3; 53 HAWAII AGRICULTURAL EXPERIMENT STATION HONOLULU, HAWAII Under the supervision of the UNITED STATES DEPARTMENT OF AGRICULTURE BULLETIN No. 53 THE HAWAIIAN TREE FERN AS A COMMERCIAL SOURCE OF STARCH BY J. C. RIPPERTON, Chemist Issued July, 1924 „/>""•' <*\*M/ifc WASHINGTON GOVERNMENT PRINTING OFFICE 1924 HAWAII AGRICULTURAL EXPERIMENT STATION, HONOLULU LUnder the supervision of the Office of Experiment Stations, United States Department of Agriculture] E. W. Allen, Chief, Office of Experiment Stations. Walteb H. Evans, Chief, Division of Insular Stations, Office of Experiment Stations. STATION STAFF J. M. Westgate, Agronomist in Charge, W. T. Pope, Horticulturist. H. L. Chung, Specialist in Tropical Agronomy. J. C. Rippebton, Chemist. R. A. Goff, In Charge of Glenwood Substation and Exten- sion Agent for the Island of Hawaii. Mabel Greene, Boys' and Girls' Club Leader. HAWAII AGR UNITED ICULTURAL EXPI HONOLULU, RAWi Under the supervision of STATES DEPARTMENT OF :riment station hi the AGRICULTURE 53 BULLETIN No. Washington, D. C. July 11, 1924 THE HAWAIIAN TREE FERN AS A COMMERCIAL SOURCE OF STARCH J. C. Ripperton, Chemist CONTENTS Page Introduction 1 Botanical description 2 Occurrence in Hawaii 3 Studies of methods of propagation Page Chemical composition of the core 9 Physical properties of tree-fern starches 10 Starch making from the tree fern 13 and growth 3 I Summary 15 INTRODUCTION Many generations ago the natives of Hawaii discovered the value of the tree fern as a source of food. They found that they could use the tree fern in place of the taro and the sweet potato, which constituted their favorite and staple food crops; and, likewise, that they could live indefinitely upon a diet of tree fern and wild game when they were defeated in battle and driven from the seashore to the mountains. Usually, they stripped the trunk of the bark 1 and baked the starchy core in an underground oven. It is not un- likely that the natives obtained starch from the tree fern, since they were familiar with the art of extracting it from the arrowroot. Many attempts have been made within recent years to produce tree-fern starch on a commercial scale. None of these proved suc- cessful, however, due to insufficient capital for the proper develop- ment of the product, until 1920 when tree-fern starch was success- fully manufactured and appeared on the local markets in a form suitable for use as food and for laundry purposes. Although some feared that the new industry would soon destroy the beautiful tree-fern forests, the Hawaii Experiment Station re- ceived many requests to aid in developing it. To satisfy those who looked unfavorably upon the industry, the station made a pre- liminary investigation to determine the effect on the forests and on water conservation of cutting over tree-fern areas. As a result of 1 In this publication the word " bark " is used to describe all that portion of the tree- fern trunk except the central starch-containing core. 94678—24 1 2 BULLETIN 53, HAWAII EXPERIMENT STATION the investigation, it was found that even a considerable thinning of the ferns for starch production is not noticeable. The tree fern falls to the ground of its own accord^ or is easily pushed over, upon reaching maturity, and since only the mature trees are utilized for starch making, relatively few trees per acre would be cut for this purpose. A rather extended program of work was therefore outlined, (1) to determine the feasibility of planting tree ferns on cut-over areas for the establishment of permanent-producing areas, and (2) to make a study of the properties and uses of tree-fern starch. When it was found that the rate of growth of the tree fern is too slow to make it commercially practicable to replant the fern and the necessity of building roads and fences to get the necessary raw material be- came apparent, hopes were abandoned of establishing the industry on a large and permanent basis in Hawaii. It is not unlikely, how- ever, that the industry might be made a permanent one under the economic conditions existing in other tropical countries in which cer- tain species of the tree fern are indigenous. This bulletin, reporting the results of certain observations, to- gether with data on the Hawaiian tree fern as they apply to its use as a source of starch, has been prepared because of the scientific interest which the industry has aroused and because of the potential importance of the tree as an emergency crop for the island popula- tion in case of interruption of shipping. BOTANICAL DESCRIPTION Rock 2 lists eight species of Cibotium, two occurring in Guatemala, one in southern Mexico, one in the monsoon districts of east Asia, one in the Philippines, and three that are peculiar to the Hawaiian Islands. The following botanical description of the two most im- portant species in Hawaii may be of interest. 