M. Zanuttini and V. Marzocchi: Alkaline Chemi-Mechanical Pulping 489 Holzforschung 57 (2003) 489–495 Alkaline Chemi-Mechanical Pulp from Poplar. Relationship between Chemical State, Swelling and Papermaking Properties By Miguel Zanuttini and Victorio Marzocchi Institute of Cellulose Technology, College of Chemical Engineering, National University of Litoral, Santa Fe, Argentina Keywords Summary Alkaline chemi-mechanical pulping Poplar wood Alkaline swelling Chemical characterisation Mechanical properties Acetyl groups In order to analyse the fundamentals of alkaline chemi-mechanical pulping of hardwoods, the chemical state of the wood was related to both the swelling level of fibres and the papermaking properties of pulp. Wafers of poplar wood were alkali treated following a factorial experimental design for two variables: temperature and alkali concentration. Treated wafers were hot defibrated in a 300-mm disk mill at 15% consistency, and then refined in PFI mill at 20% consistency. Results show how fibre swelling gradually increases as alkaline action is increased. The significant improvement in tensile and tear strength of the pulp can, in great part, be ascribed to the development of fibre bonding capacity. A limited effect of ion content on cell wall swelling was found. Swelling correlates well with deacetylation level, and is a major factor in determining the tensile strength and scattering ability of the pulp. Introduction The lower quality of hardwood mechanical pulps in comparison with softwood mechanical pulps can be ascribed to various reasons. Giertz (1977) highlighted that, besides the presence of vessels and a higher content of parenchyma-cells, which are both reduced to non-fibrillar elements during refining, the typical mechanism of fines formation that takes place in softwood pulping does not occur in hardwood pulping. However, hardwood high yield pulp, with acceptable strength and optical properties, can be produced when pretreatments with alkali, alkali-peroxide or a sulphite are applied. The fundamentals of sulphite chemi-mechanical pulping have been extensively studied for both softwoods (Atack et al. 1978, 1980; Heitner and Hattula 1988; Argyropoulos and Heitner 1991) and hardwoods (Eriksen and Oksum 1981; Heitner and Atack 1983). Comparatively, fundamentals of alkaline and oxidativealkaline chemi-mechanical pulping, which are both used specifically for hardwoods, have been discussed to a lesser extent. Although it has been shown that alkaline treatment of a hardwood strongly modifies wood dynamic mechanical properties, and improves fibre length and strength of the obtained mechanical pulp (Vikstrom and Nelson 1980). Owing to the higher reactivity of hardwood hemicelluloses, these woods particularly respond to a moderate alkaline treatment. Hemicelluloses are strongly altered, and the alkali leads to a significant swelling of the cell wall. This is restricted by the network of cellulose and lignin, which are not affected by the treatment. Two characteristics of hemicelluloses are chemically modified: the content of acid groups, able to produce cation exchange, and the content of acetyl groups. Hemicelluloses are also partially removed. It has been shown that deacetylation is the main reaction. According to Zanuttini et al. (1997), this reaction consumes a great part of the alkali, and determines the wood swelling measured as Water Retention Value (WRV). It is accepted that alkaline pretreatment of hardwoods favours the production of a mechanical pulp with more entire and flexible fibres. Alkali action induces the generation of fibrillar elements and suitable fines, thus improving fibre bonding capacity (Giertz 1977). Nevertheless, according to our knowledge, the chemical effects of alkali treatment have not been systematically related to swelling and to papermaking properties. These relationships can also be useful to analyse the fundamentals of alkali-peroxide pretreatment, where alkali plays a very important role. The underlying mechanism accepted as generator of the necessary medium-length elements and fines in mechanical pulping is the peeling of the outer layer of fibres. This mechanism was originally named “rolling sleeve” by Giertz (1977). More recently Karnis (1994) and Stationwala et al. (1996), verified the effects of this mechanism by measuring the reduction in coarseness of spruce fibres during refining. The