IAWA Journal. Vol.

19 (3), 1998: 241-264


Elisabeth A. Wheeler 2 & Pieter Baas 3
Wqod identifi cation is of value in a varie ty of co ntexts - commercial,
foren sic, archaeological and paleont ological. Thi s paper reviews the
ba s~c s of wood identification, including the problems associ ated with
dif~erent types of materials, lists co mmonly used micro scopic and mac­
ros opic features and recent wood anatomical atlases , discu sses type s
of eys (sy noptic, dichotomous, aud multiple entry), and outl ines some
work on computer-assisted wood identi fication.

Kel words: Wood identifi cation, keys, computer-aided wood identifi­
ca don .
In the lastl decad e. co mputerized keys have made the identification of uncommon
woods easier. Nevertheless, 'traditional' meth ods remain important because often they
are more efficient and convenient, particul arl y for com mon commercial woods. This
paper discusses some applications of wood identific ation, logistics, the macroscopic
and micrl coPic featnres used for wood identification (particularly for hardwood s.
i.e ., dicot ledono us angiosperms), identi fication procedures, some recent co mputer­
aided woo identification projects, and the need for additional work.
Proper pr9cessing of wood, especially drying, depend s upon correct species identifi ­
cation becfu se different species and species groups requ ire different protocols, When
probl ems arise during wood processin g (drying, machining, or finishing), one of the
first que stibns asked is whether the wood was correctly identified.
Cusro s officials need to know whether logs. timber s, or wood produ cts are cor­
rectly labe ed so that tariffs can be properly assessed and trade regnlations enforced.
The Intern tional Timber Trade Organization (ITTO ) has propo sed limiting interna­
tional trad to timber that is cut from sustainably managed concessions. Verifying the
source an identity of timber in order to enforce bans on trade in woods of endan­
gered spec'es will require wood identification skills (Baas 1994).
1) This pai l r is a modified version of a paper published earlier: Wheeler, E.A. & P. Baas.
1996. W od identification. Sci. Annals, Dept. Forestry and Natural Environment, Aristote­

lian Uni ersity, ThessaIoniki, Greece 37/ 1994: 75-104.
2) Depart ent of Wood and Paper Science, North Carolina State University, Raleigh,
NC 276 5-8005, U.S.A.
3) Rijksher arium/ Hortus Botanicus, P. O. Box 9514,2300 RA Leiden, The Netherlands.


IAWA Journal, Voi. 19 (3),1998

One means of conserving tropical forests is ensuring that any tree cut is properly
used, and its full value realized, thereby reducing waste, and resulting in fewer trees
being cut to meet demand. In tropical forests with a high species diversity, identifica­
tion is an integral part of timber grading, and often difficult because the anatomy of
many species is not well known. Oteng-Amoako (1992b) discussed the problems of
timber identification in Papua New Guinea (pNG) and the financial consequences of
the lack of personnel able to do identification.
When restoring historically significant wooden structures, restorers prefer to use
the same type of wood as used originally, and that requires identification of whatever
original wooden fragments remain. The value of a wooden object often is affected by
the type of wood used. Art historians use evidence from wooden frames and panels to
establish the authenticity and provenance of an art work .
Wood identification has forensic value, e.g .. determining whether wooden frag­
ments at the scene of a crime match those taken from the clothing or a vehicle belong­
ing to a suspect. In the U.S ., the analysis of the wooden ladder used in the infamous
Lindbergh kidnapping of the 1930s provided crucial evidence for the conviction of
the alleged kidnapper (Miller 1993).
Different tree species and different wood anatomies characterize different climates
(Wheeler & Baas 1991). Therefore, identification of ancient woods helps to recon­
struct ancient ecosystems and to document climate change. Knowing the affinities of
fossil woods can help determine the age of a geologic formation (Stone et al. 1987).
Paleontologists are interested in knowing what trees were present when dinosaurs
lived, and in what type of vegetation early primates and horninids evolved . Geologi­
cally ancient woods provide information helpful for explaining present-day distribu­
tions of plants , the history of particular families and genera, and the past distribution
and diversity of woody plants . Woody remains provide archaeologists with informa­
tion useful for reconstructing trade routes. At Pompeii and Herculaneum, identifica­
tion of woody remains helped to reconstruct the land use patterns and gardening prac­
tices of Italy in A.D. 79 (Jashemski 1990).

Levels of diffi culty
Questions of the type "Is this wood teak?" or "Is this wood Acer (maple) or Betula
(birch)?" are easier to answer than "What is this wood from northern Europe?" and
this question is easier to answer than "What is this wood? I don 't know what conti­
nent it came from."
The size of the pool of possible matches for an unknown affects the ease and also
the approach for determining a wood's identification. There are significantly fewer
tree species in the north temperate region than in the tropics. Identifying wood from a
sawmill that only uses trees from nearby is easier to do in the north temperate regions
than in the tropics. Identifying a commercially important wood whose geographic
source is known is easier than identifying woods whose geographic source is un­
known .

ray size relativ e t vessel diam eter. and depo sit s in ves­ sels and ttleir colour. and Africa (Sco tt & Wheeler ' 982). THE BASIS OF MACROSCOPIC AND MICROSCOPIC WOOD IDENTIFICATION Ma crosc ic features Macro copic features include physical features suc h as colour and lustre. ax ial parenchyma arrangement and abundance . and pulp. Some woods hav e distinctive combinations of macroscopic anatomical features so that they <life readily identified with a handlens.circles within which the numb er of vessels can be counter). Cell sizes are alter ed in CharC~ J lified woods. Cretaceous and early Tertiary dicotyledonou s fossil woods can be difficult tf. Ve small fragment s ma y show relati vely few features: veneers may not pro­ vide eno gh of all the surfaces necessary to reveal diagnostic characteri stic s.1111 in diameter are easily seen with a handlens. South America. so that they need to be compared with thou sand s of specie s of extant wood s. decay ed and petrified woods would not have the original colour or density. Types of material The type of material affect s the ease of identifi cation . char­ coal. and I mm? or 10 rnm. De cayed wood. vessels more than 100 J. vessels mpre than 200 11m are easily seen with the unaided eye . Determining vessel diameter and density (number per mrn. so quantitativ e features onl y can be used with extreme cau­ tion . particleb 9ard. Fos sil wood s (approximately 44 million years old ) from the Clarnb Nut Bed s in the western United States have nearest living relatives that today ocdur in eastern N0I1h America. to Ivarious types of composite wood products such as plywood. presence or absenc e of storied structure. Categories of vessel diameter and density (number per unit area) are also used . sawdu st. including porosity.A review 24 3 Geolo&ically ancient dicotyledonous woods may have near est living relatives that now occur on different continents. Woods treated with pre­ servatives or stained or finished would not have the original colour. fibreb oard. Baas - Wood identification . Other features e presence of included phloem .. vessel arrangem nt and grouping. ray height. temperate northern hemi­ sphere species of the elm family (Ulmaceae) have a distinctive combina tion of ring . and ana­ tomic al f atures visibl e with a handlens or unaided eye. large pieces of lumber and solid piece s of wood . For example. and to section to pro vide cross. 50 urn increments on a I mm lonk line .g . or poorly preserved petrified wood is difficult to sec tion and prepare and may not show 011 anatomical features originally present in the wood . The material needing identi­ fication i~ variable. Some wood s are unalt ered and retain the origi 1al charact eristi cs . abundance of tyloses. Asia. Unal­ tered pie es of wood that are larg e enough to provide an end-view for examination with a 10 handl ens .or per 10 rnrn-) is helped by using transparent overlays ~ith scale lines mark ed in some set increment (e. and splinters. radial.Wh eeler &j. ranging from logs . and tangential sec tions for micro scopic study are mo st likely to rev eal a diagnostic combination of features. 'identify' becau se they may repr esent extinct genera or species that have not been describ ed befor e. microscopi c and macroscopi c.