3 Cibotium menziesii. — * * * Stipes green, stout, with a ventral and two lateral furrows, tuberculate and shaggy at the base with a straightish and long brownish-yellow glossy pulu which changes higher up into stiff, long blackish hair, and as such often covers the entire stipes; frond with stipes 18 to 36 dcm. or more long and 9 to 15 dcm. or more broad, pyramidal-oblong, coriaceous, naked underneath or sometimes with minute furf uraceous dots ; the rhachis asperous with scattering tubercles : pinnae with a stipe of 25 to 50 mm., oblong, 4.5 to 7.5 dcm. long, bearing 18 to 24 pairs of free pinnules besides the pinna- tifid apex; most pinnules shortly stipitate, linear lanceolate, acute, cut half- way or more, often to the rhachis at the base, into oblong rounded or entire segments, which are separated by broad sinuses; veinlets very prominent, simple or forked; sori 8 to 14 on a lobe, also fringing the sinus. Involucre corneous, large, a little more than 1 mm. to nearly 3 mm. in width, the outer valve fornicate and large, the inner flat and narrower. C. chamissoi. — * * * Stipes 12 to 24 dcm., brownish, smooth, clothed at the base with a pale fawn-colored lustreless, matted or cobwebby pulu, fur- f uraceous or naked above; frond 12 to 24 dcm. long, chartaceous, the under face green or dull glaucous and generally covered with a pale cobwebby pubes- cence ; lowest pinnse 4.5 to 7.5 dcm. long, with 24 to 28 pairs of pinnules, these shortly stipitate, linear lanceolate 12.5 to 15 cm. by 16 to 20 mm. acute, the lower ones cut to near the rhachis into oblong, straightish, rather obtuse seg- ments with narrow sinuses, the basal segments entire and not deflected; veinlets little prominent; sori 8 to 14 to a segment, the involucre small about 1 mm. wide, chartaceous. 2 Rock, J. F. The Indigenous trees of the Hawaiian Islands, p. 89. 5 Rock, J. F. The indigenous trees of the Hawaiian Islands, pp. 91-93. HAWAIIAN TREE FERN AS A SOURCE OF STARCH 6 Hillebrand 4 lists another species, C. glaucum, but states that it is rather rare. Clbotlum chamissoi, or " Hapu " as it is popularly known, is easily recognized by its yellow pulu or hair and its comparatively short, stocky growth. In many forests it constitutes more than 50 per cent of the entire tree-fern growth. The trunk sometimes attains a height of 16 feet, but usually does not exceed 10 feet. The diameter of the trunk is usually 8 to 12 inches. (PL I, fig. 1.) C. menzitsii. or ;; Hapu Iii," is distinguished by the brownish or blackish pulu which covers the stipes and fills the crown. Occasionally it attains a height of 40 feet and frequently a diameter of 3 feet. (PI. I, fig. 2.) Another species, commonly known as the " Men," is easily recognized in the Hilo district by its very slender trunk, smaller fronds, dull, lusterless, rather scant, yellowish-brown pulu, and nearly naked stipes. (PL I, fig. 3.) The "Amau" (Sadleria cyatheoides). although a different genus of tree fern, is also of interest since it is exceedingly common in occurrence and has a starchy core. OCCURRENCE IN HAWAII The tree fern is found in all parts of Hawaii where there is an annual rainfall of 100 inches or more. It grows on nearly all the mountains, but occurs in dense forests only on the islands of Kauai and Hawaii. On Kauai the forests are too inaccessible to be of importance for starch production, but on Hawaii they occur in al- most unbroken stretches from sea level to an elevation of 6,000 feet or more. These long stretches are reached both by rail and auto- mobile roads running from Hilo to the Kilauea Volcano. The wind- ward slopes of the Manna Kea and Mauna Loa Mountains are one continuous tree-fern forest, the belt extending from the Puna dis- trict to the Hamakua district being about 10 miles wide and 40 miles long. In general it is estimated that there are 400.000 acres of tree- fern forests on the island of Hawaii alone. A very large part of this area is within the forest reserve or on Government -owned lands, from which it is illegal to cut the tree fern. There are, however, many thousands of acres of privately owned land on the island of Hawaii, which in its present state is of little value because the dense growth of tree ferns unfits it for pasture and the heavy cover- ing of leaf mold keeps the soil too wet for general agricultural pur- poses. The owners of these lands regard the tree fern as a pest and would welcome any means of removing it. STUDIES OF METHODS OF PROPAGATION AND GROWTH It was felt that before the tree-fern starch industry could be established on a large scale in Hawaii some feasible method must be found for providing a permanent source of the raw material. Many of the tree-fern areas are hard to reach, and it is almost im- possible to secure the raw material at any great distance from the established roads because of the rough topography and the heavy leaf mold. Moreover, the cost of getting the raw material from the depths of the forest, when the supply adjacent to the highway became exhausted, would reach a prohibitive figure. * Hillebrand. Wm. Flora of the Hawaiian Islands, p. 347. 