other possible mechanism was initially described by Forgacs (1963), who postulated that the unravelling Holzforschung / Vol. 57 / 2003 / No. 5 © Copyright 2003 Walter de Gruyter · Berlin · New York Brought to you by | University of British Columbia - UBC Authenticated | 172.16.1.226 Download Date | 7/30/12 6:43 PM 490 M. Zanuttini and V. Marzocchi: Alkaline Chemi-Mechanical Pulping of the fibres and the subsequent production of ribbons and fibrils is the mechanism able to produce high-quality short-fibres and fines. Nevertheless, according to Karnis (1994), the disruption of the S2 layer is not observed very often. Giertz (1977) suggested that the reduced strength of hardwood mechanical pulps, and low quality of their fines are consequences of difficulties in peeling off the primary and outer secondary layer of the cell wall. In this study, alkaline treatment effects are analysed considering: a) the visual characteristics of the pulp fractions, b) the swelling development, c) the relationship between swelling and chemical state, d) the fibre bonding capacity, e) the development of papermaking properties and f) the relationship of these properties to swelling. The possible mechanisms of generation of fines in poplar chemi-mechanical pulping in relation to the intensity of the alkaline treatment are also analysed. Material and Methods Seven-year-old poplar wood was used in this study. To achieve high homogeneity of chemical treatment, the wood was converted to wafers. These were obtained by transversal shaving of fresh, never-dried wood. The wafer preparation under these conditions produced a low effect on the fibre integrity since under microscopic observation, only 5% of fibres showed cut ends. Wafers were air dried, classified into separate fines, and mixed and divided into plastic bags for storage until used. Chemical characteristics of the wood employed have been published in previous works (Zanuttini et al. 1999). A 32 factorial experimental design was applied, that is to say two factors in three levels each: Temperature 50, 70 and 90 °C and nominal alkali concentration 0.4, 2.0 and 10.0 g NaOH l – 1 (Table 1). Two replicas for each condition were performed. The treatment time was 20 min, as previous work using highly accessible milled wood found that most of the alkali consumption and changes in chemical state of wood (acid group content and acetyl group content), as well as changes in swelling, occur within the first 20 min (Zanuttini et al.1999). Samples of 800 g, air-dried, were treated with 30 l of thermostated liquor in a closed digestor. The high liquor-wood ratio helped to reduce the variation in concentration during treatment. After a quick discharge, the material was washed, dewatered and weighed to determine the digestion yield, and, finally, wet-stored at low temperature. Based on the concentration of residual liquor, the specific alkali consumption was calculated. Refining and handsheet formation Treated wafers were refined at 15% consistency in an atmospheric discharge Sprout Waldron 300-mm disk refiner equipped with the D2A 505 plates. Similar to previous work Table 1. Conditions of the alkali treatment. The alkali concentration levels were logarithmically distributed Factors Concentration (g l – 1) Concentration (coded) Temperature (°C) Levels Low Medium High 0.4 0 50 2.0 0.5 70 10.0 1 90 (Zanuttini 1991), the mill was thermostated by continuous feeding of steam to the housing, and the pulp discharge was helped with intermittent air blowing. The number of passings was 7 for the most mechanical, and only 4 for the most intensely alkaline-treated material. Freeness was reduced to 400 –500 ml Canadian Standard Freeness (CSF). Pulps were later refined in a PFI mill at 20% consistency. Freeness was reduced to 80 –100 ml CSF. The operation in the PFI mill was done on the usual 30 g of pulp and applying the standard refining load, but it was carried out in stages. Stages of 4000 revolutions followed an initial 2000 revolution stage. In order to promote the formation of nodules and to homogenise the refining treatment, a manual mixing was applied between stages. The total number of revolutions varied from 4000 for the more severely chemically treated material up to 22 000 for the most mildly treated material, and also for the mechanical pulps. For handsheet formation, a modified SCAN standard method, already applied for other