yaguas Indians were utilizing it in the man- ufacture of curare (KKrukoff and Moldenke in Brittonia 3 (1938) 22-23, 338). It has hitherto been collected in Colombia apparently only once: the collection W. Hox 12b, which states “‘poison used for blowpipes, ’’ is labelled as coming from ‘‘ Entre Rios,’’ a now non-existent set- tlement on the Rio Karaparana in the Intendencia del Amazonas in Witoto Indian country. CotomsBiA: Comisaria del Putumayo, Rio San Miguel 6 Sucumbios, Conejo y los alrededores, frente a la Quebrada Conejo. Nombre kofan: , ee rT . ss ~ P sa-pé-pa. Wood of stem crushed to make curare poison.’’ 2—5 abril, 1942, Richard Evans Schultes 3522.—Comisaria del Amazonas, Rio Hamacayacu. Common name: quina. [April] 1943, F. J. Hermann 11309. BALANOPHORACEAE Helosis guyannensis L. C. Richard in Mém. Mus. Hist. Nat. Paris 8 (1822) 416, t. 20. This is the second collection of Helosis guyannensis known from Amazonian Colombia. The first was re- ported by Schultes in Bot. Mus. Leafl. Harvard Univ. 12 (1946) 117. Cotompia: Intendencia del Amazonas, Rio Boiauassti. “‘Mata vir- gem. Pardsita. Toda a planta e encarnada clara.’’ Ocotober 29, 1946, George A. Black & Richard Evans Schultes 46-241. NYMPHAEACEAE Victoria regia Lindley Monogr. Vict. regia (1887) 8. Occurring naturally in Colombia probably only along the bank of the Rio Amazonas in the trapécio amazénico, [ 264 ] where indeed it is very common and abundant within easy walking distance from Leticia, the magnificent V7c- toria regia has hitherto apparently never been collected in Colombian territory. It has been imported from Brazil and is established in a small pond at Ia Chorrera on the upper Rio Igaraparana, an affluent of the Rio Putumayo. It is interesting to note that the Spanish-speaking in- habitants of the trapécio amazénico know Victoria regia as horno, just as in Brazil it is called forno. The enor- mous, flat, round leaves resemble the flat, round baked- clay plates or hornos on which farina is toasted and cassabe is baked; this is the explanation of the curious common name. Cotompara : Intendencia del Amazonas, La Chorrera, Rio I[garaparana. *“Cultivated.’’? June 1942, Richard Evans Schultes 8970A.—Intenden- cia del Amazonas, Trapécio amaz6nico, near mouth of Rio Boiauasst, in swampy inlet of Rio Amazonas on Isla Zancudo, alt. about 100 m. “Leaves spiny beneath, 4-5 feet across. Flowers large, white, turning red or purplish. Horno.’’ October 1945, Richard Evans Schultes 6786. —Intendencia del Amazonas, Trapécio amazénico, Leticia. ““Beira do rio em terreno enxarcado, dentro da agua.’’ August 19, 1946, George A. Black & Richard Evans Schultes 46-18. RUTACEAE Leptothyrsa Sprucei Hooker filius in Bentham & Hooker filius Gen. Pl. 1 (1862) 284. This rare species, described from material collected by Spruce on the Rio Uaupés of Brazil, has apparently never been recorded from Colombia. Cotomp1a: Intendencia del Amazonas, |Trapécio amazonico]. **Pi- cada Cotuhé, varial aberto, beira de um corrego. Arbusto, 2 m.; flor branea.’’ November 8, 1946, George A. Black & Richard Evans Schultes 46-364. EUPHORBIACEAE Cunuria crassipes Mueller-Argoviensis in Martius Fl. Bras. 11, pt. 2 (1874) 510. [ 265 | Cunuria crassipes was first collected in Colombia by Allen in 1948 (Baldwin and Schultes in Bot. Mus. Leaf. Harvard Univ. 12 (1947) 886). His collection was made on the Rio Papuri which forms part of the boundary between Brazil and Colombia. Later, Cunuria crassipes was found in the adjacent Brazilian area of the upper Rio Negro and some of its afluents by Baldwin and by Frées (Baldwin and Schultes, loc. cit.), never at any great dis- tance from the type locality. For this reason, the species was considered to represent a rather restricted endemic. The most significant collections made recently in the Colombian trapécio amazénico are Schultes 6909 and Black & Schultes 46-371, for they establish the presence of Cunuria crassipes very far to the south of the area hitherto considered to be the exclusive home of the spe- cies. In 1945, Schultes 6909, with several opened capsules but no seeds, was collected during a preliminary study of a low caatinga-like or savanna area on the rather high divide between the Rios Hamacayacu (Amazon affluent) and Cotuhé (Putumayo affluent). The next year, Mr. George A. Black of the Instituto Agronémico do Norte visited this locality and made the collection Black & Schultes 46-371. This collection was sterile, but the large, unusually broad-ovate shape, the relatively thin-papy- raceous texture, and the lack of a waxy or glaucescent condition of the upper surface of the leaves all suggested that it might represent a distinet form of Cunuria eras- sipes. The isolation of this station from the nearest known locality of the species seemed to increase the possibility that these differences might be of significance. Accord- ingly, in September 1948, I sent my assistant, Senor Francisco Lopez, thoroughly familiar with Cunuria, to collect more material from the same locality. Lépez 4 and 5 correspond in every detail with Scehultes 6909 and Black & Schultes 46-371 and are indeed slightly different [ 266 ] from the abundant material gathered from the upper Rio Negro drainage-area. According to Lopez, the tree is unusually abundant at the locality cited. Further collections of Cunuria crassipes have been made recently in the Rio Guainia area in Colombian territory, where the species is represented in unbelievable abundance in the sandy, light-forested caatingas where it is known by the Kuripaka Indian name d-hoon-da. In July 1948, Lopez and I crossed from the Rio Uaupés to the Rio Caqueta by way of the Tikié—Ira- Igarapé and the 'Taraira—Apaporis. We fully expected to find Cunuria crassipes in Colombian territory along the Taraira, but the species was not seen on the river; in fact, it was not seen south of the headwaters of the Ira-I[garapé, an affluent of the Rio Negro drainage area. I had never succeeded in finding Cunuria crassipes in the region of La Pedrera, although there are abundant areas there where the species might be expected to occur. In November 1948, I was fortunate in being able to ac- company General Julio Londofo, of the Colombian army, on a special military flight over the Amazonia of Colombia. In flying extremely low over the La Pedrera region, I saw, in a low-forested caatinga a few miles south of the beginning of the constriction of the Caqueta below the rapids at Cérdova (Cupati), what had all the appear- ances of numerous small trees of Cunuria crassipes. We cannot definitely assert that Cunuria crassipes occurs at this point on the Caqueté until actual collections are made, but we may say that it is highly probable that it is represented there. Cotomsia: Comisaria del Vaupés, Rio Papuri, vicinity of Piracuara Mission, alt. about 200 m. “‘Trees tall, slightly buttressed, averaging 30 m. in height and 60 cm. in diameter. Bark thin, averaging less than 1 em. Latex yellowish, scant, coagulating to a non-elastic gum. Flowers greenish yellow. Trispermate Hevea-like capsule. Known locally as wah-so-né-né in Tukano.’’ August 18, 1943, Paul H. Allen [ 267 ] 3068,.—Comisaria del Amazonas, Trapécio amazoénico. Interior regions of trapécio between Amazon and Putumayo water-sheds. Alt. above 100 m. ‘““Tree 50 ft. Spreading crown, Fruit like Hevea. Latex white.’’ November 1945, Richard Evans Schultes 6909.