Multi­ ple entry keys (see below) generally use no more than seven descriptors for wood colour. yellow. 11. There is a variety of other supplementary features.000 species of woods for fluorescence.). Woods have varying shades of colour and combinations of different colours . spiral. However. Some woods with similar anatomy dif­ fer in the type of residue left after burning a splinter in still air (Dadswell & Burnell 1932.. pink ivory (Rhamnus zeyheri Sand. odour is ephemeral. 12). Classic examples of woods with distinctive colour include purpleheart (Peltogyne spp. or has streaks is important for macroscopic identification of untreated wood.). 1990). wavy). Chemical tests seem most useful for conifers (softwoods) because conifers have few useful macroscopic fea­ tures (presence or absence of resin canals. grain (straight. if the sources of variation in heartwood colour are understood.. rays of two dis­ tinct sizes. and latewood vessels in radial to dendritic patterns. colour.244 fAWA Journal. Whether or not a freshly smoothed surface of wood fluoresces and the colour it fluoresces can be diagnostic. whether or not the heartwood is dark or light. macroscopic features. even I uneven). and the heartwood colour of freshly felled trees often differs from dried wood samples (IAWA Committee 1989) . Describing colour can be difficult as individual perceptions of colour differ. and more species can be distinguished using microscopic characters than using . as can be odours and chemical tests. and woods may acquire smells from their surroundings. the physical property of specific gravity or density is used in many keys. white to grey. Fig . Avella et al. in shades of red. and Jane 1970). Refer­ ence to colour charts used for soil classification may provide some consistency while describing colour (Vetter et al. IAWA Committee 1989). Unfortunately. As recognized long ago by Bailey (1917). brown. Macroscopic features such as texture (coarse I fine. However.e.). or other colours . g. 19 (3). interlocked. Vochysiaceae) with a high aluminum content react with a distinct blue colour. and the yellow-coloured boxwood (Buxus spp. or to which family it is likely to belong . 'cigar-box' cedars (CedreZa spp. 3).). many genera have similar macroscopic fea­ tures. and odour). wood colour changes with time and exposure to light. but this has not yet been widely applied. There are other chemical tests useful for distinguishing between closely related woods (see examples in Panshin & DeZeeuw 1980. The chrome azurol-S test indicates the presence of aluminum in a wood (Kukachka & Miller 1980) and woods of families (e. but they can be very helpful as complementary diagnostic features . Most Quercus species (oak) are also distinct (ring-porous. i.). Nonetheless. commercial timbers must be handled quickly and often in large vol­ umes. raspberry jam wood (Acacia acuminata Benth. so that it is only practical to use the most obvious characters in their identifica­ tion. Although not strictly a macroscopic feature. and stink wood (Ocotea bullata E. Vol.). figure (many types) and lustre have poorly defined catego ­ ries and require experience before they can be effectively applied in wood identifica­ tion. whether the transition from early wood to latewood is abrupt or gradual. 1998 porosity and latewood vessels arranged in wavy tangential bands (ulmiform band s) (Fig. Macroscopic features often can be used to quickly establish whether a wood is correctly labeled. (1989) surveyed over] 0. Sandalwood (Santalum album L. Meyer) have distinctive odours (Jane 1970).

and to help reduce ambiguity in the description of anatomical features. . if storied . The CSI 0 macro key has 50 features (llic 1990). vessel groupings. Oteng-Arnoako 1992a). Nardi Berti & Edlmann Abbate 1988. or absence of septate fibres. type of fibre wall pitting. Macro scopic keys typically have fewer features than micro scopic keys . presence or absence of helical thickeni gs. and. til cells. tangential diameter of vessel lumina. and their cel­ lular loc~tion (in ray parenchyma and / or axial parenchyma). presence or absence of vestured pits . or cambial variants (e. The anatomital features in this list are: growth rings (distinct.. intervessel pit arrangement and size. rays per millimetre: pres­ ence or bsence of storied structure. but the larger the number of features to choose from . upright . genus. axial paren­ chyma ~ stribution . or other crystal types. presence or absence of tyloses or other deposits in vessels . vessels per square millimetre. silica bodies. Mi croscEPic f eatures The I{\WA List of Microscopic Features for Hardwood Identification (lAWA Com­ mittee 1989) was intended to be a concise list of features useful for hardwood identi­ fication. and to a lesser extent also for systematic descriptions.A review 245 macroscppi c ones. type of vessel-ray pitting. druses . whether rays are of two distinct sizes. and square cells): presence or absence of sheath cells . grain. presence or ab­ sence 0 prismatic crystals. Some recent ~orkS have included as part of the wood descriptions a list ofIAWA feature number that apply (e. presence or absence of vascular or vasicentric tracheids. to a lesser extent. the more likely it is that a combinrion of characters diagnostic of a particular family. perforated ray cells.Wheeler ~ Baas - Wood identification . fibre length. which elements are storied. or specie s will be observe . presenc1. included phloem) . while the CSIRO microscopic key has 91 1natomical features (Hie 1987). laticifers or tanni iferous tubes . fibre wall thickness. IAWA Committee (989). but was int~nded to serve as a framework for descriptions of woods for databases for wood identification.g . indistinct. texture ' j'nd. cellular compo si­ tion of ys (procumbent. porosity (ring-po ous or diffuse-porous). number of cells per axial parenchyma strand . A gi n micro scopic feature is often not any more useful or reliable (consistent from on sample to another of a particular taxon) than anyone macroscopic charac­ ter. easily visible end-grain characters than were distin­ guished y a more detailed anatomi cal examination. This IAWAlist (163 anatomical. mean vessel element length . vessel arrangement. outline of soli­ tary ves els: perforation plate type. presence or absen e of aggregate rays. absent).. Different features are useful within and between different groups . or disjunctive ray cell s. ray width. The timber dealer recognized fewer types on the basis of colour. ro one list offeatures can be expected to include all diagnostic features for all woods (Brazier 1976. The anatomical features visible and useful macroscppically are equally valuable for microscopic wood identification. 1998) on the trees of Southeast Asia provides concise descriptions of woods of approxim ately 300 genera by using a large subset of IAWA f~ature numbers . Sudo (l994) asked a timber dealer to sort a batch of Sarawakan timber to type. intercellul ar canals (radial or axial). pre s­ ence or bsence of oil /mucilage cells. g. A recen PROSEA volume (Sosef et al. 58 miscellaneous feature s) is not all inclusive.

246 IAWA Journal. These figures show most of the anatomical features that have been used in multiple entry keys. Vol.ncsu.edu/unity/lo ckers/class/wp s20 200 2/hw/hwanat. 19 (3). lllu strations and definitions of some features are available on the World Wide Web at a site providing supplementary information for a course in wood structure and properties: [uri of http://www2.html] . 1998 Figures 1-42 illustrate selected anatomical feature s.

Lithocarpus sp. 2). yet wood feature s that aretonSidered primitive in the Baileyan sense are uncommon in the dicotyledons as a wh le. Philips (1948) IFr I Fig. If the characterization of species. ­ 2: SOlit~ y ve ssels with an angular outline.7: Vessel s and parenc yma in tangential 'festoons' .3. diffuse-porons.5: Vessels in long radial multiples and in a radial pattern. .. . Therefore. Robinia has vestured pits and vessel-ray parenchyma pits that are similar in appear ce to intervessel pits . . (Proteaceae). species groups . . ceae ). (Fagaceae). Nevertheless. Amyris sylvatica (Rutaceae).10: Soli­ tary veJsels in a radial pattern. and so its usefulgess for determining phylogenetic relationships is compromised. 6. For example. Curtisia sp . and often one of the first features used in diChot9mous keys. rare in tropical woods). TIle features identification and those useful for determining evolutionary relationships useful and de~eloping phylogenetic schemes are not always the same. .1: Solitary ves sels in a diagonal arrangement. Mulberry family) . .4: Rays heteroci llular with square to upright marginal rows and sheath cells. . Citrus aurantium (Rutaceae) . .4 . Stable and reliable characters that are consistent from one sample to the next are best (B azier 1976) .. Legume family) and Moms rubra (Morae ae. 100 urn in Fig . Choosing useful characters is dependent upon systematic wood atomical studies that use large numbers of reliably identified samples. 5. 250 urn in Fig. Robinia pseudoa cacia (Leguminosae. Cel tis laevigata (Ulmaceae). latewood ve ssel angular in outline. opposite to scalariforrn pits (Fig. . and fibres with di tinctly bordered pits (Fig. This feature is highly corre­ lated wlith climate (common in temperate woods. . latewood vessels ronnded to oval in outline. understanding the range of variation within a species is of u most importance. ' t + . 11.12: Solitary vessels in dendritic pattern. Woods with 'primitive' feature s such as exclusively solitary vessels (Fig . enlarged cell s with crystal s.8: Large ray s and regular tangent al bands of parenchyma (scalariform parenchym a band s). Bumelia angustifolia (Sapotaceae). wood fling-porou s. a red oak.II: Solitary ves­ sels in radial to diagonal pattern. both of which are ring-porous with latewood vessels in small c usters. 7-12. e.. Celtis laevigata (Ulma­ ceae). Quercus [a lcata (i. 3. (Co rn aceae). e. a white oak. Eucalyptus globulus (Myr taceae). Phylogenetic rela­ tionships are established on the basis of shared derived characters. 1-112. . wood ring-porous. 1. Cardw ellia sp.3: Ring-porous wood with la ewood ve ssel s in wav y tangential bands . ac­ compa ied by herbarium vouchers . Robinia and MoTUS can be separated by microscopic feature.e . 14). while Moru s has non-vestured pits and vessel-ray pa­ renchy a pits that are coarse and with reduced borders. there is the risk that some features will be chosen that are not alway s useful for separating the taxa. 2. 12). ea ily observed feature (Fig. seal riform perforation plates. Sopranthus sp. (Annonaceae). particularl y for north temperate woods. Quercu s alba (i.9: Vessels in a diagonal to dendritic pattern. and ge era is based on only a few samples . . Faga­ Scale bar = 500 urn in Fi g. woots belonging to different botanical families may superficially appear similar because there has beeu considerable convergent and parallel evolution in wood struc­ ture.A review 247 strts What offeatures are valuable? Ther has been considerable discussion in the taxonomic literature on how to de­ termine useful features for biological identification and classification.Wheel er ] & Baas - Wood identification. Fagaceae).6: Id blasts. 41) can be relatively easy to recognize becau se they occur in relatively few familie s. . 13. Ring-porosity is a distinc­ tive.

