AV ee nee —_ ” s BOTANICAL MUSEUM LEAFLETS HARVARD UNIVERSITY PRINTED AND PUBLISHED AT THE BOTANICAL MUSEUM CAMBRIDGE, MASSACHUSETTS BOTANICAL MUSEUM LEAFLETS HARVARD UNIVERSITY VOLUME XII BOTANICAL MUSEUM CAMBRIDGE, MASSACHUSETTS 1945-1947 4 a. Ane Ole E ie Cig Rar’ Ge o ave Wusenn TABLE OF CONTENTS NuMBER 1 (June 14, 1945) PAGE The Genus Hevea in Colombia By Ricuarp Evans ScHULTES ........ 1 NuMBER 2 (August 15, 1945) The Origin and Nature of the Ear of Maize By Paut C. MANGELSDORF. ........ ..88 NuMBER 8 (October 23, 1945) African Orchids XVI By V.S. SUMMERHAYFS .......... .89 NuMBER 4 (January 25, 1946) Plantae austro-americanae III By Ricuarp Evans ScHULTES ........ 117 NuMBER 5 (February 8, 1946) Orchidaceae novae guineae By Louis O:; WintraMs* = 2 2. 4 « 2 49 Notes on the Family Corsiaceae By Louis O. WinuiamMs .......... #179 NuMBER 6 (March 27, 1946) Orchidaceae peruvianae VI By CHARLES SCHWEINFURTH ...... . .. 185 faa NumbBer 7 (April 17, 1946) New and interesting Mexican Orchids By Louts O. WILLIAMS Two Orchids from Haiti By Louis O. WILLIAMS NumBer 8 (October 25, 1946) Races of Maize in South America By Huau C. Curher Numser 9 (December 27, 1946) Rubber Production in Ceara, Brazil By Huau C. CurLer NuMBER 10 (July 18, 1947) A Conspectus of the Genus Cunuria By J.T. Batpwin, Jr. AND RicHarp Evans SCHULTES . [ vi ] 301 B25 LIST OF ILLUSTRATIONS PLATE Calochilus caerulescens L. Wms. . 2... 2. XXII Chitonanthera Brassii L. Wms. 2. 2... 2.) XXIII Chitonanthera filiformis LZ. Wms. . . . . . XXIII Chitonanthera gracilis L. Wms. . . . . . . XXIII Corsia acuminata L. Wms. . . . ..... XXIV Corsia purpurata L. Wms. ........ XXIV Cryptocentrum inaequisepalum C. Schweinf. . . XXX Cryptocentrum pseudobulbosum C. Schweinf. XXXI Cunuria australis R. #. Schultes . . . 2...) . XLII Cunuria crassipes Muell.-Arg. ...... XLIII Cunuria glabra FR. H. Schultes . . . . . . .XLIV Cunuria Spruceana Baill... . . . 2. . XLV, XLVII Cunuria Spruceana Baill. var. bracteosa (Ducke) R. I. Schultes 2... 2. XLVI Helosis guyannensis L. C. Rich. . 2... 2...) XIII Hevea brasiliensis (H BK.) Muell.-Arg. var. Bubeoncolor DIUCKE. 2. 7 a a 1 - - — _ 7 _ se = - - - 7 : >. - 7 - a ae =o M LEAFLETS HARVARD UNIVERSITY VoL. 12, No. 2 THE ORIGIN AND NATURE OF THE EAR OF MAIZE BY Paut C. MANGELSDORF Or ALL THE New World plants which excited the curiosity and wonder of the herbalists of the immediate post-Columbian period, none, perhaps, proved to be more intriguing than Indian corn or maize. The ears of this plant, and to a lesser extent the tassels, were quite baf- fling to the students of that period and have continued to perplex botanists for more than four centuries. No- where has the recognition of the peculiarities of the maize inflorescence been more vividly expressed than by Lyte in his New Herbal of 1619 (6)*. The page on which maize is described is reproduced as Plate VII of this paper. The description of the inflorescences is as follows: ‘**This Corne is a marvellous strange plant, nothing re- sembling any other kind of grayne; forit bringeth forth his seede cleane contrarie from the place whereas the Floures grow, which is against the nature and kinds of all other plants, which bring forth their fruit there, whereas they have borne their Floure. ... at the highest of the stalkes, grow idle and barren eares, which bring forth nothing but the floures or blossomes. ... ”’ The wonderment of the herbalist of the sixteenth cen- * This description, in slightly different form, first appeared in an edition entitled A Nievve Herball in 1578. [ 33] tury has persisted almost undiminished to the botanist of the twentieth. ‘True, the modern student recognizes that Lyte’s description of maize is somewhat lacking in accuracy from the technical botanical standpoint. Maize does not, as Lyte believed, bear its seeds ‘‘cleane contrarie from the place whereas the Floures grow’’ and therefore it is not, as he contended, ‘‘against the nature and kinds of all other plants.’’ Nevertheless, the botanist of today would not deny that maize is indeed a ‘‘marvellous strange plant,’’ and he would unhesitatingly agree with Lyte that its uniqueness lies in its inflorescences, partic- ularly in the pistillate inflorescence, the ear. In its general vegetative characteristics maize does not differ essentially from other grasses especially the larger cultivated species; its affinity to sugar cane and the sor- ghums, for example, is easily recognized. But nowhere among the Gramineae, indeed nowhere in the Plant Kingdom, is there a fruit which in its external aspects at least seems to be more than remotely comparable to the ear of maize. It is quite understandable, therefore, that the ear of maize has been the subject of numerous observations, investigations and conjectures. Nor is it surprising that the problem of its origin and morpholog- ical nature has remained to a large extent unsolved. This paper, despite the unqualified language of its title, does not presume to present a final solution to the problem of the origin and nature of the ear of maize. Botanical problems of this kind are seldom susceptible of complete and definitive solutions. However, in the course of extensive studies of maize-teosinte hybrids and of hybrids of pod corn with a peculiar variety of maize obtained from the Guarany Indians of Paraguay, a num- ber of interesting phenomena have been encountered which appear to shed new light upon the problem of the maize ear. Before considering the new facts, it seems de- [ B34 | sirable to review the previous evidence and hypotheses which bear upon the problem. Only the essential features of the earlier hypotheses will be considered here since the rather extensive literature on the subject has been tho- roughly reviewed in recent years by several writers. The reader is referred to the papers of Weatherwax (19), Mangelsdorf and Reeves (13), and Reeves (15) for addi- tional details. HisroricaL CONSIDERATIONS The Maize Ear Described The ear of maize, though not easily interpreted, is not difficult to describe. It is a spike upon whose thickened axis (the cob) naked grains (caryopses) are borne in lon- gitudinal or somewhat spiral rows; eight, ten, twelve or more in number. The number of rows is always even because the spikelets upon which the grains are borne are paired, a characteristic in which maize differs from its nearest relatives, teosinte and Tripsacum, whose spike- lets are solitary. Each rank of paired spikelets is clearly the equivalent of two rows of grain in the mature ear. The ear is enclosed in husks which are modified, over- lapping leaf-sheaths. The ear is obviously the terminal inflorescence of a lateral branch whose internodes have, probably during the course of domestication, become drastically contracted. Homology of Har and Tassel The general nature of the ear is sufficiently clear so that there can be little doubt that it is the homologue of the central spike* of the staminate inflorescence, the tas- sel. ‘This homology may have been vaguely suspected by some of the earlier students including Wigand, Ascher- *In one of his recent papers Weatherwax (19) calls this structure a “‘terminal raceme.’’ I prefer the older but equally appropriate **central spike.’’ [ 35 ] son and Goebel,* but it seems to have been first clearly recognized by Mrs. Kellerman (9) who postulated that the ear has developed phylogenetically from the central stem of the primitive tassel. Montgomery (14) appar- ently unaware of Mrs. Kellerman’s suggestion, inde- pendently arrived at the same conclusion, illustrating it with an interesting series of transition stages to show how the change from a staminate central spike of the tassel to a pistillate ear might have occurred. Since the appearance of Montgomery’s paper the homology of the ear with the central spike of the tassel has not been seriously challenged except, perhaps, by indirection. Strong evidence, amounting almost to proof, of the homology of the two structures has recently been furnished by Langham (11) who showed that in segre- gates of maize-teosinte hybrids there is a close correla- tion between the ear and the central spike of the tassel in the expression of distichy and polystichy. A recent paper by Bonnett (8) is illustrated with striking photo- graphs which reveal, among other things, that in the early stages of development the two structures, ear and central spike, are scarcely distinguishable. Anderson (1) has recently pointed out that some of the important characteristics of the ear are closely correlated with characteristics of the lateral branches of the tassel. These observations, as will be shown later, are not in conflict with the conception of the ear as the homologue of the central spike of the tassel. The fact that the ear is the homologue of the central spike does not, however, solve the problem of the origin of the ear. As Collins (4) pointed out many years ago, * Cf, Weatherwax (19) or Mangelsdorf and Reeves (13) for refer- ences. t Italics hers. [ 36 ] the central spike is as much in need of explanation as the ear. Hypotheses Regarding the Ear of Maize There have been three general hypotheses regarding the ear of maize: (1) That it is the product of the fusion of two or more distichous (two-ranked) spikes. (2) That it originated from the shortening and twisting of a dis- tichous spike. (8) That it is the result of the transforma- tion of a panicle to a spike through the reduction of branches. The Fusion Hypothesis. It is not surprising that botanists familiar with the compound nature of many fruits should, when confronted with an ear of maize, think immediately in terms of fusion. And when branched ears, ears apparently ‘‘disrupted’’ into their component parts, are encountered, the fusion hypothesis becomes almost inevitable. Little wonder that this hy- pothesis is the oldest of the three and has been, by all odds, the most popular. Ascherson, Hackel, Harsh- berger, Gernert, Wordsell, and Goebel * have all inter- preted the ear of maize as arising through the fusion of two-ranked spikes either like those of teosinte or Tripsa- cum, or like the lateral branches of the maize tassel. The serious weakness of the fusion hypothesis, so far as maize is concerned, is that there is no concrete evidence in support of it and considerable evidence in conflict with it. Branched ears, bifurcated or many-branched at the tip or bearing branches at the base, have repeatedly been regarded as evidence for the compound nature of the maize ear. But the majority of these branched ears prove upon examination to represent a type of branching which * Cf, Weatherwax (19) or Mangelsdorf and Reeves (18) for refer- ences. [ 37 ] is not easily attributable to the ‘‘disruption’’ of a com- pound structure into its component parts. Weatherwax (16) regards ears branched at the tip as nothing more than anomalies ‘‘giving no more clue to the past than is afforded by Siamese twins.’” And now Reeves (15) has virtually committed to the category of reductio ad ab- surdum the argument for fusion based upon branching, when he points out that the two types of branching most commonly encountered in the ear, basal branching and bifurcation of the tip, have their counterparts in the stem, the first in the well-known basal tillering or ‘‘suckering’’ of the stalk, the second in the bifurcated stalks which he described (15). Reeves contends that if branching of the ear is regarded as evidence of its compound nature, then corresponding types of branching in the culm must be regarded as evidence that it, too, is a compound struc- ture resulting from fusion. Aside from branched ears, which upon close examina- tion obviously fail to support the fusion hypothesis, what is the evidence for fusion? Apparently there is none. W eatherwax (18) states that both the young ear and the young tassel develop from ordinary growing points and that there is nothing to suggest a compound nature. Bonnett’s studies (3) support this statement, although he does not specifically discuss the question of fusion. W eatherwax has observed (19) that the vascular bundles of the cob are distributed in approximately the same manner as those of the stem and therefore furnish no in- dication of fusion. Reeves (15) too, has made a detailed study of the vascular system and the arrangement of the rachis segments and finds no structural feature of any kind which suggests that the maize ear is the product of fusion. Weatherwax (16, 18, 19) seems to regard as the most critical evidence in conflict with the fusion theory, the [ 38 | fact that the fusion of distichous branches (in which the spikelets are paired or must become paired to produce an ear of maize) should result in ears in which the num- ber of ranks of paired spikelets is always even and the number of rows of grain is always a multiple of four. Thus a fusion of two such spikes or branches would pro- duce an ear with four ranks of paired spikelets and eight rows of grain; a fusion of three spikes or branches, an ear with six ranks of paired spikelets and twelve rows of grain, etc. Weatherwax assumes that ears in which the number of ranks of paired spikelets is odd and the num- ber of rows of grain not a multiple of four (ears with ten, fourteen and eighteen rows of grain, for example) cannot have been the product of fusion of two-ranked spikes. Since such ears do occur Weatherwax concludes that the fusion hypothesis is confronted with serious mathematical inconsistencies. Kempton (10), in an attempt to reconcile these appar- ent inconsistencies, has suggested that ears in which the number of ranks of paired spikelets is odd may be the result of the abortion of a row of paired spikelets or the abortion of the pediceled* spikelets in both ranks of one of the component branches. Kempton was of the opinion that both of these phenomena are of common occurrence, but it is now doubtful if either one occurs. Weatherwax (19) found no vestiges of the supposedly aborted spikelets and Dr. Reeves and I have examined numerous ten-rowed ears without finding a trace of them, although there is no difficulty in seeing them in both teosinte and Tripsacum where normally only one spike- let in each pair is functional. Kempton (10) also suggested that ears in which the * Both members of a pair of pistillate spikelets are actually sessile, but one is potentially pediceled and is the homologue of the pediceled staminate spikelet. [ 39 ] number of rows of grain is a multiple of four are more numerous than ears in which the row number is not a multiple of four. Even before the suggestion was made, East (7) had shown that such is, indeed, the case. East offered no morphological explanation for the multimodal distribution and it still remains to be satisfactorily ex- plained. Reeves (unpublished) has suggested that the peculiar distribution reported by East may be the result of past selection for straight-rowed ears, a suggestion based on the fact that there is a tendency, discovered by Fujita (8), for ears with an odd number of ranks of paired spikelets to be twisted. But whatever the explanation of the low frequency of ears with ranks of paired spikelets in odd numbers, the fact that any ears of this kind occur has generally been regarded as evidence against the fusion theory. Perhaps more weight has been assigned to this evidence than is deserved. There are undoubtedly ways (one will be men- tioned later) in which rows of spikelets might be lost without leaving readily discoverable vestiges. Further- more if fusion is thought of in a phylogenetic rather than in an ontogenetic sense, it 1s scarcely necessary to assume that the compound structure resulting from fusion must still exhibit all of the features of its original component parts. Finally the absence of anatomical evidence of fu- sion is by no means final proof that it has not occurred. There are numerous structures in plants in which ana- tomical evidence for fusion is lacking or at least not read- ily discernible, but which are nevertheless regarded as compound structures. In these instances, however, some other kind of evidence for fusion is usually at hand; for example, a series of forms involving different species which illustrates a transition from the condition in which the components are entirely separated to that in which they are completely joined. In the case of maize there [ 40] appears to be no evidence of any kind to support the fu- sion hypothesis with respect to the ear. Yet fusion does occur in maize. It is almost certain that the slightly branched styles, the well-known “‘silks’’ are compound structures resulting from the fusion of the two stylar branches characteristic of grasses.* I have seen several cases of fusion of the pedicel of a staminate spikelet with the surface of the rachis. And an actual instance of the fusion of two of the branches of a maize tassel, the kind of fusion which might have given rise to the ear, will be described later in this paper. Nevertheless, the fusion hypothesis of the origin of the maize ear, though the most obvious, and at first glance the most plausible, actually appears to have little to commend it. Reduction of Branches. The second hypothesis, that the ear of maize has been derived from a panicle through the reduction of branches until each branch is now represented by a pair of spikelets, was suggested by Collins (4). But the general theory that spikes are more specialized than panicles and have originated from these through the reduction of branches is, as Reeves (15) has pointed out, an old one. This hypothesis would seem to be especially appropri- ate when applied to maize whose staminate inflorescence, the tassel, is a perfect illustration of a combination of spike and panicle, a structure in which the reduction of branches either has not proceeded to completion or has been confined to the upper part of the inflorescence. Furthermore it is not difficult to find extreme forms which lend credence to the hypothesis; for example, tassels which consist only of central spikes or ears which bear well-defined basal branches. Collins felt that the chief objection to this hypothesis * Cf. Weatherwax (18) p. 123. [ 41 ] lay in the fact that in most cases the change from branches to spikelet is abrupt. In the tassel there are usually only well-defined branches at the base and only spikelets on the central spike. In most cases there is no transition from branches to spikelets. Collins found some evidence of such a transition in varieties of pod corn, but the condition in pod corn was by no means as convincing as the various transition stages later described and illus- trated by Weatherwax (16) in branch (ramosa) maize. The studies of Bonnett (8) have a bearing on this prob- lem for they show that in the development of the tassel there are no discernible differences between the initials that become the lateral branches and those which be- come differentiated into spikelets. The conception of the ear of maize as a spike homolo- gous to the central spike of the tassel, and like the latter having arisen through the reduction of branches, (a re- duction usually completed in the ear, but seldom in the tassel) appears to be clear, reasonable and in complete harmony with the facts. Weatherwax (16, 19) contends, however, that this hypothesis alone is not adequate since it does not answer the question of how the polystichous condition originated in maize or other grasses. He, there- fore, interprets the maize ear in terms of spiral phyllo- taxy. The four-rowed ear (like the arrangement of the leaves on the stem) is regarded as including a single spiral ; the eight-rowed ear includes two spirals, ete. Weatherwax has made an important contribution in recognizing and describing the spiral phyllotaxy charac- teristics of some (probably not of all) maize ears, and in showing the resemblance of the maize ear to the spikes of other grasses such as Pennisetum. Phyllotaxy, how- ever, describes rather than explains the change from the distichous condition typical of the grasses in general to the polystichous condition characteristic of the inflores- [42] cences of maize. To say that the many-ranked ear of maize departs from its two-ranked prototype by virtue of its more complex phyllotaxy is scarcely more explana- tory than to say that the former has more numerous ranks than the latter. Nevertheless phyllotaxy, as will become apparent later, is immensely important. Actually, as Reeves (15) points out, the polystichous inflorescence is sufficiently common in various genera of grasses so that no special explanation of its occurrence in maize is required. Its origin is a problem of the Gramin- eae as a group rather than of maize alone. The unique- ness of maize lies not so much in its polystichous ear as in the fact that its staminate inflorescence is usually a modified panicle while its pistillate inflorescence is usu- ally a spike. Collins’ hypothesis furnishes a very satis- factory explanation of the steps involved in the trans- formation of one to the other. Anderson’s (1) observations, previously mentioned, that the characteristics of the ear are correlated with characteristics of the lateral branches of the tassel, may be interpreted to mean, not that the ear is the product of fusion and that the lateral branches are homologues of its component parts, but simply that the lateral branches are capable of revealing what kind of panicle it is that has become modified to produce the ear. Twisting of a [Two-Ranked Spike. The third hypothesis, that the polystichous maize ear arose through the shortening and twisting of a two-ranked spike, such as the spike of teosinte, was also formulated by Collins (5). This hypothesis grew out of observations made by him on the pistillate spikes of an Fg population of a maize-teosinte hybrid from which a series of specimens could be selected to illustrate the steps involved in chang- ing a structure like the spike of teosinte to one resem- [ 43 ] bling the ear of maize. These steps, not necessarily in this order, are as follows: 1. The solitary pistillate spikelets characteristic of teo- sinte become paired spikelets typical of maize as the aborted member of each pair becomes functional. 2. The axis of the rachis shortens with the result that adjacent rachis segments assume positions in the same horizontal plane and become ‘‘yoked”’ in pairs. This change, combined with the preceding one, results in a four-rowed ear. 3. A twisting of the axis causes alternate segments to assume positions in a plane at right angles to that occu- pied by adjacent segments above and below. This, com- bined with the changes already described, produces an eight-rowed ear. Ears with higher row numbers are pro- duced by further twisting of the axis. There is no doubt that the series of transition forms which Collins described do occur in segregates of maize- teosinte hybrids. Nor is there any doubt that the ear of maize could have arisen from the spike of teosinte through such a series of changes. Indeed if maize had originated from teosinte the changes which have actually occurred must have been very similar to those which Collins has described. Collins did not imply, however, that maize had actu- ally been derived from teosinte by this series of steps and in fact he carefully pointed out that these intermediate steps explained the evolution of the ear only in a me- chanical sense. Even in a strictly mechanical sense, however, the hy- pothesis has not been satisfactory. Weatherwax (19) has objected to it on the grounds that there is no evidence in maize of the supposed ‘‘yoking’’ of pairs of spikelets on opposite sides of the rachis. Collins believed that he had evidence for yoking in ears in which some of the rows [ 44 ] of grain were interrupted. He stated that among many ears examined, in which rows were dropped between the base and the tip of the ear, in no instance were the dropped rows either adjacent or separated by two rows, but always (when the dropped rows could be determined with reasonable certainty) on opposite sides of the ear. Weatherwax (17) showed, however, that when rows are dropped between the base and tip of the ear, it is always a pair of adjoining rows which is eliminated and not rows on opposite sides of the rachis. It is difficult to see how Collins, usually a keen observer, could have been so completely in error in this instance. Actually his state- ment, though highly inaccurate, has some basis in fact; for in segregates of maize-teosinte hybrids there is in some four-ranked ears a definite tendency for the spike- lets in one plane to be predominantly paired while those in the plane at right angles are predominantly single. In other words in ears in which yoking is obvious, there is a definite tendency for the spikelets diametrically op- posed to each other on the axis to be alike with respect to the abortion and development of spikelets. In considering the yoking hypothesis there is some danger, as has already been apparent in the case of the fusion hypothesis, of thinking only in rigid terms. Is it not possible that as the axis becomes thickened, the yok- ing, if it occurs at all, would become less obvious and the tendency for spikelets on opposite sides of the rachis to exhibit identical behavior would become less pronounced and would, indeed, disappear completely / Is it necessary, in other words, to conclude that yoking has not occurred in the development of the maize ear because the spike- lets on opposite sides of the axis, in ordinary ears of maize, are not alike in their behvaior? These questions will be discussed further when new evidence from maize- teosinte hybrids is considered. [ 45 ] New EviIpENCE ON THE NATURE OF THE Maize Ear Evidence for Fusion The fusion hypothesis, as has already been mentioned, has been, to a large extent, dismissed because (a) ana- tomical and other evidence for fusion has been lacking and (b) it has been thought that the hypothesis leads to mathematical inconsistencies with regard to row number. New evidence renders both of these objections somewhat less pertinent than they previously were. During the summer of 1944, Mrs. P. C. Mangelsdorf in taking notes on the number of ranks in the central spike of tassels among segregates of maize-teosinte hy- brids, encountered a clear-cut case of fusion of two of the lateral branches of the tassel. This is illustrated in Fig. 1. The two branches are joined for a distance of slightly more than half of their length. There is no doubt that this is a case of fusion rather than fasciation for the lower region in which the two parts are joined is distinctly four- ranked, a condition which I have never before encount- ered in a lateral branch of the tassel, while the two un- joined parts of the upper region are both distinctly two-ranked and differ in no important detail from the remaining branches on this tassel. The fused portion of this branch is not radially sym- metrical but exhibits a distinct dorsiventral, character since only the adjoining edges of the two component parts are fused. Had the two entire dorsal surfaces of the branches become fused, or had a third branch become fused at both of its edges to the free edges of the other two, thus forming a cylinder, the structure would have been indistinguishable from the central spike of the tassel. This single specimen makes it clear that a polystichous structure similar to the central spike of the tassel (which is the homologue of the ear) can arise in maize through [ 46 ] the fusion of distichous branches. It does not prove that the ear of maize did actually arise in this way, although it would undoubtedly have been re- garded as proof by earlier proponents of the fusion hypothesis. Also during the summer of 1944 my attention was called to some pecu- liarities of Tripsacum dactyloides (1. ) L. by Dr. Leon Croizat who suggested that they seem to have a bearing on the question of fusion. In this spe- cies the inner surfaces of the several branches of the terminal inflorescences are so sculptured that the irregularities of one branch are reciprocal to those of the adjoining branch. Because of this the branches, when brought to- gether, form a cylinder which bears spikelets on its circumference ; a struc- ture comparable to an ear of maize. I do not regard this condition as evi- dence for fusion as it is obviously noth- ing more than the result of the consid- erable compression to which the inflo- rescence is subjected while still in the sheath. The same condition in less conspicuous form is regularly encoun- tered in maize tassels where the lateral Figure 1. A case of fusion of two of the lat- eral branches of a maize tassel. The two-ranked branches are joined for slightly more than half of their length, but separated for the re- mainder. [ 47 ] branches during development have been compressed around the central spike. Indeed compression is a com- mon phenomenon in the grass inflorescence. Its impor- tance has been clearly recognized by several students of the Gramineae and has been strongly emphasized by Arber (2). The Tripsacum inflorescence is of interest, in my opinion, not in indicating fusion, but in suggesting (as pointed out by Dr. Croizat) how maize ears with odd numbers of rows of paired spikelets and rows of grain not in multiples of four mzeht have originated through fusion. This possibility is illustrated in Fig. 2. Fig. 2a _7— Spikelets Figure 2. Diagrams showing how fusion of two-ranked branches such as those of Tripsacum might give rise to ears with rows of paired spikelets in odd numbers. With the development of the abortive spike- let, the structure illustrated in diagram A would give rise to a twelve- rowed ear. But the condition illustrated in B, because one row of spikelets is abnormally oriented and ““buried,’’ would produce a ten- rowed ear. shows the condition most commonly encountered when three branches are brought together to form a cylinder. Had an ear of maize originated through the fusion of three branches such as these illustrated it would have become a twelve-rowed ear for the circumference of the axis bears six rows of spikelets which, though solitary in Tripsacum, are always paired in maize. Fig. 2b illustrates a condition encountered much less [ 48 | commonly but by no means rarely. Here one of the branches is asymmetrical with respect to the position of its spikelets. Were these three branches to fuse to pro- duce an ‘‘ear’’ it would be a ten-rowed ear for here the circumference bears only five rows of spikelets; the sixth is “*buried’’ in the cob. Its vestiges, if they remain at all, are not likely to be found on the surface. I do not intend to suggest that the ear of maize did actually originate from the fusion of two-ranked branches such as these of ‘Tripsacum, nor do I wish to explain the occurrence of ears with rows of paired spikelets in odd numbers as the result of the “‘burying’’ of a row of spikelets in the cob. It is very doubtful if the points at which the spikelets are to appear are so rigidly fixed in the undifferentiated primordium of the spike that its accidental disorientation could result in the suppression of an entire row of spikelets. On the other hand, if this is true, then it is probably also true that ears with pairs of spikelets in odd numbers are not necessarily inconsist- ent with the fusion hypothesis. For if we think of fusion in rigid terms, the joining during ontogeny of spikes each bearing the primordia of two ranks of paired spikelets, then the condition illustrated in Fig. 2b is a reasonable and valid explanation of how a row of paired spikelets might be lost without leaving clearly discernible vestiges. Add to this the fact that fusion of branches has actually been observed in maize and it becomes apparent that the possibility of fusion in the development of the maize ear cannot be dogmatically dismissed. Nevertheless the fu- sion hypothesis, in spite of the new evidence which may appear to support it, remains scarcely more satisfactory than it was before. Hybrids of Guarany Maize and Pod Corn The real nature of the ear of maize, at least of one type [ 49 ] of ear, is strikingly revealed by ears from a cross of pod corn (tunicate maize) with a variety of maize obtained from the Guarany Indians of Paraguay. The Guarany maize has many peculiar and interesting characteristics. It is one of the few varieties from the lowlands of South America whose chromosomes are knobless or almost so. It exhibits almost all of the dominant genes known in maize, a condition to be expected in a primitive variety. It has the most slender and most flexible rachis ever en- countered in a cultivated variety of maize. Finally, it has the peculiar characteristic, under some conditions, of exhibiting in the ear an indeterminate habit of growth. When this happens the ear protrudes far beyond the husks and the exposed region of the ear becomes greatly elongated to produce a lax and flexible spike. This con- dition was first encountered in 1941. An attempt made in 1942 to induce it through treatment did not succeed. In 1944 the condition again occurred spontaneously. What the factors are which are involved inits occurrence isnot known. Bonnett (8) states that both staminate and pistillate inflorescences in maize are potentially indeter- minate. Apparently the potentiality persists for a longer period in the Guarany maize than in ordinary maize and an opportunity is afforded for environment to play a part. It is of some interest in this connection to note that the indeterminate nature of the ears was much more pro- nounced in a late-planted row which flowered during a period of very favorable weather, than it was in earlier planted rows which came into bloom during a hot dry period. The stock in which the indeterminate ears occurred in 1944 was one which had been derived from a hybrid of Guarany maize and pod corn which had been twice back- crossed to Guarany and was therefore seven-eighths Gua- rany in its germplasm. ‘Tunicate and non-tunicate ears [ 50 | occurred in approximately equal numbers and both were affected by the elongation which had occurred. The ef- fects were much more striking in the tunicate ears and especially so in the ear illustrated in Plate VIII which represents an extreme example of elongation. A study of this and similar ears reveals the real nature of the ear of maize more clearly than has ever been pos- sible by anatomical studies. It should be pointed out that these ears are not abnormalities in the sense that the normal course of development has gone drastically awry —they are not teratological phenomena. Instead, the normal potentialities of the ear, apparently always pres- ent, have been more completely developed than is usu- ally the case. This, in combination with the tunicate condition, has resulted in ears so elongated and stretched out that the individual nodes of the rachis and the ar- rangement of the spikelets upon them are clearly re- vealed, as illustrated by Plate [X.