—Comisaria del Amazonas, Trapécio amazénico, Rio Hamacayacu, Quebrada Agua Negra. “‘Arvore 20 m. Latex branco.’’ November 8, 1946, George A. Black & Richard Evans Schultes 46-471 (Herb. Inst. Agron. Norte 20477).—Comisaria del Amazonas, Rio Hamacayacu, Varial at head- waters of Quebrada Agua Negra. ““Small-crowned tree 50 ft. tall. Latex white. Buttress roots slender, low. Bark shaggy, externally dark brown, internally reddish.’’ September 1948, Francisco Lopez 4. —Comisaria del Amazonas, Rio Hamacayacu, Varial at headwaters of Quebrada Agua Negra. ‘‘Buttressed tree 50 ft. tall; diameter of columnar trunk 15 inches. Latex rather thick, white. Bark dark brown.’’ September 1948, Francisco Lopez 5.—Comisaria del Vaupés, Rio Guainia, region near Sejal. “Slightly buttressed tree 45-50 ft. tall. Bark shaggy, dark brown. Latex thick, white. Wood white.’’ June 1948, Richard Evans Schultes & Francisco Lopez 10043,—Comisaria del Vaupés, same locality. ““Tree 60 ft. tall with sparse crown. Bark shaggy, dark brown outside, pinkish inside. Fruits rose-red. Latex very thick and white. Leaves very coriaceous.’’ June 1948, Richard Evans Schultes & Francisco Lopez 10058.—Comisaria del Vaupés, Rio Naquieni (affluent of Rio Guainia), at base of Cerro Monachi. “‘Tree 40 ft. tall with small buttress roots. Bark thin, brown, roughish. Latex thick, white. Leaves unusually ovate.’? June 1948, Richard Evans Schultes § Francisco Lopez 10088.—Comisaria del Vaupés, same locality. “‘Tree 50 ft. tall. Buttress roots. Trunk columnar, diameter 14-15 inches. Latex white, thick. Bark dark brown, rough.’’ June 1948, Richard Evans Schultes & Francisco Lépez 10113.—Comisaria del Vaupés, same locality. ““Tree 45 ft. tall with sparse crown. Latex white.’’ June 1948, Richard Evans Schultes & Francisco Lopez 10121. —Comisaria del Vaupés, Rio Guainia, Cafio Nani below San Antonio. In extensive caatinga. ‘‘Tree 50 ft. tall with buttress roots and col- umnar trunk. Bark shaggy, dark brown outside, light red inside. Latex white. Flowers yellowish.’? June 1948, Richard Evans Schultes & Francisco Lopez 10131.—Comisaria del Vaupés, same locality. In extensive savannah-like caatinga. ‘Bushy tree, 20 ft. tall with very small buttress roots. Bark dark brown, rough. Latex very thick, white. Leaves unusually round.’’? June 1948, Richard Evans Schultes § Francisco Lopez 10158. Cunuria Spruceana Baillon in Adansonia 4 (1863- 64) 288. Cunuria Spruceana was first collected in Colombia by [ 268 ] Allen in 1943, along the Rio Papuri, an affluent of the Rio Vaupés; in the following year, I collected it from La Pedrera on the Rio Caqueta and on the Rio Miriti- parand (Baldwin and Schultes loc. cit. 842; Schultes in Caldasia 8 (1945) 247-249). In 1948, more material was collected from La Pedrera. A most important discovery, however, is that of Lopez who, in September 1948, collected Cunuria Spruceana from the headwaters of the Rio Hamacayacu, an affluent of the Amazon above Leticia. Later, L6opez and I located a number of trees of this species in the immediate envi- rons of the town of Leticia. Known from a locality in adjacent Peru and from points on the Ica and at Sao Paulo de Olivenca on the Amazonas itself, Cunuria Spruceana was to be expected near Leticia, even though several years of intensive collecting had previously failed to disclose its presence there. In the trapécio amazénico, this species occurs sporadically, not in abundance as often is the case in other areas. Cunuria Spruceana has no dis- tinctive common name in the Spanish, Portuguese or Tikuna languages in the Leticia area, so far as we could learn. Cotoms1A : Comisaria del Vaupés, Rio Papuri, vicinity of Santa Tere- sita. ““Tall trees averaging 35 m. in height, and 80 cm. in diameter above the conspicuously developed and unique stilt buttresses which are produced toa height of 3-4 m. as laterally compressed board-like flanges which act as flying buttresses, the bases widely detached from the trunk, the arch often high enough for a man to walk upright be- neath. Leaves simple, with two basal disk-like glands. Inflorescence of small green flowers from axillary growth. Large trispermate capsule typical of Hevea. Bark very thin, 6-8 mm., with scanty latex which coagulates with difficulty, producing a non-elastic gum. Seeds col- lected by the local Indians for food. Known as wah-puh (Tukano).”’ August 15, 1943, Paul H. Allen 3063,—Comisaria del Amazonas, Rio Hamacayacu. ““Tree 80 ft. tall with huge buttress roots. Flowers yellowish. Latex very thin, white. Bark hard, brownish, thin. Wood white.’’ September 1948, Francisco Lépez 1.—Comisaria del Ama- zonas, Rio Hamacayacu. ““Tree 60 ft. tall with large buttress roots. [ 269 ] Flowers yellow. Latex white, watery. Bark thin and brittle.’’ Sep- tember 1948, Francisco Lopez 2.—Comisaria del Amazonas, Leticia. *“*Large tree with large buttress roots. Trunk columnar above buttress roots. Flowers yellowish. Latex watery, white.’’ September 1948, Francisco Lopez 3,—-Comisaria del Vaupés, Rio Guainia, near Sejal. **Enormous tree, 80 ft. tall with large flaring buttress roots. Latex thin, white. Bark thin, hard, dark tan, smooth. Buds yellowish.’’ June 1948, Richard Evans Schultes & Francisco Lépez 10057 .—-Comis- aria del Vaupés, Rio Guainia, at base of Cerro Monachi. ‘‘Enormous tree 85 ft. tall with extensive buttress roots. Latex white, thin.’’ June 1948, Richard Evans Schultes & Francisco Lopez 10135.—Comis- aria del Vaupés, Rio Guainia, Cafio Nani below San Antonio, ‘‘But- tressed tree 75 ft. tall. Trunk columnar, diameter 18—20 inches. Latex watery, white, sparse.”’ June 1948, Richard Evans Schultes & Francisco Lopez 10163.—-Comisaria del Caqueta, Rio Caqueta, La Pe- drera, at Angostura. ‘“Tree 80 ft. tall; huge buttresses; trunk col- umnar, 2 ft. diameter above buttresses. Bark hard, reddish brown. Latex very thin, white, not coagulating. Flowers yellow.’’ July 16, 1948. Richard Evans Schultes & Francisco Lopez 10221.—Comisaria del Caqueté, Rio Caqueta, La Pedrera, at Angostura. ‘‘Corpulent tree 55 ft. tall, diameter above buttresses 15 inches. Bark hard, thin, red- dish brown. Latex thin, watery, white, not coagulating. Flowers yellow.”’? July 16, 1948, Richard Evans Schultes & Francisco Lépez 10222.—Comisaria del Amazonas, Leticia. “‘Corpulent tree 75 ft. tall. Extensive buttress roots up to 6 ft. high. Bark dark brownish- tan, thin, hard. Flowers yellow.’ November 1948, Richard Evans Schultes & Francisco Lopez 10257. Senefeldera chiribiquetensis PR. 1. Schultes & L. Croizat in Caldasia 8 (1944) 122. Matching the type and topotype material of Senefeld- era chiribiquetensis pertectly, Cuatrecasas 7682 establishes the occurrence of this curious endemic slightly north- westward of the type locality which is the confluence of the Rios Macaya and Ajaju in the Apaporis drainage- area. More interesting, however, is the discovery, through Cuatrecasas 7682, that Senefeldera chiribiquetensis, hith- erto believed to be confined to cretaceous quartzite hills, also occurs on granite of a much earlier (probably Cam- brian) age. A search for Senefeldera chiribiquetensis on a number [ 270 | of granitic peaks in the upper Rio Negro and lower Rio Uaupés in Brazil (of the same age as the granite at San José del Guaviare) has indicated that it is apparently not present on Serra Curicuriari, Serra Cocui and similar mountains of the area. Indeed, during my explorations in the upper Rio Negro basin in 1947-48, I found no species of Senefeldera whatsoever. Cotomsra: “‘Rio Guaviare, San José del Guaviare. Terrenos gran- iticos, sabana, 270 m. alt. Arbol latex.’’ 