. Hypelat e trifoliata (Sapindaeea e).14: Opposite intervessel pits. . . 19 (3). Co rdia sp. 1998 Fig. Carpi nus . Andira inerm is (Leguminosae).17: Vessel-r ay paren­ chyma pits. .21: Aliform­ confluent parench yma. (Com­ bretaceae). Quercusfalcata (Fagaceae).13: Scalariform intervessel pits.23: Aggregate rays at left and right. .26.15: Altern ate and widely spaced inrervesse l pits.18: Vessel-ray parenchyma pits with rednced borders. Liriodendron tulipi fera (Mag­ noliaceae).20: Vasicentric parenchyma.22: Confluen t-banded pa­ renchyma. . Term inalia sp. Penta cletra macrophylla (Leguminosae). . and simple perforation plate viewed from the side.248 IAWA Journal.16: Alterna te intervessel pits. Hypelate trifoliata (Sapindaceae). . Vol. (Vitaceae). Halesia caroliniana (Styraca­ ceae). . . (Boraginaceae). . Ampelopsis sp.19: Vestured inter vessel pits. 13.

28). 13.A review 249 sugge~te d that ray shape was useful for distin guishing Larix from Picea. Sci rotic tyloses. However. 23. 22. 25: 50 urn in Fig. rays of two distinct sizes (Fig. . 32) (1 % each). ­ 25: AI elements storied. 1994. 20. incl uded phloem (Fig . and vesse l-ray parenchyma pit s with reduced borders (Fig.Wheel r & Baas . oil or mucil age cells (Fig . 12) (5% eac h). tangential bands of latew ood vessels. s o~1 e fea tures are 'use ful' for ident ification because they occ ur in relatively few taxa. Bartholin 1979). . 25). 38) or dru ses (Fig. axial parenchyma absent.Scale bar =500 urn in Fig. Entandrophragma cylindricum (Meliaceae).24. . 31. 3. or dicoty edonous woods (see below ) and used by Chalk while gathe rin g data for the firs t e ilion of ' Anatomy of the Dicotyledons' (Metcalfe & Ch alk 1950) are not com­ mon . . show that r y tracheid pit ting is consis tently useful for distingu ishing these two genera (Anag ost et al. 15-18: 5 prn in Fig. (Malvaceae). • cr~ta1 s in idioblasts (Fig. 250 pm in Fig. 100 urn in Fig. 36). 37) (2% each). rays with the uniseri ate portion equal in wi th to the multiseriate porti on. 33. 20 occur in 10% or less of the world ' s hardwood s. fusiform parenchym a eel s commo n (3% each). 27. 6). 2 1. 10-12). tile cells (Fig. Of those 46 features. 30). 10. • Axial ca nals (Fig. Wheeler er al. Ag rega te rays (F ig. Storied rays (Fig. 11. 26.26: Homocellular rays. It is rare that a single feature ident ifies a single group. ~ortY -S iX of the 63 anat omic al featur es they used occur in less than 25% of the world s hardw oods (percentages based on the OPCN database with 5260 entries. Se i-ring-porous woo ds (7%): Sc lariform perforati on plat es with more than 20 bars.35) (4%). Th ese uncommon featur es and their per cent occurrence are: • • • • • Presenc e of vas cular or vasice ntric tracheids ( 10%). Hibiscus sp. Acer rubrum (Acer ceae). later detailed studies of large numb ers of samples repr esenting many species showed that ra~ shape was not a reliable fea ture for distinguishing these two geuera. rays more than 10 cells wide (Fig. fibres with spiral thickenin gs. 31. rap ides (Fig. solitary vessels arranged in a radial to dendr tic pattern (Fig. Ring­ porosity. 24. tile cells appear to be co nfiried to the M al vales (Fig . 25 urn in Fig. 29. 30) ri ng-porosity (Fig . with a few of these unusual features in combination with more commo n fea­ tures ay be qui ckly identified. 4) . 17) occur in combination only in the Fagac~ae . 10-12). 3). any of the features chose n by Clark e (1938) for a mult iple entry card ke y . c aro li~ia na (Betulaceae).23). 32).Wood identification . particularly if their geographic source is known. 12) (6% each): • Ta gential vesse l arrangeme nt (Fig. 1986). (Fig.1 4. 3. 19. and rays with sheath cell s in a North Amer can wood indica te Celtis (Fig . Axial parenchyma band s more than 4 cells wide (8%).rare (9% eac h). canals or latex tubes (Fig.24: Storied rays. Recent s t u die~ incl uding one that exa mined ove r 87 samples repre sen ting 39 species.

Guazuma latifolia (Ster­ culiaceae ). 1998 Fig. .3 1: Rays with tile cell s. .30: Heterocellular rays with latex tube s. .35: Traum atic axia l cana ls. Dipterocarp us sp. . and rays with uniseri ate porti ons as wide as mult iseri ate portion s. Dryobalanops sp. Oclzrosia sp.33: Scattered axial canals. .37: Druse in . Rauwolfia ma crocarpa (Apocynaceae). . Michel ia cltampa ca (Magnoliaceae) . Avicellflia offic ina lis (Avice nniaceae ).36: Oil ce ll in ray. . . (Bornbacacae) .29: Ray with radial canal. . 27-42.250 IAWA Journal. (Dipterocarpaceae) .32 : Tile cells in radial view.34 : Ca nals in tangen tial lines. (Di pterocarpaceae).27: Are as of diffu se included phloem torn o ut during sectioning. Vol. Pisonia acu­ leata (Nyctaginaceae). Khaya sp. .28: Areas with con ce ntric included phloem torn Out during section­ ing. . (Meliaceae). 19 (3). Loxoptery gium sagotti (Anacardiaceae) .

but the combi­ nation of solitary vessels (Fig. 39. but variable within the type genus of that group Combretum (Van Vliet 1979). ash.. The level to which I an identification can be done varies within and between families . perforated ray eel s. Most Eucalyptus specie.. size and distribution). (cf. 32. 38. within Terminalia using characters such as crystal shape. but it is possible to distinguish Ulmus americana from Ulmus rubra and from the hard elm group .38: Raphides in enlarged procumbent ray cells. (Vitaceae). 250 urn in Fig . 28. and often not to a single genus. or species) a wood can be identified and when to be satisfied with an identification.42 : Septate fibres. and is typical of a few aberrant genera in the Melastornataceae and Cornbretaceae. . Highef levels of classification also show variation. but not species.41. gum. 3 .. . . 100 urn in Fig.39: Prismatic crystals abundant. (Burseraceae). In most Myrtalean families it is possible to key out in ividual genera using microscopic features.Scale bar =500 urn in Fig. it is important to use the literature and reference material to determine what level of identification is possible within particular taxa. families that are relativ Iy small). and in square ray cells . Hibiscus tilia ceus (Malvaceae).. the indi­ vidual families are only poorly defined on wood anatomical features. .31. 33-35.36. Imperforate elements with distinctly bor­ deredaits in tangential walls . genus.Wood identification . Within the order.40: Prismatic crystals in axial parenchyma. it is not possible to distinguish any 0Je ring-porous species of Celtis from another. some orders are stable for vestured pittingl. 1) and fibres with distinctly bordered pits defines most members of the Myrtaceae. Entan rophragma cyt indrlcum (Meliaceae).g. 50 urn in Fig. while others are heterogeneous (e. species group. etc. in the Ulmac[ae (Elm family) native to the United States. Euphorbiaceae. Drypetes keyensis (Euphorbiaceae).g. In the Annonaceae it usually is not possible to identify individual genera. 25 urn in Fig.41. . in the Betulaceae (Birchlfamily) it is possible to identify individual genera. Olaca eae) with as much difference between genera as exists between some families. The combination of included phloem and vestur d pits (Fig . Canella winterana (Canellaceae). Icacinaceae.Wheel er & Baas . 29.40. bloodwood. 27. 19) is largely restricted to this order. .g. None of the features cited here have universal J square1 ay cell. can only be identified to a group. Included phloem is diagnostic of a group of closel} related genera in the Combretaceae. Dadswell I 972)' IBecause of this lack of rules. 42. 37.A review 251 Levels I~f identification On of the harder things about doing wood identification is knowing to what level (familr. itis sp. Th"1 differential value of different wood anatomical features for identification at different levels of the taxonomic hierarchy can perhaps best be demonstrated by an example from a large woody plant order whose delimitation is now widely accepted: the Mt:rtales (Van Vliet & Baas 1984). while in other orders this feature varies between and within families (Van Vliet ~ Baas 1984) . g. Betulaceae. Hamamelidaceae. Magnoliaceae. Some families and genera are relatively homogeneous in their wood anatomical charaleristics (e. There are no all purpose rules. Isolated pieces of wood usually cannot be identified to species. AUCOH­ mea s . e. . and within some genera it is possi­ ble to key out individual species (e.