* There is apparently no essential difference between the structure of the ear at the base where it was completely enclosed in the husks and at the tip where it was completely free from the pressure and restraint exerted by the husks. There is a gradual transition, not an abrupt change, from one con- dition to the other. There are several definite and important conclusions to be drawn from these elongated ears of Guarany pod comm: Homologies. There is now no doubt, if there was doubt before, that the ear is the homologue of the central * The use of genetic characters in studying the nature of various structures of maize and its relatives offers important possibilities. The tunicate gene transferred to teosinte by repeated backcrossing fur- nishes a most convincing demonstration that the shell of the teosinte fruit is composed of a rachis segment and an indurated glume, as it has been described, and that it is the pedicellate spikelet which is aborted. [ 51 | spike of the tassel. Except that the spikelets are pistil- late instead of staminate and that both spikelets are ses- sile or almost so, the elongated portion of these ears is scarcely distinguishable from the central spike. Further- more, for some reason (for which mere coincidence is not a satisfying explanation) a number of these ears have basal branches. Such ears correspond, not to the central spike alone, but to the central spike plus a few of the upper lateral branches of the tassel. Fusion. The evidence against the fusion hypothesis now amounts almost to complete proof at least for this type of ear. The rachis is a simple stem-like structure exhibiting not the slightest evidence of fusion in exter- nal morphology or internal anatomy. There is no more reason for suspecting fusion in this rachis than in the rachis of wheat, oats, barley or other cereals. Phyllotaxy. The phyllotaxy is verticillate or whorled rather than spiral. The ear is a simple spike with the spikelets arranged in whorls at the nodes (Plate IX). The number of pairs of spikelets at each node varies, but it is usually two or three. On one ear, beginning with the first node at the tip, the number of pairs of spikelets was 1, 1, 2, 2.5, 8, 2.5, B respectively for the first seven nodes. Thereafter three pairs of spikelets occurred at every node up to the 67th, beyond which it was impossible to dis- tinguish the nodes. In asecond ear the number of pairs of spikelets was 2, 2, 2, 2, 2, 2, 2.5 for the first seven nodes; two pairs of spikelets each on nodes eight to twenty-four; three or four pairs of spikelets on all addi- tional nodes. There is a tendency for the pairs of spikelets at one node to alternate in position on the circumference with those at adjacent nodes above and below. In a region of the ear where two pair of spikelets at each node is the [52] predominating condition, the result is a decussate ar- rangement which, since the spikelets are paired, actually appears as a double decussate condition. Spiral phyllotaxy, when it occurs, seems to be acci- dental. There are regions of the ear in which the phyllo- taxy ostensibly is spiral, but this is apparently the result of crowding, and is not the underlying or basic arrange- ment. This will become clearer when the number of rows is discussed. Number of Rows. The number of rows is variable. In the elongated region of the ear the spikelets do not appear to be arranged in definite rows, and the orderly arrangement characteristic of most maize ears is lacking. In the basal portion of the ears, the part covered by husks, the grains are arranged in rows, but the number is not constant. It increases rapidly toward the base. In one case row numbers of eight, ten, twelve and fourteen were counted on a single ear. There is not, however, the definite dropping of a pair of rows at any point, as in the cobs described by Weatherwax (17). At one level of the ear eight rows can be counted, at another (lower) level ten, but it is difficult to determine the points where pairs of rows have terminated. The number of rows is largely independent of the un- derlying arrangement of the spikelets. In one ear in which the number of rows increased progressively from eight to ten, and from ten to twelve, the underlying ar- rangement of the spikelets remained the same, three pairs of spikelets at each node. The fact that there are rows at all seems to be largely the consequence of crowd- ing. ‘The spikelets apparently are forcibly crowded into that arrangement which is the most efficient from the standpoint of utilizing space. Certainly there is no defi- nite underlying structure which results in the orientation [ 53 | of the spikelets into clear-cut ranks as in the case of ears of many North American varieties. Straight rows in this case are probably similar to the rows seen in young in- florescences of Pennisetum (Weatherwax, 19). Pairs of Rows. Lines of demarcation between pairs of rows, commonly seen in ears of North American va- rieties, usually separate one rank of paired spikelets from the adjoining rank of paired spikelets. In the Guarany maize the lines of demarcation separate the two spikelets of the same pair and a pair of rows thus represents spike- lets from two adjacent ranks of spikelets intermeshing with each other like the teeth of gears. This is illustrated in Plate X. Compaction of the Inflorescence. Since the chief, if not the only real, difference between the upper and lower regions of these ears is one of elongation, it follows that one of the important characteristics of the normal ear of maize lies in the fact that it is a strongly compacted inflorescence. In the ears of Guarany pod corn there is a strong correlation between the amount of com- paction and other characteristics. As the ear becomes more compacted, the internodes become shorter, the rachis becomes thicker and the number of rows of grain increases. These associated changes are illustrated by the three specimens shown in Plate X. The correlation between condensation of spikelets on the lateral branches of the tassel and the number of rows on the ear which Anderson (1) has reported may repre- sent one manifestation of the relationship between com- paction and number of rows. But since the correlation which he observed is characteristic of North American maize and apparently does not hold for South American varieties (1), it is more likely that condensation of spike- lets is correlated with the number of spirals of the under- [ 54 | lying spiral phyllotaxy which, as will be shown later, is characteristic of North American maize. If these elongated inflorescences of Guarany pod corn are a valid criterion (and there seems to be no good reason to doubt that they are since they do not appear to rep- resent a teratological condition), then the ear of maize is much less complex than has generally been supposed. It is nothing more than astrongly compacted spike in which pairs of spikelets are borne in varying number in whorls at the nodes of a simple non- or weakly-articulated ra- chis. This description, however, applies only to ears of the general type exemplified by the Guarany variety. A second type of ear apparently exists and is considered below. Crosses of Maize and Teosinte The hypothesis, already discussed, of a polystichous ear developing from a distichous spike through shorten- ing and twisting of the axis was based on observations which Collins (5) had made upon segregates from maize- teosinte hybrids. In his paper he showed photographs of spikes representing the various steps assumed to have been involved; the restoration of the aborted spikelet, the yoking of adjacent segments of the rachis, and finally the change from distichy to polystichy. His figures are not too convincing. Especially lacking are specimens showing the transition from independent to yoked rachis segments. Nevertheless his description is correct in its principal details. My own material, probably because I have had much larger populations than Collins to draw from, show the transition from the independent to the yoked condition much more clearly than did his (Plate XI, figs. A, B, C). In teosinte the joints of the rachis stand almost directly above or below one another to pro- duce a structure which Collins has aptly described as [ 55 ‘‘resembling a string of triangular beads”’ (Plate XI, fig. A). As the axis shortens or the rachis segments increase in size, the segments tend to lose their linear orientation and to assume a zigzag arrangement (Plate XI, fig. B). Finally adjacent segments, whose spikelets arise on op- posite sides of the axis, become diametrically opposed to each other and yoked to each other (Plate XI, fig. C). The other steps which Collins described also occur regularly and here, too, my material is better for pur- poses of illustration than was his. The change from soli- tary to paired spikelets is shown in Plate XI, figs. A and D. This may occur either before or after the rachis seg- ments have become yoked. The change from distichy to polystichy to produce an eight-rowed ear is shown in Plate XI, figs. E and F. Collins attributed this change to a twisting of the axis and the term is satisfactory if it is not used too literally. What actually occurs is that the yoked segments become arrayed in two planes instead of one, an arrangement which makes much more efficient use of the circumference of the rachis as a spikelet-bearing surface. Collins showed diagrammatically how ears with additional row numbers would be produced by additional ‘“‘twisting’’ of the axis. These were not encountered in my material. A number of ten-rowed ears occurred, but they were, for the most part, ‘‘disharmonious’’ ears with twisted cobs. Apparently the only ‘‘normal’’ ears in which the yoking is readily discernible are four-rowed and eight-rowed ears. An important change which Collins did not note, or at any rate did not describe, is the compaction of the in- florescence. A spike which has passed through the three stages which he described is still far removed from an ear of maize. It must pass through still another stage, a change from a lax spike to acompacted spike. Plate XI, fig. G illustrates an eight-rowed spike which has become [ 56 | compacted and which is recognizable as an ear of maize. What conclusions can be drawn from this series of transition forms which occurs in maize-teosinte hybrids? Collins was cautious on this point, stating that these in- termediate forms are of value primarily in throwing light upon the morphology of the ear and explaining its evo- lution in a mechanical sense. Weatherwax (18) was even more circumspect in his statement on this subject, which follows: ‘‘Hybrids between maize and teosinté will al- ways exhibit suggestive series; but, until we are more sure of the homologies between these two genera, it is futile to expect much information from the hybrids, for they will be speaking in a language that we cannot un- derstand. ’’ Now, when the homologies of maize and teosinte are somewhat better understood than they were two decades and more ago, and when the role of genetic mechanisms in evolution has also become much clearer, speculation on the meaning of the hybrid forms may be permitted. If maize has originated from teosinte under domesti- cation, as some students of the problem have assumed, then hybrids of the two species are capable of revealing something about the nature and extent of the genetic changes which have occurred. If we accept this inter- pretation we need only to state here that the genetic changes involved have been numerous and far reaching in their effects, and that the period of domestication of maize has either been very long indeed or mutations must have occurred at an unprecedented rate, for maize differs from teosinte by numerous genes distributed among a number of different chromosomes. If on the other hand teosinte is the product of the hybridization of maize and Tripsacum, as postulated by Mangelsdorf and Reeves (18), then hybrids of maize and teosinte simply show the effects which are produced upon maize by various doses [ 57 ] of Tripsacum germplasm. This becomes especially im- portant in considering the correlative hypothesis of Man- gelsdorf and Reeves (13) that new types of maize orig- inating directly or indirectly from the hybridization of maize and Tripsacum comprise the majority of the vari- eties of Central and North America. Teosinte (Tripsacum) germplasm consistently has cer- tain definite effects upon the structure of the ear of maize. These effects vary considerably with the amount of teo- sinte germplasm involved and with the genetic level at which it is operating. In spikes approximately interme- diate between those of maize or teosinte (spikes similar to those borne on F plants of maize-teosinte hybrids), small differences in the proportion of maize and teosinte germplasm can mean drastic changes in external mor- phology, for example: from distichy to polystichy or from single spikelets to paired spikelets. As the segre- gates approach either end of the range, however, much larger increments are necessary to produce these changes. These facts, revealed by Fe segregates of maize-teosinte hybrids, are even better illustrated by stocks in which an entire teosinte chromosome or a part of a chromosome has been transferred to a uniform inbred strain of maize by repeated backcrossing accompanied by selection. Stocks developed in this way are isogenic in nine of their ten chromosomes and therefore approximately identical with the original inbred strain of maize. The remaining chromosome, however, has been substituted for, wholly or in part, by a corresponding chromosome from teosinte. The substitution of a small amount of teosinte germ- plasm for maize germplasm at this level has no apparent effect whatever upon such characteristics as paired and single spikelets; the spikelets remain completely paired. But these doses of teosinte germplasm do consistently reduce the number of rows of grain and they consistently [ 58 | tend to make the cob stiffer and more indurated, and to give the rows of grain a rigid vertical alignment some- times accompanied by spaces between pairs of rows. The end product of these various tendencies is most com- monly an eight-rowed ear in which spikelets in one plane are diametrically opposed to each other, while those in the plane at right angles to the first are also opposed to each other but in positions alternating exactly with those in the first plane. Here, then, is precisely the kind of ear which would have developed according to Collins’ yoking and twist- ing hypothesis. The only difference is that the rachis having become large, and sometimes pithy, the yoking is no longer actually apparent. But these ears also con- form completely to Weatherwax’s conception of the ear as an example of spiral phyllotaxy. Indeed the spiral phyllotaxy is so rigid that the addition of another pair of rows of grain, the equivalent of a half-spiral, results in an unbalanced phyllotaxy usually accompanied by a twisting of the cob. This type of ear is obviously quite different from that illustrated by the Guarany pod corn; and yet the one is nothing more than a modified form of the other. That there are differences in maize varieties with re- spect to phyllotaxy in the branches of the tassel has al- ready been noted by Anderson (1) who finds a spiral phyllotaxy predominating in some varieties, a whorled phyllotaxy apparently characteristic of others. Other Differences in Maize Types If there are in nature two types of ears, one with a whorled phyllotaxy another with a spiral phyllotaxy, it is reasonable to suspect that other differences between the two types also exist. [ 59 | Arrangement of Spikelets. Another important difference in the two types may lie in the arrangement of the sessile and pediceled spikelets on the ear. If the ear of maize is a spike only recently derived from a pan- icle by reduction of branches then the arrangement of the paired spikelets with respect to the position of the pedi- celed and sessile member of each pair should be largely a random one because the branches of the panicle are arranged at random with respect to the position of the pediceled and sessile spikelet of the lowest pair. If, how- ever, the ear has come under the influence of Tripsacum genes to the extent that it behaves as a structure de- rived from the distichous spike of teosinte, then the ar- rangement of the spikelets may well be a systematic one. Unless I have inadvertently overlooked a reference to it in the literature, the question of the arrangement of sessile and pediceled spikelets on the ear seems never to have been answered. Collins (5),assuming that it is always pediceled spikelets which abort when rows of grain are dropped between the base and tip of the ear, asserted that he had never seen an instance where the dropped rows were either adjacent or separated by two rows. This he regarded as evidence that the arrangement of pedi- celed and sessile spikelets around the circumference of the ear is not the one expected from the fusion of two- ranked spikes. He apparently concluded, although he certainly did not prove, that the arrangement is, instead, the one that would be expected if the ear had resulted from yoking and twisting in the distichous spike. Actually, since all pistillate spikelets are sessile (al- though one member of each pair is potentially pediceled) it is impossible to determine from the external appearance of the ear the nature of the arrangement of its spikelets. But since the central spike of the tassel is clearly the homologue of the ear, and since on the central spike the [ 60 | sessile and pediceled spikelets are often readily distin- guished, it should be possible to draw some inferences about spikelet arrangements in the ear by studying them on the central spike. The spikelets on the central spike of the tassel are borne in more or less distinct rows. These are less clear cut, to be sure, than those on the ear, but can neverthe- less be traced with a fair degree of accuracy. The posi- tion and arrangement of the spikelets on a central spike can be charted by tracing a vertical rank of spikelets from its base, determining the position of each pair by actual measurement and noting whether the pediceled spikelet is at the right or left of the sessile one. This is repeated with each vertical rank until the entire circumference of the spike has been charted. A chart of the spikelets of the central spike, prepared in this way, is almost the equivalent of peeling the sur- face of the rachis, unrolling it and spreading it flat. Plate XII, figs. D and E, shows in diagrammatic form the relative position and the arrangement of sessile and pediceled spikelets on portions of central spikes of two inbred strains of maize; the first a strain of the Guarany variety ; the second a North American strain well known to agronomists, K155. It is to be noted that in the Guarany maize the ar- rangement of the spikelets with respect to the position of the sessile and pediceled is a random one, or at least the deviation from randomness on any single row is clearly not significant. The central spike of K155 presents quite a different aspect. In one row (the center row in the diagram) the spikelets are arranged at random, but in the remaining rows the arrangement is either clearly systematic or it approaches this condition. In the second row from the left, for example, the pediceled spikelet is at the left in [ 61 ] all (seven) pairs. In the fourth row the pediceled spike- let is at the right in all (eight) pairs. A systematic ar- rangement is also closely approached in the first and fifth rows, the former having the pediceled spikelet at the right in six of the seven pairs, the latter having the ped- iceled spikelet at the left in eight of the ten pairs. It is to be noted further that the second and fifth rows in this diagram are mirror images of the first and fourth with respect to the position of the pediceled and sessile spikelets. How does the arrangement in Fig. EK compare with the arrangement theoretically expected if the ear of maize had originated from the spike of teosinte by yoking and twisting? Plate XII, fig. A shows the arrangement in teosinte when the abortive spikelet becomes functional, as it does in some cases, especially when the tunicate gene is superimposed upon teosinte by crossing and backcross- ing. Here the arrangement is completely systematic and the two ranks are mirror images of each other with re- spect to the position of pediceled and sessile spikelets. Plate XII, fig. B shows the theoretical arrangement after yoking has occurred. This arrangement has actually been seen in a segregate from a maize-teosinte hybrid in which the spikelets were yoked, both spikelets of each pair were functional and the pediceled spikelets were distinguishable from the sessile ones. Plate XII, fig. C shows the theoretical arrangement resulting from yoking combined with twisting of the axis. Here rows one and three and rows two and four (rows diametrically opposed to each other on a cylindri- cal rachis) are mirror images of each other with respect to arrangement as well as position of the spikelets. In two important characteristics, systematic arrange- ment of the spikelet and mirror imagery, K155 is like the ear theoretically developed from the spike of teosinte [ 62 ] through yoking and twisting. But in a third important feature, the sequence of pediceled and sessile spikelets, there is a distinct difference. In the ear developed from teosinte by yoking and twisting (Fig. C) the sequence is PSPSSPSP. No matter at what point this sequence begins it always involves at some point on the circum- ference two rows of sessile spikelets adjacent to each other and diametrically opposed on the rachis to two rows of pediceled spikelets adjacent to each other. In K155 (fig. FE) on the other hand (omitting the single row with random arrangement and letting the letters S and P respectively, represent the predominating condition in each row of spikelets) the sequence is SPPSSPPS. Here no matter on which row the sequence begins, the pat- tern is one of two rows of sessile spikelets alternating with two rows of pediceled spikelets. This sequence is not unique to K155. It has been ob- served in other strains where the arrangement of spikelets approaches the systematic. The sequence illustrated in Fig. C has also been ob- served in several varieties. his means, if it means any- thing at all, that the ear of maize sometimes behaves as though it had been derived from the spike of teosinte by twisting of the axis in the sense in which Collins used the term. Whether this sequence is as common as the other remains to be determined. Other evidence that both random and systematic ar- rangement of spikelets occur is found in ears which are partly or wholly staminate at the apex. When wholly staminate both spikelets are usually, though not always, sessile as they are when wholly pistillate. When the spikelets are mixed, however, the pistillate one is usually sessile, the staminate one pediceled. In the Guarany va- riety the arrangement of staminate and pistillate spike- lets, when the two are mixed, seems to be essentially [ 63 ] random. In ears derived from maize-teosinte crosses which have been repeatedly backcrossed to maize, the arrangement is definitely systematic. The entire subject of spikelet arrangement deserves more study. In the meantime it is already reasonably clear that on the central spike of the tassel, at least, the arrangement approaches a random condition in some va- rieties and is systematic in others. There is reason to sus- pect that these differences occur also on the ears. It is supposed that the random arrangement is characteristic of ‘‘pure’’ maize; the systematic of Tripsacum-contam- inated maize. Varieties which are completely random or completely systematic in the arrangement of their spike- lets are perhaps not common, the majority of varieties probably exhibiting some degree of intermediacy be- tween the two extremes. Twisted Ears and Odd Numbers of Paired Rows. Attention has been called to Fujita’s observa- tion that ears in which the number of rows of paired spikelets is odd are often slightly twisted with the rows of grain exhibiting a tendency to form weak spirals. And mention has also been made of the fact that there is, in North American maize at least, a preponderance of ears with rows of paired spikelets in even numbers as opposed to those with rows in odd numbers. The two phenomena are probably related and both may well be the result of Tripsacum influence. At any rate they are most strik- ingly displayed in segregates of maize-teosinte hybrids. The predominance of ears with rows of spikelets (in this case single as well as paired spikelets) in even numbers is well illustrated by the following frequency distribution. Population Number of Rows of Spikelets _ Total 2 3 4: 5 (Durango Teosinte < Maize) F2 65 45 58 9 177 (Nobogame Teosinte X Maize) F2 86 71 115 27 299 [ 64 | In both crosses there is a decided deficiency of three- and five-ranked ears as opposed to two- and four-ranked ears. Also in both crosses there is a tendency for the three- and five-ranked ears to be twisted. They are ob- viously the result of a structural unbalance or asymmetry which induces stresses with visible effects. These phenomena, so conspicuous in maize-teosinte hybrids, are also readily discernible in maize. There are several explanations for them: 1. Reeves (15) suggests that there is a preponderance of ears with rows of spikelets in even numbers because the basic unit of the ear is a pair of rows of spikelets rather than a single row. Thus an eight-rowed ear is much more likely to become a twelve-rowed ear than it is to change to a ten-rowed ear. He regards these units as hereditary and in a broad sense they undoubtedly are. It seems more probable, however, that the unit, a pair of rows of spikelets, is a developmental rather than an hereditary one for it is not uncommon in maize-teosinte hybrids to find two-ranked and four-ranked spikes on the same plant. This is not to say that the number of ranks is independent of heredity. There is abundant evidence that the contrary is true. But apparently the genes do little more than determine in a general way whether the spike is to be few-ranked or many-ranked, while develop- mental factors determine the size of the units by which the differences are to be attained. 2. Any ear strongly influenced by Tripsacum is likely to exhibit a spiral phyllotaxy so rigid that only those combinations which involve complete spirals will be bal- anced and symmetrical and hence are more likely to oc- cur than those involving fractional spirals. This is per- haps no more than another way of saying that spiral phyllotaxy explains and accounts for the fact mentioned [ 65 | above: that the unit of development is a pair of rows of spikelets. 3. Reeves (15) has also pointed out that with a spiral phyllotaxy the rachis segments are probably hexagonal, fitting together ‘‘like the engaged teeth of mechanical gears.’’> With an odd number of rows of paired spikelets there is a physical necessity for some kind of distortion. Either the rachis segments must change shape or the ear must become twisted. The net result of these various factors, which are close- ly related to each other, is that there is a strong tendency for rows of spikelets to occur in even numbers. When they do occur in odd numbers there is a strong tendency for the ear to be twisted. Stated in still another way ; there is a strong tendency for ears of maize to be bal- anced and symmetrical, rather than unbalanced and asymmetrical The above discussion applies especially to ears with a rigid spiral phyllotaxy and indurated tissues, the kind of ear derived from maize-teosinte crosses and the kind pre- sumably resulting in nature from Tripsacum contamina- tion. What of “‘pure’’ maize, with its whorled phyllo- taxy and random arrangement of spikelets, with respect to these phenomena! Many varieties of Bolivian and Peruvian maize, both prehistoric and modern, have rows of grain more strongly spiral than those of the twisted ears of North American varieties. Extreme forms, as Anderson (unpublished) has noted, have ‘‘spiral cross- rowing in two directions, like a pine cone.’’ But this condition in Bolivian and Peruvian maize does not give the impression of abnormality. Like the pine cone with which it has been compared, it appears to be perfectly normal. Furthermore the occurrence of spiral rows in Bolivian and Peruvian maize is usually independent of row number. Ears with pairs of rows in even numbers are [ 66 | no less subject to the occurrence of spiral rows than ears with pairs in odd number. Finally, in varieties of this kind there is not the pre- ponderance of ears whose rows of paired spikelets are in even numbers, so commonly found in North American varieties. Ears with ten, fourteen and eighteen rows are as frequent as ears with eight, twelve and sixteen rows. This, like the presence of spiral cross-rows, is character- istic of both prehistoric and modern varieties as is shown by the following frequency distribution, which includes modern ears from Isla del Sol in Lake Titicaca and pre- historic ears excavated by Dr. Julio Tello from the Para- cas Necropolis in Peru and by Dr. Junius Bird from Arica, Chile. Data on row numbers in the last named collections were kindly furnished by Dr. Edgar Ander- Collection Number of Rows of Grain Total 10. 1214. 16-238" 620. 