4 BULLETIN 53, HAWAII EXPERIMENT STATION The possibility- of securing a large tract of tree-fern forest for the establishment of a permanent starch-producing area was therefore considered. Such an area would have to be strongly fenced to keep out cattle and hogs, and roads would have to be built at intervals, with subsidiary donkey trails, to permit of the tree-fern logs being carried out. The purchase c'f such a tract would be feasible if it could be shown that the tree-fern growth could be successfully main- tained by natural methods of propagation, or by planting cuttings from different parts of the fern, thus assuring a permanent and in- creasing supply of readily accessible raw material. It was therefore decided to learn whether parts of the tree fern could be successfully planted on cut-over areas, and whether the rate of growth would be sufficiently rapid to justify the cost entailed in establishing and maintaining such an area. The first step in the solution of the problem seemed to be the acquisition of a detailed knowledge of the character and habits of the tree fern. Since the literature was found to contain only certain botanical descriptions, a series of observations was begun of its nat- ural methods of propagation and growth. NATURAL METHODS OF PROPAGATION The tree fern reproduces itself in two general ways — (1) by spore germination (PL II. fig. 1) and (2) by lateral shoot development (PL II, fig. 2). The spores are borne on the underside of the fronds. Old tree-fern trunks and moss furnish excellent seed beds for spore germination, and land that is undisturbed by cattle and wild hogs is frequently literally covered with the tiny ferns. Their growth is. however, very slow. Propagation by lateral-shoot development on the trunks of ma- ture trees is much more rapid than by spore development. One to three such shoots are found about each tree in varying stages of development in the native forests. These increase to as many as 15 when the tree fern has been injured, or the soil about it trampled upon by animals. The shoots begin to develop when the parent tree ceases to grow or falls to the ground. It is not uncommon to find, even on young, vigorous trees, one or two shoots which are almost as large as the parent. The shoot soon establishes its own root system and in a short time becomes independent of the original fern. The crown continues to grow after the tree has fallen, the fronds gradually turn in a vertical direction, another root system is established at the new base, and growth proceeds almost unin- terruptedly. PLANTING TREE FERNS It is a well-established fact that tree ferns can be successfully propagated from crowns and lateral shoots which spring from the trunk, as well as from spores, but since the entire inner core of the trunk is used for starch, planting would not be feasible if sections of the trunk were necessary for the production of new growth: and it was not known whether the undeveloped lateral shoots which are found on the average tree would develop if re- moved from the trunk. Three experimental plats were therefore Bui. 53. Hawaii Agr. Expt. Station Plate iJ,f^„ o* • CO D — hj v-> CO < S _l _J 5 d £ W n 5 CO -I z > _J £ DC < > UJ O O > £** I- :£ i^w i w 5 H o D O c El;I. 53, Hawaii Agr. Expt. Station Plate II Fig. !. — Natural Propagation from Spores. An Old Tree-Fern Trunk Literally Covered with Tiny Tree Ferns. The Spores Are Found on the Underside of the Mature Fronds Fig. 2. — Crowns and Lateral Shoots Used for Propagation Experiments HAWAIIAN TREE FERN AS A SOURCE OF STARCH 5 established at different altitudes for the purpose of determining the feasibility of planting the tree fern for starch production. These experiments were carried on at the following places: (1) Volcano plat (3,500 feet elevation) on the McKenzie ranch, at 29 Miles, Volcano Road. This plat is adjacent to the forest re- serve on windward Hawaii. It is used as a pasture and contains a scattered growth of tree ferns. The experiment was made here to de- termine whether the tree fern could be successfully planted on areas that had been denuded of their original growth and on which there was no shade. (2) Glenwood plat (2,000 feet elevation), located 2 miles above Glenwood on the Volcano Road. This plat was selected because it is in the midst of an excellent tree-fern growth