1990. and often it is relatively easy to determin e the identity of an unknown wood by comparing it with descriptions and illustrations in introductory wood anatomy texts and atlases (e. 1989). Lemmens et al. i. 1987). even though most anato­ mists do not perceive current condi tion s as supportive of anatomical studies. Atlases with good quality photographs illustrating the salient characters of woods are widel y used. 19 (3) . famil y. after studying a large sample of speci­ mens of a taxon (order. 1986. line drawings .. Fagerstedt et aI. 1996 . Brunei (Ogata & Kalat 1997). 1996). It is somewhat surprising that the number of atlases produced in the last decad e is as high as in previous decade s. • Chile (Rancusi et al. A par­ ticularly valuable atlas is the CSIRO Hardwood Atla s (IIic 1991) that has photograph s (no text descriptions ) of some 1800 species (cros s. Wagenfiihr 1996. section or species). Schweingruber 1978. • Korea (Lee Pil-Woo 1994. Many earlier atlases are listed by Gre gory (1980 ). Liu Peng et aI. Saiki 1982. Atlases prepared in the 1920s and 1930s are still useful in the 1990s. Vol. Menon 1993 . 1986) [text and photographs]. Wood anatomical atlases are the equivalent of such field guide s. Suzuki et aI. 1998 diagnostic value at the same taxonomic level in other orders. in Japanese). and useful for many years . photograph s only) . 1987). . radial . Mediterranean (Edlmann Abbate et aI. Eu­ rope . Panshin & DeZeeuw 1980 . IDENTIFI CATION PROCEDURES Comparison Identifi cation by comparison is one of the most frequently used methods of identi­ fication. Papua New Guinea (Sudo 1988. 1998 . • Israel and adjacent region s (Fahn et aI. and is the basis of most natur al history field guides (Pankhurst 1978). tribe.a revision by Sul airnan & Choon : Soerianegara & Lemmens 1993 .g. Hoadley 1980. • Commercially important woods from Africa and Tropi cal Ameri ca (Nadi Bert i & Edlmann Abbate 1988 . • Mexico (Barajas Morales & Leon Gomez 1989) [text and phot ographs]. • Himalayas (Su zuki & Noshiro 1988. • Iran (Parsa Pajouh & Schweingruber 1985).. Wilson & White 1986). and tangentiaI sections are illustrated) and was intended to be a portable wood anatomical slide collection . Japanes e text. and Japan. roots of trees and shrub s of Brit ain and northern Europe (Cutler et aI. • North America (Furono 1985. 1994) [text and photographs] . genus . • Southeast Asia and Pacific (Ogata 1985 .e . there are relatively few domestic woods that are commonly used commercially. Diagnostic value can thus only be determined 'a posteriori'. Gro sser 1977.252 lAWA Journal . In the United State s. 1995. wood descriptions are included along with descriptions of the rest of the plant. 1992. 1991). Martawija ya et al. including Peru (Acevedo Mallque & Kikata 1994/1995) [text and photographs]. Wong Yong Lee 1997). Sosef et al. A selected list of some additional recent micro scopic atlas es follows: • Europe (Schweingruber 1990).

and so would immediately refer to these keys. The longer the key. Mult iplb entry keys The simplest multiple entry key is the so-called synoptical key. 1983). Die tomous key s are useful for unknown s for which there is a small numb er of possibl matches. Co uter progr ams can help develop keys by calcul ating the information content of featu es.g. here is a need for a family key that would indicat e in which famili es particu­ lar com!binations of featur es occur. and for woods of a particular family or genus. CSIRO fa mily key ). genus. or read the short gen­ erali ze family descriptions in 'Anatomy of the Dicotyledons' (Metcalfe & Chalk 1950) . species group. Baas et al. Brazier and Frankli n (1961) provided tables useful for disti guishing related species of some genera . if they knew ey existed. it w II be necessary to use a key (e. Dichotomous keys alpartiCUlarlY useful as regional works.. group of genera or famil y). the more Ii ely there will be an error in choosing the correct descriptor at any one cou­ plet. and unrelated woods that appear similar. in Fa hn et al. The starting point and sequence in which feature s are used are predetermined by the auth or of the dichotomous key. 1983 ). related genera. with one or more ferI tures used at each dichotomy / couplet. Dichotomous keys present a series of paired contra sting choices. and often used interchangeably. (1986). Good examples of this type of key are the jists at the end of Metcalfe and Chalk ( 1950.A review 253 Data iru tabular form are particularly valuable for showing which combinations of feature~are useful for distingu ishing between smail groups of closely related woods or woo s of similar appearance. and how to use the featur es to divide species into groups (e.g .Wh eeler[ & Baa s - Wood identification . Otherwise. The unknown must match every char acteristic of the taxon as de­ fined inlthe key and there usually is only one path to one identity. Man receut systematic works contain dichotomous keys to the woods (e. At each couplet. The key user is directed to another couplet and this proce ss is continu ed until finally reaching a name (which for wood could b~ a species. and . Some experts can recognize that a wood belongs to a partie lar family or genus. both memb ers of the Meliaceae (Mahogany family).. k ys with over 200 taxa are unwieldy (Pankhurst 1978. a well-constructed dichotomou s key ca quickly lead to an identification. 1~88 ' lists in Gregory 1994). use lists giving by­ famil Y{CCUrrence of featur es (Metcalfe & Chalk 1950. 1991). and for that key to be linked to specific publica­ tions add keys for famili es and genera.. and for commercially important woods. Dallwitz & Paine I 86) . which lists for each diagnostic feature the taxa that have that feature. For example. Dichotomous keys direct the observer to look for the features the key constru tor considered useful distingui shing features. one of the two state­ ments i cho sen as applying to the unknown. to determin e the family or genus the unknown belon gs to. and for material without missing feature s.g. one table shows the combin ation of features that can be ~ sed to distinguish Swietenia and Ca rap a. Dichotomous keys Diagrostic keys have been used for centuries in biological identification (Pankhur st 1978).

. 19 (3). Verbenaceae. 7. Features used in the key are num­ bered and defined. because that wood might represent a species not included in a key. 105. either because no or little anatomical data are available for it. Usually.. and the observer can decide which features he considers most reliable in determining the affinities of the wood .254 IAWA Journal. the first application of multiple entry card keys was for wood identifi­ cation (Clarke 1938. Species descriptions are based on at least four samples. For certain situations a list rather than a single name may be a better strategy for getting to an 'accurate' identifi­ cation. 16. 14. Because of this flexibility. To identify an unknown. This key covers 380 timbers representing 800 botani­ cal species. such keys are particularly useful for unknowns in which some fea­ tures cannot be observed.. Recording these feature numbers serves as a shorthand description of a wood. Such keys have the advantage that the sequence of characters used in an identification procedure is in­ spired by the unknown wood sample. 1998 those for the Combretaceae and Terminalia (Van Vliet 1979). not by the author of the key. There are accompanying notes on distinguishing features. if a wood has a particular fea­ ture the edge of the card will be notched to indicate presence of that particular feature. or with 2) fossil woods that differ in one or more features from extant species . a needle is passed through a stack of cards at a perforation representing a feature seen in the wood. tables. but the sequence in which features are used and the total number of features used are up to the observer. Tectona grandis 1. Cards have perforated edges. Another advantage of multiple entry keys is that it is easy to add new species to a key. The Princes Risborough microscopic key to hardwoods (Brazier & Franklin 1961) has been one of the most valuable . Ilic 1990). and then for each taxon examined. The cards with feature present fall out. When using multiple entry keys it is possible to stop short of a single name.g . 3. and short dichotomous keys . so this is no longer a limitation.. there is one card per taxon. Subsequent comparison of the taxa with a combination of features similar to the unknown helps narrow down the likely affinities of the unknown. with more than 60% of the samples having herbarium vouchers (estimated from data in the first 10 pages of their publication). Multiple entry keys are easily computerized.23. Multiple entry keys are still published today. Cards of species with the feature absent stay on the needle. the perforations are numbered sequentially. Vol. In such card keys. 1) tropical woods of unknown provenance. In multiple entry keys only one feature is used at a time. The number of features used in card keys was limited by the number of holes that could be fitted along the margin of the card (generally this was less than 100). just by inserting a new card or new entry to a computer database. and this information can readily be en­ tered into a database for computer-assisted wood identification.g. 18. Apparently. and this is one of their advantages over dichotomous keys. 13. a list of the numbered features present in that wood is given (e. and each numbered perforation represents one feature . e.. The sorting process is repeated until a single or only a few cards remain. from Sudo 1992. . Pankhurst 1991).21. There are multiple entry keys based on macroscopic and microscopic features.