722 Isla del Sol, Bolivia 5 12 19 9 5 50 Paracas, Peru (Prehistoric) 4 10 5 2 2 1 24 Arica, Chile = 1 14 24 10 2 51 son. The situation is the same in all three cases; there is no deficiency of ears in which the number of pairs of rows is odd, indeed the modal number in all cases is fourteen. Discussion AND CONCLUSIONS The conclusion that there are two morphologically dis- tinct types of maize ears is not new. Mangelsdorf and Reeves (18) have previously suggested that the maize varieties of America comprise two more or less distinct groups: (1) ‘‘pure’’ maize which traces its descent directly from the original wild corn; (2) Tripsacum-contaminated maize resulting directly or indirectly from the hybridiza- tion of maize and Tripsacum. It was assumed that the differences between these two groups were reflected in [ 67 | the ears, but no attempt was made to distinguish them on the basis of differences in their basic structures beyond pointing out that many of the Andean varieties (presum- ably ‘‘pure’’ maize) are characterized by irregular rows of grain, while many Central and North American va- rieties (presumably contaminated by 'Tripsacum) exhibit straight rows frequently separated in pairs. Extreme types approaching these descriptions were found by Man- gelsdorf and Cameron (12) in a study of maize varieties of western Guatemala and are illustrated in their paper. Now the evidence that there are two kinds of maize ears basically different in origin, structure and phyllo- taxy, if not conclusive, is at least highly convincing. The first type of ear which is characteristic of the Guarany pod corn and is probably typical of ‘‘pure”’ maize may be described as a compacted spike with pairs of pistillate spikelets borne in varying numbers at the nodes of a simple, weakly articulate rachis. The ear has a whorled phyllotaxy but may assume, more or less for- tuitously, especially if the rows and kernels are crowded, the aspect of spiral phyllotaxy. The rows may _ be straight, especially if the row number is low; but straight rows like spiral phyllotaxy seem to be, as they are in cer- tain other grass spikes, fortuitous rather than the reflec- tion of a particular kind of underlying structure. Straight rows, in some cases at least, seem to represent nothing more than an efficient arrangement assumed upon crowd- ing. Pairs of rows are not distinctly separated and if there is sometimes an apparent line of demarcation be- tween pairs of rows it separates the two spikelets of the same pair and not two pairs of spikelets. ‘The number of rows is a reflection of the degree of compaction rather than of the complexity of the phyllotaxy for the number of rows can change decidedly from base to tip without any change in the underlying phyllotaxy. With respect [ 68 ] to the position of the sessile and potentially pedicellate spikelets, the arrangement of spikelets on the ear is ran- dom or nearly so. The tissues of the rachis and glumes are not strongly indurated. There is no tendency, as in many North American varieties, for ears with pairs of rows in odd numbers, ten, fourteen, eighteen, etc., to be in the minority or for such ears to be twisted. This type of ear is fundamentally identical with the spikes of certain other species of grasses and like many of them is probably derived from a panicle as the result of reduction of branches. There is not the slightest evi- dence of fusion. There is abundant evidence, however, that the ear is derived from a panicle for the central spike of the tassel, which is unquestionably the homologue of the ear, is still surrounded with basal branches. Indeed the tassel is a perfect example of a combination of pan- icle and spike and illustrates splendidly the transition from one to the other. The change from panicle to spike is obviously relatively recent, but whether it is the product of domestication, wholly or in part, there is at present no way of determining. The fact that the Guarany va- riety, which is primitive in certain other characteristics, frequently exhibits basal branching of the ear, suggests, though it does not prove, that the wild maize with which domestication began was at least moderately branched in its lateral (probably pistillate) inflorescences. This type of ear, it may be said again, differs in no single important characteristic from the inflorescences of other grasses. True, it usually is wholly pistillate, but there are other grass inflorescences, such as the pistillate forms of the dioecious species of Monantochloé, Jouvea, Buchloé and Eragrostis, which are entirely pistillate. It usually has a massive rachis and large caryopses, but there are varieties of sorghum which have the rachis thicker than the most slender maize cob and the cary- [ 69 ] opses larger than the smallest maize kernels. This ear of maize seems not to possess a single characteristic in which it is consistently unique. Its uniqueness lies rather in the particular combination of characters which it possesses. The fact that it is wholly pistillate, bears large caryopses ona massive rachis and is a strongly compacted inflores- cence, readily distinguishes it from all other grass inflo- rescences. Of these characteristics the last, compaction, is by no means the least important. Once this fact is recognized —that the ear of maize is one of the most strongly com- pacted of all inflorescences—then much of the mystery which has surrounded the ear disappears and it becomes scarcely more difficult to understand than a head of cab- bage which also is a strongly compacted but otherwise relatively simple structure. The second type of maize ear, the type which presum- ably results from ‘Tripsacum-contamination is, like the first, a compact spike. It differs from ‘‘pure’’ maize pri- marily in having a spiral rather than a whorled phyllo- taxy and asystematic rather than a random arrangement of sessile and pedicellate spikelets. Indeed a brief de- scription would appropriately term the first ‘‘whorled- random’’; the second ‘‘spiral-systematic.’” But there are other differences as well. The tissues of the rachis and glumes of the spiral-systematic type are usually in- durated ; the rachis as a whole being quite tough, though evidence of an inherent rachis fragility is sometimes dis- cernible. The rows of grain are distinct and not infre- quently separated in pairs. When this occurs a pair of rows is the equivalent of a row of paired spikelets. The tendency for a rigid vertical alignment of the spikelets is so strong, probably as the result of an inflexible spiral phyllotaxy, that ears in which the number of pairs of rows is odd are in the minority and are usually twisted. [ 70 | The spiral-systematic ear, like its putative precursor the whorled-random ear, furnishes no evidence whatever of fusion. It does, however, have the aspect of a fusion product and especially is this true of eight-rowed ears in which the pairs of rows are distinctly separated with shal- low clefts or apparent lines of fissure in the cob corres- ponding to the lines of demarcation. Furthermore the sequence of sessile and pedicellate spikelets around the circumference, SSPPSSPP, sometimes encountered, is the kind of sequence which would be expected as the re- sult of fusion of two-ranked branches. Nevertheless there is neither evidence of fusion nor necessity for assuming that it has occurred. There is perhaps no single characteristic by which these two types of maize ears can always be distinguished from each other. Actually pure forms of either are prob- ably rare and intermediate forms are more common than the basic types. There are, however, several circum- stances usually associated with the two general types. The ‘‘pure’’ maize with its derivatives is the predomi- nating type in the Andean region of Peru, Bolivia and Ecuador and in the adjoining lowlands of South Amer- ica, particularly in Paraguay. It is also the predominat- ing type in the prehistoric maize of South America. The same or asimilar ear is found at high altitudes in Guate- mala and, also at high altitudes, in some localities in Mexico. The Tripsacum-contaminated maize is typical of most of North and Central America and of most of the low- lands of South America. It is realized that there is still no actual proof for the existence of “‘pure’’ and Tripsacum-contaminated maize. Nevertheless it is now reasonably certain that there are two basically different types of maize ears with more or less distinct patterns of geographical distribution. Since Eve there is complete interfertility in maize varieties from all parts of America, it is logical to conclude that the one has been derived from the other. Segregates from maize- teosinte crosses and isogenic stocks developed by repeated backcrossing of such crosses to inbred strains of maize demonstrate that teosinte genes can produce the changes by which the two types are distinguished. All of these facts indicate, though they obviously do not prove, that a pure maize originating in South America and bearing ears characterized by whorled phyllotaxy and random arrangement of sessile and pedicellate spikelets has be- come modified by the introduction of Tripsacum (teo- sinte) germplasm to produce an ear characterized by spiral phyllotaxy and systematic arrangement of spike- lets. ACKNOWLEDGMENTS Tothe Massachusetts Society for Promoting Agricul- ture I am indebted for a grant, to be used in the study of maize varieties of Latin-America, of which this paper is an outgrowth. Professor Oakes Ames generously loaned me, from his personal library, the copy of Lyte’s Herbal from which Plate VII is reproduced. The draw- ings which illustrate the paper were made by Mr. Gordon Dillon. I wish to thank Dr. Edgar Anderson and Dr. Leon Croizat for allowing me to include certain of their unpublished observations. And I am especially indebted to Dr. R. G. Reeves for permitting me to read and to refer to his unpublished paper on the morphology of the maize ear; a paper which, except for the exigencies of war, would certainly have been published before mine. [72 ] LITERATURE CITED . Anderson, Edgar, 1944. Homologies of the ear and tassel in Zea Mays. Ann. Mo. Bot. Gard. 31: 325-342. . Arber, Agnes, 1934. The Gramineae. Cambridge, England. . Bonnett, O. T., 1940. Development of the staminate and pistil- late inflorescences of sweet corn. Journ. Agric. Res. 60: 25-37. . Collins, G. N., 1912. Origin of maize. Journ. Wash. Acad. Sci. Tai 4 Ue OF . Collins, G. N., 1919. Structure of the maize ear as indicated in Zea-Euchlaena hybrids. Journ. Agric. Res. 17: 127-135. . Dodoens, Rembert (Lyte, Henry), 1619. A new herbal, or his- torie of Plants:.... London. . East, E. M., 1910. A Mendelian interpretation of variation that is apparently continuous. Amer. Nat. 44: 65-82. . Fujita, T., 1939. Uber die Organstellung bei Maiskolben. Jap- anese Journ. Bot. 10: 113-140. . Kellerman, W.A., 1895. The primitive corn. Meehan’s Month- ly 5: 44, . Kempton, J. H., 1919. The ancestry of maize. Journ. Wash. Acad. Sci. 9: 3-11. . Langham, D. G., 1940. The inheritance of intergeneric differ- ences in Zea-Euchlaena hybrids. Genetics 25: 88-107. . Mangelsdorf, P. C. and Cameron, J. W., 1942. Western Guate- mala a secondary center of origin of cultivated maize varieties. Bot. Mus. Leaflets, Harv. Univ. 10: 217-252. . Mangelsdorf, P. C. and Reeves, R. G., 1939. The origin of In- dian corn and its relatives. Texas Agric. Exper. Sta. Bull. 574. . Montgomery, E. G., 1906. What is an ear of corn? Pop. Sci. Monthly 68: 55-62. . Reeves, R. G. (in press). Morphology of the maize ear with ref- erence to its origin. . Weatherwax, Paul, 1918. The evolution of maize. Bull. Torrey Bot. Club 45: 309-342. . Weatherwax, Paul, 1920. A misconception as to the structure of the ear of maize. Bull. Torrey Bot. Club 47: 359-362. . Weatherwax, Paul, 1923. The story of the maize plant. Chicago. . Weatherwax, Paul, 1935. The phylogeny of Zea Mays. Amer. Midland Nat. 16: 1-71. [ 73 ] EXPLANATION OF THE ILLUSTRATIONS Piate VII. Reproduction of a page of Lyte’s (1619) New Herbal, which describes the inflorescences of maize. Piate VIII. Example of extreme elongation in an ear of Guarany pod corn. The lower half of this ear was enclosed by the husk and is compacted with the kernels arranged in rows. The upper half was free of the husks and is intensely elongated. Piatre IX. The basic structure of the maize ear is illustrated by this enlarged (1.5) photograph of the elongated terminal portion of an ear of Guarany pod corn. A pair of spikelets has been removed at each node to show the simple rachis bearing spike- lets in whorls at the nodes. Piatt X. Relation between compactness and num- ber of rows. The left ear has eight rows at the tip; fourteen rows at the base. The center ear and the naked rachis at right show how the internodes be- come progressively shorter toward the base. The number of spikelets at each node remains approx- imately the same. [74] Piate XI. Transitions from teosinte to maize il- lustrated by the pistillate spikes of F2 segregates of a maize-teosinte hybrid. A. Most teosinte-like spike with single spikelets alternating in two ranks. B, C. Transition from independent to yoked rachis segments. D. Approximately like A, except that spikelets are paired. E. Yoked rachis segments combined with paired spikelets to produce a dis- tichous, four-rowed spike, bearing four spikelets at each joint of the rachis. F. Portion of spike and single rachis joint of an eight-rowed spike derived from a four-rowed spike by “‘twisting’’ of the axis. G. Most maize-like spike. It differs from F pri- marily in degree of compaction, Pirate XII. The arrangement of sessile and pedi- cellate spikelets around the circumference of va- rious spikes. A. Teosinte spike. B. Teosinte spike with yoking of rachis segments. C. Theoretical arrangement resulting from yoking combined with twisting of the axis. D. Actual arrangement of spikelets on a portion of the central spike of the tassel of an inbred strain of Guarany maize. Note that arrangement is random or nearly so. E. Ac- tual arrangement on a portion of the central spike of North American inbred K155. Note that ar- rangement is strongly systematic. [75 | PLATE VII Of Turkie Corne, or Indian wheat. The kindes. Urkith tobcat is of one,and of manp forts. A rian fhall mot find in this coun: Ts (Wi fathion and. crowing) moze than one kind, butincolour the fede 07 qrayne doth much differ : fo2 one beareth a brotune graye o2 come, the other ared,the third a pellow, and the fourth a iwhtte come o2 grape. De which colour votl Itkewife remapne both in the eares and floures. The defcription. T His Come is antarucllous range plant,nothing refembling anv other kind of arapne : for if baingety forth bts fade cleane contearie from the place whereas the Floures grew, tobhich is againt the nature and kinds of all other plants, which brine forth theirfrutt there, whereas they hauc borne their Floure. his Come beareth a high beluie o2 Femme, arid berplong, round, thicke, firme, and below to- {nards the roote of a beetwnith colour, twith fundzte knots and ioprits , trom the which Dependeth lovig,and larae leaues, like the leaues of Spire oz Polerwve : at the highett ef the ffalkes, groto tdle and barren eares , which bring forth na-hing but the floures 02 bloffomes, which are fometimes bjowne, fomvtinics red, fome- times yellow, and fometimes white, agreable with the colour of the frutt, which comet forth afterward. he fruttfull cares 80 grow, bpon the fides cf the ems amongtt the leaues, the tohich eares be great and thicke , and coucred with manp leanes, fo that one cannot (ee the fapd earcs, bppon the bppermoft partof the fapd eares there qrolwmanp long hayate thaedes, which ifue forth at the cnds 02 popnts of the leaucs, courring the care, aud doe thew themfelucs about the time that the fruit 02 care Wareth ripe. Dhe grapne 03 Cede which groweth in the cares, is about the quantitte o2 bignefle of a peale , of colour in the out-fide, fometiines Lootone, fometimes redre, and fometimes twhite, and tn the ui-fide tt ts iit colour- {white and in talle (weet, growing oderlp about the cares , Wnine o7 ten ranges 0} rowers. The place. This grapne grotweth in Lurkie, alec itis bled in the time cf dearth. The time Atis Cotwne in April, and ripein Guguff. The names. Chey doe notw call this grapne, Framentum T urcicum, and Frumentum Afiati- cum: in French, Ble de Turquie, 02 Ble Sarazin: inbigh Dutch, Wurkte iow: in bafe Almaigne, Worcklehcoren + in Cngliih , Larkilty Cone, 02 Jndian- wheat. The nature. and vertues. There is as pet no certapne erperience of the naturall bertues of thi« come. Che bead that is made thereofis eve andhard, gauing beryp {mali fatneffe 02 topfture, wherefore men may cafilp iudge, that it nourifheth but little, and ts euill of digeftion, nothing poimpstanle to the bread made of Theat, as fome haue falfly affirmed. PLATE VIII Ls PLATE he ®t — — aS ae : ~. ie << SS _ SS ra ne)» 3) ; . ' sada) bee ED “ingg PLATE X] TE XII SS ae 4s Ff F ffs ~~ NNUNNOUNUN he in a aw aw a SS P FPA 2? yf >> >>> ~\ NS ~~ SN \ 4 FF SZ HX fo r >> >>> > Oe ese | -_— ~~ _. . . gf yes LQ ly) ae eK, !,.tti“‘(‘iaS™’SC*C 5 | BOTANICALMUSEUM LEAFLETS HARVARD UNIVERSITY Campripcr, Massacunusetts, Ocroper 23, 1945 VoL. 12, No. 3 AFRICAN ORCHIDS XVI. BY V. S. SUMMERHAYES (Royal Botanic Gardens, Kew) A SUGGESTED re-classification of the genus Diapha- nanthe Schltr. together with descriptions of new species and new combinations in the genus form the greater part of this contribution. There are also descriptions of new species of other genera of Angraecoid orchids and of one species of Polystachya. The sequence, as in previous con- tributions, is that of Schlechter’s ‘‘Das System der Or- chidaceen’’ (Notizbl. Bot. Gart. Berlin, 9 (1926) 563- 591). The type specimens of all the new species are in the Kew Herbarium. Polystachya (§ Cultriformes) canaliculata Sum- merhayes sp. nov.; a P. fulvilabia Schltr. foliis angusti- oribus, inflorescentiis foliis multo brevioribus, floribus albis nec flavis, sepalis obtusis vel apiculatis, labelli ungue brevi, lobo intermedio ruguloso canaliculato; a P. Mild- braedu Kraenzl. mento breviore, labello basi magis an- gustato lobis lateralibus minoribus lobo intermedio ru- guloso satis distinguenda. Herba epiphytica; caules caespitosi, erecti, usque ad 12 cm. alti, leviter ancipites, siccitate sulcati, apice mono- phylh, basi cataphylla vaginiformi lineari acuta usque ad 7.5 em. longa instructi. Foliwm erectum, lineari-ligula- tum, apice rotundatum, basi cuneatum, 9—25 cm. longum, [ 89 ] 5-8 mm. latum, quinquenervosum. Jnflorescentia erecta, folio multo brevior, 4-7 em. alta, superne subdense pau- ciflora; pedunculus gracilis, 8-5 cm. longus, laevis, sub- teres vel leviter anceps, supra medium vagina singula instructus; bracteae subpatentes, triangulari-ovatae, acu- minatae, 1-2 mm. longae. Fores patentes, albi, rubro- notati; pedicellus (cum ovario) 7-9 mm. longus. Sepalwm intermedium lanceolato-oblongum, apice apiculatum, basi leviter dilatatum, 5.5-7 mm. longum, 8.5—-4 mm. latum; sepala lateralia oblique triangularia, subacuta, margine superiore (postice) 8.5—-9.5 mm. longa, inferiore 5.5-6.5 mm. longa, basi 7.5-8 mm. lata, mentum obtu- sum 6-7 mm. longum formantia. Petala oblique oblongo- subspathulata, apice rotundata, 6-7 mm. longa, 2.25-2.5 mm. lata. Labellum curvato-deflexum ex ungue brevi lato ambitu oblongum, infra medium trilobatum, totum 9-10 mm. longum, 6-7 mm. latum (explanatum); lobus intermedius suborbiculari-quadratus, apice recurvato- apiculatus ut videtur retusus, marginibus undulatis, 5—5.5 mm. longus, 5—5.25 mm. latus, rugulosus, farinaceo- pilosus, secus medium canaliculatus ; lobi laterales erecti, ab intermedio angulo recto sejuncti, incurvatim triangu- lares, subacuti, fere 1.5 mm. longi et lati. Coluwmna sub- teres, antice excavata, circiter 2.5 mm. longa; anthera hemisphaerica, dorso cristata, antice leviter producta; rostellum leviter productum ; stigma transverse oblongo- ellipticum. TANGANYIKA TERRITORY: Nguru Mts., Mhonda, 600 m. alt., in deep shade in fork of forest tree, April 1943, Moreau 353 (Tyrer). “*One glossy deep grass-green leaf. Flower glistening white with two narrow crimson lines running from anther base to column foot. No scent.”’ It is not easy to decide which species is the nearest relative of this addition to sect. Cultriformes. The nar- row leaf and general flower structure indicate its affinity with P. Mildbraedii Kraenzl. and P. gracilenta Kraenzl., but there is also a strong resemblance to P. fulvilabia [ 90 | Schltr. and other species with short few-flowered simple inflorescences. So far as I am aware this is the first spe- cies of sect. Cultriformes recorded from the Nguru Mountains. Rhipidoglossum densiflorum Swmmerhayes sp. nov.; affine R. obanensi (Rendle) Summerh., a quo foliis apice valde inaequaliter bilobatis lobo longiore acuto lobo breviore fere nullo, floribus paulo majoribus, sepalis lat- eralibus quam intermedio longioribus angustioribusque, labello late orbiculari vel flabellato-orbiculari apice leviter exciso basi ante ostium calcaris callo transverso perhumili instructo satis distinguendum. Herba epiphytica; caulis elongatus, usque ad 40-45 cm. longus, 2-8.5 mm. diametro, teres, superne multi- foliatus, inferne vaginis foliorum delapsorum partim cir- cumdatus, radices numerosas flexuosas ramosas laeves 3-4 mm. diametro emittens. Folia adscendentia vel fere patentia, 1-2 cm. distantia; vagina inferne arcta, superne subdilatata, leviter compressa, valde nervosa, apice trun- cata, 1-2 cm. longa; lamina oblanceolata, oblongo-vel elliptico-lanceolata vel anguste lanceolata, apice valde inaequaliter bilobata, lobulo longiore leviter incurvato acuto 2-10 mm. longo, lobulo breviore brevissimo vel fere nullo, basi angustata, 4-12 cm. longa, 8-23 mm. lata, siccitate multinervosa, tenuiter coriacea. Inflores- centiae singulae vel geminae, foliis multo breviores, 1—4 cm. longae, fere ad basin dense multiflorae, basi vaginis paucis ovatis obtusis instructae; rhachis teres; bracteae vaginantes, superne dilatatae, truncatae vel subacutae, vix 1 mm. longae, 1-2 mm. distantes. Fores patentes, albi vel pallide cremei, eos Convallariae majalis revocan- tes, tepalis = conniventibus; pedicellus cum ovario 2-38 mm. longus. Sepa/um intermedium late ellipticum, apice obtusum vel rotundatum brevissime apiculatum, 2.4—2.8 [ 91 ] mm. longum, 1.7-1.9 mm. latum, concavum, trinervium ; sepala lateralia oblique late elliptico-lanceolata, apice sub- acuta, vel obtusa, 3-8.5 mm. longa, 1.8-1.7 mm. lata, 1-3-nervia. Petala suborbicularia, apice obscure apicu- lata, 2.25-2.75 mm. longa, 2-2.6 mm. lata, trinervia, nervis lateralibus ramosis. Labellum late orbiculare, flabellato-orbiculare vel fere transverse ellipticum, apice excisum, antice subtruncatum vel rotundatum et brevis- sime bilobatum, 2.75-8.75 mm. longum, 38.5-4.7 mm. latum, multinervosum, basi ante ostium calearis callo transverso perhumili leviter emarginato instructum ; cal- car e basi angustiore dilatatum, ellipsoideum, apice ro- tundatum, 2-3 mm. longum, circiter 1.75 mm. diametro. Columna deflexa, crassa, antice excavata, apice truncata, 1-1.25 mm. longa, androclinio leviter excavato; anthera oblonga, valde convexa, antice truncata; pollinia sub- sphaeroidea vel ellipsoidea, circiter 0.5 mm. longa, stip- itibus duobus subspathulato-ligulatis deorsum angustatis 0.75 mm. longis, viscidiis duobus elliptico-oblongis antice truneatis postice rotundatis; rostellum trilobum, lobo intermedio porrecto carnoso subspathulato, lobis lateral- ibus intermedio brevioribus rotundato-triangularibus ; fovea stigmatica quadrata. Capsulae ellipsoideae, inferne angustatae, leviter curvatae, 1-1.25 cm. longae, 5 mm. diametro. Ucanpva: Bunyoro, Budongo Forest, Sept. 1933, Eggeling 1431 (Uganda Forest Service No. 1376); Sept. 1935, Eggeling 2172; same date, Hancock 13 A; Nov. 1943, in fruit, flowered Busingiro Aug. 1944, Eggeling 5475; Bugoma Forest, 1050 m. alt., Dawe 759; Ma- saka, South Buddu, Fyffe 187; Mengo, Kyagwe, Sept. 1921, Lankes- ter 29; Mabira Forest, 1200 m. alt., Sept.-Oct. 1920, Diimmer 4438. Gason: Upper Ngounyé River, waterfall on Mboumi River, at Mbigou, Nov. 1925, Le Testu d767 (Type). Ancota: Moxico Distr., River Lupula, in evergreen vegetation, Jan. 1938, Milne-Redhead 4239. This striking little species well illustrates the difficul- ties of classification of the group of genera of which it is a [ 92 ] constituent species. I have placed it, for the time being, in Rhipidoglossum, since it is clearly closely related to R. obanense (Rendle) Summerh. and R. globuloso-calca- ratum (De Wildem.) Summerh. with which it shares the short dense inflorescences and short swollen spur. The orbicular curiously veined petals are also almost identical with those in other species of Rhipidoglossum. On the other hand in the shape of the leaves and the tendency of the whole plant to blacken on drying the species re- sembles the genus Sarcorhynchus. The narrow excision at the apex of the lip may be looked upon as an inter- mediate between the entire lip-apex of S. polyanthus (Kraenzl.) Schltr. and the deeply lobed lip of S. bilobatus Summerh., described later in this paper. Another com- plication is the presence of a low transverse rim-like cal- lus just in front of the spur. Although technically this should place the species in Diaphananthe the callus is so unlike the ordinary tooth found in that genus and the plant shows so little resemblance to other species of Dia- phananthe that it scarcely seems advisable to include it therein. Further work and discoveries of yet more intermediate species may make it necessary to consider all three genera as forming a single one under the earliest name Diapha- nanthe, but I do not feel that the evidence so far warrants such a change. Rhipidoglossum microphyllum Swmmerhayes sp. nov.; a R.longicaleart Summerh. habitu graciliore, rad- icibus numerosissimis, foliis multo minoribus, petalis ellipticis, labello apice subacuto, caleari breviore facile distinguendum. Rhipidoglossum? sp. Schlechter in Engler. Bot. Jahrb. 58 (1915) 605. Herba epiphytica; caulis = erectus, gracilis, leviter [ 93 ] tlexuosus, usque ad 15 cm. longus, 1-1.5 mm. diametro, teres, apice 3-7-foliatus, radices numerosissimas flexuo- sas simplices = recurvatas saepissime compressas usque ad 20cm. longas 2-3 mm. latas griseas per totam longi- tudinem dense emittens. Folia parva, erecto-patentia vel patentia, usque ad 5 mm. distantia; vagina arcta, valde nervosa, 4-5 mm. longa; lamina oblonga vel oblongo- ligulata, apice paulo inaequaliter bilobata, lobulis rotun- datis, 7-18 mm. longa, 2-5 mm. lata, griseo-viridis. Inflorescentiae ex axillis foliorum delapsorum exortae, folia superantes, arcuato-decurvatae, 1.5-2.5 cm. longae, sublaxe 4-8-florae ; pedunculus 7-10 mm. longus, vaginis 3-4 infimis truncatis superioribus lanceolatis acutis in- structus; rhachis fractiflexa, teres; bracteae laxe ochre- atae, apice triangulares acutae, 1-1.5 mm. longae. Flores patentes, cremei, 1.5-2.5 mm. distantes; pedicellus cum ovario 3-3.5 mm. longus. Sepa/um intermedium obovato- oblongum, apice subacutum, 3 mm. longum, 1.5 mm. latum, concavum; sepala lateralia oblique et leviter cur- vatim oblongo-ligulata, apice subacuta, 3.75 mm. longa, circiter 1 mm. lata, superne dorso carinata; omnia sepala trinervia. Peta/a oblique elliptica, apice rotundata, 2.75 mm. longa, 1.5 mm. lata, trinervia, nervis lateralibus+ ramosis. Labellum elliptico-lanceolatum, subacutum, 8 mm. longum, vix 1 mm. latum, multinervosum; calcar valde incurvatum, cylindricum, apice leviter dilatatum, circiter 7 mm. longum et 0.5 mm. diametro. Colwmna brevis, crassa, antice excavata, 1 mm. longa, androclinio levissime excavato; anthera subhemisphaerica, antice vix producta; pollinia globosa, 0.3 mm. diametro, stipitibus duobus linearibus deorsum sensim angustatis 0.5 mm. longis, viscidiis duobus ellipticis; rostellum productum, lobo intermedio convexo crasso, lobis lateralibus brevior- ibus triangularibus; fovea stigmatica quadrata; ovarium papillis nigris sparsis instructum. [ 94 | TANGANYIKA Territory: Neu Langenburg Distr., Bundali Mts., Ngulu Mt., 1400 m. alt., March 1914, Stolz 2554 (Typr.) A somewhat aberrant member of the genus charac- terised by the slender stems emitting a mass of aerial roots, the very small leaves, the elliptical petals and the narrow entire lip. It is probably most closely allied to R. longicalear Summerh. in which the lip is also longer than broad. The column and pollinarium characters agree well with those in other species of Rhipidoglossum. Rhipidoglossum rutilum (Reichenbach _filius) Schlechter in Beih. Bot. Centralbl. 86, Abt. 2 (1918) 81 —Summerhayes in Blumea, Suppl. 1 (1937) 85. Mystacidium Kaessnerianum Kraenzlin in Viertel- Jahrsschr. Nat. Ges. Ziirich, 60 (1915) 395. The type gathering of Kraenzlin’s species (Adssner 741) was cited by me under R. rutilum (Rehb.f.) Schltr. in the reference given above, but I had overlooked the original description. This agrees admirably with that of FR. rutilum and there can be no doubt as to the identity of the two. Oeonia volucris (Thouars) Sprengel Syst. Veg. 8 (1826) 727. Epidendrum volucre Thouars Orch. Hes Austr.-Afr. (1822) t. 81. Until now this combination has been generally credited to Durand and Schinz, Consp. Fl. Afr. 5 (1895) 51, but the above reference has nearly 70 years’ priority. Spren- gel uses the original spelling Aeonza, but as ‘‘Oconia Lindl.’’ has been conserved against two earlier names and general usage also favours this spelling, it would ap- pear best to maintain it. Confusion with the Crassula- ceous genus Aeonium Webb & Benth. is also avoided. DIAPHANANTHE Schlechter An examination of material of practically all the spe- [ 95 ] cies of Diaphananthe, including most of the type speci- mens, has enabled me to reduce some species to synony- my while on the other hand I have added to the number by transferences from other genera and by describing several new species. The details of these changes are given below under the headings of the various species recognised. I am still doubtful as to the exact relation- ships of a few of the species and also have insufficient material of what may prove to be additional new species. The total number of species on my present reckoning is 24, which is one more than the total in Schlechter’s ac- count of the genus (Beih. Bot. Centralbl. 86, Abt. 2 (1918) 95-101). Schlechter divided the genus into three sections as follows :— 1. Eu-Diaphananthe. Stems very short ; pedicels aris- ing singly. 2. Gibbostium. Stems much elongated ; pedicels aris- ing singly. 3. Enantianthe. Stem rather short; leaves fleshy; pedicels opposite or whorled. This arrangement is to a certain extent convenient in classifying the species, but it takes no account of the two quite different types of column structure found in the genus. One group of species containing the lectotype species (D. pellucida (Lindl.) Schltr. of Beih. Bot. Cen- tralbl. 836, Abt. 2 (1918) 97) possesses a narrow acute decurved rostellum and a common viscidium to which the pollinia are attached by separate stipites. The other group possesses a column similar to that in the genus Rhipidoglossum, the rostellum being distinctly trilobed with a fleshy porrect middle-lobe while each pollinium is furnished with a separate stipe and viscidium. The species are enumerated below in accordance with what I think are the affinities of the species, but I have [ 96 | refrained from giving names to the various divisions and subdivisions as I do not feel that our knowledge of the genus and its nearest allies is anything like complete. A. Column rather short, tapering upwards, androclinium rather small and sloping upwards; rostellum beak-like, deflexed, bilobed on removal of pollinarium, lobes subulate; stipites two attached to a common viscidium. a) Stems very short with a rosette of leaves or only slightly elon- gated; flowers alternate. 1. D. pellucida (Lindl.) Schltr. 2, D. bueae (Schltr.) Schltr. 3. D. Plehniana (Schltr.) Schltr. 4. D. Quintasii (Rolfe) Schltr. b) Stems elongated, leafy along a great deal of their length; flow- ers alternate. 5. D. bidens (Sw. ex Pers.) Schltr. 6. D. subclavata (Rolfe) Schltr. 7. D. acuta (Ridl.) Schltr. 8. D. papagayi (Reichb f.) Schltr. 9. D. congolensis (De Wildem.) Summerh. 10. D. divitiflora (Kraenzl.) Schltr. c) Stems long or short; flowers opposite or verticillate. 11. D. fragrantissima (Reich. ) Schltr. 12. D. Welwitschii (Reichb f.) Schltr. 13. D. vandiformis (Kraensl.) Schltr. B. Column short and truncate, androclinium large, occupying the whole of the apex; rostellum porrect, trilobed on removal of pol- linaria, middle lobe fleshy, more or less spathulate, laterals shorter, triangular; stipites 2, each attached to a separate viscidium. a) Stems short or only slightly elongated. 14. D. kamerunensis (Schltr.) Schltr. 15. D. curvata (Rolfe) Summerh. 16. D. polydactyla (Kraenzl.) Summerh. 17. D. Mildbraedii (Kraensl.) Schltr. 18. D. tenerrima (Kraenzl.) Summerh. 19. D. pulchella Summerh. 20. D. ugandensis (Rendle) Summerh. b) Stems much elongated. 21. D. Stolzii Schltr. 22. D. Schimperiana (4. Rich.) Summerh. 23. D. tenuicalear Summerh. 24. D. subsimplex Summerh. The column in group B is almost identical with that in the genera Rhipidoglossum Schltr. and Sarcorhynchus Schltr., but in many other features the various species may resemble closely species in group A. The charac- [ 97 | teristic Diaphananthe tooth at the opening of the spur is present in all but one of the species of group B and is often identical with the similar structure in species be- longing to group A. In D. curvata (Rolfe) Summerh. this callus takes the form of a transverse ridge, but on other grounds the species seems correctly referred to Diaphananthe. Diaphananthe bidens (Swartz ex Persoon) Schlech- ter Die Orchideen (1914) 593. Limodorum bidens Afzelius ex Swartz in Kongl. Ve- tens. Akad. Handl. 21 (1800) 248, nomen tantum— Swartz ex Persoon Syn. Pl. 2 (1807) 521, descript. Angraecum ashantense Lindley in Edwards Bot. Reg. (1843) mise. p. 56. Angraecum monodon Lindley in Lindley & Paxton Flower Garden 2 (1851-52) 102, fig. 187. Angraecum Bakeri Kraenzlin in’ Mitt. Deutsch. Schutzgeb. 2 (1889) 159. Mystacidium productum Wraenzlin in Engler. Bot. Jahrb. 22 (1895) 30. Listrostachys mystacidioides Kraenzlin |.c. 28 (1900) 170. Listrostachys longissima Kraenzlin |.c. 48 (1912) 400. Angraecum subfaleifollum De Wildeman in Bull. Jard. Bot. Brux. 5 (1916) 192. Diaphananthe ashantensis Schlechter in Beih. Bot. Centralbl. 