and is easy of access. (PI. Ill, fig. i.) (3) Mill plat (2.200 feet elevation), located 4 miles north of the Volcano Road at 18 Miles. This plat is in an area now being cut over for starch production. It represents ideal conditions as to shade, soil, and the like under which the tree fern would be planted. The plantings included crowns, and large, small, and medium sized lateral shoots of each of the several species. Comparative plantings were made to determine the effect on growth of varying the length of the starch core attached to the cuttings, drying the cut surfaces before planting, planting at different depths, in different kinds of soil, and at various altitudes. Since this work was dis- continued before the effect of these various factors on the growth of the tree fern was determined, the general conclusions of the experiment only are given. Under true forest conditions, such as existed at the Mill plat, all crowns and lateral shoots were successfully propagated, regardless of variety, size, method of planting, or length of attached starch core (PL III, fig. 2). Fully 70 per cent of the small, undeveloped shoots, which had been detached from the starch core of the parent tree, grew when planted. The plantings made in the open pasture on the Volcano plat started vigorous growth during the cool, moist winter months, but died during the dry, hot summer months. Other plantings, made in a shaded area closely adjacent, lived and gave results similar to those obtained at the Mill plat. In brief, it is concluded that, under true forest conditions, an average of three successful plantings or sets can be made from each felled tree fern ; and that the starch core of the parent tree need not be attached to the plantings or sets. It seems evident, therefore, that a cut-over area could be successfully replanted without interference with starch production, and that the density of growth could be gradually increased to a maximum. METHODS OF GROWTH In order to determine how fast the tree-fern trunk grows, it was necessary to learn the nature and method of its growth. For this purpose a study was made of the lateral growth of the trunk, the vertical growth within the trunk, growth at the base of the trunk, and growth at the top of the trunk. BULLETIN 53, HAWAII EXPERIMENT STATION Fig. 1. — Graph showing the spiral arrangement of tree-fern fronds. The lateral growth of the starch core would seem to be negligible since it is immedi- ately surrounded by a very hard, brittle covering meas- uring one- fourth to one-half inch__in thickness, and appar- ently is incapable of further growth or expansion. In five out of eight trees the starch core was found to be appre- ciably larger at the top than at the bottom of the trunk, which shows that as the tree fern grows from a small lat- eral shoot or spore to a large size, the core grows corre- spondingly larger at the top only ; and that the part first formed does not increase in diameter. Many trees are found in which the starch core tapers almost to a point at the base. Trees grown from very large lateral shoots show little tapering of the core, and trees resulting from the turned-up crown of a large fallen tree show none at all. It is true that the . gross diameter of the tree, especially of the species Cibo- tium menziesii, increases with the growth of the tree, but this increase is due entirely to the increase in abundance of the air-feeding roots mak- ing up the outer bark of the tree. That there is no vertical growth within the trunk, except at its apex, is shown by the fact that the vertical distance between frond pits on the same spiral is the same regardless of whether the measurements are made at the bottom or at the top of the trunk. This vertical distance would show a gradual decrease from HAWAIIAN TREE FERN AS A SOURCE OF STARCH 7 bottom to top if there were an appreciable growth within the trunk. Many species of monocotyledonous trees show an increase of growth at the base of the trunk, as is frequently evidenced by the dead roots which are found covering 2 or 3 feet of the base of the tree trunk. That this is not the case with the tree fern is shown by the uniform occurrence of pits from which the fronds once pro- truded along the entire length of the trunk. It would seem, there- fore, that such growth in the tree-fern trunk is negligible. The method of growth at the top of the trunk, however, is very striking. In the early spring new fronds or leaves develop, a com- plete circle of new fronds averaging five in number emerging from the heart of the crown at practically the same stage of development (PL IV, fig. 1). These fronds attain full size within about three months after the time of their appearance and remain on the tree from 18 to 2-i months. An occasional