Detienne & Jacquet 1983). Computer databases and programs based on card key data Twofcomputer-aided wood identification projects that have been strongly data driven. t then is nece ssary to go back and edit the description of the unknown. after persona] computers became widely avail­ able. if the list is 10. Normand & Paquis 1976. The OPCN database (5260 entries) includes the Chalk d~a (2227 entrie s as originally described. th re were many separate projects to develop computer-assisted wood identifi­ cation p ograms and databases (discussion to follow). One project tarted with data from the Oxford cards collected by Chalk during the prepa­ ration of . LaPasha & Wheeler 1987). Oxford based key . it was envisioned that an anatomist would access ~Ilarge computer housed in one spot. and use a program and database stored in that lar e central computer.Wood identification .The GUESS program (LaPasha 1986. data fro1n the CTFT keys (929 entries . with the data transmitted on paper. or electronically via the ~11temet. g.A review Computer-assisted wood identification Whe~ computer-aided wood identification was first discussed (Miller 1980). Data collected at one institution can be shared with others. and evolved from a program that was used on the mainframe compu+ (Pearson & Wheeler 198J). and 510 entries edited to reflect new informa ion from more recent literature). ~gain with editing so that they would be comparable to the Chalk cards . Many of the identifications an instituti n does are of woods from nearby.Anatomy of the Dicotyledons' (Metcalfe & Chalk 1950) and recorded on the martinallY perforated cards developed by Clarke (1938). and has been helpful for indicating which family or genus might be related 0 an uuknown . However. The other is based on the mul iple entry keys developed by Ingle and Dadswell at CSIRO. on diskette. 1986). and reliable means of identifying the woods of a particular regiou. y using existing key card data . or if there were no matches then more I . data from the Princes Risborough Key (400 entries. accompanies the OPCN databases (Wheeler et al. and data from post-1950 literature on systematic wood anatomy (1194 entries). Because the OP~N databa se is based on multiple entry key card data . ith both DOS and Macintosh versions. then the presence or ab­ sence 0 that feature can be required as characteristic of suggested matches from the databa s1' Thi s program is u s~ful for generating lists of woods that have a particular set of fltures. Mismatches are allowed. If the presence or absence of a particul: feature is considered important or unambiguous. as a list of character states is en­ tered . easy. razier & Franklin 1961) edited so as to be comparable to the Chalk data . so that there need not be a 1:1 correspondence in the charact~s of the unknown and entries in the database. or in no matches what­ soever. then more features can be added . so that it is sensible to ensure that there is a comP1ehensive database for that region . The initial 'run' may (and often does) result in a long list of woods ith features similar to that described for the unknown . are here discussed at some length below. the number of features used is he same as on the cards Chalk used.Wheeler I& Baas . and a quick . 69 anatomical and 15 miscellaneous features The <iJUESS program functions as a batch job. be­ fore personal computers were commonplace.

The accom­ panying publications for the CSIRO keys are well-illustrated. so that the user can choose an end point when a small number of names is reached. An alternative program (for the Mac­ intosh) allows entering one or more features at a time. There is an option for suggesting 'best subsequent feature' that suggests which features might be useful for eliminating other taxa .1998 mismatches can be allowed or features deleted.CSTROm is a program developed to use the card keys developed at CSIRO (Ilic 1993). After each feature or group of features is entered. and includes the Family key (based on data collected by DadsweJl in the 1930s through '50s and for the southwest Pacific region. What seems par­ ticularly helpful with this program is that its 'features help' option provides on-screen text definitions of the features . either publications on a particular group of woods or regional atlases. and there are plans to incorporate images. and to direct the user to relevant literature. or geologically ancient woods where a one-to-one correspondence may be unlikely because ofevolutionary differences between the fossil and extant plants. To reach a single name. This reflects different levels of confi­ dence in the databases. and provides suggestions on how to use the features (Wheeler et a!. However. and the on-screen help address this prob­ lem for these keys . One of the major prob­ lems with using keys is 'learning the features' used in the key. gives their relative fre­ quency in the database. the Macro key (using data collected from the 1940s through 1960s and also emphasizing the southwest Pacific region. It includes taxa from throughout the world.256 IAWA Journal . There is a program (VIEW) for retrieving taxon descriptions from the database. the number of entries in the database matching the unknown ap­ pears. e. Ilic 1987). The OPCN database is the largest one (5260 entries) available with all taxa coded with the same set of feature numbers. The GUESS program was intended to shorten the process of comparative identifi­ cation. and somewhat different objectives from the GUESS program. The CSIRO books describing and illustrating the key features . Vol.. The entries in the OPCN database that represent information published since the first edition of 'Anatomy of the Dicotyledons' are coded to indicate the literature source for that entry. further comparative work is needed . woods whose geographic source is unknown. it probably is quicker to use keys in textbooks or compare the unknown to illustrations and descriptions in atlases . The CSIROID program does not allow for mismatched features and so does not have required absence or required presence. 1986). llic 1990). and the Eucalyptus key (Dadswell & Eckersley 1941). and a manual describing the identification program (LaPasha 1986). for common commercial woods in Europe and North America.g . This identification program is interactive and one or more features can be entered at a time. The GUESS program and OPCN databases are helpful for the ' hard to identify' or unusual woods. Neither of these programs features on-screen help for suggesting useful fea­ tures. . 19 (3) . a successful identification with the GUESS program is a list of roughly 10 or fewer taxa. For a wood whose geographic origin is unknown or is from Southeast Asia or South and Central America. There is a manual that describes and illustrates the features . CSIRO keys .

but are using additional features suitable for local wood s.ne.l datab ases to oth er datab ases allows anatomis ts to exp lore relat ionships between lOad struc ture and prop erti es (as above). It is possible to gen erate I sts of woods that sha re certain characteristics and uses. or wood structure and climate (Wheeler & Baas 1993 . There is one datas t for wood iden tification that also uses the IAWA Feature List. It woulf be a real service if there was an interface that would allow eas ier entry into the D LTA-associated programs. 1998).. naturally dur able and used for cross-ties. A rev iew 257 'New ' databa ses and keys . Lin king anatom ic. it is necessary to continue to add to and refine the exi sting d tao The TI S sys tem (Kyoung et al. Wi mann et a1. They have found discrepancies between lit­ erature descriptions and their own observations. . one or more characters can be entered at a time. They are usin g the literature as well as gathering new data. There are few wood anatornica~ dat a in DELTA format.The number of featur es in the IAWA list ( 163 anatomi­ cal and 5 ~ miscellaneou s features) exceeds the number that could be accommodated on margiljlally perforated cards . The pr grams developed to accompany the DELTAformat for recording taxonomic descripti o s (Dallwitz & Paine 1986) have been accepted by many taxon omi sts as their standard.g. They started with the IAWA feature list as a base. TIle conversion of numeric key card data has not yet been dp. In Arg~ntina . e. 1994) was de veloped for Korean woods. Ttle key is interactive. Woodcock 1994. Espin o~a de Pernia a nd Miller ( 199 1) adapted the IAWA Featur e Li st to the DELTA format anq developed a datab ase for 40 co mmercial ly important Venezuelan woods. wood s with a spec ific g~av i ty of more than 0. which indi cate s that whil e exis ting wood ana omical data are useful . There is a powerful identification key (INTKEY) for data in the DELTA form at.Wheeler '* Balls - Wood identifi cation _. and one dataset fo properties and uses (informat ion from Chudnoff 1984). and the translati on of data from the existing wood dat a ases . The DELTA system was intended as "a general system for process­ ing taxon l1mic descriptions" and its primary objective was not to help the casual or occasiona] user with identification.50 . Monteoliva and Oli vera (19 94) are working on developing a system of identificati on for Argentinian woods. wood structure and uses (Gasso n & Cutler 990 ). The program can suggest which fea­ tures maylbe useful in distinguishing between rem aining taxa. and allow s more complete descriptions of woods. which should permit distinguishing more species . and a list of po ssible matches obtained at any point.

and Chinese woods (Yang & Cheng 1990. Problems in identification of biological material The problems associated with wood identification and reasons for not reaching an accurate identification are the same for wood as those for any biological material (see McNeill 1975. The bibliographies prepared by Gregory (1980.258 IAWA Journal . wood structure and properties of these species . Reliability of data is always a concern. Stem (1988) compiled a list of the various institutional wood collections and the regional specialities of these collections. the DELTA format accommodates both nu­ meric characters and text characters . Richter. then whole samples and slides of woods can and should be compared with the unknown. Other computer-assisted wood identification projects include ones for Japanese woods (Kuroda 1987. 1993. Fortuner 1993) and include: 1) An inadequate . 1984). explanatory notes and illustrations can be acces sed. Richter and Trockenbrodt (1993 . In contrast to the programs such as GUESS or CSIROID. A database with over 200 entries (trade timbers) is currently available. but are useful information in text descriptions . There are English and German versions . personal communication 1995). 1994) are invaluable for entry into the literature on woods of a given region or of particular families and genera. reviews characters suitable for conifer identification and presents a list suit­ able for use with DELTA-lNTKEY. and is accompanied by a synoptic key. Hamburg and D . If one has access to a wood collection. More recently. When using this key. (1994) have prepared an interacti ve computer key based on the macro­ scopic feature s of 115 of the most important Guyanese tree species.CHA for producing text descriptions. and VWOODINT. Pankhurst 1978. Espinoza de Pernia and Miller have made an important contribution as they have developed a framework for other anatomists to put newly acquired data into DELTA format. Grosser. The two DELTA­ based programs they have created are VWOOD. as descriptions are based on 5 samples per spe­ cies. and information on the distribution . Zhang et al. The key is designed for use by forest products industry personnel. These characters are not used when identifying the wood. and new entries are made as time allows. it is imperative that the identification be verified by reference and comparison to reliably identified samples. A recent thesis on conifer anatomy (Heinz 1997). 1986). option INTKEY (interactive wood identification). 1998 The quality of this database is high. Germany. Izumoto & Hayashi 1990).appearance. or descriptions and illustra­ tions in atlases or other publications. 19 (3).1996) have set up the IAWA fea­ ture 1ist for use with DELTA. and the DELTA program allows one to record the number of samples on which a description is based. Munich. supervised by J . tropical woods (Tochigi et al. A program and database for South African woods is used in teaching wood anatomy and identification at the University of Stellenbosch (Malan. Vol.CHA for the identification option INTKEY. OTHER ASPECTS OF WOOD IDENTIFICATION Verifying an identification No matter what key is used . Brunner et al.