12 septiembre, 1939, J. Cuatrecasas 7682 ( U.S. Nat. Herb. 1774175). HIppocRATEACEAE Tontelea attenuata Miers in Trans. Linn. Soe. London Bot. 28 (1872) 884. According to A.C.Smith (in Brittonia 8 (1940) 474), this beautiful, low tree is known from the original Spruce collections in the Rio Negro country of Brazil, from the basin of the Rio Madeira of Brazil, and from British Guiana and eastern Peru and Colombia. Schultes 5385 represents the second collection from Colombia. The first was Klug 1967 from the Putumayo, a great distance from the Vaupés locality cited below. Cotompra: Comisaria del Vaupés, Macaya—Ajaju River confluence, Mount Chiribiquete. Quartzite base. Summit 800-1200 ft. above forest floor, 1700-2100 ft. above sea level. ‘‘Small tree. Flowers white, probably turning brown. Odour of cinnamon pronounced. Bark smooth, brown. Along banks.’’ May 15-16, 1943, Richard Evans Schultes 5386. SAPINDACEAE Paullinia scaberula FP. Ll’. Schultes sp. nov. Vitis parva. Rami graciles, fusco-rufo cum cortice et pilis nigris debilibus usque ad 6 mm. longis et dense saetigeribus armati. Ramuli subteretes, similes. Folia quinquepinnata, usque ad 80-85 cm. longa, 25 cm. lata. Rhachis nuda (exalata), firma, scaberula nigris cum pills, [ 271 ] EXPLANATION OF THE ILLUSTRATIONS Prare XXIX,. (Upper figure), Basal portion of the trunk of Septotheca Tessmannii, the individual from which the collection Schultes & Black 8612 was made, (Lower figure). View in a caatinga forest on the Rio Naquieni, Comisaria del Vaupés, Colombia. The trees with stilt roots are Cunuria crassipes; the slen- der, columnar trunks represent Hevea rigidjfolia. Oo SO a plerumque 16 cm. longa, 2 mm. in diametro. Foliola el- liptica, 10-15 em. longa, 5.5-7.5 cm. lata; superiora aequalia, apice acuminata, margine longe et leviter sin- uata, valde membranacea, utrinque aspero-pilosa, atro- viridia, lateralibus cum nervis plerumque duodecim supra et subtus elevatis, nervis praecipuis et secundariis utrinque pilosis; petioli nudi. Inflorescentiae axillares, racemi- formes, multiflorae, 6-8 cm. longae. Pedicelli robusti, 2-4 mm. longi, bracteis minutissimis, triangulari-subula- tis, 0.5 mm. longis. Flores parvi, albi. Sepala quinque ; duo exteriora crassa, extus et intus glabra, subovalia, apice rotundata, margine ciliata, 1.5 mm. longa, 1.3mm. lata; tria interiora majora, membranacea, rotundata, in- tus et extus glabra, fere 2.5 mm. longa, 2 mm. lata, valde inflexa, margine ciliata. Tori glandulosi quattuor, crassi, linguiformes. Petala quattuor, submembranacea, integra, obovata, apice rotundata, glabra, 3-38.5 mm. longa, 1.5 mm. lata. Stamina octo, ex floribus prominenter exserta. Staminum filamenta pilis longioribus lanatisque dense vestita, 1.5-2 mm. longa. Antherae glabrae, 0.4 mm. longae. Ovarium obovoideum, 1 mm. in diametro, glab- rum, cum stylo glabro 1 mm. longo coronatum. Stigma profundissime trifidum, lobis crassis, pustulatis et circin- natis. Fructus capsularis, trivalvis, subdrupaceus; cap- sula oblique ovoidea, in stipitem attenuata, rostrata (styl reliquiis coronata), ecostata et exalata, epicarpio parum crasso, praeter stipitis basim glabra, 6-7 mm. longa, 3-4 mm. lata, rostra usque ad 2 mm. longa. This new species is unusual in aspect and is distin- guished easily by its having black scaberulent hairs along the branchlets. Brazit: Estado do Amazonas, Sao Paulo de Olivenca. ‘‘Mata da terra firme. Cipozinho fino, caulifloro, flor branea.’’ April 3, 1944, A, Ducke 1628 (Tyrer in Herb. Inst. Agron. Norte, Belém). [ 275 ] BOMBACACEAE Septotheca Tessmannii Ulbrich in Notizbl. Bot. Gart. Berlin 9 (1924) 129, t. 3. This gigantic, upper Amazonian tree, very rare indeed in collections, has hitherto only been observed, never collected, in Colombian territory (Schultes in Bot. Mus. Leafl. Harvard Univ. 12 (1946) 118). The two collec- tions cited below establish beyond any doubt that this concept, described from material from the Rio Ucayali in Loreto, Peru, and found by Ducke and by Schultes near Tabatinga, Amazonas, Brazil, a few miles from Leticia (Ducke in Bol. Técn. Inst. Agron. Norte 4 (1945) 21-22), is a prominent element of the vegetation of certain areas of the Amazon band of the Colombian trapécio amazonico and must be added to our enumera- tion of Colombian plants. In November 1948, I returned to the locality of these two collections and discovered that the colony of trees from which they were gathered no longer exists. The current of the Amazon at this point has recently wrought such destruction by washing away several hundred feet of bank that the mouth of the Boiauasst has a different configuration than formerly. (Plate X XLX, upper fig.) Cotompia: Intendencia del Amazonas, Trapécio amazénico, mouth of the Rio Boiauasst, alt. about 100 m. ‘‘Enormous tree; buttress roots. Flowers yellowish green, mucilaginous. Sapotilla.’’ October 1945, Richard Evans Schultes 6888.—Same locality. “‘Sepals green- yellow. Tree 80 ft. tall, great buttress roots. Zapoterana.’’ October 27, 1946, Richard Evans Schultes §& George A. Black 8612. STERCULIACEAE Herrania albiflora Goudot in Ann. Sci. Nat. sér. 3, 2 (1844) 230, t. 5, figs. 1-10. The specimen cited below, one of the earliest collec- tions of the species, was identified as Herrania albiflora [ 276 ] at Kew by Planchon, who annotated the sheet. It is a topotype. The report of the collector of this specimen relative to his having seen ‘‘several hundreds of fruits’’ on a culti- rated tree of Herrania albiflora must be recognized as an exaggeration. Although fruits are often very numerous in most species of Herrania, | have never met, in my field studies of the genus, any condition which would in- dicate ‘‘hundreds of fruits.” CotomsiA : Departamento de Cundinamarca, **Rio Guaco, near Muzo. Cacao simaron vg. name, in some districts no other cacao is known than this; the arillus around the seed is much esteemed for its deli- cious flavour and sweetness; it bears abundantly, on one tree (culti- vated) I saw several hundreds of fruits but ina green state. I believe it to be a hardier species than 7’. cacao as it is a native of the lower mountain, while 7’. cacao is a native of the hotter plains. This species is often of the same habit and size as 7’. cacao. Hab. in dense woods .... Of this I was able to obtain ripe fruit.’’ December 1845, Wil- liam Purdie sine num. (Tororyre in Herb. Kew). Herrania kanukuensis PR. 1’. Schu/tes in Caldasia no. 6 (1943) 11. A duplicate type of Herrania kanukuensis at Kew (A.C. Smith 3541) has several leaflets which tend to be slightly irregularly dissected, suggesting a condition which approaches that of some specimens of HZ. lemnis- eata in which the leaflets are abnormally developed. In the original description of Herrania kanukuensis, the leaflets are described as ‘‘margine subintegra, saepe prope apicem subundulato-sinuato.’’ In their fruit, however, these two concepts are clearly distinct. It is obvious that Herrania lemniscata represents a link between HZ. lacini- ifolia of Colombia and HZ. kanukuensis of British Guiana which is, in turn, probably allied to the several Ama- zonian species. Herrania laciniifolia Goudot ex Triana & Plan- eh EXPLANATION OF THE ILLUSTRATION Piatre XXX. Schomburgk’s field painting of Lightia lemniscata, tab. XLI in the Schomburgk collection of water colors in the British Museum. [ 278 ] PEATE Nous. Oo chon apud Garcia-Barriga in Caldasia no. 2 (1941) 55, t. 1, 4. Collections of Herrania laciniufolia are not frequent. 