immature frond can be found on the tree after the first five fronds develop. METHOD OF DETERMINING THE RATE OF GROWTH "When the outer bark is stripped from the tree fern, or when an old log in which the starch core has decayed is split open, the fronds are observed to be arranged in definite spirals. Figure 1 represents graphically the surface of a tree-fern log from which the outer bark has been removed to disclose the location of the frond attachments, regarded as a hollow cylinder cut lengthwise and laid out flat. The ellipses represent frond pits or the openings in the bark from which the fronds protruded. A number of different spirals are apparent in the diagram. Among the more obvious are those parallel to the lines o-d% n-d\ and m^a', respectively. The arrangement of fronds was found to be the same on a large number of stripped tree-fern logs, and the spiral combination of 3, 5, and 8 fronds could readily be counted. The only difference noted was in the direction of growth of each spiral, which was clockwise on some trees and counterclockwise on others. Since probably a circle averaging five fronds is developed each spring, each of the five spirals would seem to be annually represented by one frond. If this were true, or if the average yearly number of fronds per tree showed little variation in the different trees, it would be necessary only to measure the vertical distance between two frond pits on one of the five spirals to determine the annual growth of the tree (PI. IV, fig. 2). To determine the correctness of this assumption, the fronds of a number of trees growing at different altitudes were marked with copper tags at each frond-setting period and the trees were visited once a year. Usually, it was not difficult to distinguish the new fronds. The species Hapu (Cibotium chamissoi) . which is the only one used to any extent for starch production, was selected for this study. 8 BULLETIN 53, HAWAII EXPERIMENT STATION Table 1 gives the annual number of fronds set per tree during four successive years : Table 1. — Annual number of fronds set per tree {Cibotium chamissoi) during four successive years Tree No. Number of new fronds de- veloped Tree Xo. Number of new fronds de- veloped 1920 1921 1922 1923 Aver- age 1920 1921 j 1922 1923 Aver- age Elevation, S, 500 feet 1. 5 6 4 7 5 6 5 3 8 7 .5 5 7 5 7 5 8 5 5 5 7 5 5 7 5 4 4.33 7 1. ? Elevation, 2 ,200 feet 5 5 5 5 6 7 6 4 6 6 5 5 .=; 2.1 2 3 4. 67 3. 6. 33 4. 6 4 6 1 5 5 7 5 ' 3 6 5 5 5. 25 4 6 4. 75 8 5.67 6.33 5 5.25 5. ?: 8 5 5. 5 6 5.75 5 4. 5 8. 5 4 5.5 6 4.66 i Average 5 5. 25 A verage 5.52 j While the data given in Table 1 do not conform to the theory that each of the five spirals is represented annually by a frond, they do verify the observation that the average annual number is a fraction over five, and show that this number is practically the same regardless of altitude. It would seem, therefore, that the rate of growth of the tree fern could be closely determined simply by measuring the vertical dis- tance between any two successive frond pits of the same spiral and by multiplying this distance by 5.39 and dividing by the spiral num- ber, that is by 3, 5, or 8. For example, if this distance on the 5 spiral averaged 4 inches, the yearly vertical growth would be 5 39 -^ — , or 4.31 inches; or, if measured on the 3 spiral, it was 2.4 4X 5.39 inches, the yearly vertical growth would be 2.4 X -^ — , or 4.31 inches. o In order to ascertain the average vertical distance between the frond pits, eight trees {Cibotium chamissoi) , growing near the Mill plat, were stripped of their bark and measurements were made of their diameter, length, weight, and the like. These data, together with the rate of growth as computed by the foregoing method, are given in Table 2 : Bui. 53, Hawaii Agr. Expt. Station Plate III fFiG. I. — Tree Ferns Grown from Crowns at Glenwood Substation (Hawaii i Fig. 2. — Tree-Fern Hedge Grown from Crowns Bui. 53, Hawaii Agr. Expt. Station Plate IV * c » OC W * < CO ^ CQ 5 O uj D I o: u o I- UJ CO — DI-)U. Oh < HAWAIIAN TREE FERN AS A SOURCE OF STARCH Table 2. — The rate of growth of the tree fern and amount of starch core it produces annually l Tree No. (a) Weight of starch core (b) Length of starch core Diameter of starch core (C) Vertical distance between successive fronds (d) Annual vertical growth 2 (e) Esti- mated length of time required for growth 3 (0 Starch core produced annually 4 Top Bottom 5-spiral 1 2.. 3 Pounds 34.5 64 76 87 58.5 72.5 15 11 Inches 66 65 80 128 72 101 33 41 Inches 5.4 7.3 5.4 4.8 6.1 5.7 4.1 3.8 Inches 3.8 6.7 6.1 5.4 4.1 4.5 4.5 2.43 Inches 4.12 4.06 4 4.57 4.5 3.49 4.13 3.41 Inches 4.44 4.38 4.32 4.93 4.85 3.76 4.45 3.68 Years 14.9 14.8 18.5 26 14.8 26.9 7.4 11.1 Pounds 2.31 4.32 4.11 4 5 6 3.35 3.95 2.7 2.03 8 .99 4.35 2.97 1 The 5-spiral measurements are used in these computations because the coefficient of error in measure ment is less than in case of the 3-spiral. * c multiplied by 5.39, divided by 5. 