36, Abt. 2 (1918) 98. Diaphananthe monodon Schlechter l.c. Diaphananthe mystacidioides Schlechter l.c. 99. Diaphananthe producta Schlechter |.c. Diaphananthe subfalcifoha Schlechter |.c. 100. Mystacidium Duemmerianum (Kraenzlin in Viertel- jahrsschr. Nat. Ges, Ziirich 74 (1929) 100. An examination and careful analysis of nearly sixty [ 98 | gatherings from an area extending from Sierra Leone to Uganda, leads to the conclusion that the above ‘‘species”’ are merely forms of one widely distributed species. This shows considerable variation in leaf size and shape and in the detailed structure of the lip; there is also a less marked variation in flower size. Unfortunately it has not been possible to examine the types of some of the above names and there seems little likelihood that such examination ever will be possible. These species therefore will have to be interpreted in the light of other gatherings from the same general areas. The descriptions of Angraecum Bakeri, Listrostachys mystacidioides and L. longissima do not include any fea- tures, or combination of characters, which cannot be found in other gatherings both from the east and from the west. The spur appears to be longer in relation to the lip than in most other specimens, but even as regards this character Kraenzlin’s own descriptions are somewhat contradictory. From the description the flowers in Listrostachys sub- JSaleifola are larger than in any other gathering, but some collections from the Belgian Congo (e.g. Corbisier 503) have flowers almost as large and are otherwise evidently D. bidens. Mystacidium Duemmerianum Kraenzl. is a puzzling case. The type gathering is Diimmer 4437 of which there are duplicates at Kew and at the British Museum. These specimens are certainly referable to D. bidens, the only remarkable feature being the consistently short inflores- ecences which otherwise are quite normal. Kraenzlin’s description, although it fits the Kew and British Museum material in most respects, seems to have been drawn up from a plant bearing considerably smaller flowers with shorter and broader tepals. In view of the poor state of preservation of the duplicates it seems probable that [ 99 | Kraenzlin was unable to discover a well-preserved flower on the type sheet except in the bud stage when preser- vation is often better. This would explain the short per- ianth members since it is a well-known fact that there is considerable growth in length as the flower develops. There is, however, a possibility that Diimmer 4437 1s a mixture of two species and that Kraenzlin’s type differs from the material seen. So far as can be judged from the description, Diapha- nanthe subclavata (Rolfe) Schltr. (Angraecum subclava- tum Rolfe in Bolet. Soc. Broter. 9 (1892) 140) seems also to be referable to D. bidens. Unfortunately no specimen could be discovered in the herbaria at the British Mu- seum or at Coimbra. Admittedly Rolfe described the pollinia originally as having only a single stipe, but he stated that the species is allied to Listrostachys ashanten- sis. However, in the Flora of Tropical Africa (vol. 7, p. 160) he transfers the species to Listrostachys which, as understood by him there, possesses two separate stipites in the pollinarium. D. didens has been recorded from Fernando Po and there seems no reason why such a widely spread species should not also occur in Sao ‘Tomé. Arrangement of the specimens in a geographical se- quence from west to east shows that those occurring to the west of the Cameroons Mountain have on the whole shorter and broader leaves than those to the east and south. There is no absolute distinction, however, mod- erately long-leaved forms being found in Sierra Leone and short-leaved ones in Gabon. The flowers in the ‘‘Congo’’ group are on the whole slightly larger and the spur relatively shorter than in gatherings from *‘Upper Guinea,’” but again there is no hard and fast line of de- marcation. With regard to variations in the labellum, which have been used by some authors for differentiating purposes, it is obvious that considerable variations occur. [ 100 | The lip is usually more or less quadrate with rounded base, the apex truncate and somewhat emarginate with a short central apiculus, and the margins normally irreg- ularly dentate especially in the anticous part. Sometimes the lip is narrowed towards the front, sometimes it is somewhat pandurate, while in other cases there is no ob- vious emargination. The spur is much narrowed at the base and obtuse at the apex. It varies from being the same length as the lip to a little more than half as long again. On the whole the western or ‘‘Upper Guinea’’ specimens have a relatively longer spur than those from the Congo basin region, but here again there is no sharp distinction. On the whole the evidence so far available does not appear to be sufficient to justify the recognition of two geographical varieties. Diaphananthe congolensis (De Wildeman) Sum- merhayes comb. nov. Mystacidium congolense De Wildeman Not. Pl. Utiles Congo 1 (1903) 151—Summerhayes in Blumea Suppl. 1 (1937) 80, in obs. Rhipidoglossum rutilum Schlechter in Beih. Bot. Cen- tralbl. 36, Abt. 2 (1918) 81, partly. This species is closely allied to D. divitiflora (Kraenzl.) Schltr. and may eventually prove to be conspecific. At present I[ have not seen sufficient material of either spe- cies to make a definite decision. Diaphananthe fragrantissima (Reichenbach filius) Schlechter in Beih. Bot. Centralbl. 36, Abt. 2 (1918) 100. Lastrostachys fragrantissima Reichenbach filius in Flora 48 (1865) 190. Listrostachys Kirkii Rolfe in Dyer FI. Trop. Afr. 7 (1897) 164. Lustrostachys falcata De Wildeman Not. Pl. Utiles Congo 1 (1908) 147. [101 J Listrostachys fimbriata Rolfe in Bull. Mise. Inform. Kew (1906) 115. Angraecum Muanse Kraenzlin in Orchis 2 (1908) 99. Diaphananthe falcata Schlechter in Beih. Bot. Cen- tralbl. 36, Abt. 2 (1918) 100. Diaphananthe fimbriata Schlechter l.c. Diaphananthe Kirku Schlechter |.c. 101. Diaphananthe Muansae Schlechter l.c. I fail to see any differences between the above species on comparison of the descriptions and 80 gatherings in- cluding type specimens, although there is certainly some variation in flower size and particularly in lip structure. The lip varies in the degree of laceration or fimbriation of the margins and in the length of the apiculus although this is usually at least one quarter of the length of the rest of the lip and sometimes as much as two-thirds. ‘The spur is variously thickened in the distal part, acute or obtuse, and either shorter or longer than the lip. So far I have been unable to correlate the variations in any of the above characters, while in general facies, including habit, inflorescence, shape of tepals, general lip construc- tion and column structure all the specimens exhibit a remarkable uniformity. Judging from the material available and the notes of collectors the stem is upright at first while still short and the leaves are then recurved downwards and may almost meet below. As the stem elongates the root system is not strong enough to hold it up and it gradually falls over, the leaves produced subsequently then becoming more and more hanging and consequently straighter. D. Welwitschu (Reichb.f.) Schltr. which is, so far as I am aware, only known from the type gathering, has elongated upright stems with short closely imbricate al- most erect leaves which are rounded and unequally bi- lobed at the apex. The inflorescences and flowers closely [ 102 ] resemble those of D. fragrantissima. D. vandiformis (Kraenzl.) Schltr., judging from the photograph in Schlechter Die Orchideen, fig. 202, has apparently a habit similar to that in D. Welwitschii, but the leaves are much longer and spreading. Until it is possible to see material of this species from the Cameroons it had best be left as a separate species. As now understood D. fragrantissima occurs in Ou- bangui-Chari, Eastern Belgian Congo, southern Sudan, Uganda, northern Tanganyika Territory, Angola and North-Western Rhodesia. From the data available it appears generally to inhabit gallery forests and dryish woodlands in savanna regions, either epiphytic or on rocks. Diaphananthe pulchella Summerhayes sp. nov. ; a D. Stolzi Schitr. caule multo breviore, foliis longioribus angustioribusque, inflorescentiis longioribus, floribus mi- noribus, tepalis obtusissimis vel rotundatis; a D. ugan- densi (Rendle) Summerh. floribus majoribus, labello subflabellato distinguenda. Herba epiphytica, pulchella; caulis brevis vel + elon- gatus, = dependens, usque ad 13 cm. longus, 3-4 mm. diametro, superne dense 4—9-foliatus, inferne vaginis fol- iorum delapsorum fere omnino circumdatus, radices flex- uosas ramosas, griseas, laeves, 4-6 mm. diametro per totam longitudinem emittens. Mola circiter 7.5 mm. distantia, patentia; vagina arcta, leviter compressa, valde nervosa, 5-10 mm. longa; lamina oblongo-ligulata, lig- ulata vel lineari-ligulata, apice inaequaliter bilobulata, lobulis rotundatis vel obtusissimis longiore 4-11 mm. longo, basi leviter angustata, 4.5-15 cm. longa, 7-11 mm. lata, + coriacea. Inflorescentiae ex axillis foliorum vel foliorum delapsorum exortae, singulae, simpliciter racemosae, dependentes, 6-15 cm. longae, sublaxe mul- [ 108 J tiflorae; pedunculus 1-2 cm. longus, vaginis paucis ova- tis subacutis instructus; rhachis teres, circiter 1 mm. diametro; bracteae 6—9 (-13) mm. distantes, ochreatae, ovato-triangulares, acutae, 1.5-8 mm. longae. Flores patentes, pallide viridi-lutei, cremei-viridescentes vel viridescentes, translucentes; pedicellus cum ovario 8-4 mm. longus. Sepalum intermedium erectum, concavum, ellipticum, oblongo-ellipticum vel obovato-ellipticum, apice obtusissimum vel rotundatum, 4.75—-7.25 mm. lon- gum, 2.5-8.75 mm. latum; sepala lateralia leviter cur- vatim oblonga vel elliptico-oblonga, apice rotundata, 5.75-7.5 mm. longa, 2-3.5 mm. lata; omnia sepala tri- nervia, nervis lateralibus saepius ramosis. Peta/a oblique ovata vel lanceolato-ovata, apice breviter acuminata, marginibus superne irregularibus vel fere denticulatis, 4,25-7.25 mm. longa, 3-4.75 mm. lata, 83—5-nervia, ner- vis lateralibus ramosis. Labellum late flabellato-suborbic- ulare, apice = excisum apiculo saepe interjecto, antice obscurissime tri-vel quadrilobatum, marginibus irregu- lariter denticulatis vel excisis, totum 6—-9.25 mm. lon- gum, 7.25-10 mm. latum, nervis numerosis + radiantibus, ante orificium calearis callo erecto dentiformi instructum ; calear incurvatum, ex ore latiore angustatum, deinde fusiformi-cylindricum, apice angustatum, 8.5-11.5 mm. longum. Colwmna incurvatim porrecta, crassa, subteres, 1.8-2.8 mm. longa, apice truncata, androclinio leviter excavato; anthera subhemisphaerica, antice producta truncata; pollinia subsphaeroidea vel pyriformi-sphae- roidea, 0.65-0.8 mm. diametro, stipitibus linearibus su- perne leviter dilatatis inferne subulatis 1.05-1.4 mm. longis, viscidiis duobus distinctis orbicularibus ; rostellum convexum, antice productum, trilobum, lobo intermedio leviter spathulato carnoso, lobis lateralibus dimidio brevi- oribus incurvatim triangularibus acutis; fovea stigmatica elliptico-quadrata. [ 104 | Kenya Cotony: Mt. Elgon; Kitale, Caves of Elgon Farm, 2010 m. alt., May 1931, Tweedie 10; same locality, 2160 m. alt., on Acacia at edge of forest, May 1941, Tweedie 571; Suam Saw Mill, 2370 m. alt.,edge of forest, high up on Podocarpus gracilior, May 1941, Tweedie 570; no local., 1800 m. alt., in forest, Feb. 1933 Napier, Coryndon Mus. No. 2524; Ngong, 10 miles W. of Nairobi, 1800 m. alt., Nov. 1638, Cunningham-van Someren 512. TANGANYIKA Territory: Masai-land, 50 miles S. of Moshi, Lolbeni Mt., in rich primary forest, April 1943, Page-Jones in Moreau 569; W. Usambara Mts., eastern edge, Mazumbai, 1350 m. alt., forest edge, March 1943, Moreau 611; E. Usambara Mt., Amani, 900 m. alt. in rain-forest canopy, April 1942, Moreau 64 (Tyre); Tanga Prov. Handeni, 45 miles S W of Korogwe, 720 m. alt. in semi-ever- green bush, March 1943, Moreau 610. Diaphananthe pulchella Swmmerh. var. genicu- lata Summerhayes var. nov.; a typo inflorescentiis sae- pius longioribus, floribus paulo majoribus, sepalis later- alibus longioribus angustioribusque ligulatis basin versus lateraliter geniculatis vel abrupte curvatis; pollinii stip- itibus 1.5-1.6 mm. longis differt. Ucanva: Toro: Bwamba, Sempaya, Aug. 1937, Eggeling 2392; no local., 1500 m. alt., in forests, Lankester in Snowden 739. Bunyoro: Bugumolo, in savannah, usually on Acacia Sieberiana rarely on Com- bretum Binderianum, Aug. 1935, Eggeling 2153 (Tyrer); Hoima, 1200 m., Aug. 1940, Purseglove 983; same local., Dec. 1942 (in fruit), flowered at Busingiro, Aug. 1943 & July-Aug. 1944, Eggeling 5360. Busoga: Kitumbazi, Sept. 1921, Lankester 26. Also seen (flowers matched) by W. J. Eggeling at Aduku in Lango Prov. and at Metu, West Nile Prov. The variety may be distinguished from the type of the species especially by the narrow strap-shaped lateral se- pals which are suddenly bent in the posticous direction about one-third way from the base and by the longer stipites to the pollinia. Each stipes is narrowly strap- shaped with a broader rounded apex and tapers gradually towards the base where it is narrowest. In the type the basal part of the stipes is somewhat thickened and the tapering from above downwards is not so marked. In both type and variety the pollinia are of the same size. There is some indication from habitat data that the va- | 105 | riety is a savanna plant, whereas the type appears to oc- cur mainly in forests or at the forest margins. D. pulchella is evidently most closely related to D. ugandensis (Rendle) Summerh. The latter has smaller flowers with a differently shaped lip. This latter organ is more or less orbicular, the basal margins being rounded but entire whereas in the apical part they are irregularly toothed. At the apex is a broad sublunate truncated sinus with a small central apiculus. In D. pulchella the basal part of the lip is distinctly cuneate-flabellate while the front margins are irregularly dentate and more or less obscurely trilobed, the middle lobe being narrowly and sometimes rather deeply incised to form two sub-lobes which give a four-lobed appearance to the whole. The narrow apical sinus or cleft has sometimes a small central apiculus. So far as the lateral sepals and pollinarium are concerned D. ugandensis more closely agrees with the Kenya and Tanganyika type form of D. pulchella than with the Uganda variety genicu/ata with which it occurs. Diaphananthe Schimperiana (4. Richard) Sum- merhayes comb. nov. Dendrobium ? Schimperianum A. Richard Tent. FI. Abyss. 2 (1851) 282. Angraecum Schimperianum Reichenbach filius in Walp. Ann. 8 (1852-8) 573. This species is omitted by Schlechter from his account of the Angraecoid Orchids. From dissections made of a rather withered flower on a duplicate of the type, taken together with Reichenbach’s description, it is evident that the species should be referred to Diaphananthe. Here it is clearly related to D. Stolzu Schltr. and D. pulchella Summerh. From D. Stolz it may be distin- guished by the much longer leaves with a more unequal apex in which the shorter lobe is scarcely evident. In [ 106 ] D. Stolz the leaves are widest towards the apex, the latter being unequally bilobed with both lobes rounded and quite distinct. The long stem and relatively short spikes distinguish the Abyssinian species from D. pul- chella. It differs from both D. Stolzii and D. pulchella in the shape of the lip which is broadest near the base, slightly pandurate and quite distinctly, if shortly, lacer- ate or fimbriate on the anterior angles. The column, al- though much swollen following fertilisation, is clearly of the Rhipidoglossum type, being trilobed with a central fleshy and porrect middle lobe. Diaphananthe subsimplex Swmmerhayes sp. nov. ; a D. Stolz Schltr. foliis angustioribus, floribus multo minoribus, labello suborbiculari-quadrato apice obscuris- sime trilobo; a D. tenuicaleart Summerh. foliis apice nec angustatis sed lobulis rotundatis, floribus minoribus, cal- cari multo breviore differt. Herba epiphytica, glaberrima; caulis elongatus, + de- pendens, pauciramosus, 10-35 cm. longus, 2.5-8 mm. diametro, superne laxe foliatus inferne vaginis foliorum delapsorum fere omnino circumdatus, radices flexuosas simplices usque ad 85 cm. longas et 4 mm. diametro laeves griseas vel brunnescentes per totam longitudinem emittens. Molia erecto- ad recurvato-patentia, 1-2 cm. distantia; vagina arcta, valde nervosa, apice truncata, 7-15 mm. longa, leviter compressa; lamina ligulata (si mavis lineari-oblonga), apice inaequaliter bilobulata, lob- ulis rotundatis longiore usque ad 8 mm. longo, basi + abrupte angustata, 2.5-8.5 cm. longa, 5-12 mm. lata, subearnosa. Inflorescentiae axillares, simpliciter racemo- sae, patentes, 1.5-4 cm. longae, subdense 5—138-florae; bracteae 2-7 mm. distantes, ochreatae, triangulares, sub- acutae, 1-2.5 mm. longae. Flores patentes, cremei ad ochracei, + viridi-tincti; pedicellus cum ovario 2.5-8 { 107 ] mm. longus. Sepalum intermedium oblongo- vel ovato- ellipticum, apice rotundatum vel obtusissimum, 2.7—4 mm. longum, 1.5-2 mm. latum, saepissime trinervium ; sepala lateralia oblique oblonga vel oblongo-elliptica, apice obtusa, 3.3-4.5 mm. longa, 1.3-1.6 mm. lata, 1— 3-nervia. Peta/a oblique triangulari-ovata, acuta, mar- gine antico inferne dilatato, 2.8-3.5 mm. longa, 1.7—2.7 mm. lata, trinervia, nervis lateralibus saepe ramosis. La- bellum suborbiculari-quadratum, apice obtusum vel ob- scurissime bi-vel trilobulatum, basi rotundato-cuneatum, 2.5-3.6 mm. longum, 2.8-3.5 mm. latum, basi ante ori- ficium calcaris callo obtuso dentiformi instructum, disco multinervoso; calear leviter incurvatum, cylindricum, apice leviter angustatum, 4.4-6.75 mm. longum, circiter 1 mm. diametro. Coluwmna deorsum porrecta, crassa, brevis, apice truncata, 1.2—-1.5 mm. alta, androclinio levi- ter excavato; anthera hemisphaerica, antice producta truncata; pollinia fere globosa, circiter 0.5 mm. diame- tro, stipitibus duobus linearibus superne dilatatis apice rotundatis 0.6 mm. longis, viscidiis duobus separatis or- bicularibus; rostellum productum, lobo intermedio con- vexo carnoso apice subspathulato rotundato, lobis later- alibus duplo brevioribus triangularibus subacutis; fovea stigmatica quadrata, margine inferiore = prominente. Ucanpa: Karamoja, Napak, 2250 m. alt., in ravine forest, occa- sional, May 1940, Thomas 3644. Kenya Cotony: Mt. Elgon, Suam River Valley, 1950-2100 m. alt., Nov. 1934, Tweedie 289; July 1936, Tweedie 329; 330; Aberdare Mts., Kinangop, 2640-2670 m. alt., close to river, April 1938, Chand- ler 2400; Western Aberdare Mts., Kinobop Forest Station, 2580 m. alt., June 1931, Dale 2863; Chyulu Hills, Central part, 1650 m. alt., in rain forest, June 1938, Bally, Coryndon Mus. No. 7853; Nanyuki, N W of Mt. Kenia, 2040 m. alt., on bole of large tree by river, June 1943, Moreau 578 (Tyre). This species resembles strongly Rhipidoglossum wan- thopollinium (Reichb.f.) Schltr., not only in the general habit and flower size, but also in the structure of the col- [ 108 ] umn which is of the Rhipidoglossum type. It possesses, however, a distinct tooth in the mouth of the spur and, moreover, the petals are markedly acute and similar in shape to those in other species of Diaphananthe. As I pointed out in my revision of Rhipidoglossum (Blumea Suppl. 1, 78-9: 1937) the two genera are very closely allied, but it still appears advantageous to maintain them as separate entities for the time being. Diaphananthe tenerrima (Kraenzlin) Summer- hayes comb. nov. Listrostachys tenerrima Kraenzlin in Vierteljahrsschr. Nat. Ges. Ziirich 68 (1923) 424. This species was reduced to D. Mildbraedu (Kraenzl.) Schltr. by Schlechter in the Berlin Herbarium but on comparison of the two type gatherings the differences, especially in the floral characters, warrant their separa- tion as distinct species. The sepals and petals in D. ten- errima are broader and rounded at the apex while the lip is transversely elliptical and quite rounded in front in- stead of obovate and apiculate as in D. Mildbraedii. In vegetative characters there is admittedly a close resem- blance between the two species and both came from the volcanic area near Lake Kivu. Diaphananthe tenuicalcar Summerhayes sp. nov. ; affinis D. Stolzii Schltr. et D. Schimperianae (A. Rich.) Summerh., ab hae foliis plus duplo minoribus, ab illa foliis apice angustioribus lobulis acutis, floribus minori- bus sed calcari longiore tenuissimo, ab utraque statura minore, inflorescentiis multo brevioribus floribus densior- ibus distinguitur. Herba verosimiliter epiphytica, glaberrima; caulis elongatus, 7-16 cm. longus, circiter 8 mm. diametro, apice circiter 3-foliatus, inferne vaginis foliorum delap- sorum + circumdatus, radices numerosas reflexo-arcuatas [ 109 ] laeves griseas 8 mm. diametro per totam longitudinem emittens. Folia 6-8 mm. distantia, adscendentia vel + patentia; vagina subarcta, 6-8 mm. longa, nervosa; lam- ina oblique vel curvatim oblongo-lanceolata, basi abrupte angustata, apice breviter et inaequaliter bilobulata, lob- ulo longiore 4-6 mm. longo acuto, lobulo breviore usque ad 1 mm. longo subacuto, 4-5 cm. longa, 1—-1.3 cm. lata, textura verosimiliter tenuis. Inflorescentiae ex axillis foli- orum natae, solitariae, erecto-patentes, 1—-1.7 cm. longae, 3—4-florae, basi vaginis paucis instructae; bracteae 2.5—4 mm. distantes, ochreatae, triangulares, acutae, 1.5—-2 mm. longae. Flores albi; pedicellus cum ovario 0.8-1.2 em. longus. Sepalum intermedium oblongo-ellipticum, apice rotundatum, 4.5 mm. longum, fere 3 mm. latum, concavum, trinervium; sepala lateralia leviter curvata vel obliqua, ligulato-oblonga, obtusa, 5 mm. longa, 2.5 mm. lata, sub-trinervia. Petala oblique et irregulariter ovata, apice acuta, basi margine antico rotundato-dilatata, 4.5-5 mm. longa, 3 mm. lata, trinervia, nervis lateralibus ramosis, saccis crystalliferis numerosis distincte notata. Labellum flabellatim obtriangulare, apice anguste retu- sum apiculo interjecto, 7.5 mm. longum, antice 6.5 mm. latum, multinervium, basi ante ostium calcaris callo hum- ili dentiformi instructum; calear ex ore lato filiformi- cylindricum, apice sensim angustatum, leviter incurva- tum, fere 2.5 cm. longum, circiter 0.75 mm. diametro. Columna crassa, subteres, 1.6 mm. longa; anthera sub- hemisphaerica, antice breviter producta truncata; pol- linia sphaeroidia, 0.6 mm. diametro, stipitibus linearibus superne subspathulatis inferne subulatis 1.5 mm. longis, viscidiis duobus distinctis oblongis postice acutis 0.8 mm. longis; rostellum porrectum, lobo intermedio rostrum anatis revocanti, lobis lateralibus multo brevioribus cur- vatim linearibus apice truncatis ; fovea stigmatifera quad- rata. { 110 } Ucanpa: Karamoja, Napak, 2250 m. alt., in ravine forest, occa- sional orchid with masses of small white flowers, May 1940, Thomas 3645, Another member of the group with elongated stems and Rhipidoglossum type of column bearing separate vis- cidia on each side of a fleshy often more or less clavate obtuse rostellum middle-lobe. The present species is characterised by the short few-flowered inflorescences and the relatively long slender spur. Diaphananthe ugandensis (Rendle) Summerhayes comb. nov. Mystacidium ugandense Rendle in Journ. Linn. Soc. London Bot. 87 (1905) 220. This species is also omitted from Schlechter’s mono- graph of the Angraecoid Orchids, but an examination of the type specimen shows it to be a typical member of Diaphananthe with a well-developed tooth in the mouth of the spur. Vegetatively the plant resembles Rhipido- glossum xanthopollinium (Reichb.f.) Schltr. but may be readily distinguished by the floral characters. Material collected in Uganda by Mr. W. J. Eggeling and agree- ing well in floral structure and other respects with the type, (Bagshawe 425), has larger and more laxly flowered inflorescences. The column is of the Rhipidoglossum type found in other members of Diaphananthe (D. Stolzii Schltr., for example), there being two separate vicidias and stipites and a trifid rostellum with a fleshy central lobe. From D. pulchella Summerh., which is probably its nearest ally, the species differs in the narrower leaves, slightly smaller flowers and differently shaped lip. This is relatively broader, and lunately truncate at the apex with a distinct central apiculus. Sarcorhynchus bilobatus Swmmerhayes sp. nov. ; a S. polyantho (Kraenzl.) Schltr. planta majore, foliis [111 ] duplo latioribus, tepalis latioribus, labello profunde bi- lobato, caleari multo longiore incurvato facile distin- guendus. Planta epiphytica vel rarius terrestris; caulis elonga- tus, pauciramosus, usque ad 50 cm. longus, verosimiliter plus minusve dependens, circiter 4-6 mm. diametro, fere teres vel sectione ellipticus, radices 8-6 cm. distantes ad 15 em. longas et 4mm. diametro laeves griseas per totam longitudinem emittens. Folia disticha, 1.5-8 cm. dis- tantia, fere patentia; vagina arcta, dorso leviter carinata, nervosa, 1.5-2.5 cm. longa, apice truncata; lamina el- liptico-oblonga vel fere elliptica, apice oblique inaequal- iter bilobata, lobulis subconniventibus lobulo breviore vix distincto, basi subrotundata vel rotundata, 5—9.5 em. longa, 1.5-4.25 em. lata, siccitate nigrescens. Inflores- centiae 1—8-natae, folia aequantes vel paulo superantes, 5-9 cm. longae, fere ad basin subdense multiflorae, basi vaginis paucis ochreatis infimis truncatis superioribus lan- ceolatis acutis instructae; rhachis teres, leviter fracti- flexa; bracteae laxe vaginantes, ochreatae, truncatae, 1-2 mm. longae, 8-4 mm. distantes. Flores patentes, pellucidi, albidi vel viridescentes ; pedicellus (cum ovario) 2-3 mm. longus. Sepalum intermedium elliptico-oblon- gum vel oblongo-obovatum, apice rotundatum, 3.75- 4.75 mm. longum, 2-38.25 mm. latum; sepala lateralia oblique vel leviter curvatim ligulato-oblonga, apice ro- tundata, 4.5-5.5 mm. longa, 1.75-2.5 mm. lata, dorso humiliter carinata; omnia sepala trinervia. Petala oblique elliptico-oblonga vel elliptico-ovata, apice breviter acu- minata, basi subrotundata, 3.5-4 mm. longa, 2-8 mm. lata, binervia vel trinervia, nervis lateralibus ramosis. Labellum ambitu late subpandurato-oblongum, trienti- bus duobus apicalibus recurvatis, triente apicali proftunde trilobatum, in toto 4.75—5.75 mm. longum, 8.5—-4.8 mm. latum; lobus intermedius brevissimus, dentiformis; lobi [112 ] laterales subconniventes, oblique lanceolato-oblongi, 1.5-1.7 mm. longi, 1.25-—2 mm. lati, marginibus leviter incisis; calear incurvatim dependens, cylindricum, api- cem versus leviter angustatum, subacutum, 7-8.25 mm. longum, 1.75-2 mm. diametro. Columna horizontalis, semiteres, 1.3-1.75 mm. longa; anthera compresse hem- isphaerica, antice breviter producta; pollinia subsphaer- oidea, stipitibus subspathulato-ligulatis deorsum angus- tatis 0.6—0.8 mm. longis, viscidiis separatis fere orbicu- laribus antice truncatis; rostellum porrectum, lobo in- termedio carnoso lanceolato, lobis lateralibus dimidio breviore leviter incurvatis obtusis; stigma quadrato- ellipticum. Ueanpa: Kigezi, Rutenga, Oct. 1940, Eggeling 4212 (Tyrer); Toro; Ruwenzori, Namwamba Valley, 1875 m. alt., in dense forest by river, Jan. 1935, Taylor 2807 a; Toro, near mouth of Mpanga River, on rock close to river, Sept. 1906, Bagshawe 1205; Ankole, 1 mile from Mbarara on shorter Kabale road, April 1944, Eggeling 5496. An outstanding addition to this small but interesting genus, and occupying an intermediate position as regards geographical distribution, the two other species coming from the Cameroons and Tanganyika Territory respec- tively. 8. bilobatus has a longer stem and broader leaves than the other species, but differs most markedly in the lip being deeply incised in the apical third with two large somewhat connivent lateral lobes and a very much small- er triangular middle lobe; the spur is also much longer. As in the other species the leaves blacken somewhat dur- ing drying while all three have similarly shaped tepals. In column and pollinarium characters the species agree well. Mystacidium tanganyikense Swmmerhayes sp. nov.; a M. venoso Harvey ex Rolfe foliis apice acutiori- bus, petalis latioribus, labelli calcari breviore, rostelli lobis lateralibus crassioribus haud barbatis sed tantum papil- [113 ] losis, pollinii stipitibus brevioribus latioribusque; a M. caffro Bolus foliis majoribus, perianthii segmentis longi- oribus angustioribusque, labello basi amplhiato apice acu- minato, caleari haud apice inflato, pollinii stipitibus lati- oribus viscidiis minoribus satis distinguitur. Planta parva, epiphytica; caulis brevissimus, vix 1 cm. longus, radices numerosas flexuosas tenues emittens. Folia circiter quarto, plus minusve dependentia, ligulata vel lanceolata-ligulata, apice breviter inaequaliter biloba- ta, lobulis subacutis, 83-6 cm. longa, 8-14 mm. lata. In- florescentiae dependentes, foliis duplo vel fere duplo lon- giores, dimidio apicali laxiuscule 5-12-tlorae, 5-10 cm. longae; pedunculus gracilis, vaginis paucis distantibus instructus; rhachis gracilis, interdum levissime fracti- flexa; bracteae vaginis similes, ovatae, acutae vel acumi- natae, 1-2 mm. longae, 3-7 mm. distantes. lores sub- patentes, virides, flavido-virides vel cremei, fragrantes ; pedicellus tenuis, cum ovario 6-9 mm. longus. Sepalum intermedium recurvatum, lanceolatum, leviter acumina- tum, apice ipso obtusum, 5-7.75 mm. longum, 1.5—2 mm. latum; sepala lateralia recurvata, oblique lanceolato- ligulata, apice obtusa, 6-8.5 mm. longa, 1.25—1.75 mm. lata; omnia sepala trinervia. Petala deltoideo-lanceolata, dimidio superiore valde angustata, apice sub-acuta, 4.75— 7.25 mm. longa, prope basin 1.5-2.5 mm. lata, inferne trinervia; omnia tepala saccis crystalliferis minutis nu- merosis instructa, dimidio basali tenuiora et pallidiora. Labellum basi valde ampliatum ita ut videtur obscure trilobatum, lobis lateralibus rotundatis semiorbicularibus, in calear sensim angustatum, dimidio apicali (lobus inter- medius) valde angustatum, lineari-ligulatum, sectione V- forme, apice anguste obtusum; totum labellum 4.7—6.6 mm. longum, 2.75-3 mm. latum; calcar anguste cylin- dricum, leviter incurvatum, 17-21 mm. longum. Coluwmna brevis, crassa, circiter 1.5 mm. alta, androclinio leviter [114 ] excavato. Anthera hemisphaerica, antice leviter pro- ducta; pollinia subsphaerica, stipitibus duobus ligulatis deorsum angustatis circiter 1.5 mm. longis, viscidiis duo- bus lineari-oblongis antice acutis 0.8—0.9 mm. longis. Rostellum valde deorsum productum, viscidiis amotis tri- fidum; lobus intermedius subspathulato-ligulatus; lobi laterales breviores, lineares, acuti; omnes lobi dense papillosi. TANGANYIKA Territory: Lake Daluti, 10 miles east of Arusha, in complete shade on underside of forest liane, June 1942, Moreau 317 (Tyrer); Uluguru Mts., west side of Lukwangule Plateau, 2400 m. alt., epiphytic in dry evergreen forest on Xymalos monospora Baill., Jan. 1943, Moreau 474. A characteristic member of the genus differing from its allies rather in the combination of characters than in any outstanding feature. It is more closely allied to sev- eral South African species than to its Tropical African congeners. Angraecopsis holochila Swmmerhayes sp. nov.; species distinctissima, ab aliis speciebus labello fere inte- gro, caleari plus 2 cm. longo, pollinii viscidio communi, stipitibus longis divaricatis facile distinguenda. Herba parva, epiphytica; caulis brevissimus, fere 0.5 em. longus, paucifoliatus, radices flexuosas griseas glab- ras emittens. Folia saepius duo, oblongo-ligulata vel ob- lanceolato-ligulata, apice rotundata, leviter inaequaliter obscureque bilobata, basi angustata, 38.5—5.5 cm. longa, 3-7 mm. lata. Inflorescentiae dependentes vel deflexae, foliis aequilongae vel breviores, laxiuscule 3-—5-florae ; pedunculus gracilis, teres, 1.5-2.5 cm. longus, vaginis duobus vel tribus, acutis, 3-8.5 mm. longis instructus; bracteae ochreatae, ovatae, acutae, 2-8 mm. longae, 5-10 mm. distantes. Fores patentes, sordide flavi; pedicellus (cum ovario) gracilis, 5-8 mm. longus. Sepalwm interme- dium ellipticum, apice rotundatum vel leviter retusum, 3.25 mm. longum, 1.75 mm. latum; sepala lateralia par- [115 ] ellela, incurvatim oblanceolato-subspathulata, apice ro- tundata, 4.5 mm. longa, superne 1.5-1.75 mm. lata. Petala basi cum sepalis lateralibus breviter connata, ob- lique ovata, subacuta, margine antico inferne dilatata, 3 mm. longa, vix 2 mm. lata; omnia tepala trinervia. La- bellum subquadrato-ovatum, obscurissime trilobatum, 8 mm. longum, 1.75 mm. latum; calcar incurvatim depen- dens, ex ore lato filiformi-cylindricum, apice leviter an- gustatum, 1.75-2.25 cm. longum, 0.3 mm. diametro. Columna incurvata, crassiuscula, 0.7 mm. longa. Anthera subhemisphaerica, antice vix producta; pollinia subsphae- rica, 0.45 