mechanical pulps (Zanuttini 1991), was followed. Grammage was increased to 95 g m – 2, a 200 mesh wire was used, and the discharge of the sheet former was restricted by increasing the discharge time of water from 4.0 to 10.0 s. By this procedure, a high retention of fines was obtained. Determination of Water Retention Value (WRV) WRV was determined on the +100 Bauer-McNett classifier fraction of the pulp with freeness near to 150 ml CSF. Chelating agent was added to classification water. To convert the fibrous fraction to its acid form (H-form), it was soaked in 0.01 mol l – 1 HCl for 1 h and then washed with demineralised water until minimal conductivity was achieved. To convert it to sodium form (Na-form), the pulp in H-form was soaked in 0.01 N NaCl for 4 h and washed with demineralised water until minimal conductivity. The centrifugation conditions were 1500 g force and 30 min. Determination of acid and acetyl group contents The content of acid groups was determined by Katz’s conductimetric titration technique (Katz et al. 1984). For the determination of acetyl groups, the method proposed by Solár et al. (1987) was used, consisting in the deacetylation in oxalic acid and G-L chromatographic determination of the released acetic acid using the propionic acid as internal standard. More details are given elsewhere (Zanuttini et al. 2000). Results Chemical consequences of treatment Figure 1 shows the response surface of alkali consumption as a function of alkali concentration and temperature. The adopted experimental design resulted in a chemical consumption that varies from 0.5 to 6.0% NaOH on wood. The levels of consumption here obtained are in accordance with published data corresponding to the treatment of meal (40/60 mesh) of the same wood (Zanuttini et al. 1999). The geometric progression adopted for alkali concentration (0.4, 2.0 and 10 g NaOH l – 1), respectively coded as 0, 0.5 and 1.0, gives a gradual increase in chemical consumption (Fig. 1). In Figure 2, wood which originally contained 3.22% acetyl groups is increasingly deacetylated as intensity of alkali treatment, expressed by alkali consumption, is increased. The specific deacetylation reaction and the dis- Holzforschung / Vol. 57 / 2003 / No. 5 Brought to you by | University of British Columbia - UBC Authenticated | 172.16.1.226 Download Date | 7/30/12 6:43 PM M. Zanuttini and V. Marzocchi: Alkaline Chemi-Mechanical Pulping 491 consider the WRV of fibre to be representative of the wood swelling brought about by the alkali treatment. Figure 3 demonstrates that swelling (WRV-Na) increases as temperature and alkali concentration are increased. A maximum WRV at the level of 180% on fibre is reached by the more intense treatment conditions. Fig. 1. Response surface of alkali consumption as function of treatment conditions. Alkali concentration is expressed in coded units. See Table 1. Fig. 3. Fibres WRV in the sodium form and in the undissociated form versus temperature for the three levels of alkali concentration. 0.4 g l – 1 ; 2 g l – 1 ; 10 g l – 1 . Fig. 2. Chemical consumption of the treatment versus deacetylation degree. The original acetyl content is indicated. 0.4 g 1 – 1 ; 2 g 1 – 1 ; 10 g 1 – 1 . Fig. 4. Swelling level versus acid group content. 50 °C ; 70 °C ; 90 °C . solution of hemicelluloses combine to reduce acetyl content. At the most severe conditions, almost the total elimination of acetyls is achieved. Approximately one relationship between alkali consumption and deacetylation degree exists. This agrees with the results previously obtained on poplar wood meal (Zanuttini et al.1997). Relationship between swelling and treatment conditions Figure 3 shows the swelling of the material, as expressed by the WRV of the long-fibered fraction (+100 Bauer-Mc Nett classification) in its Na-form. In order to analyse the swelling development, we determined WRV on pulps with similar freeness levels (150 ml CSF). Nevertheless, it has been shown that refining does not significantly change WRV of the long-fibered fraction of a mechanical pulp (Chang et al. 1981; Law et al. 1998). This means that, irrespective of the refining degree of the pulp, we can Fig. 5. Swelling degree versus deacetylation. 