1 It is unfortunate that a precise locality for MKalbreyer 2047 is not available. Cotompia: ‘‘New Granada, 4000 ft. Shrub, stem simple 10-12 ft. If., yellowish, woody, rough, 34 ft. 1. 4 br., fl. on the stem below, red brown with bunch of greenish white petals, 7-8 in., threadlike, brownish white, pendulous.’? May 18, 1881, Kalbreyer 2047 (Herb. Kew, 2 spec.). Herrania lemniscata (Schomb.) R. EH. Schultes in Caldasia no. 6 (1948) 18. In 1943, when I validated the Schomburgk concept and made the new combination Herrania lemniscata, European collections were unavailable for study. Re- cently, additional and extensive material which I have seen in England indicates that Herrania lemniscata is a very distinct species. Schomburgk’s original water colors, made in the field, are preserved at the British Museum of Natural History. Included in the collection is an excellent painting (tab. XLI) of this concept (as Lightia lemniscata) which is herein reproduced because it depicts with unusual accuracy the habit of the small tree. (Plate XXX) Furthermore, the material at Kew has excellent fruit- ing specimens, so that it is now possible to describe the capsule: Fructus longe pedunculatus (pedunculus robustus, glaber vel glabrescens, usque ad 4.5 cm. longus, 8 mm. in diametro), perfecte ovoideus, 7 cm. longus, 4 cm. in diametro, apice brevissime et abrupte cuspidatus, basi subrotundatus, omnino dense et minute stellato-pilosius- culus, sine pilis urticantibus ut videtur, decem-costatus quinque cum costis primariis, angustis subhumilibusque, 2 vel 4 (siccitate) mm. altis, 1-2 mm. crassis, subcultri- [ 281 ] formibus, atque quinque cum costis secundariis similibus, minoribus, pericarpio inter costas longitudinales valde reticulato-costato tenuissimo, maturitate probabiliter luteo. Semina plura quam sexaginta ut videtur, in cir- cuitu triangularia vel angulato-ovata, complanata, 12x 11X11 mm., plusminusve 4 mm. crassa, in pulpa alba inclusa. Of the specimens cited below, fruits in mature condi- tion are preserved in im Thurn sine num. (which I have chosen as the type of the fruit), Rusby & Squires 252 and Forest Department Brit. Guiana Record No. 4500. The last of these three collections has a 6-costate fruit, an un- usual and probably abnormal departure from the 5-parted condition of the genus. Bririsu Gurana: “Banks of Barima River’? Schomburgk (Herb. Kew, 3 spec.)—|Everard] im Thurn sine num. (Herb. Kew, fructus typus, 4 spec.).—‘‘Upper Arawau River, Aruka River, N.W.D. Mixed forest in swamp and on rather swampy soil. Small tree, 20-30 ft., but told by Indians it grows larger. Trunk 4 in, diam. at 6 ft. from ground. Leaves alternate, large, compound, palmate, 6-lobed in specimens taken. Slender paired stipules at base. Flowers cauliferous at base of trunk. Arawak dialect: bur-oo-ma.’’ May 1929, Martyn 61. (Herb. Kew, 2 spec.).—‘*‘Essequibo River, White Creek, Groete Ck. 15 ft. high, soft-wooded tree from clearing with terminal, peltate, long- stalked lvs., pubescent below, br. velvety on petiole and midrib be- neath; fr. cauliflorus, fascicled, 1-4 together, drooping, long-stalked ovoid-globose, acuminate with 6 major and 6 minor ribs and numerous lesser oblique ridges between ribs, greenish, the ribs darker, thinly bony with an inner pithy layer, seeds forming a soft, white oblong- globose mass divided into 6 segments shaped like orange flakes, each flake consisting of 3 seeds embedded in thin, sugary pulp with a very sticky coating to testa—testa pale brown finely pitted, wrinkled and veined—seed trigonous.’’ Forest Dept. British Guiana Field No. F1764, Record No. 4500 (Herb. Kew). Venezugkta: *‘Catalina, Lower Orinoco’? 1896, H. H. Rusby & Roy W., Squires 252 (Herb. Kew, 2 spec.). Herrania purpurea (Pittier) R. 1. Schultes in Cald- asia no. 9 (1944) 8383. The following collections of Herrania purpurea em- Ea phasize the common occurrence of this representative of a South American genus in adjacent areas of Central America, as far north as the Republic of Nicaragua. The species appears to be concentrated especially in Panama. The Colombian collection cited below is the second collection from Colombia and the South American con- tinent. The first, Schultes 5754, (reported in Caldasia no. 11 (1944) 28, t. 1, 2), was from the region near Turbo. Schultes & Lopez 10464, while from the same general area (the Golfo de Uraba), places Herrania purpurea much farther to the south. Cotomsi1a: Departamento de Antioquia, Golfo de Uraba, Villa Arte- aga, Municipio de Pavorandocito. **Small tree 8 ft. tall.’ December 1948, Richard Evans Schultes & Francisco Lopez 10464. NicaraGua: 1867, Seemann sine num. (Herb. Mus. Brit.). Panama: ‘‘Chiriqui’’ [Rec’d.] June 1886, J. H. Hart sine num. (Herb. Kew). Sutton Hayes sine num. (Herb. Kew; Herb. Mus. Brit. ) Canal Zone, near Madden Dam, “‘near sea level. Forest. Shrub, simple stem 10 ft.; fruits yellow.’’ June 9, 1939, 4. G. H. Alston 8861 (Herb. Mus. Brit.; Herb. Gray). DILLENIACEAE Saurauia rigidissima R. 7. Schultes sp. nov. Arbor parva et pulcherrima ut videtur. Rami lignosi strigillosique cum pilis subhamulatis, fusco-ferrugineis, 4 mm. longis dense obtecti. Folia rigide coriacea, obo- vata, apice rotundata, basi cuneata, margine remotissime subdenticulata, 29-41 em. longa, 12.5—16 cm. lata; infra (siccitate) aureo-brunnea, dense et minutissime granu- losa, mMaximopere minutissime et subremote stellato- pilosa; supra (siccitate) brunneo-viridia, subnitida; venis crassis, hamulato, strigosis, supra planis, infra promi- nenter elevatis; petioli robustissimi, hamulato-strigosi, 4.5 em. longi, 5 mm. crassi. Inflorescentiae erectae, foliis aequales vel longiores, multiflorae, dense paniculatae ; bracteae pedicellos subtendentes minutissimae, triangu- [ 283 ] lares, 0.7 mm. longae, hispidulae. Alabastra globosa. Flores parvi, aperti 8-10 mm. in diametro. Sepala quin- que, inaequalia; exteriora tria membranacea, ovata, apice rotundata, 5-6 mm. longa, 8 mm. lata, glaberrima, alba, inflexis cum marginibus; interiora duo membranacea, suborbicularia, 5.5 mm. longa, 4.5 mm. lata, alba. Petala quinque, membranacea, alba, rotundato-ovata, 6 mm. longa, 4.5 mm. lata. Stamina circiter 28, basi albo-bar- bata; filamenta 2 mm. longa. Antherae plusminusve 1.8 mm. longae. Ovarium globosum, stylis quinque filiformi- bus, 8 mm. longis, apice in stigmata leviter dilatatis. Fructus maturus adhue ignotus. The peculiar fragile texture of the highly coriaceous and rigid leaf of this new species distinguishes Saurauia rigidissima from the other known Colombian members of the genus. According to the field notes, the leaves are fragile even in life. There are also minor distinguishing characters in the hairs of the petiole and branch. Cotompia: Departamento del Valle, “‘Cordillera Occidental, Los Farallones, vertiente oriental: abajo de Almorzadero, bosque 2850- 2880 m. alt. Arbolito. Hoja grande, coriacea, rigida, fragil, verde grisacea haz, verde claro envés. Caliz verde padlido. Corola blanca,’’ July 25, 1946, J. Cuatrecasas 21716 (Tyrer in Herb. Gray). FLACOURTIACEAE Mayna amazonica (Martius ex Hichler) J. F. Macbride in Field Mus. Publ. Bot. 18, pt. 4 (1941) 16. Rather widespread in the Amazon drainage-area, Mayna amazonica extends up to the very slopes of the Andes in eastern Colombia. This is the second collection from the Comisaria del Putumayo (Schultes in Caldasia 3 (1945) 439). Among the Kofiin Indians, the bark of the trunk of this small tree is used as an ingredient of curare. Cotomsia: Comisaria del Putumayo, Rio San Miguel 6 Sucumbios, [ 284 | Conejo y alrededores frente a la Quebrada Conejo, alt. 300 m. ‘‘Nombre kofan: tzd-he-vee-ko. Tree.’’ 2-5 abril, 1942, Richard Evans Schultes 3515. Mayna pacifica Cuatrecasas var. pusilla R. EF. Schultes var. nov. Frutex parvus, usque ad tres pedes altus, ab Mayna pacifica specialiter foliis minoribus (usque ad 85 cm. longis et 11 cm. latis), infra glabris atque fructu minore (solum 4 em. longo et 2.5-8 cm. in diametro) cum alis membranaceis profunde fimbriato-dentatis differt. In provincia Choc6é nomen vulgaris: ‘‘clavo.”’ This newly described variety differs from Mayna pa- cifica Cuartr. in being much smaller in all its parts and, most conspicuously, in having elaborate and deeply fim- briate wings on the fruit. In Mayna pacifica, the fruit is two and one half times as long as in var. pusilla and the wings are strongly and irregularly dentate, not approach- ing the echino-fimbriate condition found in var. pusilla. Mayna pacifica var. brachycarpa Cuatr. is also larger than var. pusilla, and the wings of its fruit are barely conspicuous. Mayna pacifica and the var. brachycarpa are native to the Pacific slopes of the cordillera in the Departamentos del Cauca and Valle, respectively, slightly south of the general region of Quibd6, where the type of var. pusilla was collected. Cotomsp1a: Intendencia del Chocé, Headwaters of Rio Tutunend6, east of Quibd6. “*Clavo. Slender shrub 2-3 ft. Flw. and fruit white. Flw. faintly scented.’’ May 20-21, 1931, W. A. Archer 2195. (Tyre in U.S. Nat. Herb. No. 1518794). THYMELEACEAE Schoenobiblus peruvianus Standley in Field Mus. Publ. Bot. 11 (1936) 169. [ 285 ] EXPLANATION OF THE ILLUSTRATION PLare XXXI. Scuornoprp_us peruvianus Standley. 1, fruiting branch, approximately one half natural size. 2, fruit, approximately natural size. [ 286 ] PLaTE XXXI SCHOENOBIBLUS peruvianius Si za nde Schoenobiblus peruvianus has hitherto not been re- corded from Colombia. The only other representative of this genus in Colombia is Schoenobiblus coriaceus Domke, which has been found on Sierra Santa Marta at an altitude of 4,500 feet. Schoenobiblus peruvianus 1s avery important ingredient in the arrow-poisons of the Kofin Indians of the Putu- mayo. Both the fruits and the root are employed. This is apparently the first report of a member of the Thymel- eaceae as an ingredient of curare. Cotomsra: Comisaria de] Putumayo, Rio San Miguel 6 Sucumbios, Conejo y alrededores, frente a la Quebrada Conejo, alt. 300 m, “Large shrub. Root furnishes one of the poisons for curare.’’ 2-5 abril, 1942, Richard Evans Schultes 3486.—Same locality. ‘Small shrub. Poison: crushed fruits an ingredient of curare.’’? 2-5 abril, 1942, Richard Evans Schultes 3521.—Same locality. ‘‘Poison. Fruits used in making curare. Shrub.’’ 2-5 abril, 1942, Richard Evans Schultes 3655,—Comis- aria del Putumayo, Rio San Miguel 6 Sucumbios, Santa Rosa y los al- rededores, alt. 380 m. (?). “‘Roots and fruits used in Kofaén curare preparation.’’ 7-8 abril, 1942, Richard Evans Schultes 3613 SAPOTACEAE Lucuma dolichophylla Standley ex Llewelyn Wil- liams (sine diagn. lat.) in Field Mus. Publ. Bot. 15 (1936) 411. Arbor 45-70 pedes alta, recto cum trunco plusminusve cylindrico, 8-20 poll. in diametro, ramo inferiore 12- 35 pedes ab humo. Cortex extus atropurpureo-fuscus, dulce cum latice, intus grosse fibrosus. Flores parvi al- bique. Fructus rotundatus, indumento brunneo velu- tinoque maturitate lutescenti obtectus, esculentus. This concept was described in English without a Latin diagnosis in Williams’ ‘* Woods of northeastern Peru.” The name is here validated by rendering the English description into Latin. [ 289 ] LOGANIACEAE Strychnos brachiata Ruiz & Pavon FI. Peru. et Chil. 2 (1799) 80. This species, known apparently only from subandean Peru and Bolivia (Krukoff and Monachino in Brittonia 4 (1942) 269), is now registered from Colombia (Krukoff and Monachino in Bol. Téen. Inst. Agron. Norte 12 (1947) 10) on the basis of the following collection. CotomsB1A: Comisaria del Putumayo, Rio San Miguel 6 Sucumbios, Santa Rosa y los alrededores. Nombre kofan: se-hé-pa. ‘“Vine. Root used in preparation of arrow poison.’’ 7—8 abril, 1942, Richard Evans Schultes 3602. Strychnos Erichsonii Richard Schomburgh Faun. FI. Br. Guiana (1848) 1082, nomen; ex Progel in Mar- tius Fl. Bras., 6, pt. 1 (1868) 274. An Amazonian species, Strychnos Hrichsoni was col- lected in Brazil (Ducke in Bol. Técn. Inst. Agron. Norte 83 (1945) 14) near the Colombian frontier, and was re- ported by Krukoff and Monachino (in Caldasia 4 (1946) 46) as one of the species to be expected in the Amazon- ian regions of Colombia. The collection cited below es- tablishes the occurrence of the species in the southwest- ernmost sector of the Colombian Amazon area. It is known also from British and Dutch Guiana and Vene- zuela. Cotoms1a: Comisaria del Putumayo, Rio San Miguel 6 Sucumbios, Conejo y los alrededores, en frente de la Quebrada Conejo. Nombre kofan: ir-rro-chee, 2-5 abril, 1942, Richard Evans Schultes 3524. Strychnos Jobertiana Bail/lon in Adansonia 12 (1879) 367. This species of Strychnos has apparently not been col- lected hitherto from Colombia. Known from Amazonian Peru, Ecuador and Brazil, it was first indicated as prob- [ 290 ] ably occurring also in Colombia by Krukoff and Mona- chino in Caldasia 4 (1946) 45. The Canelos Indians of Ecuador use the bark of the stem and roots in preparing curare (Krukoff and Monachino in Brittonia 4 (1942) 282). Cotomsra: Comisaria del Putumayo, Rio San Miguel 6 Sucumbios, Conejo y los alrededores, frente a la Quebrada Conejo. Nombre ko- fan: ir-rro-chee. ‘‘Bark of root used in curare formula.’’ 2-5 abril, 1942, Richard Evans Schultes 3523.—Intendencia del Amazonas, Rio Loretoyacu, November 1946, George A. Black & Richard Evans Schultes 46-262. Strychnos Peckii B. L. Robinson in Proc. Am. Acad. 49 (1918) 504. This collection establishes the occurrence of Strychnos Peckii in the Colombian Amazonas (Krukoff and Mona- chino in Bol. Téen. Inst. Agron. Norte 12 (1947) 18). Hitherto known only from Colombia through one col- lection from the Valle del Cauca (Krukoffand Monachino in Lloydia 9 (1946) 65), it was to be expected as an ele- ment of the Amazonian flora. It ranges very widely, from Guatemala to Brazil and Bolivia (Ducke loc. cit. 14). Cotompra: Comisaria del Putumayo, Rio San Miguel 6 Sucumbios, Santa Rosa y los alrededores. Nombre kofan: se-hé-pa. ‘*Root used in curare preparation. Small tree.’’ 7-8 abril, 1942, Richard Evans Schulles 3601. Strychnos rondeletioides Spruce ev Bentham Journ. Linn. Soc. 1 (1856) 104. Strychnos rondeletioides has been known from Colom- bia hitherto through material collected along the Rio Vaupés (Krukoff and Monachino in Caldasia 4 (1946) 43). Schultes 8829 was collected in the southernmost sector of the Colombian Amazon drainage area. ‘The species also occurs in Amazonian Bolivia, Brazil and Peru, as well as in Venezuela (Ducke loc. cit. 11). { 291 | Cotomsia: Intendencia del Amazonas, Rio Karaparand, entre las boecas y El Encanto, alt. ca. 150 m. “‘Bush. Fruit blue-black or orange-red.’’ 22-28 mayo, 1942, Richard Evans Schultes 3829. SOLANACEAE Solanum apaporanum