A review 259 basis for co mpar iso n because a) the unknown lacks c ritical features. Trunk wood and root wood often differ in both qu antit ativ e (cell diameters and lengths) and qu alitative features (pa renchyma a bundance and distribntion . porosity. Error ih fea ture recog nition is as likely a so urce of misidentification as any othe r. and with linkages to tabul ar data . ray type). Intermedi ates between two features may be interpreted differ­ ently by different peopl e (what one anat om ist terms semi-ring-porous. Th e IAWA Features List (IAWA Co mmittee 1989) is elping to incr ease consisten cy in the nse of wood anatomi cal terminology. and pertinent literature references . trunk. use of fibre-trac eid. with inform ation o n sample prep aration .0t kn own . users are apt to try wood identification pro­ gram s wit out reading the explanatory mat erial on the accompanying databases . images of wood s. branch. sapwood or heartwood) or age of the wood material (Ifrom ju venile or mature stems) .r. and 2) Difficulty in using keys because of errors in feature recognit ion.. Brazi er & Franklin 1961 . An o bjfctive for the new computer keys is to provid e a means of access ing illu s­ trated feaf re definitions directly from the identi fication program (Richter & Trocken­ brodt 199 ). I Mo st of the mult iple entry keys are accompanied by introductory sections that ex­ pla in the Ifeatures used (e. and so the range of vari ability of some species and genera is.g . dichptom ou s keys. Juvenil e wood of trees (wood near the pith ~d formed by a youn g cambium) and branch wood usually ha ve smaller cells than ma re trunk wood and ray ce llular composition and ray siz e ofte n vary with position i the tree (cf. Sinc e computer zation of multiple entry keys. Detienne 1& Jacquet 1983). most user s would take the ti[e to read the introductory materi al before attempting to use the key. use illu strated prompts while describing an unkn own (some of these a. Jane 1970: Pan shin & DeZ eeu w 1980). Descriptions of many woo ds are only based on o ne or a few samples . When the se keys were on punch cards. person al communi arion).Wheeler Ba as - Wood identification .g . as in decayed wood o r oor ly preserved foss il woo d. if desired. Normand & Paqui s 1976 . or. becau se not all sp cies are includ ed in a key and the full range of variabil ity of a taxo n might not be kn wn. Not only is there the variability exp ected when comp aring dif­ ferent in9ividual s of the same species from different localities and / or with different genotypef' but also the huge amount of wood ana tomical variation that depends on position ir the tree (root. so that the boundaries betwe en some features are not sharp . and woods of so me have never been describ ed or incorp orated int o any key. Multimedi a presentations on wood ide ntificatio n are likely to be devel ope d in the next few years . I1 ic. An int eract ive CD-ROM of the informatio n o n wood anat omy and woo d properties of the 400+ genera of South­ . Thi s is likel y tt result in mi suse and misinterpretati on of features. reticulate parenchym a). Th ere t e so me terms that wood ana tomists have applied differently (e. illu strated featnre descriptions . There fire thousand s of species of woody plants. or b) an inadequ ate reference base.. Many wood anatomica l features are not discrete. or interpretin g a featu re differentl y from the way a key cons truct r intend ed . ano ther might term diffuse-porou s or rin g-po rou s). peers have been incorporated in the new version of CSIROID . either inc rrectl y interpreting a feature in an unkn own becan se of the obser ver 's lack of backg ound or training. intera ctive keys.

M. FDA a la Cuenta No. A J .S. 1: 68-70. IAWA Bull. 1994/1995 . Anatornia de mad eras de Mexico..M. Th e Kebun Raya Bogor. 1979 .J. M . Root identifi cation manu al of tree s and shru bs . Div. Zurich .E. & Y. For.W. Detienu e. Meth od for tile identifi cation of colour ed woods of the genu s Euca lyptus. Suhirman. Wood an atomy of the Oleaceae . H. No. Bolt on & D..s. 1938. Chern. 19 (3) . 14: 266-280. App!. Serie s Nr. Carling (eds. espec. Paine .H.. 9 : 103-182. 3. Bartholin. Clarke. Fuaddini. 1989. Counc. 1988. Bastin . Forestry 15 (2): 1-13 [Ill Plates]. n. of Entomology. M. Dechamps & M. Mexico. CSIR. Wood Strncrure in Biological and Techn ological Resear ch: 102-106.M. Res. & M . Bra zier. S. Chapman & Hall. Prod. Bog or. Bull. IAWA Bull . J . Bailey.M. Vol. Ed.C. Meyer & C. Ind. Zandee. 206. Prod . A guide ro the anat omy of roots of trees and shrubs hardy in Britain and North ern Europ e. M . 1972. trop. New Phytol. . 1983. P.F. Richard son . 1917. ZUrcher.s. In: P. The Picea-Larix problem. Shendar). V. Atlas d'id entific aton des bois de I' Amazonie et des regions voisines.1. Pfeiffer. the Netherl ands. Confirm ation and significanc e of Bartholin's method for the ident ification of the wood of Picea and Larix.E. T.D. Washin gton. Dallwitz. Burn ell . Report No. H. Tech . Fluore scence study of 10. The Kebun Ray a Bogor Confer­ ence Proceedings (Eds .s. Rud all. 1994 . Dad swell .E. P. Paper. 15: 171-184. User's guide to the DELTA system: a general system for process ing taxonomic descript ions. P. Tropenb os Series 10. IAWA J. London. l. Tropical timbers of the world. R. but also on compact disks. Jacquet. Wood structure and quality research . Atlases probably will be available not only as printed copies. & T.: Noge nt-sur-Marne. Baas. Wageningen & Swi ss Federal Institute of Techn olog y. C utle r. 1932.J.5 . van derWesten & M. 1994 .). Dadswell . G.J . REFERENCES Acevedo Mallqu e. Major timb er trees of Guyana . Div. BioI. Lond on. Baas . Butler. Pub 1. J. & A.260 IAWA Journal. 1998 east Asian wood documented in Volume 5 (1-3) of the PROSEA Series is currently being produced by the Expert Center of Taxonomic Identificati on (ETr) in Amster­ dam. Idenrification of hardwoods : a microscopic key.E.L. 1986. Ecke rsley. Bra zier. The role of the microscope in the identificati on and cla ssification of the 'Timbers of Comm erce' . 46: HMSO . Res. Univ... Dadswell. Fund acion para el desarrollo Agrario.E. 1984 . especie s de una selva baja cadncifo Jia. NO.D. and on the world wide web. Pulifor. 38: 369-374 . J. Atlas of Peruvian woods. Avell a. 9: 346-352 . . Chudnoff. Brunn er. Th e Trop enbos Foundation. L. J. Card sorting method appli ed to the identification of tile commercial timber of the genu s Eucalyptu s. 1987. lnst. 66: 1-28. & G. n.. Kikata. 1961. S. Baas. nacoauton. Australia. & C. For. 1994.W. n.E. Ana gnost. Kuc era & E. Esser. Leon Gomez. Ob servation s on so me anatomica l featu res used in identificati on and tax­ onomy. Handb ook 60 7. A multipl e-entry perforat ed-c ard key with speci al referenc e to the identi fi­ ca tion of hardwoods. R. Leiden University Press. Technical Paper No. 1976. Gale. 3. D. P. CSIRO. 1989.610 woody speci es fro m the Ter vnren (Tw) collecti on.T. The ana tomy of encalypr woo ds. CSIR Div.DA Fo res t Service . P. Leiden Bot. P. H. & P. Gas son & R. M.A .A lens key.M. deZ eeuw . 13. J.recip es for conse rvation or overutilization? In: Str ategie s for Flora Conservatio n in Asia: 143-149. J. M. Barajas Morales. Belgium. Sci.O. 1941. Ce ntre techn oFor. T. D. IAWA Bull. Fran klin . Agric.