3 b divided by d. 4 a divided by e. Table 2 shows that the average vertical growth of the tree fern is fairly uniform and averages 4.35 inches annually, and that the an- nual production of starch is subject to wide variations due to differ- ence in diameter of the starch core. The annual increment in starch core averages 2.97 pounds. Sections of the bark of a large number of trees growing at vary- ing altitudes were slabbed off to permit of measuring the distance between successive fronds for a distance of about 1 foot. The data so secured, while only approximate, would seem to show that the rate of growth of the tree fern is practically constant between sea level and an altitude of 3,500 feet. Measurements of other species of tree ferns showed that they make about the same rate of growth as does Cibotium chamissoi. The very slow rate of growth of the tree fern brings into very serious question the feasibility of planting cuttings of different parts for starch production. It would require at least 20 years to grow a tree fern of sufficient size to cut for starch, since it has been found unprofitable to cut trees having less than a 60-pound starch core. Twenty years is a rather prohibitive length of time, especially for a small industry, to wait for replanted areas to furnish new material. As a result of the investigations it was concluded that, while the tree fern can be successfully planted on cut-over areas, its rate of growth is too slow to make such a procedure advisable under present conditions. CHEMICAL COMPOSITION OF THE CORE Preparatory to analyzing the core the outer fibrous sheath and the hard inner bark were stripped from it. The yellowish-white core was then shredded and the nonreducing and reducing sugars were determined. The methods of analysis recommended by the Associa- 10 BULLETIN 53, HAWAII EXPERIMENT STATION tion of Official Agricultural Chemists 5 were employed in making" the rest of the analysis. The results are given in Table 3. which includes also, for purposes of comparison, the chemical composition of the potato and edible canna. both of which are grown for commercial starch production : Table 3. — Comparison of the chemical composition of the tree-fern core with that of the potato and edible canna tubers Constituent Tree fern ( C. Pntatn chamis- Folato soi) Edible Constituent Tree fern ( C. p t t Edible soi) Water. 73.39 .95 .06 1.04 78.30 2.20 .10 1.00 72.62 .98 .11 1.40 Fiber N-free extract.. .. 1.57 : 22.99 i .24 3.99 0.40 18.00 61 Protein 24 28 Fat Xonr educing sugars Reducing sugars Ash... __ From Table 3 it will be seen that in chemical composition the core of the tree fern is very similar to the ordinary tuber crops, and, as- might be expected, it contains rather large amounts of reducing sugars. In view of the peculiar manner of growth of the tree fern, it is very probable that the sugar content varies considerably in the different sections of the core as well as in different seasons. The tree-fern core is apparently equal to other starch crops as a stock feed. It is eaten readily by cattle and hogs, probably because of its high sugar content. The new, tender fronds form the most important part of the forage for stock pasturing in a tree-fern for- est ; and the heart of the trunk is also eaten when it can be reached. Were it not for the prohibitive amount of labor required to split open these trunks, the tree fern undoubtedly would have an impor- tant place among the stock feeds of the tree-fern forest districts. The core of the tree is not palatable as human food. It has a peculiar flavor and is rather fibrous. The tender, undeveloped fronds- are sometimes cooked as a vegetable, but the core as such has never become of practical importance. PHYSICAL PROPERTIES OF TREE-FERN STARCHES Samples of starch were prepared from the species Hapu (Cibotium chamissoi) . Men {Cibotium sp.), Hapu Iii or Heii (c7. menziesii) T and Amau (Sadleria cyatheoides) . 6 Microscopic examination and photomicrographs were then made of each. The characteristics of each starch as they appeared when magnified 220 diameters were found to be as follows : (1) Hapu (C. chamissoi). (PL V, fig. 1.) Size. 0.01 to 0.05 millimeter. Varying sizes in about equal proportion; shape, irregular oval, occasionally kidney-shaped, in large sizes, but circular disks, with a few truncated forms in the small sizes ; hilum, annular, with a few longitudinal rifts, depressed, usually central in small sizes ; rings, complete, very distinct, with a very pronounced ring in most of the larger granules about midway between the hilum and the outer edge; polarizer, very brilliant dark cross at central axis. 5 Methods of analysis of the Association of Official Agricultural Chemists. Sec. VII, p. 71. Revised to Nov. 1. 