0.4 g l– 1 ; 2 g l – 1 ; 10 g l – 1 . Holzforschung / Vol. 57 / 2003 / No. 5 Brought to you by | University of British Columbia - UBC Authenticated | 172.16.1.226 Download Date | 7/30/12 6:43 PM 492 M. Zanuttini and V. Marzocchi: Alkaline Chemi-Mechanical Pulping Fig. 6. Light-micrograph of 50/100 Bauer-McNett fraction of the pulp corresponding to medium degree of alkali pretreatment, i.e., 2.0 g l – 1 and 70 °C, and refined up to (150 ml CSF). Ribbon-like elements can be observed. Fig. 7. Light-micrograph of 30/50 Bauer-McNett fraction of the pulp corresponding to the most drastic alkali treatment, i.e., 10 g l – 1 and 90 °C, and refined up to 150 ml CSF. Surface of fibre shows a partial detachment of fibrils. Holzforschung / Vol. 57 / 2003 / No. 5 Brought to you by | University of British Columbia - UBC Authenticated | 172.16.1.226 Download Date | 7/30/12 6:43 PM M. Zanuttini and V. Marzocchi: Alkaline Chemi-Mechanical Pulping For chemical pulps, it is well known that the content of acid groups in the ionic form has an important effect on swelling (Scallan and Grignon 1979; Scallan 1983). Nevertheless, for high yield pulps it has been shown that the effect of the presence of acid groups is variable, but generally clearly lower (Scallan and Grignon 1979). In order to assess the contribution of the ions to swelling of the poplar chemi-mechanical pulp, in this paper we compare the WRV of the fibre in the Na-form (WRV-Na) with the corresponding WRV in the undissociated proton form (WRV-H). We consider the last form as indicative of the swelling of the wall without ions. It is worth mentioning that there were not appreciable differences between WRV of the original pulps and the WRV of pulp exchanged to its Na-form. Figure 3 shows that the effect of ion content is minimal. There is some difference at medium levels of treatment where WRVNa reaches 10 points more than WRV-H, but this favourable effect of the ions disappears for the lowest and highest levels of treatment. Figure 4 plots WRV-Na as a function of acid group content. Note that there is not only one relationship between them. The existence of a maximum for each acid group content curve is ascribed to the fact that the acidgroup-containing material (xylan) is partially dissolved from the cell wall. Figure 5 shows swelling as a function of deacetylation. Only one relationship can be established between these two parameters. This is in accordance with results previously obtained for poplar wood meal (Zanuttini et al.1999). 493 fibre length classifier) has been employed by other authors to analyse the bonding capacity of chemi-mechanical pulp fibres in relation to variables of the chemical treatment (Argyropoulos and Heitner 1991). Figure 8 shows tensile strength of the 30/50 fraction corresponding to pulps at 150 ml CSF as a function of fibre swelling. Fibres begin to achieve certain strength at a medium level of swelling which corresponds to the conditions at the centre of the experimental design (2 g l – 1 and 70 °C), and they reach a considerable strength (20 Nm g – 1) for the most severe treatment condition. This high bonding capacity of the fibres means that the contribution of fines to the strength of the whole pulp is not as essential as it is for low-level treatment pulps or for purely mechanical pulp. Microscopic observation of the fractions The observation of the pulps through the light-microscope allows us to detect clear differences in relation to the degree of alkaline treatment. The 50/100 fraction of the RMP pulp (not shown here) showed rigid short fibres and cut fibres. On the other hand, at similar pulp freeness, the 50/100 fraction of the pulp corresponding to medium level of alkali treatment, clearly shows the presence of ribbon-like elements (Fig. 6). The “rolling sleeve” mechanism or the disruption of the cell wall could bring about these ribbons. Owing to their slenderness and flexibility, these elements surely play an important role in the interfibre bonding process of this pulp. Figure 7 shows that the 30/50 fraction of the pulp corresponding to the most intense treatment. Here, fibres show a partial detachment of fibrils from the outer layer. The mechanism of peeling-off is evident in this pulp and it appears as a particular external fibrillation. This was not observed in pulps with a lower level of alkali treatment. Fig. 8. Tensile strength of 30/50 fraction of Bauer Mc Nett fibre-length classifier as a function of the swelling degree. 