1986 . Shi man e Univ. J.Wh eeler & Baas - Wood ident ificat ion . Japan. Fager ste t.F. Wood ana to my an d id entification o f trees and shrub s fro m jIsrael and adjacent region s. n.F. n.L 997 .s. London. Development of a woo d auarorny and wood use s dat ab ase at Kew. [Jap anese. I : 104-109. Th e structure of wood . 12 : 34-50. Relazione e morografie agrarie sub tro picali e tro picali : nueva serie No . Kim & W. 1990. de Luc a & S. Insti tu te for Wood Biol og y. B. IAWA J. ed s.). Ja shem slli .B .s. J. Th esis (unpublished) .. Colorado. IAWA Bul l. Jeru sal em . A ustralia. Identifi cation sys te m of wo od as sisted by micro computer II. n. 1: 3-4 1. 1970. Die HOlzer M ittel europa s. An atlas of light and scanning elec tro u rnicrographs.'TK EY. ~' S ' 8: 69. Ilic.B. Identifyin g wood . Taunt on Pre ss . A natomy of North American woods. C omputer Vision: 137-164. 1977. Baas. Lockl ey & A. Sp eci al Pap er 244 . Bould er.A review 26 1 Edlman n Abbate. W. S upp eme nt 1. Kukachkf . Rijksherbariurn / Hortus Bot ani cu s. T imon en . Adam & Charles Bl ack . 1980 . Accurate res ults w ith simple to ol s . Greg ory. M . 79 . P. Baas & P Ga ssen. Lond on . ser. Y. H oadley. 1. IAWA J . L. 14: 333-340. B. PeJlin en . Bal­ timor . IAWA list o f micro scop ic fea tu es for hardwood iden tificati on . M.8. Greg OrY ' ~M . Wilson & DJ . 1985.s . 1994. H ayash i. White). 1980 . IAWA Bull. Au stral ia. n. Arrific ial Int el­ ligcnce. Entwicklung von Systemkornponenre fur die Co mpu teru nters tiitz te Best immung vo n adelh ol zern in DELTA / fr'. Ev ide nce of flora and fauna in the garden s and cultivated land de­ stroyed by Vesn vius in A. 1994. USA. 1990 . Ili c. Miller. IAWA Bull . J . 1 87.A. Ce nter for Studies of San 'in Regi on. 1 90 . lIic. Rice (eds. A ' E.B . 199 1. 1989 . Co mputer aided wood ideutifi cari on us ing CS IRO ID . P. Yl io istop ain o . Furono. Datab ases. ~ . 1990 . In : R. nw. He lsiuki. Fortuner (ed. M ika puu . . Hard w ood id entificati on using a microcomputer and IAWA codes.IIAWA Bull. 1993 . A che mical spo t-tes t for alumi nu m and its value in wood identIfica tio n. Research Data and Sourc e Ma­ teriall No .s . 10: 2 19-332 . & R . J.F. M . A tla nte anato rnico degli al beri ed ar­ bust] pe lla maccchia medit errane a. Adva nce s in Computer M eth od s for Sy stemati c Biol og y. Izumoto.] Jane. Springer-Verlag : Berlin . & R. Kyoun g .misia puu sta. Lazzeri . Mem Osaka Ky oiku Uni v. E S P i noz~e Pern fa. Gasson. For tuner t R. Firenze.of mic roscopic feat ure s for h dw ood ide ntifica tion to DELTA.G. J .L. 1996 . Grosse r. Bibli ography of sy stem atic wood an at omy of d icot yledons . IAWA Bull. 114.-S. Wheel er. n.lnstitu to Ag rono mico per J'Oltrem are. Geological Society of A me rica. Volc anism and fo ssil biotas : J 13-15 3. Germany.). IAWA C mm ittee (E. The TIS S syste m: a co mpute r search ro utine for wood ide ntifica tio n an d ph ysi cal or m ech anical pro perty info rma tio n. IAWA Journal . Werker & P. 1994 . BUI I.77 . Ei n mikrophorograph ischer Lehratl as . . Cutler. K . 15: 2 12­ 213 ( bstr act) . K.s. Th e NEMISYS so lut io n to pr oblems in nemat od e identifi c ati on. & D. 1987. Th e Jehus Hopkins Un iver sit y Press. Ill. Lee. Ili c. Wood identification : a n annotated bibli ography. CSIRO atlas of hardwoods. . CSIR O. Israel Acade m y of Scienc es . Spri nger­ Verl . Ch em ical and Wood Te eh ­ n010! y Paper No . . & S.). New wn. N. 11: 125 . 1 9 1. Ado pting the IAWA Jist. CS IRO Macro key for hard wood ide u tificatio n. Mill er. Leiden .. Studies of the Sanin Region . 1990. 39 (I ) : 87-102. D. l AWA f. Leu s hnerstrasse 9 1. H amburg. Th e CS lRO fami ly key for hardwood identification . P. 1 93 . Saranp aa & T. Heinz. Fahn . Berl in . In: M . Kurod a. e. CSIRO.L H ardwoods. (revised by K.

Papna New Guinea. Indonesian wood atlas. Nogent-sur-Marne.S. Lee Pil -Woo.. R. A program package for com­ put er-assisted identific ation . General unkn own entry and search system. Kore a. 1976. & Develop . & M. Res. Bogor. & H.7. c . VA: 91-111. 1. Manuel d'identification des bois commerciaux. 1983. Kart asujana. Afrique guineo-co ngolaise. 1985.] McNeill . 2. 1998 LaPa sha. I. Towards impro ved wood identification for better utilization of timbers: Papua New Guinean experie nce. VoL 2. K. A.e. Men ou. Prawira. Indonesian wood atlas. NY. Chinese Fore stry Pub­ lishers Beijing. Plant resources of South-east Asia. Dept. The Netherland s. Olivera.A. Clarendon Pres s. Suppl .M. IAWA J. [Japanese . Forest Products Research and Development Institute. Quanti co. [lndonesian. For.Korean timber atla s. Miller .O. Philippines. Malaysia. In: R.5 (2). & Develop .R. Centre techn oFor.with IAWA microscopic hard wood identifi catiou featur es. Metcalfe. Kartasujaua . Norm and. 1975. A. shrubs and climbers in Brunei Darussalam .U . Wood Technol ogical Assocation of Japan . 1989. A microc ompu ter based system for computer-aided wood identification .s. No.A. 1993. Res. 1993. J. R. Oteng-Amoako .] Og ata.). 1994. II. I. Wong (eds. 1994 . Africa . Anatomy of the dicotyledons . Ad ay. Nardi Berti. Mala yau Fore st Records No. 1950 . Dept. 1997. Bio­ logical ident ificatiou with computers: 28}'-'-289. P. C. Oteng-Amoako .B . R. Res. Structur e and identifica­ tiou of Malay an woods. 1992b. Jiang Xiaomei & Zhang U Fei . Vol. In: J.K. 1986. N. A. I.R. Y. 1992a. Brunei Forestry Research Project. R.A. Legnarni tropic ali irnportati in Italia: anatomia e identificaz ione. After-Care Programme. No. K. Leiden.B. 1995 . PNG Forest Re­ search Institute . Identificati on of woods of South East Asia and the Pacific Region. The System atic s Association Special Vol­ ume No. Metcalfe. 1988. ll.B. FRIM . Intr oduction to the use of computer systems for wood identi fication in Argentina. & E. Chalk.A. Box 3 14. I. Mill er. 1992 .. & A. IAWA Bull. CNR: Firenze (1992). Centre. 19 (3) . 1987 . 2nd ed. R. Vol. K. Japan International Cooperation Agenc y (JIC A) and Forestry Department. Pro ceedings 2nd Pacific Regional Wood Anat omy Confe rence: 235-241 . Structure of Korean woods . Wood anatomy of some trees. e.K. & J.A.S. 1980 . In: Pr oceedin gs of the Iuternatioual Symp osium on the Forensic Aspects of Trace Evidence. 3. Lae. Publi catiou No.J.H. 1996. Lemm en s. Anat omy of the dicotyl edons. Jeong Min Sa. Clarendon Pres s. Darusima Tradin g and Printin g Co. D. . 3. Rojo. Prawira & K. ITTO & Re­ search Institute of Wood Indu stry. Baril e. Centre. America (OOs. IAWA Bull..M. Araral & W. Pankhurst (ed. Academic Press. Timber s from Tropical Africa . & L. Identification of wood fragments in trace evidence. Phot omicrographic atlas of Papu a New Guine a timbers . S. Ag.e. U. Vol. ServoBull. Liu Peng. Mini stry of Industry and Primary Resources. Monte oliva.P.. M artawij aya. Chalk. Brun ei Darus salam . Timber trees: minor commercial timbers . Speci al Publ. 25. Seoul. America latin a. 15: 329 (Abstract). J. Ogata. Laguna. Paqui s. Edlrnann Abbate . Chinese Aca demy of Forestry. trop . Kadir. & L. (revised by Ani Sulaiman & Lim Seng Choo n).L.A. 474.n. S . Bull . 1: 154-160. Ribera Editore: Mil an (1988 ). Backhuy s Publishers . A botanist' s view of automatic identifi catiou .). Wheeler. Wood identi fication via computer. Soerianegara & W. A.). n. E. Mand ang. 474A. Ox­ ford.] Mart awija ya. Oxford . 2 Vol. P. [Chinese. C. June 24-2 8. For. Kalat. n.262 IAWA Journal.J.A. Bogor. 8: 347-354. Department of Ju stice Federal Bureau of Investigation. Kepou g. 1986. Kadir & S. LaPa sha.