P. LE. Schultes sp. nov. Since I have been unable to find any material of Sola- num from the Amazon Valley to match this curious concept, I am describing it as new. Fruticulus scandens. Rami teretes, glabri, spinosi, tenul cum cortice cinereo-fusco. Ramuli sordide fusco- pulverulenti. Folia mediocria, brevissime _ petiolata, subchartacea, elliptica, integra, apice longissime acumi- nata, basi cuneata, aequalia, 38.5—-7.5 cm. longa, 1.5—2.3 em. lata, supra sordide fusco-viridia, aspera, omnino dense et minute stellato-pilosa, subtus viridia pallida, densissime stellato-pilosa, nervo centrali retrorso-spinoso ; petiol 2-4 mm. longi. Inflorescentia umbellata, pauci- flora, pedicellis comparate robustis, rectis, 7-10 mm. longis, densissime fusco-pilosis. Alabastrum 4 mm. lon- gum. Calyx omnino densissime stellato-pilosus, lobis breviter acutis, stellato-pilosis. Corolla profunde lobata extus stellato-pilosa, lobis anguste triangulari-acutis, 8 mm. longis, basi 2 mm. latis, membranaceis, albis, valde patentibus. Antherae conspicuae, flavae, 6-8 longae, apice poriferae. Bacca perfecte globosa (sine rostro), vivo 1—1.8 cm. in diametro, nitidissima, aureo-rubens. Cotompra: Comisaria del Vaupés, Macaya-Ajaju River confluence. Mount Chiribiquete. Quartzite base. Summit 800-1200 ft. above forest floor, 1700-2100 ft. above sea level. ‘‘Spiny vine-shrub. Fruit orange. Flowers white, anthers yellow.’’ May 15-16, 1943, Richard Evans Schultes 5406. (Tyrer in Herb. Gray.) [ 292 | BOTANICAL MUSEUM LEAFLETS HARVARD UNIVERSITY CAMBRIDGE, MassacuuseTts, JUNE 24, 1949 VoL. 13, No. 10 PLANTAE COLOMBIANAE XI DE PLANTIS PRINCIPALITER COLOMBIAE VALLIS AMAZONICAE OBSERVATIONES Bye RicHarp Evans ScHuutes’ Stupy of recently made collections of plants, chiefly from the Amazon watershed of Colombia, has brought to light a number of data which contribute to our knowl- edge of the flora of the northwesternmost reaches of the Amazon system. May 22-28, 1942, Richard Evans Schultes 3845. [ 296 ] Habenaria monorrhiza (Sw.) Reichenbach fil. in Ber. Deutsch. Bot. Gesell. 8 (1885) 274. Known from Peru (where it is very common), Ecua- dor, Venezuela and Puerto Rico, this beautiful Haben- aria has apparently hitherto been collected but twice in Colombia: Klug 1662 from Umbria in the Putumayo (only a few miles from the locality of Schultes & Smith 3005) and Lawrance 206 from Mt. Chapoén in Boyaea, the former from an altitude of 825 meters, the latter from 1086 meters. CoLomsia: Comisaria del Putumayo, Mocoa y alrededores al norte, alt. 750-850 m. ‘‘Flowers white. Growing in sand on bank.’’ De- cember 3-7, 1942, Richard Evans Schultes & C. Earle Smith 3008. Maxillaria rufescens Lindley in Bot. Reg. (18386) t. 1848. A widespread species, Maaillaria rufescens has appar- ently not been collected previously from Colombia. Known from Trinidad, Honduras, Guatemala, British Honduras, Nicaragua, Costa Rica and Cuba, it is ap- parently very common in Central America and the An- tilles. It is known also from Brazil, but collections are rare. Cotomspra: Intendencia del Amazonas, Rio Putumayo, carretera entre Caucaya y La Tagua. “‘Epiphyte.’? May 17, 1942, Richard Evans Schultes 3791. Odontoglossum coronarium Lindley Fol. Orch. Odontog. (1852-59) 21. Odontoglossum coronarium has hitherto been known from Panama, Colombia, Peru, and, probably, Ecuador. CotomstaA : Comisaria del Putumayo, Valle de Sibundoy, Porotoyaco, alt. about 2250 m. “‘Sepals and petals red-brown, yellow in spots. Lip yellow; lower half of lip red-brown. Column red-brown and yel- low.’’ December 16, 1942, Richard Evans Schultes & C. Earle Smith 3150. [ 297 ] Sobralia rosea Poeppig & Endlicher Nov. Gen. ac Sp. 1 (18386) 4, t 98, Schultes 3559 is the second collection from southern Colombia of this orchid which is widespread on the east- ern slopes of the Andes. Cotompra: Comisaria del Putumayo, Rio San Miguel 6 Sucumbios, Santa Rosa y alrededores. ‘‘Flowers white, tip of column purple. Kofin Indian name: ku-pu-fii-mem-ba. April 7-8, 1942, Richard Evans Schultes 3559. Xylobium squalens (Lindl.) Lindley var. gracile (Schitr.) C. Schweinfurth in Bot. Mus. Leafl. Harvard Univ. 11 (1944) 198. This is apparently the first collection from Colombia of Xylobium squalens var. gracile, hitherto known only from Ecuador. Cotompra: Comisaria del Putumayo, Paramo de Tambillo, nordeste del Valle de Sibundoy. “‘Flowers cream with red specks.’’ December 13-14, 1942, Richard Evans Schultes & C. Earle Smith 3094. MorackEakE Perebea lecithogalacta (R. EL. Schultes) R. EL. Schultes comb. nov. Castilla Ulei W arb. forma lecithogalacta R. E. Schultes in Bot. Mus. Leafl. Harvard Univ. 12 (1946) 128. Further investigation has indicated that this concept, which was referred to Castilla Ulei as a form, represents a new species of the related genus Perebea. The genus Perebea has hitherto not been considered as a commercial source of rubber, but P. /ecithogalacta is felled indiscriminately with Castilla Ulei in the upper reaches of the Vaupés River in Colombia for the produc- tion of ‘‘caucho negro.”’ The latex, when it flows from the wound, is first a deep cream color but rapidly becomes a bright canary yellow. It is evident that the ‘ 3 ‘caucho negro”’’ of this area of [ 298 ] Colombia is a mixture of Castilla and Perebea latex, and it may well be that other genera furnish a part of this rubber so eagerly sought for during the recent war. PROTEACEAE Roupala colombiana RP. E. Schultes sp. nov. Arbor quindecim ad viginti quinque pedes alta, appar- enter ramosa. Ramuli teretes, densissime et molliter rufo- vel cinereo-tomentosi. Folia alterna, elliptico-ovata, apice rotundato-subacuta (sed nune subrotundata vel obscure acuta), basi late cuneata vel rotundata (Killip & Smith 19737), valde marginata, integra, vel saepe obscure sinuata (Kx. & S. 19737), valde coriacea, maturitate 7-10 em. longa, 4.5-6.5 em. lata (13 em. longa, 10 cm. lata in K.& S. 19737), supra minute fusco-tomentella, nervorum centralium nervos secundarios versus dense sordide cin- ereo-tomentosa, venis plerumque elevatis et bene con- spiculs, infra omnino densissime et molliter rufo-tomen- tosa (folia juniora multo rufiora), venis non valde elevatis; petioli dense tomentosi, 1—2.5 cm. longi. Racemi multi- flori, axillares, quam folia multo longiores, aliquid rigidi, usque ad 15 cm. longi vel longiores, rhachide floribusque dense rufo-tomentosis. Bracteae minutissimae, caducae. Alabastri densissime rufo-tomentelli. Calycis lobi valde clavati, usque ad 11-12 mm. longi, contorti. Stylus lon- gior, usque ad 12 mm. longus. Ovarium subcompresso- ovoideum, densissime et grossiuscule tomentosum, 2X 1.7X1.5 mm. Folliculus obliquus, elliptico-oblongus, valde compressus, apice attenuato-acutus, cinereo- vel fusco-tomentellus. Vegetatively, Roupala colombiana resembles R. saai- cola, but the reddish or rusty indumentum of the former is very much denser and softer than that of the latter. There is likewise a difference in the shape of the leaves. The flowers and the peculiarly strict, very long and [ 299 ] rather remotely flowered inflorescence exhibit further dif- ferences, and the fruits of the two species are entirely distinct, those of Roupala saxicola being fully rounded, apically obtuse, and glabrous, while those of R. colom- biana are complanate, apically acute, and tomentulose. Cotomp1a: Departamento Norte de Santander (Eastern Cordillera) between Mutiscua and Pamplona; alt. 2700 m. ““Tree 15-25 ft., sepals cream white. Edge of woods.’’ February 23, 1927, EF. P. Killip §: Albert C. Smith 19762 (Type in Herb. Gray).—Same locality and date. E. P. Killip & Albert C. Smith 19737.—Departamento Norte de Santander, Cordillera oriental, vertiente oriental, Pamplona, Quebrada de Cariongo, matorrales, 2500 m. alt., 26 julio 1940, J. Cuatrecasas § H. Garcia-Barriga 10242. Roupala saxicola PR. H. Schultes sp. nov. Frutex usque ad duodecim vel quindecim pedes altus, ramosus. Rami teretes, nigro-cinereo cum cortice; ram- uli densissime et molliter rufo-tomentosi. Folia pinnati- partita, 25-28 cm. longa, oppositi- vel suboppositi-foliola, cum 6-10 foliolis, petiolulata (petioli usque ad 5 mm. longi) ; foliola elliptico-oblonga, apice rotundata, basi cu- neata, marginata, integerrima, coriacea, maturitate 7.5— 11 cm. longa, 4.5—5.5 em. lata, supra viridia, minute re- ticulata, minute tomentella nervorum centralium nervos secundarios versus dense rufo-tomentosa, venis vix eleva- tis, infra omnino densissime et molliter rufo-tomentosa, venis valde elevatis. Racemi densiflori, axillares, quam folia breviores, rigidissimi, usque ad 14 cm. longi, cum rhachide floribusque dense rufo-tomentosis. Bracteae mi- nutissimae, caducae. Calycis lobi valde clavati, cucullati, 8-8.5 mm. longi, valde contorti et patentes. Stylus robus- tus, 9-10 mm. longus, paulo arcuatus; stigma capitato- clavatum. Ovarium sessile, grosse rufo-villosum, elongato- ovoideum. Folliculus (Schultes 5457) ovali-oblongus (noncompressus), glaber, rufo-aureus vel aureus, apice rotundatus, pedicello brevissimo, 8.2-3.5 2.22 em. Roupala saxicola differs from other species which have [ 800 ] a soft, dense, red indumentum by having perfectly entire, slightly marginate leaves. It is also striking in having a short and very dense inflorescence. The unusually broad and rounded leaflets of this species and also the fully rounded (not flattened) fruits are distinct in the genus. Cotomsia: Comisaria del Vaupés, Upper Apaporis Basin, Macaya River, Mount Chiribiquete. Sandstone. Xerophytic conditions. Sa- vannah, Alt. 400-1200 ft. above forest floor or 1300-2100 ft. above sea-level. “‘Shrub up to 12 ft. Flowers greenish yellow. Growing on summit.’* January 18, 1944, Richard Evans Schultes 5744. (Tyrr in Herb. Gray).—Same locality. ‘‘Small tree. Fruit yellow.’’ May 15- 16, 1943, Richard Evans Schultes 5457. POLYGONACEAE Triplaris Pavonii Messner in DC. Prodr. 14 (1856) 172. This widely distributed tree is one of the commonest species on floodable river-banks in the upper Apaporis River drainage-area. CotomsiA: Comisaria del Vaupés, confluence of Macaya and Ajaju Rivers, Puerto Hevea. Alt. about 300 m. ““Bracts of calyx red. Tree.’’ July 1943. Richard Evans Schultes 5576. LEGUMINOSAE Cynometra Zamorana RP. EL. Schultes sp. nov. Arbor magna et robustissima, usque ad centum pedes alta. ‘Truncus columnaris, duo pedes in diametro, brun- neo cum cortice crasso. Foliola unijuga, obliqua, ellip- tica, apice breviter cuspidato-acuta, basi cuneata, saepe obscure marginata, coriacea, supra nitidissima, infra ni- tida, viridia, utrinque glabra maxima pro parte 13.5 em. longa, 5—5.5 cm. lata; venae secundariae saepissime quin- decim vel sedecim, supra non conspicuae sed infra valde elevatae, nervo centrali rufo; petioli crassiores. Racemi axillares, foliis saepius aequales, minutissime villosi pubi molli patenti, rhachide robusta, 6-7 cm. longa, pedi- [ B01 ] cellis crassis 8-10 mm. longis. Bracteae subtendentes ‘aducissimae. Flores ignoti. Iegumen subsessile, com- pressum, oblique ovoideum, minutissime cuspidatum, basi valde fusco-tomentosum, rufo-brunneum, conspicue rugosum, sublignosum vel lignoso-coriaceum, plerumque 3.7-4.8 em. longum, 2.2—-8 em. latum, 1.8 cm. crassum. Semen transverse oblongum, arillatum, 2.5 cm. longum, 1.8 cm. latum, 1.2 em. crassum, nigrum, nitidum. G+ ab «286+. a . PLaTE XXXIII ROUPALA saxicola 5 ce Ae Oe tn » 2 Mia > PLaTE XXXIV Ss SS CYNOMETRA Zamorana R.E. Schultes PLATE: XX XV ee GRAFFENRIEDA vy alee 4. fantastica R.E. Schultes & LB Smith XXXVI . 4 PLATE ee PLATE XNXVII Bn NEW CONCEPTS IN STYRAX FROM EASTERN COLOMBIA BY Jesus M. [prozo! anp Ricuarp Evans Scour res’ THe identification of a number of plants from the iso- lated sandstone mountains of the upper Apaporis River basin in eastern Colombia has revealed a collection which represents a species of Styrax hitherto apparently un- described. A distinct form of this species is represented by a collection from the ecologically similar sandstone mesa at Yapoboda at the headwaters of the Cuduyari, in the middle Vaupés River basin. In the uppermost reaches of the Apaporis River, where the Ajaju and Macaya join, there are a number of gro- tesquely eroded quartzite mountains which rise abruptly from the flat, extensive Amazonian forest surrounding them. They represent the westernmost remnants of an ancient, now disrupted, mountain-mass which has its center in southern Venezuela and the Guianas. They can be followed westward and southwestward throughout the Colombian Amazonia where, across the Comisarias del Vaupés and Caqueta, there are extensive flat-topped ridges, isolated from each other often by many hundreds of kilometers. These ridges extend west as far as the upper Apaporis basin and San José del Guaviare, and southwest to the Falls of Araracuara on the Caqueta River (famous as Martius’ westernmost collecting station and the type locality of many endemics) and La Chor- ‘Instituto de Ciéncias Naturales, Bogota. ? Botanist, Bureau of Plant Industry, Soils and Agricultural Engi- neering, U.S. Department of Agriculture; Research Fellow, Botanical Museum, Harvard University. [ 825 | rera on the Igaraparana River. Yapoboda, part of this broken range, is of special interest because, in the vast and sparsely explored Comisaria del Vaupés, it is located about midway between the westernmost mountains in the Apaporis and the Duida mass in Venezuela. In many other localities in Amazonian Colombia, one finds distant and very indistinct outliers of this ancient formation in the form of small areas of white sand with a low caatinga-forest or savanna vegetation. So indistinct topographically are some of these formations that it might be difficult geologically to identify them with the ancient Venezuela-Guiana land-mass were it not for the fact that their flora is so strikingly different from that of the surrounding Amazon jungle and so similar to that of a number of analogous areas in southern Venezuela and in the Guianas. It can be shown that these isolated patches have an ancient remnant flora which is most closely allied to the flora of the Duida-Roraima ranges and of other peaks in the same far-distant mountains on the northern rim of the Amazon Valley. As an example of these indistinct outliers, we may cite the rather extensive caatinga-forest between the headwaters of the Hamacayacu and Cotuhé Rivers in the interior of the trapécio amazénico (near Leticia). Similar areas occur near Mishuyacu and at Iquitos in Peru and at Sao Paulo de Olivenga on the upper Amazon River in Brazil. It is, therefore, apparent that a thorough study of the floras of these isolated mountains in eastern Colombia is of the utmost importance for a complete understanding of the origin and composition of the flora, not only of the Amazonas of Colombia, but also of large sectors of south- ern Venezuela, the Guianas and northern Amazonian Brazil. There would seem, likewise, to