0.4 g l– 1 ; 2 g l – 1 ; 10 g l – 1 . Bonding capacity of the fibres Tensile strength of the handsheets formed from individual fibre fractions (R14, 14/28 and 28/48 Bauer-Mc Nett Fig. 9. Tensile strength of the pulps corresponding to different treatment conditions as a function of the refining degree. 0.4 g l – 1 ; 2 g l – 1 ; 10 g l – 1 ; RMP . Holzforschung / Vol. 57 / 2003 / No. 5 Brought to you by | University of British Columbia - UBC Authenticated | 172.16.1.226 Download Date | 7/30/12 6:43 PM 494 M. Zanuttini and V. Marzocchi: Alkaline Chemi-Mechanical Pulping Pulp strength strength development brought about by the refining. Strength increases following lines with similar slopes. Tensile and tear strength interpolated at 250 ml CSF are presented in Figure 10. A strong effect of alkali treatment can be observed. It should be highlighted that alkali concentration levels, which were adopted in geometric progression, give a gradual increase in both strengths. For the nine treatment conditions employed, and also for the mechanical pulp, Figure 9 displays the tensile Relationship between pulp properties and swelling Figure 11 shows tensile strength and light scattering ability of the pulp at 150 ml CSF as a function of fibre swelling. Starting from a WRV degree of 130 g g – 1, the tensile index continuously increases from 8.0 to 58.0 N m g – 1 as fibre WRV increases. On the other hand, scattering coefficient shows a maximum at a medium level of WRV and, later on, an abrupt drop from 58.0 to 44.0 m2 kg – 1. It can be observed that swelling degree of pulps is a very significant factor in determining their strength and optical properties. Fig. 10. Tensile and tear strength of the pulp at 250 ml CSF as function of the temperature for the three levels of alkali concentration. 0.4 g l – 1 ; 2 g l – 1 ; 10 g l – 1 . Conclusions For treatment conditions corresponding to alkaline chemi-mechanical pulping, the deacetylation degree of poplar wood is a good indicator of the degree of the applied chemical action. A limited favourable influence of the ion content on the swelling of the cell wall exists. A small difference between H-form and Na-form exists at a medium level of alkali treatment, but ion content shows no influence at the lowest and the highest levels. As alkali treatment is intensified, mechanical action of refining leads increasingly to the generation of good quality fines, and also the visible detachment of fibrils from the outer layers of the fibres. The swelling of the treated fibres, as measured by WRV-Na, which correlates well with the deacetylation degree, is a major factor in determining the strength and optical properties of the chemi-mechanical pulps produced. Fig. 11. Tensile strength and light-scattering ability of pulp at 150 ml CSF as a function of the swelling degree. 0.4 g l – 1 ; 2 g l – 1 ; 10 g l – 1 ; RMP  . Table 2. Results of Anova tests and multiple regression Source Response Constant A c Consumption (%) –2.19 Acetyl groups (%) 3.7 WRV-Na (%) 80.7 –14.6 Tensile (Nm g – 1) Acid groups (mEq kg – 1) 118 –5.84 Tear (Nm2 kg – 1) 45.34 Scattering (m2 kg – 1) –5.9 30/50 Tensile (Nm g – 1) 3.553 –2.07 50.66 32.47 76.4 0.600 24.08 –2.633 A2 B F c F c 183 0.0548 70 – 98 –0.0202 15 – 143 0.6000 32 – 106 0.3300 18 – 46 – – – 287 0.1883 42 2.933 22 0.2125 0.8 –16.67 38 0.0831 12 – B2 F c – – – – – – – – – – 16 –0.001(1) 6 – – – A*B F c F – – – – – 6 – – – – – – – – –0.225(2) 0.24(3) – – – – – – 3 3 d.f. R2 6 6 6 6 7 4 5 5 97.7 95.0 96.7 95.4 86.8 98.9 90.0 91.5 A: concentration in coded units; B: temperature; c: regression coefficients; F: F-ratio; p: probability; d.f.: degree of freedom. Sources of p > 0.05 were discarded, except for (1) p = 0.071, (2) p = 0.093 and (3) p = 0.147. Holzforschung / Vol. 57 / 2003 / No. 5 Brought to you by | University of British Columbia - UBC Authenticated | 172.16.1.226 Download Date | 7/30/12 6:43 PM M. Zanuttini and V. Marzocchi: Alkaline Chemi-Mechanical Pulping Acknowledgements Financial support from UNL (Universidad Nacional del Litoral), CONICET (Consejo Nacional de Investigaciones Científicas y Técnicas) and ANPCyT (Agencia Nacional de Promoción de Ciencia y Tecnología), all from Argentina, is gratefully acknowledged. The authors appreciate the technical assistance of E. Comini Stiefel. We also thank M. Rudi for assistance with experiments as well as for helpful discussions. R. Althaus helpfully assisted with the analysis of the statistical design. Thanks are due to Professor Aldo Lossada for