1919. Washington, D. C, 1920. 6 The individuals of each species selected for these samples were as nearly typical speci- mens as could be found. No botanical identification was made of them, since they exist in comparatively pure strains, and no difficulty was bad in differentiating between them. HAWAIIAN TREE FERN AS A SOURCE OF STARCH 11 Men (Cibotium sp.). (PI- V, fig. 2.) Size. 0.01 to n.Oo millimeter (Ion? axis) : shape, usually elongated disk in the large sizes, but round to oval in the small sizes, and a few truncated and kidney-shaped in all sizes; hilum, usually annular, depressed, eccentric; rings, complete, very distinct ; polarizer, distinct cross at hilum. (3) Hapu Iii or Heii (C. menziesii). (PI. VI, fig. 1.) Size, 0.01 to 0.04 millimeter : shape, round to oval, occasionally truncated and angular : hilum. annular, central, depressed ; rings, very pronounced, con- centric, complete : polarizer, very distinct cross at hilum. in Amau (Sadleria cyatheoides). (PI. VI. fig. 2.) Gore, mostly colloidal dextrins. with starch grains of extremely minute size. The starch can not be separated from the grated mass by sedimentation ; size, 0.01 to 0.03 millimeter: shape, round to oval: hilum. usually central, annular, depressed ; rings, distinct on large granules, and complete ; polarizer, well- marked cross at hilum. The illustrations and morphological descriptions show that the four starches possess many of the same characteristics, the chief differences being in size and shape. Considered from the standpoint of the physical characteristics of the starch granules, each of the four species could be used for starch production. As a matter of fact, the starch of Cibotium chamissoi is very much to be preferred to that of the other three species. The starch of both C. menziesii and Men, in addition to being of small diameter, contains such large quantities of dextrins and other colloidal matter as to make the separation of the starch difficult. The starch of Sadleria cyatheoides is manifestly unsuitable. VISCOSITY Since the viscosity curve of a starch, when transformed by boiling water into " soluble starch," is useful in showing its general proper- ties, determination was made of the viscosity of tree-fern starch, and likewise of corn and arrowroot starches for purposes of compari- son. The method of procedure was as follows: Varying amounts of starch were weighed into 100 cubic centimeter flasks graduated at 80° C. with 10 cubic centimeters of cold water. Boiling water was added with vigorous shaking, and the flasks were made up to the mark at 80° C. with hot water. The flasks were then placed in boil- ing water for one hour without agitation, after which they were quickly cooled to 80° C. with as little agitation as possible, and the viscosity was determined with a Saybolt universal viscosimeter at that temperature. Duplicate determinations by the above arbitrary procedure agreed with fair accuracy. Any variation in procedure, however, caused very large differences in the result. For example, vigorous shaking during cooking decreased the viscosity as much as 50 per cent. Variations in temperature and time of cooking also caused appreciable deviations. The results are graphically given in Figure 2. The concentration of starch solution necessary to produce a defi- nite hydrogel when cooled was determined by pouring 10 cubic cen- timeters of the hot starch solution used for determining the viscosity into test tubes one-half inch in diameter. The tubes were placed in water at about 18° C. and allowed to remain unagitated for one hour. They were then inverted. The minimum concentration neces- sary to keep the mass from flowing down the inverted tube was termed its ; * gelling strength." The gelling strength of cornstarch, tree-fern starch, and arrowroot starch was found to be 4.25, 5, and per cent, respectively. 12 BULLETIN 53, HAWAII EXPERIMENT STATION The above data on viscosity and gelling strength bring out a num- ber of important differences among the starches. The concentra- tions necessary to cause any appreciable increase in viscosity were 1.5 per cent of arrowroot, 2 per cent of tree fern, and 3 per cent of cornstarch. Above 3 per cent the curve for cornstarch shows a very sharp increase. A 1-gram increment causes an increase in viscosity of 1 to 6, whereas a like increment of tree fern and arrowroot over the concentrations of 2 and 3 per cent, respectively, causes increases of 1 to 3. As the concentration increases, the tree fern and corn- starch curves practically coincide. The arrowroot curve, although -