be some connec- tion between this ancient flora rimming the north of the [ 826 ] Amazon Valley and that of the geologically old plateau which gives rise to most of the great southern tributaries of the Amazon. This might explain the apparently anomalous occurrence in northwesternmost Amazonia of rare elements which have their greatest diversification in southeastern Brazil; or, as in the case of the concept herein described, the close relationship of some species known from these two widely separated areas of central South America. Mount Chiribiquete, the type locality of the Styrax which is herein described, has yielded a number of strik- ing endemic plants and numerous species which exhibit a definite relationship with elements known only from Mounts Duida or Roraima. Styrax rigidifolius is one of the few species from the shrubby cover of the flat sum- mit of Chiribiquete which does not seem to have a close ally inthe northern Amazon region or on the Venezuela- Guiana land-mass. It isnot common but grows hidden with the more abundant Graffenrieda fantastica Schultes, & Smith, Micus chiribiquetensis Dugand, Hevea nitida Muell.-Arg. var. towicodendroides (Schultes & Vinton) Schultes, Roupala saxicola Schultes, Senefeldera chiri- biquetensis Schultes & Croizat, Bombax coriaceum Mart. & Zuce., Vellozia phantasmagoria Schultes, several Vo- chysias, numerous species of Clusza, an interesting Tern- stroemia, and an undescribed species of Letgebia. All of these plants are completely adapted to xerophytic conditions and high radiation, and, in this connection, it is interesting to note that Styraw rigidifolius itself has unusually rigid-coriaceous leaves with a very strongly inrolled margin. There has been no species of Styrax recorded from Mount Duida. The only member of the genus reported from Mount Roraima is Styraxv roraimae Perk., an en- demic of this peak; it bears no relationship to Styrax [ 827 ] rigidifolius. Styrax rigidifolius would not seem to be closely allied to any known species, but it approaches most closely 8. ambiguus Seub. of southeastern Brazil. If we are to judge from the description and a photograph of the type, Styrax ambiguus is very similar vegetatively to S. rigidifolius in having rather coriaceous or thick- papyraceous leaves of a small size with revolute margins, characters not common in the genus. There are, how- ever, very appreciable differences in the size of the flower, the texture and indumentum of the calyx, the shape and size of the corolla lobes, the shape and size of the sta- mens, the pilosity at the point of union of the filaments with the corolla tube, and the form and pilosity of the ovary. Styrax rigidifolus belongs to the section Mustyraa, series Valvatae. In the key to the species in Perkins’ monograph (Pflanzenr. [V, 241 (1907) ), the specimen which is here described as new traced out to the vicinity of Styrax calvescens Perk., an East Indian species with no relationship to S. rigidifolius. Styrax rigidifolius [drobo & R. I. Schultes sp. nov. Frutex parvus, usque ad octo pedes altus. Rami te- retes, cortice griseo, Juniores fusco-pulverulenti, demum glabrescentes, usque ad 5-6 mm. in diametro. Folia peti- olata, oblongo-elliptica, longe acuminata, apice ipso acu- ta, basi subrotundata vel saepe rotundato-cuneata, rigide coriacea, integra, fusco-pulverulenta, maturitate 6-7 cm. longa, 2.3-8 cm. lata, margine valde revoluta, supra gla- bra (ochraceo-viridia in siccitate) vel in nervo centrali basim versus paulo sordide pulverulenta, nitida, subtus cinnamomea, juniora rufidula, dense et molliter praecipue nervos versus tomentosa, pilis albis minutissimis stellatis remotius brunneis cum pilis aloéformibus intermixtis ves- tita, supra nervo centrali immerso, subtus valde elevato, [ 328 ] venis lateralibus sex ad novem, leviter arcuatis, minoribus subparallis et obscure reticulatis; petioli in vivo sulcati ut videtur, 6-10 mm. longi. Inflorescentiae racemosae, axillares, brevissimae, usque ad octoflorae, rhachide mi- nute ferrugineo-tomentosa, bracteis caducis, minutis, linguiformibus, 2 mm. longis. Flores (specimine viso) 15 mm. longi, pedicello 5 mm. longo; calyx cupulifor- mis, coriaceus, margine truncatus, minutissime quinque- denticulatus, extus asper, valde minutissime stellato- tomentosus, 5 mm. altus, 4mm. latus, margine obscure subundulato; corolla quinquepartita, tubo 8 mm. longo, quam lobi multo breviore, lobis in aestivatione valvatis ut videtur, vivo recurvis, membranaceis, apice utrinque den- sissime albo-stellato-tomentosis, apice paulo subeuspida- tis, leviter marginatis, 18 mm. longis, 2.5 mm. latis; stamina decem, filamentis parte libera aliquid planis corol- lae tubo conniventibus et pilis fasciculatis albis grossius- cule atque dense obstita; antherae 7 mm. longae, 0.7 mm. latae, dorsaliter glabrae, ventraliter thecas versus albo- pilosae; ovarium superum, pyriforme, dense albo-tomen- tosum, 2 mm. in diametro; stylus stamina longitudine aequans. Fructus maturitate cinereo-viridis, quadrangu- lari-ovoideus, 8-10 mm. longus, 4-5 mm. in diametro, minute aspero-tomentellus, apice truncatus, breviter mu- cronatus, calyce persistenti; semen unum, interdum duo (2), rufo-brunneum, glabrum, oblongo-ovoideum, 7 mm. longum, 4 mm. in diametro. Cotompra: Comisaria del Vaupés, Macaya-Ajaju River confluence, Mount Chiribiquete. Quartzite base. Summit 800-1200 feet above forest floor; 1700-2100 feet above sea-level. *‘Low shrub. Flowers: petals white, recurved, anthers yellow. Fruit cupped, green.’’ May 15-16, 1943, Richard Evans Schultes 5468 (Typrr in U.S. Nat. Herb. ; Dupuicate type in Herb. Nac. Colomb. ). [ 829 | EXPLANATION OF THE ILLUSTRATION Prare XXXVIII. Sryrax riawirotius Idrobo & R. EF. Schultes. A. Flowering branch, one half nat- ural size. B. Bud, twice natural size. C. Open flower, twice natural size. D. Dissection of flower, twice natural size. EK. Young fruit, twice natural size. F. Young fruit showing embryo, twice natu- ral size. S. Riaipirotius forma yarosovensis Idrobo & R. E. Schultes. G. Portion of leaf, one half natural size. H. Inflorescence with bud, twice natural size. Drawn by Jesus M. [propo [ 380 ] PuatTeE XXXVITI STYRAX S tyrax ie ieg bpeieed a. odensis vi 3k et Schultes E rigidifolius Th oe ih Schultes { Styrax rigidifolius forma yapobodensis Jdrobo & BR. Ie. Schultes forma nova. A Styrace rigidifolo foliis majoribus, marginibus rev- olutioribus, cum apice acuto non longe acuminato, supra opacioribus, subtus rufioribus ; alabastris minoribus ; caly- cibus grossiuscule quinque-sinuato-denticulatis; atque foliorum et calycum indumento asperiorio et rufiore differt. This forma appears to be an ecological adaptation to the extreme xerophytism of the sandstone mesa at Yapo- boda. An examination of the interesting plants collected at Yapoboda by Allen and a comparison of them with those of Mount Chiribiquete indicates that an even more severe xerophytism exists at the former than at the latter locality. As an example, the collection Allen 3190, ref- erable to Bombax coriaceum Mart. & Zuce. but with ex- traordinarily reduced leaves, may be cited. The differences enumerated above in the description of the form would, for the most part, seem to be related to adaptation to extreme xerophytism and radiation. Cotomata : Comisaria del Vaupés, | Cuduyari River], Yapobod4. ““Tree 5m. Flowers white.’’ December 10, 1943, Paul H. Allen 8232 (Tyrr in U.S. Nat. Herb. 1951983). [ 382 ]