valuable discussions and criticism of the manuscript. Appendix Statistical analysis Table 2 gives the regression coefficients and the F-ratio as well as R2 values of the Anova test after discarding the sources of p > 0.05 for all the responses, i.e., alkali consumption, acetyl and acid group content, swelling degree, properties of the pulp and strength of the 30/50 fraction. Note that alkali consumption, acetyl group content, WRVNa and tensile strength show, with acceptable values of R2, only a linear effect of the variable A (alkali concentration in coded variables) and B (temperature). For the same degree of freedom of the error, i.e., 6, F-ratio values are comparable, which confirms that these four responses are strongly connected. References Argyropoulos, D.S. and C. Heitner. 1991. Ultra-high-yield pulping. Part VII: The effect of the pH during impregnation on the quality of lightly sulphonated CTMP. J. Pulp Paper Sci. 17(5), J137– J143. Atack, D., C. Heitner and M.I. Stationwala. 1978. Ultra high yield pulping of eastern black spruce. Svensk Papperstidning 81(5),164 –176. Atack, D., C. Heitner and A. Karnis. 1980. Ultra high yield pulping of eastern black spruce. Part 2. Svensk Papperstidning 83(5), 133 –141. Chang, H-M, D. Abson and G.Y. Pan. 1981. Improving the bonding strength of the pulps by physical and chemicals treatments. In: International Mechanical Pulping Conference. Oslo, Norway, June 1981. pp. 269 –274. Eriksen, J.T. and J.A. Oksum. 1981. Bright chemi-mechanical pulps from birch. In: International Mechanical Pulping Conference. Oslo, Norway, June 1981. pp. 277 –282. Forgacs, O.L. 1963. The characterization of mechanical pulps. Pulp Paper Magazine of Canada 64 (C),T89– T118. Giertz, H.W. 1977. Basic wood raw material properties and their significance in mechanical pulping. In: International Mechanical Pulping Conference, Helsinki. pp. 37 –51. Heitner, C. and D. Atack. 1983. Ultra-high-yield pulping of as- 495 pen, effects of ion content. Pulp and Paper Canada 84(11), T252– T257. Heitner, C. and T. Hattula. 1988. Ultra-high-yield pulping. Part VI: The effects of sulphonation on the development of fibre properties. J. Pulp Paper Sci. 14(1), J6– J11. Karnis, A. 1994. The mechanism of fibre development in mechanical pulping. J. Pulp Paper Sci. 20(10), J280– J288. Katz, S., R.P. Beatson and A.M. Scallan. 1984. The determination of strong and weak acidic groups in sulfite pulps. Svensk Papperstidning 87(6), R48– R53. Law, K.N., J.L. Valade and K.C. Yang. 1998. Fibre development in thermomechanical pulping: Comparison between black spruce and jack pine. J. Pulp Paper Sci. 24(2), 73 –76. Scallan, A.M. 1983. The effect of acidic groups on the swelling of pulps. A review. Tappi J. 66 (11), 73 –75. Scallan, A.M. and J. Grignon. 1979. The effect of cations on pulp and paper properties. Svensk Papperstidning 82(2), 40 –47. Solár, R., F. Kacik and Y. Melcer. 1987. Simple semimicro method for the determination of O-acetyl groups in wood and related materials. Nordic Pulp Paper Res. J. 2(4), 139–141. Stationwala, M.I., J. Mathieu. and A. Karnis. 1996. On the interaction of wood and mechanical pulping equipment. Part I: Fibre development and generation of fines. J. Pulp Paper Sci. 22(5), J155– J160. Vikstrom, B. and P. Nelson. 1980. Mechanical properties of chemically treated wood and chemimechanical pulps. Tappi J. 63(3), 87 –91. Zanuttini, M. 1991. Effect of alkali charge in bagasse chemimechanical pulping. Part 2. Handsheet properties. Appita 44(4), 257 –260. Zanuttini, M., V. Marzocchi, M. Citroni and M.J. Martìnez. 1997. CMP from poplar wood. fundamentals of the alkaline action (in Spanish). In: ATIPCA Conference, Buenos Aires, Argentina. p. 1. Zanuttini, M., V. Marzocchi and M. Citroni. 1999. Alkaline treatment of poplar wood. Holz Roh- Werkst. 57, 185 –190. Zanuttini, M., M. Citroni and V. Marzocchi. 2000. Pattern of alkaline impregnation of poplar wood at moderate conditions. Holzforschung 54(6), 636 –639. Received June 26th 2002 Miguel Zanuttini1) Victorio Marzocchi Institute of Cellulose Technology College of Chemical Engineering National University of Litoral Santiago del Estero 2654 3000 Santa Fe Argentina mzanutti@fiqus.unl.edu.ar 1) Corresponding author Holzforschung / Vol. 57 / 2003 / No. 5 Brought to you by | University of British Columbia - UBC Authenticated | 172.16.1.226 Download Date | 7/30/12 6:43 PM View publication stats