mm. diametro, stipitibus duobus divaricatis corniformibus 0.7 mm. longis, viscidio suborbiculari con- vexo antico breviter fisso. Rostellum breviter productum, latum, obtusum, viscidio amoto bifidum. Ucanva: Karamoja, Napak, 2250 m. alt., in ravine forest, occa- sional, May 1940, Thomas 3646. This interesting little species possesses the dwarf habit, small leaves and delicate inflorescence characteristic of many species of Angraecopsis. ‘The somewhat spathulate lateral sepals curve forward parallel to one another on each side of the lip and for a short distance at the base are united to the anticous margin of the petals, both of which features are found in several other species of the genus. The lip, however, is outstanding in being practi- cally entire with only the suspicion of a small rounded lateral lobe on each side about half way between the base and apex. The pollinia have a broad common viscidium, from the sides of which the two stipites diverge in a man- ner recalling the horns on a cow’s head. This structure of the pollinarium is quite distinct from that in 4. par- viflora (Thou.) Schltr. on the one hand and from that of A. breviloba Summerh. and its allies on the other. On the whole, however, the species seems best placed in Angraecopsis in spite of the distinctive characters mentioned. [116 ] iy \ 501 BOTANICAL MUSEUM LEAFLETS HARVARD UNIVERSITY j ~“" i ee, H ~ { CAMBRIDGE, MassaAcHUsETTS, JANUARY 25, 1946 VoL. 12, No. 4 PLANTAE AUSTRO-AMERICANAE III * BY RicHarp Evans SCHULTES DE PLANTIS PRINCIPALITER COLOMBIAE ORIENTALIS OBSERVATIONES A NUMBER OF recently made collections, chiefly from eastern Colombia, represent hitherto undescribed con- cepts or range extensions. It appears advisable to present the following notes on the significance of these collec- tions as a contribution to our rapidly growing under- standing of the flora of the northwesternmost reaches of the Amazonian system. It is with pleasure that I acknowledge the interest of Dr. Lyman B. Smith of the Gray Herbarium in identi- fying the bromeliads which are enumerated and of Dr. Leon Croizat of the Arnold Arboretum for his determi- nations of the species of Mediocactus and Sapium which are discussed. The photographs were taken by the writer. BALANOPHORACEAE Helosis guyannensis L.C. Richard in Mém. Mus. Hist. Nat. Paris 8 (1822) 416, t. 20. This curious, red, fungus-like saprophyte, of rather wide occurrence in the Amazon Valley, has apparently * Plantae Austro-Americanae I and II were published in Caldasia 6 (1943) 11-26 and 9 (1944) 325-336 respectively. [117 ] not been reported hitherto from eastern Colombia. In the trapecio amazonico of Colombia, it is known by the Brazilian name flor da terra, and a strong decoction of the entire plant is highly esteemed in the treatment of dysenteric disturbances and diarrhea. CoLomsiA: Amazonas, Loretoyacu River, Lake Pichuna, small, scarlet root-parasite growing in clumps in wet soil in forests, Novem- ber 1944, R. FE. Schultes 6196, BOMBACACEAE Septotheca Tessmannii Ulbrich in Notizbl. Bot. Gart. Berlin 9 (1924) 129, t. 3. I have noticed individuals of this gigantic tree on some of the sandy islands of the Amazon River in the Colom- bian trapecio amazonico, but no fertile specimens were available. The collection cited below, from the Brazilian island of Aramaca near Tabatinga, only a few kilometers from the Colombo-Brazilian boundary, is a close match for the type which came from a not-too-distant locality on the Rio Ucayali in Loreto, Peru. This species has apparently not been reported hitherto from Colombia or Brazil. Brazit: Amazonas, Ilha Aramaca, near Tabatinga. Enormous tree, 90 ft. tall, with buttress roots, flowers greenish yellow. Zapoterana. October 1944, R. EF. Schultes 6150. BROMELIACEAE Aechmea Mertensii (Meyer) Schultes filius in Roe- mer & Schultes Syst. 7 (1830) 1272. This very widespread Amazonian species is rather abundant in the Vaupés. Cotompra: Vaupés, Alto Rio Vaupés, vecindades de Miraflores, mas o menos 300 m. de altura. Pardsita con flores de color rojo; frutos blancos con el apice rojo, Enero 4, 1944, G. Gutierrez 515. Navia acaulis Martius ex Schultes filtus in Roemer & Schultes Syst. 7 (1880) 1196. [ 118 ] For more than a century, Navia acaulis was known only from the type collection made by Martius in Ara- racuara on the Rio Caqueta. Recently, several additional collections have been made in eastern Colombia which extend the known range of this interesting species. In Caldasia 5 (1942) 8, Dr. L. B. Smith reported the second collection which was made by Cuatrecasas on the gran- itic ridges at San José del Guaviare (Cuatrecasas 7707). This represents a considerable range extension north- ward. Later, Navia acaulis was discovered and collected on several of the sandstone mountains of the Upper Ap- aporis Basin; these collections are cited below. It is an interesting ecological note that Cuatrecasas found Navia acaulis growing on granitic terrain at San José (where sandstone also occurs), whereas at both the type locality at Araracuara and on the Apaporis mountains, the plant is endemic on quarzite ridges. CotomBia: Vaupés, Upper Apaporis Basin, Ajaju River, Mount Campana, sandstone shelf at base of mountain, alt. 900-1500 feet above sea-level, June 1—6, 1943, R. E. Schultes 5570. Caqueta, upper Apaporis Basin, Apaporis River, Mount Castillo (10 km. below Ajaju- Macaya confluence), sandstone, xerophytic conditions, shady crevasses and faults, alt. 350-1000 ft. above forest floor, July 27, 1943, R. E. Schultes 5656. Navia bicolor L. B. Smith sp. nov. Subacaulis, solitaria vel 6-7 aggregatae; foliis multis, dense rosulatis, 9-14 cm. longis, vaginis quam laminis paulo latioribus, omnino occultatis ; laminis lineari-oblon- gis, acutis vel acuminatis, 15-19 mm. latis, supra glabris viridibusque, in sicco nervatis, subtus dense adpresseque albo-lanatis, margine densissime minuteque serrulatis sed prope apicem integris; scapo ad 8 cm. longo, glabro; in- florescentia dense capituliformi sed spicis apices versus distinctis, bracteis exterioribus paulo floccosis exceptis glabra, ad 35 mm. diametro; bracteis florigeris late [119 ] ovatis acuminatis, sepala subaequantibus, integris, pun- gentibus; floribus subsessilibus; sepalis oblongis, acutis, 5.5 mm. longis; petalis albis, laminis lanceolatis, 8 mm. longis, apice valde inflexis. Dr. Smith writes: ‘‘Navia bicolor derives its specific name from the sharp contrast between the green gla- brous upper side of the leaf and the white felt-like cover- ing of the under side. This and the relatively great width of the leaf quickly distinguish it from all other known species of Navia.”’ Cotomspia: Vaupés, Cerro de Chiribiquete, a un lado del Rio Maca- ya. “‘Hierba de flores amarillas y bracteas blancas.’’ Enero 17, 1944, G. Gutierrez 653 (Tyrer in Gray Herb.); upper Apaporis Basin, Maca- ya River, Mount Chiribiquete, sandstone, xerophytie conditions, sa- vanna, alt. 1300-2100 ft. (390-630 m.), May 15-16, 1943, R. E. Schultes 5444 (Corypr in Gray Herb.). Navia graminifolia L. B. Smith sp. nov. Caulibus ramosis, densissime foliatis; foliis persisten- tibus; vaginis parvis, ovatis; laminis linearibus, acumi- natis, ca. 10 cm. longis, infra 8 mm. latis, nervatis, mox omnino glabris, margine subdense minuteque serrulatis ; scapo gracili, ad 15 mm. longo, glabro; inflorescentia dense capituliformi, 15-18 mm. diametro, glabra; brac- teis florigeris late ovato acuminatis, quam sepalis multo brevioribus; floribus subsessilibus; sepalis oblongis, acu- tis, 5 mm. longis; petalis flavis, laminis lanceolatis, 3 mm. longis, apice valde inflexis. Dr. Smith writes: ‘‘Mavia graminifolia appears to be closely related to N. Schultesiana, but its longer and much more persistent leaves make the branches practi- cally indistinguishable and give the plant an entirely different habit.”’ Cotompia: Vaupés, upper Apaporis Basin, Macaya River, Mount Chiribiquete, sandstone, xerophytic conditions, savanna, alt. 1300- 2100 ft. (390-630 m.), May 15-16, 1943, R. E. Schultes 5492 (Tyre in Gray Herb.). [ 120 ] Navia Schultesiana L. B. Smith in Caldasia 12 (1944) 131. Diagrammatic drawings of critical parts of the leaf, flowers and inflorescence were published with the re- cent description of this concept. In view of the appre- ciable contribution which the Apaporis collections have made to our knowledge of this very small genus, it seems advisable to publish the accompanying habit photograph of a complete colony of Navia Schultesiana. Pitcairnia patentiflora L. B. Smith in Contrib. Gray Herb. 127 (1989) 18, t. 1, fig. 4. A xerophytic species of sandstone exposures, Pitcair- na patentifiora is known only from the type locality, Mount Duida in Venezuela, and from Mount Chiribi- quete. Cotompia: Vaupés, Cerro Chiribiquete, a un lado del Rio Macaya, Hierba de hojas acaules y frutos rojos, Enero 17, 1944, G. Gutierrez 672. Vriesia chrysostachys FH. Morren in Belg. Hor- tic. 31 (1881) 87. This species of Vriesta occurs in extraordinary abun- dance in rocky places under light forest in the creeks and brooks which drain the Cerros Pact, Cacuri and Circasia on the Vaupés River above Mitu. It is known from Trinidad to eastern Peru where the type was collected. CotombBia: Vaupés, Lower Vaupés River, Cafio Pacti near Tayast and Circasia, alt. 280 m., savannah, forming dense masses on rocks, March 6, 1944, R. E. Schultes 5826. Vriesia socialis L. B. Smith sp. nov. Verisimiliter acaulis, ad 5 dm. alta; foliis ignotis; scapo recto, 6 mm. diametro, glabro; scapi bracteis erec- tis, quam internodiis bene brevioribus, ellipticis, acutis, membranaceis, obscure punctulato-lepidotis; inflores- centia erecta, simplicissima, laxissime secundeque 6—7- [ 121 ] flora, 10-12 cm. longa, glabra; rhachide flexuosa; bracteis florigeris cum floribus secunde versis, ovatis, acutis, nullo Vriesta socraLis. Inflorescence, one half natural size. Sepal, natural size. modo carinatis, ad 25 mm. longis, quam inter- nodiis paulo longioribus, membranaceis; floribus subpatentibus ; pedicellis robustis, 7 mm. longis; sepalis suboblongis, ad 4 em. longis, bracteas cap- sulasque multo superan- tibus, 9 mm. latis, ecari- natis, tenuibus; petalis ignotis; capsulis crasse ellipsoideis, ad 24 mm. longis. Dr. Smith states: ‘‘In my treatment of the spe- cies of Vriesia with a sim- ple secund-flowered in- florescence in ‘‘ Lilloa’’ VI, 888 (1941), Vriesia socialis would run down to the vicinity of V’.bras- sicoides and V’. Platzman- ni, both of them native of southeastern Brazil over two thousand miles distant. Vriesia socialis has much longer and narrower sepals and both its sepals and floral bracts are much thinner than in either Brazil- ian species. ”’ Cotomp1a: Vaupés, forming dense masses with Vriesia chrysostachys on rocks, savanna along banks of Cafio Pact, affluent of Vaupés near Tayast and Circasia, alt. 280 m., March 6, 1944, I. E. Schultes 5826A (Type in Gray Herb.). [ 122 } BURMANNIACEAE Gymnosiphon cornutus (Benth.) Bentham & Hooker filius Gen. Pl. 8 (1888) 458. The collection cited below establishes the occurrence of Gymnosiphon cornutus in the Colombian Amazonas. This species is much rarer in herbaria and is apparently much more restricted in its distribution than the closely related Gymnosiphon tenellus (Benth.) Urban which oc- curs from Guatemala to Brazil. Cotoms1aA: Amazonas, Loretoyacu River, Lake Pichuna, pale bluish white root-parasite found with Triuris and Leiphaimos, November 1944, R. EB. Schultes 6180 A. CACTACEAE Mediocactus megalanthus (Schum.) Britton & Rose Cactaceae 2 (1920) 212, fig. 292. The occurrence of this beautiful epiphytic cactus within the boundaries of Colombia is established by the collection Schultes 5811. Croizat (Caldasia 9 (1944) 350) had previously included it in an enumeration of cacti which were to be expected in Colombia. The type was collected at T'arapoto, Peru, far from the Vaupés. Mediocactus megalanthus is said to have the largest flowers of any of the Cactaceae. Cotomsia: Vaupés, extensive vine-like epiphyte. Leaves triangular in cross section, ash-grey with very small spines. Joints 3-4 feet long. Flower large and showy. Petals white, sepals light brownish red and yellowish. Pedicel reddish green, fleshy with red-tipped bracts, each with a small spine in the axil. Hanging on gigantic trees. Rio Vaupés, 5 kilometers above Cerro Cireasia, March 4, 1944, R. E. Schultes 5811. EUPHORBIACEAE Hevea viridis Huber var. toxicodendroides R. E. Schultes & EH. L. Vinton in Caldasia 11 (July, 1944) 25. [ 123 J ** Hevea tovicodendroides R. Kk. Schultes*” ex P. H. Allen in Mo. Bot. Gard. Bull. 82 (February, 1944) 50; nomen nudum. It is unfortunate that the nomen ‘‘ Hevea toawicoden- droides’’ was published inadvertently in a letter which Mr. Allen wrote from the field and which appeared in print before the description of the new variety. Sapium Cuatrecasasii Croizat in Journ. Arnold Arbor. 24 (1948) 172. This concept was described from material collected in the Valle de Sibundoy in the Alto Putumayo not remote trom the locality of Schultes 5145 which is the second collection which has been made. The close relationship of the flora of the Sibundoy Mountains and the so-called Villalobos Range southwest of Pitalito in Huila is ob- vious from numerous collections of species which are known only from these two areas. Among these, we may cite Ficus sibundoya Dugand and Saurauia putumayonis R. E. Schultes & H. Garcia-Barriga, both of which were described from material from the Sibundoy area and later turned up in the region of the Villalobos. CotompiA: Huila, area of Rio Villalobos, southwest of Pitalito, 1500-2000 m. alt. *‘ easily coagulate. Bark grey-red, smooth. Tree 80 ft. high, diam. 1.5 ft. Jan. 3-13, 1943, R. EF. Schultes 5145. Cauchillo blaneo.’’ Latex thin, yellow, does not FLACOURTIACEAE Mayna integrifolia (Auwh/m.) R. FE. Schultes comb. nov. Carpotroche integrifolia Kuhlmann in Mem. Inst. Os- waldo Cruz 21 (1928) 396, t. 68. In 1912, Dr. Adolpho Ducke collected the type of Mayna integrifolia, an exceedingly interesting and very distinct species, “‘in silvis non inundatis prope Puerto Cordoba, Rio Caqueta, Colombia, juxta fines Brasiliens. ”’ [ 124 } This region is very poorly known and is rich in endemic or restricted species, as is indicated by the classic collec- tions of Von Martius from La Pedrera. Mayna integrifolia has since been collected by Dr. Ducke on the Upper Amazon River (Solimées) at Sao Paulo Olivenga, near the Colombian border. I have ex- amined three excellent collections (Ducke 387, 405, 24013) which are unquestionably referable to this con- cept. Mayna longifolia Poeppig var. phasmatocarpa R. EE. Schultes var. nov. Arbuscula vel arbor parva, usque ad 6 m. sed saepis- sime circa 3.5 m. alta, in parte superiore parce ramifi- cata foliosaque. Truncus erectus, usque ad 6 cm. (saepius 3-4 cm.) in diametro, nigro vel atrogriseo-fulvo cum cor- tice terete protectus. Rami, ramuli, petiolique striati, fulvo-rubescentes, aliquid scabro-scariosi vel minute squamiferi; ramuli juveniles obscure et primo pulveru- lento-puberuli. Folia alterna, simplicia, ampla, firmissime chartacea, pallide viridia, subconcoloria. Petioli 6-8 em. longi, robusti, obscure puberuli. Stipulae 5 mm. longae, deciduae. Foliorum laminae obovato-lanceolatae, margi- ne integrae, apice breviter acuminatae, basi attenuato- cuneatae, nervis secundariis viginti ad viginti-quinque marginem versus potius arcuatis, pertiis reticulatis; supra glabrae, nitidae, nervis satis prominentibus, pilis minutis et deciduis prope nervos; infra pallidiores, epidermide et nervis satis dense sed minutissime et molliter pilosiuscu- lis, nervo medio robusto, valde elevato; 40-50 em. lon- gae, 11-14 cm. latae in statu adulto. Monoecia. Flores in fasciculis brevibus caulinaribus congesti, inflorescentiis paucifloris. Alabastra globosa, 5-6 mm. in diametro, minute et dense pulverulento-tomentulosa. Bracteae mi- nutae, subsquamiformes. Flores staminiferi albi, 1.5 [ 125 ] (rarenter 1.3)- 1.8 cm. in diametro, brevissime pedicel- lati; pedicelli usque ad 5 mm. longi, pulverulenti. Sepala tria, saepissime aliquid inaequalia; majus rotundato-ova- tum vel late ovato-ellipticum, 6 mm. longum, 5 mm. latum, apice subacutum, extus dense pubescens, intus glabrum, carnoso-subcoriaceum; minora elliptica, apice obtusa, 7-8 mm. longa, 4 mm. lata, extus praeter prope marginem pubescentia, intus glabra. Stamina cirea tri- ginta quinque usque ad quinquaginta, petalis multo bre- viora, libera, erecta, antheris sessilibus vel subsessilibus, lineari-ellipticis, utrinque obtusis, apice poricidis, densis- sime pilosis vel hirsutis, 4 mm. longis. Petala septem, subaequalia, alba, membranacea, extus pulverulenta, in- tus glabra, elongato-elliptica, apice obtusa, margine sub- sinuosa, 6-7 mm. longa, 83-4 mm. lata. Capsulae ovoideae vel aspectu subsphaericae, basi obscure acuminatae vel saepe rotundatae (numquam umbilicatae), apice obscure et breviter acuminatae sed aspectu praeter costas alasque obtuse rotundatae, sepalis persistentibus, 4-5 cm. longae, 3.5-4 em. in diametro maturitate, longe pedunculatae (pedunculi robusti, pulverulenti, usque ad 2 cm. longi), extus omnino albae luridae, subdense sed minute et prop- ter colorem inconspicue puberulentae, saepius novem- costatae, costis longitudinalibus suturalibus et dorsalibus prominentissimis cum alis membranaceis irregulariter et protunde lacerato-fimbriatis, usque ad 8 vel 12 mm, altis. Semina aequalia, obovato-rhombica, plus minusve quin- quaginta, 10 mm. X6 mm. *5 mm., in sicco straminea, extus valde carnosa tamquam pulpa arillata involuta, cum pulpa atrocrocea vel forte miniana, acidula vel de- mum in maturitate dulcia, cum testa tenuiter papyracea crustaceaque, puberulenta et grossiuscule ruguloso-gran- ulosa, raphe elongata fulva in pulpa recepta. Crescit in silvis umbrosis humidisque in terra fertili qua quotannis diluviem profundam sustinet. In provincia Vaupesens} [ 126 ] ab agrestibus ‘‘cacaoito’’ et ‘‘cacao blanco,’’ ab indigenis tribubus Karijonorum ‘‘ha-pe’-ta-ke’’ appelatur. Mayna longifolia var. phasmatocarpa differs from M. longifolia in the length of the petioles, the size and shape of the leaves, their margin (entire, instead of sinuate), the pilosity (pubescent only on the under side, instead of usu- ally on both sides), the size and alation of the capsules, and in several floral characters, especially in the sepals and petals. This new variety, known only from the Up- per Vaupés River where it is abundant, is apparently the northwesternmost representative of the widespread Am- azonian Mayna longifolia. To the best of my knowledge, Mayna longifolia has not been collected in eastern Colom- bia, but the extensive range of this species and its ap- parent abundance in adjacent Peruvian and Brazilian localities (in Loreto and at Tabatinga respectively ) would seem to indicate that it may yet be discovered in Colombia. The varietal epithet phasmatocarpa has reference to the ghastly white color of the fruit. I collected fruits which were ripe almost to the point of opening, and they showed no sign of darkening into a brown or a green as has been reported for some species. Furthermore, natives of the region state that the fruit is always white. The resemblance of the plant, especially the fruit, to Her- rania (a member of the Sterculiaceae) is interestingly borne out by the common names cacaoito and cacao blanco which the settlers in the Vaupes apply to Mayna longi- folia var. phasmatocarpa. We may note also that in Bra- zil, Mayna longifolia is known as cacau branco and in Loreto, Peru, as cacahuito. The brilliant orange-vermillion pulp surrounding the seeds of this plant is usually somewhat acidulous, but, according to natives, it is sweet when the fruit is com- pletely ripe. The Karijona Indians, who call the species [ 137 | ha-pe'-ta-ke, when on hunting trips, often place a few of these seeds in the mouth for the sweet taste. There is apparently no other use made of the plant in the Vaupés. CotomsiA: Vaupés, Vaupés River, near confluence of Unilla and Itilla Rivers, at ‘Las Bocas,’’ alt. c. 200 m., in forest. Treelet 12 ft. tall. Bark grey and brown mottled or black, thin. Basal diameter 14-2 inches. Fruit cauline, pure white, 9-ribbed with two fringes deeply fimbriated on each rib. Inflorescence in fascicles. Stems have strong odor of cyanide. Leaves firmly papyraceous. Seeds with ver- million aril, Flowers cream-white. January 138, 1944, R. EF. Schultes 5728 (Tyre in Econ. Herb. Oakes Ames); Vaupés River, Puerto Naré. Small treelet 18 ft. tall. Basal diameter 3 inches. Bark smooth, brown. Fruits pure snow-white. Seeds numerous, brownish yellow with bril- liant orange aril. ‘“Cacao blanco,’’ cacaoito.’’ Karijona: ‘‘ha-pe/-ta- ke,’’ April 10, 1943, R. EF. Schultes 5378. Ibid. Flowers white. Tree- let 11-15 ft. tall. Basal diameter 3 inches. April 10, 1943, R. EF. Schultes 5366; Vaupés River, near Miraflores, February, 1944, G. Gutierrez V. 797. LORANTHACEAE Phthirusa magdalenae (Cham. & Schlecht.) Eichler ex Martius Fl. Brasil. 5, pt. 2 (1866-68) 55. This parasite is very commonly found attacking Hevea trees in the Upper Apaporis Basin. Cotompia: Vaupés, Upper Apaporis Basin, confluence of Ajaju and Macaya (Puerto Hevea), forest, alt. ca. 900 ft., July 23, 1944, R.E. Schultes 5641. MoracRkaE Castilla Ulei Warb. forma lecithogalacta R. L. Schultes forma nov. A Castilla Ulei foliis amplioribus, subtus densus mol- liter ferrugineo-tomentosis, margine integris sine setis denticulatis, fructibus multo majoribus (diametro paene <2 ad 24), fructuum bracteis apice longissime (4 mm.) acuminatis, cortice fere albo, vulgo tenuiore, latice luride flavo-luteo differt. [ 128 ] In reality, little is known about the genus Castilla in the Amazon for there is relatively little good material in our herbaria. Under Castilla Ulei, moreover, one can find material which would appear to represent several species. Even less is known of the subspecific variations of Cas- tilla Ulei. This is due chiefly to the lack of extensive and systematic field studies devoted primarily to this tree. Furthermore, the general collector is handicapped in making comparative studies and collections of Castilla Ulei because of the size and habit of the tree. It is my belief, after having seen a relatively large number of individuals of this species in the Vaupés and Apaporis basins of eastern Colombia, that there are many unde- scribed subspecific variants in Castilla as is the case in Hevea. Old caucheros (rubber workers) who have cut Castilla all their lives have distinguishing names for the different ‘‘types’’ of trees which produce the so-called ecaucho negro of commerce. These ‘‘types’’ probably would be found to represent taxonomically distinct sub- specific variants, for it is my experience that the cauch- eros of eastern Colombia are rather keen in their percep- tion of differences in plants. These subspecific variants of Castilla Ulei cannot be fully understood, however, until some investigator carries out field studies on thousands of individuals over a rather wide area. The form described here, lecithogalacta, is one of the most common and widely recognized of the ‘‘types’’ of caucho negro in the Upper Vaupés River. The very ap- propriate name yema de huevo (‘‘egg-yolk’’) refers to the bright yellow color of the latex. As the latex, which is extremely thick, flows out of the cuts, it is a deep cream ; on contact with the air, however, it almost immediately deepens in color and very shortly becomes a yellow so brilliant that one finds it difficult to realize that one is [ 129 ] actually dealing with Castilla Ulei. The latex of this spe- cies is usually a pure white or, at the most, a light cream. That the color of the latex, in the concept which I have just described, is not due to soil or other site-factors is borne out by the facts that (1) there are numerous mor- phological characters which likewise differentiate forma lecithogalacta from typical Castilla Ulei; and (2) 1 en- countered this form in several rather widely separated localities, in each locality growing in association with in- dividuals of Castilla Ulei which yielded the usual white latex. The epithet which I have applied to the new form re- fers to the unusual color of the latex and is a rendition into Greek of the common name yema de huevo. Cotoms1a: Vaupés, Rio Vaupés near Puerto Naré, alt. 900 ft. Tree 100 ft. tall, 2 ft. in diameter. Latex brilliantly yellow. Bark light ashy grey or almost white, smooth. Common name: yema de huevo. February 18, 1944, R. FE, Schultes 5798 (Tyrer in Econ. Herb. Oakes Ames). TRIURIDACEAE Triuris hyalina Miers in Proc. Linn. Soc. London 1 (1841) 96. Schultes 6180 represents the first collection of a mem- ber of this rare family from Colombia and, apparently from the entire northwestern sector of South America. Cotoms1a: Amazonas, Rio Loretoyacu, Laguna Pichuna. Very small, white root-parasite. November, 1944, R. E. Schultes 6180. VELLOZIACEAE Vellozia phantasmagoria PF. 1. Schultes sp. nov. Frutex usque ad quattuor pedes altus. Caudex fibroso- lignosus, erectus, basi usque ad 8 cm. in diametro, in- divisus vel saepius bi- vel trifureatus, foliorum vaginis cinereo-stramineis persistentibus arcte adpressis, spiraliter { 130 | imbricatim dispositis obtectus, in parte superiore dis- tincte lineatis. Folia in apice ramorum subrosulate con- ferta, non numerosa, rigide erecta sed exteriora etiam Juniora saepe aliquid subpatentia, plana, valde sicca, cori- acea, dense et minute albo-hirsuta, pilis in marginibus irregulariter plurichotome ramosis, linearia, margine in- tegra, apice Jongissima et sensim acuminata, 250-300 mm. (plerumque 280-290 mm.) longa, 10-11 mm. lata. Flores duo usque ad quinque, pseudoterminales, grandes, speciosissimi, albi, quam foliis breviores, longissime pe- dunculati; pedunculo filiformi, triquetro, superne mi- nute squamoso-echinato, inferne glabro, rubro-fulvo, usque ad 8 em. longo. Perigonii tubus tenuis gracilisque, eylindricus, 50-55 mm. longus, 2.5 mm. in diametro, dense viscoso-glandulosus; limbus infundibuliformis, vivo 60-80 cm. in diametro, segmentis elongate lanceo- lato-obovatis vel paene pseudospathulatis, 50-60 mm. longis, 8-9 mm. latis, apice acutis, extus in parte inferi- ore glandulosis. Stamina octodecim, tepalis multo brevi- ora; antherae elongatissime ellipsoideae, glabrae, albae, quam filimenta multo longiores, antheris lateralibus plus minusve 9 mm. longis, mediana 7 mm. longa, sed omnes subaequales. Ovarium oblongo-clavatum, apice trunca- tum, paleis iis pedunculi similibus sed multo robustiori- bus, viscoso-glandulosis, sordide luteis densissime orna- tum. Stylus filiformis, 6-8 cm. longus, trigonus, inclusus, stamina superans. Stigma trilobatum. Capsula adhue ignota. Crescit in montibus siccis areno-saxosisque in provincia colombiana Vaupesensi. Vellozia phantasmagoria is the first representative of the genus known from northwestern South America. It is apparently not closely related to any previously de- scribed species. The specific epithet has reference to the peculiar habit of this showy and anomalous plant. This species occurs [ 131 ] on some of the flat, dry sandstone expanses on the sum- mit of Mount Chiribiquete in dense stands which, at first sight, seem fanciful and unreal to the observer because of the oddity of the shrub. A description of the unusual habitat of Vellozia phantasmagoria has been published in Caldasia 12 (1944) 124-1380. Vellozia phantasmagoria has its flowering period in December. When the type was collected on January 18, only a few flowers were found, although several thou- sand plants were seen. Cotomsia: Vaupés, Upper Apaporis Basin, Macaya River, Mount Chiribiquete, sandstone, xerophytic conditions, savannah, alt. 400- 1200 ft. above the forest floor or 1300-2100 ft. above sea-level, Jan- uary 18, 1944, R. E. Schultes 5741 (Type in Herb. Gray; Dupticate rypr in Herb. Nac. Colomb.). { 132 | PLATE XITI Henosis GuyaANnensis LC. Rich, A photograph of the plant which furnished the collection Schultes G196, Lore- toyacu River, Amazonas, Colombia. Prarve ALY Navid Breotor LL. B, Smith. A photograph of the plant from which the collection Sehultes 5444 was obtained. Mount Chiri- biquete, Vaupés, Colombia. PLATE XV Navia Scuutresiana L. B. Smith. A photograph of the colony from which the type collection was made. Mount Castillo, Ca- queta, Colombia. PLATE XVI Merptocacrus MEGALANTHUS (Schum. ) Britt. & Rose. A habit photo- graph of the plant from which the collection Sehu/tes 5811 was taken, Vaupés River, Vaupés, Colombia. a3 | “ I PLATE ESTAMPA 08 MEMORIAS DO INSTITUTO OSWALDO CRUZ i TOMO XXI—1922, SSX oO Number of rows of grain 9 12 14 8 14 20 8 Mid-cob width in mm. 12 23 21 15 23 26 22 Kernal width divided by kernel thickness 1.65 2.3 _ O & - io 2) _ © a9) ° Kernel length in mm. 9 11 11 14 12° «15 Number of tassel branches 43 32 14 24 386 84 34 Sterile zone length, mm. 5 4 3 6 8 14 12 Spikelet-node index 13 teh 1 Tet We 28 3 Percent subsessile spikelets 4 6 10 6 4 36 8 The percent of sub-sessile spikelets is another measure of the amount of disturbance to the basic spikelet pair arrangement (Table I). It is taken from the same basal- most secondary branch as the spikelet-node index. Sub- sessile spikelets result from the Tripsacoid tendency to shorten pedicels or indicate irregularities in the order of the spikelets. In a series of spikelets with the pedicellate spikelet on the left, a sub-sessile pair of spikelets is often followed by several pairs with the pedicellate spikelet on the right. The Har The shape of the ear varies greatly from one race to [ 270 } another, but is remarkably constant within the races (Plates XX XV-XXXVII). It has not been studied as well as it should have been because it is dependent upon numerous factors and measurements and comparisons are difficult to make. One of the peculiarities of Tripsacoid maize, which has high numbers of chromosome knobs in Central and North America, is the long naked cob-tip, which is also found in some Guarani maize of Paraguay. Andean maize and Guatemalan highland maize with few chromosome knobs have the tip of the ear completely covered by grains. Ears of Longfellow flint and some ears from the Iroquois tribes of New York also have this grain-covered tip. Anderson (1944) has indicated a correlation between high condensation index and high row number in North American maize, and it is probable that the high row number in Bolivian sweet (Plate XX XVII, G, H) is brought about by the same fasciation which increased the number of tassel branches. Row number is dependent upon many factors and little is known about the manner in which it is regulated. It is, however, relatively con- stant within each race (Table I) and with the exception of the two lowland races, Coroico and Guarani, usually does not vary much within a race. Mangelsdorf has shown that there probably are two types of arrangement for the ear and tassel (1945). His studies were made on tassels and tunicate ears of Guarani maize. More recently a method! of distinguishing be- tween the paired spikelets of an ear has been developed and the ears examined support his contention that Trip- sacoid types tend to have a systematic arrangement while ’ Plate XXXIV shows an alicole of a maize ear after the lower glume has been cut away and the upper glume and the parts above it torn off. Note that one spikelet arose from a hair-fringed callus located in the bottom of the alveolus while the other was attached to the side of the alveolus at a point farther from the center of the cob. [ 271 ] purer maize types have a somewhat whorled or random arrangement, although there occasionally is a suggestion of spiralling. Let us suppose that we slit a corn ear down one side and spread it flat upon this page. Then an ordinary ear with eight rows of grains could be shown as in Figure 3 A. Each pair of grains arises from a pair of spikelets and comprises, with the related parts of the cob, the al- icole. The alicoles are arranged in vertical rows, the ad- joining rows of alicoles shifted slightly (Figure 4 A). The arrangement may be interpreted as being derived from a spiral, from a whorled arrangement in which com- pression and the forcing of the alicoles into the most com- pact order results in the position shift of the vertical rows, or even from the fusion of vertical rows of paired spike- lets. In ordinary maize, then, ears with eight rows of grains have four rows of alicoles, for each alicole bears two grains (diagrams Figure 8, and sketches Figure 4). In the Coroico race, however, the arrangement differs in that, although an ear has eight rows of grains, it also has eight rows of alicoles as shown in Figure 4 B. The alicoles are placed like bricks, each alicole being covered by half of an alicole from the right and half of an alicole from the left so that in any row of grains, one-half of the row will come from one row of alicoles and the other half will come from an adjacent row of alicoles (Figure 4B). The arrangement of the alicoles in Coroico maize and the location of the spikelet pairs upon a fibrous ped- estal which is nearly twice as tall as wide, instead of upon the ligneous lower ridge of the alveolus as in nearly all Ficure 3 (shown on opposite page). Alicole and grain arrangements diagramed as though the ear was split open and spread out. A. Normal arrangement of 8-rowed ear. B. Coroico arrangement of 8-rowed ear. C. Coroico arrangement of 9-rowed ear. [ 272 | gag gogoBase 988 .92a8 Babe Bee Bees Bak gag B8e8 Sad other races of maize, suggests an arrangement which could be derived from a shortened or condensed panicle. This arrangement leads to two curious situations in num- bers of rows of grains. Aye hice RT Pp y, i SANUS Ficure 4. Arrangement of alicoles on the maize ear. The spikelets have been cut away in some of the alicoles and only the bases or pedestals left. A. Normal arrangement of alicoles. a= é ee B. Coroico arrangement of alicoles. Most unusual is the occurrence of ears with an odd number of rows of grains. It is true that ears are occa- sionally found in which part of the ear has an odd num- ber of rows, but in the one case which was examined, an ear of the inbred R4 obtained from Dr. William L. Brown, this results from abortion of one of the paired grains in some of the alicoles, and an odd number of rows of grains is found only on a portion of the ear. On some of the Coroico ears there were 9 rows of grains [ 274 | (Figure 8 C) over the entire ear except for a small por- tion at the base where there was a slightly irregular ar- rangement, based apparently on nine rows. Some of the progeny of these ears had 9 rows of grains and of alicoles. A second novelty is the potential increase in number of rows of grains which is possessed by the Coroico race. If the ear is condensed longitudinally or spread out lat- erally, the grains and the alicoles assume the position of those in ordinary ears with the eight rows of. alicoles arranged to form sixteen rows of grain. There is good evidence that this change occurs, for ears from localities near Coroico usually have row numbers near 16 or less (Plate XX XV, E, F,G), and have a high degree of tes- sellation or interlocking of the grains. Is it possible that ears with row numbers less than 16 are affected by contamination or by selection for larger and softer grains which also reduced row number? The oldest South American prehistoric ears have 14-16 rows of small grains while later prehistoric ears have fewer rows of larger and softer grains or may have more rows and still have small hard seeds. This probably represents the development of maize for special uses from an orig- inal 16-rowed type, one type for boiling, parching or grinding, the other for popping. It is significant that much of the oldest maize in South America bears the paired female spikelets upon a pedestal similar to that in Coroico maize. Prehistoric and pop corn are not included in this paper because they present special problems. Pop corn has primitive types as well as recent developments which owe much of their character to mixture with the races described here. In prehistoric maize there is much of the same problem, with some primitive types which persisted until quite recently and other types which may either represent developments in the region where their remains [ 275 ] are now found or which may in prehistoric times have been introduced from another region. The study of pre- historic maize must be based largely upon the ears, for even where other material is found, it is practically never associated with an ear. For this reason a special set of characters must be utilized. ‘The most useful ones are in the glumes and alveoles of the ear, but before valuable prehistoric material can be studied with maximum pro- fit, a technique must be developed which will not destroy the specimens being studied. Thickness of cob varies greatly and is quite indepen- dent of the number of rows of grains. The cob may be fibrous and flexible or lignified, hard and stiff. The paired female spikelets may be attached by a broad base to the lower margin of the alveolus so that a definite line of sep- aration between glumes and cob is lacking, or the spike- lets may be borne on a pedestal as much as 2 mm. tall. Glumes vary from horny, thickened and sculptured masses to delicate membranes, although the variation within each race is rarely very great. Lemmas and paleas vary much less. Grains range from bird-shot sized spheres through pointed, beaked, polygonal, dented and other forms to the broad Cuzco grain. Most difficult to interpret is the occurrence of denting, and until some information on the cause and control of denting is available, it is impossible to discuss the distribution intelligently. Denting is found occasionally in all the races although it is often so slight as to be barely perceptible. It is strongly present in some Valle and Cuzco maize. Isodiametrical polygonal grains which are widest at their tips are usually correlated with tesselation. This arrangement and shape of the grain is characteristic of Coroico and Guarani maize, which might lead one to sus- pect that they have been contaminated with Tripsacum, [ 276 | since a similar arrangement and shape occurs in ears from maize-teosinte crosses. Ears of Cuzco maize with more than the usual number of rows of grain often have iso- diametrical grains in a tesselated arrangement. If this had been the result of Tripsacum contamination, one would expect to find it in the ears with lower row num- bers instead of becoming more marked in the higher row numbers. Salpor of Guatemala and Cacahuatzintle of Mexico are similar to the many-rowed forms of Cuzco and occasionally have isodiametrical tesselate arranged grains. Kernel width divided by thickness gives a positive number which is an index to the grain cross section (Table I). Variations in grain size are common in some of the races studied and when prehistoric specimens and pop corns are compared with races described here, it may be noticed that, although there are often vast differences in the sizes of grains, there are similarities in the shapes. Colors have been used very little in this study for the examples found in most collections are the extremes. Colors are controlled by numerous factors which may interact, be modified or inhibited, be restricted in action to certain areas of the plant, or have a broad effect over nearly the entire plant. The most significant feature of colors in South American maize is the frequent occur- rence of browns and reds or purples. These colors are common in grasses. Brown grains are not only colored by a combination such as yellow or orange endosperm plus blue or purple pericarp, but by true brown pigments. The common maize of many districts is often divided into two types, a yellow and a white. This is true of the altiplano, where there is a flint with yellow endosperm and a white flour maize; of the valleys, where there are yellow and white forms of Cuzco; and of the Paraguay basin lowlands, where there is a flour maize with a yellow [ 277 ] (‘‘brown’’) aleurone and a white flint which belong to the Guarani race of maize. Highland maize is susceptible to smut just as the va- rieties classified by Mangelsdorf and Cameron (1942) as Andean were susceptible. Perhaps there is some connec- tion between the use of corn smut as food, a common practice in the highlands, and its occurrence. Lowland maize grown at an altitude of 2,500 meters in Cocha- bamba, Bolivia, was rarely attacked by smut. On the other hand, lowland maize grown in Cochabamba was badly damaged by rust. Coroico maize was most severely infested, then Guarani and least of all the Coastal Flints and the highland races. Several commercial yellow dent inbreds from the United States were so badly covered with rust that they died before tasseling. Races or SourH AMERICAN MAIZE Coroico maize Coroico maize is the most unusual race known so far. Some of its characteristics are found in other races, es- pecially in Guarani (Plate XX XV, C, D, EK, F, G) but these are encountered less frequently in areas remote from Coroico, Bolivia, and no ears of this race have been discovered more than a short distance from Coroico. At present the race is restricted to those ears which have the alternate arrangement of alicoles or which approach this condition and have enough of the other characters (slender flexible ear, spikelet pairs on a pedestal, brown- orange aleurone and brownish cob) to distinguish them definitely from Guarani which is to the south and east, and from mixtures with the Coastal Tropical Flints, which are found in parts of eastern Ecuador, and in Brazil as far east as the states of Goias, Maranhao and Ceara. The town of Coroico is in a transition area close to the division between the highland and lowland Indian groups [ 278 | on the margin between the lowlands and the highlands, and not far from the separation of the Amazon and Para- guay River basins. The most peculiar characteristic of Coroico maize (Plate XXXV, A, B) is the arrangement of the ali- coles which has been described on page 272. The long, slender and flexible ears have a light brown cob with very little pith and the shallow alveoles and slender pedestals do not stiffen the ears like the trusswork system of deep alveoles with the horny attached glumes of North Amer- ican and most South American cobs. The isodiametrical grains are nearly always brown-orange in color due to the presence either of the brown aleurone characteristic of Guarani maize or to a brown-orange aleurone hitherto unknown to students of maize. The presence of a dom- inant inhibitor for other aleurone colors makes grains of any color except brown-orange infrequent in this race. The plant of Coroico maize is as distinct from the other races of South America as is the ear. When grown in Cochabamba it tillers abundantly and has narrow leaves with a distinct hairy channel over the midrib. All of the leaf sheaths have hairs which are stiffer and straighter than those found in the pubescent types of Central America. Guarani maize This is the maize grown throughout most of the low- lands and plains of the Paraguay River basin, an area inhabited mainly by the Guarani Indians and related groups. The ears (Plate XX XV, C, D) usually have 12 or 14 rows of grains, are of good size, nearly cylindri- cal, with the naked cob protruding beyond the grains as in Tripsacoid types. The number of chromosome knobs in the plants Mangelsdorf studied was extremely low and some plants had no knobs. Although the cobs are [ 279 ] cream-colored, firm, quite stiff and possess moderate pith cavities and alveoles, some characters, especially in those ears from Santa Cruz Province on the southeastern Bolivian border, suggest Coroico maize. The grains in Santa Cruz Province are smaller, so strongly tesselated at times that the pairs almost overlap, and a dominant aleurone color inhibitor is found in most ears. There are two types of Guarani maize, grown for sep- arate purposes and planted separately: yellow soft flour, ‘alled abati moroti, and crystal-white flint, called abati tupi. The yellow color of the flour maize is formed by a brown aleurone, other colors being absent when the dominant color inhibitor is present. The flour types ap- pear to be the older form. The ears of flint are stiffer, more ligneous, more cylindrical; the alveoli are deeply sunk into the cob, and the grains are rounded at the top, less flattened and less compressed by the husks. The apparent uniformity of Guarani maize must not be interpreted as a lack of potential variation. Concealed within this race, in part by modifiers and inhibitors and by human selection of two definite varieties, is an amaz- ing amount of potential variation. From selected Guarani ears it might be possible to develop many of the varieties of South American highland maize and still have some characters which are not found in other varieties. Coastal Tropical Flint The orange-yellow tropical flint which is the most common maize in Kurope, Cuba and throughout the Caribbean area, is also found in eastern Ecuador, Brazil and Argentina. This variety belongs to the race of trop- ical flints described by Anderson and Cutler (1942) and probably has been spread by the Arawak, Carib and ‘Tupi- Guarani groups which populated coastal areas from Cuba to Argentina. In Brazil, where it is called Cateto, it was [ 280 | previously limited to the coast, but is now spreading rapidly to the interior where Guarani maize was grown (Plate XX XV, H, I). The tight husks prevent much damage to the growing or stored ears and the hard flint grains resist weevils. There has been much active and intelligent work carried out in the selection of improved varieties of Cateto. Much of the seed grown by that name throughout Brazil originated in the State of Sao Paulo and may be contaminated with North American maize. The ears have 12 to 16 rows of flattened, not isodiametrical, grains, are slightly tapered and have an enlarged base, although this may not be evident in some ears. Altiplano maize Altiplano maize is widely distributed throughout the higher or less favorable parts of the Andes from Argen- tina and Chile to Ecuador and probably Colombia. Altiplano maize plants are small and early, with few leaves and tassel branches, occasionally with one or two tillers, and nearly always with red or purple plant color on leaves and culms. The ears are small and nearly spherical, averaging about 14 rows of grains although often these rows cannot be distinguished until the grains are removed and the arrangement of the alicoles studied. The grains vary from hard pop to soft flour, with a wide assortment of endosperm, aleurone and pericarp colors, and grain shapes from nearly spherical through ovoid, pointed, beaked, imbricated and, very rarely, minutely dented. When plants are grown under adverse conditions or near their growth limits, it is difficult to distinguish be- tween the effects of environment and heredity. The small plants and nubbin-like ears of Altiplano maize (Plate XXXVI, A, B) resemble those produced under ad- [ 281 ] verse conditions, yet when seeds are planted in a more favorable environment, as in material from Lake Titicaca planted under irrigation in Cochabamba, there is little or no change. The specimens of maize Anderson described from Rio Loa, Chile (1948), as well as most of the pre- historic material discovered within the present range of the race, falls within the range of variation of Altiplano maize. Like most of the prehistoric material, Altiplano maize probably does not represent a primitive type but may be one of the early developments which could with- stand the environment in which it is grown and may in- dicate the nature of the primitive type from which it was developed. Uchukilla maize The name, Uchuhilla, is applied in Bolivia and Peru to a small crystal-white flint grown on the slopes of valleys at about 2600 meters altitude. The race includes, how- ever, some flour and semi-flint varieties, usually yellow and often red, which are known locally by other names. The plants are small and mature earlier than other races in the same district, usually requiring 50 days from time of planting to the appearance of the silks. The ears (Plate XXXVI, C, D, E) are small, with 8 or 10 straight rows of grains. These grains are widest about two thirds of the distance from the base to the tip, and frequently very slightly beaked, pointed or dented. Ears of Altiplano maize which have low row numbers and a less spherical ear shape than usual approach Uchukilla in appearance. It is nearly always possible, however, to dis- tinguish the few and distinct rows of nearly diamond- shaped flattened grains of Uchukilla from the larger number of somewhat irregular rows of rounded or slight- ly pointed and imbricated grains of Altiplano maize. [ 282 ] Valle maize The favored places for human habitation in Bolivia and Peru are the highland valleys and it is in these val- leys that agriculture is changing most rapidly. No longer are these valleys isolated centers where the people live shut off from their neighbors, and conserve their customs, habits and crop plants inviolate. No longer do feudal landowners and the state church, be they Incaic or mod- ern, hold the workers to the soil, toiling on designated crops and laboring in a limited area. Now restrictions are few, travel is easy and rapid, and an influx of foreign materials and methods has changed the passive Indian to a contused citizen. With the loss of native customs, there has been less rigidity in selection of seed for planting; with the introduction of foreign customs and methods, such as drinking beer instead of chicha, or using animals instead of human labor, new varieties of plants are re- quired. Maize in the valleys is in a volatile condition, with old types being lost and abandoned or modified to suit the newer concepts. An example of this has been given in the growing of two special types of Valle maize, sweet and Culli, or cherry pericarp, maize. The four divisions of Valle maize, hanka sara or the mixture for toasting, Cudli or the cherry pericarp chicha maize, morocho or the ordinary field and general use maize, and chuspillo, the sugar maize, differ greatly in length of growing season, size of plant and number of rows of grains. All, except chuspillo, have long grains which are widest slightly above their center, and usually dented and medium-sized strongly tapering or pyramidal ears with enlarged butts. As in all the races of maize discussed in this paper, with the exception of Guarani and Coastal Tropical Flint, the tip of the ear is rounded and covered with grains. Heterogeneity in Valle maize is increased by the plant- [ 283 ] ing of a special corn for parching. The seed for this plant- ing is obtained from ears which vary in color or grain shapes from the varieties normally planted, and which are floury or nearly so. This seed, known as hanka sara or sechys in Bolivia, thus contains the most varied assort- ment of grain shapes, colors and markings which can be found anywhere (Plate XX XVII, A, B). Dr. Martin Cardenas, rector of the Universidad Autonoma Simon Bolivar at Cochabamba, suggested that a critical study of this group would reveal many varieties no longer in cultivation in a pure form. Most of these varieties, however, would fall into the Valle race of maize and probably within the variation, except for color and some extremes in grain shape, of the other three divisions. Culli or morado recalls the purple dye corn of the Hopi Indians. The plant is the earliest of the Valle maize, smaller, the grains shorter and thicker, and the row num- ber usually lower (10 or 12) than others of this race. It suggests a close relationship with the very similar Uchu- killa. Morocho or huilcaparo is the most important maize of the Valley of Cochabamba and similar forms are the most common types of maize in other valleys of Bolivia, Peru and Ecuador. Although the present form usually has 14 or 16 rows of slightly dented flint grains and a brownish color, there is a tendency to select ears which are more variable in row number, softer and more dented and yellow-orange in color. The cob is nearly always a rust- red color. Chuspillo or chulpi is the sweet corn, eaten only in the form of dried toasted grains or used by the wealthier na- tives in the preparation of a special chicha, but never eaten in the form of green fresh corn. The plant is very simi- lar to that of morocho, but the leaves are a lighter green. The most curious feature of chuspillo is that the number [ 284 | of tassel branches and the number of rows of grain are approximately double those of morocho. The high spike- let-node index of chuspillo suggests that it differs not only because it is sweet, but because there has been some fasciation. There are frequent branched ears and some splayed tips or slight bear-pawing, but the flattened and bear-pawed ears which are associated with high row num- bers and high spikelet-node indices in North American maize are not found in South American highland sweet corn, nor is condensation usually found in the tassel branches, though multiplication is common and charac- teristic (Table I). The gene for sugary endosperm in Valle maize is prob- ably the same as that involved in North American sweet corn, but there is some suggestion that there may be modifiers, for frequently the grains are not completely translucent, but are opaque for part of their length when crossed with North American sweet. It is also possible that there is more than one gene for sugar present. Multiplication of row number has crowded the grains so that they are very slender, almost nail-like and the ear has become more rounded, with less tapering to a point and no visible enlargement of the butt. The grains are usually light yellow or nearly white, though occasion- ally red. In the purer forms the cob is white, but some cobs are red or rust-red, especially where there has been opportunity for crossing with morocho. ‘Two type of ears, one tunicate (called paca sara or hidden maize in Quechua) and the other with the double- grained spikelets such as are found in Country Gentle- man (called cuti sara or turned maize in Quechua), are used as medicine in the Bolivian highlands. When planted these gave many ears which were of Valle type, but others had characters of Coroico or Guarani mingled with those of Valle maize. It is impossible to say where [ 285 | these peculiar and primitive types arose, but it is proba- ble they will be found commonly in Valle maize because of the preservation of the heterogenous hanka sara or mixed toasting corn group. v ; ° Cuzco maize This is the most famous South American maize for the grains are so much larger than any others that they have been collected more often and introduced in many places. It is likely that the extremes of Cuzco represent a com- paratively modern development, for in the valleys near Cochabamba, Bolivia, with conditions as favorable as those in Cuzco, Peru, the two forms (Plate XXXVI, F, G, H, I) of this race are known by Spanish descrip- tive names, mais amarillo and mais blanco instead of the Quechua names by which they are known in Cuzco. The soft texture of the floury endosperm resembles that of salpor of Guatemala and cacahuazintle of Mexico. Like these two varieties, Cuzco maize may also be flinty, a form in which it can be transported or stored in areas where insects and mold would damage floury ears. The ears are moderately long, tapered from a frequent- ly enlarged and irregular butt to a round grain-covered tip. Typical ears of Cuzco maize have eight rows of grains with spikelets so strongly paired that the cob in cross section appears like a cross, with deep sulci between the rows of paired spikelets. Frequently the second and third ears on a plant or ears grown under difficult con- ditions are distichous, with a flattened cob bearing one row of paired spikelets on each edge, somewhat resem- bling, even to the turned up lower glumes, some of the progeny of teosinte-maize crosses. The grains on these ears, as on ears which have not been completely polli- nated, are elongated spheres with a slightly pointed tip (the shape of the grains in teosinte and Tripsacum). Cuz- [ 286 } co grains are usually wide, flattened, slightly dented and with a small beak or overhang on the upper edge similar to that found in much Valle maize, in Mexican pyram- idal dents, and some North American dents. Just as in Guarani maize, planters made two main divisions of Cuzco maize, white and yellow, but unlike Guarani, there usually is no difference in the endosperm and both are floury. The large haciendas of Bolivia are selecting ears with 10 or 12 rows for their plantings and most of these show clearly by their shape and color that they have been crossed with morocho, the most common Valle maize. The yellow ears show most contamination with other races. Both white and yellow Cuzco may have calico pericarp, but most of the ears which exhibit other colors are contaminated by other races of maize. [ 287 ] LITERATURE CITED Anderson, Edgar, 1944. A variety of maize from the Rio Loa. Ann. Mo. Bot. Gard. 30: 469-474. Anderson, Edgar, 1944. Homologies of the ear and tassel in Zea Mays. Ann. Mo. Bot. Gard. 31: 325-342. Anderson, Edgar and Cutler, Hugh C., 1942. Races of Zea Mays: I. Their recognition and classification. Ann. Mo. Bot. Gard. 29: 69-88. Graner, E, A. and Addison, George, 1944. Meiose em Tripsacum australe Cutler e Anderson. Anais da Escola de Agricultura, Pira- cicaba, Brazil. 1944, 213-224. Kozuchoy, I. V., 1935. Results of investigations of world sorts of maize (in Russian). Bull. Appl. Bot. Gen. & Pl. Breed. Ser. A, Loe 0-1. Kuleshov, H. N., 1930. The maize of Mexico, Guatemala, Cuba, Panama and Colombia. Bull. Appl. Bot. Gen. and Pl. Breed., Supplement 47: 117-141. English summary 492-501. Mangelsdorf, P. C., 1945. The origin and nature of the ear of maize. Bot. Mus. Leafl. Harv. Univ. 12: 33-75. Mangelsdorf, P. C. and Cameron, J. W., 1942. Western Guatemala a secondary center of origin of cultivated maize varieties. Bot. Mus. Leafl. Harv. Univ. 10: 217-252. Mangelsdorf, P. C. and Reeves, R. G., 1939. The origin of Indian corn and its relatives. Texas Agric. Exper. Sta. Bull. 574. Weatherwax, Paul, 1935. The phylogeny of Zea mays. Amer. Midl. Nat. 16+ 1-71. White, O. E., 1945. The biology of fasciation. Journ. Hered. 34: il — 22, [ 289 ] EXPLANATION OF THE ILLUSTRATIONS Pirate XXXIV. Three alicoles of a maize ear with the lower glumes cut away and the upper glumes and parts of the spikelets above the upper glumes torn off to show the attachment of the pairs of spike- lets. One spikelet of each pair arose from a hair- fringed callus in the depths of the alveolus while the other spikelet was attached to the side of the alveolus further from the center of the cob. Pirate XXXV. Representative ears of South Amer- ican maize. A. Coroico maize, Coroico, Bolivia. B. Coroico maize, Coroico, Bolivia. C. Guarani maize, yellow flour, Concepcion, Paraguay. D. Guarani maize, white flint, Concepcion, Paraguay. EK. Guarani maize, with characters approaching Coroico, near Robore, Bolivia. F. Guarani maize, with characters approaching Coroico, near Robore, Bolivia. G. Guarani maize, with characters ap- proaching Coroico, near Magdalena de Itenez, Bolivia. H. Coastal Flint maize, State of Maran- hao, Brazil. I. Coastal Flint maize with Guarani influence, State of Ceara, Brazil. Pirate XXXVI. Representative ears of South American maize. A. Altiplano maize from shores of Lake Titicaca, Bolivia. B. Altiplano maize from shores of Lake Titicaca, Bolivia. C. Uchukilla maize, Cochabamba, Bolivia. D. Uchukilla maize, Cochabamba, Bolivia. E. Uchukilla maize, Tarija, Bolivia. F. Cuzco maize, white, Cliza, Bolivia. G. Cuzco maize, white, Cochabamba, Bolivia. H. Cuzco maize, yellow, Cochabamba, Bolivia. {. Cuzco maize, yellow, Sacaba, Bolivia. PrarE XXXVII. Representative ears of South American maize. All the ears shown in this plate are Valle maize. A. Toasting maize, a speckled ear in the mixed group of Valle maize, Sucre, Bo- livia. B. Toasting maize, a white-capped red ear in the mixed group of Valle maize, Cochabamba, Bolivia. C. Culli, Cliza, Bolivia. D. Culli, Cliza, Bolivia. E. Morocho or huilcaparo, the common field corn, Cliza, Bolivia. F. Morocho or huilcaparo, the common field corn, Sacaba, Bolivia. G. Chuspillo, sweet corn, grown among morocho at Cochabamba, Bolivia. H. Chuspillo, sweet corn, Cochabamba, Bolivia. [ 291 J 1g ss a. SG BG yy :F,, Shs Kr } a PLATE XXXV ‘URRELEN EST pa RAAIAII I aaMa®' re) £ u% ; tae 1473 . ddad \e il y i : , . ; ‘ a Sees Veen * > PEP mew Mae. & id tt * ¥ ev ‘ pepe VP ae rl PrAtre AAA YV / ‘ aagtatts ¢ «* (| Heel: ' re i ia “ CREEL) 9 dulldigundaravan XXXVII y 4 PLATE reece rece rett. Wiam Cte re er cs 4! ‘ : J j Pa a ve ¥ ; : A * + wee, ARIAL ERR x ) ‘? 4 Fea Cer eee : rink) ay ew 4 Pie er Lehi: vik he eh metitere ean dtay 44) ¥) 49 ‘i . "e patios pont guretat aH TANASE TAG 4o G97 Q? : s' eeee® RPS | Fer sis VAIS: 19059 rep agddadeaas PI4as F ab) As, aA» Pee a aaery : > VV FINA AS Aly i@é 922 eis ‘sv bake 6A be Pree dcedid ddldl see’ bf ® e™) PP a Me Be be | oN HARVARD \ UNIV ivy » 4 mnt RA Ro BOTANICAL ‘MUSEUM LEAFLETS HARVARD UNIVERSITY OL. 12, No. 9 Campnipoe, “Massacuuserrs, December R 27, 1946 y RUBBER PRODUCTION IN CEARA, BRAZIL BY Hueu C. Cutter’ ALTHOUGH Various commissions of the American gov- ernment have been investigating rubber sources in the Americas for many years, especially after measures were taken by foreign governments to control rubber produc- tion and prices, there still were rubber yielding plants, as late as 1943, which had not been adequately studied. Most of these, to be sure, had been tapped during the rubber boom years, but the rubber buyer often did not know which type of plant yielded a particular grade of rubber, or from what district it had come. In one north- eastern Brazilian state alone, Ceard, nine types of trees were being tapped by the end of 1948; in many areas it was impossible to find a rubber producing tree which was not being utilized. However, little was known about the production methods which were being used or whether or not they could be improved upon. Studies of these rubber yielding plants and the methods of extraction and preparation of their rubber had been made earlier (Ule 1908, Zehntner 1914), but these were out of date by 1948. Investigations carried on during 1943-1945 in Ceara 'The data for this paper was gathered during 1943-45 while the author was employed as Field Technician for the Rubber Develop- ment Corporation, [ 301 ] by field technicians of the Rubber Development Corpo- ration, a United States government agency, determined the identity of the various rubber sources and disclosed that the local methods of production then in use were so well developed that only slight changes and improve- ments were necessary to extract the maximum amount of rubber from the available trees. The various sources of rubber in Ceara and the methods of rubber production will be discussed in detail in this paper. The surveys showed that in Ceara only two genera were important, Hancornia and Manihot, occurring wild with the exception of some small plantings of Manihot in northern Ceara made during the last rubber boom. A few isolated trees of Castilla (eaucho) and two small colonies of Hevea were planted in damp mountains near Baturité, the Hevea seed coming from the territory of Acre. While extremely variable in yield, most of the few Hevea trees are of good quality and are now being tapped by the method used on most Eastern and African plantations (Plate XX XVIII, A) whichis by means of a panel carefully cut with a jebong knife (KMlippert 1942). For several months in 1944 smoked sheets of Hevea rubber comprised the larger part of Ceara rubber ex- ports. These were Far Eastern plantation sheets (Plate XXXVIII, B) produced under Japanese control and were being transported in German ships which were sunk near the coast by American planes and boats. Many fish- ermen earned more money during these months by sal- vaging the floating rubber than they had ever earned by fishing. HaANCORNIA This genus occurs from Paraguay and Bolivia through the states of Mato Groso, Sio Paulo, Goiaz and Minas Gerais to Ceara, Piauf, Maranhio and Para in Brazil [ 302 ] (Schery 1942). Although the genus Hancornia has been split into many species and varieties, the present ten- dency is to reduce the number. There are few herbarium specimens, and for this reason the latest review of the genus (Monachino 1944) can only be considered as a pre- liminary treatment. Most of the material is called man- gabeira locally and may be classified as Hancornia spec- iosa Gomes. In Ceara (map, figure 1) mangabeira is found in small areas along the sandy coast, on the high sandy chapada or wooded savanna of the Serra de Araripe, and in one small isolated colony in the Serra de Ibiapaba near Campo Grande. The trees vary in size from thick- trunked, aged specimens, which resemble old apple trees, HANCORNIA Fee MANIHOT YW FORTALEZA CANINDE ye. Ficure 1. Map of the State of Ceara, Brazil, with inset map of South America showing location of the state. (Drawn by Gordon W. Dillon) { 803 } to slender saplings which frequently die after the first tapping. Seedlings are absent from most areas because the local custom of burning the chapadas, even though prohibited by law, has exterminated most perennial plants which do not have a protective covering of bark or heavy leaf bases. Only the poorest and most primitive of the Indian- White mixtures live in the areas which mangabeira pre- ters. These people are strongly resistant to innovations and, even where it is possible to demonstrate positive superiority of a new method, there must be an inertia- overcoming incentive before the change will even be considered. While this conservatism is characteristic of most people dependent upon the soil, and equally true of American farmers as well as Brazilian agriculturists and woods-people, this conservatism is especially evident in the Cearé coastal mangabeira areas, because most of the tapping is done by women who are even more con- servative in their outlook than men. The trees are usu- ally tapped three times each year with deep V-notches cut with a small paring knife. Because the latex flow is of short duration and cups are seldom on the tree for more than half an hour, mangabeira may be tapped throughout the year, in both the rainy and the dry sea- son. For cups, tough and leathery araticum leaves (An- nona coriacea Mart.) are folded into cones (Plate XL, A), pinned with splinters or cactus spines and slipped under a flap cut in the soft bark. If the tree yields well, the shallow cup overflows; if the wind blows, it over- turns. Even though the additional latex gathered by using tin cups would pay for all the necessary cups in less than two days, only a small portion of the workers have been induced to adopt them. Yet, even leaf cups are an advance over the method which, though rare in Ceara, is common in Maranhao, a state to the north- [ 804 ] west. In this method, the bark is smoothed with a ma- chete to within 4 mm. of the wood on about a third of the circumference of the trunk to a height of 70 cm. Many pricks and small cuts are made in this panel with the tip of the machete, and the latex which exudes is scraped off with a spoon and collected in a gourd. Many of the trees tapped in this manner are left with huge gaping wounds and cannot be retapped even if they survive. The best methods of tapping mangabeira are those used in the Serra de Araripe. The workers organized to tap on the high chapada use a special knife with one end bent into a U for cutting the latex groove, the other end sharpened to a point to make an incision cutting all the latex vessels underlying this groove down to the wood. Below this V-cut a tin cup is pushed into the bark to receive the pink latex. Cuts are made at 40 cm. intervals from the ground as high up on the tree as can be reached (Plate XX XVIII, C). Even branches as small as a man’s arm may be tapped and some trees support as many as twenty-five cups. When a basal cut is made before dawn, the latex may overflow a 120 ce. cup, yet the same cut made at 10 a.m. would yield only 5 cc. A few workers in the Serra de Araripe use the same style of two-ended knife to cut a deep spiral groove on the trunk, from the highest point they can reach to the ground. Here a leaf is inserted in the cut to funnel the latex into a bottle. While there have been reports of trees which will fill a 750 ce. bottle by this method, these trees must be extremely rare, for even with cups the most skilled worker seldom secures more than 400 cc. of latex even from the rare large virgin trees. Claims of five to eight liter yields must be received with skepticism. Most trees can be tapped three times in good years, but only once if rains are scarce. Mangabeira latex usually coagulates very slowly in [ 305 ] comparison with the latex of Hevea and manisoba (Man- thot) which coagulates rapidly when it becomes slightly acid. 