Stuttgart.G . IH . n. Au atlas of sca nning elect 'ou micro graph s. Sudo. co m­ put e vision: 125.J.M. Hamiltou & K. Iu : J. D. Textbook of wood techn ology. Cretaceous age of the upp er part of the cCoy Mountains Formation. & C. 1978.11994. Lagun a. Geo ogy 15: 561. & M. 1981. pankhurtt.. Tropi cal timb er status in As ia and Pacific reg ions o bserved und er the micro­ sco pe aud a need to mi nimise lesser-kn own species . Baltim or e. Sudo . F. I ~92. Co ntributions to t. [En gli sh. Aca emia Scie ntific Book Inc. Ibarakl . Res.M. Wh eeler. An atomi cal character s and ideurification of Papu a New Gui nea timbe r spec ies . Wagenin gen .U. P. R. I. Fossil Wood s from the Eoce ne Clarno Formation of 01'­ ego . & M . Howard . Prin cipl es and problems of identi fication.L. 1993. Pud oc Scienrific Publish ers. 350.a demou str ation. Co mputer ident ificat ion of hardwood species. Araral. & E. For. . Fores t Prod­ ucts es. Wood anat omical charac terist ics ?f tropical species from the Pacific Region and sm. Mikroskopische Holzanatomie. Xylot om y of import ant Chilean woods. ] Scott. 4. Sa iki. Sosef. Birmensdorf. 1990.158 + plates 8-66. 3: 135-154 . VM .G. W. Prawi rohatrnodjo (eds . Rancu si. Fr ench.] Schwein uber. Lo ndo n.I AWA J.G . pear son . H. Le rnmen s (eds . Nishid a & H. Plant Resour ces of South-Eas t As ia. Soerianegara. F. & E.136.s. 199 1. R. .A review 263 I Pankhurst. P.J . Auatomie europaischer Holzer. Ad­ vane s in co mputer method s for syst ematic biology. Ha urg. Swiss Fed. 1993. Ar tificia l intell igence. Vol.JR. BUlll Forest ry For est Produ crs Re search Institute. Troc ken brodr.). IAWA Bull n. Switzerl and .Wh eeler l& Baas .). Richter.. Par sa Pa~ouh. Bull. 2: 37-40. Zug. Verlag paU~Haupr.H. Th e principles and practice of identifi cati on m et~od s in biolog y. J3 irmens­ dorf. H 1982. Edition Ziircher. Japan Forest Techni cal Associ ation . Wheeler. Richt er. [ishida (ed .he botany in the Andes II: 68. Page. In stituti on al wood co llec tio ns of rhe world . 3. Trockenbrod t. R. IAWA Bull n. Am enca (ed s. & R. Philip piue s. Federal Research Ce ntre for Forestry aud Fore st Product s. A narorny of Euro pean woods . S.J. Hong & S. Schw eingruber. Rokubaucho. No. No. Sc hwe in ruber. 15: 22 2 (Abstrac t) . Ideutifi cation of softwoods by their microscopic struc ture.zerland. Baril e. Baltimore . 1987. 5 (l )1. S. Index Xyl ariorum. Fores t Pr oduc ts Research and Developm ent Institut e . 194 8.J. H. 9: 20 3-252. University Park Press. and its t ctonic significance : Reconcili ation of paleob otani cal and paleom agnet ic ev ide nce . E nglish. Biological iden tification. S. 1987. Cambridge Unive rsity Press. Pan khurst. The struc ture of domestic and imported wood iu Japan. databases.-M .s. Japau. Tokyo. Cambridge.). K. R A. 17 : 26 2. IAWA J3ull. Prac tical taxonomi c co mputing .S . Gerrnan. Panshiu' IA . Ed. 22. Tokyo.A. Nishida. L.A . Lo ndo n. [Germ an . 1998.H. Leide n. In: M. Procee di ngs 2nd Pacific Regioual Wood Aua torny Co nfere nce : 193-208. Inst.5 64 .Wood identification . Stem . southeas tern Californ ia and southweste rn A rizona. & F. The Joh ns Hopkins Universit y Press.). 1993. M cGr aw-H ill B oo~ Company. Timb er trees: major co mmercial timb ers. M . and W.s. 1.H. Japan. E.T. Co mputer-aided wood identifi cat ion wi th DELTA ! INT Y .. HMS O. 1988. Institut e Fede ral de Rpcherches fo resti eres. Tsukuba. 1996. Backh uys Publi shers. Aday. deZe euw. Sud o. J. New York . 1985 Atlas des bois du nord de 1'1ran . Bern . . 1988. Plaut Re sources of South-East Asia No 5 (3) Timber rrees: Lesser-known timb ers. In: R Fortuner (ed. 1980. S tone.M. 1978. R J. IAWA list of featu res fo r hardwood identificat ion adapted to the DELTA syste m.). Swit.. 1AWA J. No . Phlll iPst E. M. 1982.J .A. Raja .R.w.

] Zhang. I: 341-370 + plates 52-96. Wh ite. 1991. Baas. Bull ..C.s. 22 : 21 3-217. Yang.A. T he str uctu re of Korean woods . Pal aeograph y. Takahashi . G. The Himalayan plants. N. Jap an: 174-1 76. Vol. C.s. Cheng. 474 .C. S. 1984 . In: H. J. The Himalay an plants.. 1998. 2: 17-65 + plates 5. Noshiro . Wh eeler.C. & P. Missou ri Bot. T. D. Pearson . M. Beijing For.A. Leipzig. Wheeler. Seo ul. Muse um.A. Nos hiro. Malia (eds. C.E . Wheeler. 1993. 1986. Martin o & J. In: H . Ma da mba.. Zack . A nn. 71 : 783-800. Microcomputer-assisted woo d identificati on sys tem WIP-89. Fachbuch verlag. The Identification of Ti mber s]. No . Tsukuba. 12: 271. Pacifi c Region al Wood Anatomy Co nference. E. Shiokura. W. E. The anatomy of wood: its diver sity and variability. Lian .62. Hy ang Mun Sa Publishing. Oh ba & S.. 4. E.C. Coradin. Vessel s per squ are millim etre or vessel group s per sq uare mil limetre ? IAWA Bull. IAWA Bull. & S. B. Vol. Tokyo.264 IAWA Journ al . E. R.R . S tation Bull . N. 1997. 1988. Q. Cam argos . Lon don.A.332 . n.. Salud & C.G .B . Ma ncheste r & K. 19 (3). & DJ. 1986. Wood colour . Co mputer as ­ sisted tropica l wood identificati on (CATWI ). J. Ed .M .. IAWA Bull. Raleigh. Wiemann .). Wh eel er. Suzuki. [Ch apt er 10. Van Vliet. Univ. 1998 Su zuki. Vetter. 1990 . K. Malia (eds . R. A surve y of the fossil record for dicotyled onous wood and its sig nificance for evo lutionar y and ecol ogical wood anatomy. 12 (4) : 88-94. Portier.A. Univ. Baas. Wood anatomy and classification of the Myrtales. & P. Tokyo Press. Co mpute r-aide d wood ide ntification. Lan tica n. 1990.C. Ohba & S . Cheng & YH. c .B . n. .B. 1991. Wood SlJUClUre of Himalayan plant s. Wood anato my of the Com bretaceae. F. Korea.223. 7: 73-74. & F. Paleobiology 19 : 486-497.. Joshi . Tochigi.11Je poten tials and limitations of dico tyledono us wood anato my for c lima tic reco ns truc tions. Palaeoclimat ol ogy. Blumea 25 : 141.R. IAWA J. S tobart & Son Ltd.). 1994. Dicotyledono us woo d anato mical c ha racte rs as pr ed ic tors of cli ma te. In: Proc. H olzatlas.M . Woo dcock .A . 1986 .An a tlas of Kore an woo ds . Wh eeler. Van Vlie t.a compari­ son betwe en determination meth ods. No.A. LaPasha. Univ. E. Vol. Ag .C. M. Gard .M . E. n.A.3!.s. Baas. Res. 11: 429--439. Germ any. Se rv. English summary. 1979. [Ch inese . & P. Wagenftihr.c. B. Palaeoecology 139: 83-100. Hatley & T. 1996. Wilson . R. 1986.A.C. Wood struct ure of Himalayan plants. Microcomputer identi fication of hard wood species. Univ. Occurrence of woods with a gra dation of vessel diamete r across a rin g.J. Wong Yong Lee . 15: 3 77~386 . M. Yoda & L.s. 1984 . Scientia Si lvae Sin. K. S. V. T. GJ.