'wo and three months old samples of mangabeira latex may remain liquid even when they are slightly acid. The strength and vulcanization properties of man- gabeira rubber are affected by the methods of coagulating the latex and handling the coagulum. It was a difficult task, second only to that of inducing the natives to tap the trees, to introduce acceptable methods for rubber preparation. The natives have long used the dried strings of rubber found in crude cuts on the tree for making small hard balls and the latex for making waterproof sheets or shoes. In one good native recipe for coating cotton cloth a tablespoon of powdered sulfur and the white of an egg are mixed with a large cup of water, added to a liter of pure latex and spread thinly over the fabric. When dry this is placed in the sun for a day to cure. The first mangabeira rubber produced tor export was made by heating the pure latex in a bow] over a fire, adding a variable amount of salt or alum, if either were available, and removing the mass when it had coagulated. A worker can do this so that all of the original water in the latex remains entrapped in the thick soft coagulum, which then has the same 20 to 83 per cent rubber con- tent of the original latex. The most practical method of preparing the rubber in Ceara is to add an equal amount of warm water to the latex and then a smal] amount of a solution of table salt. Coagulation takes place within half an hour. Locally made clay or wood basins are more satisfactory for coagulation trays than the usual kerosene cans, for the tannic acid in the latex does not then en- counter rust and consequently the blue-black ink caused by the rust combining with the tannin does not stain the rubber. [ 306 | The coagulum, about 3 cm. thick, is placed on a board and rolled by hand to a sheet about 6 mm. thick, then washed in a basin with water and hung in the shade to dry. Home made rolling machines which resemble wringers with wooden rollers are used in many places to produce sheets of uniform thickness. 'To corrugate or to smoke the sheets is inadvisable in Ceara because prepared mangabeira rubber deteriorates rapidly, especially when nearly dry and exposed to the air. There are better methods for the preparation of man- gabeira rubber, yet none is as satisfactory for field use. By the use of exact quantities of expensive chemicals and careful heating it is possible to produce a mangabeira rubber which approaches Hevea in quality. However, the native tappers lack the skill and the careful habits necessary to accomplish this, and it was even difficult to introduce the simple scheme outlined above. Maninor Manihot is a large genus with its center of distribution from Ceara to central Baia, Brazil. It is best known to Americans in the form of tapioca which is obtained from Manihot esculenta Crantz, a species native to Brazil, and like Hevea, extensively grown in the East Indies. As in the genus Hancornia many species have been de- scribed, but the paucity of specimens in herbaria, the scanty material upon which many species were based, and the difficulties encountered in the field in ascribing many of the variable trees to a definite species suggest that a monograph of the genus based upon field study in this part of Brazil is needed. Manihot Glaziovii Muell.-Arg., the large rubber yield- ing species which has been tried on plantations in Africa and the East but replaced by Hevea, is native to the Serras of Baturité and Maranguape, Ceara. Most of the [ 307 ] Ficure 2. Manisoba (Manihot) tree types and tapping methods in Ceara. A. Piaui manisoba, tapped on the root with latex gathered in a dust- lined hole in the ground. B. Serra de Uruburetama manisoba with hand-sized pieces of bark removed down to the wood. The latex dries and coagulates upon the bark and exposed wood surface in strings and globules called choro. C. Low-yielding manisoba typical of region west of the Serra de Maranguape and of that about Sobral, Pereiro and Assaré. The trunk is hacked irregularly and the latex strings dry and coagu- late upon the bark. D. Brava manisoba which yields practically no latex, but has been hacked experimentally to see if it will yield. E. Manisoba of the Serra do Machado, west of Canindé. Below, it has been tapped on the saliences of the trunk and the latex col- lected in dust-lined holes in the ground ; above, it has been tapped experimentally with an Amazonas knife, a method which proved unsatisfactory for general use. F. Manisoba of the Serras de Baturité and Maranguape. Above, it is tapped with the daily paired upward cuts and the latex flows into the tin cups; below, a panel cut with the jebong knife, a method which is excellent only if the tappers are careful and conscientious. [ 308 ] trees are tapped by two upward cuts with a short ma- chete or a narrow-bladed hatchet (Figure 2 F). A tin cup is slipped under a slit in the bark directly below the vertex of the cuts (Plate XLI, A). The tapper begins each Monday inserting one to eight cups about the cir- cumference of each tree. On ‘Tuesday he taps about six inches below the previous cuts and affixes the same cup with its now coagulated rubber into a slit below the sec- ond tapping. This is continued until Saturday when the lumps of rubber coagulated during the week are collected and spread out to dry (Plate X LI, B). These lumps are often cut into smaller pieces to speed drying or are sold at a low price (while still fresh) to a merchant who dries the rubber to secure a higher price. Even though the latex flow increases when the rains begin, tapping is dis- continued, partly because cups frequently fill with rain water before the latex coagulates, but mainly because farming is done at this time. In the damp mountains of Baturité and Maranguape where the latex flow is greater, Manihot Glaziovit may be tapped by cutting with a jebong knife in a manner (Figure 2 F) similar to that used for Hevea. A sharp knife must be used and care taken to avoid injuring the cambium. This method was introduced on two small properties and in both cases the yield was greater per man-day, much greater per unit of trees, and, when some care was used in tapping, the trees were scarcely dam- aged. ‘The method was abandoned after a few weeks, however, for the workers did not sharpen their knives and would not use care to avoid cutting the cambium, nor would they utilize the bark to greatest advantage. The narrow-bladed hatchet which is used to prune coffee trees in the mountains cuts a very small gash deep into the wood and severs only a few latex vessels in the bark. The short machete used on the mountain slopes and [ 309 ] plains barely cuts to the wood, but opens more latex vessels. By substituting the machete for the hatchet or inducing the producers to use wider-bladed hatchets, it was possible to decrease the damage to the trees, preserve a bark surface satisfactory for retapping, and at the same time increase the yield of rubber with no extra labor. A modified Amazonas knife to cut a deep groove was un- satisfactory because cuts which were too close together left the bark between them dry and dying, a ready shel- ter for termites. In the same area, the Serras of Maranguape and Ba- turité, a type of manisoba tree occurs which yields practically no latex. This is called manisoba brava, wild manisoba (Figure 2 D). There is a complete series of intergrades between the best yielding manisoba and the driest of these wild trees. It is difficult to say whether there are two intergrading species or one variable species. Frequently one can distinguish the extremes by their bark, which is gray and thickened in the rubber-yielder, and reddish or silvery (usually smooth and thin) in the wild type. The natives believe that by hacking them regularly, it is possible to stimulate or ‘‘tame’” (@mansar) some of the wild trees so they will produce latex (Plate XLI, C). If the latex does not flow after three or four weeks of this treatment, the trees are abandoned. The basis for this belief in ‘‘taming”’ is a phenomenon known as wound-response. In response to the stimulation of tapping, manisoba and Hevea both increase their yield slowly for about three weeks when they reach a maxi- mum at which they remain unless overtapped or injured. West of the Serra de Maranguape the trees resemble the brava type, although in damp places they still yield enough latex for cup-tapping. Trees of the drier places are cut and hacked with the machete so the latex flows over the trunk (Figure 2 C) and coagulates and dries in [ 310 ] strings or globules which are collected two weeks later when the tree is hacked again. This type of rubber is called choro, or tears, and resembles the lacy strips left in cuts made by other methods and sold under the name of sernamby. Still farther to the west, about the Serra de Urubure- tama, excellent choro is produced by slicing the bark off the trunks in hand-sized patches (Plate X LI, D and Fig- ure 2 B). After the cuts are made, the latex is allowed to drip over the trunk and dry for twelve to thirty days before it is collected and a new series of cuts made. Most of these trees are small, spindling, gnarled and scarred, like the legs of undernourished children. There are some larger trees which yield more latex and can be tapped profitably with cups, but practically all of the rubber produced in this area is choro. Several unsuccessful ex- periments were conducted to find practical methods which would yield as much or more latex than the local choro method and still not injure the tree. However, the large exposed surfaces of clean wood left by slicing off the bark are less harmful to the trees than the termite infested areas of dried and dead bark about the cuts made by other methods, such as those used in the Serra de Maranguape. In the Serra de Uruburetama the horny outer layer of the manisoba bark made it difficult to use the jebong or Amazonas knife, and the low yields, scat- tered stands, and rough terrain made it impractical to tap daily. Even farther west, in hills near Sobral, only poor- yielding trees are found and none of the spindling but good-yielding Serra de Uruburetama trees are seen. Choro is produced by both of the methods described: by hacking cuts over the surface of the trunk, or by chip- ping away chunks of the wood. About Assaré in southern Ceara and Pereiro in east- [311 ] ern Ceara, there are many manisoba trees which, like those near Sobral, greatly resemble brava trees (Figure 2 C) and like them yield a resinous, though much more abundant latex. The Assaré and Pereiro trees are hacked severely every two to four weeks. After the dried rubber is collected any part of the trunk which can be reached may be hacked again. Most of the rubber is a dark and sticky choro, but some of the cuts yield so much latex that globules an inch or more in diameter are frequently found. These are pressed out to form small plates about 5 mm. thick and 80 mm. long and are sold at a higher price than choro. These plates resemble the rubber pro- duced in central Ceara, west of Canindé. West of Canindé rises a low mountain range, the Serra do Machado, which is poorly represented on maps. In this area there are many trees which yield a small amount of latex. Some of the trees could be classified as brava, while still others have a latex flow sufficiently large to make the use of cups possible. Nearly all of the trees are tapped from a point about a meter high to the ground (Plate XL, B and Figure 2 E) by slicing away the bark on the four to eight saliences of the trunk. A small hole is dug in the ground and lined with a thin layer of dry dust (called tabatinga, taud, or simply pd), which is care- fully prepared by pounding up calcareous clay and siev- ing it. A sack of this dust is as much a part of the equip- ment of a good rubber gatherer (borracheiro) as the smal] tapping hatchet (Plate XX XIX, B). The latex flows over the surface of the powder which lines the holes and, like raindrops on a dusty road, does not soak in but coagulates into a small plate called chapa. 'These trees are tapped at infrequent and irregular intervals, the suc- cessive tappings merely removing a thin layer of the bark on both sides of the original vertical cut, a procedure which, although it cuts away the cambium, has much in { 812 | common with the jebong cut panel. The rubber is gath- ered after a few days and the tree may be tapped again, although a tree is seldom tapped more than six times in one season. Well prepared chapa is a good rubber, because what- ever little dirt normally adheres to it can be easily re- moved before selling. There are several tricks employed by the natives to add weight to the rubber. The most common of these is to turn over the coagulated latex before making a new cut and then to allow the second flow to cover the dirty surface. When this chapa is re- moved it presents two smooth and clean surfaces. A wise buyer soon learns to be suspicious of rubber which is too smooth and clean on both surfaces, so the tappers occa- sionally leave the first chapa in the hole, kick in a small amount of dirt and allow the new latex to flow in and cover the dirt with rubber. This is more difficult to de- tect unless the piece of rubber is suspiciously large, since one surface issmooth and the other slightly rough. Care- ful buyers have all the large pieces cut up because this practice not only allows them to calculate the discount which must be made for the dirt, but enables them to dry the rubber faster, sell it at a better price as a grade with higher rubber content, and reduce the freight and taxes which are based on weight. Many property owners in the Serra do Machado are planting manisoba seeds with other crops. Snr. José Velosa of Sitio Valtiburi has planted 900 kilos, about 810,000 seeds. He plants beans, corn, cotton, mandioca, castor beans and manisoba at the same time in his field, and the harvest is made in this order. At the end of two years only a few plants of castor bean and cotton remain and the mangabeira is ready to be tapped. Most of the rubber produced in Piaui, the state west of Ceara, is chapa. It is collected in a hole dug in the [ 313 ] ground and lined with dust, just as in the Serra do Ma- chado. The tree and the tapping method differ. The tree is scarcely more than a large bush, branched and spreading from a short trunk (Figure 2 A and Plate XL, C). It grows very rapidly and some one year old trees which were planted on fairly good soil in northwestern Pernambuco have been tapped. In Ceara the only native Piaui type trees are found along the southwestern margin of the state, and there are scarcely any plantings. ‘To tap the tree the roots are exposed, a hole is dug to receive the latex, lined with tabatinga (Plate XL, D), and a groove is cut into the root down to the wood. Every three or four days the coagulated rubber chapa is collected and a new cut made below the old one until it is difficult to reach into the hole with the tool. At this time the tree is abandoned for the year and, since the hole is rarely ever refilled, the uncut bark frequently dries and the tree dies. Well-prepared clean manisoba and mangabeira rubber can be used instead of Hevea in many small rubber in- dustries. To improve low grades like choro, which make up the bulk of the Ceara production, washing plants have been established in northeastern Brazil. In wash- ing, crude rubber is passed between two large rollers. These revolve at differing speeds and separate the rubber so that the streams of water which flow over the rollers can remove any dirt. If the washing is done well, the rubber leaves the rollers as a clean sheet of crépe which needs only to be dried before it can be utilized by the manufacturers. Washing low-grade rubber is difficult and expensive. For this reason, and because of the great diversity and variability of the Ceara rubber and the un- predictable fluctuations in production, it is unlikely that exportation will increase. Ceara rubbers cannot compete in either quality or cost with the Hevea produced on [ B14 ] plantations during the last decade. Industry will con- tinue to prefer the uniform reliability of cleaner and more abundant Hevea. LITERATURE CITED Klippert, W. E., 1942. Small-farm rubber production. Agriculture in the Americas 2: 48-53. Monachino, Joseph, 1945. Revision of Hancornia. Lilloa 11: 19-48. Schery, R. W., 1944. Some secondary rubbers in Cearaé, Brazil. Mo. Bot. Gard. Bull. 32: 41-49. Ule, Ernesto, 1908. Kautschukgewinnung und Kautschukhandel in Bahia. Notizblatt d. Kénigl. bot. Gart. u. Mus. Berlin-Dahlem n. 41a, Bd. 5. Zehntner, Leo, 1914. Estudo sobre as manicobas do Estado da Bahia, em relacao ao problema das secas. Inspectoria Federal de Obras Contra as Secas (Brazil), Pub. 41, Ser. I, A. 115 pp. [ 815 ] EXPLANATION OF THE ILLUSTRATIONS Prater XXXVIII. A. Tapping Hevea with the je- bong knife on Sitio Irapuru near Baturité, Ceara. This panel has been tapped for an entire season. B. Smoked sheets of oriental Hevea rubber recov- ered from the sea. C. Tapping Hancornia near Axixa, Maranhao. Even the branches are tapped and some trees support as many as twenty-five cups. PLrate XXXIX. A. Preparing rubber from latex of Hancornia near Axixa, Maranhao. B. Serra do Ma- chado workers filling their bags with the fine dust used to line the holes in which the latex is col- lected. Pirate XL. A. Native women making a cup from an araticum (? Annona coriacea Mart.) leaf for gath- ering the latex, near Cascavel, Ceara. B. Serra do Machado manisoba tree tapped with a small hatchet on the saliences of the trunk. The latex flows into a dust-lined hole. C. One year old planting of Piaui manisoba at Belmonte, Pernambuco, ready for tapping. D. Lining a hole with dust to receive the latex flow, at Belmonte, Pernambuco. Pirate XLI. A. Inserting a tin cup ina slit below the pair of tapping cuts in a manisoba near Maran- guape, Ceara. B. Drying coagulated cup-lumps of manisoba rubber at Maranguape, Ceara. C. Brava, or wild manisoba near Maranguape, Ceara, which has been hacked to see if it will yield latex. D. Manisoba trees in the Serra de Uruburetama near Iracuba, Ceara. [ 316 | XXX VITI PLATE PLATE. So LS Prate Nh XLI PLATE na pa | Cay S BOTANIC LEAFLETS HAR UNIVERSITY ey Cammripce, Massacnuserrs, JuLy 18, 1947 VoL. 12, No. 10 A CONSPECTUS OF THE GENUS CUNURIA BY J.T. Batpwin, Jr.’ anp RicHarp Evans ScHULTES’ THE writers of the present article became interested in the euphorbiaceous genus Cunuria as the result of in- dependent field studies of Hevea, the genus of the Para Rubber Tree. These genera are related, and, indeed, it has been suggested (Baldwin in Journ. Heredity 38 (1947) 54; in Am. Journ. Bot., in press, ‘‘Hevea rigid- ifolia’’) that they possibly have hybridized. A critical classification of the several concepts of Cunuria and an understanding of their geographic distribution may con- tribute greatly to an interpretation of the complex of genera to which Cunwria shows affinities. I Cunuria belongs to the tribe Jatropheae. The genus appears to be related to Micrandra, Hevea, Joannesia, and Nealchornia, and has especially close affinities with Micrandra and Hevea. Bentham (in Journ. Linn. Soc., Bot. 17 (1880) 185-268) observed that the first four of 1Professor, William and Mary College; Collaborator, Rubber Plant Investigation, Bureau of Plant Industry, Soils and Agricultural En- gineering, U.S. Department of Agriculture. *Botanist, Rubber Plant Investigations, Bureau of Plant Industry, Soils and Agricultural Engineering, U.S. Department of Agriculture ; Research Fellow, Botanical Museum, Harvard University. [ 325 ] these genera ‘‘may to acertain degree be related to each other.’ Huber later described the monotypic fifth genus. Croizat, however, in a manuscript report on the Hu- phorbiaceae of the Tafelberg expedition, has stated of Cunuria: ‘*This genus would seem to be improperly placed in the Huphorbiaceae Gelonieae. 1 should be in- clined to think that Garcia Vohl, Sagotia Baill., Ostodes BL.,and Cunuria Baill., are not too far distantly related.’ From Micrandra, Cunuria may be distinguished by its glandular stipules; by differences in the form of the petiole glands, as well as of the disks and staminodes ; and by the number and form of the stamens. Cunuria is at once set apart from Hevea by having simple rather than compound leaves; by having stipules; and by lack- ing a staminate disk. It likewise differs from Joannesia by having entire leaves as well as by being apetalous. Joannesia appears to be diploid, whereas Cunuria, Mi- erandra, and Hevea are, in great measure, tetraploid (Baldwin in Journ. Heredity, oc. cit.). From Nealchor- nia, Cunuria can very readily be separated by its having a pistillate disk; by the presence of a rudimentary ovary in the staminate flower; and by differences in the form of staminodes and styles. Cunuria, at first considered to be completely dioecious, was recognized by Ducke (in Notizbl. Bot. Gart. Berlin 11 (1982) 586) as having monoecious representatives. As now known and interpreted, Cunuria comprises four species and one variety. Two species are established in the present paper, and one concept, recently described as a species, is here reduced to varietal rank. II The distribution of Cunuria is indicated on the ac- companying map (Fig. 1). Strictly speaking, Cunuria is not a characteristic element of the flora of the great [ 826 ] e Cunuria Spruceana a C.Spruceana var. bracteosa aC. australis xC. crassipes ¢C. glabra Fic. 1. Map showing the distribution of the species of Cunuria. oe : i ae ee “ss C”™CUC~«< Amazonian planada. The genus is typical of higher ele- vations. It could be referred to as an element of the ‘‘upland”’ flora of the Amazonian region, but that ex- pression might connote the eastern slopes of the Andes, where the genus apparently is not represented. Since Cunuria prefers the higher areas in the Amazonian re- gion, it may be interpreted as a very old genus that has spread from the ancient Venezuela-Guiana land-mass. Both Schultes (in Caldasia 8, no. 12 (1944) 124-1380; in Chron. Bot. 9, no. 2/8 (1945) 123-127) and Baldwin (in Journ. Heredity and in Am. Journ. Bot., /oc. cit.) have suggested that the hills rising sharply out of the Amazonian plain in the Rio Negro—Rio Uaupés’ area of Brazil and Colombia possibly represent a route, now dis- continuous, along which plants migrated southwestward from the Venezuela-Guiana region. There may also have been another route southward from the Guianas. Bald- win (doc. cit. and in Am. Journ. Bot. 88 (1946) 215-216) has considered the Uaupés area to be the center of vari- ability of both Hevea and Cunuria. Representatives of Cunuria are known from the an- cient mountains in southern British Guiana and Surinam. The genus is also represented in southern Venezuela along the Casiquiare. It is common along the upper Rio Negro in Brazil and along the lower Rio Uaupés in Bra- zil and Colombia as well as in the area around Sio Paulo de Olivenea (one of the highest localities along the Rio Solim6es) and in the vicinity of Mandos (a relatively high area near the confluence of the Rio Negro with the So- lim6des, with a vegetation rather like that of the region of Serra de Sapucud, mentioned below). Cunwria is ex- ceedingly abundant on the slopes of Cerro de La Pedre- ® The Rio Uaupés, flowing through Colombian as well as Brazilian territory, is known as the Rio Vaupés in Colombia. [ 829 ] ra on the Rio Caqueta (Japura) in eastern Colombia. It is known from a single Peruvian locality near Iquitos in Loreto and has been collected from the high plateau be- tween the Rio Livramento and the Rio Ipixuna, tribu- taries of the Madeira, in Brazil. To these records we may add reports which suggest other areas where Cunuria possibly exists. South of the mountains in the Guianas where the genus has been col- lected, ‘‘there are some hills off the Rio Trombetas and near the Lago de Sapucua, known as Cunuri mountains, .... probably reaching four hundred feet’’ (Brown, C. B. and W. Lidstone: Fifteen thousand miles on the Amazon and its tributaries (1878) 241). Since the name cunurt is commonly applied to various species of Cunu- ria in Brazil and since the genus often dominates the slopes of such low mountains, as observed on Cerro de La Pedrera by Schultes and at Montepelago on the Rio Uaupés by Baldwin, we presume that Cunuria is com- mon on the hills designated by Brown and Lidstone. These hills are probably the ones now known as Serra de Sapucua. In 1943, a native rubber producer wished to demonstrate to Baldwin a new kind of buttressed rubber tree which might well be a representative of Cunuria. In the Rio Madeira region, Schultes heard repeated ref- erences to gigantic, buttressed rubber trees which were locally known as pae da seringueira—“*‘ father of rubber trees.’’ These are reputedly abundant west of Humayta, between the Rio Madeira and the Rio Purts, and they may belong to Cunuria. If the theory be accepted that an inland sea, opening westward, once covered a great part of what is known as the Amazonian planada, Cunuria might be considered to be an element of the flora of the heights fringing the shores of the sea. Accordingly, one would reasonably expect additional collections of Cunuria to help indicate [ 330 | the limits of that sea, for which now only a hazy outline can be suggested. Ifl CLAVIS SPECIERUM VARIETATISQUE GENERIS CUNURIAE: A. Folia chartacea, leviter marginata. Petioli graciles, 1/3 foliorum partem aequantes vel longiores. B. BB. Folia circiter 2/3 lata quam longa, plerumque 11 cm. longa, 7T em, lata, sicco brunneo-fusca, nitida sed non glauca, venis subtus elevatis, brunneis. Florum bracteae minutissimae. Stip- ulae usualiter caducae. Inflorescentiae saepe compactiores. (4) Cunuria Spruceana Folia circiter 1/2 lata quam longa, plerumque 15-16 cm. longa, 7-8 em. lata, sicco straminea, utrinque glauco-nitidissima, venis non elevatis, subimmersis, luteis. Florum bracteae magnae, usque ad 6 vel 7 mm. longae. Stipulae valde persistentes. In- florescentiae saepe laxae. (5) Cunuria Spruceana var. bracteosa AA. Folia firme coriacea, valde marginata. Petioli crassi, 1/5 folio- rum partem aequantes vel breviores. C. Petioli crassi. Capsulae parvae, usque ad 3.5 cm. longae vel breviores. D. Folia elliptica, siceo laminis discoloribus, apice obscure acuminata, basi usualiter cuneata vel subcuneata. Venae tertiae venis secundariis paralleles, immersae. (2) Cunuria crassipes DD. Folia late ovata, sicco laminis subconcoloribus, apice abrupte acuminata, basi usualiter rotundata. Venae ter- tiae venis secundariis angulatim 90°, omnes elevatae. (1) Cunuria australis CC. Petioli crassissimi. Capsulae magnae, 5 cm. longae vel lon- giores, (3) Cunuria glabra Ly: Cunuria Baillon in Adansonia 4 (1863-64) 287; Mueller Argoviensis in DC, Prodr. 15, pt. 2 (1866) 1123; Mueller Argoviensis in Martius Fl. Bras. 11, pt. 2 (1874) 507 ; Baillon Hist. Pl. 5 (1874) 190; Baillon Nat. Hist. Pl. 5 (1878) 190; Bentham et Hooker fil. Gen. Pl. 3 (1880) 289; Baillon Dict. Bot. 2 (1886) 299; Pax in [ 331 ] Engler & Prantl Pflanzenfam. 8, Abt. 5 (1890) 77; Pax in Engler Pflanzenr. LV. 147 (Heft 42) (1910) 16; Lemée, Dict. Genre Pl. Phan. 2 (1980) 418; Record in Trop. Woods 54 (1988) 18; Record & Hess, Timb. New World (1943) 157. Clusiophyllim Mueller Argoviensis in Flora 57 (1864) 518. Descriptio archetypa: ‘*F lores dioeci. Mase. ignoti. Foem. calyx 5-partitus, lacinils persistentibus imbricatis. Discus hypogynus (an androcaeum abortivum /) cupulaeformis glandulosus 6—8- lobus; lobis cuspidatis ovarii loculis dum 6 adsint) per paria oppositis. Ovarium liberum 8-loculare ; loculis uni- ovulatis; ovulo pendulo obturato; micropyle extrorsum supera. Stylus sessilis 8-partitus; laciniis crassis 2-fidis reflexis ovario adpressis. ‘*Arbor ¢ foliis alternis, stipulis infraaxillaribus folia juniora (ut in Artocarpeis) involventibus caducis; limbo simplici basi glandulis 2 sessilibus concavis munito. ”’ Descriptio amplificata: Arbores magnae, elatae, parco cum succo lacteo, foliis alternis, petiolatis, stipulis infraaxillaribus folia juniora involventibus caducis. Folia integra, penninervia, coria- cea vel firme chartacea, pagina supra basim biglandulosa, costis secundariis angulo amplo decurrentibus, arcuatis, tertis reticulatis vel subparallelis. Pedunculi laterales vel axillares, foliis multo breviores, rigidi, staminati simplices vel parce ramosi, floribus in apice cymarum pauciflorum ramorum sessilibus. Flores dioeci vel monoeci, apetali. Floris staminati sepala quinque, valde imbricata, quin- cuncialia; discus nullus; stamina decem receptaculo el- evato affixa; filamenta brevia libera; antherae dorsifixae, loculi connectivo latiusculo adnati, longitudinaliter de- [ 332 ] hiscentes, rudimentum ovarii tripartitum. Floris pistillati sepala quinque, laciniis eis maris similia; discus hypogy- nus cupuliformis, glanduloso-lobatus vel simplex, ova- rium liberum triloculare, loculis uniovulatis; stylus ses- silis, tripartitus cum divisionibus bifidis, crassus, reflexus, ovario adpressus. Fructus capsularis; capsula magna, subglobosa vel saepe quasi ovoidea, in coccas bivalves partiens, carnoso cum epicarpio et lignoso cum endocar- plo crasso. Semina magna, ovoidea, testa crustacea et nitidissima, immaculata, ecarunculata. Albumen carno- sum; cotyledones planae, latae. Species typicus: Cunuria Spruceana. Nominis generis significatio: ‘‘Cunuri’’ in regione typica nomen vulgare est. V 1. Cunuria australis R. EL. Schultes sp. nov. Arbor magna, usque ad 20 m. alta. Rami robusti, cinereofusco cum cortice crasso. Ramuli aequales, sub- teretes, glabri, apice ipso autem cum stipulis caducis- simis, minute adpresso-pubescentibus. Stipulae triangu- lares, minutae, 6mm. 2mm. Petioli crassissimi, 30-35 mm. longi. Folia coriacea, marginata, late elliptica, apice abrupte acuta, basi rotundata, 14-16 em. longa, 7.5-9.5 cm. lata, venis supra nitidis glabrisque, non elevatis, sicco stramineis, infra prominenter elevatis, fuscis, glandulis duabus comparate parvis, 1 mm. latis, fusco-nigricanti- bus limbi basi. Flores ignoti. Capsulae subglobosae, sicco usque ad 38.5 cm. longae, 3.5 cm. in diametro; epi- carpio glabro, tenul, superne plus minusve 8 mm. crasso, non conspicue fibroso; endocarpio lignoso, tenui, usque ad 2 mm. crasso; valvis regularibus, dehiscentibus non contortis, 2.8 em.longis, 1.2 em. latis, intus aureo-brun- neis et striolatis. Semina ignota. [ 388 ] Brazit: Estado do Amazonas—Basin of Rio Madeira, Municipality Humayta, on plateau between Rio Livramento and Rio Ipixuna. ‘Tree 60 ft. high. Cipoal.’’ B. A. Krukoff’s 5th Exped. Braz. Amaz. 7201, November 7-18, 1934. (Typus Herb. Arnold Arb. ; Herb. N.Y. Bot. Gard. ; Herb. Jard. Bot. Rio.) This is the only representative of the genus which occurs at any significant distance south of the Amazon River, a consideration which has suggested the specific epithet. This new species of Cunuria is of considerable interest because it has a capsule rather comparable to that of C. crassipes, whereas in vegetative characters it is extremely similar to C. Spruceana. The petioles of Cunuria australs are much thicker than those of C. Spruceana, and the leaves appear to have been, in life, a bit more coriaceous in texture. The stipules, if indeed they are present on Cunuria australis, are very caducous, Just as they are in C. Spruceana. Cunuria australis is probably more closely allied to C. Spruceana than to C. crassipes, and the specimens have, in the past, been identified as representing C. Spruceana. The ligneous valves of Cunuria australis apparently do not open violently, as in the other species of the genus; the lack of the characteristic twisting of the valves is in- terpreted as indicative of slow and gradual dehiscence. A parallel example can be cited in the case of Hevea minor Hemsl., the only species in the genus in which the valves open slowly, not in a violently explosive manner, and persist for some time after the seeds have dropped. The fruit of Hevea Spruceana (Benth.) Muell. Arg. open explosively, but the valves remain on the tree. The capsule of Cunuria australis is only about half as large as that of C. Spruceana; the woody endocarp is rather thin, and the fleshy epicarp is much thinner and measur- ably less fibrous. The difference in size of the fruit is so great that, even lacking the other characters of shape and [ 384 | texture, it would be obvious that the Madeira collection is specifically distinct from Cunuria Spruceana. One could easily suspect that the Rio Madeira material represents a combination of characters from two species, in which case there might be apprehension that a specific name is being assigned to a segregant from interspecific crossing. Cunuria crassipes, apparently an extremely localized endemic of the Rio Negro area, could be con- sidered one parent of such an hypothetical hybrid, and C. Spruceana, also known only from a region far to the north of the Rio Madeira, the other. Possibly future ex- ploration may turn up material of Cunuria Spruceana and C. crassipes in the Rio Madeira. These species grow in caatingas, often in association with Hevea rigidifolia, Hf, viridis and H. pauciflora var. coriacea (possibly cor- rectly called H. confusa). All these plants, when ‘‘pure,”’ seem to have precise ecological requirements, and these requirements are similar for each of the species cited. It is interesting, therefore, that Hevea viridis is known from Borba on the Rio Madeira and that HY. camporum (which Ducke is considering as possibly a form of H. pauciflora var. corvacea and which Baldwin would refer to HZ. con- JSusa) is known from the great plateau at the headwaters of the Rio Marmellos, an affluent of the Madeira. These localities are not distant from the general region where the type material of Cunuria australis was found. It would not be surprising, therefore, to discover in this area additional species of Cunuria or any of the already recognized species. Certainly the requisite ecological con- ditions exist in the Madeira Valley. 2. Cunuria crassipes Mueller Argoviensis in Mar- tius FJ. Bras. 11, pt. 2 (1874) 510; Corréa Diccion. PI. Uteis Brasil. 2 (1981) 482; Pax in Engler Pflanzenr. IV. 147 (Heft 42) (1918) 17. [ 885 ] Clusiophyllum Sprucei Mueller Argoviensis in Flora 57 (1864) 518. Cotomp1a: 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-elastiec gum. Flowers greenish yellow. Trispermate Hevea-like capsule. Known locally as wah-so-né-né in Tukano,’’ Paul H. Allen 3068, August 18, 1943. (Herb. Gray). Venezue.a: Territorio del Amazonas—‘‘Arbor 70-pedalis. In syl- vis ad basin montis Cocui,’’? July 1853. Spruce 3029 (Herb. Kew); “*Prope San Carlos ad Rio Negro’’ Spruce 3029 (3474) (Herb. Kew; Herb. Brit. Mus.; Herb. Univ. Cambr.); “*Prope San Carlos ad Rio Negro, 1853—54’’ Spruce 3473 (Herb. Brit. Mus.); ‘* Arbor lactescens 60-pedalis. San Carlos, in sylvis.’? April 1854 Spruce 3474 (Herb. Kew; Herb. Univ. Cambr.)—Rio Negro, near San Carlos. Richard Spruce 3474, 1853-54 (Lecrorypus: Herb. Gray; Herb. N.Y. Bot. Gard.).—Same locality. Richard Spruce 3029. (Photo.spec. Hort. Berol. in Herb. Gray; Herb. Field Mus. 646435). Brazit: Estado do Amazonas — Rio Uaupés, Montepelago. “‘In a sort of caatinga-forest. Tree to 60 ft., buttressed. Fruit cherry- colored, just as in Hevea confusa and in contrast to the green fruit of C. Spruceana and Cunuria sp. at Foz do Uaupés. Chromosomes: 2 = 36.°? Nom. vulg. cunuri, J. T. Baldwin Jr., 3673, March 17, 1944. (Herb. Arnold Arb.; Herb. Instit. Agron. Norte; U.S. Nat. Herb. ; Herb. Kew.)—Same locality. ‘‘In a sort of caatinga-forest. Tree to 40 ft., buttressed. Latex white. Fruit flushed with cherry red. Chro- mosomes: 2 »=36.°’ Nom. vulg. cunuri. J. T. Baldwin, Jr. 3676. March 17, 1944. (Herb. Arnold Arb.; U.S. Nat. Herb. )—Same lo- eality. ‘“Tree to 40 ft., buttressed. Fruit cherry red. In a sort of caatinga-forest. Chromosomes: 2 »=36.’’ Nom. vulg. cunuri. J. T. Baldwin, Jr. 8681, March 8, 1944. (Herb. Arnold Arb. ; Herb. Instit. Agron. Norte; U.S. Nat. Herb.; Herb. Kew.)— Foz de Uaupés. “*Total height of 60 ft., buttress of 3 ft. Wood of this tree in Har- vard collection. Growing not far from C. Spruceana, Micrandra, Hevea viridis, and H. guianensis. Chromosomes : 2 n= 36"’, Nom. vulg. cunurt da caatinga. J. T. Baldwin, Jr. 3698, March 13, 1944, (Herb. Arnold Arb. ; Herb. Instit. Agron. Norte; U.S. Nat. Herb.; Herb. Kew.) —TIraruca, Icana. ‘*Arvore 15 ms., 20 em. (diam.) Latex. Madeira dura, folhas quebradicas, verde negro no dorso; claro no ventre. Cunuri. [ 336 | Caatinga.’’ Ricardo de Lemos Frées 21396, Nov. 16, 1945. (U.S. Nat. Arb. Herb, 182206.) Enough collections of Cunaria crassipes are now avail- able to assure us that this species concept is compara- tively constant. Apparently the species is a very localized endemic of the upper Rio Negro area and the adjacent lower Uaupés. This region is known to have high en- demism, a fact which is emphasized occasionally, as when Allen rediscovered Cunuria crassipes, one of the endem- ics which Spruce collected there almost a century ago. No other known species of Cunuria has the venation of C. crassipes. Its coriaceous, strongly marginate leaves are supplied with a relatively strong central nerve, rather prominently elevated on the under surface but only slightly so above; eight or nine secondaries which are extremely slender and hardly elevated; and tertiaries which anastomose but which are, in general, parallel to the secondaries. In all other species, the tertiaries are set at right angles to the secondaries and are not parallel. All specimens of Cunuria crassipes which have been available for study have the upper and lower surface of the leaf very unequally discolored when dry, the upper usually con- serving a very characteristic bluish-grey glaucescence. The flowers of Cunuria crassipes are still unknown. The specimen of F’rées 21396 in the U.S. National Ar- boretum Herbarium has the remnant of one flower in a poor state of preservation. It is possible to ascertain from it, however, that the flower is much larger than in Cunuria Spruceana. According to the collector’s notes, Baldwin 3673, 3675 and 3681 have a capsule with a cherry-red epicarp when ripe. Cunuria Spruceana, as stated on the label of Bald- win 3673, has a green epicarp. Baldwin 3698, which is here included under Cunuria crassipes, is larger than the typical material. The collec- [ 337 ] tor has indicated that the tree was more robust than the other representatives of the species which he examined and that there were visible differences in wood-texture. The leaves of Baldwin 3698 are larger and more broadly elliptic than is characteristic of the species. Furthermore, the fruit is green instead of the usual cherry-red. While it may be true that Baldwin 3698 does not represent a ‘pure’ line of Cunuria crassipes, there are no good dis- tinguishing morphological characters to set it apart as a separate taxonomic entity, and little of a definitive na- ture may be stated regarding the possibility of admixture of genes—(perhaps) from C. Spruceana—until much more field work is carried out. It is interesting to note that Allen 3068 very closely approaches the type in the size and shape of the leaves. The leaves of this collection are somewhat smaller than those of the Spruce material ; those of the several Baldwin collections tend to be some- what larger. In no specimens of Cunuria crassipes are we able to find traces of large foliaceous stipules. It is probable that these do not occur in this species, or, if they do occur, they are extremely caducous. All infor- mation indicates that the buttress roots are much smaller in Cunuria crassipes than they are in C. Spruceana, and the former is generally a much smaller tree than the latter. The epithet Cunuria crassipes is being conserved, in conformity with Recommendation XIV of the Inter- national Rules of Botanical Nomenclature, over the ear- lier Clusiophyllum Sprucet. Were the indicated combi- nation to be made, a new name would enter into the taxonomic literature. This name would be unfortunate because of its resemblance to Cunuria Spruceana, and endless confusion would be the result. In this connec- tion, it should be noted that the name Clusiophyllum Spruce? was based upon Spruce $029 and 3474. Although [ 338 ] these two collections were, in the early period of inves- tigation of the genus, referred to Cunuria Spruceana, there is no Justification for Pax’s citation of Clusiophyl- lum Sprucet as a synonym of C. Spruceana (Engler Pflanzenr. 1V. 147 (Heft 42) (1910) 16). Little is known about the latex of Cunuria crassipes. In Allen 3068, it was ‘‘yellowish, scant, coagulating to a non-elastic gum’’; while in Baldwin 3673 and 3675, the color of the latex is reported to be white. In the opinion of the writers no known species of Cu- nuria appears to be promising as a commercial latex plant. 3. Cunuria glabra R. E. Schultes sp. nov. Arbor monoecia, usque ad 75 pedes alta, radicibus cum tabularibus. Folia valdissime coriacea, perfecte ovata vel rarenter elliptico-rotundata, omnino glaberrima, in speci- mine typico 18-21 cm. longa, 12-18 cm. lata (in speci- minibus aliis multo minora: 10-16 em. X8-11.5 cm.), apice rotundata vel rarenter obscurissime subacuminata, basi rotundata vel subcordata, valde marginata, supra brunnea vel glauco-brunnea, venis omnino (praecipue superfice inferiore) conspicue elevatis, secundariis octo arcuato-adscendentibus, sub marginem ipsum tenuiter anastomosantibus, tertiis prominentibus reticulatis sub- parallelis, glandulis comparate magnis. Petiolus crassissi- mus, sicco striato-fibrosus, 8.5-4 em. longus, 4 mm. crassus. Inflorescentia subapicalis, laxa sed pedicellis rigi- dis crassisque, parce et minute adpresso-tomentellis. Flores staminati alabastro subglobosi, 2X8 mm. in di- ametro, calycibus ovatis, usque ad 5 mm. longis, 8 mm. latis, concavis. Florum pistillatorum calyces similes sed crasslores majoresque, 6 mm. longi, 4mm. lati. Bracteae magnae, concavae, ovatae, 6 mm. longae, 4 mm. latae. Fructus (Pinkus 236) vivo usque ad 5.5 cm. longus, 3 [ 889 | cm. latus, endocarpio tenuissimo, epicarpio tenui ut videtur, glabro, pedunculo robusto usque ad 8 em. longo. Durcn Guiana: Tafelberg (Table Mountain). “‘Common tree 25 m. tall, 40 em. diam., rooting at base to 2 m.; scant white latex oxidiz- ing to yellow, leaves chartaceous; fl. white; dom. high tree, north of savanna 2.’’ Bassett Maguire 24279, August 10, 1944 (Typus Herb. Arn. Arb.). British Guiana: Membaru Creek, upper Mazaruni River, “Tree in mixed forest, 70 ft. high; trunk 16 in, diam. ; latex white; fruit green without, white within; seed edible.’’ Albert S. Pinkus 236, February 16, 1939. (Herb. Arn. Arb.; U.S. Nat. Herb. 1776073; Herb. Mo. Bot. Gard. 1175213, 1175214; Herb. Field Mus. 1001560; Herb. N.Y. Bot. Gard.; U.S. Nat. Arb. Herb. 156385). It would appear that Cunuria glabra is allied to C. Spruceana. Croizat (in Bull. Torr. Bot. Club 57 (1940) 289) cited Pinkus 236 as representing Cunuria Spruce- ana. He pointed out, however, that the specimens ‘‘have slightly larger leaf blades which are truncate or fairly cordate at the base and larger capsules to 6 cm. long.”’ When sufficient material is amassed, it is at once evi- dent that Cunuria glabra is distinct from C. Spruceana. Besides the size and shape of the leaves, there are impor- tant differences in the texture and margin; in Cunuria glabra the leaves are extremely thick, coriaceous and very conspicuously marginate. The floral bracts are rela- tively large in Cunuria glabra, the flowers are extraordi- narily large, and the petioles are very thick and rigid. 4. Cunuria Spruceana Baillon in Adansonia 4 (1863-64) 288; Mueller Argoviensis in Martius Fl. Bras. 11, pt. 2 (1874) 510, t. 14; Corréa Diccion. Pl. Uteis Bras. 2 (1931) tab. on p. 482; Le Cointe A Amazonia Bras. 3 (1984) 145. ‘* Micrandra et Pogonophora ¢ Cunurt H. Bn. ol. in exs. Spruce’’ in Adansonia 4 (1863-64) 288. Micrandra Cunuri Baillon ex Mueller Argoviensis in DC. Prodr. 15, pt. 2 (1866) 1123. [ 340 ] Brazit: Estado do Amazonas— Foz do Uaupés. ‘*Cunuri da terra Jirme, 60 ft. bole above 6 ft. buttress to crown rising 30 ft. and with 30 ft. spread. Wood fairly soft; sample of this tree in Harvard wood collection. Seeds eaten by natives; see Spruce for preparation. With Hevea and Monopteryx in unflooded forest. Chromosomes: 2 n= 36.’ J. T. Baldwin Jr. 3683, March 13, 1944. (Herb. Gray; Herb. Instit. Agron. Norte; U.S. Nat. Herb.; Herb. Kew.)—Serra Comande, Umarituba (above Santa Isabel). ““Cunuré da terra firme.”’ J. T. Bald- win Jr. 83689, March 22, 1944. (Spec. cum fructu unico: Herb. Arn. Arb.; Herb. Instit. Agron. Norte; U.S. Nat. Herb.)—Sao Paulo de Olivenca. ‘‘Mata. Arvore grande. Cunuri.’’ Ricardo de Lemos Frées 20752, April 1945 (U.S. Nat. Arb. Herb. 182190, 182204),—Porto Curucui, Sao Gabriel, Rio Negro. ‘‘Terreno silicoso, beira rio nao inundavel.’’ Ricardo de Lemos Frées 21129, Oct. 9, 1945. (U.S. Nat. Arb. Herb. 182193, 182205). Cotompia: Comisaria del Vaupés— Rio Papuri, vicinity of Santa Teresita. ‘Tall trees averaging 35 m. in height, and 80 cm. in di- ameter above the conspicuously developed and unique stilt buttresses which are produced to a height of 3-4 m. as laterally compressed board-like flanges which act as flying buttresses, the bases widely de- tached from the trunk, the arch often high enough for a man to walk upright beneath. Leaves simple, with two basal disk-like glands. In- florescence of small green flowers from axillary growth. Large trisper- mate capsule typical of Hevea. Bark very thin, 6-8 mm., with scanty latex which coagulates with difficulty, producing a non-elastic gum. Seeds collected by the local Indians for food. Known as “‘ wah-puh’’ (Tukano).’? Paul H. Allen 3063, August 15, 1943. (Herb. Gray). Peru: Departamento de Loreto— Mishuyacu, near Iquitos, alt. 100 m., forest. “‘Tree 15 m. high, fl. light yellow.”’ G. Klug 1312, May- June 1930. (Herb. Field Mus. 627499).—‘‘Tree 20 m. high; fl. yellow.’ G. Klug 1325, May-June 1930. (Herb. Field Mus. 627508). —‘**Tree 18 m. high; fl. white.’’ G. Klug 340, October-November 1929. (Herb. Field Mus. 624153). Venezueta: ‘“Cunuri, ling Barré. Arbor vasta, 100-140 pedalis lac- tescens, in radicibus exsertis late arcuatis suffulta. Pericarpium exter- num subcarnosum, Ad Casiquiarem, in sylvis. Jan. 1853. Per partes superiores fluviorum Nigri et Pacimonis ut etiam per totum fluv. Uaupés vulgata.’’ Spruce 3299 (Herb. Kew (two specimens); Herb. Brit. Mus. ; Herb. Univ. Cambr.)—‘‘ Ad flumina Casiquiari, Vasiva et Paci- moni. Cunuri.’’ Richard Spruce 3299, 1853-54. (Typus) (Herb. Gray ; Herb. N.Y. Bot. Gard. Photo spec. Herb. Geneva). [ 341 ] In addition to the specimens cited above, we may refer to several collections from the Rio Caqueta drainage area in eastern Colombia. These collections— Schultes 5869 and 5989 from the slopes of Cerro de La Pedrera and Schultes 5895 and 5932 from the Rio Miritiparana—were studied in the field, and critical notes, written in the field, were published (Schultes in Caldasia 8, no. 13 (1945) 247-249). Although the specimens, together with several hundred others from the same region, were un- fortunately lost in shipment, there is no question about the identity of these four collections. The fruit was very large, 6 em. long or longer, and was, in all specimens, a perfect match for the fruit of Baldwin 3689. A photo- graph of the typical enormous buttressed roots was like- wise published (doc. cit. fig. p. 248). Along the lower Caqueté and Miritiparana rivers, especially in the vicinity of the town of La Pedrera (an important Martius type locality formerly known and cited as ‘‘Cupati’’), Cunu- ria Spruceana is one of the dominant trees on high land, especially where there are rock outcrops. The Miranha Indians know the tree as ko-nd-ko; the Yukunas as yé- cha. Schultes (loc. cit.) reported: ‘‘Where it occurs in the lower Caqueta, Colombia, it is abundant, in some places. . .. crowding out other common forest trees and forming pockets of almost pure stands. It does not occur over wide extensions, but seems to be localized in swampy or well-watered areas where rock outcrops (sandstone as well as granite) are frequent.”’ Martius apparently did not collect this species in the Cupati locality, where it is so well represented, because he happened to pass in the interim between the flowering and the fruiting season—in January and February (see Dugand in Rev. Acad. Colomb, Ciéne. Exact. Fisie. Nat. 5, no. 18 (1942) 212 ff.). Ducke, who collected rather extensively in this same locality in late 1922, ap- [ 842 ] parently did not collect Cunuria Spruceana, although it should have been in flower at the time (Ducke in La Géograph. 80 (1914-15) 865-872). This serves to indi- ‘cate the extremely localized and disrupted occurrence of the species, a phenomenon which has been noted in other regions for this as well as for other species of Cunuria. A large tree probably representing Cunuria Spruceana was observed by Schultes in numerous discontinuous lo- calities in the lower Rio Apaporis in eastern Colombia (Schultes in Chron. Bot. doc. cit.) during a survey of Hevea rubber which was made along that river in 1943. Cunuria Spruceana is an important food plant of many Indians, for the seeds, like those of Hevea, are a favorite food when properly prepared. Spruce (in Hook. Journ. Bot. 6 (1854) 333-337) reported that from the seeds of ‘“cunuri, abundant on the Alto Rio Negro, Orinoco, Casiquiare, Pacimoni, etc. the Indians prepare a paste resembling cream-cheese in appearance and taste. The seeds are first boiled and then steeped for some days un- der water, after which they are broken up by the hand. In the boiling, a quantity of oil is said to be collected ; . it is said to be as bitter as andiroba oil, but to afford an excellent light.’’ There is a handwritten nota- tion on the herbarium specimen at Kew which gives much greater detail concerning the native uses of the seeds of Cunuria Spruceana: ‘‘On the Uaupés and around Sio Gabriel, a large tree, obviously allied to S7- phonia | Hevea], called by the Indians Cunuré is frequent in the forest. It has large arched buttresses at the base, like the uacu [Monopterya Uaucu Spruce ex Benth. }, from which it is distinguished by milk flowing from it when wounded. I have not yet seen its flowers or fruits, but the Indians describe the latter as tricoccous, quite as in Siphonia, and they use the seeds in a similar man- ner. These being boiled 24 hours yield a small quantity [ 843 ] of oil, which serves for lamps. The pulpy mess into which the seeds have now fallen is packed in a basket and kept under water 8 days to sweeten; when taken out, it has a pleasant taste and no ill smell. It is eaten without the addition of anything else and may be kept a long time, but if the seeds have not been well boiled, it is a quick poison, and Indians have fallen victims to its incautious use.’’ Schultes found the Indians of the lower Caqueta in Colombia similarly utilizing Cunuria Spruceana as a food, and stated (/oc. cit.) that the seeds ‘apparently contain a cyanide and, according to the na- tives, are extremely poisonous when taken internally in the crude state.’’ The Indians there ‘‘consume large quantities of the seeds in the form of a greyish mash which is prepared by boiling the pulp in three waters to remove the poison. This mash has a peculiar taste, some- what like burnt potato. According to the natives, salt must not be added to this mash.’? Ducke reported this use for Indians of the Rio Negro (in Le Cointe Joe. cit.). Allen 3063 records a comparable use by the natives of the Rio Papuri, an affluent of the Vaupés which forms part of the Colombo-Brazilian boundary. And, in 1944, Baldwin found the seeds used, Just as reported in 1858 by Spruce, in the upper Rio Negro—Rio Uaupés region. Cunuria Spruceana was the first species of the genus to be described. An extended description appeared in Flora Brasiliensis. The only two illustrations of Cunuria have been of this species; the second is a poor copy of the first which was published in Flora Brasiliensis. Prob- ably because of this emphasis on one species, collections of Cunuria have almost always been referred to C. Spruceana. This has led to the erroneous assumption that Cunuria Spruceana is more widely distributed than it really is. As we interpret it, this species occurs, discon- tinuously, in an are or crescent from the Casiquiare and [ B44 | Rio Negro southwestward across the easternmost por- tion of the Colombian Amazonia to a locality near Iqui- tos in eastern Loreto, Peru, and to the upper SolimGes in Brazil. The Peruvian specimens which are here referred to Cunuria Spruceana are slightly atypical in a number of minor characters of the flowers and in having stipules which are not readily caducous. In some respects, it would seem that the Peruvian specimens are somewhat intermediate between typical Cunuria Spruceana and C. Spruceana var. bracteosa. Until more collections are available from this westernmost station for the genus, we feel that it is unwise to interpret these variations as representing a distinct taxonomic concept. 5. Cunuria Spruceana Baillon var. bracteosa (Ducke) R. EH. Schultes comb. nov. Cunuria bracteosa Ducke in Notizbl. Bot. Gart. Ber- lin 11 (1982) 586, in synon.; in Arch. Jard. Bot. Rio Janeiro 6 (1988) 57. Brazi_: Estado do Amazonas—Rio Solimoes, Sao Paulo de Olivenca. “‘Frecuens. Silva non inundabili, saepius locis humidis vel ad rivulos. Arbor maxima, flor, albido-viridibus.’’ A. Ducke Herb. Jard. Bot. Rio 23519, August 20, 1929 “‘florif., fructus maturi tempore pluvioso.”’ (Typus). (U.S. Nat. Herb. 1516538, 1517678; Herb. Jard. Bot. Rio; Herb. Kew).— Rio Negro super ostium flum. Curicuriary. “‘ Silva non inundabili. Arbor magna radicibus tabularibus magnis, flor viridi- bus.’’ Nom vulg. Cunury. A. Ducke Herb. Jard. Bot. Rio 24873, De- cember 24, 1931. (U.S. Nat. Herb. 1617664; Herb. Jard. Bot. Rio 24873; Herb. Kew).—Manaos, estrada do Aleixo. ““Silva non inun- dabili, prope rivulum. Arbor magna; flores masculi albi, odore ut in Hevea.’’ A. Ducke Herb. Jard. Bot. Rio 24874, December 18, 1941. (U.S. Nat. Herb. 1617665; Herb. Jard. Bot. Rio 24874; Herb. Kew). —Manaos. ‘‘ Matta da t. f. silicoso-humosa, loga humido perta d’um riachinho. Arv. gr. com sapupemas altas, latex branca, fl, masc. braca- centa com cheiro que lembra Hevea.’’ Nom. vulg. cunuri. A. Ducke 77, June 1932. (Yale School Forestry Mus. Ser. 21336; Herb. Field Mus. 657675).— Rio Negro, Sao Gabriel. ‘Silva non inundabili. Arbor magna flor. viridibus.’’ Nom. vulg. cunuri. A. Ducke 146, February [ B45 | 16, 1936. (Herb. Arn. Arb.; Herb. Field Mus. 902328; Herb. Mo. Bot. Gard. 1158515; Herb. N.Y. Bot. Gard.; U.S. Nat. Herb. 1693155, 1694262; Herb. Jard. Bot. Rio 35437; Herb. Kew).— Manaos, Estrada do Aleixo, ‘‘Silva non inundabili. Arbor magna flor- ibus albidis.’? 4. Ducke 848, December 18, 1941. (Herb. Mo. Bot. Gard. 1255541; U.S. Nat. Herb. 1875676; U.S. Nat. Arb. Herb. 167007).—Manaos, Igarapé do Crespo. “‘Ad marginem igapo. Arbor magna floribus albidis. 4. Ducke 1087, December 12, 1942. (Herb. Mo. Bot. Gard. 1263939; U.S. Nat. Herb. 1832363; U.S. Nat. Arb. Herb, 167008, 167009).—Tonantins. ‘‘Mata da terra firme, logar um- ido. Arvore muito grande; flor. brancacenta. Arbor maxima floribus albidis.’? 4. Ducke 1554, February 10, 1944. (Herb. Arn. Arb.; U.S. Nat. Herb, 1906514; U.S. Nat. Arb. Herb. 167213).—Manaos. ‘*arvore. Mata.’’ Ricardo de Lemos Frées 20505, Feb. 22, 1945 (U.S. Nat. Arb. Herb. 182200).—Sao Paulo de Olivenca. “‘ Mata, caatinga. Arvore.’’ Ricardo de Lemos Frées 20697, April 1945. (U.S. Nat. Arb. Herb. 182201). Cunuria Spruceana var. bracteosa is one of the most widespread concepts of the genus. It is found in the upper Rio Negro in close proximity to Cunuria Spruce- ana; it is apparently frequent in the vicinity of Mandos; and it is known from the Rio Solimdées. On the highland forest floor near San Pedro on the Rio lea (Putumayo), between the Colombian boundary and the mouth of the river, Schultes, in 1946, saw fruit and seed of what ap- parently is C. Spruceana var. bracteosa. This is the west- ernmost known locality for the variety. Cunuria Spruceana var. bracteosa is very easily recog- nized and separated from C. Spruceana by having re- markably large floral bracts and large foliaceous stipules which are extremely persistent. In herbarium specimens, the leaves of the variety invariably dry a glossy straw- color, while those of the species always turn a dull, dark brown. Furthermore, the nerves of the former are yel- low when dry and are not elevated; those of the latter are dark brown when dry and are very prominently ele- vated. The capsule and seed of var. bracteosa are very [ 346 | similar to, but always smaller than, those typical of the species itself; the capsule usually measures about 4 cm. long. Ducke published Cunuria bracteosa in 19838, but a year previously he had said (in Notizbl. /oc. cit.) that ‘‘zu dieser Art [C. Spruceana]| gehért wohl sicher das Herb- armaterial das unter dem Namen Cunuria bracteosa nov. spec. von mir verteilt wurde; wahrscheinlich ist iibrigens diese Gattung monotypisch.’’ It is obvious that Ducke continued to regard his species, Cunuria bracteosa, as synonymous with C. Spruceana, because his more recent collections of the former concept have all been deter- mined by him as representing the latter, and, in 1934, he reiterated his opinion that C. bracteosa should be re- garded as a synonym of C. Spruceana (in Arq. Instit. Biol. Veget. Rio Jan. 1, no. 2 (1984) 91, footnote 1). A letter from Ducke (April 1982) attached to A. Ducke 23519 in the Kew Herbarium states: ‘‘I distrib- uted the no. 23519 with the name Cunuria bracteosa n.sp. (Sao Paulo de Olivenca, Solimées River), but lately I collected similar material on the upper Rio Negro, typi- cal region of Cunuria Spruceana. Cunuria bracteosa dif- fers of the description of Cun. Spruceana by bracteae well developed and by the glandulae of the exterior of the male flowers receptacle, but I now presume that Spruce did not collected material in good conditions and that the Cunuria might be only of one single species. Still, I dare to say that the flowers as well as the cap- sules and seeds may vary in size on the same tree enor- mously. Consequently, I resolved not to publish mean- while my supposed new species.’ VI CONCEPTUS EXCLUDENDI: Anomalocalyx Uleanus (Paw et Hoffman) Ducke [ 347 | in Notizbl. Bot. Gart. Berlin 11 (1982) 344; in Arch. Jard. Bot. Rio Janeiro 6 (1938) 60, tab. 6, fig. 8. Cunuria Uleana Pax et Hoffman in Engler Pflanzenr. IV. 147. xIv (Heft 68 in part) (1919) 51. When Pax and Hoffman described this concept and referred it to Cunuria, they had only fruiting material at hand. From flowering and fruiting material which he had collected at the type locality, Ducke described the genus A nomalocalyx to accommodate the concept which, up to that time, had been accepted as Cunuria Uleana. We are in agreement with Ducke that this genus has no relationship with Cunuria. A study of the remarkably complicated flowers of Anomalocalyx Uleanus, well fig- ured by Ducke, indicates this fact. The structure and texture of the capsule likewise substantiates such a conclusion. Cunuria ? casiquiariensis Croizat in Journ. Arnold Arb. 26 (1945) 192. When Croizat described this concept, placing it with reservation in Cunuria, he stated clearly: ‘*‘Cunuria is suggested by the intangibles of habit and the characters of the foliage, but the inflorescence is somewhat uncon- ventional.’” The inflorescence would appear to diverge so widely from that which is usual for Cunuria, that we feel that this plant cannot be included, even provisional- ly, in the genus. For several valid reasons, Croizat has excluded it from Conceveiva and Concevastrum. In many respects, it suggests Pogonophora, but the lack of stam- inate flowers makes it impossible to be certain of proper generic reference. In a number of other respects, it 1s extremely similar to Micrandra. Cunuria Gleasoniana Croizat in Bull Torr. Bot. Club 57 (1940) 289. [ 348 ] The presence of a definite caruncle on the seed of this concept renders its inclusion in Cunuwria untenable. VII The authors wish to express their appreciation to the following institutions for the loan of material and for other courtesies: Arnold Arboretum, and Gray Herbar- ium of Harvard University: Missouri Botanical Garden: Chicago Museum of Natural History; New York Bo- tanical Garden; Yale School of Forestry: U.S. National Herbarium and U.S. National Arboretum Herbarium. The authorities of the Jardin Botanico de Rio de Janeiro and the Herbario do Instituto Agronémico do Norte in Belem do Parad, both in Brazil; the British Museum (Natural History) and the Royal Botanic Gardens, Kew are thanked for their kindness in permitting the speci- mens in their care to be studied. [ 849 | EXPLANATION OF THE ILLUSTRATIONS Pirate XLII. Cunuria austrauis R. E. Schultes. 1, terminal branch, one half natural size. 2, two valves of the capsule, natural size. Drawn by G. W. Ditton Pirate XLIII. Cunurta crasstprs Muell. Arg. 1, terminal branch, one half natural size. 2, two valves of the capsule, natural size. 3, seed, natu- ral size. Drawn by G. W. Diti0N Pirate XLIV. Cunuria arapra R. E. Schultes. 1, terminal branch, one half natural size. 2, two valves of the capsule, natural size. 3, staminate flower, four times natural size. Drawn by G. W. Ditton Pirate XLV. Cunuria Spruceana Baill. 1, termi- nal branch, one half natural size. 2, two valves of the capsule, natural size. 3, seed, natural size. 4, pistillate flower, four times natural size. Drawn by G. W. Ditton Pirate XLVI, Cunuria Spruceana Baill. var. prac- TEosA (Ducke) R. E, Schultes. 1, terminal branch with inflorescence, one half natural size. 2, two valves of the capsule, natural size. 3, seed, natural size. 4, pistillate flower, four times natural size. 5, staminate flower, four times natural size. Drawn by G. W. Ditton Pirate XLVII. Cunuria Spruceana Baill. Photo- graph taken at La Pedrera, Colombia, showing the enormous buttress roots so characteristic of the species, Photograph by R. E. Scuvuites [ 851 ] Pirate XLII RE Schulies KK TAN NGS NIRS | gS ZARA RS 5 aol USES oN ; ETTORE 2 Sil i ‘ ° ECCT y ~ ETRY Vee rete XLII CUNURIA crassioes Muell Arg. loo > AU le £47 WU ‘ li, Y cg Z ike 17 Z [ NG - “fe 4 4 ON iting x Se. fe Le: gd tt§ PiuateE XLIV glabra RE. Schultes OS AUT SOA WO CUNURIA PLatE XLV CUNURIA Spruc eana Baill. 4 ee S KS \ AN iH } 1 =, Wa PLATE p.< LVI CUNURIA OWN SW <— Can As ae: EN a 39a WAH f V, \\ \ ucke ruceana Baill .Gracteosa (O Sp var XLVII PLATE > _ < e: ;